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 |
|---|---|---|---|---|---|
The cell envelope of mycobacteria is a highly unique and complex structure that is functionally equivalent to that of Gram-negative bacteria to protect the bacterial cell . Defects in the integrity or assembly of this cell envelope must be sensed to allow the induction of stress response systems . The promoter that is specifically and most strongly induced upon exposure to ethambutol and isoniazid , first line drugs that affect cell envelope biogenesis , is the iniBAC promoter . In this study , we set out to identify the regulator of the iniBAC operon in Mycobacterium marinum using an unbiased transposon mutagenesis screen in a constitutively iniBAC-expressing mutant background . We obtained multiple mutants in the mce1 locus as well as mutants in an uncharacterized putative transcriptional regulator ( MMAR_0612 ) . This latter gene was shown to function as the iniBAC regulator , as overexpression resulted in constitutive iniBAC induction , whereas a knockout mutant was unable to respond to the presence of ethambutol and isoniazid . Experiments with the M . tuberculosis homologue ( Rv0339c ) showed identical results . RNAseq experiments showed that this regulatory gene was exclusively involved in the regulation of the iniBAC operon . We therefore propose to name this dedicated regulator iniBAC Regulator ( IniR ) . IniR belongs to the family of signal transduction ATPases with numerous domains , including a putative sugar-binding domain . Upon testing different sugars , we identified trehalose as an activator and metabolic cue for iniBAC activation , which could also explain the effect of the mce1 mutations . In conclusion , cell envelope stress in mycobacteria is regulated by IniR in a cascade that includes trehalose .
Mycobacterium tuberculosis , the causative agent of tuberculosis disease ( TB ) , is currently the most deadly infectious agent , causing over 1 . 8 million deaths annually [1] . Although TB can be cured with a six-month treatment regimen of different antibiotics and chemotherapeutics , this disease is still a huge public health burden in large parts of the world [1] . One of the major issues in combating TB is the emergence of multi-drug ( MDR ) and extensively-drug resistant ( XDR ) strains . To combat TB caused by these new strains novel anti-tubercular agents are urgently needed . A prominent drug target is the mycobacterial cell wall , a non-canonical and lipid-rich structure [2] . One of the unique elements of the mycobacterial cell wall is the so-called arabinogalactan layer that is covalently linked to the peptidoglycan . This layer is composed of a chain of galactofuranose residues and side chains of ( branched ) arabinofuranose units . The terminal arabinofuranose residues of these side chains can be substituted with long ( C60-C90 ) carbon chain fatty acids , commonly known as mycolic acids . These mycolic acids form the inner layer of another unique element , the mycobacterial outer membrane . Mycolic acids can also be linked to trehalose , resulting in trehalose mono- ( TMM ) and trehalose dimycolates ( TDM ) . These glycolipids are hypothesized to form the outer layer of the outer membrane . The mycobacterial outer membrane also contains a variety of other ( glyco ) lipids , many of which contain a trehalose unit [3] . The success of targeting mycobacterial cell envelope biogenesis with antitubercular drugs is best illustrated by first-line antibiotics ethambutol ( EMB ) and isoniazid ( INH ) and recently discovered antibiotics like benzothiazinone ( BTZ ) . Both EMB and BTZ disrupt the formation of the arabinogalactan layer; EMB targets arabinosyltransferases EmbA , EmbB and possibly EmbC , which effectively abolishes the biosynthesis of arabinogalactan and lipoarabinomannan [4 , 5] . BTZ , in turn , inhibits arabinan biosynthesis by binding to the decaprenylphosphoryl-β-D-ribofuranose-2'-epimerase enzyme DprE1 [6 , 7] . INH enters the mycobacterial cell as a pro-drug and , after intracellular activation , binds and inhibits InhA , an essential NADH-dependent enyol-ACP reductase that is required for mycolic acid biosynthesis . Inhibition by INH causes severe cell wall deformations , ultimately leading to bacterial cell death [8 , 9] . Mycobacterial cell wall composition and biosynthesis have been studied extensively . However , the induction and function of mycobacterial cell envelope stress responses are largely unexplored . Mapping bacterial stress pathways is important , because key components that are essential for dealing with stress , for instance as a result of antibiotic treatment with cell-wall targeting antibiotics or chemotherapeutics can be identified . This knowledge can subsequently shed light on possible resistance routes that subvert the current treatment regimen . Moreover , pinpointing these stress-associated networks may lead to new targets for anti-TB therapy or lead to the discovery of antimycobacterial agents that can synergize with currently used antibiotics . Also , identification of important antibiotic-associated regulatory networks can lead to the discovery of sensitizing agents , as shown previously by work from Peterson et al . [10] . A recent example of a sensitizing agent is the recently identified ethR2-inactivating agent , which leads to reversion of ethionamide resistance [11] . Studying stress responses can also provide fundamental insights into the bacterial heterogeneity observed in patients receiving treatment . To fill the gap in our knowledge on mycobacterial cell envelope stress responses , we study iniBAC operon induction . This operon is the most abundantly induced gene cluster upon incubation with sub-lethal concentrations of EMB and INH [12] . We and others have previously shown that this operon is specifically induced as a result of cell wall stress [12 , 13] . For our studies we used the causative agent of fish tuberculosis , Mycobacterium marinum , and confirmed our findings for M . tuberculosis . A previous transposon mutagenesis study performed by our group revealed that the iniBAC operon was strongly induced by mutations in vitamin B12 biosynthesis genes or in genes encoding the vitamin B12-dependent enzyme methylmalonyl-CoA mutase ( MutAB ) [13] . Because MutAB is a crucial component of the proprionate-degradation pathway , we hypothesized that iniBAC upregulation may be linked to coping with otherwise toxic levels of metabolic intermediates or that it may influence the production of ( trehalose-containing ) branched chain fatty acids [14] . However , the genes directly involved in regulation of the iniBAC operon remained unknown . Other groups have speculated on possible regulators of the iniBAC operon , but none of them can explain the high , specific upregulation of this single operon during antibiotic stress conditions [15 , 16 , 17] . In this study , we aimed to identify the regulator of the iniBAC operon . We used a transposon mutant screen to identify the activator that is essential and specific for iniBAC regulation .
Previously , we have shown that mutants affected in vitamin B12-biosynthesis and the vitamin B12-dependent enzyme MutAB already displayed a high degree of iniBAC induction in the absence of any antibiotic [13] . These transposon mutants with upregulated iniBAC expression did not show an apparent phenotype , e . g . we did not observe an altered susceptibility to first-line antibiotics , nor a growth defect in culture [13] . As an indicator for iniBAC induction we used a construct containing the iniBAC promoter cloned in front of a promoterless gene encoding the fluorescent protein mEos3 . 1 . To identify the iniBAC regulator , we used one of the previously described vitamin B12 biosynthesis mutants , cobC::tn , carrying an integrative variant of the iniBAC reporter plasmid and performed a second round of transposon mutagenesis [13] . This time , mutant colonies that resulted from transposon mutagenesis were selected for the absence of fluorescence on 7H10 plates , as assessed manually , by fluorescence microscopy . Next , these mutants were streaked on fresh 7H10 plates and compared to a WT M . marinum containing the reporter , as a negative control . We termed these non-fluorescent mutants ‘on/off mutants’ . In total , 27 transposon mutants that repressed iniBAC induction were characterized , of which 2 could be retraced to mutations in the mEos3 . 1 gene of the reporter construct . The remaining 25 on/off mutants were all mapped to specific transposon insertion locations within the M . marinum genome ( Table 1 ) . With 5 independent transposon insertions , the mce1 operon was most prominently present in our mutant list . Although this operon is not directly involved in gene regulation , we decided to characterize these mutants in more detail because they could reveal clues on the signal transduction pathway that is involved in iniBAC induction . It has been reported that disruption of the mce1 operon in M . tuberculosis results in the intracellular accumulation free mycolic acids ( FMAs ) [18] , indicating that Mce1 is involved in the reuptake of FMAs from the cell envelope [18 , 19] . To identify whether M . marinum mce1 transposon mutants also show an aberrant lipid profile , thin layer chromatography ( TLC ) was performed to measure FMA levels . In line with these previous findings , disruption of both mce1D and yrb1EB of the mce1 operon resulted in a substantial increase in FMAs , as compared to WT M . marinum and the parent cobC::tn strain ( Fig 1A ) [18] . As a control , a double mutant affected in the cyclopropanation of mycolic acids , cmaA2::tn was included . This mutant seemed to be unaffected in the level of FMAs . In addition to FMAs , the amount of bound mycolic acids were also analyzed . Interestingly , both mce1 mutants displayed a large decrease in the amount of bound mycolic acids , as compared to the parent strain ( Fig 1B ) . We hypothesized that a decrease in bound mycolic acids might influence membrane integrity and thus cell envelope stress . To test this hypothesis , we performed an MIC assay with a selection of first and second line antibiotics ( ciprofloxacin , ethambutol , isoniazid and rifampicin ) . In the experiment , a WT M . marinum MUSA , the cobC::tn parent strain and the double mutants affected in yrbE1B and mce1D were included . However , no major differences in the MICs between the mce1D , yrbE1B transposon mutants and the respective controls were observed ( S1 Table ) . To assess cell wall permeability with a different method , we also performed an ethidium bromide uptake assay with a WT M . marinum , the cobC::tn parent mutant and the mce1D::tn + cobC::tn double mutant . We used a WT M . marinum that expresses M . smegmatis porin MspA as a positive control , because it has previously been shown to have an increased outer membrane permeability [20] . As expected , we observed a high increase in permeability for our positive control ( S1 Fig ) . However , both cobC::tn and the mce1::tn on/off mutant showed highly similar rates of ethidium bromide uptake when compared with our WT control . Therefore , we conclude that although mce1 mutants seem to be affected in the mycolic acid salvage pathway and show reduced iniBAC induction , this mutation does not seem to affect cell envelope permeability . The second most abundant set of mutations in our transposon screen were three unique insertions in one gene , MMAR_0612 . The gene product of MMAR_0612 has a predicted DNA-binding domain , suggestive of a role as transcriptional regulator ( as determined by Phyre 2 [21] ) and is located upstream of the iniBAC operon ( Fig 2 ) . Together , these characteristics make this gene a major candidate to code for the iniBAC regulator . To examine this , we tested whether the reporter construct present within the MMAR_0612 on/off mutants could still be induced by EMB ( 1 μg/ml ) or INH ( 10 μg/ml ) in culture . As a control , we included a selection of other on/off mutants ( e . g . we display one mce1 transposon mutant and one iniR::tn mutant ) . The on/off mutants were cultured with or without 1x MIC EMB ( 1 μg/ml ) or INH ( 10 μg/ml ) , both concentrations that were previously established to cause a strong induction of our reporter plasmid [13] . The response was quantified by measuring fluorescence induction with a flow cytometer ( a representative selection is shown in Fig 3 ) . Out of the 25 mutants identified we found only three mutants that were unresponsive and showed no induction of iniBAC upon antibiotic challenge . These mutants were all affected in MMAR_0612 , indicating that this is indeed the iniBAC regulator . Consequently , we propose to name the MMAR_0612 gene product IniR for iniBAC Regulator . Next , a MIC assay was performed with first- and second-line antimycobacterial compounds ciprofloxacin , ethambutol , isoniazid and rifampicin on one of the iniR::tn mutants to address whether disruption of this gene altered antibiotic susceptibility . Comparing the MIC values of iniR::tn to the cobC::tn parent strain and an sdhA1::tn ‘on/off mutant’ revealed that there are no major differences in antibiotic sensitivity ( S1 Table ) . This finding is in line with previous evidence that constitutive upregulation of the iniBAC operon did not change antibiotic susceptibility [13] . To confirm our previous results , as well as to exclude possible side-effects of the cobC::tn mutant background , we generated a targeted knockout of MMAR_0612 in M . marinum ( from here onwards termed iniRMm ) . A complementation vector that contained the putative native iniRMm promoter and iniRMm was also constructed and introduced into the ΔiniRMm strain . Subsequently , the iniBAC reporter plasmid was introduced into the ΔiniRMm and the complemented ΔiniRMm strain and cultures of the resulting strain were treated with either 1x MIC EMB or 1x MIC INH and compared to a control culture without antibiotics . On day 3 , cultures were washed and bacterial cells were analyzed for fluorescence induction by flow cytometry . The gating strategy for this experiment can be found in Fig 4A . As shown in Fig 4B–4D , the ΔiniRMm mutant showed no induction upon treatment with EMB or INH , confirming previous findings with the iniRMm::tn mutants in the cobC::tn background . Upon complementation iniBAC induction levels were restored back to near wild-type levels . In addition , we also observed a mild ( 2-fold ) repression of iniBAC in the knockout mutant grown without antibiotics , indicating that IniRMm is also required for basal production levels of the iniBAC operon ( Fig 4E ) Bioinformatic analysis of the locus of iniR and iniBAC reveals that the synteny is conserved in several mycobacterial species , including M . tuberculosis , M . bovis and M . smegmatis ( Fig 2 ) . In all these species , the gene encoding the IniR regulator is separated from the iniBAC operon by a small gene designated iniB* , because this gene shows weak homology to the 5’ end of iniB ( 32% identity over a length of 50 amino acids for the M . marinum proteins ) . The major differences in the iniBAC locus are caused by the presence and the size of the iniB gene; M . smegmatis lacks an iniB gene , whereas the iniB gene of M . marinum is , with 3000 bp , more than double the size of M . tuberculosis iniB . Moreover , M . smegmatis contains a dnaK-like gene directly downstream of iniR . Bioinformatic analysis ( NCBI BLAST and Phyre2 [21] ) of the 818 amino acids-long IniR of M . marinum revealed three distinct domains ( Fig 5 ) . The first domain is an AAA+ ATPase domain , located at the N-terminal side of the protein . This domain is predicted to be structurally highly similar to sso_1545 from Sulfolobus solfataricus , an archaeal ATPase [22] . A second domain ( from AA 470 to AA 731 ) , shows structural similarity ( Phyre2 ) to domain III of MalT , a transcription factor found in E . coli . This domain is responsible for binding maltotriose , which subsequently triggers multimerization of the transcription factor MalT [23] . In E . coli MalT is coupled to the maltose uptake system and , upon binding maltose , positively regulates transcription of the genes encoding the maltose transporter ( mal operon ) [24] . Lastly , the C-terminal portion of IniR contains a putative DNA-binding domain , showing the highest similarity to SdiA , a LuxR-like regulator in E . coli . In conclusion , the bioinformatics analysis indicates that IniR is a multidomain regulator that is responsive to external effectors . Based on our results and the bioinformatic analysis , we hypothesized that IniR is a direct activator of iniBAC . To test this , we created an anhydrotetracyclin-inducible ( ATc ) expression vector that encodes a FLAG-tagged M . marinum IniR ( IniRMm ) and introduced the construct in an M . smegmatis strain that also contained the iniBAC reporter plasmid . The resulting strain was grown in presence or absence of 10 ng/ml ATc and fluorescence induction was assessed with flow cytometry . At 2 days after induction a strong , ATc-dependent , mEos3 . 1-signal was observed and this induction was specific for the presence of the vector containing iniRMm ( named piniR in Fig 6A ) . The induction of fluorescence by ATc compared very well to that of the known inducer EMB . Moreover , addition of EMB combined with ATc boosted induction to approximately 1 . 5 times the observed induction with ATc alone ( Fig 6B ) . In order to investigate whether Rv0339c ( iniRMtb ) of M . tuberculosis is indeed the functional orthologue , we used an ATc-inducible vector encoding iniRMtb-FLAG and repeated the same experiment . With the FLAG-tagged IniRMtb we observed similar induction pattern after addition of ATc ( Fig 6C and 6D ) , indicating that IniRMtb is indeed the functional orthologue . To confirm that IniRMtb binds to the promoter region of the iniBAC operon to induce transcription , we also performed a chromatin immuno-precipitation ( ChIP ) sequencing experiment . Briefly , M . tuberculosis containing the ATc inducible vector encoding IniRMtb-FLAG was cultured in the presence of ATc . Cultures were crosslinked and bacteria lysed . Subsequently DNA-bound iniRMtb-FLAG was immuno-precipitated with an anti-FLAG antibody . DNA fragments were de-crosslinked , purified and sequencing was used to determine the binding DNA regions . The resulting reads were mapped onto the H37Rv genome and showed a very strong binding peak of IniRMtb directly in front of the iniBAC operon ( Fig 7 ) . Together these experiments show that expression of IniR is sufficient to induce the iniBAC operon . Moreover , IniRMtb can very clearly bind directly upstream of the iniBAC operon , indicating that IniR is the iniBAC activator protein in both M . marinum and M . tuberculosis . To determine the IniR regulon we used a previously generated M . tuberculosis H37Rv iniRMtb mutant . Using this strain , we performed mRNA sequencing using different growth conditions , comparing the RNA profiles of untreated , EMB- or INH-treated cultures of WT H37Rv to those of the iniRMtb knockout strain . Consistent with prior observations , the iniBAC operon was highly induced in WT H37Rv M . tuberculosis after treatment with either ethambutol or isoniazid . Knocking out iniRMtb completely abrogated the induction with both antibiotics ( Fig 8A and 8B ) , similar to our results for iniRMm ( Fig 4 ) . When comparing the expression in WT versus the iniRMtb knockout mutant , iniBAC levels were reduced ~6 fold on average in untreated conditions ( S2 Table ) , a similar repression as seen for iniRMm disruption . Values in S2 Table are fold change in RNA levels for a triplicate of samples comparing WT H37Rv to ΔiniRMtb . Treatment with EMB or INH does not increase transcription of the iniRMtb activator itself , indicating that this element is not regulating its own expression by upregulation of iniRMtb transcripts . An important observation is that only the expression of the iniBAC operon was significantly altered when iniRMtb was deleted , which implies that IniR is a highly specific and dedicated activator , only regulating this single operon . Because IniR has structural homology to the MalT domain III that is responsible for maltotriose binding , we hypothesized that the presence or uptake of ( di ) saccharides may be triggering IniR multimerization and subsequent activation of iniBAC . To test this hypothesis , WT M . marinum containing an integrated iniBAC reporter plasmid was grown in the presence of maltose , sucrose or trehalose . The introduction of the iniBAC reporter on the chromosome was chosen to simulate a more natural expression level of iniBAC , even though an integrated variant reduces the dynamic range of the induction levels than can be observed . After exposure to maltose , sucrose and trehalose , fluorescence induction was measured by flow cytometry after one , two and three days of incubation . Both 1% maltose and 1% sucrose did not induce iniBAC . However , addition of 1% trehalose caused a small but reproducible 2 . 5 fold induction ( Fig 9A and 9C ) , indicating that trehalose can induce iniBAC expression . Trehalose plays an important role in mycobacterial cell envelope biogenesis . A well-known example is the release of trehalose when TDMs are produced from two TMM molecules . To determine whether the levels of free mycolic acids were influencing iniBAC induction by trehalose , we assessed an cobC::tn + mce1D::tn mutant for induction of our fluorescent iniBAC reporter . However , the mce1D::tn double mutant showed similar induction levels as the control transposon mutants ( S2 Fig ) , indicating that the sensing and signaling of iniBAC induction is separate from the reimport of the trehalose-containing lipid TMM . We hypothesized that outer membrane permeability could be a major factor in the limited induction levels observed upon addition of 1% trehalose . Therefore , a plasmid containing the gene coding for the M . smegmatis porin MspA , was introduced [20] . Introduction of MspA has previously been shown to increase the outer membrane permeability in slow-growing mycobacteria , which was also confirmed by our EtBr uptake experiment ( S1 Fig ) [20] . The resulting indicator strain with increased outer membrane permeability showed a significant increase in the ability to induce iniBAC in response to 1% trehalose ( Fig 9B and 9C ) ; a 12-fold induction was observed after three days . Importantly , the other disaccharides tested , sucrose and maltose , showed no response . These observations indicate that trehalose can serve as a trigger for iniBAC induction when it is taken up by the cell . To substantiate these results , we screened for transposon mutants that induced iniBAC induction at low trehalose levels . For this , we created another transposon mutant library in a WT M . marinum containing the integrated iniBAC reporter . Resulting mutants were plated on 2% trehalose , which only mildly induces a GFP signal in WT bacteria . Mutants that highly expressed iniBAC were identified by fluorescence microscopy , isolated and plated on 7H10 plates containing 2% trehalose and regular 7H10 plates to select for a trehalose-specific induction phenotype . In total , 25 mutants were found to be upregulated only on plates containing 2% trehalose . Strikingly , upon analysis of the transposon insertion sites by LM-PCR and sequencing , all mutations were mapped to the PDIM biosynthesis locus ( Table 2 ) . Our group has previously shown that PDIM mutations in M . marinum result in increased permeability of the mycobacterial outer membrane [20] . In conclusion , an increased outer membrane permeability by either introducing MspA or by disrupting PDIM biosynthesis results in increased iniBAC induction by the addition of trehalose . Previously , Winder and Brennan reported that INH-treated M . smegmatis accumulates free trehalose [25] . Because we show that addition of trehalose to mycobacterial cell can induce iniBAC , we hypothesized that addition of both EMB and INH , cause the accumulation of trehalose levels . In addition , we also included the drug candidate PBTZ169 . BTZ has a similar effect as EMB , disrupting biosynthesis of the arabinogalactan layer by inhibiting DprE1 [6 , 7] . Because we did not know whether BTZ induces the iniBAC operon we first used our reporter assay to address this question . As can be seen in S3A Fig , 1 ng/ml BTZ caused a mild growth inhibition ( S3A Fig ) . Subsequently , we used this concentration as well as a ten-fold dilution ( 0 . 1 ng/ml ) to treat cultures of WT M . marinum containing our iniBAC reporter construct and measured fluorescence induction with flow cytometry . After addition of 1 ng/ml BTZ a high induction of fluorescence was observed over time , when compared to an untreated control ( S3B–S3D Fig ) . To measure the effect of chemotherapeutic agents on free trehalose levels , we compared untreated cultures to cultures treated with either EMB , INH or BTZ and isolated trehalose with a hot water extraction after 3 and 6 hours . Cells were adjusted by OD600 , as well as normalized to protein content . As shown in S4A Fig , there is a visible increase in the amount of free trehalose for INH , EMB and BTZ when compared with untreated cultures . However , the signal for INH was highest for both time points ( S4B Fig ) . Because we observed that an increase in permeability increased the ability of trehalose to induce iniBAC we assume that the trehalose enters from the outside into the periplasm . To address whether uptake of trehalose over the inner membrane is required for stress signal transduction , we used a transposon mutant , sugA::tn , that is defective in the only known inner membrane transporter of trehalose , i . e . the LpqY-SugA-SugB-SugC complex [26] . To measure whether induction is hampered in a sugA::tn mutant , we introduced our exosomal iniBAC reporter into this strain and assessed whether exogenous trehalose can still induce the iniBAC operon . We used a WT M . marinum with the exosomal stress reporter as a control and the gating strategy for our flow cytometry experiment can be found in S5A Fig . These experiments showed that the sugA::tn mutant can still readily induce iniBAC in response to 1% trehalose ( S5B Fig ) . We also noticed that the operon was still inducible by EMB and INH ( S5B Fig ) . When comparing the fold induction levels ( corrected for the MFIs of the respective untreated controls ) over time , the sugA::tn mutant induces iniBAC with at a comparable level as a WT M . marinum . On day 3 the sugA::tn mutant shows a higher fold induction ( 12 . 8 fold ) compared with WT ( 9 fold ) , perhaps reflecting a higher concentration of periplasmic trehalose ( S5C Fig ) . These combined results show that antimycobacterial compounds that target the cell wall can indeed lead to an increase in intracellular trehalose . We show that trehalose probably does not have to be transported across the cytoplasmic membrane , indicating that the sensing system required for induction is located in the periplasm . Based on bioinformatics analysis ( BLAST and Phyre2 [21] ) we found that IniR contains two domains ( the ATPase domain and the MalT domain III ) that suggested possible multimerization . We hypothesized that trehalose could be required for multimerization , as maltose is required for MalT tetramerization in E . coli [23] . To address this , we constructed a Strep-tagged iniRMtb and cloned it into an ATc-inducible vector ( to generate pEXCF-iniRMtb-Strep ) . We transformed the plasmid into M . smegmatis and induced expression . Purified protein fractions were separated on an SDS-PAGE gel and stained with Coomassie ( Elution fractions ( 1–5 ) can be found in S6 Fig ) . As shown in Fig 10 , higher order structures can be identified in all tested conditions . The extra bands are approximately at the expected heights of dimers ( 180kDa ) and tetramers ( 360kDa ) . Multimerization did not seem to be dependent on the exogenous addition of trehalose in this experimental setup . We also performed the experiment with a crude cell lysate of M . smegmatis that overexpressed FLAG-tagged IniRMtb . However , also in this experiment we observed multimers independent of the presence of trehalose or ATP ( S7 Fig ) . In summary , we see that IniR readily forms oligomeric structures similar to MalT in E . coli [23 , 24] .
In this work , we describe the identification of the iniBAC regulator IniR , by performing transposon mutagenesis in a strain that highly expressed iniBAC by default . In our first publication on the regulation of the iniBAC operon we identified a strong link between upregulation of the iniBAC operon and the vitamin B12 dependent enzyme MutAB . This enzyme is a central component of the methylmalonyl-CoA pathway , a pathway that is involved in the degradation of fatty acids [13 , 14] . At the time , we attempted to identify a metabolic cue causing iniBAC activation . However , exogenously adding different kinds of fatty acids or small metabolites present or linked in this pathway did not influence iniBAC expression . In this work , trehalose was identified as a metabolic signal that activates the iniBAC operon . Moreover , increasing the permeability by introducing MspA ( Fig 9B ) or blocking PDIM production ( Table 2 ) strongly increased the induction potency of trehalose . This finding also confirms our previous work that showed that PDIM mutations can allow for the essential uptake of nutrients when the ESX-5 system is deleted [20] . In mycobacteria , trehalose is used to synthesize a range of important outer membrane lipids , including trehalose monomycolates , trehalose dimycolates , penta-acyl trehaloses ( PATs ) , diacyl trehaloses ( DATs ) and sulfolipids ( only in members of the M . tuberculosis complex ) [27 , 28] . Because of its importance , trehalose biosynthesis and metabolism has been extensively studied [29] . These studies showed that trehalose is recycled from TMMs , TDMs , PATs and DATs and import occurs via the LpqY-SugA-SugB-SugC system , identified and described Kalscheuer et al [26] . This system , upon knockout , causes attenuation of M . tuberculosis in vivo . Mycobacteria possess three separate biosynthesis routes for trehalose , a redundancy that also illustrates the importance of this disaccharide . The best studied route is the OtsA-OtsB pathway that synthesizes trehalose via glucose-6-phosphate and UDP-glucose intermediates [29] . Another pathway is the TreY-TreZ pathway that can synthesize trehalose from α ( 1→4 ) glucose polymers [30] . However , the glyoxylate shunt can also produce trehalose de novo from sucrose or maltose via the TreS pathway [26 , 31] . The most intriguing mutants linking iniBAC induction to trehalose are the multiple mce1 operon transposon mutants . These mutants show repression of iniBAC in the highly expressing cobC::tn strain . The mce1 operon , suggested to be involved in the salvage pathway mycolic acids , was reported to show a high accumulation of FMAs in M . tuberculosis [18] . We not only confirmed this finding for M . marinum , but also identified a strong decrease in bound mycolic acids , suggesting that mce1 disruption has a pleiotropic effect in M . marinum . This multitude of effects of mce1 disruption on the mycobacterial lipidome has also been illustrated in M . tuberculosis by previous work by Queiroz et al . who found that disruption of mce1 significantly changes the abundance of over 400 lipid species [32] . An explanation for the decreased amounts of bound mycolic acids could be that the enzymatic reaction leading to increased amounts of FMAs is influenced by the amount of free mycolic acids . Another possible explanation comes from recent work by Ekiert et al . , who provided the first structural insights into the function of the Mce proteins [33] . These authors showed that the E . coli mce homologues encode proteins that can from a hexameric complex that mediates lipid transport . They propose , based on their structural studies , that the mycobacterial mce orthologues may very well perform a similar function in lipid transport [33] . These structural predictions may explain the absence of FMA transport by mce1 mutants . Without the shuttling of FMAs , the recycling of trehalose-containing lipids and the concomitant transport of free trehalose would also decrease . In line with this theory , we observed that the mce1D::tn mutant is still inducible with free trehalose ( S2 Fig ) . More evidence for links between TMM recycling and iniBAC induction comes from a recent paper by Degiacomi et al . , who examined the effect of a knockdown of the essential inner-membrane TMM-flipase MmpL3 [34] . This procedure resulted in the upregulated expression of the iniBAC operon . Whether the induction is caused because of a lack of TMMs or an overflow of trehalose by breakdown of TMMs that are not flipped over the inner-membrane is unknown . These combined observations could also explain why iniBAC induction is observed when EMB and INH are used , in both cases there is a strong disruption of outer membrane biogenesis and a concomitant activation of the trehalose salvage pathway . Interestingly , we managed to confirm previous findings by Winder and Brennan who established that there is an accumulation of free trehalose as a result of isoniazid treatment [25] . We extended this observation to BTZ and EMB , further supporting that iniBAC induction could be mediated through an increase in trehalose . The IniR protein contains three domains that together make it a member of the small family of signal-transduction ATPases with numerous domains , with a domain composition similar to MalT of E . coli ( 62% ) . In E . coli MalT is bound to the maltose uptake system and released upon the binding of maltose [23 , 24] . Binding maltotriose causes MalT to multimerize and leads to binding and transcription of the genes encoding the maltose transport system . For IniR the signaling molecule could be trehalose . These findings reveal trehalose as a unique cell wall stress signaling molecule that has not previously been described as such in gram negative bacteria . Gram negatives are known to activate their cell wall stress signaling cascades through the detection of mistranslated proteins or aberrant folding of membrane proteins , but not through disaccharide metabolites [35] . Future experiments should address the question whether iniR can sense or bind trehalose . Our data on the oligomerization of IniR at least suggests that it occurs in dimers and/or tetramers and that trehalose is not required for oligomerization of IniR . Understanding the interaction of IniR with trehalose or other binding partners will shed light on the function of the iniBAC operon and will allow us to address why IniR is so specific , regulating only the iniBAC genes . The discovery of IniR also sparks further speculation on IniBAC function . Both IniA and IniC contain the same ATPase domain ( P-Loop NTPase domain ) that is found also in the membrane-associated MalK , the ATP hydrolyzing unit of the maltose transporter that opens upon binding of maltotriose [36] . Studies on the localization of the IniBAC proteins and their association with IniR will allow us to address possible homology between the E . coli maltose uptake system and IniBAC . Our RNA sequencing data indicates that the stoichiometry of the system might favor a high amount of IniB molecules , as compared to the IniA and IniC counterparts of the operon . This high production of IniBAC proteins is also suggestive of a functional link between IniB , IniA and IniC , where all three are needed to mount an effective stress response . Another extrapolation of IniA and IniC protein function comes from the high similarity of the GTPase domain to dynamin family proteins . This may suggests that both proteins can function as vesiculation proteins , shielding the cell from possible toxic cell wall or lipid intermediates . Bacterial dynamins have already been shown to form tubulations in lipid bilayers in a GTP-dependent manner in Nostoc punctiforme [37] . The role of IniB in this process is unclear , especially because the presence and size of this protein is varying extensively between species . Sequence analysis indicates that the N-terminal part of IniB is highly conserved and possibly contains a domain that is required for interaction or secretion . Studying these iniBAC genes , but also other genes that are strongly induced by antibiotics may very well lead to the discovery of essential stress coping routes in mycobacteria that can be targeted by novel anti-TB therapeutics .
The E . coli strain ST08 ( Clontech ) was used to propagate plasmid DNA . Bacterial cultures were routinely grown at 37°C in LB broth with the addition of antibiotics kanamycin ( 25 μg/ml ) , hygromycin B ( 50 μg/ml ) or streptomycin ( 30 μg/ml ) , where required . M . marinum wild type MUSA , as described by Abdallah et al . [38] , was used for the transposon mutagenesis experiments and generation of the iniR ( MMAR_0612 ) knockout . Cultures were routinely grown in Middlebrook 7H9 , supplemented with Middlebrook ADC and 0 , 05% Tween-80 . M . smegmatis MC2155 was used for the modified reporter assay and cultured identically to M . marinum . M . tuberculosis H37Rv WT and an isogenic ΔRv0339c mutant , generated using specialized transduction with a phage kindly provided by Michelle Larsen and Bill Jacobs , were used for RNA sequencing experiments and cultured in 7H9 , supplemented with ADC , 0 , 05% Tween-80 and 0 . 2% glycerol [39 , 40] . Additional antibiotics isoniazid ( Sigma ) and ethambutol ( Sigma ) were added , where indicated , at mid-logarithmic phase . For isoniazid 10 μg/ml was routinely added , whereas for ethambutol 1 μg/ml was used ( both written as 1x MIC [13] ) . For experiments with PBTZ169 1 ng/ml was used as 1x MIC . The PBTZ169 was provided by Stewart Cole ( EPFL . Lausanne ) . For the modified reporter assay anhydrotetracycline , or ATc , ( Sigma ) was added to indicated final concentrations . For mycobacterial experiments on solid medium , Middlebrook 7H10 solid agar supplemented with Middlebrook OADC was used . Trehalose , maltose and sucrose were purchased from sigma and added to the final concentrations indicated in the figures . Both mycobacterial cultures and plates were grown at 30°C . All the strains that were used in this study , besides the ones derived from our transposon screens , can be found in S5 Table . All the primers used in this study can be found in S3 Table . All plasmids used in this study can be found in S4 Table . Plasmids pSMT3-iniB4-mEos3 . 1 and pMV-iniB4-mEos3 . 1 were previously described by Boot et al . [13] . Plasmid pEXCF-Rv0339c-FLAG has been previously described in [41] . For clarity within this manuscript we refer to this construct as pEXCF-iniRMtb-FLAG For complementation experiments , iniRMm ( MMAR_0612 ) was amplified from M . marinum MUSA gDNA . A fragment of 2786 bp was generated with primers iniRMm-Comp_FW and iniRMm-Comp_RV containing the promoter and iniRMm gene . Primers were designed to include 15-bp overlapping regions with the target vector , a pMV361 derivative described previously [13 , 42] . All PCRs were performed with iProof master mix ( Bio-Rad ) . The pMV vector was digested with HindIII . Subsequently , InFusion ( Clontech ) was used according to manufacturer’s protocol to generate the complementation construct pMV-priniRMm . The construct pEXCF-IniRMtb-FLAG was digested with BsrGI and used for the introduction of iniRMm . Primers iniRMm-FLAG_FW and iniRMm-FLAG_RV were used to amplify iniRMm from M . marinum MUSA gDNA with 15bp overlapping regions with the pEXCF vector backbone . The resulting construct , pEXCF-iniRMm-FLAG was generated with InFusion reagents . The plasmid containing iniRMtb-Strep ( Rv0339c ) was constructed similarly . pEXCF- iniRMtb-FLAG was digested with BsrGI . Subsequently iniRMtb was amplified with primers iniRMtb-Strep-FW and iniRMtb-Strep-RV , containing a strep-tag in a first round of PCR . This product was purified and used as a template for primers with a 15bp overhang ( iniRMtb-BsrGI-FW and iniRMtb-BsrGI-RV ) with the pEXCF vector backbone . This product was again purified and ligated into the digested pEXCF backbone with InFusion , resulting in construct pEXCF- iniRMtb -Strep The selected transposon mutants cobC::tn , cobC::tn+mce1D , cobC::tn+yrbE1B::tn and cobC::tn+cmaA2::tn and a WT M . marinum MUSA were pre-cultured , diluted and grown to an OD600 of 1 . 0 . Subsequently , 50 OD units were collected by centrifugation and the resulting pellets were washed three times with PBS . The pellets were weighed and adjusted accordingly to provide equal pellet masses . Subsequently , extraction of the cell envelope lipids was performed in three steps to harvest different lipid fractions: the apolar , polar and mycolic acids as previously described by Carrère-Kremer et al . and Minnikin et al . [43 , 44] . Both FMAs and bound mycolic acids were analyzed by 1D-TLC . For each sample , equal amounts of lipid fractions were loaded onto a silica-60 TLC plate ( Merck ) and separated by 1D-TLC . For MAMEs and FAMEs analysis running solvent hexane:ethylacetate 19/1 ( v/v ) was used . FMA analysis running solvent hexane:di-ethyl ether:HAc 70/30/1 ( v/v/v ) was used ( as previously described by Ritu Bansal-Mutalik and Hiroshi Nikaido [45] . Subsequently , the lipids were visualized with 5% molybdophosphoric acid ( MPA ) in methanol and TLC-plate charring at 160°C for 10 minutes . For MIC assays M . marinum MUSA WT and the indicated transposon mutants were grown in 7H9 supplemented with ADC to an OD600 of 0 . 6–1 , containing the appropriate antibiotics . Serial two-fold dilutions of ciprofloxacin , ethambutol , isoniazid , streptomycin and rifampicin ( all from Sigma ) were added per well in a 96-well plate . Per well , 104 bacterial cells were added . The 96-wells plate was inoculated for 7 days before the minimal inhibitory concentration ( MIC ) was determined . This concentration is described as the concentration of antibiotics that showed visible growth inhibition . Wells without the addition of antibiotics and 7H9 medium itself were used as a growth control and experiments . All the growth assays were performed in triplicate . The uptake of ethidium bromide in transposon mutants cobC::tn and mce1D::tn + cobC::tn was compared to WT M . marinum MUSA . As a positive control , we overexpressed mspA in WT M . marinum MUSA using plasmid pSMT3-hsp60-mspA [20] . Strains were grown in 7H9 with ADC and 0 . 02% tyloxapol , cell were washed with PBS containing 0 . 02% tyloxapol and diluted to OD600 of 1 . 0 . Then 180 μl of bacterial cells were distributed per well in a 96-wells microtiter plate . Subsequently , ethidium bromide was added to a final concentration of 5 μg/ml . The fluorescence was determined at 30°C using a plate reader ( Biotek , excitation: 300 nm , emission: 605 nm , bottom-reading mode ) . Measurements were performed in quadruplicate . Transposon mutagenesis was performed in M . marinum MUSA containing pMV-iniB4-mEos3 . 1 The strain was infected with the mycobacterial phage phiMycoMarT7 that contains the Himar1 transposon with kanamycin resistance , as described previously by Sassetti et al . [46 , 47] . Transposon mutants were assessed for increased mEos3 . 1 expression , indicating induction of iniBAC , manually by fluorescence microscopy . For the permeability mutagenesis screen , resulting mutants were plated on 7H10 plates containing 2% trehalose . Colonies that showed a vast increase in fluorescence as assessed by fluorescence microscopy were isolated and the transposon insertion site was determined by ligation-mediated PCR ( LM-PCR ) as described previously [38] . A cobC::tn mutant was used for a second round of mutagenesis , this time with a Himar transposon containing a hygromycin resistance cassette [46 , 47] . Colonies that showed no or severely decreased mEos3 . 1 expression on 7H10 plates were selected manually by fluorescence microscopy . These colonies were streaked on fresh 7H10 plates and compared to WT M . marinum containing the reporter construct for comparison of fluorescence . Mutants that showed no iniBAC expression were isolated and LM-PCR was used to identify the transposon insertion site , utilizing the cobC::tn parent strain as a control . The induction of the iniBAC was routinely assessed by flow cytometry on a BD Accuri C6 flow cytometer ( BD biosciences ) . For induction experiments bacteria were grown to a mid-log phase and diluted to an OD of 0 . 2 . Subsequently , antibiotics or disaccharides were added at indicated concentrations for a specified amount of time . Time point analysis was performed by sampling 1 ml of culture . The sample was washed in PBS containing 0 . 05% Tween-80 . The bacteria were then spun down and the resulting bacterial pellet was in PBS containing 0 . 05% Tween-80 . Flow cytometry analysis was performed with a 488 nm laser and 530/30 nm filter for mEos3 . 1 . Per sample , 30 . 000 gated events were analyzed per sample per time point and data was analyzed and visualized using BD CFlow software and Graphpad Prism 6 . For samples the mean fluorescence intensity ( MFI ) was used to quantify fluorescence intensity . The iniRMm ( MMAR_0612 ) knockout was created via allelic exchange with the phAE159 temperature-sensitive phage method in M . marinum MUSA . Details on the origin of the phage system can be found in the manuscript by Jain et al . and further details on the method can be found in Phan et al . [40 , 48] . The knockout construct was generated by PCR by amplification of left ( iniRMm_L-FW+ iniRMm_L_RV ) and right flanking ( iniRMm_R_FW + iniRMm_R_RV ) regions of iniRMm , covering a total of 82% of the gene ( see S1 Table ) . The deletion was confirmed by PCR analysis and sequencing . Subsequently , the hygromycin resistance cassette was removed by γδ-resolvase ( TnpR ) and counter-selected with SacB for sucrose sensitivity . An M . smegmatis strain containing the plasmid with pEXCF-iniRMtb-FLAG was grown in 7H9 liquid medium supplemented with ADC , 0 . 05% Tween-80 until mid-logarithmic phase , after which the cells were washed and inoculated in 7H9 medium with or without 10 ng/ml ATc to an OD600 of 0 . 2 and grown for another 18 h . Subsequently , bacterial cultures were spun down for 10 min at 6 , 000 × g , washed with phosphate-buffered saline ( PBS ) , and resuspended in PBS to a concentration of 20 OD units/ml . Cells were lysed by bead-beating with 0 . 1mm silica beads 3 times for 20 seconds each . Then 25 mM MgSO4 was added , along with DNaseI ( Thermo Fischer ) , to degrade chromosomal DNA and incubated for 1 hour at 37 degrees Celsius . Samples were spun down for 15 minutes at 16 . 000 x g to isolate soluble proteins . Subsequently , samples were split in equal volumes and trehalose ( 1% ) and/or ATP ( to 1 mM ) were added to the appropriate samples . The resulting mixture was crosslinked with 1% formaldehyde for 30 minutes , on ice and quenched with 100 mM cold glycine for 30 minutes . The purified IniRMtb-Strep protein was acquired by expressing the ATc-inducible vector containing pEXCF- iniRMtb-Strep to 10 ng/ml ATc for 18 hours . 50ml cultures were spun down and cell pellets were resuspended in 100 mM Tris-HCl pH 8 . 0 , 150 mM NaCl . The cells were lysed at 1 kilobar with the Stansted cell homogenizer . Unbroken cells were removed by centrifugation ( 3000 x g ) , after which the lysate was centrifuged for 45 minutes at 200 . 000 x g to remove insoluble cell components . Strep-tagged IniRMtb was purified from the soluble fraction by StrepTactin beads ( iba-lifesciences ) and eluted with 10 mM desthiobiotin , 10% glycerol , 50 mM Tris-HCl pH 8 . 0 and 150 mM NaCl . The protein fractions were separated on any kD SDS-PAGE gels ( BioRad ) and transferred to a nitrocellulose membrane . Membranes were stained with anti-FLAG M2 ( Abcam ) antibodies or anti-Strep-tag-II ( Abcam ) The secondary antibody was a goat anti-mouse antibody coupled to a peroxidase ( Abcam ) as secondary antibody for the anti-FLAG M2 and a goat anti-rabbit antibody coupled to a peroxidase ( Abcam ) as secondary antibody for the anti-Strep antibody . Nitrocellulose blots were developed with ECL Western Blotting substrate ( Pierce ) and chemiluminescence was visualized with a Amersham 600 imager ( GE healthcare ) . ChIP was done following protocols previously used ( PMID: 25581030 and PMID: 23823726 ) from [41] . Briefly , 50 ml of log phase M . tuberculosis ( containing pEXCF-iniRMtb-FLAG ) culture was cross-linked with 1% formaldehyde for 30 minutes , then quenched with 250 mM glycine . Cells were pelleted , washed in PBS with protease inhibitor ( Sigma ) and resuspended in ChIP buffer ( 20 mM HEPES pH 7 . 9 , 50 mM KCl , 0 . 5 mM DTT and 10% glycerol ) with protease inhibitor . Samples were lysed in Lysing Matrix B tubes with three rounds of bead beating at maximum speed for 30 s . Beads were pelleted , supernatants removed , and ChIP buffer added to a final volume of 500 μl . A Covaris S2 ultrasonicator was used to shear chromatin to ~200 bp fragments . Samples were adjusted to buffer IPP150 ( 10 mM Tris-HCl—pH 8 . 0 , 150 mM NaCl and 0 . 1% NP40 ) and immuno-precipitated by incubating samples overnight rotating at 4°C with 10 mg M2 anti-FLAG antibody . Samples were then incubated with protein G-coupled agarose beads for 30 min at 4°C and then 90 min at room temperature . Samples were pelleted for 2min at 2 , 000xg , supernatant discarded , washed five times in IPP150 buffer , then twice with TE , pH 8 . 0 . Protein complexes were eluted off the beads with elution buffer ( 50 mM Tris-HC pH 8 . 0 , 10 mM EDTA and 1% SDS ) for 15 min at 65°C . Eluted protein–bead complexes were treated with TE pH 8 . 0 and 1% SDS for 5min at 65°C , then digested and de-crosslinked with 1 mg/ml Pronase for 2 h at 42°C followed by 9 h at 65°C . Finally , immuno-precipitated DNA was purified using QiaQuick PCR columns . Transcriptional RNA profiles between H37Rv wild type and a iniRMtb ( ΔRv0339c ) were compared in triplicate . Cultures were grown until an OD600 of 0 . 2 and at that point INH or EMB was added to the cultures and allowed to respond to drug stress for 24 hours . Cultures without antibiotic treatment served as a control . Subsequently RNA was isolated using standard methods described previously [49] and rRNA was depleted using RiboZero Gold ( Illumina ) RNA sequencing was done using Illumina Nextseq with libraries generated using the NEBNext directional library kit ( NEB ) . Alignment to the H37Rv genome was done using Bowtie 2 . 0 [50] , read depth was determined using DuffyNGS tools ( available at https://github . com/sturkarslan/DuffyNGS ) , and visualized using Arraystar ( DNASTAR ) . For isolation of trehalose from WT M . marinum cells were grown in 7H9 containing 0 . 2% glycerol , ADC and 0 . 05% Tween-80 . Subsequently , cells were washed thrice with 7H9 containing 0 . 2% glycerol and 0 . 05% Tween-80 and diluted to OD 0 . 1 and split into 50 ml cultures ( again in 7H9 containing 0 . 2% glycerol and 0 . 05% Tween-80 ) . Antibiotics were added to final concentrations of 1 ng/ml for pBTZ169 , 1 μg/ml for EMB and 10 μg/ml for INH . An untreated control was taken along for reference . Directly after inoculation , one culture per tested condition was spun down at 4 . 000 x g and the pellet was washed with PBS and stored at -20°C . This was time point was considered 0 hours . Subsequently , after 3 hours and 6 hours cultures were sacrificed and pellets were isolated , washed with PBS . The dry pellets were further adjusted on a microbalance according to pellet weight . Subsequently bacterial pellets were resuspended in distilled H2O ( 20 OD units/ml ) . To further normalize samples , protein content was measured and equalized for each sample before loading it on TLC plates . To this end , 250 μL of cell suspension was mixed with glass beads ( 0 . 1 mm diameter , zirconia/silica ) and cells were disrupted with a beadbeater for 1 min . After briefly settling the beads ( by leaving them for 1 min at RT ) , the protein content of the supernatant was determined using a BCA protein assay kit ( Pierce ) . The free sugars of the bacterial cells were extracted by boiling of 250 μL of the pellet suspension in water at 99°C for 10 min . Afterwards , the cellular debris were removed by centrifugation at 16 , 000 x g for 5 min . The supernatant containing the hot water extracts was transferred into new tubes and normalized according to protein content as measured with the BCA protein assay kit . Hot water extracts were spotted on TLC plates ( HPTLC , silica gel 60 ) and addition of trehalose and glucose served as reference ( 1 μl of a 1 mM solution of both was spotted ) . The sugars were separated using the solvent system acetonitrile–water ( 7:3 ) . The sugars were visualized by spraying anthrone solution ( 10 mM anthrone in concentrated sulfuric acid ) and subsequent charring at 110°C until the controls became visible . The quantification of the trehalose signals were calculated using GelQuant . NET V1 . 7 . 8 ( available at http://www . biochemlabsolutions . com ) . Intensity of the trehalose signals were corrected for background and the intensities of treated samples were divided to their corresponding untreated time point controls to get fold change intensities . | The mycobacterial cell wall is a complex and unique structure that protects extremely well against harmful compounds . Understanding the biogenesis and functioning of this cell envelope is essential to be able to effectively target mycobacteria . One way to uncover cell envelope functionality is to study stress mechanisms that are induced when the cell envelope is damaged . Here , we describe the identification of a major cell envelope stress regulator and the inducing signal . As stress inducers we have used antimycobacterial drugs that target the biogenesis of the mycobacterial cell envelope , as these have previously been shown to specifically induce the major cell wall stress operon iniBAC . We have identified a multi-domain regulator that is essential for the induction of this operon to transduce cell envelope stress and named this IniR . Importantly , we were also able to show that cell envelope stress signaling was induced by free trehalose . Trehalose is a central unit in many mycobacterial lipids and mycobacteria have a dedicated trehalose salvage pathway that is used when lipids are degraded and recycled . We hypothesize that lipid turnover and concomitant release of free trehalose in the cell envelope is a signal for cell envelope stress in mycobacteria . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"antimicrobials",
"medicine",
"and",
"health",
"sciences",
"chemical",
"compounds",
"disaccharides",
"drugs",
"microbiology",
"operons",
"carbohydrates",
"organic",
"compounds",
"regulator",
"genes",
"antibiotics",
"genetic",
"elements",
"gene",
"types",
"dna",
"pharmacol... | 2017 | Cell envelope stress in mycobacteria is regulated by the novel signal transduction ATPase IniR in response to trehalose |
Variable numbers of tandem repeats ( VNTR ) typing is widely used for studying the bacterial cause of tuberculosis . Knowledge of the rate of mutation of VNTR loci facilitates the study of the evolution and epidemiology of Mycobacterium tuberculosis . Previous studies have applied population genetic models to estimate the mutation rate , leading to estimates varying widely from around to per locus per year . Resolving this issue using more detailed models and statistical methods would lead to improved inference in the molecular epidemiology of tuberculosis . Here , we use a model-based approach that incorporates two alternative forms of a stepwise mutation process for VNTR evolution within an epidemiological model of disease transmission . Using this model in a Bayesian framework we estimate the mutation rate of VNTR in M . tuberculosis from four published data sets of VNTR profiles from Albania , Iran , Morocco and Venezuela . In the first variant , the mutation rate increases linearly with respect to repeat numbers ( linear model ) ; in the second , the mutation rate is constant across repeat numbers ( constant model ) . We find that under the constant model , the mean mutation rate per locus is ( 95% CI: , ) and under the linear model , the mean mutation rate per locus per repeat unit is ( 95% CI: , ) . These new estimates represent a high rate of mutation at VNTR loci compared to previous estimates . To compare the two models we use posterior predictive checks to ascertain which of the two models is better able to reproduce the observed data . From this procedure we find that the linear model performs better than the constant model . The general framework we use allows the possibility of extending the analysis to more complex models in the future .
Mycobacterium tuberculosis , the bacterial pathogen that causes tuberculosis , latently infects one third of the world's population and is responsible for the highest mortality rate of any single bacterial pathogen [1] . Recent advances in genotyping techniques have increased our ability to discriminate among M . tuberculosis isolates , helping to shed light on the genetic diversity , demographics and evolution of this pathogen [2] , [3] . For instance , Pepperell et al . [4] , [5] suggested that the restricted diversity in this bacterial species is likely the result of population bottlenecks and founder effects . Genotyping or fingerprinting also refines our understanding of the epidemiological characteristics of the disease in a population , for example by revealing the extent of local transmission and factors associated with this transmission ( e . g . , [6] ) . Frequently used methods for genetic fingerprinting of M . tuberculosis include restriction fragment length polymorphism typing based on mobility of the insertion sequence IS 6110 [7] and spoligotyping which exploits variation at the Direct Repeat or CRISPR locus [8] . More recently , a multilocus typing method based on variable numbers of tandem repeats ( VNTR ) has been developed for M . tuberculosis [9]–[11] . These loci are minisatellites , and are also known as mycobacterial interspersed repetitive units ( MIRUs ) . We will refer to these as “VNTR loci” . VNTR-based methods are increasing in importance and efforts are being made to standardise the loci used [9] . The larger the number of loci used , the greater the discrimination among isolates resulting in a large number of smaller clusters of identical profiles in a sample . The early standard of 5 locus VNTR typing lacked the discriminatory power of IS6110-typing but comparative studies have shown that using at least 12 loci can have comparable or better discrimination relative to IS6110 [12]–[14] . An advantage of using VNTR is that if the mutation rate is low there is the possibility of adding more loci to increase discriminatory power [10] . Inferences about transmission are sensitive to the degree of genetic clustering , which is a function of the mutation rate of the marker [15] . It is therefore important to have accurate estimates of the mutation rate of VNTR loci . Knowledge of the mutation rate of VNTR also allows calibration of the molecular clock to make inferences about the evolutionary history of M . tuberculosis , for instance , the time until the most recent common ancestor of a clade [3] . A standard model for the evolution of VNTR loci is the stepwise mutation model [16] , [17] , which has successfully been used to describe microsatellite evolution in eukaryotes ( e . g . [18] ) . The stepwise mutation model has also been applied to VNTR evolution in M . tuberculosis [19] , leading to estimates of the rate of mutation . Such estimates in the literature vary widely from per locus per year [19] to per locus per year [3] to – [20] . This wide variation in estimates has led to debate in the literature [21]–[24] . Taking a model-based approach can help to resolve this question . It allows our understanding of biological mechanisms underlying VNTR evolution to be incorporated into the analysis , while providing a natural framework for model validation and criticism . Similarly , examination of multiple data sets under the same models and methods could provide support or otherwise for resulting estimates . In this study we estimate the mutation rate of VNTR markers by developing a stochastic stepwise mutation process of the evolution of genotypes through gains and losses of repeat numbers [16] , [19] embedded in a model of disease transmission [25] . We consider and evaluate two alternative formulations of the stepwise mutation model under a Bayesian statistical framework , applying our methods to four geographically distinct data sets . Our study provides a posterior estimate of the VNTR mutation rate under an explicit model of evolution placed within an epidemiological context .
In the model of disease transmission we use , tracks the number of individuals who are susceptible to infection and tracks infectious individuals , where is time measured in years . For simplicity , we assume a population of fixed size . Let be the rate of transmission and be the rate of death or recovery . First consider a deterministic model where the dynamics are given by ( 1 ) We start the process with a single infected individual ( ) . Define to be the basic reproductive ratio , that is , the number of cases resulting from a single infectious case in a wholly susceptible population . For this model , . The analytical solution of Equation ( 1 ) can be written as ( 2 ) The steady state of the infectious population is We use this deterministic model as the basis for a continuous-time stochastic model that incorporates mutation at VNTR loci . The transition rates of this model , summarised in Table 1 , are as follows: the rate of new infections is and the rate out of the infectious class from death or recovery is . An infection event increases by 1 while a death-or-recovery event decreases by 1 . Each infection is associated with a bacterial genotype by which we mean the set of repeat states across all loci considered in a VNTR typing technique , determined for a particular isolate . Let be the number of individuals infected with bacterial genotype so thatwhere is the number of distinct genotypes in the population at time . We apply the stepwise mutation model to describe VNTR mutation [16] , [17] , [19] in which an event results in a unit increase or decrease in the number of repeats at a locus . We define to be the mutation rate per infectious case for genotype so that the transition rate for mutation of genotype is . A mutation event results in either a new genotype , or a pre-existing genotype in the population ( i . e . , homoplasy ) . In the event of mutation to a new genotype , the number of individuals from the mutating genotype decreases by 1 and the number of individuals in the new class becomes 1 . In the case of homoplasy , the number of individuals in the mutating genotype decreases by 1 while the number of individuals in the existing class increases by 1 . In either case the total number of infected cases , , does not change . We consider two alternative ways to specify VNTR mutation . In the first model , the mutation rate at a locus is proportional to the number of repeats at that locus . In this linear model , the per-locus mutation rate increases linearly with the number of repeats at the locus . In the second constant model , the mutation rate the per-locus mutation rate is constant and thus not dependent on repeat number . Defining to be the number of loci , to be the number of repeats at locus for genotype , and to be the rate of mutation at a locus with a single repeat , under the linear modelUnder the constant modelwhere is the per locus mutation rate and where the indicator function if is true and 0 otherwise . In both models the boundary condition is an absorbing state in that a locus with zero repeats cannot gain or lose repeats . The process starts at time with a single infected individual and the population evolves until time . The initial individual has genotype given by , which we call the founding genotype . At time a sample of size is taken from the population . We simulate this process using the Gillespie exact algorithm [26] so that the time between events is distributed exponentially , with parameter , whereGiven an event , the probability of a specific outcome is proportional to the rate of that outcome , so thatGiven a mutation event , the probability of mutation in an individual with genotype isand given a mutation event in genotype , the probability that it occurs at locus under the linear model isand under the constant model isWe assume that given a mutation event at locus in genotype , the probability of repeat gain is equal to the probability of repeat loss , following [3] , [19] . We implement a standard Bayesian analysis of model parameters using approximate Bayesian computation ( ABC ) [27]–[29] . ABC methods permit approximate Bayesian inference when numerical evaluation of the posterior distribution is either computationally prohibitive or not available , and have been successfully applied to problems in molecular epidemiology [30]–[34] . Intuitively , given a candidate parameter vector , , prior distribution and model likelihood with observed data , ABC methods proceed by generating an artificial dataset from the model and then reducing the dataset to a low dimensional vector of summary statistics , . If is similar to the same vector of statistics obtained from the observed data , , then could have credibly reproduced the observed data under the model . As such , the parameter vector is then retained as part of the approximate posterior , otherwise it is discarded . More precisely , the posterior obtained under ABC methods is given by ( 3 ) where is a standard smoothing kernel with scale parameter . As becomes small , the approximation ( 3 ) becomes increasingly accurate , although computational overheads increase . If the vector of summary statistics are informative for the model parameters , then this posterior distribution approximates the true posterior distribution so that . See e . g . [30] , [31] , [35] , [36] for further description of ABC methods . The parameter vector for the constant model above is where is the repeat structure of the founding genotype in the simulation . For the linear model we have . Except where this may cause confusion , we will refer to a non-model-specific parameter vector as . Conditional on the parameter vector , and following simulation under the model , a sample of size individuals is drawn from the resulting population . Summary statistics , , are then computed , determined as quantities expected to be highly informative regarding the model parameters . Using lower case letters ( e . g . ) to denote sample-based values of the population-level counterparts ( e . g . ) , the summary statistics include the number of distinct genotypes in the sample , , and the set of sample means of repeats at each locusfor , which is expected to contain information about the initial repeat numbers for some time after the founding case . Here , denotes the number of individuals in the sample with genotype , and denotes the within-sample number of repeats at locus for genotype . The final two statistics are based on the ANOVA decomposition given bywhere , from which and can be computed . These two statistics are expected to be informative about the mutation rate between and within loci . The complete vector of summary statistics is then given byTo complete the model specification , we set the parameter to , following [32] , [37] . This death/recovery rate is the sum of the death rate due to tuberculosis , the death rate due to other causes , and the recovery rate from tuberculosis . We chose an informative prior distribution for based on the study of the basic reproductive value of tuberculosis by Blower et al . [38] . We use a distribution approximating the histogram in Figure 3a in reference [38] which has a mean of 5 . 16 and a standard deviation of 2 . 82 , and in particular define the prior of to be a gamma distribution with a shape parameter of and a scale parameter of . The priors for , , and are uniform with wide ranges as shown in Table 2 . We examine the effectiveness of the ABC inference procedure by evaluating its ability to recover accurate estimates of the mutation rate based on data generated under the constant and linear models We simulated a population of individuals with loci , , , and considered a range of mutation rates under each model varying across orders of magnitude and . The number of repeats of the founding genotype were initialised as ( determined as random draws from ) , where denotes loci with repeat number . Based on a sample of size we generated data under each mutation rate value , and obtained weighted samples from the ABC posterior approximations ( c . f . 3 ) using a population-based ABC algorithm , following [32] , [39] , [40] . The technical algorithmic details are given in Text S1 . The estimated posterior distributions of and using the simulated data are shown in Figure 1 . These results indicate that mutation rates can generally be recovered accurately , with the true parameter values lying in regions of high posterior density close to the posterior mode , and with a clear location shift in the density with varying mutation rate . Higher precision can be attained by using a larger sample size , although already represents a sample larger than the real datasets used for this study ( c . f . Table 3 ) . In the ABC setting , posterior precision can also be improved by reducing the kernel scale parameter in ( 3 ) or by the inclusion of more summary statistics [30] , [31] , [35] , [36] , although each of these can substantially increase computational overheads . Improving the precision of posterior parameter estimates for given summary statistics is currently an area of active ABC research [41] . We selected recently published VNTR loci data sets from studies undertaken in four countries: Albania [42] , Iran [43] , Morocco [11] and Venezuela [44] . We chose data sets with a high number of isolates largely from the same clade , a high number of VNTR loci in the typing method , and relatively short periods of isolate collection . The data from Albania and Venezuela are based on 24-locus typing , and the data from Iran and Morocco are based on 15 and 12 loci respectively . A summary of these data are provided in Table 3 , along with the incidence of tuberculosis for each country . As an initial exploratory examination of these data , we computed gene diversity [45] ( also known as virtual heterozygosity ) , for each locus in each data set . This statistic is given by where is the number of isolates with repeat size at locus . Figure 2 ( left plots ) shows the empirical cumulative distribution function of gene diversity across loci for each of the data sets . There is no obvious bimodality in these distributions . This feature is consistent with a common process generating diversity , compared to , for example , the potential bi- or multi-modality in the empirical cumulative distribution function arising from a multi-modal distribution of mutation rates . Similarly , plotting the proportion of VNTR states per locus per repeat ( right plots of Figure 2 ) reveals that while some loci are more variable than others , there is no obvious separation between loci exhibiting high and low variation .
Figure 3 shows the marginal posterior distribution of the mutation rate of VNTR loci for each of the four data sets analysed . In the case of the linear model we also show ( middle panel of Figure 3 ) the posterior of , the per-locus mutation rate at repeat size 1 scaled by the average repeat number of each dataset to provide estimates of the mean per-locus mutation rate in a population with the same distribution of repeats as found in each sample . The posterior means of the mutation rate under the two models , along with 95% central credibility intervals are given in Table 4 . The mean per-locus mutation rate at a locus with a single repeat from the four data sets under the linear model is , and under the constant model the mean per-locus rate is . Note that the prior distributions of the mutation parameters are uniform on a logarithmic ( base 10 ) scale , and so Figure 3 displays the posterior distributions on this scale . To evaluate the suitability of the constant and linear models to describe the observed data , we follow [36] , [46] , [47] and implement posterior predictive model checks . This approach examines the predictive distribution of specified validation statistics ( based on data-generation under the fitted models ) expected to be informative about various model aspects . Comparing the predictive distribution of these statistics with the same statistics derived from the observed data , enables some degree of discrimination between models . To avoid confusing model fitting with model assessment , these statistics should be different from those used in the ABC model fitting process . Unlike the constant model , the mutation rate increases with repeat number under the linear model , and so we expect variation in repeat numbers to increase with repeat numbers . Our model assessment statistics aim to capture these differences from the data . Specifically , we focus on measures of the spread of repeats over the loci . Definingwhere , andwhere , and indexes loci as before , we consider the maximum ( over loci ) range ( ) , the difference between maximum and minimum range ( ) , maximum variance ( ) and the difference between maximum and minimum variance ( ) . Under the linear model , the distributions of these statistics are expected to be shifted to higher values compared to the constant model . We also fit a simple linear regression to each data set with the standard deviation of repeat number at a locus as the response variable and the mean repeat number at a locus as the predictor variable . Based on this fit , we considerwhere is the fitted standard deviation in repeats at a locus with a mean repeat number of one . These statistics are expected to be informative in that the slope should be positive under the linear model and near zero under the constant value , and the intercept should be low under the linear model and high under the constant model . Figure 4 displays the predictive distributions of versus under both models . The observed data statistics are indicated by a cross ( ) . If the cross does not lie within the body of the predictive distribution , this suggests that the model and data are inconsistent with respect to aspects of the data captured by these statistics . The lower four panels present these diagnostics for artificial data generated under both models . The linear data ( lower images ) can be seen to be inconsistent with the constant model , but consistent with the linear model . The constant data ( middle images ) appear to be consistent with both models . As such , these diagnostics are able to reject the constant model when the data is generated by the linear model . In terms of the actual empirical data , the top plots in Figure 4 are based on the data from Albania . Clearly , the constant model is insufficient to describe the variation in repeat numbers inherent in the data . The linear model is better able to account for the observed pattern of repeat variation , although it is still imperfect . The posterior predictive distributions using the data sets from the other three countries were very similar to those of the Albanian data set ( not shown ) . The question of whether the linear model is adequate is examined further in Figure 5 which shows a posterior predictive check of versus under the linear model for each of the analysed data sets . In each case , the observed data lie on the periphery of the predictive densities . Although the linear model is partially able to reproduce these statistics , this analysis shows that there is room for improvement .
We have analysed VNTR data from four tuberculosis studies using a model combining marker mutation and disease transmission processes , within a Bayesian framework . Our analysis shows that the VNTR mutation rate is likely to be relatively high – the posterior mean is higher than some previous estimates obtained in the literature [3] , [19] and closer to more recent estimates [20] . The four data sets , which are from different geographic regions , yielded very similar estimates . Such agreement of estimates is expected if there is a common mechanism of mutation across data sets . Previous work by two of us [20] used standard equilibrium results of the infinite alleles model to describe mutation at multiple VNTR loci , and used estimates of other markers ( IS6110 and spoligotyping ) to calibrate the VNTR rates . That population genetic approach did not account for evolution of VNTRs as a stepwise mutation process . It therefore did not account for homoplasy , though this problem is mitigated by the inclusion of multiple VNTR loci . Further , the underlying dynamics did not include any epidemiological details . Nevertheless , it allowed us to analyse a large number of data sets in the literature to provide a ballpark estimate of VNTR mutation rates . In contrast to that and other prior work , here we used a model that explicitly and simultaneously accounts for the mutation process of the marker and the disease dynamics , and we explored two alternative models of mutation . In addition , the stepwise mutation model used here allows mutation events to re-generate existing VNTR profiles , thereby accounting for homoplasy [48] . In the debate over the magnitude of VNTR mutation rates [3] , [21]–[24] it has been noted that if loci are classified as less variable and more variable , then lower values would be estimated from the former category of loci . This raises the question of whether classification of loci into two categories of rates is supported by an underlying bimodal distribution whose modes correspond to low and high levels of polymorphism . In examining gene diversity , which is a measure of polymorphism , across loci in each data set ( Figure 2 ) we did not observe any obvious break separating less and more variable loci . We have therefore pooled all loci and obtained an estimate of the rate of an arbitrary locus , rather than for a subset of slow or fast evolving loci . If hypermutable VNTR loci exist and are excluded from estimation procedures , using the remaining loci would clearly yield a lower mutation rate . Our use of the linear model is a step towards resolving this issue . The linear relationship by which more units of a repeat are more prone to mutation naturally creates variation in rates . In fact , in assessing the ability of each of our two mutation models to describe the data , we found that the linear model performs better than the constant model ( Figure 4 ) . We note that the average mutation rate under the linear model was estimated to be very close to the mutation rate in the constant model; in this sense our analysis is robust to the exact form of the mutation model . Despite the linear model outperforming the constant model , a posterior predictive goodness-of-fit analysis revealed some evidence that the linear model did not fit the data perfectly ( Figure 5 ) . While previous studies of eukaryote minisatellites agree with a linear relationship between repeat number and mutation rate [49] , some studies of eukaryote microsatellites indicate a more complex relationship between repeat number and mutation rate [50]–[53] . We investigated a third model in which the mutation rate increases exponentially with repeat number , but the results are very similar to those of the linear model ( Figure S3 in Text S1 ) . Future work might adopt a per locus mutation rate that grows non-linearly with repeat number . A drawback of this possibility would be the added complexity and dimensionality of the model with the need to estimate further parameters in a framework that is already computationally intensive . An alternative approach might be to construct a hierarchical Bayesian model of mutation rates in which each locus is associated with its own rate according to some distribution , akin to the analysis of Bazin et al . [54] . We have used a simple model to avoid overfitting the data . However , it is possible to extend the model in future studies to incorporate further complexity and realism . One such detail is the reactivation of latent infection , which could be described by a susceptible-exposed-infected ( SEI ) model in which a proportion of cases progress directly to disease [38] . We performed preliminary simulations from a stochastic version of such a model ( details in Text S1 ) . We consider the number of distinct genotypes since this is one of the statistics we use in the inference and it is known to be informative for mutation rate in similar models [55] , [56] . Figure S2 in Text S1 shows how the number of distinct genotypes in a sample varies with the mutation rate under both models . The latent reactivation model was able to generate statistics close to the observed statistic . The points in the region of the observed statistic are near the posterior density generated under the original model . While this is suggestive that a latency model would produce similar estimates , a full Bayesian analysis would be required to address this issue . The lack of latency is a limitation of our study which should be addressed in future research . Migration is another factor which a more realistic multi-deme population model might incorporate . The interplay between migration and mutation may affect the resulting estimates of the mutation rate . For example , migration from regions with genetically very different clades of M . tuberculosis occurs at a high rate would lead to over-estimation of the mutation rate . Our approach based on the approximate Bayesian computation framework makes future directions such as this and those relating to the mutation process feasible . | Genetically typing the bacterium responsible for tuberculosis is useful for understanding the evolutionary and epidemiological characteristics of the disease . Typing methods based on variable number tandem repeat ( VNTR ) loci are increasingly being used . These loci , which are composed of repeated units , mutate by increasing or decreasing in the number of these repeats . Knowledge of the mutation rate of molecular markers facilitates the epidemiological interpretation of the observed genetic variation in a sample of bacterial isolates . Few studies have examined the rate of mutation at these markers and estimates to date have varied considerably . To address this problem we develop a stochastic model of evolution of these markers and then estimate their mutation rate using approximate Bayesian computation . We examine two alternative forms of the mutation process . The observed data are from four published data sets of tuberculosis bacterial isolates sampled in Albania , Iran , Morocco and Venezuela . We find that these markers have fairly high rates of mutation compared with estimates from previous studies . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"mathematics",
"microbial",
"evolution",
"statistics",
"biology",
"microbiology",
"biostatistics"
] | 2012 | A Model-Based Bayesian Estimation of the Rate of Evolution of VNTR Loci in Mycobacterium tuberculosis |
Upon reactivation from latency and during lytic infections in neurons , alphaherpesviruses assemble cytosolic capsids , capsids associated with enveloping membranes , and transport vesicles harboring fully enveloped capsids . It is debated whether capsid envelopment of herpes simplex virus ( HSV ) is completed in the soma prior to axonal targeting or later , and whether the mechanisms are the same in neurons derived from embryos or from adult hosts . We used HSV mutants impaired in capsid envelopment to test whether the inner tegument proteins pUL36 or pUL37 necessary for microtubule-mediated capsid transport were sufficient for axonal capsid targeting in neurons derived from the dorsal root ganglia of adult mice . Such neurons were infected with HSV1-ΔUL20 whose capsids recruited pUL36 and pUL37 , with HSV1-ΔUL37 whose capsids associate only with pUL36 , or with HSV1-ΔUL36 that assembles capsids lacking both proteins . While capsids of HSV1-ΔUL20 were actively transported along microtubules in epithelial cells and in the somata of neurons , those of HSV1-ΔUL36 and -ΔUL37 could only diffuse in the cytoplasm . Employing a novel image analysis algorithm to quantify capsid targeting to axons , we show that only a few capsids of HSV1-ΔUL20 entered axons , while vesicles transporting gD utilized axonal transport efficiently and independently of pUL36 , pUL37 , or pUL20 . Our data indicate that capsid motility in the somata of neurons mediated by pUL36 and pUL37 does not suffice for targeting capsids to axons , and suggest that capsid envelopment needs to be completed in the soma prior to targeting of herpes simplex virus to the axons , and to spreading from neurons to neighboring cells .
Many pathogens spread within the nervous system by fast intracellular axonal transport mediated by microtubule motors , and progeny particles are transmitted via synapses from neuron to neuron [1 , 2] . Such life style allows pathogens to hide from the extracellular complement system , from antibodies , and from immune cells . But it comes with the challenge of finding the tiny exit gates from the neuronal somata , and of specific targeting to dendrites or to axons . It has been estimated that the cross section of the axonal outlet occupies less than 2 thousandth of the surface area of the soma [3 , 4] . The molecular post codes achieving this remarkable intracellular trafficking control are not well understood; neither for host nor for pathogen cargoes . Alphaherpesviruses such as herpes simplex viruses ( HSV ) replicate in the skin , the eyes , and the oral , genital , or nasal mucosa , and then enter local axon endings of sensory or autonomic neurons in which they establish latency with little gene expression . Latent HSV-1 genomes have been detected in human cranial ganglia , especially in the trigeminal ganglia , while dorsal root ganglia ( DRG ) can harbor both , HSV-1 and HSV-2 [5–7] . Viral gene expression can be reactivated upon stress and a decline in the local immune responses , and progeny viruses are transported back to the periphery , where they cause lytic infections . Efficient targeting of alphaherpesviruses to the axons is essential for axonal spread and thus disease . Severe outcomes are potentially blinding herpes keratitis upon spread within the eyes or life-threatening herpes encephalitis upon spread within the brain [reviewed in 8 , 9] . The HSV-1 virion contains a large double-stranded DNA genome encased in a viral capsid , numerous tegument proteins and an envelope with many viral membrane proteins . Furthermore several host proteins and mRNAs have been detected in highly purified inocula [10–12] . HSV-1 assembly begins in the nucleus with genome packaging into capsids , which then traverse the nuclear envelope [reviewed in 13 , 14] . Cytosolic capsids associate with the inner tegument proteins pUL36 and pUL37 for their intracellular transport along microtubules to cytoplasmic membranes , where they meet other tegument and viral membrane proteins for secondary envelopment and virion formation [15–24] . In addition to pUL36 and pUL37 , other structural proteins are required for efficient capsid envelopment . HSV-1 mutants lacking either pUL36 or pUL37 , or the membrane proteins pUL20 or glycoprotein K ( gK ) are severely impaired in cytoplasmic envelopment , and accumulate cytosolic capsids instead [15 , 16 , 18 , 20 , 25–32] . HSV1-pUL20 and gK form functional complexes that connect with the capsids via pUL37 which in turn binds to pUL36 and the small capsid protein VP26 [33–36] . Another prominent tegument linker is VP16 , with binding sites for pUL36 , VP11/12 , VP22 and gH [22 , 34 , 37–39] . VP22 in turn can bind ICP0 , pUL16 , gD , gE and gM [reviewed in 21 , 22] . The resulting vesicles are transported to the cell periphery and fuse with the plasma membrane to release infectious virions [reviewed in 13 , 14 , 22] . HSV-1 cytosolic capsids and complete virions within transport vesicles are targeted from the neuronal somata to axons to a varying extent; this has led to different hypotheses on the mode of neuronal alphaherpesvirus assembly [reviewed in 23 , 40 , 41] . According to the married model , capsid envelopment occurs exclusively in the soma , and only transport vesicles harboring complete virions enter axons . The separate model refers to different cargo structures being targeted independently of each other to the axons; namely capsids with associated tegument proteins as well as vesicles harboring viral envelope proteins and tegument proteins associated to their cytosolic tails . Many structural proteins contribute to the neuronal spread of alphaherpesviruses , but the molecular determinants that are required for microtubule motor recruitment and for targeting from the soma to the axon gate have not been fully dissected . Purified HSV-1 capsids with inner tegument proteins can recruit the microtubule motors kinesin-1 , kinesin-2 , dynein and its cofactor dynactin to their surface , are translocated along microtubules in vitro , and can dock to nuclear pores [11 , 42–44] . pUL36 of HSV-1 and of the porcine pseudorabies virus ( PRV ) harbor potential binding motives for kinesin-1 light chains , and PRV-pUL36 has been shown to bind dynein and dynactin [45–47] . For these reasons and based on many additional functional studies , pUL36 and pUL37 are considered the most likely candidates for microtubule motor recruitment [reviewed in 23 , 45 , 46 , 48] . We therefore hypothesized that pUL36 and pUL37 might be sufficient for axonal capsid targeting and axonal capsid transport . As capsid envelopment may be a fast process , and since therefore the steady-state concentration of cytosolic capsids might be low , we used HSV-1 mutants to enrich for assembly intermediates . We infected neurons with HSV1-ΔUL20 , HSV1-ΔUL37 or HSV1-ΔUL36 that accumulated cytosolic capsids decorated with pUL36 and pUL37 , only with pUL36 , or lacking both . Our previous data indicate that acquisition of both pUL36 and pUL37 is essential to enable capsids to enlist microtubule transport in epithelial cells [20] . However , while these inner tegument proteins enabled intracellular capsid motility of HSV1-ΔUL20 in epithelial cells and the somata of DRG neurons , they did not suffice to target cytosolic capsids to the axons . Our data indicate that HSV-1 particles have to acquire additional tegument and envelope proteins , and suggest that cytoplasmic envelopment needs to be completed prior to axonal transport which is in accordance with the married model for assembly of alphaherpesviruses in neurons .
To investigate HSV-1 axonal targeting , we cultured primary neurons derived from dissociated DRG of adult mice until they had developed mature neurites . Within 3 to 5 days of culture in vitro ( div ) , the neurons expressed the axonal microtubule-associated protein tau ( not shown ) , phosphorylated neurofilament , un-phosphorylated neurofilament , and ankyrinG . In the somata , there were short β-III-tubulin microtubules and careful analysis often revealed a perinuclear microtubule-organizing center ( S1Aii Fig , arrow ) , but individual microtubules could not be discerned in the neurites ( S1A Fig ) . There were less phosphorylated neurofilament H and M in the somata than in the neurites ( S1B Fig ) , while non-phosphorylated neurofilament H epitopes were distributed more evenly ( S1C Fig ) , as reported for rat nervous tissue [49] . Likewise ankyrinG , another axonal marker was targeted to the neurites ( S1D Fig ) . In situ , peripheral sensory axons also contain ankyrinG along the entire axons up to the dermal-epidermal junction [50] . To evaluate the microtubule polarity in these neurites , we transduced the neurons at 1 div with a lentiviral vector expressing end-binding protein 3 ( EB3 ) , which associates with dynamically growing microtubule plus-ends [51] . At 4 div , live-cell imaging as well as time-projections showed that all EB3-GFP comets moved away from the soma , suggesting that most , if not all , microtubule plus-ends faced towards the distal ends of the neurites ( not shown ) . These results indicate that the DRG neurons had regrown neurites with axonal features and with a unipolar microtubule polarity . We have shown previously that such neurons are productively infected , and that they transmit HSV-1 to co-cultured epithelial cells upon infection with the HSV1 ( 17+ ) Lox-Che ( c . f . Table 1 ) , which had been cloned into a bacterial artificial chromosome ( BAC ) , and which expresses mCherry as a reporter [52 , 53] . To generate cytosolic capsids with different tegument protein composition , and thus to determine whether the requirements for intracellular capsid motility parallel those for axonal targeting , we constructed HSV1 ( 17+ ) Lox-CheVP26-ΔUL36 [20] , -CheVP26-ΔUL37 [20] , and -CheVP26-ΔUL20 ( this study ) in the same genetic background ( c . f . Table 1 ) . We deleted the ATG start codon and a second ATG , and inserted three stop codons into the UL20 gene , since it includes the promotor of UL19 that codes for the major capsid protein VP5 . Restriction analyses of pHSV1 ( 17+ ) Lox-ΔUL20 and -CheVP26-ΔUL20 indicated the addition of mCherry to VP26 ( AscI and BamHI , C in S2A Fig ) , and that a resistance gene inserted with the UL20 mutations had been removed ( XhoI , asterisks in S2A Fig ) . Further HindIII ( S2A Fig ) , EcoRI , or EcoRV digestions resulted in the expected fragment sizes ( not shown ) . HSV1 ( 17+ ) Lox-ΔUL20 and -CheVP26-ΔUL20 were recovered by transfecting the corresponding BACs into Flp-In-CV-1-cells that express pUL20 in trans [54] . Sequencing of the mutated region confirmed the introduction of the intended changes ( not shown ) . The lack of the ATGs and the introduced stop codons prevented the expression of pUL20 , whereas pUL37 expression was unchanged ( S2B Fig ) . The intra- and extracellular titers of HSV1 ( 17+ ) Lox-ΔUL20 and -CheVP26-ΔUL20 were about 1 , 000-fold lower than their parental strains in non-complementing Vero cells , but higher in a pUL20 complementing cell line ( S2C and S2D Fig ) . Similar results have been reported for HSV1-ΔUL20 mutants in other genetic backgrounds [25 , 26 , 54–56] . There were little differences between the parental Lox and the -CheVP26 strains , indicating that tagging VP26 with mCherry ( Che ) did not impair HSV-1 replication , as reported before [24 , 57] . Using conventional electron microscopy , we next analyzed virus morphogenesis . Upon infection of Vero cells with HSV1 ( 17+ ) Lox ( Fig 1A ) or -CheVP26 ( not shown ) , viral particle maturation proceeded as expected with the formation of nuclear capsids , the appearance of cytosolic capsids , partially and completely enveloped cytoplasmic capsids , and extracellular virions associated with the plasma membrane . With secondary envelopment , capsids acquired an electron dense tegument layer , and the electron dense genomes were well preserved ( Fig 1Ai and 1Aiii; asterisk ) . Therefore the morphology of capsids and tegument of virions inside authentic transport vesicles was more similar to that of extracellular virions ( Fig 1Aii; black arrow ) than to that of cytosolic capsids ( Fig 2Aiii; white arrowhead ) even if they were partially enveloped ( Fig 1Ai and 1Aiii , black arrowhead ) . Nuclear capsid egress of the mutants HSV1 ( 17+ ) Lox-ΔUL20 ( Fig 1Bi and 1Bii ) and -CheVP26-ΔUL20 ( Fig 1Biii ) was unaffected , resulting in many cytosolic capsids , and several of them were closely associated with cytoplasmic membranes ( Fig 1Bi and 1Bii; black arrowhead ) . We detected neither capsids with the characteristic morphology of completed secondary envelopment nor virions bound to the plasma membrane in the absence of pUL20 . Furthermore , in contrast to infection with HSV1 ( 17+ ) Lox-ΔUL36 or -ΔUL37 [20] , we did not detect any cytoplasmic or extracellular L-particles , which are viral envelopes with an electron dense tegument but lacking capsids [58–60] . Next , we analyzed DRG cultures infected with HSV1 ( 17+ ) Lox-CheVP26 , -CheVP26-ΔUL36 , -CheVP26-ΔUL37 , -CheVP26-ΔUL20 ( Fig 2 ) or their respective untagged strains ( not shown ) by electron microscopy . Neurons identified by the characteristic morphology of their nuclei and infected with the parental HSV-1 strains contained nuclear capsids , primary envelopment intermediates ( Fig 2A , white star ) , cytosolic capsids ( Fig 2A , white arrowhead ) , and enveloped virions ( Fig 2A , black star ) . Neurons infected with HSV1 ( 17+ ) Lox-CheVP26-ΔUL36 ( Fig 2B ) , -CheVP26-ΔUL37 ( Fig 2C ) , or -CheVP26-ΔUL20 ( Fig 2D ) contained non-enveloped cytosolic capsids ( Fig 2 , white arrowhead ) , but did not reveal any extracellular virions bound to their plasma membranes . HSV1 ( 17+ ) Lox-CheVP26-ΔUL20 infected neurons contained in addition wrapping intermediates ( Fig 2D , black arrowhead ) . A quantification of the number of cytoplasmic HSV-1 assembly intermediates revealed that cells infected with HSV1 ( 17+ ) Lox-ΔUL20 or -CheVP26-ΔUL20 had a similar ratio of cytosolic capsids to membrane-associated capsids as upon infection with the respective parental strains HSV1 ( 17+ ) Lox or -CheVP26 ( Table 2 ) . To characterize the surface protein composition of the cytosolic capsids , Vero cells were infected with HSV1 ( 17+ ) Lox or -ΔUL20 for 16 h , fixed , and ultrathin cryosections were prepared for quantitative immunoelectron microscopy . We have reported similar experiments for -ΔUL36 and -ΔUL37 before [20] , and show their results here again for a direct comparison ( Fig 3Ai and 3Aii ) . Cytosolic capsids of both , Lox [20] and -ΔUL20 were labeled with anti-pUL36 ( Fig 3Aiii ) and anti-pUL37 ( Fig 3Aiv ) . There were on average 2 . 2 gold particles per capsid for HSV1 ( 17+ ) Lox , 2 for -ΔUL37 and -ΔUL20 , but only 0 . 5 for -ΔUL36 ( Fig 3Bi ) using anti-pUL36 antibodies . Upon labeling with anti-pUL37 , there were 1 . 7 or 1 . 5 gold particles per capsid for HSV1 ( 17+ ) Lox or -ΔUL20 , but only 0 . 5 for -ΔUL36 and -ΔUL37 ( Fig 3Bii ) . These results indicate that cytoplasmic capsids of Lox-ΔUL20 had recruited pUL36 and pUL37 to a similar extent as those of the parental HSV1 ( 17+ ) Lox , while capsids of -ΔUL37 lacked pUL37 but still recruited pUL36 , and capsids of -ΔUL36 lacked both , pUL36 and pUL37 . Thus , the HSV1 ( 17+ ) Lox-ΔUL36 , -ΔUL37 , and -ΔUL20 as well as HSV1 ( 17+ ) Lox-CheVP26-ΔUL36 , -ΔUL37 , and -ΔUL20 mutants generated cytosolic capsids exposing different proteins on their surface , namely no inner tegument , only pUL36 , or both pUL36 and pUL37 . To characterize the intracellular capsid motility in the presence of different inner tegument proteins , we infected Vero cells with HSV1 ( 17+ ) Lox-CheVP26 , -CheVP26-ΔUL36 , -CheVP26-ΔUL37 , or -CheVP26-ΔUL20 and acquired confocal fluorescence time-lapse movies with a temporal resolution of 5 images per second . For a direct comparison , we analyzed in parallel movies with -CheVP26 and -CheVP26-ΔUL20 recorded in this study as well as movies that we had recorded with -CheVP26-ΔUL36 and -CheVP26-ΔUL37 previously and published in Sandbaumhüter et al . [20] . The time-projections of the tracks show that capsids of Lox-CheVP26 ( Fig 4Ai ) and -CheVP26-ΔUL20 ( Fig 4Aiv ) exerted short and long range transport towards the nucleus and towards the cell periphery ( S1 and S4 Movies ) . In contrast , the tracks of -CheVP26-ΔUL36 ( Fig 4Aii ) and -CheVP26-ΔUL37 ( Fig 4Aiii ) revealed only random , undirected motility ( S2 and S3 Movies ) . While the average track length ( Fig 4Bi ) and the maximum step velocity ( Fig 4Bii ) were similar for the capsids of -CheVP26 , -CheVP26-ΔUL36 , -CheVP26-ΔUL37 , and -CheVP26-ΔUL20 , there were more tracks with a length of more than 5 μm and with a velocity exceeding 1 μm/s for the capsids of the parental Lox-CheVP26 and the -CheVP26-ΔUL20 than for the capsids of -CheVP26-ΔUL36 and -CheVP26-ΔUL37 . We furthermore determined the mean square displacement exponent ( MSDex ) for each track ( Fig 4Biii ) . The MSDex is a characteristic feature for the displacement mode of a single motile particle from its starting position over time . The MSD is defined as the square of the travelled distance from the starting point of the track , which is calculated for each time point , and plotted against the time . The slope of such a curve in a log-log plot defines the MSDex and indicates whether a particle has undergone active transport ( MSDex > 1 ) , free diffusion ( MSDex = 1 ) , or confined diffusion [MSDex<1; 61 , 62 , 63] . While the MSDex of the majority of the tracks denoted free or confined diffusion as expected for any intracellular cargo , the MSDex was greater than 1 for more tracks of Lox-CheVP26 and -CheVP26-ΔUL20 than for -CheVP26-ΔUL36 or -CheVP26-ΔUL37 . We furthermore infected DRG neurons with HSV1 ( 17+ ) Lox-CheVP26 , -CheVP26-ΔUL36 , -CheVP26-ΔUL37 , or -CheVP26-ΔUL20 and acquired spinning disk microscopy time lapse movies of the cell bodies with a temporal resolution of 20 images per second . Similar as in the epithelial cells , capsids of the parental ( Fig 5Ai; S5 Movie ) and -CheVP26-ΔUL20 ( Fig 5Aiv; S8 Movie ) were transported over short and longer distances as indicated by the time projection lines . In contrast , the capsids of -CheVP26-ΔUL36 ( Fig 5Aii; S6 Movie ) and -CheVP26-ΔUL37 ( Fig 5Aiii , S7 Movie ) only moved in an undirected fashion . Quantification revealed that the transport characteristics in DRG neuronal cell bodies were the same as in epithelial cells ( Fig 5Bi to 5iii ) . Thus , the motility features of the cytosolic capsids of -CheVP26-ΔUL36 or -CheVP26-ΔUL37 in epithelial cells and sensory neurons were indicative for intracellular diffusion which is also the mode of motility in the absence of microtubules after nocodazole treatment [62] . In contrast , the capsids of -CheVP26 and -CheVP26-ΔUL20 had transport characteristics typical for directed , active microtubule-dependent transport . After we had characterized the intracellular motility of capsids in the presence of pUL36 and pUL37 ( -ΔUL20 ) , of only pUL36 ( -ΔUL37 ) , or lacking both ( -ΔUL36 ) , we asked which capsid types could be targeted to the axonal outlet . We infected primary DRG neurons with HSV1 ( 17+ ) Lox , -ΔUL36 , -ΔUL37 , or ΔUL20 ( Fig 6 ) or the respective HSV1 ( 17+ ) Lox-CheVP26 strains ( S3 Fig , Fig 7 ) . At 24 to 26 hpi , the cells were fixed , permeabilized and labeled with antibodies directed against the capsid protein VP26 ( Fig 6i and 6iv panels , green in 6iii ) , and the tegument proteins VP22 ( Fig 6A–6D , ii and v panels , red in iii and vi ) , the tegument protein VP13/14 ( Fig 7 , ii panels , green in iii ) , the envelope protein gD ( Fig 6E–6H , ii and iv panels , red iii and vi ) , the envelope protein gB ( Fig 7 , vi panels , green in viii ) , or the neuron specific beta-3-tubulin ( Fig 7vii ) . Representative images from at least 3 independent experiments for the different strains are shown in Figs 6 , S3 and 7 . After infection with HSV1 ( 17+ ) Lox ( Fig 6A and 6E ) or HSV1 ( 17+ ) Lox-CheVP26 ( Figs S3A and 7A ) , the nucleoplasm ( Figs 6Ai , 6Ei and S3Ai ) , the cytoplasm ( Figs 6Ai , 6Ei and S3Ai ) and the axons ( Fig 6Aiv , 6Eiv and 7A ) were filled with capsids; in the cytoplasm , the capsids were often clustered . Infection with -ΔUL36 ( Figs 6B , 6F , S3B and 7B ) resulted in a dispersed and more random distribution of capsids in the cytoplasm of the somata ( S3B Fig ) , with fewer capsids being targeted to the axons ( Figs 6Biv , 6Fi , 6Fv and 7B ) . After infection with -ΔUL37 ( Figs 6C , 6G , S3C and 7C ) , capsids clustered within the cytoplasm and the cell morphology was often altered; again very few capsids had been targeted to the axons ( Figs 6Civ , 6Giv , 7Ci and 7Cv ) . The stronger cytopathic effects as indicated by membrane blebbing upon infection upon infection with the HSV1-ΔUL37 strains ( S3C Fig ) might be due to the lack of pUL37 targeting RIG-I and blocking RNA-induced activation [64] , but we did not investigate this further . Unexpectedly , there were also only very few axonal capsids upon infection with -ΔUL20 ( Figs 6Hiv and 7D ) . The capsid distribution of -ΔUL20 in the somata was similar to that of the parental HSV1 ( 17+ ) Lox , but with reduced accumulation at the plasma membrane ( Fig 6Div , S3Di ) . After infection with -ΔUL36 , -ΔUL37 , or -ΔUL20 , the outer tegument proteins VP22 ( Fig 6Bv , 6Cv and 6Dv ) and VP13/14 ( Fig 7Bii , 7Cii and 7Dii ) as well as gD ( Fig 6Fv , 6Gv and 6Hv ) and the envelope proteins gB ( Fig 7Bvi , 7Cvi and 7Dvi ) had still been targeted to the axons as for the parental strain ( Figs 6A , 6E and 7A ) , indicating that axonal transport per se had not been impaired . Nevertheless , the deletion mutants -ΔUL36 , -ΔUL37 , and -ΔUL20 were incapable of efficient axonal capsid transport . For quantitation , we scaled up two experiments to image random axonal regions and determined the number of different viral structures per axon length using a novel automated image analysis algorithm ( S4 Fig ) . We determined the imaged axon length , the total number of capsids as detected by anti-VP26 , and the fraction of these capsids colocalizing with gD or VP22 ( c . f . S1 Table ) . After infection with the parental strain HSV1 ( 17+ ) Lox , we counted 20 to 30 capsids per 100 μm axon length ( Fig 8A; total VP26 , black bars ) of which 75% colocalized with gD ( Fig 8B; VP26 + gD ) . Furthermore , there was a similar number of membrane structures containing gD and colocalizing with VP26 or not ( Fig 8C , gD ) ; the size of these structures varied considerably ( Fig 8D ) . After infection with Lox-ΔUL36 ( Fig 8; light grey columns ) , or -ΔUL20 ( Fig 8; dark grey columns ) , there were only few capsids targeted to the axons ( Fig 8A ) , which again mostly co-localized with gD ( Fig 8B ) . However , the number of gD-containing membrane structures was only moderately reduced for the HSV-1 mutants suggesting that axonal transport per se had not been inhibited ( Fig 8C ) . In uninfected neurons , there was also some anti-gD background signal ( Fig 8C , mock ) , but these cross-reacting structures were much smaller ( Fig 8D ) . We also quantified the number of capsids detected by anti-VP26 ( Fig 8E ) , and the fraction of these capsids colocalizing with the outer tegument protein VP22 ( Fig 8F ) of a parallel set of samples . After infection with the parental strain HSV1 ( 17+ ) Lox , we counted again 20 to 30 capsids per 100 μm axon length ( Fig 8E; total VP26 , black bars ) , of which 60% to 70% colocalized with VP22 ( Fig 8F; VP26 + VP22 ) . Furthermore , there was a similar number of structures containing VP22 colocalizing with VP26 or not ( Fig 8G; VP22 ) which also varied considerably in size ( Fig 8H ) . After infection with Lox-ΔUL36 , or -ΔUL20 , again fewer capsids had been targeted to the axons ( Fig 8E ) , of which almost 80% co-localized with VP22 ( Fig 8F ) . The number of VP22 containing structures was moderately reduced for the HSV-1 mutants supporting the notion that axonal transport per se had not been inhibited ( Fig 8G ) . In uninfected neurons , there was almost no anti-VP22 background signal ( Fig 8G , mock ) , and again these cross-reacting structures were much smaller ( Fig 8H ) . These experiments show that the inner tegument proteins pUL36 and pUL37 on the capsids of -ΔUL20 were not sufficient for axonal targeting . Instead , association with membranes and outer tegument proteins , and thus possibly completion of secondary envelopment seem to be a prerequisite for efficient capsid targeting into the axons .
Embryonic and neonatal neurons dissociated from trigeminal ganglia , DRG , or superior cervical ganglia of chicken , mouse , rat , or human have been used to study axonal trafficking and egress of HSV particles [71–78] . However , primary cultures from embryonic tissue do not present optimal models for age-related changes in differentiation , physiology , or late-onset disease [79–81] . We and others [82] therefore used neurons from adult mice to complement results obtained by infecting adult animals with HSV-1 [74 , 83] and to obtain functional insights into the process of axonal targeting of HSV-1 viral structures . The neurons from the DRG of adult mice formed neurites with axonal features which contained ankyrinG and microtubules of uniform polarity with the plus-ends pointing towards the axon endings . Furthermore , we have shown previously that HSV1 ( 17+ ) Lox strains that have been cloned into a bacterial artificial chromosome and therefore lack the OriL [84 , 85] infect such neurons and spread the infection to neighboring epithelial cells [52 , 53] . Hence murine adult DRG neurons provide a versatile system to study productive HSV-1 infection and transport in axons with unipolar microtubules in vivo and in vitro . While there is a growing consensus that the swine alphaherpesvirus PRV relies on the married model for axonal transport in all neuron types , the picture is less clear for HSV-1 . According to the married model , capsids are enveloped exclusively in the somata , and thus it would be sufficient to expose a host or a viral postal code on the cytosolic membrane surface to target vesicles harboring fully assembled virions to axons [reviewed in 23 , 40 , 41] . But for HSV-1 two types of axonal cargo , free cytosolic capsids as well as capsids surrounded by an envelope and a transport vesicle membrane , have been detected to a varying degree [71 , 73 , 76 , 78 , 86–88] . Complementary methods have been used to image viral assembly intermediates . Antibodies have less access to capsids associated with membranes than to cytosolic capsids [89 , 90] . While fluorescent protein tags circumvent this challenge , they may shift the relative abundance of different assembly intermediates [57 , 72 , 73 , 84 , 91–94] . Furthermore , the resolution limit of confocal microscopy is larger than a capsid diameter of 125 nm leading to an overestimation of capsid colocalization with membrane markers . On the other hand , conventional electron micrographs provide sufficient contrast to distinguish hexagonal capsids from membrane vesicles or fully enveloped capsids within transport vesicles but only upon sufficient heavy metal deposition and in very thin sections of 50 nm , which results in an underestimation of membrane-associated capsids [47 , 71 , 87] . In embryonic rat hippocampal neurons , many cytosolic HSV-1 capsids are indeed not associated with cytoplasmic membranes as shown by 3-dimensional cryoelectron tomography [88] . The latter is in fact the best imaging method for this question , but limited to specialized centers and not amenable to medium-throughput [95] . For these reasons , we developed novel image quantitation algorithms and novel mutants to characterize HSV-1 axonal targeting , and used both , fluorescent tagging and antibodies to detect the capsid protein VP26 , in combination with antibodies against the tegument proteins VP22 and VP13/14 , and the envelope proteins gD and gB . The separate model of alphaherpesvirus egress postulates that different subassemblies , namely cytosolic capsids as well as membrane vesicles conveying envelope proteins , are targeted to axons to be assembled in the axons or just prior to trans-synaptic spread or even during budding at the axonal plasma membrane [reviewed in 23 , 40 , 41] . Neuronal cargoes enlist the so-called smart motors that utilize stabilized microtubules which lead to the tiny outlets to dendrites or to axons [96 , 97] . One such smart motor is kinesin-1 that preferentially binds to de-tyrosinated and acetylated microtubules , modifications that are typical for axonal microtubules while others , such as kinesin-5 or kinesin-13 prefer tyrosinated microtubules [97 , 98] . Based on extensive genetic , biochemical and cell biology experiments , pUL36 and pUL37 are considered the most likely candidates for motor recruitment to cytosolic capsids [reviewed in 45 , 48] . Alphaherpesviruses lacking pUL36 or pUL37 are impaired in microtubule-mediated intracellular transport [15 , 16 , 18–20 , 30–32 , 99–101] . HSV-1 capsids exposing the inner tegument proteins pUL36 and pUL37 but not naked capsids translocate along microtubules in vitro and recruit kinesin-1 and kinesin-2 from a brain cytosolic extract [11 , 43] . Finally , pUL36 of HSV-1 and PRV contains binding sites for dynein and dynactin , and possibly the light chains of kinesin-1 [46 , 47] . We therefore generated HSV-1 mutants that assembled un-enveloped cytosolic capsids with different tegument proteins . Consistent with their proposed function to recruit microtubule motors , the capsids of HSV1-ΔUL20 harboring pUL36 and pUL37 but not the ones of HSV1-ΔUL36 or HSV1-ΔUL37 were capable of active transport in epithelial cells or the somata of DRG neurons . The capsids of HSV1-ΔUL20 therefore resemble capsids which recruit the smart motor kinesin-1 from a cytosolic brain extract in vitro [11] . Our data indicate that capsid-associated pUL36 and pUL37 were sufficient for microtubule transport , but not for efficient axonal targeting . It is possible that the capsids of HSV1-ΔUL20 had been associated with a larger amount of outer tegument proteins covering potential motor-binding sites on the inner tegument . Capsids with lower amounts of outer tegument recruit more kinesin-1 from a cytosolic brain extract than capsids with more outer tegument [11] . Furthermore these outer tegument proteins might lead to an association with cytoplasmic membranes , although our quantitative electron microscopy data indicate that while HSV1-ΔUL20 was impaired in completing secondary envelopment , there was no higher association with wrapping membranes when compared to the respective parental strains . Those capsids associated with wrapping membranes might have been connected to larger membrane systems in neurons , which might have reduced their motility , and thus their chances to find the axonal exit . Axonal transport in general had not been impaired upon infection with either HSV-1 mutant , since the envelope proteins gB and gD as well as the outer tegument proteins VP13/14 and VP22 had been targeted to axons . Our data are consistent with the notion that cytosolic capsids might rely on the inner tegument for kinesin-1 mediated transport in epithelial cells and the somata , but that transport vesicles harboring complete virions are more efficiently targeted to axons than such cytosolic capsids . Our data indicate that in addition to pUL36 and pUL37 , other HSV-1 components and processes depending on pUL20 are required for axonal targeting in adult neurons . The simplest interpretation of our data is that HSV-1 relies also on the married model for axonal targeting and trafficking , and that the membrane of the transport vesicles harboring complete virions harbors a viral or a host receptor for smart axonal microtubule motors on its cytosolic surface . Indeed , kinesin-1 can co-traffic with PRV and HSV-1 tegument and envelope proteins [102–104] . In the absence of pUL20 , secondary envelopment was not completed , and apparently also the tegument adopted a different conformation or composition , as our electron microscopy analysis revealed a different tegument contrast for capsids associated with wrapping membranes than for complete virions . Furthermore without pUL20 , the envelope proteins gD and gH/gL , and possibly also their interaction partners VP22 and VP16 [38] , are not properly targeted to cytoplasmic virus assembly sites [105] . Of the alphaherpesvirus envelope proteins , gE/gI and pUS9 are also necessary for efficient axonal spread but not for cytoplasmic capsid envelopment [23] . pUS9 of PRV associates with the neuron-specific kinesin-3 subunit KIF1A , and contributes to initial axonal sorting of PRV particles [106–108] . Similarly , pUS9 of HSV-1 is required for efficient axonal targeting , and its large cytosolic domain interacts with the KIF5B subunit of conventional kinesin-1 [74 , 109] . If the intracellular trafficking and targeting of gE/gI or pUS9 had been impaired in the absence of pUL20 , this would explain why even capsids associated with wrapping membranes could not be targeted to the axons upon infection with HSV1-ΔUL20 . According to the loading hypothesis , capsids require in the somata in addition to the inner tegument proteins pUL36 and pUL37 also the membrane proteins gE/gI and pUS9 to be loaded onto kinesin-1 [13 , 109 , 110] . Vesicles harboring gD and possibly VP22 did neither require pUL36 , pUL37 nor pUL20 for axonal targeting . Future work elucidating which assembly intermediates of wild-type HSV-1 as well as -ΔUL36 , -ΔUL37 , -ΔUL20 and other deletion mutants recruit which smart microtubule motors will reveal specific viral and specific host components that need to be assembled onto an alphaherpesvirus cargo for efficient axonal targeting . Our data are consistent with the notion that cytosolic capsids rely on the inner tegument for dynein and kinesin-1 mediated transport prior to secondary envelopment in epithelial cells and in the neuronal somata , but that transport vesicles harboring complete virions are more efficiently targeted to axons than such cytosolic capsids .
Cell lines were cultured in a humidified incubator at 37°C and 5% CO2 and passaged twice per week . BHK-21 cells ( ATCC CCL-10 ) were maintained in MEM ( Cytogen , Wetzlar , Germany ) supplemented with 10% ( v/v ) FBS ( fetal bovine serum; PAA Laboratories GmbH , Cölbe , Germany ) , Vero cells ( ATCC CCL-81 ) in MEM supplemented with 7 . 5% ( v/v ) FBS , and HEK-293T cells ( ATCC CRL-11268 ) and Flp-In-CV-1 cells in DMEM ( Invitrogen , Karlsruhe , Germany ) supplemented with 2 mM L-glutamine and 10% FBS ( v/v ) . The pUL36 trans-complementing Vero-derived HS30 cell line [31] was provided by Prashant Desai ( Johns Hopkins University , Baltimore , USA ) and the pUL37 trans-complementing rabbit skin 80C02 cell line [15] by Frazer J . Rixon ( University of Glasgow , Scotland , UK ) . These complementing cells were maintained in MEM containing 1% non-essential amino acids ( Cytogen ) and 7 . 5% or 10% ( v/v ) FBS , respectively . Every 5th passage was cultured in the presence of G418 ( 500 μg/ml; PAA Laboratories GmbH ) . The trans-complementing Flp-In-CV-1-derived cell line expressing pUL20 under the control of the HSV-1 gD promoter [54] was provided by Konstantin G . Kousoulas ( Louisiana State University , Louisiana , USA ) and maintained in DMEM ( Invitrogen ) supplemented with 10% ( v/v ) FBS and 2 mM L-glutamine and 125 μg/ml hygromycin B ( Invitrogen ) . Primary neurons from dorsal root ganglia ( DRG ) of adult C57Bl/6JHanZtm mice were cultured using established protocols [111 , 112] . Briefly , mice were sacrificed , and the DRG from the cervical , thoracic and lumbar level were dissected and collected in 1x HBSS-complete buffer ( Hank’s balanced salt solution , pH 7 . 4 with 5 mM HEPES and 10 mM D-Glucose ) . The DRG of three mice were pooled and treated with 20 mg/ml papain ( Sigma-Aldrich , Schnelldorf , Germany; in 0 . 4 mg/ml L-Cysteine , 0 . 5 mM EDTA , 1 . 5 mM CaCl2 , pH 7 . 4 ) for 20 min at 37°C , and with 10 mg/ml collagenase IV ( Invitrogen ) and 12 mg/ml dispase II ( Sigma-Aldrich ) in 1x HBSS-complete buffer for 20 min at 37°C . DRG and cells were sedimented and re-suspended in 1 ml 1xHBSS-complete buffer and triturated using Pasteur pipettes with narrowed ends . The suspensions were spun for 8 min at 381 x g through 20% ( v/v ) Percoll ( Sigma-Aldrich ) cushions in CO2-independent medium ( Life Technologies Gibco , Carlsbad , CA , USA ) containing 10 mM D-glucose , 5 mM HEPES , 10% FBS , 100 U/ml penicillin and 0 . 1 mg/ml streptomycin . The cells were washed with 2 ml CO2-independent medium , sedimented 2 min at 1 , 000 x g , suspended in Ham’s F-12 nutrient mix medium with 10% FBS , 50 ng/ml 2 . 5S nerve growth factor ( Promega Corporation , Fitchburg , WI , USA ) , 100 U/ml penicillin and 0 . 1 mg/ml streptomycin , and seeded onto cover slips of 20 mm diameter in 24-well plates or glass bottom dishes ( Nunc LabTek II Chambered cover glass 4-chamber #155382 , #1 . 5 borosilicate glass , Thermo Scientific ) . The cover slips and glass bottom dishes had been pre-coated with 0 . 01% ( w/v ) poly-L-lysine ( 150 , 000–300 , 000 g/mol , Sigma-Aldrich ) and 7 ng/μl murine laminin ( Invitrogen ) . The cells were cultured in a humidified incubator at 37°C and 5% CO2 , and the media were replaced twice a week . 1-β-D-arabinofuranosylcytosine ( Sigma-Aldrich ) was added at 1 to 2 div to a final concentration of 2μM to suppress proliferation of dividing , non-neuronal cells , but removed prior to HSV-1 infection . We used rabbit polyclonal antibodies ( pAb ) to detect VP26 [VP26aa95-112; 113] , pUL36 [#147; 43] , pUL37 [32] , pUL20 [114] , VP13/14 [R22 , 115] , gB [R69 , 116] or VP16 ( BD 3844–1 , Becton-Dickinson , Franklin Lakes , NJ , USA ) . Mock infected neurons showed little binding to antisera raised against HSV-1 proteins that had been cleared by pre-adsorption [89] on uninfected neurons . Mouse monoclonal antibodies ( mAb ) were used to detect VP22 [mAb22-3; 117] or gD [mAb DL6; 118] . To detect host antigens , we used mAb 1501 for actin ( Millipore , Billerica , MA , USA ) , mAb 5564 for β-III-tubulin ( Millipore ) , mAb 106/36 for ankyrinG ( E9PE32; UC Davis/NIH NeuroMab Facility ) , SMI-310 for phosphorylated 200 kDa and 160 kDa neurofilament ( ab24570 , Abcam , Cambridge , UK ) , and SMI-320 for non-phosphorylated 200 kDa neurofilament ( ab28029 , Abcam ) . We used the clinical isolate HSV1 ( 17+ ) [119] and its derivatives HSV1 ( 17+ ) Lox [20] , HSV1 ( 17+ ) Lox-CheVP26 [20] , HSV1 ( 17+ ) Lox-ΔUL36 [20 , 47] , HSV1 ( 17+ ) Lox-CheVP26-ΔUL36 [20 , 47] , HSV1 ( 17+ ) Lox-ΔUL37 [20] , HSV1 ( 17+ ) Lox-CheVP26-ΔUL37 [20] , HSV1 ( 17+ ) Lox-ΔUL20 ( see below ) and HSV1 ( 17+ ) Lox-CheVP26-ΔUL20 ( see below ) . Our pHSV1 ( 17+ ) Lox BAC plasmids contain a BAC cassette with a chloramphenicol resistance gene , a Cre recombinase gene with an intron under the control of a eukaryotic promoter , a single flippase recognition target site ( FRT ) and LoxP sites at both ends inserted between the genes UL22 and UL23 , and an almost complete HSV1 ( 17+ ) genome lacking only the OriL [84 , 85] . The Cre recombinase excises the BAC cassette upon transfection into eukaryotic cells . HSV-1 stocks were prepared in BHK-21 , HS30 ( for -ΔUL36 ) , 80C02 ( for -ΔUL37 ) or Flp-In-CV-1-pUL20 ( for -ΔUL20 ) , and extracellular virus was harvested by sedimentation from the supernatant of infected cells as described previously [84 , 89 , 90] . After DNase treatment , the inocula of HSV1 ( 17+ ) Lox-ΔUL37 and Lox-CheVP26-ΔUL37 had genome to PFU ratios below 2 , 300 while that of the other strains had genome to PFU ratios of below 108 [120] . To express EB3 in neurons , we used lentiviral transduction . The spleen focus forming virus promoter was replaced by the human cytomegalovirus immediate early promoter in the plasmid pRRL . PPT . SF . GFPpre [121 , provided by Axel Schambach , Hannover Medical School , Hannover , Germany] via PCR by generating 5’PstI and 3’BamHI restriction sites adjacent to the HCMV immediate early promoter of pCMV-Tag 2B ( Agilent Technologies , Santa Clara , California , USA ) using the primers 5’-GAACCTGCAGCGTATTACCGCCATGCATTAGT-3’ and 5’-GAACGGATCCCCAGCTTTTGTTCCCTTTAGTG-3’ . Furthermore , with the restriction sites 5’NdeI and 3’AgeI , parts of the human cytomegalovirus immediate early promoter and EB3 were cloned from pEGFP-N-EB3 [51 , provided by Marco van Ham , Helmholtz Centre for Infection Research , Braunschweig , Germany] to generate pRRL . PPT . HCMV . GFPEB3pre . HEK 293T cells ( 5 x 106 per 10 cm dish ) were transfected with 5 μg pRSV_Rev ( provided by Axel Schambach ) , 2 μg pMD2 . g ( Addgene Inc . , Cambridge , MA , USA , Cat . No . 12259 , deposited by D . Trono , provided by Axel Schambach ) , 10 μg pCDNA3 . GP . CCCC ( provided by Axel Schambach ) , and 10 μg transfer plasmid as described previously [122] . The supernatants were harvested and spun in a Beckman SW 28 rotor at 23 , 000 rpm or a SW32 . Ti rotor at 24 , 000 rpm for 90 min at 4°C ( Beckman Coulter , Krefeld , Germany ) . The resuspended lentiviral particles were snap frozen in liquid N2 and stored at -80°C . For lentiviral transduction , DRG neurons were prepared and seeded as described and at 1 div neuronal growth medium was replaced by one containing 20 mM HEPES and the lentiviral particles . HSV-1 inocula titers were determined by plaque assays [89 , 90 , 120] . Vero , Flp-In-CV-1 or the derivative Flp-In-CV-1-UL20-expressing cells were cultured in 6-well dishes for 16 to 20 h to almost confluency . The respective inocula were diluted in CO2-independent medium ( Life Technologies Gibco ) containing 0 . 1% ( w/v ) cell-culture grade bovine serum albumin ( BSA; Capricorn Scientific , Ebersdorfergrund , Germany ) , added to the cells for 1 h on a rocking platform at room temperature , and then replaced by growth medium containing 20 μg/ml pooled human IgGs ( Sigma-Aldrich ) to neutralize HSV-1 in the culture medium . At 3 dpi , the cells were fixed with ice-cold , water-free methanol and air-dried prior to staining with 0 . 1% ( w/v ) crystal violet in 2% ( v/v ) ethanol for 1 min . After removing the excess of crystal violet , the cells were air-dried , and plaques were counted while using a binocular loupe ( Nikon , Tokyo , Japan ) to calculate the virus titer as plaque forming units ( pfu ) per ml . We used the BACs pHSV1 ( 17+ ) Lox and pHSV1 ( 17+ ) Lox-CheVP26 to construct HSV1 ( 17+ ) Lox-ΔUL20 and HSV1 ( 17+ ) Lox-CheVP26-ΔUL20 . To prevent expression of pUL20 , we mutated the ATG-start codon and a second ATG codon of UL20 to CTG followed by an immediate insertion of 3 stop codons . Recombinant PCR fragments of pEPkan-S2 [provided by B . Karsten Tischer and Nikolaus Osterrieder , Freie Universität Berlin , Germany , 123] were amplified to mutate the 5’ region of UL20 using traceless Red recombination , and transformed for homologous recombination into E . coli GS1783 ( provided by G . Smith , Northwestern University , Chicago , IL , USA ) harboring pHSV1 ( 17+ ) Lox or pHSV1 ( 17+ ) Lox-CheVP26 [85 , 124] . We used the forward primer 5’-CCTTGCGGTTTCGGTCTCCCCACCTCCACCGCACACCCCCTGACCCTGTAGTAATAGCGGGATGACCTTCCTCTGGTTAGGGAAACAGGTAATCGATTT-3’ , the reverse primer 5’-TCGTCGACCAGATCTCGATCACCAGAGGAAGGTCATCCCGCTATTACTACAGGGTCAGGGGGTGTGCGGTGGCAGGTGGTGCCAGTGTTACAACCAATTAACC-3’ , the forward sequencing primer 5’-AAAGACCGGCTGGGTATG-3’ , and the reverse sequencing primer 5’-GGGCGTAGGCGTAAATTC-3’ . The mutated start codons are underlined and the inserted stop codons are shown in bold . The BAC plasmids were digested with 15 U/μg DNA of AscI , BamHI , XhoI or HindIII for 3 . 5 h , and analyzed on 0 . 6% ( w/v ) agarose gels in 0 . 5x TBE buffer ( 0 . 44 M Tris-HCl , 0 . 44 mM boric acid , 10 mM EDTA , pH 8 ) run at 66 mA for 17 h ( Peqlab system , Erlangen , Germany ) . To reconstitute viruses , sub-confluent Vero or Flp-In-CV-1-UL20-expressing cells were transfected with 10 μg BAC-DNA ( MBS mammalian transfection kit; Stratagene , La Jolla , CA , USA ) per 6 cm dish and cultured until cytopathic effects had developed . Cells and medium were pooled , virus was released by three cycles of freeze-thawing and the resulting cell pellet was used to further propagate the virus according to standard protocols . The genomes of the novel HSV1 ( 17+ ) Lox strains were sequenced around the pUL20 start site that had been targeted , and the engineered mutations were confirmed . For the growth curves , sub-confluent Vero , Flp-In-CV-1 or the derivative Flp-In-CV-1-UL20-expressing cells were inoculated at 5 pfu/cell , the cells and the supernatants were harvested at the indicated time points , and virus titers were determined by plaque assay . Vero cells were seeded on cover slips , 3 . 5 cm cell culture dishes or glass-bottom chambers ( Nunc LabTek II Chambered cover glass 4-chamber #155382 , #1 . 5 borosilicate glass , Thermo Scientific ) and infected 16 to 20 hours after the seeding as described before [16 , 20 , 29 , 84 , 89 , 120 , 125] . For synchronous infections , the cells were pre-cooled for 20 min on ice , and inoculated with 10 pfu per cell or mock treated as a control in CO2-independent medium containing 0 . 1% ( w/v ) BSA for 2 h on ice while rocking . The cells were then shifted to regular growth medium at 37°C and 5% CO2 for 1 h . Non-internalized virus was inactivated at 4°C by a 3 min acid wash ( 40 mM citrate , 135 mM NaCl , 10 mM KCl , pH 3 ) . Neurons cultured on cover slips in 24-well plates were pre-incubated at room temperature with CO2-independent medium for 20 min , and inoculated with 1 to 5 × 106 PFU in 200 μl per well in CO2-independent medium . After 30 min , the virus-suspension was replaced by 500 μl F-12 medium and cells were incubated again at 37°C and 5% CO2 . Cells were lysed with hot sample buffer ( 50 mM Tris-HCl , pH 6 . 8 , 1% [w/v] SDS , 1% [v/v] β-mercaptoethanol , 5% [v/v] glycerol , 0 . 001% [w/v] bromophenol blue ) containing protease inhibitors AEL ( aprotinin , E-64 , leupeptin , Sigma ) , ABP ( antipain , bestatin , pepstatin , Sigma ) and PMSF ( Roth , Karlsruhe , Germany ) . The proteins were separated by SDS-PAGE in 12 . 5% gels and transferred in 48 mM Tris , 380 mM glycine , 0 . 1% [w/v] SDS and 10% [v/v] methanol to nitrocellulose membranes ( Pall Corporation , Pensacola , FL , USA ) . After blocking with 5% ( w/v ) low-fat milk powder in PBS containing 0 . 1% ( v/v ) Tween 20 , the membranes were incubated with primary antibodies and secondary antibodies coupled to alkaline phosphatase ( Dianova , Hamburg , Germany ) , transferred to 100 mM Tris-HCl , pH 9 . 5 , 100 mM NaCl , 5 mM MgCl2 , and stained with 0 . 2 mM nitroblue tetrazolium chloride and 0 . 8 mM 5-bromo-4-chloro-indolyl-3-phosphate . For documentation , the membranes were imaged with a digital scanner ( ScanJet 6300 , Hewlett Packard , Wilmington , DE , USA ) . Cells were fixed at room temperature with 3% ( w/v ) paraformaldehyde ( PFA ) in PBS for 20 min , followed by 50 mM NH4Cl for 10 min and permeabilization with 0 . 1% Triton-X-100 for 5 min , or at 37°C with PHEMO fix [3 . 7% ( w/v ) PFA , 0 . 05% [w/v] glutaraldehyde , 0 . 5% [v/v] Triton-X-100 in PHEMO buffer with 68 mM PIPES , 25 mM HEPES , pH 6 . 9 , 15 mM EGTA , 3 mM MgCl2 , 10% [v/v] dimethyl sulfoxide] , washed two times with PHEMO buffer followed by the NH4Cl treatment . The HSV-1 Fc-receptor and unspecific protein binding sites were blocked with 10% ( v/v ) human serum of HSV-1-seronegative healthy volunteers and 0 . 5% ( w/v ) BSA . Samples were labeled with primary and pre-adsorbed secondary antibodies to prevent cross reactivity to antibodies of other species , namely goat-anti-mouse coupled with rhodamine X , carbocyanine 5 or AlexaFluor488 , and goat-anti-rabbit coupled with AlexaFluor488 ( Invitrogen ) . DNA was stained using To-Pro-3-iodide or DAPI dyes ( Invitrogen ) , and the cover slips were mounted in Mowiol 4–88 containing 10% ( w/v ) 1 , 4-diazabicyclo-[2 . 2 . 2]octane . The specimens were analyzed by confocal fluorescence microscopy ( LSM 510 Meta , software LSM 510 version 4 , ZEISS , Göttingen , Germany ) . Pseudo-coloring , brightness and contrast were adjusted identically across each set of images using Adobe Photoshop CS4 . To analyze systematically the degree of co-localization of tegument ( anti-VP22 , red channel ) or envelope proteins ( anti-gD , red channel ) with capsids ( anti-VP26 , green channel ) , and the targeting efficiency of the different viral particles to the axons , we developed a novel image processing pipeline . The semi-automated approach consists of three steps: ( i ) manual annotation of regions of interest ( ROIs ) to identify the axons ( cf . S4 Fig ) , ( ii ) automated particle detection of the capsids , and ( iii ) automated detection of the co-localizing tegument or envelope protein signal using pixel-wise image classification . ( i ) ROI selection: To distinguish the axon from other structures , the regions of interest were manually selected . To simplify the process of region annotation , a ROI is defined by positioning a few control points in the image and adjusting the region width at the control points , as shown in S4 Fig . Usually , three to four control points were sufficient for accurate annotation . ( ii ) Capsid particle detection: The capsids have a spherical shape with a characteristic diameter . Due to blurring caused by the image acquisition process , the fluorescence signal of a single capsid was modeled by an isotropic Gaussian distribution . Template matching was applied to discriminate between well-separated viral particles and other signals such as background noise or agglomerations of multiple , non-separable capsids . In detail , a detector response map Rσ ( x , y ) was computed for each pixel ( x , y ) of the image using normalized cross-correlation of the image signal Icapsid ( x , y ) with a Gaussian shaped template model Tσ ( u , v ) =exp[−u2+v22σ2] with variance σ2 and filter size of 1 + 2 ⋅ ⌈σ ⋅ s⌉ . Candidate particles were extracted from the detector response map Rσ ( x , y ) by searching for local maxima in Rσ ( x , y ) using an 8-neighborhood . In case of a plateau , a morphological shrinking operator was used to only consider the central position of the plateau as a detection candidate . To discriminate from noise , the candidate points pi = ( xi , yi ) were only accepted as capsids if the detector response Rσ ( xi , yi ) was greater than a threshold τcorr and the fluorescence signal Icapsid ( xi , yi ) was greater than a threshold τintens . The parameters σ , s , τcorr , and τintens were trained on manually annotated test images using grid search . The final parameter values used throughout the experiments were σ = 0 . 7 , s = 1 . 5 , τcorr = 0 . 4 , and τintens = 0 . 5 . ( iii ) Tegument or envelope protein signal detection: We used pixel-wise image classification in order to discriminate between signal and background noise . The classifier used various features derived from the gD or VP22 fluorescence signals IgD ( x , y ) : Gaussian blurred intensity signal IσgD=IgD*Tσ , non-linearly distorted intensity Es;σgD=exp ( s⋅IσgD ) normalized cross correlation ( NCC ) of IgD to a Gaussian kernel RσgD=NCC ( IgD , Tσ ) , and a high-pass filter clamped to values between 0 and 1 . Finally , the feature vector is F=[R0 . 5gD , R1gD , R2gD , R4gD , IgD , E10;0gD , I2gD , H2gD , I5gD , H5gD , E10;5gD , I10gD , H10gD , E20;10gD , I20gD , H20gD , E50;20gD] . We used a logistic regression classifier which was trained on manually annotated test images . The gD or VP22 regions were extracted from the resulting pixel-wise classification images using a connected components analysis with an 8-neighborhood . A capsid was said to co-localize with gD or VP22 if the capsid center position coincided with a gD or VP22 region dilated by 1 pixel . The dilation operation accounted for uncertainties in both the capsid position estimation and the pixel-wise gD or VP22 classification . Several statistical parameter were derived from the automatically detected capsids and gD or VP22 regions . The number of capsids or gD and VP22 regions per 100 μm was derived by projecting the particle locations orthogonally to the axial center line of the ROI as illustrated in S4 Fig . For differential interference contrast imaging and to avoid evaporation at 37°C , a glass lid was attached to the chambers with vacuum grease ( Dow Corning , Midland , Michigan , USA ) . Movies of infected Vero cells were recorded at 37° with a high temporal resolution of 5 images per second and a pixel size of 79 nm using a confocal laser scanning microscope equipped with a heating unit ( PeCon , Erbach , Germany ) as described before [20] . Movies of infected DRG neurons were recorded at 37°C ( Incubator from PeCon , Erbach , Germany ) with high temporal resolution 20 images per second on a Nikon Ti microscope equipped with a Yokogawa ( Ratingen , Germany ) CSU-X1 spinning disk and an Andor iXion Ultra 897 EMCCD camera ( Belfast , UK ) . Automated tracking of cytoplasmic capsid motility was performed using the ImageJ plugin MOSAIC [63 , 126] . First , the nuclei were identified by the confined nuclear capsids mobility and the nuclear capsid fluorescence was removed . Second , the Gaussian blur filter of ImageJ was applied ( Sigma radius 2 pixel ) . No further image processing was performed . Third , automated tracking was performed using the Particle Tracking plugin with the following settings: Kernel radius 3 , Cutoff radius 3 , percentile 0 . 5 , displacement 20 , and link range 1 . Of the tracks identified , we only considered further tracks with a length of 10 frames or longer . Track length and maximum step velocity were calculated from tracks with a MSD exponent of 1 . 2 or larger , representing non diffusive transport events . Cells were infected as described above and fixed at the indicated time points . For conventional electron microscopy cells were fixed with 2% glutaraldehyde in cacodylate buffer ( 130 mM ( CH3 ) 2AsO2H , pH 7 . 4 , 2 mM CaCl2 , 10 mM MgCl2 ) for 1 h at room temperature . Cells were washed and contrasted with 1% ( w/v ) OsO4 in cacodylate buffer ( 165 mM ( CH3 ) 2AsO2H , pH 7 . 4 , 1 . 5% ( w/v ) K3[Fe ( CH ) 6] followed by 0 . 5% ( w/v ) uranyl acetate in 50% ( v/v ) ethanol overnight . The cells were embedded in Epon plastic ( Serva , Heidelberg , Germany ) and 50 nm ultrathin sections were cut parallel to the substrate . Cytoplasmic capsids were counted and the respective cellular areas that had been sampled were measured using an Image J plugin . For immunoelectron microscopy , infected cells were fixed at the indicated time points with 2% ( w/v ) PFA and 0 . 2% [w/v] glutaraldehyde , in PHEM-buffer ( 60 mM PIPES , 25 mM HEPES , pH 6 . 9 , 10 mM EGTA , 2 mM MgCl2 ) , embedded , frozen and sectioned . Sections were labeled using specific antibodies and protein-A gold of 10 nm ( Cell Microscopy Centre , Utrecht School of Medicine , The Netherlands ) . Sections were contrasted using 0 . 5% ( w/v ) uranyl acetate in 2% methylcellulose ( Merck ) [127] . Images were acquired with an electron microscope at 200 kV equipped with an Eagle 4k camera ( Tecnai G2; FEI , Eindhoven , The Netherlands ) . Immunogold labeling was quantified by counting each gold particle within a radius of 100 nm around the center of a cytoplasmic capsid [20] . Data were analyzed by using Kruskal-Wallis followed by Dunn’s post testing , and the p values were adjusted for multiple testing ( software Prism , version 6; Graphpad , San Diego , CA , USA ) . The mice ( strain C57Bl/6JHanZtm , not genetically modified ) were bred and maintained without any perturbation . On the day of the experiment they were picked up from the animal facility and within 3 h sedated with CO2-inhalation prior to killing by cervical dislocation without any prior experimental perturbation . DRG were dissected afterwards . According to the German Animal Welfare Law §4 , killing of animals does not need approval if the removal of organs serves scientific purposes and the mice had not undergone experimental treatment before . The animal care and sacrifice was performed in strict accordance with the German regulations of the Society for Laboratory Animal Science ( GV-SOLAS ) , the European Health Law of the Federation of Laboratory Animal Science Association ( FELASA ) and the German Animal Welfare Law . According to the German Animal Welfare Law , this study does not contain animal experiments that require pre-approval , but the total number of killed mice was reported at the end of each year to the animal welfare deputy of Hannover Medical School . The number of animals killed according to §4 of the German Animal Welfare Law was registered with the animal welfare application number 2012/20 at the LAVES ( Niedersaechsisches Landesamt fuer Verbraucherschutz und Lebensmittelsicherheit , Oldenburg , Germany ) , and the experiments were performed before 2013 . Human sera of adult , healthy , HSV-1 seronegative volunteers were obtained after written informed consent by the blood donors . Permission was granted by the Institution Review Board ( Hannover Medical School; Approval Number 893 ) . | Human and animal alphaherpesviruses establish lifelong latent infections in neurons of the peripheral nervous system and cause many diseases upon primary infection as well as following reactivation from latency . The highly prevalent human herpes simplex viruses HSV-1 and HSV-2 are responsible for facial and genital herpes , potentially blinding eye infections , and life-threatening encephalitis and meningitis . Here , we asked how these viruses master the bottleneck of being targeted from the neuronal somata to the axons . Our data suggest that only transport vesicles harboring fully matured virus particles can enter axons for spreading infection to the brain or the peripheral organs . Our data imply that the limiting membrane of the transport vesicles must expose viral or host receptors to recruit the microtubule motors required for axonal transport . Inhibiting such viral factors on the surface of the transport vesicles might provide novel therapeutic approaches to prevent the spread of alphaherpesviruses in the nervous system . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"vesicles",
"microtubules",
"cell",
"processes",
"microbiology",
"viral",
"structure",
"neuroscience",
"nerve",
"fibers",
"cellular",
"structures",
"and",
"organelles",
"cytoskeleton",
"animal",
"cells",
"axons",
"viral",
"packaging",
"viral",
"replication",
"virions",
... | 2017 | Inner tegument proteins of Herpes Simplex Virus are sufficient for intracellular capsid motility in neurons but not for axonal targeting |
It has recently been estimated that a single HIV-1 infected cell produces between and more than viral particles over its life span . Since body-wide estimates of the ratio of free virus to productively infected cells are smaller than and much smaller than , individual virions must be cleared rapidly . This seems difficult to reconcile with the fact that most of the total body virus is trapped on follicular dendritic cells where it can survive for many months . It has also been difficult to reconcile the vast difference in the rates at which the virus is cleared from the blood in rhesus macaques and in chronically infected patients . Here we attempt to reconcile these seemingly contradictory observations by considering the virion clearance rate in various organs and the virion exchange rates between them . The main results are that the per capita clearance rate of free virus in lymphoid tissue should be fast , the virion exchange rate between lymphoid tissue and the blood should be slow , and the comparatively slow previous estimates for the virion clearance rate from the blood correspond to the rate of virion efflux from the blood to other organs where the virus is ultimately cleared .
The major targets of HIV and simian immunodeficiency virus ( SIV ) infection are CD4 T cells [1] . During the acute stage of infection large numbers of resting and memory CD4 T cells disappear from the lymphoid tissues and mucosal layers , particularly in the gut , by direct infection and by bystander effects . Lymphoid tissue remains the primary site of infection after acute infection has resolved and the viral load approaches a steady state called the “set-point” [1] . Chronic infection is characterized by a set-point viral load , and rapid turnover of productively infected cells [2]–[4] . To maintain this steady state requires a balance between virus production and clearance , and between target cell production and death . Combining image analysis with in situ hybridization in lymphoid tissue from patients chronically infected with HIV-1 , the total number of productively infected CD4 T cells has been estimated to be cells , and the total number of HIV-1 particles has been estimated to exceed virions [5] , [6] . At steady state , one could naively conclude that a single productively infected CD4 T cell should therefore account for a viral load of approximately 500 virions . We shall show below that the situation is more complex . To understand viral production and clearance better , one needs to consider the current quantitaive estimates of viral production and clearance , as well as where these processes are occurring in the body . Most of the total body virus is located in lymphoid tissues , typically in association with follicular dendritic cells ( FDCs ) [1] , [5] . FDCs trap virus and retain it on their surface for many months [7]–[11] . The FDC associated virus pool fills up during early infection , i . e . , not later than a few days after onset of symptoms , and does not expand over the course of chronic HIV-1 infection [12] . Although most of the virus resides in this fairly constant storage on FDCs , the store rapidly declines during antiretroviral treatment ( ARV ) [13] , suggesting the existence of a quasi steady state between free and FDC associated virus in the lymphoid tissue [8]–[10] , [14] . In a mouse model it was shown that a small fraction of HIV-1 persisted on FDCs , and remained infectious over a period of 9 months [7] . A recent study confirmed the long-lived nature of this reservoir in humans [11] . Viral clearance rates have been estimated in the blood , and different techniques have yielded a variety of estimates [4] , [15]–[17] . Rapid clearance rates with half-lives of 3–4 minutes were found after infusion of SIV into the blood of uninfected and infected rhesus macaques [16] , [17] . In patients chronically infected with HIV-1 , more than 10-fold slower clearance rates were found using plasma apheresis to increase viral clearance , and hence perturb the viral set point [15] . By plasma apheresis approximately particles were removed over a period of two hours , and this reduced the viral load in blood , with a nadir that was approximately half of the original viral load [15] . The mere fact that the removal of less than 1% of the total body virus lead to an observable decline in the plasma virus load [15] , suggests that the exchange of virus between the lymphoid tissues and the blood cannot be rapid [14] . Further , sequence analysis of virus in splenic white pulps suggests that virus trapped on FDC is produced locally [18] , supporting the notion of slow viral exchange between blood and lymphoid tissue . Virus production rates have also been estimated by several techniques . Because in other lentivirus infections , most notably visna virus , intracellular viral DNA levels increase approximately exponentially and then virus appears to be released rapidly [19] , the term burst size is commonly used to describe the total amount of virus produced by an infected cell [20] . If one knows the burst size , , and the average lifespan of a productively infected cell , , then the viral production rate , , is given by . Note that is then the average rate of virion production over the lifespan of a productively infected cell . Current estimates suggest d [21] . Thus , the burst size corresponds to the daily viral production rate . Recent studies have combined image analysis with in situ hybridization to estimate burst size . Assuming that the maximal HIV RNA count in a cell corresponds to the burst size , Haase et al . [5] estimated a production rate of approximately a hundred particles over the life span of a productively infected cell . Hockett et al . [22] quantified more precisely the amount of viral RNA ( vRNA ) per cell by a PCR technique . They found an average of 3900 ( range 3162–5011 ) vRNA copies per infected cell , and because of limited variation in the number of copies per cell , they concluded that viral production is a few thousand virions per cell , and also assumed that bursting was an all-or-none phenomenon [22] . However , the estimates by Haase et al . [5] and Hockett et al . [22] are based on measuring HIV RNA at a single time point . If infected cells continue to produce virus over an extended period , then these estimates would be underestimates of the true total cellular production of virus . One can also attempt to measure burst size by directly imaging the extracellular viral particles surrounding an infected cell [23] . Using this method , Reilly et al . [23] found and copies of HIV-1 RNA surrounding infected activated and resting CD4 T cells , respectively , in the lymphoid tissue of acutely SIV-infected rhesus macaques . They then fitted a five parameter model to this data , with three of the parameters describing the rate of viral production as a function of time since infection ( see Methods ) , and the remaining parameters describing the rate of exponential decay of cells producing virus , and the rate of loss of viral particles . Using this model , they estimated that the half-life of virus located around CD4 T cells producing virus in lymphoid tissue was approximately three hours . However , this half-life combined diffusion of virus out of the local area and true virion clearance [23] . Even if all loss was due to clearance , a three hour half-life corresponds to a per virion clearance rate of d . With this estimate , Reilly et al . [23] calculated median production rates of approximately 1500 and 1400 viral particles per activated cell , and of approximately 650 and 3400 viral particles per resting cell , depending on two different assumptions for the half life of productively infected resting cells ( see Methods ) . Finally , the most direct estimates for the total amount of virus produced per infected cell was achieved using single-cycle SIV to infect PBMC which were placed back in uninfected rhesus macaques . By measuring the total amount of virus produced and accounting for clearance , this experiment yielded a total production of approximately ( range – ) virions per infected cell [24] . Because productively infected cells have a lifespan of about one day , the cellular burst size estimates of Chen et al . [24] imply daily production rates of approximately virions . Summarizing , the latest production rate estimates converge on a few thousand to approximately virions per productively infected cell [22]–[24] . The production rate estimates of Reilly et al . and Chen et al . depend on the viral clearance rate , . The 10-fold range in the estimated production rates is at least partly due to differences in the clearance rate used in the calculations . Reilly et al . [23] estimate that d in lymphoid tissue , while Chen et al . [24] used a previous estimate of d in the blood [15] . Since , our main result will be that the clearance of free virus in lymphoid tissue should be fast , and that the observed clearance from the blood is not clearance but the rate of efflux to other organs , we will vary the production rate in our analysis and consider to particles per cell as potential realistic estimates . Finally , note that different cell types , e . g . , infected macrophages , may have different production rates than infected T cells . Here we consider that the vast majority of virus is produced by infected CD4+ T-cells [4] , [25] , and hence use estimates of production from those cells . In one earlier modeling study a production rate of several thousand particles per cell was shown to be consistent with a viral half-life of 3–4 hours in the lymphoid tissue [14] , suggesting that the recent estimates of 10-fold higher production rates [24] imply even shorter half-lives . However , large total viral production per infected cell [22]–[24] and the short viral half-lives they imply [14] , seem difficult to reconcile with the suggestion that most of the virus in the lymphoid tissue is long-lived and in association with FDCs [5] . The problem of balancing production with clearance can be introduced by a simple calculation that assumes the body is a single well-mixed compartment . An order of magnitude estimate for the total number of productively infected cells in a human is cells [5] , [6] . For a human with a viral load of approximately particles ml of plasma , and an estimated total of 15 liters of extracellular body water in which virus could distribute , one estimates that there are a total of free virus particles in extracellular fluids ( i . e . , only about 2% of the estimated total body load ) [4] . Requiring steady state in the conventional model for virus production , i . e . , , with a production rate of viral particles per infected cell , , per day , and a steady state of free virus particles and productively infected cells , one would need a per virion clearance rate of d , which is much higher than published estimates for the viral clearance rate in humans [15] , but resembles the rapid clearance rate observed in rhesus monkeys [17] . Even if d then d is needed to balance production , which is still larger than the current clearance rate estimate in humans [15] . In this paper we attempt to reconcile the various estimates for the viral clearance rate , the viral production rate , and the amount of long-lived virus trapped on FDCs within one modeling framework in order to test whether there is a consistent interpretation explaining all observations . To do so we introduce compartmental models to analyze recent experimental data on viral clearance in various organs , and estimate the rates at which virus is exchanged between them .
A simple and direct approach to estimate the clearance rate of virus from the blood is to infuse virus particles into the blood , and monitor their disappearance by taking frequent blood samples . Zhang et al . [16] , [17] administered large amounts of SIV to infected and uninfected rhesus macaques by an intravenous bolus injection ( to viral particles ) , or by constant intravenous infusion ( viral particles min ) . From the rate at which virus was lost from the plasma afterwards , plasma half-lives of 3–4 minutes were estimated [16] , [17] . These half-lives were similar in infected and uninfected monkeys . Virus did not appear to be lost from the plasma by binding to erythrocytes , PBMCs , granulocytes , or platelets because there was no evidence of virion binding to these cell types [16] . In another experiment , viral clearance in various organs was tracked by injecting radioactively labeled SIV into macaques , and measuring the percentage of radioactivity and of SIV RNA persisting in various organs after two hours [17] . Because 30% of the radioactivity , and only 0 . 053% of the injected SIV RNA , was recovered from the liver ( see Table 1 ) , it was concluded that the liver plays a major role in viral degradation [17] . If most of the viral degradation indeed takes place in organs such as the liver [17] , most of the measured clearance from the blood would be efflux from the blood into the organs . Therefore , we write the following simple model for the amount of radioactive virus in the plasma , , and in a particular organ , , ( 1 ) where is the daily efflux from the blood , is the fraction that arrives in the particular organ , and is rate of clearance in the organ . We neglect the flux from the organ back to the blood during this 2 hr experiment because this simplifies the analyses , but also because the results of the bolus injection and the continuous virus infusion experiments [16] , [17] , yielded similar disappearance of virus from plasma in uninfected and already infected monkeys , indicating that this flux is very small in these short-term experiments . If is the total amount of virus injected into the plasma , then ( 2 ) After two hours , the fraction of infused viral RNA found in the organ is ( 3 ) where the converts 2 hrs into a per day timescale . The fraction of radioactivity ending up in the organ after the two hour experiment , , is assumed to be equal to the fraction of virus entering ( and presumably degraded in ) that organ , i . e . , ( 4 ) Using the estimated clearance rate from the blood of 0 . 2 min [17] as the efflux rate , d , we estimate the fraction , , and the clearance in the organ , , from the Zhang et al . [17] data shown in Table 1 . Since in Eq . ( 4 ) the term , the fraction of measured radioactivity in the organ , , determines the parameter in the model . Substituting the fraction of SIV RNA in the organ , , the estimated efflux , , and the fraction of radioactivity , , into Eq . ( 3 ) , one can numerically solve for the clearance rate constants , , in the four organs ( Table 1 ) . The estimated clearance rates in the various organs vary from d in the liver to d in the lymph nodes . The latter is only 2-fold faster than the clearance rate of d estimated by Reilly et al . [23] for lymphoid tissue . Only 40 . 5% of the total radioactivity was recovered in the monkey 2 hr after injection . This could be due to a loss of radioactivity by viral degradation and removal of labeled molecules , or to accumulation of SIV in other body compartments , such as the gastrointestinal tract , that were not examined [17] . The former we can correct for by re-normalizing the radioactivity data so that the total is 100% . This correction doubles the estimate clearance rate in lymph nodes , and has a smaller effect on the other clearance rates ( see Table 1 ) . If the virus unaccounted for by the radioactivity data is ending up in other organs , the clearance rates based upon the uncorrected radioactivity data should be valid . Interestingly , percentages of SIV RNA were also measured in ileum , cecum , duodenum and rectum in other monkeys , and adding these data from the gut to the “Others” class hardly increased the amount of SIV RNA in that class [16] . This would argue that only a minor fraction of the injected virus ends up in the gastrointestinal tract , and/or that the clearance rate in the gastrointestinal tract is much larger than in the other organs so that SIV RNA is not found there . Unfortunately , there is no radioactivity data for the gastrointestinal tract to distinguish between these two possibilities , and we estimate the viral half-lives in the various organs by the ranges indicated in Table 1 , as obtained from the corrected and the uncorrected radioactivity data , respectively . Summarizing , modeling the radioactivity data provides estimates of viral clearance rates between d and d in various organs ( Table 1 ) . HIV-1 clearance rates from plasma have been estimated in chronically infected patients by plasma apheresis over a period of two hours . Plasma was removed at a rate of 39 mL per min , and was replaced by an equivalent volume of isotonic saline containing 5% albumin . On average , this procedure removed a total of approximately particles , and resulted in a nadir of virus equal to half the initial viral load [15] . The following model , formally equivalent to the one presented in Ramratnam et al . [15] , was used to fit the data: ( 5 ) where is the rate of virus influx from the lymphoid tissue , is the normal efflux from the blood , and is the additional rate of virus removed from the blood due to plasma apheresis . It was assumed that over the course of the two hour experiment the flow of virus into the blood , , remains constant , and it was found that the rate of viral efflux in four patients ranged from to 36 d , with an average of d [15] . The rate of virus influx , , at steady state is estimated by multiplying the plasma efflux rate , , by the initial virus load , and varies from to particles d , with an average of particles d . The mere fact that the removal of less than 1% of the total body virus over a period of two hours led to significant declines in the plasma virus load ( Table 2 and Ramratnam et al . [15] ) , suggests that the plasma virus pool is not rapidly replenished from the lymphoid tissue or other organs[14] . This observation also supports our neglecting virus return from organs back into the plasma in Eq . ( 1 ) . It is worth noting that the estimate for the efflux rate in humans of HIV-1 from plasma is more than 10-fold slower than the estimated plasma efflux rate of SIV in rhesus monkeys [15]–[17] , which could reflect a true difference between these two species . Alternatively , it could be that the plasma clearance rate in the four patients studied by Ramratnam et al . [15] , all of which had high viral loads , is slower than in the monkeys studied which had much lower viral loads [16] , [17] . A potential mechanism for the more rapid clearance in monkeys with low viral load could be the rapid attachment of virus to various receptors on blood born cells , whereas in patients with chronic high viral loads these receptors could be saturated and bind less virus ( see Discussion ) . Since most virus production takes place in the lymphoid tissue , we modify a previously published compartmental model [14] to rewrite the fixed source in Eq . ( 5 ) into a term depending on the amount of virus in lymphoid tissue ( LT ) . We proceed by considering four viral compartments: virus in organs other than LT , , virus in the plasma , , free virus in the lymphoid tissue , , and virus bound to FDCs in lymphoid tissue , . The model has two clearance rates , and , for the rate of clearance of free virus in LT , and in other organs ( like the liver and lung ) , respectively . As before , there is no clearance in blood . Virus bound to FDCs is considered to be long-lived [7] , and virus in the plasma is considered to be lost by migration to organs or LT . We allow for influx of free virus into the plasma from the LT with rate constant , because now we are modeling a long term process , and efflux from the plasma with rate constant . A fraction of the virus leaving the plasma will return to the LT , the rest is cleared in organs like the liver and lung . This compartment model is represented by the following equations: ( 6 ) ( 7 ) ( 8 ) ( 9 ) where is the number of productively infected cells in the LT . In this model we make a distinction between clearance and efflux . We speak of efflux when the number of virus particles is conserved , and speak of viral clearance in an organ only if the virus is being degraded there . Thus , these equations assume that there is no viral clearance from the blood; there only is efflux to the lymphoid tissue and to other organs that have viral clearance rates , and , respectively . Mixing SIV with fresh blood taken from monkeys provided no evidence for viral degradation within plasma ex vivo [17] , and for several viruses most of the clearance takes place in liver and spleen [16] . The model can easily be modified to allow for viral clearance from the blood , e . g . , by decreasing the term in Eq . ( 6 ) . This would not affect our results , however , because Eq . ( 6 ) represents a sink that does not affect the other three compartments of the model . Technically , the “viral clearance” rates from the blood estimated previously by the apheresis [15] and infusion [16] , [17] experiments , cannot distinguish between clearance and efflux , and we will let these estimates represent the efflux rate , , in Eq . ( 7 ) . The parameter in Eqs . ( 8 ) and ( 9 ) is the average dissociation rate of virus bound to FDCs . is the maximum number of binding sites on FDCs that HIV-1 can attach to , and is the rate constant for the association of virus with an FDC binding site . These binding sites include complement receptors and Fc receptors [8] , and possibly other receptors like DC-SIGN [26] . The dissociation process is complicated and depends on the number of bonds by which virus is bound to the FDC [8] . When most virus particles have multiple bonds holding the virus to FDC , which makes the dissociation slow , and probably accounts for the long half-life of a fraction of the bound virus [7] , [8] , [11] . When is large most viruses will have few bonds holding them to FDC and dissociation should be more rapid . We can describe this phenomenologically with a Hill-function such that the effective virion dissociation rate constant increases with ( 10 ) where and are constants with , and when . In a typical patient the FDC pool seems saturated , i . e . , , and most virus is expected to be monovalently bound with a dissociation rate estimated as /sec [8] . Image analysis combined with in situ hybridization suggested that most of the virus in the lymphoid tissue is associated with the FDC network [5] . The FDC associated virus , , fills up early in infection [12] . During chronic infection we therefore assume , and this large pool of FDC associated virus is viewed as a filled store in quasi steady-state that contributes little to the total body viral clearance . When the FDC pool is close to being saturated , the steady state of Eqs . ( 6 ) to ( 9 ) corresponds to ( 11 ) where if , and if . To consider the total virus load we add Eqs . ( 6 ) – ( 9 ) yielding , ( 12 ) where is the total body virus load . For a chronically infected patient assumed to be in a steady state we substitute from Eq . ( 11 ) to obtain ( 13 ) where the total body clearance rate , , is the total steady state rate of clearance taking place in organs like the liver and the lymphoid tissues . Since our aim is to estimate , we rewrite Eq . ( 13 ) as ( 14 ) where we have again used from Eq . ( 11 ) . We can use the relationships between and the steady state levels in Eq . ( 14 ) to study what clearance rates would be required to balance production in a number of typical situations . Since there is ambiguity on the rate of efflux from the blood , , we use the efflux estimates from the plasma apheresis experiments [15] , i . e . , to 36 d , and from the rhesus monkeys [16] , [17] , i . e . , d , as lower and upper bounds to create examples of how viral production and clearance could be balanced in hypothetical patients . If we use a middle value for the fraction of plasma virus returning to the LT , e . g . , , we obtain an upper estimate of d and take a lower estimate of d . To estimate the ratios between the variables in Eq . ( 14 ) we pick an example of a patient in a chronic steady state with total body counts of productively infected cells , virus particles in the peripheral blood , and virus particles in the lymphoid tissue . Finally , because Hockett et al . [22] did not detect virus associated with FDCs in almost half of their patients , and measured 10-fold higher total amounts of virus in lymph nodes from those patients where they could detect virus on FDC , we consider two possibilities . To model a “typical” patient where most of the virus is associated with FDCs , we let 90% of the lymphoid tissue virus be associated with FDCs , and obtain that . To model patients with a smaller pool of FDC associated virus , we also consider the possibility that half of the LT virus is bound to FDCs , i . e . , , which amounts to . This allows us to study how the estimates for the viral clearance rate in LT depend on the fraction of virus bound to FDCs . For the more “realistic” example , Fig . 1a , where most of the virus is associated with FDCs , our estimate of the per capita clearance rate in LT , , depends strongly on the production rate , and a large production rate , e . g . , virions per cell per day , requires rapid clearance of virus in the lymphoid tissue , i . e . , per day to maintain a steady state level of virus ( Fig . 1a ) . The recently proposed production rates of more than viral particles per infected CD4 T cell would require LT clearance rates of per day ( Fig . 1a ) . When Chen et al . [24] estimated these high production rates they were conservatively assuming that d . Because their estimated production is proportional to the assumed clearance rate , our new results suggest that the true production could be even higher . In cases where less virus is associated with FDCs ( Fig . 1b ) , we find a similar relation between the clearance rate in LT , , and the production rate , , but the required clearance rate is approximately 5-fold smaller because we allow for 5-fold more free virus , i . e . , ( Fig . 1b ) . In this case , for realistic virion production rates per cell , e . g . , per day , the estimated clearance rate is fairly independent of rate of efflux from the blood ( ) ; see Fig . 1 . Analysis of the plasma apheresis experiments in humans provided estimates for the influx of virus from the LT into the blood ( Table 2 ) . This can also be calculated from the quasi steady state of Eq . ( 7 ) , i . e . , for the influx of virus from the LT into the blood , one obtains that ( 15 ) where the efflux has the two estimates of d and d in patients and macaques , respectively . For the case when 90% of the lymphoid tissue virus is associated with FDCs , i . e . , , this means that d . Assuming an equal distribution of free and FDC-bound virus in lymphoid tissue , i . e . , , one obtains d . Taking these two cases as extremes , the daily influx of virus from the lymphoid tissue into the blood would be between virus particles per day , i . e . , about to particles per hour . These estimates are in good agreement with the daily influx estimated from the plasma apheresis experiments ( Table 2 ) . Finally , for these estimates of , the clearance rate in LT should approach as the term in Eq . ( 14 ) is negligible . To test whether the full model ( Eqs . 6–9 ) is consistent with the plasma apheresis experiments , we make reasonable guesses for the other parameters of the model . Allowing rapid filling of the pool of virus bound to FDCs we set the number of FDC binding sites and d . With these parameters the initial “on” rate when all FDC sites are free is d ( 10 s ) . To have 100-fold more free virus in LT than in blood , i . e . , ( see Eq . ( 11 ) ) , we set d and d . To have about free virus particles in the LT , we set particles d and d ( see Eq . ( 11 ) ) . During a two hour plasma apheresis experiment we transiently add an estimated efflux of d during apheresis to Eq . ( 7 ) ( like we do in Eq . ( 5 ) ) . For these parameters the model results mimic the plasma apheresis experiments in the blood ( Fig . 2a ) , reducing the viral load in plasma by 30% and accumulating a total of virus particles . Free and bound virus in LT are hardly affected ( Fig . 2b ) . Very similar results are obtained when we increase production and viral clearance in the LT 10-fold to particles d and d ( not shown ) . Explaining the plasma apheresis experiments in humans therefore indeed requires a viral efflux half-life from the plasma of about half an hour [15] . Choosing the efflux rates estimated in monkeys [16] , [17] , we have also set d and d , which delivers a very similar steady state as that shown at time zero in Fig . 2 . Simulating plasma apheresis by setting d for two hours a total of virions are removed , but the plasma virus load decreases by 5% only ( not shown ) . This is a natural result because the additional clearance of d is small compared to the normal efflux from the plasma of d . We therefore predict that if plasma apheresis experiments were repeated in SIV infected rhesus macaques , the viral load in the plasma would hardly be affected .
We have shown that the balance between viral production and viral clearance implies rapid per capita viral clearance rates in lymphoid tissues . The larger the viral production rate per productively infected cell , and the more virus that is bound to FDCs ( see Fig . 1 ) , the larger the per capita clearance rate of free virus in lymphoid tissue must be to balance viral production . Recent estimates of a burst size up to virions per cell and a productively infected cell life span of about a day [21] , [24] , imply viral clearance rates in the lymphoid tissue of to d ( Fig . 1 ) . In our modeling work the rate of virion clearance , , is assumed to be a constant . This is equivalent to assuming that clearance occurs by a first-order process or that virus clearance can be described by an exponential decay . However , it is possible that viral clearance obeys more complex laws . Comparing viral loads in lymph nodes and plasma from 9 patients at relatively advanced stages of disease , Hockett et al . [22] demonstrate that the plasma viral load increases faster than proportional with the number of productively infected cells . One possible explanation is that the viral clearance rate decreases or saturates when the viral load increases . However , this explanation remains speculative as it requires a 100-fold decrease in the clearance rate at their highest viral load , and it seems unlikely that there is such a tremendous variation in the clearance rate [22] . The efflux and/or clearance rate of SIV from the blood of uninfected monkeys and of infected monkeys with a low viral load [16] , [17] is about 10-fold higher than that of HIV-1 measured by plasma apheresis experiments in chronically infected patients [15] . Because the additional removal ( in Table 2 ) realized in the plasma apheresis experiments is also 10-fold smaller than these efflux rates in monkeys , plasma apheresis would have hardly any effect if humans were to have efflux rates similar to these monkeys ( this expectation was confirmed by computer simulation ) . As discussed above this 10-fold difference in the estimated efflux rate from the plasma could reflect a true species difference . A speculative alternative is that efflux from the plasma hinges upon attachment of virus to various receptors on blood born cells , like CCR5 and CD4 on various cell types , gp340 on macrophages [27] , DC-SIGN on dendritic cells [26] and DARC on red blood cells [28] , [29] . Because the monkeys in these experiments had much lower viral loads than the four patients studied by apheresis [15]–[17] , most of the receptors could be free in monkeys and occupied with HIV-1 in chronically infected humans with a high viral load . However , this remains speculative because Zhang et al . [16] found negligible amounts of virus on erythrocytes , peripheral blood mononuclear cells ( PBMC ) , granulocytes , and platelets . Moreover , note that the estimated clearance rate in the liver , the organ that appears to be responsible for most of the peripheral viral degradation [17] , is reasonably close to the high clearance rates of free virus in lymphoid tissue that we estimate to be required for balancing the total body virus production . Finally , it may seem that the notion of a large FDC store of bound virus is incompatible with the rapid viral clearance rates seen in spleen and lymphoid tissues of uninfected macaques [16] , [17] . In other words , if the FDCs were trapping and maintaining the virus in lymphoid tissues , then after radiolabeled virus was injected , the percentages of radioactivity and SIV RNA found in these tissues should have been similar , whereas a 2-fold difference was observed , i . e . , 3% vs . 1 . 4% ( Table 1 ) . There are at least two possible explanations for this difference . First , virus is reversibly bound to FDC [8] so radiolabeled virus could dissociate and the virus could then be degraded . Alternatively , the clearance after bolus infusion of SIV in monkeys was studied over a time window of just two hours [16] , [17] , and one could speculate that most of the added virus in these short experiments fails to bind FDCs , and could therefore be cleared rapidly . Rapid viral clearance in lymphoid tissue is not surprising . The lymphoid tissue contains more than CD4 T cells , i . e . , the ratio of virus to CD4 cells in the LT is approximately 1∶1 , and virus particles will bind CD4 T cells and macrophages , and defective virus particles will be “cleared” by non-productive infection . Phagocytic cells that are also abundantly available in LT may clear virus via various types of receptors , like Fc , complement , and DC-SIGN on dendritic cells [26] . Since the process of virus binding cell surface receptors is relatively fast , it can readily account for the rapid clearance rates that we derive from balancing total body production with total body clearance . Finally , during chronic HIV-1 infection the long-lived pool of virus on FDCs should be in steady state , and thus not contribute to the actual clearance rate of the virus in lymphoid tissue ( see Eq . ( 12 ) ) . In a previous paper we showed that the rate of viral clearance in lymphoid tissue would markedly affect the estimated life span , , of productively infected cells deduced from antiretroviral drug therapy ( ART ) experiments , if the clearance rate in lymphoid tissues were sufficiently slow [14] . Since we now estimate even higher clearance rates than we did previously , it becomes even more likely that the clearance from lymphoid tissue is sufficiently fast to not affect the accuracy of current estimates of , the death rate of productively infected cells . Having , one expects the loss of productively infected cells to be the dominant slope of viral decline during the first week or two of ART [14] . The amount of virus in the blood , free virus in lymphoid tissue , and virus on FDCs should be in quasi steady state with the loss of productively infected cells , which is in good agreement with the rate of about 0 . 5 per day at which virus in lymphoid tissue declines during ART [13] . When the amount of virus on FDCs has dropped significantly , most of the remaining virus will be attached by multiple bonds [8] , which can account for the observed long half lives of virus on FDCs during ART [7] . Summarizing , we have provided new estimates of viral efflux and clearance rates in various organs , including blood and lymphoid tissue . We have confirmed that the exchange rate from lymphoid tissue to the blood should be slow [14] . Whenever viral production rates exceed virus particles over the life time of a productively infected cell [22]–[24] , we estimate clearance rates in lymphoid tissue of 10–100 d for typical situations where most of virus in lymphoid tissue is associated with FDCs .
Reilly et al . [23] fitted data obtained by in situ hybridization in lymphoid tissue , including measurements of the number SIV RNA copies on the surface of and in the vicinity of activated and resting CD4 T cell in lymphoid tissue of acutely SIV-infected rhesus macaques , with averages of and copies per cell . Although the data were static , i . e . , they were single snap-shots of different cells not containing any information on the time since infection of the cells , the data was fit to a dynamic model with a Bayesian approach using simulated annealing to find the most likely parameter values [23] . The model consisted of a three parameter viral production function ( see Fig . 3 , i . e . , intercept , up-slope , and saturation time , and two parameters for the exponential decay of cells producing virus , and loss of viral particles , respectively [23] . The prior distribution for the half-life of activated infected cells was fixed to a mean of 1 . 5 days , whereas having little information on the expected life span of productively infected resting cells , two different prior distributions for the half-life of resting infected cells were chosen , with means of 4 days ( Fig . 3a & b ) and 14 days ( Fig . 3c & d ) , respectively . To estimate viral production rates per virus producing cell from this data , one has to integrate the virus production function over the life span of the cells . Reilly et al . [23] estimated the median production rate by integrating up to the estimated half-life of the cells ( see the heavy lines in Fig . 3 ) , and this yielded median production rates of 1479 and 1395 viral particles per activated virus producing cell for the two prior distributions ( Fig . 3b & d , respectively ) , and of 644 and 3405 viral particles per resting virus producing cell ( Fig . 3a & c , respectively ) . Note that the estimates obtained in Fig . 3 are independent of the estimated time to saturation , and largely depend on the intercept and slope of the virus production function ( Fig . 3 ) . In three of the four panels the heavy black line , reflecting the period over which virus production was integrated , stops well before the saturation point . This is reassuring because the 95% credible intervals on the saturation time are very large . Indeed , these data can hardly support the existence of a saturation time in the production curve because the estimated saturation time is much larger than the estimated half-life of the cells . Thus , the data must have had virtually no cells that became old enough to breach the saturation time , and therefore the data hardly contains any information on possible saturation effects . This implies that the estimated saturation times were largely determined by the prior distribution of the Bayesian parameter estimation procedure . Fortunately , for the interest of this paper , eliminating the saturation barely affects the estimated production rates . | A human cell that is infected with the AIDS virus HIV-1 may produce more than new viral particles over its short life span . In patients chronically infected with HIV-1 , one can estimate that on average there are much less than free viral particles per productively infected cell . This suggests that the rate at which individual virus particles are cleared from the body must be fast . Most of the virus is long-lived , however , because it is trapped on follicular dendritic cells . We attempt to reconcile these seemingly contradictory observations by estimating the virion clearance rate in various organs , and the virion exchange rates between them , using a mathematical modeling approach . We find that individual virus particles are cleared rapidly from the lymphoid tissue , and that the rate at which virus is exchanged between lymphoid tissue and the blood is slow . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"mathematics",
"virology/immunodeficiency",
"viruses"
] | 2010 | Current Estimates for HIV-1 Production Imply Rapid Viral Clearance in Lymphoid Tissues |
Tuberculosis ( TB ) treatment is hampered by the long duration of antibiotic therapy required to achieve cure . This indolent response has been partly attributed to the ability of subpopulations of less metabolically active Mycobacterium tuberculosis ( Mtb ) to withstand killing by current anti-TB drugs . We have used immune modulation with a phosphodiesterase-4 ( PDE4 ) inhibitor , CC-3052 , that reduces tumor necrosis factor alpha ( TNF-α ) production by increasing intracellular cAMP in macrophages , to examine the crosstalk between host and pathogen in rabbits with pulmonary TB during treatment with isoniazid ( INH ) . Based on DNA microarray , changes in host gene expression during CC-3052 treatment of Mtb infected rabbits support a link between PDE4 inhibition and specific down-regulation of the innate immune response . The overall pattern of host gene expression in the lungs of infected rabbits treated with CC-3052 , compared to untreated rabbits , was similar to that described in vitro in resting Mtb infected macrophages , suggesting suboptimal macrophage activation . These alterations in host immunity were associated with corresponding down-regulation of a number of Mtb genes that have been associated with a metabolic shift towards dormancy . Moreover , treatment with CC-3052 and INH resulted in reduced expression of those genes associated with the bacterial response to INH . Importantly , CC-3052 treatment of infected rabbits was associated with reduced ability of Mtb to withstand INH killing , shown by improved bacillary clearance , from the lungs of co-treated animals compared to rabbits treated with INH alone . The results of our study suggest that changes in Mtb gene expression , in response to changes in the host immune response , can alter the responsiveness of the bacteria to antimicrobial agents . These findings provide a basis for exploring the potential use of adjunctive immune modulation with PDE4 inhibitors to enhance the efficacy of existing anti-TB treatment .
Despite effective chemotherapy available for over 50 years , and development of a control strategy of directly observed therapy short-course ( DOTS ) , tuberculosis ( TB ) remains the leading cause of adult mortality attributable to a single infectious disease [1] . M . tuberculosis ( Mtb ) , the causative agent of TB , is an intracellular pathogen that is well adapted to survive in host phagocytes within lung granulomas in humans and experimentally infected animals [2]–[4] . The outcome of Mtb infection is largely determined by a delicate balance between the host immune response and bacterial evasion and/or subversion of this response , resulting in successful control of the infection or manifestations of active disease of different severity [5] , [6] . In the presence of an optimal host immune response , growth of the infecting Mtb is controlled efficiently and the bacilli often cannot be detected in infected tissues by the conventional colony-forming unit ( CFU ) assay . However , during growth arrest , not all the infecting bacilli are necessarily killed . Rather , they can adapt to survive in a viable latent state , serving as a reservoir for potential reactivation TB in the host , when immunity weakens [7]–[10] . Following Mtb infection of a host with a suboptimal immune response , the bacilli , internalized into the phagosome of macrophages and dendritic cells , replicate and grow [11] , [12] . Within the phagosome , Mtb must resist the bactericidal molecules of the host cell , including abundant reactive oxygen ( ROS ) and nitrogen species ( RNS ) , hydrolytic enzymes and an acidic pH [13]–[15] . The invading bacteria counteract this hostile host environment by dampening the process of phagosome maturation , inhibiting lysosome-phagosome fusion and limiting acidification , and shifting from a reliance on high oxygen and a predominantly carbohydrate carbon source for growth . Several mycobacterial genes involved in the bacterial response to phagosomal stresses , such as hypoxia , starvation , iron depletion , acid shock and alternate carbon metabolism , have been reported to contribute to the metabolic adaptation of the bacteria to intracellular survival and growth [16] . Activation of the immune response following Mtb infection affects the expression of many host genes that are involved in the production of cytokines , chemokines , surface receptors , and molecules associated with intracellular signaling [6] , [17] . These early changes affect subsequent cellular events , including extravasation of leukocytes from the circulation , migration of immune cells to affected tissues and lymphoid organs , and proliferation of effector cells of the innate and adaptive immune response , in an orchestrated response to fight the infection [18]–[20] . An association between the level of macrophage activation and Mtb intracellular survival has been reported previously [12] , [21] . In particular , RNS produced by activated macrophages inhibit the growth of Mtb and have been implicated as an environmental cue directing the physiologic shift of the bacilli towards a state of dormancy [10] , [22] , [23] . Thus , both the host cells and the bacilli alter their gene expression during Mtb infection . However , the specific nature and the interdependence of these events , and the links between the host-pathogen crosstalk and the outcome of Mtb infection , are not fully understood . Tumor necrosis factor-alpha ( TNF-α ) , produced by activated macrophages and other cells of the immune system , is required for the protective host response against Mtb infection . This inflammatory cytokine renders the macrophages more capable of controlling the growth of or killing intracellular Mtb [24]–[26] . In addition , TNF-α plays a vital role in coordinating and driving the host inflammatory response [27] . While complete inhibition of TNF-α production results in exacerbation or reactivation of TB in animals and humans , excessive TNF-α levels can lead to severe inflammation and damage to host cells and tissues [28] , [29] . Since intracellular cyclic AMP ( cAMP ) regulates TNF-α production , the total cellular TNF-α level and the resulting tissue inflammation are determined in part by the level of this molecule [30] . Consequently , agents that increase cellular cAMP levels have anti-inflammatory properties [31]–[33] . Consistent with these findings , inhibitors of PDE4 ( PDE4i ) that increase cellular cAMP levels have been shown to reduce TNF-α production and dampen inflammation [34]–[36] . Pharmacologic inhibition of TNF-α production has recently been considered as a therapeutic modality in inflammatory diseases [37]–[39] . Inhibition of TNF-α production by thalidomide treatment in vivo and in vitro has been reported earlier [40] , [41] . Additional studies have shown that thalidomide treatment , in combination with antibiotics , reduces TNF-α production in patients with pulmonary TB and can improve treatment outcome [42] . To avoid the side effects of thalidomide , analogs of the drug have been synthesized and screened for their safety and TNF-α inhibitory capacity [43] , [44] . One class of synthetic compounds acts as effective PDE4i , which reduce TNF-α production by increasing intracellular cAMP levels . These PDE4i are non-emetic , non-teratogenic , and several of them have been found to be well tolerated by humans in Phase I and II clinical studies [45] . One of these PDE4i , used in the present study , CC-3052 , is water soluble , more stable in human plasma and ∼200-fold more potent in reducing TNF-α production , compared to the parent drug [46] . In peripheral blood mononuclear cells ( PBMC ) , the effect of CC-3052 on TNF-α inhibition is dose-dependent [46] . Importantly , at pharmacologically active doses , CC-3052 affects mainly monocyte and macrophage TNF-α production , i . e . the innate immune response , and does not have a significant effect on T cell activation , suggesting a highly cell-specific mechanism of action [47]–[49] . Isoniazid ( INH ) is one of the first-line antibiotics used in TB treatment . Though INH efficiently kills actively growing Mtb , dormant and non-replicating bacilli are killed only poorly [50] , [51] . This observation has been reported in humans , mouse and guinea pig models of experimental pulmonary TB , as well as in vitro , using broth culture and Mtb infected macrophages [52] , [53] . Exposure to INH results in the expression of a number of Mtb genes involved in multiple stress and/or toxic responses [54]–[56] . In addition , a specific group of INH-responsive Mtb genes , involved in fatty acid metabolism and cell wall biosynthesis , have been reported to be differentially expressed between actively growing versus dormant bacilli , during acute and chronic mouse infection , respectively [57] . These findings highlight the close links between the regulation of Mtb gene expression , bacterial metabolism and the response of the bacilli to INH treatment . As mentioned above , activation of macrophages by Mtb infection creates a more hostile intracellular environment for the bacilli to survive and grow [18]–[20] . It has been postulated that robust macrophage activation drives a sub-population of infecting organisms into a metabolic state that is recalcitrant to antibiotic killing [10] , [22] , [23] . Consequently , we hypothesize that dampening macrophage activation will alleviate the stress on the intracellular bacilli , thereby creating a more permissive environment for bacterial metabolism , resulting in improved efficacy of antibiotic killing . To test this hypothesis , using a rabbit model of pulmonary Mtb infection , we examined the impact of CC-3052 treatment on the interactions between the host and pathogen , as revealed by the changes in the expression of host and pathogen genes . We used the rabbit model because , upon Mtb infection , rabbits develop progressive pulmonary cavitary TB that , in contrast to the mouse , is similar to the pathologic process seen in humans . We determined which immune response genes in the infected rabbit were specifically affected by treatment with CC-3052 . In addition , we evaluated corresponding changes in the expression profile of bacterial genes , with particular emphasis on the stress and INH response genes . Finally , we examined the impact of CC-3052 treatment on bacillary load in the lungs of Mtb infected rabbits with and without INH co-treatment . The results of our study provide data to support the idea that combining anti-TB drugs with an adjunctive immune modulator may enhance the efficacy of current TB therapy regimens and shorten the duration of treatment if applied appropriately to humans .
Owing to the current paucity in immunologic reagents to analyze the cellular function and soluble mediators of immunity in rabbits , we used microarray technology to investigate the rabbit immune response to pulmonary Mtb infection . Ingenuity Pathway Analysis ( IPA ) was used for functional classification and pathway construction of differentially expressed rabbit genes during Mtb infection and CC-3052 treatment . Since an annotated rabbit genome database for functional analysis is currently unavailable , we utilized the homologues of rabbit genes from the annotated human , mouse and rat genomes available in the IPA knowledge base for our analysis . Rabbits were infected with Mtb by the respiratory route , and the development of the host immune response in the lungs was evaluated at 4 , 8 and 12 weeks post-infection using microarrays ( Table 1 ) . Of the 43 , 603 oligonucleotide probes represented on the rabbit array and analyzed by global gene expression patterns , 5 , 332 were significantly differentially expressed ( 3 , 075 up; 2 , 257 down ) in response to 4 weeks of Mtb infection of the lungs , compared to uninfected , naïve animals ( significant change defined as ±2 fold with p≤0 . 05 ) . Progressive up-regulation of host genes was noted as the infection progressed for 8 weeks and 12 weeks , with 8 , 597 ( 5 , 344 up; 3 , 253 down ) and 13 , 783 ( 4 , 260 up; 9 , 523 down ) genes differentially regulated , respectively . It is important to note that with time , as the granulomas mature and differentiate following Mtb infection in rabbit lungs , the profile of gene expression also changes . In addition , of the total number of differentially regulated genes only a subset overlap ( 1 , 269 , 5 , 212 and 1 , 584 genes respectively for 4 vs . 8 , 8 vs . 12 and 4 vs . 12 weeks , data not shown ) , consistent with the histologic observation of heterogeneity of the differentiating granulomas [58] . CC-3052 treatment was started at 4 weeks post-infection , and lung tissue was collected after 4 and 8 weeks of treatment ( i . e . at 8 and 12 weeks post-infection ) . To study the changes in host gene expression induced by CC-3052 treatment , we compared the gene expression profiles in the Mtb infected lungs of CC-3052 treated rabbits to untreated infected control animals at the same time points . The complete list of rabbit genes differentially expressed by CC-3052 treatment at the time points tested has been submitted to the Gene Expression Omnibus ( GEO ) database ( GSE27992 ) . Among the rabbit genes differentially expressed by Mtb infection , statistically significant changes in the expression of 1 , 055 ( 34 genes up; 1 , 021 down ) and 1 , 272 ( 117 genes up; 1 , 155 genes down ) genes were observed after 4 and 8 weeks of CC-3052 treatment , respectively ( Table 1 ) . Expression of 20 overlapping genes was affected at both 4 ( 1 up; 19 down ) and 8 weeks ( 9 up; 11 down ) of CC-3052 treatment . The distribution , among various cellular immune response pathways , of the genes differentially expressed in response to 4 and 8 weeks of CC-3052 treatment , is shown in Table 2 . About 30% of the genes differentially expressed following 4 weeks of CC-3052 treatment , compared to no treatment , were involved in host immune response-related pathways . These included immune cell growth and proliferation , cell morphology and movement , cell death/apoptosis , immune cell trafficking and hematologic development and function ( Table 2 ) . At 8 weeks of CC-3052 treatment , somewhat higher numbers of genes , involved in immune cell growth/proliferation , cell death , molecular transport , hematologic development/function , cell trafficking , were modulated compared to 4 weeks of treatment . In contrast , the number of rabbit genes involved in tissue morphology/development and immune cell signaling/interaction were reduced at 8 weeks compared to 4 weeks of CC-3052 treatment . In summary , the microarray analysis reveals differential expression of many host genes during Mtb infection in the rabbit lung as well as the modulation of a subset of those genes , involved in key host immune response pathways , following CC-3052 treatment . Since inhibition of PDE4 reduces TNF-α expression , a key regulator of innate and acquired immunity , we studied the effect of CC-3052 treatment on the genes of the TNF-α network . The expression patterns of a subset of genes that directly regulate or are regulated by TNF-α , in Mtb infected rabbit lungs , at 0 , 4 and 8 weeks of CC-3052 treatment were analyzed by microarray and compared to levels of expression in lungs of infected untreated animals ( Figure 1 ) . The fold changes in gene expression between CC-3052 treated and untreated animals were used for the pathway construction using IPA . In order to determine absolute expression levels , we set no cut-off values for the expression levels of individual genes in the network . In the untreated rabbits , Mtb infection up-regulated more than half of the genes ( 18 out of 32 ) in the TNF-α network , including TNF-α itself ( Figure 1A ) . Only 14 out of 32 genes were down-regulated at 4 weeks post-infection compared to uninfected , control rabbits . After 4 and 8 weeks of CC-3052 treatment , the level of expression of the majority of these genes , specifically those involved in tissue inflammation and the innate immune response , including TNF-α , were progressively reduced ( about 54% of genes at 4 weeks and 75% at 8 weeks of CC-3052 treatment ) compared to untreated animals ( Figure 1B ) . Not all of the genes in the TNF-α network were affected by CC-3052 treatment . Eight genes that are indirectly related to the TNF-α network ( CD1c , CD1e , FMO-1 , EXT-1 , MGST-3 , and SERPIN-D1 ) were expressed at comparable levels in untreated and CC-3052 treated , infected rabbits at 4 and 8 weeks post-treatment . Overall , the expression pattern of key genes of the TNF-α network impacted by CC-3052 treatment suggests a gradual reduction in the activation of the TNF-α network in the lungs of Mtb infected rabbits . To evaluate the impact of CC-3052 treatment on the expression of host genes associated with PDE4 , the target of CC-3052 , we analyzed mRNA levels from the lungs of Mtb infected rabbits by quantitative real-time PCR ( qRT-PCR ) ( Figure 2 ) . At 4 weeks post-infection , low basal levels of expression of TNF-α and PDE4A were noted in comparison with uninfected animals . TNF-α mRNA levels increased to about 30- and 45-fold , respectively , in response to 8 and 12 weeks of infection , but were not significantly different between these two time points . The expression of PDE4A was similarly elevated at 8 weeks and was somewhat lower at 12 weeks post-infection . Treatment of infected rabbits with CC-3052 ( for 4 or 8 weeks ) significantly reduced the expression levels of both TNF-α and PDE4A ( Figure 2 ) . On the other hand , while the lungs of Mtb infected animals showed lower PKA expression at all time points in comparison with uninfected controls , CC-3052 treatment reversed this effect significantly , leading to about 3-fold higher levels of PKA mRNA at 4 and 8 weeks of treatment ( i . e . 8 and 12 weeks post-infection ) ( Figure 2 ) . We then evaluated by qRT-PCR the mRNA levels of a subset of genes , including cytokines ( IFN-γ , IL10 , IL13 , IL6 and TNF-β ) , growth factors , chemokines and their receptors ( CXCR3 , GMCSF , MIF and CCL4 ) , signaling molecules and receptors ( NFκB , APRIL1 , PI3K3 and TLR2 ) and intracellular trafficking and apoptosis-related genes ( RAB7 , LAMP2 , BCL-X , FAS and RAB11 ) ( Figure 3A and B ) . Many of these gene products are known markers of macrophage activation and maturation , endosomal trafficking and/or Th1/Th2 immunity [59]–[60] . In untreated rabbits , most of these genes were progressively induced by Mtb infection from 4 to 8 weeks , and many had reached a plateau ( IFN-γ , IL13 , IL6 , MIF , CCL4 , NFκB , APRIL , TLR2 , RAB7 and LAMP2 ) or were even reduced ( IL10 , CXCR3 , GMCSF and PI3K3 ) by 12 weeks post-infection . In contrast , expression of the anti-apoptotic gene BCL-X was progressively down-regulated by infection , and there were no significant changes in the expression of other selected genes , including TNF-β and RAB11 ( Figure 3A and B ) . Following 4 or 8 weeks of CC-3052 treatment , the expression of many of these same genes were significantly reduced in comparison to Mtb infected untreated animals . The expression of genes that were not regulated by Mtb infection , for example RAB11 and TNF-β , were also unaffected by CC-3052 treatment ( Figure 3 ) . In summary , the expression profile of the host immune response genes in the lungs of Mtb infected rabbits suggests that CC-3052 treatment of rabbits with pulmonary TB resulted in a dampening of macrophage activation in the lungs . We analyzed the impact of CC-3052 treatment on gene expression of the infecting bacilli . Total RNA from Mtb isolated from infected rabbit lungs was prepared , and the expression levels of selected hypoxic/oxidative stress response genes ( narX , narK2 , devR , sodA and sodC ) were evaluated by qRT-PCR . Expression of these genes was induced progressively in the rabbit lungs from 4 to 12 weeks post-infection , with the exception of narK2 which reached a plateau after 8 weeks ( Figure 4 ) , consistent with our observation of progressive activation of macrophages in the granulomas . This notion is also supported by previous demonstrations that rabbit granulomas are hypoxic [61] and exposure of Mtb to hypoxia differentially regulates several genes of the dosR regulon [62] , [63] . Treatment of Mtb infected rabbits with CC-3052 significantly reduced the expression of narX , narK2 , devR , sodA and sodC from 5- to 150- folds ( p<0 . 05 ) , compared to infected but untreated animals . These results are consistent with our hypothesis that CC-3052-mediated reduction in macrophage activation would be accompanied by changes in the extent of the Mtb stress response during intracellular survival ( Figure 4 ) . However , treatment of log-phase Mtb culture with equimolar amount ( compared to in vivo ) of CC-3052 had no impact on the mRNA levels of these and other dormancy/stress response-related Mtb genes , compared to untreated controls ( Supporting Information Figure S1 ) , indicating that the effect of CC-3052 on Mtb gene expression was specific to the in vivo conditions . Recently , a comprehensive analysis of the gene expression profiles of Mtb isolates , representing diverse lineages , grown in vitro within activated and resting murine macrophages was reported [64] . We selected for study a subset of genes from the Homolka et . al . report , including genes previously reported as differentially expressed by Mtb in response to environmental stresses , such as metabolic adaptation to dormancy [65] , [66] . The genes were functionally grouped as follows: a ) protein synthesis ( rpsT , rpsR , rpsP , rpsL , rpsG , rplL and rplU ) ; b ) iron metabolism ( mbtI , mbtF , mbtH , mbtE , mbtD , mbtC , mbtG , mbtA , mbtB and bfrA ) ; c ) cell wall/lipid metabolism ( ppsD , drrA , papA5 , dfrA , fbpB , pckA , lipF , mmpL8 , pcaA , tgs1 , icl and drrC ) ; d ) general stress response ( hspX , sigF , sigH , dnaE2 , relA , mprA , groEL1 , groEL2 , groES and dnaJ ) ; e ) ESX-3/secretion system ( Rv0284-Rv0286 , Rv0289 , Rv0290 and Rv0292 ) ; and f ) histone-like proteins ( hns and lsr2 ) ( Figure 5 ) . Comparative gene expression profiling revealed that many of the Mtb genes involved in protein synthesis , ESX-3/secretion , iron metabolism and histone-like proteins were up-regulated by 4 weeks of CC-3052 treatment in rabbit lungs ( Figure 5 ) . In contrast , many of the general stress response and cell wall/lipid metabolism associated genes were down-regulated by CC-3052 treatment . When the Mtb gene expression pattern from the lungs of CC-3052 treated rabbits was compared to the functional classifications of Homolka et . al . , the bacteria from the rabbits showed a gene expression profile closer to that seen in resting , rather than activated , macrophages ( Figure 5 ) [13] , [64] , [67] . Thus , our functional analysis of the Mtb gene expression profile in the lungs of infected rabbits during CC-3052 treatment suggested that the bacilli were in a different metabolic state than in the lungs of untreated , control rabbits , and that this metabolic state corresponded with that observed in suboptimally activated macrophages . To study the effect of CC-3052 treatment on the ability of INH to kill Mtb in the lungs of infected rabbits , animals were treated from 4 weeks post-infection with high dose INH ( 50 mg/kg body weight/day ) or CC-3052 ( 25 mg/kg body weight/day ) , neither or both for 4 or 8 weeks ( i . e . , 8 or 12 weeks post-infection ) . In the untreated animals , implantation of about 3 . 2 log10 bacilli ( on day 0 ) into the lungs resulted in progressive , active disease with exponential bacterial growth up to 4 weeks of infection , after which the bacterial counts stabilized and then declined slightly . The control of bacillary load in the lungs of rabbits was not significantly affected by CC-3052 treatment , as indicated by the similar numbers of CFU in untreated and CC-3052 treated animals at all time points tested ( Figure 6 ) . There were also no significant differences in numbers of CFU in the liver and spleen between these two groups ( data not shown ) . INH treatment of infected rabbits did not significantly reduce the CFU numbers in the lungs of treated rabbits for the first 4 weeks . However , 8 weeks of INH treatment ( i . e . 12 weeks post-infection ) , the bacillary load in the lungs was reduced by about 1 log10 ( Figure 6 ) . The limited efficacy of INH monotherapy in Mtb infected rabbits during the first 4 weeks prompted us to evaluate the bioavailability of the drug . To assess the extent to which INH was absorbed into the circulation , we measured the plasma INH concentration in animals that received either 25 mg/kg or 50 mg/kg of INH per day by gavage administration . Both doses of the drug resulted in similar plasma levels up to 8 hours post-administration ( Figure 7A ) . However , only the 50 mg/kg dose of INH resulted in detectable INH levels in the plasma up to 24 hours . To compare INH availability when administered in drinking water to that of the gavage route of delivery , rabbits were treated with 50 mg/kg of INH by either route , and the pharmacokinetics of INH in the plasma were compared ( Figure 7B ) . The parameters for INH in drinking water were calculated , assuming a steady state dosing during a 24-hour interval . Compared to INH in drinking water , a single dose of 50 mg/kg of INH administered by gavage showed about 100-fold increase in plasma drug levels within 2 hours of administration . However , both routes of administration showed similar INH levels in the plasma at 24 hours post-administration and similar bioavailability , as measured by the area under the curve ( Figure 7B ) . This observation supported the use of gavage administration for appropriate dosing of INH , thereby avoiding the uncertainty of drug dosing through drinking water which arises from the fact that rabbits do not consume reproducible volumes of water daily . When rabbits were treated with CC-3052 in combination with INH for 4 weeks , a statistically significant reduction in the numbers of CFU was noted , compared with INH alone ( p = 0 . 027 , Figure 6 ) . An additional 4 weeks of co-treatment resulted in a 10-fold reduction in bacillary load compared to INH alone ( p = 0 . 003 , Figure 6 ) . Thus , while the rate of Mtb killing was similar during the second month of treatment , the absolute numbers of CFU in CC-3052 co-treated animals were significantly lower than in the INH alone group by the end of experiment . It is important to note that treatment of infected rabbits with INH alone or in combination with CC-3052 for up to 2 months did not produce INH resistant bacilli , as determined from the CFU assay by plating the lung homogenates in the presence and absence of INH . The effect of CC-3052 , either alone or in combination with INH , on Mtb was specific to the in vivo conditions present in the lungs of infected rabbits , since the addition of up to 50X molar excess of CC-3052 and 0 . 2 µg/ml of INH ( the minimal inhibitory concentration ) to a growing Mtb culture in vitro did not affect the growth or INH killing significantly ( Figure S2 ) . Future experiments , including prolonging the duration of treatment of infected rabbits ( more than 8 weeks ) and varying the starting time of treatment , will determine if combined administration of INH and CC-3052 can fully eliminate the bacteria from rabbit lungs . We next analyzed the impact of INH treatment on the expression of INH-associated Mtb genes in the presence and absence of CC-3052 in rabbit lungs ( Figure 8 ) . Total Mtb RNA was isolated from the lungs of infected rabbits and analyzed by qRT-PCR . Similar to results reported in several studies , after 4 weeks of INH treatment , we observed significant increases in the mRNA levels of the Mtb genes , katG , ahpC , inhA , kasA , iniB and efpA ( Figure 8 ) . [55] , [57] , [68] . There was no statistically significant differences in the level of expression of fadD26 between Mtb RNA pools from INH treated and untreated animals . The gene induction profile was specific for INH treatment , since these genes were not affected by treatment of similarly infected rabbits with rifampicin ( RIF ) , another important antibiotic for TB treatment ( Figure 8 ) . The fold change in expression for these genes , after normalization to the 16S rRNA , ranged from 2- ( for inhA ) to about 20-fold ( for katG and ahpC ) ( p≤0 . 05 ) in comparison to levels observed in untreated animals . The levels of expression of these genes in the lungs of untreated and CC-3052 alone-treated animals were comparable and significantly less than what was observed in the INH only treated animals . Thus , although 4 weeks of INH treatment of infected rabbits did not reduce the bacillary load significantly , exposure to the antibiotic clearly affected the physiology of the infecting bacteria , as manifested by differential gene expression , consistent with previous reports [7] , [8] , [55] , [57] , [68] . Interestingly , co-treatment with INH plus CC-3052 reduced the expression of INH–induced genes in Mtb to levels similar to those observed in untreated rabbit lungs .
In the present study , we describe the effect of PDE4 inhibition on the immune response during Mtb infection in a rabbit model of pulmonary TB and show how changes in the expression of immune response genes affect the crosstalk between the host and pathogen ( Figure 9 ) . During the course of Mtb infection in the lungs of untreated rabbits , expression levels of many host genes involved in the innate immune response were up- regulated . Corresponding changes were observed in the expression of many Mtb genes known to be responsible for successful intracellular survival of the bacilli , including genes involved in protecting the bacteria against ROS- and RNS-induced damage and bacillary response to stress induced by nutritional deprivation and acid shock . These genes were up-regulated gradually with disease progression in the lungs of untreated , Mtb infected rabbits . Treatment of the Mtb infected rabbits with CC-3052 led to significantly reduced expression of many of the host innate immune response genes , including TNF-α and IL-6 . Concomitant with these PDE4-induced changes in host gene expression , the levels of expression of many bacterial stress response genes were reduced . Importantly , CC-3052 treatment of infected rabbits was associated with a loss in the ability of the bacilli to withstand INH killing , and bacillary clearance was improved starting from the first month of treatment compared to INH alone . These results contrast with our previously reported study in the murine TB infection model [49] , in which we showed that co-treatment with CC-3052 extended the duration of INH-mediated bacillary clearance from the lungs at later time points , when INH effectiveness began to fail , but did not significantly affect the CFU numbers during early treatment [49] . The differences in the kinetics of improved INH-mediated Mtb clearance from the lungs in the presence of immune modulation likely reflect differences in the pathogenesis of Mtb infection in these two animal models . While rabbit lungs develop fully matured granulomas by 4 weeks post-infection that progress to necrotizing granulomas and cavities , similar to humans , the mouse immune response to Mtb infection involves the accumulation of mononuclear leukocytes that do not differentiate into structured granulomas [69] . Our observations suggest that , in the rabbit , INH fails to kill the bacilli efficiently in the early well-differentiated granulomas of the lungs ( 4 to 8 weeks post-infection ) . Moreover , CC-3052 co-treatment improves INH-mediated bacillary clearance at these early time points . Thus , in the present study , we have demonstrated the effect of PDE4 inhibition on Mtb killing within a mature granuloma , in contrast to our findings in the mouse , where the impact of immune modulation was observed on residual persisting bacilli towards the end of treatment . It is possible that , in the rabbit model , CC-3052 treatment would further improve Mtb killing at later time points ( after 12 weeks ) when INH alone fails to kill persisters . Indeed , it has been reported that the proportion of physiologically quiescent bacilli and their vulnerability to INH killing determines the overall response to INH treatment [70] , [71] . Hence , the differential outcome of INH treatment with and without CC-3052 could be attributed to changes in the metabolism of a sub-population of bacilli [56] , [72] . This suggests the presence of a higher percent of bacilli that are dormant and/or non-responsive to INH at 4 weeks post-infection of the rabbit lung , compared to the mouse lung . Since the nature and extent of the host immune response to Mtb infection significantly affect the physiology of the infecting bacilli [73] , and rabbit granulomas are hypoxic and more mature compared to lesions in the mouse lung , it is expected that the number of bacilli that are non-responsive to INH treatment in the 4-week rabbit granulomas may be relatively high . Taken together , studies in the rabbit provided a useful model that is more similar to human disease than the infection seen in the mouse lung . The changes observed in host gene expression during CC-3052 treatment of Mtb infected rabbits support a link between PDE4 inhibition and specific modulation of the innate immune response . So far , at least 11 families of class 1 PDEs ( PDE1-11 ) have been identified in humans [74] . However , only 3 of the 11 ( PDEs 4 , 7 and 8 ) are involved in hydrolyzing cAMP [75] . Of these , PDE4 is the predominant isoform of phosphodiesterase found in monocyte/macrophages [76] . Thus , inhibition of PDE4 would be expected to target monocytes and macrophages , but not T cells and other cells of the immune response . The ability of the rabbits to control bacillary growth in the lungs in the presence of CC-3052 supports the conclusion that this drug does not suppress the acquired immune response . Moreover , in our previous study , the mouse TB model enabled us to directly demonstrate that CC-3052 does not inhibit T cell activation in Mtb infected animals [49] . PDE4 hydrolyses cAMP in the phagocyte , decreasing the levels of this important second messenger , which regulates essential cell signaling events that determine many cell functions [77] , [78] . Previous studies have shown that increased levels of cAMP modulate immune cell functions , including the respiratory burst , chemotaxis , phagocytosis and phagosome maturation [79] , [80] . Consistent with these findings , treatment of Mtb infected rabbits with CC-3052 significantly reduced the steady state mRNA levels of TNF-α and other pro-inflammatory , Th1 and Th2 cytokines , chemokines and their receptors , apoptosis-associated genes , surface receptors , and signaling molecules in macrophages . TNF-α plays a central role in the host innate immune response and specifically in the activation of macrophages [81] . This cytokine is primarily regulated via TLR signaling through activation of the NFκB pathway [82] , [83] . Thus , the reduced level of expression of both the TNF-α and NFκB genes in the lungs of Mtb infected rabbits treated with CC-3052 suggests a suboptimal activation of macrophages in the lesions of these animals [84] . Suboptimal activation of macrophages has been shown to affect the cellular organization in the lung granulomas of Mtb infected animals [49] . Briefly , in Mtb infected animals , there were no significant differences either in the total number of subpleural lung granulomas or in the area of lung involved , between the untreated and CC-3052 treated rabbits up to 12 weeks post-infection . However , the granulomas in the lungs of CC-3052 treated rabbits appeared less necrotic , less fibrotic and more cellular compared to the untreated rabbits , most strikingly at 12 weeks post-infection ( 8 weeks of CC-3052 treatment ) . In mice , CC-3052 treatment affected the distribution of T cells in the lung lesions without affecting their total numbers . When Mtb infected animals ( both mice and rabbits ) were treated with INH plus CC-3052 , a significant reduction in the size and number of lung granulomas was noted , compared to INH treatment alone . Whether these morphologic changes affected drug penetration and/or exposure of the bacilli to INH in the granulomas is as yet unknown and will be investigated in the future . Macrophage activation during infection is associated with phagosome maturation , as indicated by the acquisition of RAB7 and LAMP2 on the organelle surface [59] , [85] . The reduced expression of RAB7 and LAMP2 in response to CC-3052 treatment suggests that there may indeed have been impaired or suboptimal macrophage phagosome maturation in the immune modulated rabbits . In addition , macrophage activation during Mtb infection is characterized by the production of a number of pro-inflammatory cytokines , chemokines , growth factors and their receptors . Increased TNF-α production by activated macrophages leads to increased expression of apoptotic genes , such as FAS , and reduced expression of anti-apoptotic gene expression , such as BCL-X [60] . The reduced expression of FAS and increased expression of BCL-X observed in rabbits treated with CC-3052 suggests reduced cell apoptosis during treatment with the PDE4i . Overall , the pattern of change in host gene expression in the lungs of Mtb infected rabbits treated with CC-3052 , compared to untreated rabbits , was similar to that seen in resting macrophages cultured in vitro , where expression of the genes encoding for proinflammatory cytokines , chemokines , and their receptors was relatively low [86]–[88] . The modulation of host innate immunity by CC-3052 treatment appeared to be sensed by the infecting intracellular Mtb , resulting in corresponding changes in the expression of many bacterial genes . How would changes in macrophage activation be expected to affect the physiology of the infecting bacilli ? First , the observation that expression of many of the Mtb stress response genes induced during infection was significantly reduced by CC-3052 treatment supports our hypothesis that Mtb adapted to perturbations in the host environment [14] , [23] , [73] , [89]–[92] . The metabolic shift-down of selected Mtb genes during non-replicating persistence is mediated by the activation of devR , a global transcriptional regulator , that up-regulates at least 53 Mtb genes involved in the bacterial adaptation to dormancy [93] , [62] . Among these are the Mtb stress response genes , narX and narK2 , which are induced when the infecting bacilli enter into dormancy [65] , [94] , [95] . In the present study , CC-3052 treatment reduced the expression of devR , narX , and narK2 in the infected rabbit lungs , suggesting that the bacilli sensed the reduction in environmental pressure within suboptimally activated macrophages . Treatment with the PDE4i also lead to reduced expression of the transcriptional regulator relA , which is essential for Mtb survival under stress conditions , as well as for infection , persistence and dissemination in mice and guinea pigs [96] , [97] . Consistent with our findings , loss of relA expression in Mtb has been shown to be associated with reduced lung inflammation and pathology in infected guinea pigs [97] . Similarly , mycobacterial genes encoding superoxide dismutases ( sodA and sodC ) , heat shock protein ( hspX ) and stress response sigma factors ( sigF and sigH ) have been shown to be associated with adaptation of the bacilli to a stressful environment , such as hypoxia , within activated macrophages [98]–[100] . The mRNA levels of these mycobacterial genes ( sodA , sodC , sigF and sigH ) were down-regulated during CC-3052 treatment , consistent with adaptation of the bacilli to a less stressful environment . Differential regulation of the sigma factors ensures the regulation of a distinct set of Mtb genes essential for survival under various environmental conditions such as availability of nutrients , oxygen status , and the presence of antibacterial molecules . [101] . Moreover , the up-regulation of a set of Mtb genes involved in protein synthesis , cell wall/lipid metabolism , iron metabolism and ESX-3/secretion pathways in response to CC-3052 treatment was similar to the pattern observed for intracellular Mtb inside resting macrophages [64] . Interestingly , cycloproponation of Mtb cell wall mycolates by pcaA has been shown to be essential for the immune recognition of Mtb through trehalose dimycolate ( TDM ) , an inflammatory glycolipid unique for pathogenic Mtb [102] , [103] . Furthermore , lack of mycobacterial pcaA gene expression was shown to reduce the level of TNF-α and other proinflammatory cytokines in Mtb infected murine bone marrow macrophages [102] . Taken together , the results of our study support the existence of mutually regulated host-pathogen interactions during Mtb infection . Both the infected host cells and the infecting bacilli appear to fine-tune their gene expression during adaptation to a state of intracellular infection . When the natural infection-induced macrophage activation was dampened in CC-3052 treated animals , as evidenced by the reduced expression of innate immune response genes , the bacilli sensed the change and reset their metabolism by altering their own gene expression . Remarkably , changes in the gene expression profile of the bacilli , albeit limited in magnitude , appeared to affect the ability of INH to kill the bacteria . It is conceivable that , when the transcriptional regulation of Mtb genes , such as those governed by devR , relA and alternate sigma factors , associated with the dormant/non-replicative state is altered , the bacilli are forced out of the non-replicating , persistent drug tolerant state [104] , [105] . Thus , a change in the microenvironment wherein the infecting Mtb reside would significantly influence the bacterial vulnerability to antibiotic killing . Consistent with this hypothesis , we observed improved bacterial killing in the lungs of Mtb infected rabbits treated with INH plus CC-3052 compared to INH alone . Increased expression of the INH-induced Mtb genes following 4 weeks of INH treatment was an indication of bacterial exposure to INH , despite the minimal bactericidal activity observed at this time point . In contrast , 4 weeks of INH treatment in the presence of CC-3052 led to reduced levels of expression of these genes . Though the exact mechanism of down-regulation of INH-responsive Mtb gene expression during CC-3052 co-treatment is currently unknown , this observation may be linked to the improved killing of Mtb by INH plus CC-3052 . The results of a number of previously published studies may offer alternative non-exclusive explanations for our findings . INH is a pro-drug that is activated by the bacterial catalase-peroxidase ( KatG ) to form functional adducts with NAD , which inhibit InhA , an important enzyme involved in the cell wall synthesis of Mtb [106]–[107] . In addition to INH activation , KatG also possesses peroxynitritase and NADH oxidase activities that are vital for bacterial defense against oxidative stress [108]–[109] . Thus , as we have observed in the CC-3052 treated rabbits , under reduced oxidative stress conditions , katG expression would be expected not to be up-regulated . Furthermore , induction of mycobacterial efflux pumps and their regulators , in response to intracellular stress conditions during macrophage infection of Mtb , has been reported [110]–[112] . Recently , increased expression of an efflux pump was reported to be associated with tolerance of M . marinum to INH in a zebrafish model of mycobacterial infection [113] . Moreover , up-regulation of efpA , encoding for an efflux pump in Mtb , has been reported in MDR-TB [114] . In the present study , the reduced expression of efpA in CC-3052 treated rabbits may contribute to the improved INH-mediated killing observed . Recently , by screening a chemical library of 300 compounds for their inhibitory activity against enoyl reductase ( InhA ) , Vilcheze et . al identified two potential targets , CD39 and CD117 , that exhibited significantly increased Mtb killing when combined with INH , compared to treatment by either of the compounds separately or INH alone [115] . This finding suggests that chemicals that reduce/inhibit InhA could facilitate improved Mtb killing when co-administered with INH . A similar mechanism may be operative in our studies where inhibition of expression of inhA by CC-3052 treatment of the rabbits was associated with an increase in INH-killing of Mtb . Studies from our group and others have shown that generalized immune suppression by anti-TNF-α antibody treatment in the mouse model of pulmonary TB compromised the ability of the animals to control bacillary growth , and exacerbated the pathology in the lungs [84] , [116] , [117] . Unlike anti-TNF-α antibody treatment and similar to our data in the mouse , CC-3052-mediated PDE4 inhibition had no effect on bacillary growth in the lungs of infected rabbits [49] . The use of PDE4 inhibition as a means to reduce , but not abolish , TNF-α levels has the selective advantage of targeting the level of production of the cytokine in monocytes and macrophages . Thus , T cells and other cells of the immune response would retain their ability to produce TNF-α , facilitating control of Mtb infection in the lungs . In contrast , complete neutralization of TNF-α production by treatment with an antibody against the cytokine would be expected to render the animals immune suppressed , thereby leading to uncontrolled bacillary growth and death of the animals [118]–[121] , [94] . In addition , in vitro culture of Mtb in the presence of CC-3052 had no effect on the growth of the bacilli , confirming that the PDE4i had no direct effect on bacillary growth or killing , but rather affected the bacilli in vivo via modifying the host cell ( Figure S2 ) . Thus , dampening innate immunity without interfering with the development of the acquired immune response was associated with changes in the metabolic activity of Mtb but not with loss of control of bacillary growth . Previous reports have demonstrated the usefulness of TNF-α inhibition by CC-3052 treatment , both in acute and chronic HIV-1 infection in vitro [122] , [123] . In addition , the safe administration of several PDE4i has been demonstrated in clinical trials for the treatment of other inflammatory diseases , including asthma and chronic obstructive pulmonary disease [124]–[126] . An adjunctive therapeutic approach , such as the one described in this study , could be used to treat humans with TB , providing a means for shortening treatment and improving clinical outcome in patients with active disease . The absence of significant immune suppression or other toxicities supports the idea that such an approach may safely contribute to improved treatment of TB . Taken together , our observations support the hypothesis that changes in the physiology of the bacteria , in response to changes in the host immune response can alter the susceptibility of the bacteria to antimicrobial agents . This is a novel strategy to combat Mtb infection , facilitating the use of the existing drugs more efficiently . As different anti-TB drugs target both overlapping and diverse bacterial functions , it is important to evaluate immune modulation during other single as well as multidrug treatments . Currently we are testing this approach by combining CC-3052 with other anti-TB drugs in our rabbit model of pulmonary TB .
All rabbit experiments were performed according to the procedures and policies of the Animal Welfare Act guidelines for housing and care of laboratory animals and conducted in accordance with Public Health Service Policy Institutional regulations . Animal ethics approval for rabbit aerosol infection with Mtb , treatment , post-treatment care , euthanasia and necropsy was obtained from the Institutional Animal Care and Use Committee ( IACUC ) and Institutional Biosafety Committee ( IBC ) of the Public Health Research Institute ( PHRI ) at University of Medicine and Dentistry of New Jersey ( UMDNJ ) ( IACUC approval numbers 070 , 124 and 125 ) . All procedures with the infected animals/tissues were performed in a Biosafety level three ( BSL-3 ) containment facilities according to the approved protocols . Mycobacterium tuberculosis HN878 ( Mtb ) ( a gift from Dr . Musser , TX , USA ) , a member of Beijing strains , was grown to mid-log phase ( OD600 = 0 . 6–0 . 7 ) in Middlebrook 7H9 liquid media ( Difco , MI , USA ) supplemented with 0 . 5% glycerol , 10% OADC enrichment ( oleic acid , albumin , dextrose and catalase; BD Biosciences , MD , USA ) and 0 . 25% Tween 80 at 37°C and 5% CO2 as static culture . The culture was gently mixed once a day for proper aeration . Serial dilutions of the culture were enumerated for the number of colony forming units ( CFU ) by plating on 7H10 agar plates ( Difco , MI , USA ) followed by incubation of the plates at 37°C and 5% CO2 for 4–5 weeks . All the stock cultures were stored as aliquots at −80°C until use . A vial of frozen stock culture was thawed at 37°C , sonicated three times in 5-seconds pulses on ice to disrupt bacterial clumps , diluted to 5×106 CFU/ml in sterile saline with 0 . 05% Tween 80 and used for rabbit aerosol infection [127] . The PDE4 inhibitor used in this study , CC-3052 , was a kind gift from Celgene Corporation ( Celgene Corporation , NJ , USA ) . All other chemicals were purchased from Sigma ( St . Louis , MI , USA ) unless otherwise mentioned . To study the effect of CC-3052 and/or INH on Mtb growth in vitro , bacterial pellets from mid-log culture were washed with sterile PBS and resuspended in complete 7H9 media . About 1×106 CFU/ml of the bacteria in complete 7H9 media was seeded in to 24 well plates and various concentrations of INH ( 0 . 0125 µg or 0 . 2 µg per ml ) , CC-3052 ( 4 or 40 or 200 µM ) or a combination of both ( CC-3052+INH ) were added every day up to 4 days . The number of bacterial CFU was enumerated for the initial inoculum and after every 24 hours of culture by plating serially diluted bacterial suspensions ( treated and untreated ) on 7H10 agar plates followed by incubation of the plates at 37°C and 5% CO2 for 4–5 weeks . A total of 84 male and female , specific pathogen-free , New Zealand White rabbits of approximately 2 . 5 kg in weight , purchased from Millbrook Farms ( Millbrook Farms , MA , USA ) were used for all the experiments reported in this study . The animals were acclimatized for a week after arrival at the PHRI Research Animal Facility before exposure to aerosol challenge . Groups of 6 rabbits ( per round of infection ) were infected with Mtb , using a nose-only aerosol exposure system ( CH Technologies , Inc . , NJ , USA ) as described earlier [127] . Twelve to twenty-four rabbits were infected for each experiment . After 3 hours post-exposure , one group of infected animals were sedated with a combination of Ketamine and Xylazine and euthanized by Euthasol and necropsy performed to determine the bacillary load ( T = 0 ) implanted in the lungs as described earlier [128] . For all the experiments , the bacterial inoculum , exposure time and conditions were standardized so as to implant approximately 3 . 2 log10 CFU in the lungs at T = 0 . Infected rabbits were housed individually in a BSL-3 facility , with an unrestricted food and water supply . At defined time points ( T = 4 , 8 and 12 weeks post-infection ) , groups of 2–4 rabbits , per treatment and time point , were euthanized and the lung , liver and spleen aseptically removed and portions of each organs used for CFU enumeration and isolation of host and bacterial RNA . The tissue segments for RNA isolation were frozen immediately in liquid nitrogen and stored at −80°C . The Mtb infected rabbits were classified into the following treatment groups: 1 . CC-3052 Treatment . The immune modulatory drug used in this study , CC-3052 , is an analog of thalidomide . Solutions of CC-3052 were prepared , freshly every day , in sterile water and administered through gavage at a dose of 25 mg/kg body weight using a flexible rubber feeding tube 5 days per week . Treatment with CC-3052 started at 4 weeks post-infection and continued until 12 weeks after infection ( 8 weeks of treatment ) . 2 . Isoniazid ( INH ) Chemotherapy . Freshly prepared INH , at 25 or 50 mg/kg body weight was administered to rabbits through oral administration using a flexible rubber feeding tube 5 days a week . The INH treatment was started at 4 weeks post-infection and continued until 12 weeks after Mtb infection ( 8 weeks of treatment ) . 3 . INH plus CC-3052 Combination Therapy . Rabbits were treated with a combination of INH ( 50 mg/kg body weight ) plus CC-3052 ( 25 mg/kg body weight ) using a flexible rubber feeding tube 5 days per week . Treatment was initiated concurrently at 4 weeks post-infection and continued until 12 weeks after Mtb infection ( 8 weeks of treatment ) . All three treatments were carried out in parallel . One group each of uninfected and infected but untreated rabbits served as controls for each time point of the experiment . Bacterial loads in the lungs , liver and spleens of the Mtb infected rabbits were enumerated by CFU assay in respective organs at each time point of the experiment . Briefly , random portions of lungs ( about 1/3 of the entire lung ) , liver ( about 1/10 ) and spleen ( about half ) tissue homogenates in sterile saline were serially diluted and plated onto Middlebrook 7H11 agar plates ( Difco , MI , USA ) with or without INH supplement ( 0 . 2 µg/ml ) . The plates were incubated at 37°C for 4 to 5 weeks . Colonies were counted , and results were calculated for total numbers of CFU in the whole organ . Total RNA of the host and Mtb were isolated from the lung tissues of all four groups of rabbit ( infected , infected and CC-3052- , INH- or INH+CC-3052- treated and uninfected ) at each time point of necropsy ( T = 0 , 4 , 8 and 12 weeks post-infection ) . Tissue sections from each rabbit were processed separately for host and bacterial RNA extraction . To prepare host RNA , lung tissue slices were homogenized in 10 volumes ( wt/vol ) of TRIzol ( Invitrogen , CA , USA ) using a PolyTron homogenizer ( Kinematica , Lucerne , Switzerland ) , and extracted with 0 . 3 vol ( vol/vol ) of chloroform . The mixture was centrifuged at 13 , 000 rpm for 20 minutes at 4°C and the cleared supernatant was added to an equal volume of precipitation solution ( Macherey-Nagel , GmbH ) and eluted through the NucleoSpin kit as per the manufacturer's protocol ( Macherey-Nagel , GmbH ) . For bacterial total RNA isolation from infected rabbit lungs , a modified differential lysis method was used [129] . Briefly , the tissue sections were homogenized in 10 volumes of sterile 0 . 01% SDS solution , followed by centrifugation at 13 , 000 rpm for 20 minutes at 15°C to pellet the bacteria . The bacterial pellet was resuspended in 10 volume of TRIzol ( wt/vol ) and the mixture was subjected to bead beating with Ribo-lyser ( MP Biosciences , OK , USA ) for 2 minutes in 30-seconds pulses with 1-minute ice-incubation in between the pulses . The lysate was extracted with equal volume of chloroform ( vol/vol ) followed by centrifugation and the cleared supernatant was eluted using NucleoSpin kit as per the manufacturer's protocol ( Macherey-Nagel , GmbH ) . Both host and bacterial RNA were subjected to DNase I digestion before final purification through RNeasy mini kit ( Qiagen , MD , USA ) . The quantity and quality of the total RNA was estimated by NanoDrop ( NanoDrop Products , DE , USA ) and agarose gel electrophoresis as reported elsewhere [130] . The rabbit microarray slides and the recommended reagents were purchased from Agilent Technologies ( Agilent Technologies , CA , USA ) . According to the manufacturer , each rabbit microarray slide contains quadruplicates of 43 , 604 probes ( 4 arrays per slide ) of rabbit genome , derived from public domain databases , corresponding to 43 , 604 open reading frames ( ORFs ) . The total lung RNA from individual rabbits ( 2–4 animals per time point per group ) was processed separately for each microarray experiment . Dye-swap was done for every group , to avoid dye bias during cDNA labeling , hybridization and post-hybridization procedures . The conditions for cDNA synthesis , labeling and hybridization were according to the standard operating protocol of the Center for Advanced Genomics of PHRI ( refer <http://www . cag . icph . org/downloads_page . htm> ) . For the 4 weeks time point , the gene expression values of the Mtb infected rabbits were calculated relative to uninfected rabbits ( Infected vs . Uninfected ) and for all other time points , the gene expression values of CC-3052 treated rabbits were calibrated against untreated ( but Mtb infected ) rabbits at the same time point ( CC-3052 treated vs . untreated ) . The slides were scanned and analyzed by Agilent Scanner and Feature Extraction Software ( Agilent Technologies , CA , USA ) and the acquired data was loaded into Partek Genomics Suite ( PARTEK , MO , USA ) for further analysis after appropriate statistical analysis to adjust for the errors . A 2 . 0 fold difference in expression value and a p value of less than 0 . 05 was set as cut-off to select differentially expressed genes between various comparison groups . The selected list of differentially expressed rabbit genes from PARTEK Genomics Suite was further analyzed for functional classification including derivation of networks and pathways using Ingenuity Pathway Analysis ( IPA ) software ( Ingenuity Pathway Analysis , CA , USA ) . Since the information on the functional characterization of rabbit genes is currently unavailable with any of the commercial software including IPA , we used the gene information from human , mouse and rat database , from various microarray platforms , available in IPA . We searched the IPA knowledge base for conserved homologous of rabbit genes that perform similar functions and mostly share common pathways , across the 3 different species ( mouse , rat and human ) . The filtered list of rabbit genes , containing “pathway eligible genes” was classified according to their functions in specific pathways and used to construct corresponding networks in IPA . The complete list of rabbit genes differentially expressed by CC-3052 treatment has been submitted to the GEO database ( GSE27992 ) . Total RNA of the host and Mtb , isolated from the uninfected/infected rabbit lungs were subjected to cDNA synthesis using Sprint RT Complete kit as described by the manufacturer ( Clontech , CA , USA ) . The cDNA was amplified with gene specific primers and SYBR green mix as per the manufacturer's instructions ( Clontech , CA , USA ) in a MxPro4000 real time PCR machine ( Stratagene , CA , USA ) . The SYBR green qRT-PCR mix also contains ROX as an internal reference dye . The primers for specific rabbit and Mtb genes were designed using Clone Manager Suite ( Sci-Ed software , NC , USA ) . The DNA ( for Mtb ) and mRNA ( for rabbit ) sequences of specific Mtb and rabbit genes were obtained from Tuberculist ( for Mtb genes ) or GenBank ( for rabbit genes ) data base . The DNA sequences of the primers used for qRT-PCR can be found in Table S1 and S2 . The threshold cycle ( Ct ) for each amplified target gene was calculated using MxPro4000 software ( Stratagene , CA , USA ) . Uniform baseline fluorescence was set for all the genes in each experiment and across different experiments . The Mtb gene for 16S rRNA and the transcripts for rabbit GAPDH gene were used to normalize the Ct values of the target genes . Fold change was calculated using the formula 2−ΔΔCt and represented either as absolute transcript levels or as relative expression after normalization to uninfected or untreated groups . The experiments were repeated at least 3 times with RNA samples from two to four animals per group per time point per treatment group . The concentration of INH in the plasma of Mtb infected and treated rabbits were measured using Sciex API4000 ( Applied Biosystems , CA , USA ) mass spectrometer coupled to Symbiosis Pharma HPLC system ( Spark Holland B . V , The Netherlands ) . Briefly , one part of the plasma or homogenized lung tissue was extracted with nine parts ( vol/vol ) of acetonitrile containing 0 . 2% acetic acid . Ten microliter of the supernatant was injected into a Phenomenex Gemini C6-Phenyl column ( 4 . 6×150 mm ) ( Phenomenex , CA , USA ) and eluted with a gradient of mobile phase A ( 0 . 2% acetic acid in deionized water ) and B ( 0 . 2% acetic acid in methanol ) at a flow rate of 1 ml/min . Presence of the INH analytes and quantification of drug levels were performed by monitoring multiple reactions of parent/daughter transitions in electro-spray positive ionization mode . The test samples , standards and quality control samples were spiked with Warfarin , which was used as an internal standard . The calibration curve was designed to cover the expected concentration range ( 10 . 2 ng/ml to 10 . 4 µg/ml ) of samples delivered and was derived from standard solutions of INH . The data obtained was processed using Analyst software v 1 . 4 . 2 ( Applied Biosystems , CA , USA ) and regression for pharmacokinetic parameters was performed as non-compartmental analysis using WinNonLin 5 . 0 ( Pharsight , CA , USA ) . The independent Student t-test or the Mann-Whitney test for nonparametric independent data was used for analysis ( SPSS software ) . p≤0 . 05 was considered significant for all the experiments . | Tuberculosis ( TB ) caused by Mycobacterium tuberculosis ( Mtb ) is a leading infectious cause of morbidity and mortality . Although current antibiotic regimens can cure TB , treatment requires at least six months for completion . Recent studies indicate that bacteria in a less metabolically active state are less responsive to antibiotic killing and suggest that this may partly explain the long duration required for TB treatment . In this study , using a rabbit model of pulmonary TB , we show that immune modulation of Mtb infected animals with CC-3052 , a phosphodiesterase-4 ( PDE4 ) inhibitor that reduces tumor necrosis factor alpha ( TNF-α ) production by increasing intracellular cAMP levels , resulted in the down-regulation of host genes involved in the innate immune response . Bacteria from the lungs of CC-3052 treated rabbits displayed differential expression of those genes associated with stress responses . In addition , co-treatment of INH with CC-3052 abolished the INH-induced Mtb gene expression in the infected rabbits . Importantly , bacillary clearance from the lungs of rabbits co-treated with CC-3052 and INH was improved over that in animals treated with INH alone . The results of this study provide a basis for novel use of immune modulation to improve the efficacy of antibiotic therapy and to shorten the duration of TB treatment . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"bacteriology",
"medicine",
"microbial",
"metabolism",
"cellular",
"stress",
"responses",
"functional",
"genomics",
"medicinal",
"chemistry",
"immune",
"cells",
"pathology",
"immunologic",
"subspecialties",
"drugs",
"and",
"devices",
"immunology",
"microbiology",
"host-path... | 2011 | Phosphodiesterase-4 Inhibition Alters Gene Expression and Improves Isoniazid – Mediated Clearance of Mycobacterium tuberculosis in Rabbit Lungs |
The NALCN/NCA ion channel is a cation channel related to voltage-gated sodium and calcium channels . NALCN has been reported to be a sodium leak channel with a conserved role in establishing neuronal resting membrane potential , but its precise cellular role and regulation are unclear . The Caenorhabditis elegans orthologs of NALCN , NCA-1 and NCA-2 , act in premotor interneurons to regulate motor circuit activity that sustains locomotion . Recently we found that NCA-1 and NCA-2 are activated by a signal transduction pathway acting downstream of the heterotrimeric G protein Gq and the small GTPase Rho . Through a forward genetic screen , here we identify the GPCR kinase GRK-2 as a new player affecting signaling through the Gq-Rho-NCA pathway . Using structure-function analysis , we find that the GPCR phosphorylation and membrane association domains of GRK-2 are required for its function . Genetic epistasis experiments suggest that GRK-2 acts on the D2-like dopamine receptor DOP-3 to inhibit Go signaling and positively modulate NCA-1 and NCA-2 activity . Through cell-specific rescuing experiments , we find that GRK-2 and DOP-3 act in premotor interneurons to modulate NCA channel function . Finally , we demonstrate that dopamine , through DOP-3 , negatively regulates NCA activity . Thus , this study identifies a pathway by which dopamine modulates the activity of the NCA channels .
Heterotrimeric G proteins modulate neuronal activity in response to experience or environmental changes . Gq is one of the four types of heterotrimeric G protein alpha subunits [1] and is a positive regulator of neuronal activity and synaptic transmission [2–4] . In the canonical Gq pathway , Gq activates phospholipase Cβ ( PLCβ ) to cleave the lipid phosphatidylinositol 4 , 5-bisphosphate ( PIP2 ) into diacylglycerol ( DAG ) and inositol trisphosphate ( IP3 ) , which act as second messengers . In a second major Gq signal transduction pathway , Gq directly binds and activates Rho guanine nucleotide exchange factors ( GEFs ) , activators of the small GTPase Rho [5–8] . Rho regulates many biological functions including actin cytoskeleton dynamics and neuronal development , but less is known about Rho function in mature neurons . In C . elegans , Rho has been reported to stimulate synaptic transmission through multiple pathways [9–11] . We recently identified the C . elegans orthologs of the NALCN ion channel , NCA-1 and NCA-2 , as downstream targets of a Gq-Rho signaling pathway [12] . We aim to understand the mechanism of activation of this pathway . The NALCN/NCA ion channel is a nonselective cation channel that is a member of the voltage-gated sodium and calcium channel family [13–15] . The NALCN channel was proposed to be the major contributor to the sodium leak current that helps set the resting membrane potential of neurons [16] , though there is controversy whether NALCN is indeed a sodium leak channel [17–19] . In humans , mutations in NALCN or its accessory subunit UNC80 have been associated with a number of neurological symptoms , including cognitive and developmental delay [20–33] . In other organisms , mutations in NALCN/NCA or its accessory subunits lead to defects in rhythmic behaviors [16 , 34–42] . Specifically in C . elegans , the NCA channels act in premotor interneurons where they regulate persistent motor circuit activity that sustains locomotion [43] . In addition to the Gq-Rho pathway described above , two other mechanisms have been reported to regulate the activity of the NALCN channel: a G protein-independent activation of NALCN by G protein-coupled receptors [44 , 45] and a G protein-dependent regulation by extracellular Ca2+ [46] . Here we identify a molecular cascade downstream of dopamine in the nematode C . elegans that involves the D2-like dopamine receptor DOP-3 and the G protein-coupled receptor kinase GRK-2 to modulate activity of the NCA-1 and NCA-2 ion channels . G protein-coupled receptor kinases ( GRKs ) are protein kinases that phosphorylate and desensitize G protein-coupled receptors ( GPCRs ) . Mammalian GRKs have been divided into three groups based on their sequences and function: 1 ) GRK1 and GRK7 , 2 ) GRK2 and GRK3 , and 3 ) GRK4 , GRK5 and GRK6 [47] . C . elegans has two GRKs: GRK-1 and GRK-2 , orthologs of the GRK4/5/6 and GRK2/3 families respectively [48] . Mammalian GRK2 is ubiquitously expressed [49 , 50] and GRK2 knock-out mice die as embryos [51] . In C . elegans , grk-2 is expressed in the nervous system and required for normal chemosensation [52] and egg-laying [53] . In this study , we find that C . elegans grk-2 mutants have locomotion defects due to decreased Gq signaling . We identify the D2-like dopamine receptor DOP-3 as the putative GRK-2 target and find that GRK-2 acts through DOP-3 to inhibit Go signaling . This in turn leads to activation of the NCA channels through the Gq-Rho signaling pathway . We also find that GRK-2 and DOP-3 exert their effect by acting in the premotor interneurons , where the NCA channels also act to regulate persistent motor neuron activity [43] . The D2-like receptors are GPCRs that couple to members of the inhibitory Gi/o family [54] . In mammals , GRK2 has been connected to the regulation of D2-type dopamine receptors , but the reported results are based mainly on effects of GRK2 overexpression in heterologous expression systems [55–59] . The results reported here provide a direct connection between GRK-2 and D2-type receptor signaling in a behaviorally relevant in vivo system . In C . elegans , dopamine , through dop-3 , causes the slowing of the worm’s locomotion rate on food [60]; DOP-3 signals through Go to inhibit locomotion [61] . Here we find that dopamine , through activation of DOP-3 , negatively modulates the activity of the NCA channels . This suggests a model in which dopamine signaling negatively regulates NCA channel activity and sustained locomotion through G protein signaling acting in premotor interneurons .
To identify regulators of Gq signaling , we performed a forward genetic screen in the nematode C . elegans for suppressors of the activated Gq mutant egl-30 ( tg26 ) [62 , 63] . The egl-30 ( tg26 ) mutant is hyperactive and has a tightly coiled “loopy” posture ( Fig 1A and 1B ) . These phenotypes were suppressed by the yak18 mutation isolated in our screen ( Fig 1A ) . When outcrossed away from the egl-30 ( tg26 ) mutation , yak18 mutant animals are shorter than wild-type animals , have slow locomotion ( Fig 1C , Right ) , and are egg-laying defective . We mapped yak18 to the left arm of Chromosome III and cloned it by whole-genome sequencing and a complementation test with the deletion allele grk-2 ( gk268 ) ( see Methods ) . yak18 is a G to A transition mutation in the W02B3 . 2 ( grk-2 ) ORF that leads to the missense mutation G379E in the kinase domain of GRK-2 . GRK-2 is a serine/threonine protein kinase orthologous to the human GPCR kinases GRK2 and GRK3 [52] . The deletion allele grk-2 ( gk268 ) also suppresses the loopy posture and hyperactive locomotion of activated Gq ( Fig 1A and 1B ) and causes defects in locomotion , egg-laying , and body-size similar to grk-2 ( yak18 ) ( Fig 1C Left , S1A and S1B Fig ) . We also found that grk-2 mutant animals are defective in swimming ( S2 Fig ) , a locomotion behavior that has distinct kinematics to crawling [64] . Additionally , grk-2 mutants restrict their movements to a limited region of a bacterial lawn , whereas wild-type animals explore the entire lawn ( S1C Fig ) . Our data suggest that GRK-2 regulates locomotion and is a positive regulator of Gq signaling . The standard model of GRK action is that GPCR phosphorylation by GRK triggers GPCR binding to the inhibitory protein beta-arrestin; binding of arrestin blocks GPCR signaling and mediates receptor internalization [65] . We tested whether loss of arrestin causes defects similar to loss of grk-2 by using a deletion allele of arr-1 , the only C . elegans beta-arrestin homolog . We found that arr-1 ( ok401 ) mutant animals do not have slow locomotion ( S3A Fig ) . To test whether an arr-1 mutation suppresses activated Gq , we constructed an egl-30 ( tg26 ) mutant strain carrying an arr-1 mutation in trans to a closely linked RFP marker ( that is , an egl-30 ( tg26 ) ; arr-1/RFP strain ) . Surprisingly , this strain segregated few viable non-red animals , suggesting that egl-30 ( tg26 ) ; arr-1 double mutants are subviable . The few egl-30 ( tg26 ) ; arr-1 viable animals looked similar to the egl-30 ( tg26 ) single mutant ( S3B Fig ) , but died as young adults . These results suggest that GRK-2 acts independently of arrestin to regulate locomotion rate and Gq signaling . In addition to phosphorylation of GPCRs , mammalian GRK2 can also regulate signaling in a phosphorylation-independent manner [66 , 67] . Thus , we tested whether the kinase activity of GRK-2 is required for proper locomotion and Gq signaling by assaying whether a kinase-dead GRK-2[K220R] mutant [48 , 68] is capable of rescuing the grk-2 ( gk268 ) and egl-30 ( tg26 ) ; grk-2 ( gk268 ) mutants . Wild-type GRK-2 rescued the locomotion defect of grk-2 ( gk268 ) mutants ( Fig 1C , Left ) , but the kinase-dead GRK-2[K220R] , although it was properly expressed ( Fig 2G ) , did not rescue either the locomotion defect or the suppression of activated Gq ( Fig 1C and 1D ) . We conclude that GRK-2 acts as a kinase to regulate locomotion rate and Gq signaling . To examine whether GRK-2 acts as a GPCR kinase to control locomotion , we took a structure-function approach ( Fig 2A ) . We took advantage of previously-described mutations that disrupt specific activities of GRK-2 , but do not disrupt GRK-2 protein expression or stability [48] . These mutations all affect conserved residues in well-characterized domains of GRK-2 [48] . Although GRKs act as kinases for activated GPCRs , mammalian GRKs have been shown to interact with and phosphorylate other molecules as well [66 , 67] . Therefore , although the kinase activity of GRK-2 is required for locomotion , it is possible that the relevant targets are proteins other than GPCRs . To examine whether phosphorylation of GPCRs is required for GRK-2 function in locomotion , we expressed GRK-2 with mutations ( D3K , L4K , V7A/L8A , and D10A ) that have been shown to reduce mammalian GRK2 phosphorylation of GPCRs , but that do not affect phosphorylation of other targets [69] . These N-terminal residues of mammalian GRKs form an amphipathic α-helix that contributes specifically to GPCR phosphorylation [70–74] . grk-2 ( gk268 ) mutants expressing any of these mutant GRK-2 constructs had slow locomotion like grk-2 ( gk268 ) ( Fig 2B , 2C and 2G ) , indicating that GPCR phosphorylation is required for GRK-2 function in locomotion in vivo . In mammalian GRKs , interaction of the N-terminal region with the kinase domain stabilizes a closed and more active conformation of the enzyme , important for phosphorylation of GPCRs and other substrates [70–72] . Specifically , mutation of mammalian GRK1 Arg191 disrupted phosphorylation of target substrates in addition to GPCRs , suggesting that this residue is critical for conformational changes important for GRK function as a kinase [71] . To determine whether the analogous residue in GRK-2 is required for its function in locomotion , we expressed GRK-2[R195A] in grk-2 ( gk268 ) mutants . GRK-2[R195A] did not rescue the grk-2 ( gk268 ) locomotion phenotype ( Fig 2D and 2G ) , further supporting the model that GRK-2 acts as a GPCR kinase to regulate locomotion . The RH ( Regulator of G protein Signaling Homology ) domain of mammalian GRK2 ( Fig 2A ) does not act like other RGS domains as an accelerator of the intrinsic GTPase activity of the Gq subunit , but instead interacts with Gq and participates in the uncoupling of GPCRs linked to Gq via a phosphorylation-independent mechanism [67 , 74] . To examine whether the Gq-binding residues of the RH domain are needed for GRK-2 function in locomotion , we expressed GRK-2[R106A] , Y109I , and D110A that correspond to mutations previously shown to disrupt mammalian GRK2 binding to Gq/11 [75] . All three mutant GRK-2 constructs rescued the slow locomotion defect of grk-2 ( gk268 ) ( Fig 2E and 2G ) . These results suggest that GRK-2 binding to Gq and phosphorylation-independent desensitization of GPCR signaling are not required for GRK-2 function in locomotion . The pleckstrin homology ( PH ) domain of mammalian GRK2 ( Fig 2A ) mediates interactions of GRK2 with membrane phospholipids and Gβγ subunits [67 , 76–78] . To examine whether these activities are required for GRK-2 function in locomotion , we expressed GRK-2[K567E] that disrupts phospholipid binding [79] and GRK-2[R587Q] that disrupts binding to Gβγ [79] . Neither of these GRK-2 mutants rescued the locomotion defect of the grk-2 ( gk268 ) mutant ( Fig 2F and 2G ) , suggesting that both phospholipid and Gβγ binding through the PH domain of GRK-2 are required for GRK-2 function in locomotion . GRK-2 is broadly expressed in body and head neurons [52] . To determine where GRK-2 acts to control locomotion , we expressed the grk-2 cDNA under the control of neuron-specific promoters . Expression of grk-2 under the pan-neuronal ( Prab-3 ) or acetylcholine neuron ( Punc-17 ) promoters fully rescued grk-2 ( gk268 ) mutant locomotion ( Fig 3A ) . Interestingly , expression in ventral cord acetylcholine motor neurons ( Pacr-2 ) did not rescue the locomotion phenotype , but expression driven by an unc-17 promoter derivative that is expressed mainly in the head acetylcholine neurons ( Punc-17H [80 , 81] ) rescued the locomotion phenotype ( Fig 3A ) . Additionally , expression driven in a number of interneurons and head motorneurons by the glr-1 promoter did not rescue ( Fig 3A ) . To exclude the possibility that the described role of GRK-2 in chemosensation [52] contributes to the slow locomotion phenotype of grk-2 mutants , we expressed grk-2 under ciliated sensory neuron promoters ( Pxbx-1 and Posm-6 ) . Expression of grk-2 in ciliated sensory neurons did not rescue the slow locomotion of grk-2 mutants ( Fig 3A ) . We conclude that grk-2 acts in head acetylcholine neurons to regulate locomotion . To determine if grk-2 also acts in head acetylcholine neurons to regulate Gq signaling , we expressed the grk-2 cDNA in the head acetylcholine neurons of egl-30 ( tg26 ) ; grk-2 double mutants . Expression in head acetylcholine neurons reversed the grk-2 suppression of the loopy posture and hyperactive locomotion of activated Gq−that is , the egl-30 ( tg26 ) ; grk-2 double mutants expressing grk-2 cDNA in the head acetylcholine neurons resemble the activated Gq single mutant ( Fig 3B and 3C ) . These results suggest that grk-2 acts in head acetylcholine neurons to positively regulate Gq signaling . To confirm that grk-2 is expressed in the head acetylcholine neurons , we coexpressed tagRFP fused to GRK-2 driven by the grk-2 promoter ( grk-2::tagRFP ) and GFP driven by the head acetylcholine neuron promoter ( Punc-17H::GFP ) . We observed that grk-2::tagRFP is expressed broadly in head neurons and colocalizes with GFP in several head acetylcholine neurons ( Fig 3D ) . We conclude that GRK-2 is expressed in head acetylcholine neurons to regulate locomotion and Gq signaling . Our results suggest that GRK-2 acts as a GPCR kinase to regulate locomotion . If GRK-2 were a kinase for a GPCR coupled to Gq ( EGL-30 in C . elegans ) then we would expect GRK-2 to negatively regulate Gq , which does not agree with our data . Alternatively , GRK-2 could be a kinase for a GPCR coupled to Go ( GOA-1 in C . elegans ) . The C . elegans Gq and Go pathways act in opposite ways to regulate locomotion by controlling acetylcholine release [82] . EGL-30 is a positive regulator of acetylcholine release whereas GOA-1 negatively regulates the EGL-30 pathway through activation of the RGS protein EAT-16 and the diacylglycerol kinase DGK-1 . egl-30 loss-of-function mutants are immobile whereas egl-30 gain-of-function mutants are hyperactive and have a loopy posture [83 , 84] . goa-1 and eat-16 mutants have locomotion phenotypes opposite those of egl-30; they are hyperactive and have a loopy posture [85–87] . dgk-1 loss-of-function mutants are hyperactive but do not have a loopy posture [88] . To test whether GRK-2 acts on a Go-coupled GPCR , we examined whether goa-1 mutations suppress grk-2 mutants . We found that the goa-1; grk-2 double mutant is hyperactive and has a loopy posture like the goa-1 single mutant ( S4A , S4C and S4D Fig ) , indicating that GRK-2 acts upstream of goa-1 . This result suggests that GRK-2 could be acting on GPCR ( s ) coupled to GOA-1 . To further dissect the GRK-2 pathway , we examined whether grk-2 mutations suppress the hyperactive phenotypes of eat-16 and dgk-1 mutants . The eat-16; grk-2 double mutant is hyperactive and has a loopy posture like the eat-16 single mutant ( S4A , S4C and S4D Fig ) indicating that eat-16 , like goa-1 , acts downstream of GRK-2 . By contrast , the grk-2; dgk-1 double mutant is similar to grk-2 ( S4B Fig ) . Expression of the kinase-dead GRK-2[K220R] in grk-2; dgk-1 mutants does not restore dgk-1 hyperactive locomotion ( S4E Fig ) . In addition , expression of GRK-2 under a head acetylcholine neuron promoter in grk-2; dgk-1 mutants restores dgk-1 hyperactive locomotion ( S4F Fig ) . Thus , GRK-2 regulation of the locomotion rate , Gq signaling , and DAG signaling all depend on the GRK-2 kinase activity and a function of GRK-2 in head acetylcholine neurons . In a search for potential Go-coupled GPCR targets for GRK-2 , we considered the Go-coupled D2-like dopamine receptor DOP-3 . In C . elegans , dopamine is required for the “basal slowing response” , a behavior in which wild-type animals slow down when on a bacterial lawn [89] . This behavior is mediated by the mechanosensory activation of dopamine neurons caused by physical contact of the worm body with bacteria . cat-2 mutants that are deficient in dopamine biosynthesis [90] or dop-3 mutants that lack the D2-like dopamine receptor DOP-3 , are defective in basal slowing [61 , 89] . DOP-3 has been proposed to act through Go in ventral cord acetylcholine motor neurons to decrease acetylcholine release and promote the basal slowing response [61] . If grk-2 acts in the dopamine pathway to mediate proper locomotion , possibly by phosphorylating and inactivating DOP-3 , then mutations in dop-3 and cat-2 should suppress the grk-2 locomotion phenotype . Indeed , the grk-2 mutant slow locomotion phenotype was suppressed by mutations in dop-3 and cat-2 ( Fig 4A ) . A dop-3 mutation also suppressed the swimming defect of the grk-2 mutant ( S2 Fig ) . In addition , the dop-3 and cat-2 mutations reversed the grk-2 suppression of the loopy posture and hyperactive locomotion of activated Gq−that is , the triple mutants resemble the activated Gq single mutant ( Fig 4B–4E and S5 Fig ) . These results suggest that GRK-2 acts in the dopamine pathway to regulate locomotion and Gq signaling by negatively regulating the D2-like dopamine receptor DOP-3 . Our results suggest that GRK-2 acts in head acetylcholine neurons to regulate locomotion . To test if DOP-3 acts in the same neurons as GRK-2 , we expressed the dop-3 cDNA under a pan-neuronal promoter ( Prab-3 ) , an acetylcholine neuron promoter ( Punc-17 ) , a head acetylcholine neuron promoter ( Punc-17H ) , and an acetylcholine motor neuron promoter ( Pacr-2 ) in the grk-2; dop-3 double mutant . Expression driven by the pan-neuronal , acetylcholine neuron , and head acetylcholine neuron promoters reversed the dop-3 suppression of the slow locomotion of grk-2 ( gk268 ) mutant animals—that is , grk-2; dop-3 mutants expressing dop-3 cDNA by these three promoters resemble the grk-2 mutant ( S6A Fig ) . By contrast , expression of the dop-3 cDNA by an acetylcholine ventral cord motor neuron promoter did not reverse the grk-2; dop-3 locomotion phenotype ( S6A Fig ) or the hyperactive locomotion and loopy posture of egl-30 ( tg26 ) ; grk-2; dop-3 mutant animals ( S6C–S6E Fig ) . We conclude that dop-3 , like grk-2 , acts in head acetylcholine neurons , consistent with the model that GRK-2 acts directly on DOP-3 . Moreover , dop-3 expression under the grk-2 promoter reversed the dop-3 suppression of the slow locomotion of grk-2 mutants ( S7A Fig ) , supporting the idea that GRK-2 and DOP-3 act in the same neurons . We observed that grk-2; dop-3 and cat-2; grk-2 double mutant animals still retain some of the characteristic grk-2 phenotypes: the animals have shorter bodies and are egg-laying defective . In addition , grk-2 mutants do not fully explore a bacterial lawn and this behavior remains in the grk-2; dop-3 double mutant ( S1C Fig ) . Thus , GRK-2 has additional neuronal functions that do not depend on dop-3 . The D1-like dopamine receptor DOP-1 has been shown to act antagonistically to DOP-3 to regulate the basal slowing response: dop-1 mutations suppress the dop-3 basal slowing phenotype [61] . By contrast , we found that DOP-1 is not involved in the GRK-2 and DOP-3 pathway that regulates locomotion rate because dop-1 mutations do not affect the locomotion rate of the grk-2; dop-3 double mutant ( S6B Fig ) . Thus , the role of DOP-3 in GRK-2-regulated locomotion is independent of its role in the basal slowing response . Exposure of C . elegans to exogenous dopamine causes DOP-3-dependent paralysis—dop-3 mutants are significantly resistant to the paralytic effects of exogenous dopamine [61] . If GRK-2 negatively regulates DOP-3 , then grk-2 mutants might be hypersensitive to dopamine due to increased DOP-3 activity . Indeed , we found that grk-2 mutants are hypersensitive to dopamine and this hypersensitivity depends on dop-3 ( Fig 4F ) . In an effort to dissect the molecular mechanism by which grk-2 regulates DOP-3 activity , we expressed GFP-tagged DOP-3 under the grk-2 promoter in dop-3 and grk-2; dop-3 mutant animals and examined the levels of expression of DOP-3::GFP both by Western and by fluorescence microscopy ( S7 Fig ) . Although Pgrk-2::DOP-3::GFP fully reversed the dop-3 suppression of the slow locomotion of grk-2 mutants ( S7A Fig ) , we did not observe any difference in the level of DOP-3 expression in a grk-2 mutant ( S7B–S7D Fig ) nor did we observe any obvious change in the subcellular localization of DOP-3::GFP in a grk-2 mutant ( S7C Fig ) . However , one caveat is that we do not have the resolution to distinguish between DOP-3 localization on the plasma membrane or in an intracellular compartment . In addition to genes within the canonical Gq-PLCβ pathway , our screen for suppressors of activated Gq also identified the Trio RhoGEF ( UNC-73 in C . elegans ) as a new direct Gq effector [8] . Recently , we identified the cation channels NCA-1 and NCA-2 as downstream targets of this Gq-Rho pathway . Specifically , we found that mutations in genes that encode accessory subunits of the NCA channels ( unc-79 , unc-80 ) or in the NCA channels per se ( nca-1; nca-2 ) suppress the neuronal phenotypes of activated Gq and activated Rho [12] . Moreover , mutations in the Rho-NCA pathway suppress the loopy posture of the activated Gq mutant more strongly than do mutations in the canonical PLCβ pathway [12] . Like mutations in the Rho-NCA pathway , grk-2 mutants also strongly suppress the loopy posture of an activated Gq mutant ( Figs 1A , 4D and 4E ) , suggesting that grk-2 may affect signaling through the Rho-NCA pathway . To further examine whether grk-2 affects Rho-NCA signaling , we built double mutants of grk-2 with an activated Rho mutant ( G14V ) , referred to here as Rho* , expressed in acetylcholine neurons . Rho* has a loopy posture and slow locomotion and a grk-2 mutation partially suppresses these phenotypes ( Fig 5A and 5B ) , consistent with grk-2 affecting signaling through the Rho-NCA pathway . We also built double mutants of grk-2 and a dominant activating mutation in the NCA-1 channel gene , nca-1 ( ox352 ) , referred to here as Nca* [12 , 24] . Like Rho* , Nca* mutants have a loopy posture and slow locomotion . However , grk-2 mutants do not suppress either of these phenotypes because Nca*; grk-2 double mutants behave identically to Nca* mutants ( Fig 5C and 5D ) . This suggests that grk-2 acts upstream of NCA . C . elegans has two proteins that encode pore-forming subunits of NCA channels , NCA-1 and NCA-2 . Mutations that disrupt both NCA-1 and NCA-2 channel activity cause a characteristic “fainter” phenotype in which worms suddenly arrest their locomotion and acquire a straightened posture [35] . Our genetic experiments indicate that GRK-2 affects Rho-NCA signaling , but grk-2 mutants are not fainters . Given that grk-2 partially suppresses Rho* , we hypothesized that GRK-2 is not absolutely required for Rho-NCA signaling , but provides modulatory input . To test this hypothesis , we built double mutants between grk-2 and nlf-1 , which is partially required for localization of the NCA-1 and NCA-2 channels and has a weak fainter mutant phenotype [12 , 91] . A grk-2 mutation strongly enhanced the weak fainter phenotype of an nlf-1 mutant so that the double mutants resembled the stronger fainter mutants that completely abolish NCA-1 and NCA-2 channel activity ( Fig 6A and 6B ) . Moreover , double mutants between grk-2 and the RhoGEF Trio unc-73 were also strong fainters , supporting the hypothesis that GRK-2 modulates the Rho-NCA pathway ( Fig 6C ) . By contrast , double mutants between grk-2 and the egl-8 PLCβ do not have a fainter phenotype ( Fig 6C ) . These results suggest that GRK-2 is a positive modulator of NCA-1 and NCA-2 channel activity . If GRK-2 modulates the NCA channels by acting as a negative regulator of Go then we would expect that mutations in other proteins that act as negative regulators of Go might enhance the fainter phenotype of nlf-1 mutants . Indeed , a mutation in egl-10 , encoding the RGS that negatively regulates Go [92] , strongly enhances the nlf-1 fainter phenotype ( Fig 6D ) . As controls , mutations in genes involved in dense-core vesicle biogenesis ( eipr-1 and cccp-1 ) , that cause locomotion defects comparable to grk-2 or egl-10 [63 , 80] , did not enhance the nlf-1 fainter phenotype , indicating that the interactions of grk-2 and egl-10 with nlf-1 are specific . grk-2 acts in head acetylcholine neurons to mediate locomotion . We recently used the same Punc-17H promoter construct to show that nlf-1 also acts in head acetylcholine neurons and not in ventral cord motor neurons to regulate locomotion [12] . Therefore , we predicted that expression of an activated Go mutant under a head acetylcholine neuron promoter would enhance the fainter phenotype of nlf-1 mutants . Indeed , we found that expression of the activated Go mutant GOA-1[Q205L] in head acetylcholine neurons makes the animals slow ( Fig 6E ) and significantly enhances the fainter phenotype of nlf-1 mutants ( Fig 6F ) . These results support the model that GRK-2 negatively regulates Go , and that Go negatively regulates NCA-1 and NCA-2 channel activity . Our results are consistent with the model that GRK-2 acts in locomotion by negatively regulating DOP-3 and that GRK-2 is a positive modulator of NCA-1 and NCA-2 activity . These data predict that DOP-3 would be a negative modulator of NCA-1 and NCA-2 channel activity . Consistent with this model , mutations in cat-2 and dop-3 almost fully suppress the nlf-1 fainter phenotype during forward movement ( Fig 7A and 7B ) . Additionally , dop-3 mutants partially suppress the strong grk-2; nlf-1 fainter phenotype , consistent with the model that DOP-3 is a substrate for GRK-2 ( Fig 7C ) . These results suggest that dopamine , through DOP-3 , negatively modulates NCA-1 and NCA-2 channel activity . To more directly test whether grk-2 and dop-3 modulate the NCA channel per se , we created double mutants between grk-2 and the pore-forming subunit gene nca-1 . nca-1 mutants have a low penetrance , very weak backward-fainting phenotype that is strongly enhanced in a grk-2 mutant background ( S3C and S6F Figs ) . arr-1 mutants , on the other hand , do not enhance the nca-1 phenotype , further supporting the conclusion that arrestin does not play a role in this pathway ( S3C Fig ) . As expected , dop-3 suppresses the enhanced fainting phenotype of the grk-2; nca-1 double mutant ( S6F Fig ) . Our data suggest that GRK-2 and DOP-3 play modulatory and not essential roles in the regulation of NCA-1 ( and possibly NCA-2 ) channel activity . By contrast , UNC-80 is necessary for the stability and function of NCA-1 and NCA-2 , so unc-80 mutants are strong fainters [36 , 39] . As expected for a modulatory role in regulating NCA-1 and NCA-2 activity , mutations in dop-3 and cat-2 do not suppress the strong fainter phenotype of unc-80 mutants ( S8A and S8B Fig ) . We showed above that grk-2 mutants are hypersensitive to the paralytic effects of dopamine . We also found that low concentrations of dopamine do not paralyze grk-2 mutants but instead cause them to faint , and that this effect depends on dop-3 ( Fig 7D ) . This is consistent with the model that dopamine acts through DOP-3 to negatively modulate NCA-1 and NCA-2 . In C . elegans , the NCA channels act in premotor interneurons [12 , 43 , 91] . To determine whether grk-2 acts in a cell autonomous way to regulate NCA , we identified the head acetylcholine neurons where GRK-2 is expressed . We coexpressed GRK-2 fused to tagRFP driven by the grk-2 promoter ( grk-2::RFP ) and nuclear YFP driven by the choline transporter cho-1 promoter ( Pcho-1fosmid::SL2::YFP::H2B ) , which is expressed in all acetylcholine neurons [93] . We found that grk-2 is expressed in the following head acetylcholine neurons: the AVA , AVB , AVD , and AVE premotor interneurons; SMD and RMD head motor neurons; and in the AIN , AIY , SIA , SIB , and SAA interneurons ( Fig 8A ) . To further determine where GRK-2 acts to control locomotion , we expressed the grk-2 cDNA under additional neuron-specific promoters in grk-2 mutants . We used a cho-1 promoter fragment for expression in the SMD and RMD head motor neurons [94] , the ceh-24 promoter for expression in the SIA and SIB interneurons , and a ttx-3 promoter fragment for AIY-specific expression [95] . For expression in premotor command interneurons we used a combination of the nmr-1 promoter for AVA , AVD and AVE ( also PVC and RIM ) expression together with the sra-11 promoter for AVB ( also AIY and AIA ) expression , as previously described [91] . We found that grk-2 expression in command interneurons fully rescued the slow locomotion of grk-2 mutants , but expression in the other neuron types failed to rescue ( Fig 8B–8E ) . However , expression of grk-2 in only sra-11 or only nmr-1 expressing neurons did not rescue the slow locomotion defect ( S9 Fig ) . Additionally , grk-2 expression in the command interneurons was sufficient to rescue the enhanced fainting phenotype of grk-2; nlf-1 mutants ( Fig 8F ) . Similarly , dop-3 expression in the command interneurons was sufficient to reverse the dop-3 suppression of the slow locomotion of grk-2 mutants ( Fig 8G ) . Given that the fainting phenotypes of nlf-1 mutants and nca mutants were also rescued by expression in command interneurons [43 , 91] , our results suggest that GRK-2 , DOP-3 , and the NCA channels act in the same neurons . However , we did not see rescue of the grk-2 locomotion phenotype using the glr-1 promoter ( Fig 3A ) , which in principle is also expressed in the command interneurons , and similarly we did not observe statistically significant rescue of the nlf-1 fainting phenotype using the same glr-1 promoter [12] . The different results seen between the glr-1 and the nmr-1 + sra-11 promoters may be due to different levels of expression or because of expression in some different neuron types .
In this study we identified a pathway that modulates the activity of the NCA-1 and NCA-2 channels through dopamine and Gq signaling ( Fig 9 ) . We found that dopamine acts through the D2-like receptor DOP-3 to negatively modulate NCA-1 and NCA-2 . Furthermore , we identified the GPCR kinase GRK-2 as a positive ( indirect ) regulator of Gq and negative regulator of NCA-1 and NCA-2 . Our results suggest that GRK-2 mediates its regulatory effects by inhibiting DOP-3 . In C . elegans , GRK-2 was previously found to act in sensory neurons to regulate chemosensation [52] . Here we found that GRK-2 acts in command interneurons to regulate locomotion and Gq signaling . Using a structure-function approach , we found that GPCR phosphorylation , Gβγ-binding , and membrane-binding are required for GRK-2 function in locomotion , but binding to Gq is not required . Similar results were reported for the function of GRK-2 in chemosensation [48] , suggesting that in both cases GRK-2 acts as a GPCR kinase and that membrane localization is critical for its function . Additionally , GRK-2 seems to act independently of arrestin to regulate both locomotion and chemosensation [52] . Because cat-2 and dop-3 mutants are hypersensitive to the aversive odorant octanol [96–98] and grk-2 mutants are insensitive to octanol [52] , GRK-2 might act as a GPCR kinase for DOP-3 in chemosensory neurons as well . GRK-induced phosphorylation of GPCRs induces endocytosis , which leads to their sorting to either lysosomes for degradation or to recycling endosomes . GRK-dependent recruitment of arrestins to the phosphorylated receptor is typically required for endocytosis , but GRK2 was also reported to utilize arrestin-independent mechanisms to mediate receptor internalization [66] . GRK2 associates with a large number of proteins with known roles in receptor internalization and signaling . For example , the C-terminus of GRK2 directly binds clathrin and this interaction has been proposed to be involved in arrestin-independent internalization [99] . Our data suggest an arrestin-independent role for C . elegans GRK-2 in GPCR regulation , supporting the idea that the role of GRK-2 extends beyond just the recruitment of arrestin . The D2-type dopamine receptors , like DOP-3 , are GPCRs that couple to members of the inhibitory Gi/o family . Mammalian GRK2 and GRK3 ( the orthologs of GRK-2 ) have been connected to the desensitization , internalization , and recycling of D2-type dopamine receptors [55–59 , 100] . Interestingly , some of the effects of GRK2 on D2 receptor function may be independent of receptor phosphorylation [57 , 58 , 100] , though one caveat of these studies is that they involve GRK2 overexpression in heterologous cells . Our structure-function approach indicates that GPCR phosphorylation is important for GRK-2 function in locomotion and Gq signaling in C . elegans , although we cannot exclude the possibility that phosphorylation of additional substrates may also be required . In vivo studies of the role of mammalian GRKs in the regulation of dopamine receptors have focused on the analysis of behaviors that are induced by psychostimulatory drugs such as cocaine that elevate the extracellular concentration of dopamine [59] . Mice with a cell-specific knockout of GRK2 in D2 receptor-expressing neurons have altered spontaneous locomotion and sensitivity to cocaine [101] , though the cellular mechanisms underlying these behavioral effects are not known . Our findings provide evidence of a direct association between GRK-2 and D2-type receptor signaling that regulates locomotion in an in vivo system . In C . elegans , the Gq and Go pathways act in opposite ways to regulate locomotion by controlling synaptic vesicle release [82] . Gq acts as a positive regulator of acetylcholine release while Go negatively regulates Gq signaling , through activation of the Gq RGS EAT-16 and the diacylglycerol kinase DGK-1 . DGK-1 phosphorylates the second messenger DAG and thus inhibits its action . Using genetic epistasis , we demonstrated that GRK-2 acts upstream of GOA-1/Go and EAT-16 to positively regulate locomotion and body posture . Given this result , our cell-specific rescue data , and our data indicating that GRK-2 acts as a GPCR kinase for a locomotion-related GPCR , we propose that GRK-2 acts as a kinase for the Go-coupled GPCR DOP-3 in premotor interneurons . In this model , GRK-2 driven phosphorylation of DOP-3 reduces Go signaling and thereby promotes Gq signaling ( Fig 9 ) . Inhibition of Go by GRK-2 could promote Gq-Rho signaling by two mechanisms: ( 1 ) by inhibiting the Gq RGS EAT-16 and thus activating Gq itself , and ( 2 ) by inhibiting DGK-1 which acts in parallel to Gq-Rho to regulate DAG levels ( Fig 9 ) . Interestingly , a grk-2 mutant is suppressed by mutations in goa-1 and eat-16 , but not by dgk-1 . This finding supports other literature that suggests that goa-1 and eat-16 have similar interactions with Gq signaling , but that dgk-1 is distinct [87 , 102] . GOA-1 and EAT-16 act upstream of Gq to inhibit Gq signaling . DGK-1 , on the other hand , acts downstream of Gq to reduce the pool of the Gq-generated second messenger DAG . Adding to the complexity , DAG levels may be controlled by both the PLCβ and Rho branches of the Gq pathway ( Fig 9 ) . Previously , it has been shown that mutations in dgk-1 partially suppress the strong locomotion defect of egl-30/Gq loss-of-function mutations [102] . Surprisingly , we found that a grk-2 mutation fully suppresses a dgk-1 mutant . This suggests that the effect of GRK-2 on locomotion is more complex and may be partially independent of Gq signaling and of Gq-generated DAG . This agrees with our data showing that GRK-2 has additional neuronal functions that do not depend on DOP-3 . Gq signaling regulates several genetically separable aspects of locomotion behavior including locomotion rate and waveform . The Gq-PLCβ signaling pathway has been reported to act in ventral cord motor neurons to regulate acetylcholine release and locomotion rate [103] , whereas the Gq-Rho pathway has been reported to act in at least two different classes of neurons including head acetylcholine neurons to regulate locomotion rate , waveform , and fainting behavior [12] . Our data here further suggest that DOP-3 , GRK-2 , and the Gq-Rho pathway all act together in the premotor command interneurons to regulate activity of the NCA channels . The command interneurons have been previously shown to regulate several aspects of locomotion behavior including the propensity to go forward or reverse [104 , 105] and the tendency of the worm to sustain persistent locomotion [43] . Our data here suggest that the command interneurons also regulate the locomotion rate and the posture of the animals . As we reported previously , mutations in the Rho-NCA pathway suppress both the locomotion rate and loopy posture of activated Gq mutants whereas mutations in the PLCβ pathway suppress mainly the locomotion rate [12] . Thus , Gq acts through both the PLCβ pathway and the Rho-NCA pathway to regulate locomotion rate , probably by acting in different neurons . By contrast , Gq acts primarily through the Rho-NCA pathway to regulate the posture of the worms . This agrees with our data showing that grk-2 mutations , which affect signaling through the Rho-NCA pathway , strongly suppress the loopy posture of activated Gq . The identification of GRK-2 as a putative DOP-3 kinase and positive modulator of Gq-Rho signaling connects dopamine signaling to modulation of the NCA channels ( Fig 9 ) . NCA channels have been shown in recent years to be important for neuronal excitability and a number of rhythmic behaviors [16 , 34–42] . In humans , mutations affecting the NCA channel NALCN cause neurological diseases [20–33] . However , despite the relevance of this channel to neuronal function it is unclear how it is gated and activated . Two studies have shown that NALCN-dependent currents can be activated by G protein-coupled receptors in a G protein independent way [44 , 45] whereas another study showed that NALCN can be activated by low extracellular calcium via a G protein-dependent pathway [46] , but the specific mechanisms remain unknown . Our results suggest that dopamine acts through the DOP-3 G protein-coupled receptor and downstream G protein signaling pathways to modulate activity of the NCA channels in a physiologically relevant setting . This is the first study connecting dopamine to the activation of these important channels .
Strains were maintained at room temperature or 20° on the OP50 strain of E . coli [106] . The Supplementary Information contains full genotypes of all the strains we used ( S1 Table; List of strains ) . The grk-2 ( yak18 ) mutant was isolated in an ENU screen as a suppressor of the hyperactive locomotion and loopy posture of the activated Gq mutant egl-30 ( tg26 ) [63] . We mapped the yak18 mutation to the left arm of Chromosome III ( between -27 and -21 . 8 m . u . ) using a PCR mapping strategy that takes advantage of PCR length polymorphisms due to indels in the Hawaiian strain CB4856 ( Jihong Bai , personal communication ) . Using whole-genome sequencing ( see below ) , we found that yak18 is a G to A transition mutation in the W02B3 . 2 ( grk-2 ) ORF that creates a G379E missense mutation in the kinase domain of GRK-2 . We confirmed the gene identification by performing a complementation test between grk-2 ( yak18 ) and the grk-2 ( gk268 ) deletion mutant , finding that they fail to complement for the slow locomotion phenotype . Genomic DNA from grk-2 ( yak18 ) animals was isolated and purified according to the Worm Genomic DNA prep protocol from the Hobert lab website ( http://hobertlab . org/wp-content/uploads/2013/02/Worm_Genomic_DNA_Prep . pdf ) . The sample was sequenced using Ion Torrent sequencing ( DNA Sequencing Core Facility , University of Utah ) . The sequencing data were uploaded to the Galaxy web platform and were analyzed as described [107] . The Supplemental Information contains a complete list of constructs used ( S2 Table; List of plasmids ) . All constructs made in this study were constructed using the multisite Gateway system ( Invitrogen ) . Specifically , a promoter region , a gene region ( cDNA ) , and an N- or C-terminal 3’UTR or fluorescent tag ( GFP or tagRFP ) fused to a 3’UTR were cloned into the destination vector pCFJ150 . For the cell-specific rescuing experiments , an operon GFP was included in the expression constructs downstream of the 3’UTR [108] . This resulted in expression of untagged grk-2 , dop-3 , or goa-1 , but allowed for confirmation of proper promoter expression by monitoring GFP expression . The cho-1 fosmid reporter construct otIs534 carries an SL2-spliced nuclear localized YFP::H2B immediately after the stop codon of the cho-1 gene [93] . Extrachromosomal arrays were made by standard injection and transformation methods [109] . In all cases we injected 5–10 ng/ul of the expression vector and isolated multiple independent lines . At least two lines were tested that behaved similarly . We made a construct driving expression of the grk-2 cDNA fused to tagRFP under the grk-2 promoter and generated worms with extrachromosomal arrays . For the grk-2 promoter region , we PCR amplified 2892 bp upstream of the start codon using genomic DNA as a template and the following set of primers: forward primer 5’cacgacagtttccatagtgattgg3’ and reverse primer 5’tttttgttctgcaaaatcgaattg3’ . grk-2 was expressed in neurons in the head , ventral cord , and tail , consistent with the published expression pattern [52] . Neurons were identified by the stereotypical positions of cells expressing the acetylcholine neuron reporter cho-1fosmid::SL2::YFP::H2B [93 , 110] that colocalized with grk-2::tagRFP . For most experiments , we measured locomotion rate using the body bend assay . Specifically , first-day adults were picked to a three-day-old lawn of OP50 and stimulated by poking the tail of the animal with a worm pick . Body bends were then immediately counted for one minute . A body bend was defined as the movement of the worm from maximum to minimum amplitude of the sine wave [102] . Specifically for the experiment described in Fig 5D we used a radial locomotion assay . Animals were placed in the center of 10 cm plates with thin one to two-day-old lawns of OP50 and left for one hour . The position of each worm was marked and the radial distance from the center of the plate was measured ( cm travelled/h ) . Egg-laying assays were performed as described [80] . L4 larvae were placed on plates with OP50 at 25°C overnight . The next day , five animals were moved to a fresh plate and allowed to lay eggs at 25°C for two hours . The number of eggs present on the plate was counted . First-day adult animals were placed on an OP50 plate and allowed to move forward until when they had completed five to ten tracks . Each animal's tracks were imaged at 40X magnification using a Nikon SMZ18 microscope with the DS-L3 camera control system . Period and 2X amplitude were measured using the line tool in Image J . For each worm , five period/ amplitude ratios were averaged and five worms were used per experiment . The fainting phenotype is characterized by frequent arrest of locomotion , accompanied by a straightening of the anterior part of the body . We scored fainting as a sudden halt in movement accompanied by a straightened posture . First-day adults were transferred to plates with two to three-day-old lawns of OP50 and left undisturbed for one minute . Animals were then poked either on the head ( for backward movement ) or on the tail ( for forward movement ) , and we counted the number of body bends before the animal faints . If the animal made ten body bends , the assay was stopped and we recorded ten as the number . Thus , animals that never faint ( for example , wild-type ) are scored as 10 in these experiments . Specifically for the experiment described in Fig 7D the number reported was the percentage of animals that fainted before making 10 body bends . Single , first-day adults were transferred to a 25 ul drop of M9 buffer at the center of an empty NGM plate and video recorded for 30 sec . The swimming behavior was analyzed as described [38 , 64] . First-day adults were mounted on 2% agarose pads and anesthetized in M9 buffer containing 50 mM sodium azide for ten minutes . The image of each animal was obtained using a Nikon 80i wide-field compound microscope . Body size was measured using ImageJ software . We used a method similar to the one described [61] . Specifically , first-day adults were transferred to plates containing dopamine ( 5 mM , 10 mM , 15 mM , 20 mM , 40 mM ) and incubated for 20 min at room temperature . Animals were then poked using a worm-pick and the number of body bends was counted , stopping the assay at 10 body bends . We report the percent of animals that moved 10 body bends without stopping ( Percent of animals moving ) . A body bend was defined as the movement of the worm from maximum to minimum amplitude of the sine wave . Dopamine plates were prepared fresh just before use , as described [61] . For the Western analysis shown in Fig 2G , worm lysates were prepared as follows . Ten transgenic animals from each strain were transferred to a 6 cm OP50 plate and grown until most of their progeny had reached adult stage . Animals from five such plates were washed off with M9 , collected in a 15 ml Falcon tube , and spun down at 2000 rpm for 3 min . Animals were washed twice with M9 . The pelleted worms were then resuspended in 2X SDS loading dye and lysed by incubation at 95°C for 20 min . For the Western analysis shown in S7B Fig , worm lysates were prepared as follows . Two hundred transgenic worms were individually picked and transferred in a microfuge tube in 10 ul M9 . An equal volume of 2X SDS loading dye was added to the tube and the animals were lysed by incubation at 95°C for 20 min . Samples were resolved on 10% SDS-polyacrylamide gels and blotted onto PVDF membranes . To detect the desired proteins , we added the following primary antibodies: monoclonal anti-GRK2/3 , clone C5/1 . 1 ( 1:1000 , EMD Millipore #05–465 ) , monoclonal anti-beta-tubulin antibody ( 1:1000 , ThermoFisher , BT7R , #MA5-16308 ) , rabbit polyclonal anti-GFP ( 1:1000 , Santa Cruz #sc-8334 ) , and monoclonal anti-mCherry ( 1:50 , a gift from Jihong Bai and the Fred Hutchinson Cancer Research Center antibody development shared resource center ) . The secondary antibodies were an Alexa Fluor 680-conjugated goat anti-mouse antibody ( 1:20 , 000 , Jackson Laboratory #115-625-166 ) and an Alexa Fluor 680-conjugated goat anti-rabbit antibody ( 1:20 , 000 , Jackson Laboratory #111-625-144 ) . A LI-COR processor was used to develop images . For fluorescence imaging , first-day adult animals were mounted on 2% agarose pads and anesthetized with 50 mM sodium azide for ten minutes before placing the cover slip . The images shown in Fig 3D and S7C Fig were obtained using an Olympus FLUOVIEW FV1200 confocal microscope . The images shown in Fig 8A were acquired using a Zeiss confocal microscope ( LSM880 ) with Z-stack analysis and reconstruction performed using the ZEN software tool . For pictures of worms , first-day adult animals were placed on an assay plate and photographed at 50 or 60X using a Nikon SMZ18 dissecting microscope with a DS-L3 camera control system . The images were processed using ImageJ . P values were determined using GraphPad Prism 5 . 0d ( GraphPad Software ) . Normally distributed data sets requiring multiple comparisons were analyzed by a one-way ANOVA followed by a Bonferroni or Dunnett test . Normally distributed pairwise data comparisons were analyzed by two-tailed unpaired t tests . Non-normally distributed data sets with multiple comparisons were analyzed by a Kruskal-Wallis nonparametric ANOVA followed by Dunn’s test to examine selected comparisons . Non-normally distributed pairwise data comparisons were analyzed by a Mann-Whitney test . For the experiments shown in S3C and S6F Figs a chi-square test for multiple comparisons was used . | Dopamine is a neurotransmitter that acts in the brain by binding seven transmembrane receptors that are coupled to heterotrimeric GTP-binding proteins ( G proteins ) . Neuronal G proteins often function by modulating ion channels that control membrane excitability . Here we identify a molecular cascade downstream of dopamine in the nematode C . elegans that involves activation of the dopamine receptor DOP-3 , activation of the G protein GOA-1 , and inactivation of the NCA-1 and NCA-2 ion channels . We also identify a G protein-coupled receptor kinase ( GRK-2 ) that inactivates the dopamine receptor DOP-3 , thus leading to inactivation of GOA-1 and activation of the NCA channels . Thus , this study connects dopamine signaling to activity of the NCA channels through G protein signaling pathways . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"neurochemistry",
"acetylcholine",
"chemical",
"compounds",
"neuroscience",
"organic",
"compounds",
"biological",
"locomotion",
"hormones",
"motor",
"neurons",
"amines",
"neurotransmitters",
"interneurons",
"catecholamines",
"g",
"pr... | 2017 | Dopamine negatively modulates the NCA ion channels in C. elegans |
The Pneumococcal serine-rich repeat protein ( PsrP ) is a pathogenicity island encoded adhesin that has been positively correlated with the ability of Streptococcus pneumoniae to cause invasive disease . Previous studies have shown that PsrP mediates bacterial attachment to Keratin 10 ( K10 ) on the surface of lung cells through amino acids 273–341 located in the Basic Region ( BR ) domain . In this study we determined that the BR domain of PsrP also mediates an intra-species interaction that promotes the formation of large bacterial aggregates in the nasopharynx and lungs of infected mice as well as in continuous flow-through models of mature biofilms . Using numerous methods , including complementation of mutants with BR domain deficient constructs , fluorescent microscopy with Cy3-labeled recombinant ( r ) BR , Far Western blotting of bacterial lysates , co-immunoprecipitation with rBR , and growth of biofilms in the presence of antibodies and competitive peptides , we determined that the BR domain , in particular amino acids 122–166 of PsrP , promoted bacterial aggregation and that antibodies against the BR domain were neutralizing . Using similar methodologies , we also determined that SraP and GspB , the Serine-rich repeat proteins ( SRRPs ) of Staphylococcus aureus and Streptococcus gordonii , respectively , also promoted bacterial aggregation and that their Non-repeat domains bound to their respective SRRPs . This is the first report to show the presence of biofilm-like structures in the lungs of animals infected with S . pneumoniae and show that SRRPs have dual roles as host and bacterial adhesins . These studies suggest that recombinant Non-repeat domains of SRRPs ( i . e . BR for S . pneumoniae ) may be useful as vaccine antigens to protect against Gram-positive bacteria that cause infection .
Streptococcus pneumoniae is a leading cause of otitis media ( OM ) , community-acquired pneumonia , sepsis and meningitis . Primarily a commensal , S . pneumoniae typically colonizes the nasopharynx asymptomatically , however in susceptible individuals such as infants , the elderly , persons who are immunocompromised , and those with sickle cell anemia , the pneumococcus is often able to cause opportunistic diseases [1] , [2] , [3] , [4] . Worldwide , S . pneumoniae is responsible for up to 14 . 5 million episodes of invasive pneumococcal disease ( IPD ) and 11% of all deaths in children [5] , [6] . In the elderly the mortality-rate associated with IPD can exceed 20% and for those in nursing homes may be as high as 40% [7] . Thus , the pneumococcus has been and remains a major cause of morbidity and mortality . psrP-secY2A2 is a S . pneumoniae pathogenicity island whose presence has been positively correlated with the ability to cause human disease [8] . Analyses of the published S . pneumoniae genomes has demonstrated that psrP-secY2A2 is present and conserved in a number of globally distributed invasive clones , in particular those belonging to serotypes not covered by the heptavalent conjugate vaccine [9] . To date , numerous studies have shown that deletion of genes within psrP-secY2A2 attenuated the ability of S . pneumoniae to cause disease in mice . psrP-secY2A2 mutants were shown to be unable to attach to lung cells , establish lower respiratory tract infection , and were delayed in their ability to enter the bloodstream from the lungs . Importantly , the same studies found that psrP-secY2A2 did not play an important role during nasopharyngeal colonization or during sepsis following intraperitoneal challenge [10] , [11] , [12] , [13] . Thus psrP-secY2A2 is currently understood to be a lung-specific virulence determinant . In TIGR4 , a virulent serotype 4 laboratory strain , psrP-secY2A2 is 37-kb in length and encodes 18 proteins . These include the Pneumococcal serine-rich repeat protein ( PsrP ) , which is a lung cell adhesin , 10 putative glycosyltranferases , and 7 proteins homologous to components of an accessory Sec translocase [14] . To date , the latter 17 genes remain uncharacterized; however , based on their homology to genes found within the Serine-rich repeat protein ( SRRP ) locus of Streptococcus gordonii , the encoded proteins are putatively responsible for the intracellular glycosylation of PsrP and for its transport to the bacterial surface [8] , [15] , [16] , [17] , [18] . PsrP in TIGR4 is composed of 4 , 776 amino acids , has been confirmed to be glycosylated , and separates at an apparent molecular mass of 2 , 300 kDa on an agarose gel [13] . It is one of the largest bacterial proteins known . PsrP is organized into multiple domains including a cleavable N-terminal signal peptide , a small serine-rich repeat region ( SRR1 ) , a unique non-repeat region ( NR ) , followed by a second extremely long serine-rich region ( SRR2 ) , and a C-terminal cell wall anchor domain containing an LPXTG motif ( Figure 1A ) . The SRR1 and SRR2 domains of PsrP are composed of 8 and 539 serine-rich repeats ( SRR ) of the amino acid sequence SAS[A/E/V]SAS[T/I] , respectively , and are the domains believed to be glycosylated . The NR domain of PsrP has a predicted pI value of 9 . 9 , for this reason it is called the Basic Region ( BR ) domain . S . pneumoniae is surrounded by a polysaccharide capsule that protects the bacteria from phagocytosis but also inhibits adhesion to epithelial cells [19] . Based on the size and domain organization of PsrP we have previously hypothesized that the extremely long SRR2 domain serves to extend the BR domain through the capsular polysaccharide to mediate lung cell adhesion ( Figure 1B ) [12] , [13] . Consistent with this model , we have previously shown that PsrP is expressed on the bacterial surface , that the BR domain , in particular amino acids 273–341 , was responsible for PsrP-mediated adhesion to Keratin 10 ( K10 ) on lung cells , and that complementation of psrP deficient mutants with a truncated version of the protein ( having only 33 SRRs in its SRR2 domain ) restored the ability of uncapsulated but not capsulated PsrP mutants to adhere to A549 cells , a human type II pneumocyte cell line [13] . It is now recognized that biofilms play an important role during infectious diseases . Briefly , bacteria in biofilms are more resistant to host-defense mechanisms including phagocytosis and serve as a recalcitrant source of bacteria during antimicrobial therapy [20] , [21] . For S . pneumoniae , pneumococcal biofilms have been shown to occur in the middle ears of children with chronic otitis media and is thought to contribute to its refractory nature [22] . Likewise , biofilms have been detected in the nasopharynx of infected chinchillas [23] . However , until now biofilm structures have not been described in the lungs during pneumococcal pneumonia . This is in contrast to other respiratory tract pathogens , such as Pseudomonas aeruginosa and Bordatella pertussis , for which in vivo biofilm production is now recognized to be an important pathogenic mechanism [21] . Herein , we demonstrate for the first time that S . pneumoniae forms biofilm-like aggregates in the lungs . We show that this phenomenon is PsrP-dependent and mediated by its BR domain . Using recombinant protein and SRRP mutants , we show that the SRRPs of S . gordonii and Staphylococcus aureus , GspB and SraP , respectively , also promote bacterial aggregation , thus describing a previously unrecognized role for members of the SRRP family . Collectively , these findings suggest an important dual role for PsrP and other SRRPs during infection , host cell and intra-species bacterial adhesion , both of which may be targeted for intervention with antibodies against recombinant ( r ) NR .
To test whether PsrP contributed to biofilm or microcolony formation in vivo mice were infected with TIGR4 and its isogenic psrP deficient mutant , T4 ΔpsrP , and whole lung sections were examined using scanning electron microscopy ( SEM ) . As would be expected for both wild type and the mutant , the majority of bacteria present were in the form of diplococci . However , for TIGR4 we also observed the presence of large bacterial aggregates attached to ciliated bronchial epithelial cells as well as to alveolar epithelial cells ( Figure 2 ) . For quantitative analysis of this phenomenon , nasal lavage fluid and bronchoalveolar lavage ( BAL ) fluid from mice was collected two days post-challenge . Aliquots from each biological sample were heat-fixed to glass slides , Gram-stained , and examined with a microscope ( Figure 3A ) . In all , the number of bacterial aggregates composed of 2–9 , and ≥10 diplococci were significantly greater for mice infected with TIGR4 than T4 ΔpsrP in both the nasopharyngeal and BAL elute fluids ( Figure 3B , C ) . Moreover , the largest aggregates , those composed of >100 bacteria , were observed only in mice infected with TIGR4 . Fluorescent imaging of bacteria in frozen lung sections confirmed this phenotype; large bacterial aggregates were only detected in the lungs of TIGR4 infected mice ( Figure S1 ) . Thus we determined that PsrP promoted the formation of biofilm-like aggregates in vivo , including in the nasopharynx , a site previously shown not to require PsrP for bacterial colonization [12] . Given the previous results , moreover to develop an in vitro model that was amendable to manipulation , the ability of TIGR4 and T4 ΔpsrP to form early biofilms was tested using microtiter plates [24] . As shown in Figure 4A , no differences were observed between wild type and the mutant , suggesting that PsrP does not play a role in pneumococcal attachment to polystyrene or the formation of early biofilm structures , in particular the bacteria lawn . The role of PsrP was next tested in 3-day old mature biofilms using the once-through continuous flow cells as described previously by Allegrucci et al . [25] . In this system , a stark difference in the architecture of TIGR4 and T4 ΔpsrP biofilms was observed ( Figure 4B ) . Wild type biofilms displayed a dense cloud-like morphology with extremely large aggregates that covered the glass surface . Closer inspection revealed that these aggregates were composed of tightly clustered pneumococci . In contrast , T4 ΔpsrP biofilms displayed a less intimate phenotype characterized by smaller aggregates , gaps , and the formation of columns , resulting in an overall patchier phenotype . Quantitative analysis of the biofilm structures using COMSTAT software confirmed that TIGR4 biofilms had significantly greater total biomass and average thickness than those formed by the T4 ΔpsrP ( Figure 4C ) . No differences in either the maximum thickness of the biofilms or the roughness coefficient ( a measure of biofilm heterogeneity ) were observed ( Figure 4C; data not shown , respectively ) , indicating that T4 ΔpsrP could still form biofilms , although with distinct architecture . Importantly , T4 ΩpsrP-secY2A2 , a mutant deficient in the entire psrP-secY2A2 pathogenicity island , behaved identically to T4 ΔpsrP , forming patchy biofilms with small aggregates and less intimate associated bacteria ( Figure S2 ) . Bacterial biofilms were also grown under once through conditions in silicone tubing . After a designated time , the biofilms were extruded from the line and examined for biomass both visually and quantitatively . After 3 days of growth , differences between TIGR4 and T4 ΔpsrP in opacity of the exudates were visible to the eye ( Figure 5A ) and could be confirmed using a spectrophotometer which showed a >3-fold difference in optical density ( Figure 5B ) . Microscopic visualization of the line exudates following crystal violet ( CV ) staining revealed that TIGR4 had formed large aggregates whereas T4 ΔpsrP exudates were composed of small clusters or of individual diplococci ( Figure 5C ) . Increased biofilm biomass was supported by measurement of total protein concentrations that showed TIGR4 biofilm exudates had 2–3 fold more protein than those corresponding to T4 ΔpsrP ( Figure 5D ) . Of note , during planktonic growth TIGR4 , T4 ΔpsrP , and T4 ΩpsrP-secy2A2 were indistinguishable , growing either as short chains or diplococci with a marked absence of aggregates ( data not shown ) . This led us to examine psrP transcription using Real-Time PCR and the finding that TIGR4 expressed psrP at levels 47-fold greater during biofilm versus planktonic culture ( P = 0 . 04 using a Student's t-test ) . Thus low expression of psrP may be one reason TIGR4 did not form aggregates during liquid culture . To date a number of groups , including our own , have shown that SRRPs mediate bacterial adhesion to host cells primarily through their NR domain [13] , [26] , [27] . For this reason we sought to test whether the BR domain of PsrP was also involved in biofilm/bacterial aggregation . To do this we first utilized a pre-existing collection ( described in Figure S3 ) of encapsulated ( T4 ΩpsrP ) and unencapsulated ( T4R ΩpsrP ) S . pneumoniae mutants deficient in PsrP that either expressed a truncated version of PsrP with 33 SRRs in its SRR2 domain ( PsrPSRR2 ( 33 ) ) , a similar truncated version lacking the BR domain ( PsrPSRR2 ( 33 ) -BR ) , or carried the empty expression vector pNE1 [13] . These strains were tested for their ability to form biofilms in silicone lines under once through conditions . Complementation of T4 ΩpsrP with PsrPSRR2 ( 33 ) , but not PsrPSRR2 ( 33 ) -BR or the empty pNE1 vector , partially restored the ability of T4 ΩpsrP to form large aggregates in the lines when examined microscopically ( Figure 6A ) . However , measurement of other biofilm markers such as optical density and total protein concentration showed no differences between any of the complemented mutants and the negative controls ( Figure 6B–C ) . Complementation of T4R ΩpsrP with PsrPSRR2 ( 33 ) , also partially restored the ability of T4R ΩpsrP to form aggregates ( Figure 6A ) . In this instance , line exudates from T4R ΩpsrP with PsrPSRR2 ( 33 ) had significant more biofilm biomass than the negative controls ( Figure 6B–C ) . Importantly , the truncated version of PsrP lacking the BR domain failed to restore , even partially , T4 ΩpsrP or T4R ΩpsrP suggesting that the BR domain was responsible for the intra-species aggregation . This was subsequently confirmed by Far-Western blot analyses that showed that Gst-tagged recombinant BR ( Gst-BR ) bound only to S . pneumoniae cell lysates that contained a truncated PsrP with the BR domain ( Figure 6D ) and a control experiment showing that a Gst-tagged Chlamydia trachomatis protein did not interact with these lysates ( Figure S4 ) . To further explore the role of the BR domain in the observed bacteria to bacteria interactions , the ability of His-tagged BR constructs ( rBR; Figure 7A ) , purified from Escherichia coli and Cy3 labeled , were tested for their ability to bind to the surface of TIGR4 and T4 ΔpsrP . Full-length rBR interacted with TIGR4 but not with T4 ΔpsrP ( Figure 7B ) , confirming not only that PsrP bound to pneumococci , but also suggesting that its ligand was another PsrP . Furthermore , only rBR . A retained the ability to attach to PsrP on the pneumococcal surface . This suggested that the binding domain of PsrP was possibly located within AA 122–166 , the section not shared between rBR . A and rBR . B . Hereafter , BR to BR interactions were tested for by Far Western and co-immunoprecipitation . Far Western blot experiments using assorted E . coli cell lysates from bacteria expressing assorted rBR constructs , confirmed that only lysates containing PsrP constructs with AA 122–166 bound successfully to Gst-BR ( Figure 7C ) . This was also observed in co-immunoprecipitation experiments , whereby Gst-BR was tested for its ability to bind whole cell lysates from E . coli expressing versions of PsrP ( Figure 7D ) . Far Western blots using purified proteins showed that Gst-BR had affinity to purified rBR , rBR . A , and a synthesized peptide corresponding to AA 122–166 , but not rBR . B , BR . C , or the control his-tagged Streptolysin O ( Figure 7E ) . Hence , using numerous assays it was determined that the BR domain , most likely AA 122–166 , had self-interacting properties that might be responsible for the observed bacterial aggregation . Of note , because the BR constructs were purified from E . coli and PsrP is normally glycosylated , the above observations may have been an artifact of the unglycosylated constructs used . To address this possibility a glycosyated truncated PsrP construct was purified from S . pneumoniae ( PsrPSRR2 ( 33 ) -HIS; Figure S5 ) and tested for its ability to bind S . pneumoniae cell lysates containing either native PsrP or assorted constructs . As shown in Figure 7F , it was determined that a glycosylated PsrP probe maintained specificity for the BR domain even in the context of glycosylated recipient protein . A finding that supports the notion that PsrP to PsrP interactions occur in natural setting when PsrP is always glycosylated . To determine whether the BR aggregation ( AA 122–167 ) and the K10 binding subdomains ( AA 273–341 ) of BR had functionally independent roles , competitive inhibition assays were performed using rBR constructs . Bacterial adhesion to A549 cells was tested following incubation of cells with the AA 122–166 peptide , rBR , and rBR . C ( Figure 8A ) . Pre-treatment of A549 cells with AA 122–167 had no impact on adhesion . In contrast and consistent with the location of the K10 binding domain within BR . C: 1 ) TIGR4 adhered significantly less to cells treated with rBR or rBR . C , 2 ) TIGR4 adhered to BSA treated cells better than T4 ΔpsrP . In complementary biofilm experiments the opposite result was observed . Addition of 1 µM peptide AA 122–167 to media reduced the aggregation phenotype observed for TIGR4 ( Figure 8B ) and modestly lowered the optical density of the biofilm exudate and the total biomass collected from the continuous flow lines versus addition of BR . C ( Figure 8C–D ) . Thus these findings suggested that the aggregation and K10 subdomains of PsrP had distinct roles that did not overlap during host cell adhesion or biofilm formation . Finally we sought to determine a biological effect for the aggregation phenotype . We observed that after 1 hour , 69±2% of J477 macrophages incubated with planktonically grown TIGR4 were associated with FITC-labeled bacteria whereas only 51±5% of macrophages mixed with biofilm grown TIGR4 were positive ( P = 0 . 024 ) . Macrophages exposed to biofilm grown TIGR4 also took up less bacteria than macrophages mixed with planktonic ( 74±1%; P = <0 . 001 ) and biofilm ( 60±1%; P = <0 . 001 ) cultures of T4 ΔpsrP . Interestingly , a 10% reduction in macrophage uptake was observed for the biofilm versus planktonic grown T4 ΔpsrP cultures ( P = 0 . 077 ) ; and no difference was observed between macrophage uptake of TIGR4 and T4 ΔpsrP when taken from planktonic cultures . These findings suggest , that in addition to PsrP , other bacterial factors expressed during growth in a biofilm also affect opsonophagoyctosis . Previously we had shown that antibodies against the SRR1-BR domains of PsrP neutralized the ability of S . pneumoniae to attach to lung cells and that vaccination with rBR protected mice against pneumococcal challenge [12] , [13] . For this reason we tested the ability of polyclonal antiserum against rBR and against a SRR motif peptide to block bacterial aggregation in the biofilm line model . Todd Hewitt Broth ( THB ) supplemented with a 1∶1000 dilution of antiserum against the BR domain inhibited the formation of bacterial aggregates as observed by microscopic visualization of the biofilm line exudates . In contrast , bacteria in media supplemented with antiserum to the SRR motif peptide or that from naïve animals , formed aggregates similar to wild type bacteria grown under serum free conditions ( Figure 9A ) . Biofilm exudate optical density and protein concentrations supported these microscopic observations ( Figure 9B–C ) . To determine whether the effect of the BR antiserum on biofilm formation was specific for TIGR4 , we tested the ability of antibodies to the BR domain to block biofilm formation in unrelated clinical isolates ( Figure S6 ) . Antiserum against rBR from TIGR4 inhibited biofilm formation in two unrelated clinical isolates that carried PsrP . The same sera had no effect on biofilm formation by an invasive serotype 14 isolate that lacked PsrP . Therefore these studies confirmed previous observations that increased bacteria aggregation in biofilm models can occur independently of PsrP , but that if present , antiserum against BR can block the contribution of PsrP to these processes . To determine whether other SRRPs also mediated intra-species aggregation we tested the effect of gspB and sraP deletion on S . gordonii and S . aureus biofilm architecture , respectively . Deletion of gspB and sraP negatively impacted biofilm formation in the microtiter biofilm model at 24 hours ( Figure 10A , B ) . Growth of wild type and mutant bacteria in the line models also demonstrated that both proteins contributed to the formation of large aggregates during surface attached growth; although this property was much more dramatic for S . gordonii than for S . aureus which did not show a significant difference in the optical densities of the exudates ( Figure 10C , D ) . Of note , S . aureus biofilm experiments were stopped after 1 day due to bacteria overgrowth and blockage of the lines . Subsequent Far Western analysis using Gst-BR from S . pneumoniae as well as recombinant SRR1-NR from SraP and recombinant NR from GspB showed that these proteins have affinity for cell lysates from their parent strain but not for cell lysates from isogenic SRRP deficient mutants ( Figure 10E ) . This supports the notion that these proteins might be involved in intra-species aggregation . For PsrP BR from S . pneumoniae , no affinity was observed for cell lysates from either S . gordonii or S . aureus suggesting that PsrP does not play a role as an inter-species adhesin ( Figure 10E ) . In contrast , the NR constructs from S . aureus and S . gordonii bound to cell lysates from the other bacteria , even in the absence of the SRRP ( Figure 10E ) . The discrepancy between PsrP and the other SRRPs might be explained by the fact that certain SRRPs have been described to have lectin activity [26] , [27] . In contrast PsrP adhesion has been shown to be independent of lectin-activity [13] .
To date , SRRPs have been described in at least 9 Gram-positive bacteria and have been shown to function as adhesins that contribute to virulence . For example , deletion of sraP and gspB in S . aureus and S . gordonii , respectively , decreased the ability of these bacteria to bind to platelets and form vegetative plaques on heart valves of catheterized rats [27] , [28] . Similarly , Srr-1 of Streptococcus agalactiae has been shown to bind human Keratin 4 , mediate adherence to mucosal epithelial cells , and promote invasion of bacteria through human brain microvasculature endothelial cells [29] , [30] . SRRPs also mediate acellular attachment , a role important for colonization of the dental surface by oral streptococci . Froelinger and Fives-Taylor showed that Streptococcus parasanguis containing mutations of Fap1 failed to attach to saliva-coated hydroxyapatite [31] . Likewise , deletion of srpA significantly diminished the ability of Streptococcus cristatus to attach to glass slides [32] . Thus , while it was well established that SRRPs play an important role in bacterial attachment to cells or surfaces , until this report their role as intra-species adhesins remained unrecognized . A dual role , host and bacterial adhesin for bacterial surface proteins is not unprecedented . For example , in Streptococcus pyogenes and S . agalactiae , the pilus proteins mediate adhesion to epithelial cells and promote microtiter biofilm formation [33] , [34] . Likewise , for Neisseria meningitidis , PilX , also a pilus-associated protein , mediates adhesion to epithelial cells and facilitates bacterial aggregation [35] . For the pneumococcus , some evidence existed that bacterial adhesins may also have dual roles . In 2008 , Munoz-Elias et al . found that the pneumococcal adhesins Choline binding protein A and the pilus protein RrgA were both required for robust biofilm formation on microtiter plates and efficient nasopharyngeal colonization [36] . However , the attenuated biofilm phenotype was observed only with unencapsulated bacteria and encapsulated mutants formed biofilms normally . Other pneumococcal proteins shown to affect biofilm formation in vitro include Neuraminidase A , which possibly alters the extracellular matrix [37] , [38] , [39] , competence proteins , which suggest an altered protein profile [40] , [41] , and capsule synthesis enzymes , which were determined to be down regulated in biofilms [36] , [42] , [43] . Unlike PsrP , which would be expected to bridge cells directly , these proteins most likely act indirectly by altering gene expression , the extracellular milieu , or the surface availability of other adhesins , including possibly RrgA and CbpA . Our studies determined that the self-aggregating subdomain of PsrP was located in the BR domain and involves amino acids 122–166 . Recombinant BR constructs containing these amino acids were able to bind S . pneumoniae carrying PsrP , had an affinity for the BR domain in other PsrP constructs , and could modestly inhibit biofilm formation when added to media . Importantly , adhesion assays using pretreated cells and biofilm assays with rBR . C showed that the AA 122–166 was not responsible for adhesion to lung cells and that the K10 binding subdomain ( AA 273–341 ) was not involved in bacterial aggregation . Thus these subdomains appeared to have independent roles during the conditions tested . Further studies are warranted to delineate the specific AAs responsible for these adhesive properties , also to determine the structure of the BR domain and clarify how these subdomains interact with PsrP on other pneumococci and K10 on lung cells . GspB and SraP have been previously shown to bind platelets [27] , [28] . While the ligand for SraP is unknown , it has been determined that GspB binds to Sialyl T-antigen on platelet membrane glycoprotein Ibα [26] , [27] . The observation that the NR domains of GspB and SraP bound to cell lysates containing their respective SRRPs but not to their mutants and that the mutants had diminished aggregative properties suggests that SRRPs in other bacteria might also mediate aggregation in vivo . One could imagine that SraP on S . aureus or GspB on S . gordonii mediating attachment to platelets and cells in an endocarditic lesion while at the same time mediating adhesion of individual bacteria to each other . Similarly , one could envision a microcolony of the pneumococcus in the lungs with some bacteria attached to host cells via PsrP/K10 interactions and other bacteria attached to these bacteria through PsrP/PsrP interactions . Presumably , this is what was observed in the lungs of the infected mice . Interestingly , the finding that GspB and SraP NRs bound to cell lysates from other bacteria suggests that these proteins may also mediate inter-species biofilm formation . For S . gordonii , this would be relevant as the dental plaque is now recognized to be a multi-species biofilm . Importantly , neutralization of pneumococcal aggregation in biofilms with BR antiserum suggests that SRRPs might have utility as vaccine antigens . One caveat is that SRRPs would have to be one-component of a multi-valent vaccine because not all strains of S . pneumoniae , S . aureus , or the oral streptococci carry these proteins . In previous studies we had found that the length of the SRR2 domain was important for adhesion to K10 when capsule was present . Consistent with these findings , the inability of truncated PsrP to fully complement capsulated mutants supports our hypothetical model that the SRR2 domain serves to extend the BR domain away from the cell to mediate bacterial interactions . This model is also indirectly supported by Munoz-Elias et al . , who showed that down-regulation of capsule allowed CbpA and RrgA to contribute to biofilm production [36] . It is also noteworthy to state that Munoz-Elias et al . did not identify PsrP in their screen for biofilm mutants although they used TIGR4 which carries PsrP . This can be explained by the fact that we observed no contribution for PsrP in the microtiter plate early biofilm model . We observed that PsrP-mediated bacterial aggregation occurred in the nasopharynx , despite earlier studies demonstrating that K10 was absent from this site and that PsrP was not required for nasopharyngeal colonization . Aggregation of S . pneumoniae in the nasopharynx may serve as a mechanism to resist opsonophagocytosis as shown herein , or we speculate a way to resist desiccation during transmission of infectious particles . The observation that aggregates were present at an anatomical site that lacked K10 , further supports an independent role for these PsrP subdomains . In regards to opsonophagoyctosis , one important consideration is that the pneumococcus most likely has different gene expression profiles in vivo as an aggregate attached to a cell versus in vitro as a biofilm [44] . Thus caution is warranted in applying our vitro observations , such as resistance to opsonophagoyctosis or enhanced PsrP expression during biofilm growth , with events that occur in vivo . Polyclonal antibodies against the BR domain , but not the SRR motif , neutralized the ability of TIGR4 and clinical isolates carrying PsrP to form aggregates in the line model . These findings were consistent with previous studies showing that antibodies against BR also neutralized its ability to mediate adhesion to host cells and protected mice against pneumonia [12] , [13] . One possible reason that antibodies against the SRR motif peptide failed to have a neutralizing effect is that PsrP is glycosylated and antibodies against the peptide failed to recognize the native version of the protein . Alternatively , antibodies to the SRR motif may bind away from the BR domain and therefore do not inhibit the ability of the BR domain to self-interact . Interestingly , polyclonal antibodies to surface proteins often promote aggregation . This did not occur for unknown reasons . Finally , our finding that antibodies against rBR neutralized bacterial aggregation in the biofilm line model suggests that the same antibodies might also neutralize bacterial aggregation in vivo . This remains to be tested , however , the protection that was observed in mice following immunization with rBR [13] , may have been in part due to inhibition of bacterial aggregation in addition to blocking interactions with K10 . Importantly , because rNR domains produced in E . coli are not glycosylated , yet for the tested SRRPs were able to aggregate , immunoprecipitate , and bind to native protein in cell lysates , it seems that the BR domain does not require glycosylation to function as a self-adhesin . This is supported by the observation that addition of antibodies against unglycosylated rBR and that synthetic peptide AA 122–166 both inhibited bacterial aggregation in the biofilm line . In contrast to the latter concept , Wu et al . demonstrated that monoclonal antibodies specific for the glycan motifs of the serine-rich repeat motifs of Fap1 were capable of blocking attachment to saliva coated hydroxyapatite by Streptococcus parasanguis [45] . Importantly , Fap1 is the most divergent of the SRRPs and has 2 NR domains . Fap1 adhesion to saliva coated hydroxyapatite is mediated by glyconjugates on the serine-rich repeat domain [46]; as evidenced by the fact that inactivation of one of the glycosyltranferases known to modify the glycan moieties of Fap1 , drastically altered the ability of S . parasanguis to form biofilms [45] . Thus Fap1 is interesting because it suggests an NR-independent mechanism for SRRP adhesion , which is distinct from those discussed for GspB , SraP , or PsrP . Future studies need to further examine the differences between these diverse SRRPs and to determine if the two NRs of Fap1 play a role in bacterial aggregation . This is especially true given that the NR domain of SraP has a pI of 5 . 6 , in contrast to the basic NRs of GspB ( 9 . 5 pI ) and PsrP ( 9 . 9 pI ) [47] . In summary , we have described for the first time the presence of a pneumococcal biofilm-like structure in the lungs of infected mice . We have determined that PsrP mediates a more intimate bacterium to bacterium interaction that contributes to the presence of large bacteria aggregates in vivo and increased biofilm biomass and aggregates in vitro . This property appears to be shared among other SRRPs including those of medically relevant bacteria such as S . aureus and S . gordonii , suggesting that it is a conserved function for this class of proteins . How these interactions contribute to pathogenesis remains to be fully determined , however , studies with other bacteria indicate that biofilms serve to inhibit phagocytosis , protect against defensin-mediated killing , and serve as a focal point of infection during early stages of disease . Future experiments will be required to determine the extent to which this may apply for SRRP-mediated aggregates in vivo .
Wild type strains used in this study included S . pneumoniae strain TIGR4 and the previously described clinical isolates IPD-5 , TNE-6012 , and TBE-6050 [8] , [12] , [14] . T4R is an unencapsulated derivative of TIGR4 [48] . S . aureus ISP479C and S . gordonii M99 and their corresponding isogenic mutants ISP479C ΔsraP , and M99 ΔgspB have also been previously described [17] , [27] . All of the S . pneumoniae mutants used in this study including T4 ΔpsrP , T4 ΩpsrP-secY2A2 , T4 ΩpsrP , and T4R ΩpsrP have been shown not to have polar effects on upstream and downstream gene transcription [12] , [13] . S . pneumoniae and S . gordonii were grown in Todd-Hewitt broth ( THB ) or on blood agar plates at 37°C in 5% CO2 . S . aureus were grown in Tryptic-Soy Broth ( TSB ) or on blood agar plates at 37°C . Stocks for the PsrP mutants were grown in media supplemented with 1 µg/mL of erythromycin , those complemented with the expression vector pNE1 were grown on media supplemented with 250 µg/mL of spectinomycin . SraP and GspB mutant stocks were grown in media supplemented with either 15 µg/mL of erythromycin or 5 µg/mL chloramphenicol respectively . E . coli strain DH5α ( Invitrogen , Carlsbad CA ) expressing recombinant PsrP constructs were grown with 50 µg/mL of kanamycin . Recombinant proteins were purified as previously described [13] , [26] . To avoid stress effects on the bacteria , no antibiotics were added to the media during any of the experiments . Female BALB/cJ mice , 5–6 weeks old , were obtained from The Jackson Laboratory ( Bar Harbor , ME ) . Mice were anesthetized with 2 . 5% vaporized isoflurane prior to challenge . Exponential phase cultures of S . pneumoniae were centrifuged , washed , and suspended in sterile phosphate buffered saline ( PBS ) . For each experimental cohort at least 6 mice were instilled with either 107 cfu of TIGR4 or T4 ΔpsrP in 20 µL of PBS into the left nostril . After two days mice were sacrificed for tissue collection . For imaging experiments the intact lungs were collected and processed as described below . For enumeration of bacterial aggregates , nasal lavage fluid was collected from anesthetized mice by instillation and retraction of 20 µl PBS . The same mice were subsequently asphyxiated with compressed CO2 , and BAL fluid collected by flushing the lungs twice with 0 . 5 ml of PBS using a sterile catheter . All animal experimentation was conducted following the National Institutes for Health guidelines for housing and care of laboratory animals . Animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee at The University of Texas Health Science Center at San Antonio . Lungs were cut in a sagital orientation , fixed for 2 hours with 2 . 5% glutaraldehyde in PBS , and then rinsed twice for 3 min in 0 . 1 M phosphate buffer ( pH 7 . 4 ) . Lungs were submerged in 1% osmium diluted in Zetterquist's Buffer for 30 minutes then washed with the same buffer for 2 minutes [49] . This was followed by step-wise dehydration with ethanol ( i . e . 70% , 95% , and 100% ) ; the first two steps for 15 minutes , the last for 30 minutes . Samples were treated with hexamethyldisilizane for 5 minutes prior to drying in a desiccator overnight . The next day samples were sputter coated with gold palladium and viewed with a JEOL-6610 scanning electron microscope . From each mouse BAL and 1∶10 PBS diluted nasopharyngeal lavage elutes were smeared onto glass slides , heat fixed , and Gram-stained . Since the nasopharyngeal samples were mucoid , dilution of the samples was warranted . Bacteria were visualized using a CKX41 Olympus microscope at 200× magnification . For each biological sample 100 CFU were randomly selected , taking note of the approximate number of diplococci composing each CFU , either 1 , 2–10 , or >10 . Images of the bacteria were acquired at 400× magnification to better show the multiple bacteria composing the aggregates . Lung tissues were excised and frozen in Tissue Tek O . C . T solution ( Miles Scientific ) . 5 µm thick lung sections were cut at the University of Texas at San Antonio Histopathology Core and stored at −80° C . Bacteria in the lung sections were detected by immunofluorescence using antibody against the capsular polysaccharide . Sections were thawed , fixed with ice-cold acetone for 20 minutes , and then rehydrated with 70% ethyl alcohol and then PBS . Samples were permeabilized with 0 . 1% Triton-X-100 for 5 minutes then blocked with 10% fetal bovine serum ( FBS ) in F12 media for 1 hour . Sections were incubated with 1∶1 , 000 rabbit anti-serotype 4 pneumococcus antiserum ( Statens Serum Institut , Denmark ) overnight at 4°C . After washing for three times with 0 . 5% Tween-PBS , sections were covered with FBS-F12 containing goat anti-rabbit FITC conjugated antibody ( Invitrogen ) at 1∶2 , 000 and DAPI ( 5 µg/ml; for DNA ) and the sections incubated for 1 hour at room temperature . Tissue sections were washed and mounted with FluorSave ( Merck Biosciences ) . Images were acquired at 1 , 000× using a Nikon AX-70 fluorescent Microscope and images processed with SimplePCI software . Early biofilm formation was examined by measuring the ability of cells to adhere and accumulate biomass on the bottom of a 96-well ( flat-bottom ) polystyrene plates ( Costar , Corning Incorporated , Lowell MA ) [24] . Microtiter wells with 200 µl THB were inoculated with 106 CFU of S . pneumoniae taken from cultures at mid-logarithmic phase growth ( OD620 = 0 . 5 ) . Plates were incubated at 37° C in 5% CO2 . S . aureus and S . gordonii biofilm formation on microtiter plates was done in a similar manner , with the exception that TSB was used for S . aureus [50] , [51] . Bacteria were grown for 2 , 4 , 6 , 8 , 18 , and 24 h , after which the biofilms were washed gently with PBS and stained with 100 µL of 0 . 1% CV . Biofilm biomass was subsequently quantified by image capture using an inverted microscope at 15× and 100× magnification and measuring the corresponding optical density ( A540 ) of the supernatant following washing of the bacteria and solubilization of CV in 200 µL of 95% ethanol . Mature S . pneumoniae biofilms were grown under once through conditions in a glass slide chamber using a continuous-flow through reactor [25] . The flow cell was constructed of anodized aluminum containing a chamber ( 4 . 0 mm by 1 . 3 cm by 5 . 0 cm ) having two glass surfaces , one being a microscope slide and the other being a glass coverslip serving as the substratum . S . pneumoniae cells grown to mid-logarithmic phase served as the inoculum and were injected into a septum 4 cm upstream from the flow cell . Bacteria were allowed to attach to the glass substratum for 2 hours prior to initiating flow . The flow rate of the system was adjusted to 0 . 014 ml/min . Flow through the chamber was laminar , with a Reynolds number of <0 . 5 , having a fluid residence time of 180 min . Biofilms were grown at 37°C in 5% CO2 for 3 days under once through conditions . Biofilms were then visualized by confocal laser microscopy as described below . Biofilms were also grown on the interior surface of a 1-meter long , size 16 Masterflex silicone tubing ( 0 . 89mm Internal Diameter , Cole Parmer Inc . ) using once-through continuous flow conditions . The line was inoculated with 5 mL of a mid-logarithmic culture and the bacteria were allowed to attach for 2 hours . The flow rate of the system was adjusted to 0 . 035 ml/min and bacteria were grown for 3 days at 37°C in 5% CO2 . Bacterial cells were harvested from the interior surface by pinching the tube along its entire length , resulting in removal of the cell material from the lumen of the tubing . Following extraction , exudates were gently suspended in 1 ml of PBS and the optical density ( OD620 ) was measured . For light microscopy pictures , 50 µl of line exudate in saline was stained by the addition of 50 µL of 1% CV . A volume of 5 µl of stained line exudates was applied to glass slides , coverslipped , and images taken at 200× magnification using a light microscope . Viable cell counts were determined by plating serial dilutions of exudates following the disruption of each sample by vortexing . Biofilm biomass was determined by measuring the total protein concentration of the exudates by BCA following the complete lysis of S . pneumoniae with saline containing 0 . 1% deoxycholate and 0 . 1% sodium-dodecyl sulfate , which activates the murein hydrolase autolysin , or use of French press for S . gordonii and S . aureus cultures . For studies testing whether antibodies or recombinant protein inhibited bacteria aggregation media was supplemented with BR antiserum at 1∶1 , 000 or spiked with recombinant protein at a final concentration of 1 . 0 µM . Confocal scanning laser microscopy was performed with an LSM 510 Meta inverted microscope ( Zeiss , Heidelberg , Germany ) . Images were obtained with an LD-Apochrome 40×/0 . 6 lens and the LSM 510 Meta image acquisition software ( Zeiss ) . To visualize the biofilm architecture of 3-day-old biofilms , biofilms were stained using the Live/Dead BacLight stain from Invitrogen ( Carlsbad , CA ) . Quantitative analysis of epifluorescence microscopic images obtained from flow cell-grown biofilms at the 6-day time point was performed with COMSTAT image analysis software [52] . Recombinant full-length BR and truncated versions ( BR . A , BR . B , BR . C ) were expressed and purified from E . coli as previously described [13] . Glycosylated PsrPSRR2 ( 33 ) -HIS was purified in the same manner from TIGR4 ( Figure S3 ) , with the exception that cultures were induced with 1% fucose and lysed with 1% SDS in PBS . Far Western analysis was carried out as described by Takamatsu et al . with minor modifications [53] . Nitrocellulose membranes were spotted with either 1 µg of whole cell lysate of S . pneumoniae , S . gordonii , S . aureus or E . coli expressing various PsrP constructs or with purified recombinant proteins in PBS . Membranes were incubated overnight in PBS with 4% bovine serum albumin and 0 . 1% Tween 20 ( T-PBS ) at room temperature . The next day , membranes were washed with T-PBS three times for 5 minutes , and incubated overnight at 4°C on an orbital platform rocker with T-PBS containing 1% bovine serum albumin ( TB-PBS ) with 1 µg/mL of Gst-BR , PsrPSRR2 ( 33 ) -HIS , or the designated NR constructs from S . gordonii and S . aureus . Membranes were washed and incubated with monoclonal mouse anti-Gst antibody ( 1∶5 , 000 dilution ) ( Proto-Tech ) overnight at 4°C in TB-PBS . Antibody binding was detected by incubating the membranes for 1 h with HRP-conjugated anti-mouse IgG ( 1∶10 , 000 dilution ) ( Sigma ) , followed by development with the Super Signal chemiluminescent detection system ( Thermo Scientific ) . As a control for inadvertent interactions with the Gst tag , Far Western blots were also performed using an unrelated Gst-tagged Chlamydia trachomatis protein ( TC0109; Figure S4 ) . No interactions were observed . Co-immunoprecipitation of Gst-BR with the truncated versions of rPsrP was carried out as previously described by Shivshankar et al . [13] . Protein G Sepharose beads ( Amersham ) were incubated overnight at 4°C with mouse monoclonal penta-His antibody ( 1∶50; Qiagen ) in 500 ml of F12 media supplemented with 10% fetal bovine serum . Beads were incubated with 400 µl of whole bacterial lysates from E . coli expressing penta-His tagged recombinant versions of PsrP spiked with 200 µg of recombinant Gst-BR full length and incubated overnight at 4°C with gentle agitation . Beads were washed with RIPA buffer , then boiled in sample buffer for 10 min [54] . Samples were separated on 12% SDS-PAGE gels and electrophoretically transferred to nitrocellulose membranes . Membranes were blocked with T-PBS containing 4% bovine serum for 30 min at room temperature . Membranes were then incubated overnight at 4°C with mouse anti-Gst ( 1∶7500; Proto-tech ) in blocking buffer . Following incubation , membranes were washed with T-PBS three times for 5 minutes . HRP-conjugated goat anti-rabbit Immunoglobulin G ( 1∶10 000; Sigma ) was used as the secondary antibody , followed by development with the Super Signal chemiluminescent detection system ( Thermo Scientific ) . For labeling of bacteria , TIGR4 and T4 ΔpsrP were pelleted and suspended in 1 ml of carbonate buffer ( pH 9 . 0 ) containing FITC ( 1 mg/ml ) and incubated in the dark at room temperature with constant end-to-end tumbling . FITC-labeled bacteria were washed with PBS ( pH 7 . 4 ) and centrifuged , until the supernatant became clear . rBR fragments were labeled using a FluorLink-Ab Cy3 labeling kit ( Amersham ) using the instructions provided by the manufacturer . Labeled bacteria were suspended in serum-free F12 media containing the labeled constructs for 1 hour and gently mixed . Subsequently , pneumococci were washed and suspended in F12 medium . Labeled bacteria and bound recombinant protein were visualized using an AX-70 fluorescent microscope and the images were captured at 0 . 1112–0 . 8886 ms exposure time for Cy2 and Cy3 filters . The magnification used for capture of digital images was 1000× . Captured images were processed using Simple PCI software . A549 cells ( human alveolar type II pneumocytes; ATCC CRL-185 ) , were grown to 90% confluence on 24-well plates ( ∼106 cells/well ) . Prior to use , cells were washed with cell F12 media to remove serum . For competitive inhibition binding assays , A549 cells were incubated with 1µM of either rBR , rBR . C , a synthesized peptide corresponding to AA 122–167 , or BSA for 1 HR . Following incubation , cells were exposed to media that contained 107 cfu/mL of bacteria and incubated for 1 h at 37°C in 5% CO2 . Nonadhering bacteria were removed by washing the cells 3 times with T-PBS and the number of adhering bacteria was determined by lysis of the monolayer with 0 . 1% Triton X-100 and plating wells per experiment . Bacterial cultures were centrifuged and suspended in 0 . 1M sodium carbonate buffer ( pH 8 . 0 ) at an OD620 of 0 . 2 . Care was taken to cause minimal disruption of the biofilm aggregates . The diluted cultures were labeled with fluorescent isothiocyanate ( 1mg/ml ) for 30 min at room temperature in the dark . Following labeling , cultures were gently washed three times with sterile PBS to remove free FITC and suspended in PBS . FITC-labeled bacteria were opsonized with 3% control rabbit serum for 30 minutes at 37°C with mild periodical tapping . Mouse J774 . 1 , macrophage cultures maintained in 10% FBS containing DMEM were used for phagocytosis of the opsonized pneumococci . Macrophages were harvested , washed and diluted with opsonophagocytosis buffer ( PBS containing 0 . 2% BSA ) . FITC-labeled bacteria in 100 µl were added to 106 macrophage cells in 400 µl and incubated for 1 hour at 37°C with periodic shaking . Afterwards , the macrophages were pelleted and washed twice in the assay buffer . Cells were suspended in 400µl of 2% paraformaldehyde until flow cytometric analysis . A2-Laser BD FACSCaliber Analyzer ( Becton Dickinson , NJ; Institutional Flow Cytometry Core Facility at the Health Science Center ) was employed to analyze percent phagocytic uptake of the labeled bacteria by the macrophages . A minimum of 20 , 000 events were counted for each sample at 480 nm excitation and 530nm detection wavelengths . Background fluorescence was nullified by subjecting negative control macrophages in assay buffer without any fluorescent bacteria to FACS analysis . Data were processed using CellQuest software . For pair-wise comparisons of groups statistical analyses were performed using a Student's t-test . For multivariate analyses a 1-Way ANOVA followed by a post-priori test using Sigma Stat software was used . | Serine-rich repeat proteins ( SRRPs ) are a family of surface-expressed proteins found in numerous Gram-positive pathogens , including Staphylococcus aureus , Streptococcus pneumoniae , Group B streptococci , and the oral streptococci that cause infective endocarditis . For all of these bacteria , SRRPs have been demonstrated to play pivotal roles in adhesion to tissues and the development of invasive disease . It is now known that biofilm formation is an important step for bacterial pathogenesis . Bacteria in biofilms have been shown to have differences in metabolism , gene expression , and protein production that contribute to enhanced surface adhesion and the persistence of an infection . Herein we describe a novel role for PsrP , the S . pneumoniae SRRP , as an intra-species bacterial adhesin that promotes bacterial aggregation in the lungs of infected mice during pneumonia . In vitro we show that the Basic Region domain of PsrP promotes self-interactions that result in denser biofilms , greater biofilm biomass , and altered architectures of surface grown cultures; these interactions could be neutralized by antibodies to PsrP that are protective against pneumococcal infection . We also demonstrate that the SRRPs of S . aureus and Streptococcus gordonii also function as intra-species bacterial adhesins . Therefore we conclude that SRRPs have dual roles as host-cell and intra-species bacterial adhesins . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"infectious",
"diseases/bacterial",
"infections",
"microbiology",
"infectious",
"diseases/respiratory",
"infections"
] | 2010 | The Pneumococcal Serine-Rich Repeat Protein Is an Intra-Species Bacterial Adhesin That Promotes Bacterial Aggregation In Vivo and in Biofilms |
The proto-oncoprotein Raf is pivotal for mitogen-activated protein kinase ( MAPK ) signaling , and its aberrant activation has been implicated in multiple human cancers . However , the precise molecular mechanism of Raf activation , especially for B-Raf , remains unresolved . By genetic and biochemical studies , we demonstrate that phosphorylation of tyrosine 510 is essential for activation of Drosophila Raf ( Draf ) , which is an ortholog of mammalian B-Raf . Y510 of Draf is phosphorylated by the c-src homolog Src64B . Acidic substitution of Y510 promotes and phenylalanine substitution impairs Draf activation without affecting its enzymatic activity , suggesting that Y510 plays a purely regulatory role . We further show that Y510 regulates Draf activation by affecting the autoinhibitory interaction between the N- and C-terminal fragments of the protein . Finally , we show that Src64B is required for Draf activation in several developmental processes . Together , these results suggest a novel mechanism of Raf activation via Src-mediated tyrosine phosphorylation . Since Y510 is a conserved residue in the kinase domain of all Raf proteins , this mechanism is likely evolutionarily conserved .
The Raf serine/threonine kinase is a key component of the evolutionarily conserved signal transduction module that also includes the Ras GTPase , the mitogen and extracellular signaling-regulated kinase kinase ( MEK ) , and the extracellular signaling-regulated kinase ( ERK ) [1 , 2] . In the canonical model , receptor tyrosine kinase ( RTK ) activation by extracellular signals such as peptide ligands leads to , via a series of adaptor proteins , the activation of Ras , which switches from GDP- to GTP-bound form . Ras-GTP binds to and thus causes the translocation of Raf to the plasma membrane , where it is activated by mechanisms that are still not completely resolved . It has been reported that somatic mutations in B-Raf are found in 60% of malignant melanomas and are also associated with other types of human cancers [3 , 4] , which underscores the importance of this signaling pathway in tumorigenesis . The mammalian Raf family consists of A-Raf , B-Raf , and C-Raf ( also known as Raf-1 or c-Raf ) , which share three highly conserved regions ( CR1–3; see Figure S1 ) [5 , 6] . The sole Raf homolog present in the Drosophila genome , Draf , is encoded by lethal ( 1 ) pole hole ( phl ) ; it contains all three conserved domains and is most homologous to B-Raf and secondarily to C-Raf , with 61% and 54% overall sequence similarity , respectively [7 , 8] . Draf shares similar substrate specificity and regulatory mechanisms with C-Raf and very likely also with B-Raf , as expression of human C-Raf in Draf mutants can restore Draf signaling [9 , 10] . CR1 and 2 are located in the N-terminal regulatory region , which acts to repress the catalytic activity of kinase domain located in the C-terminal CR3 . Two subdomains in CR1 , a Ras-binding domain ( RBD ) and a cysteine-rich domain ( CRD ) , are responsible for binding to Ras-GTP [5 , 6] . The mechanism of Raf activation is complex and , to date , not fully understood . Numerous factors , including Ras , Ksr , CNK , 14-3-3 , Src , and others , have been reported to regulate Raf activation [5 , 6 , 11] . Binding to Ras-GTP , an important first step leading to Raf activation , serves to translocate Raf to the membrane and subject Raf to activation by other factors localized at the membrane [12] . Following Ras binding , modifications of Raf by phosphorylation that occur at the plasma membrane appear essential for Raf activation [13 , 14] ( Figure S1 ) . C-Raf is mainly phosphorylated on S338 , Y341 , T491 , and S494 following Ras-induced activation , and numerous factors have been implicated in phosphorylating these sites [6 , 11] . S338 and Y341 reside in the “N region , ” a negatively charged regulatory region located at the N-terminus of the kinase domain , about 20 amino acids from the ATP-binding domain [15 , 16] . Phosphorylation of these sites is believed to play a role in relieving the inhibitory N-terminus from the C-terminal kinase domain [17 , 18] . The p21-activated kinase 3 ( PAK3 ) has been identified as a kinase capable of phosphorylating S338 [19 , 20] . Y341 is phosphorylated as a result of overexpression of activated Src family kinase ( SFK ) Src in cultured cells and in vitro , and substitution of this residue with the phosphomimetic aspartate increases C-Raf activity [15 , 21–23] . In addition , it has also been shown that Src activity is required for rapid activation of mitogen-activated protein kinase ( MAPK ) signaling [24] , and that Src can function downstream of RTK to induce Shc/Grb association , leading to Ras activation [25 , 26] . However , studies of mouse cells lacking three SFKs ( Src , Yes , and Fyn ) indicate that SFKs are mostly dispensable for RTK signaling [27] . Thus , the interplay between SFKs and RTKs is complex , and whether endogenous Src plays a direct role in Raf activation remains unclear . In addition to the N region , T491 and S494 in the activation segment of the kinase domain are also phosphorylated in a Ras-dependent manner [16] . Recent studies of B-Raf suggest that phosphorylation on these activation loop residues may be important for catalytic activity and/or stabilization of the active conformation of the kinase [28] . However , the kinases responsible for their phosphorylation remain elusive . Interestingly , in B-Raf , the equivalent of S338 is constitutively phosphorylated and the equivalent of Y340/341 are occupied by aspartic acids; consequently , B-Raf exhibits higher basal activity and was shown not to be regulated by Src [15] . Similar to B-Raf , Draf also contains two acidic residues ( glutamates ) at positions equivalent to Y340/341 of C-Raf , and the equivalent to S338 in C-Raf is also constitutively phosphorylated [29] ( Figure S1 ) . Thus , it has been unclear how B-Raf or Draf are regulated in vivo . In an effort to identify potential activators of Draf , we have previously conducted a genetic screen for genes involved in Draf signaling , taking advantage of a Draf hypomorphic allele , DrafSu2 , which causes partial lethality to flies [30] . DrafSu2 encodes a Draf variant with two point mutations that abolish its Ras-binding ability , making it more sensitive to reductions in Ras-independent Draf activators [8 , 30] . This genetic screen identified Src64B as a potential Draf activator , as reducing the gene dosage of Src64B dominantly enhanced the lethality associated with DrafSu2 flies [30] . To determine the role of Src64B in Draf activation in vivo , we genetically and biochemically investigated the function of Src64B in Draf activation . Here , we show that Src64B behaves as a direct Draf activator in vivo . An activated form of Src64B induces Draf target genes in the absence of RTK or Ras in vivo , and associates with and phosphorylates Draf . Moreover , we identified a novel tyrosine ( Y510 ) within the Draf kinase domain that mediates Draf phosphorylation by Src64B in vitro . Interestingly , the role of Y510 appears to be purely regulatory , as mutating it to phenylalanine or glutamate had no significant impact on the enzymatic activity of a Draf kinase domain fragment . However , mutating Y510 to glutamate resulted in activation of full-length Draf and reduced affinity between N- and C-terminal Draf fragments . These results suggest that Y510 phosphorylation plays an essential role in Draf activation by interfering with the association of the C-terminal kinase domain with the inhibitory N-terminal regulatory region .
To investigate whether Src64B plays a direct role in Draf activation , we first tested whether it can induce Draf activation in the absence of Ras1 . We examined the effects of expressing an activated form of Src64B ( referred to as Src64Bact; a . k . a . Src64BΔ540 ) [31] on activities of the Torso-Ras1-Draf signaling pathway using the Torso target gene tailless ( tll ) as a readout [32 , 33] . Src64Bact ( or Src64BΔ540 ) is truncated from amino acid ( a . a . ) 540 and lacks the C-terminal autoinhibitory domain [31] . The posterior expression domain of tll in the early embryo reflects quantitatively the strength of Torso or Draf activation [8 , 33–35] . tll is expressed from 0% to 15% of egg length ( EL ) from the posterior in wild-type embryos and is absent or little detected in embryos lacking Torso , Draf , or Ras1 ( Figure 1A; also see Hou et al . , 1995 [35]; Li et al . , 1997 [34]; see Materials and Methods for mutant embryo production ) . We have previously shown that heat-shock induction of Src64Bact can induce ectopic tll expression in wild-type embryos [30] . If Src64B plays a role in activating Draf , we expect that Src64Bact would induce tll expression in the absence of Torso or Ras1 , but not Draf . Indeed , we found that a brief induction of Src64Bact expression in the early embryos was capable of inducing expression of tll in embryos lacking Torso or Ras1 , but not in those lacking Draf ( Figure 1A ) . Consistent with its effects on tll expression , expression of Src64Bact was able to rescue to a certain extent the cuticular defects associated with torso or Ras1 , but not Draf mutant embryos ( Figure 1A ) . It has been shown that activated Src can induce the formation of a complex between Shc and Grb2 , an event upstream of Ras in the activation of this signaling pathway [25] , and that the effects of Src64Bact on eye differentiation are sensitive to the dosage of Ras1 [31] . However , our results indicate that Src64B is able to function downstream or in parallel to Ras1 but upstream of Draf , suggesting it might be able to directly activate Draf . To test whether endogenous Src64B is required for Draf activation in vivo , we sought to analyze the phenotypes of existing mutant alleles of Src64B . Src64BΔ17 and Src64BPI are the strongest mutant alleles isolated to date , although both are hypomorphs [36] . Homozygous mutant animals for these alleles exhibit defects in oogenesis [36] . We focused on Src64BΔ17 , which is associated with a deletion that removes the first two noncoding exons of Src64B [36] . Other than the oogenesis defects , which result in partial female sterility , Src64BΔ17 homozygotes are morphologically normal and exhibit no discernable developmental or behavioral defects ( unpublished data; also see [36] ) , and all aspects of RTK signaling examined appear normal ( see below ) . This suggests that the residual activity in the mutant suffices for development , or that Src64B is functionally redundant . The role of Ras in Raf activation is mainly attributed to Ras binding to the inhibitory N-terminus of Raf and targeting Raf to the membrane [8 , 37 , 38] . Since the Draf allele , DrafSu2 , harbors point mutations that abolish its interaction with Ras [8] , it is thus an ideal tool to test whether Src64B is required to directly activate Draf in vivo . To determine whether the activation of Draf , especially the activation of DrafSu2 , is impaired in Src64B mutants , we measured Draf kinase activity in protein extracts from different mutant combinations . Using protein extracts and in vitro kinase assays with the downstream suppressor of Raf 1 ( Dsor1; a MEK homolog ) as substrate , we found that extracts from Src64BΔ17 homozygotes exhibited reduced levels of Draf activity ( Figure 1B , lane 2 ) , and the reduction in Draf kinase activity was even more pronounced in extracts from DrafSu2; Src64BΔ17 double homozygotes ( Figure 1B , lane 4 ) . The Draf activity detected in fly embryo extracts in vitro was indeed dependent on the presence of Draf , as depletion of Draf from the protein extracts prior to kinase assay by an anti-Draf antibody abolished Dsor1 phosphorylation ( Figure 1B , lane 5 ) . Thus , even though Src64B hypomorphic mutants are morphologically normal , their Draf activity is severely compromised . We next determined whether addition of Src64Bact could restore the activity of Draf immunoprecipitated from Src64BΔ17 mutant embryos . Draf immunoprecipitated from Src64BΔ17 and wild-type fly embryos had undetectable and low kinase activities , respectively ( Figure 1C , lanes 2 , and 3 ) . However , addition of bacterially purified Src64Bact ( see Figure S2 for activity ) significantly increased the kinase activities of immunoprecipitated Draf proteins from Src64BΔ17 and wild-type embryos to comparable levels ( Figure 1C , lanes 4 and 5 ) . Src64Bact did not directly cause Dsor1 phosphorylation ( Figure 1C , lane 1 ) . A kinase-dead version of Src64Bact had no effect on immunoprecipitated Draf ( unpublished data ) . These results suggest that the deficit in Draf kinase activity exhibited by Src64BΔ17 flies might be due to reduced phosphorylation by Src64B . Finally , consistent with the idea that Src64B activation increases Draf activity in vivo , we found that expressing Src64Bact rescued the lethality associated with DrafC110 hemizygous males ( Figure S3 ) . DrafC110 is a hypomorphic allele , and DrafC110 hemizygous males die as late pupae [7] . In summary , the above independent pieces of evidence suggest that Src64B is required for Draf activation in vivo , that activated Src64B can mediate Draf activation independent of Ras1 , and that Src64B might directly activate Draf by phosphorylation . To investigate the mechanism of Src64B-mediated Draf activation , we first determined whether Src64B binds to or phosphorylates Draf . By transfection and coimmunoprecipitation experiments , we found that Src64Bact is indeed able to bind to Draf , mainly to the N-terminal half , and very weakly to the C-terminus ( Figure 2A ) . Moreover , we found that when expressed in Drosophila S2 cells , both Src64Bact and its kinase-dead version Src64BKR were able to bind to endogenous Draf , as the endogenous Draf was coimmunoprecipitated with transfected Src64B molecules ( Figure 2B ) . However , tyrosine phosphorylation was detected only in Draf coimmunoprecipitated with Src64Bact , suggesting that Src64Bact can directly phosphorylate Draf ( Figure 2B ) . Indeed , using bacterially expressed proteins and in vitro kinase assays , we found that GST-Src64Bact phosphorylated a C-terminal Draf fragment , Draf-C , consisting of the kinase domain ( see below ) . It has previously been reported that Fyn/Src binds to and phosphorylates C-Raf on tyrosine residues [39] . Interestingly , unlike other Src substrates that bind to Src through SH2-phosphotyrosine interaction , binding to C-Raf by the SH2 domain of Src does not require phosphotyrosine residues on C-Raf [39] . Our results are consistent with this finding and suggest that Src64B associates with Draf N-terminal region in a phosphotyrosine-independent manner and that Src64B can phosphorylate Draf . We next searched for candidate tyrosine residues in Draf that can be phosphorylated by Src64Bact . Sequence comparison revealed two tyrosine residues within the kinase domain ( Y510 and Y538 ) that are conserved among all Raf proteins ( Figure 2C ) . We mutated these tyrosine residues to phenylalanine , a non-phosphorylatable neutral substitution for tyrosine , and tested whether any of the mutations would affect Draf phosphorylation by Src64B . By comparing tyrosine phosphorylation levels of Draf-CWT with those of Draf-CY510F and Draf-CY538F in the presence or absence of GST-Src64Bact , we found that the Y510F substitution greatly diminished phosphorylation by GST-Src64Bact , whereas Y538F had little effect ( Figure 2D ) . Since Src64B binds to Draf mainly through the N-terminal region ( Figure 2A ) and Src can bind to Raf independent of phosphotyrosine residues ( Figure 2B ) [39] , Y510 of Draf-C is unlikely responsible for binding with Src64B . Thus , a plausible explanation for the above results is that Y510 is the major Src64B phosphorylation site on Draf . We confirmed that Src64B phosphorylates Draf on Y510 by producing polyclonal antibodies specific for phospho-Y510 of Draf ( anti-pY510; see Materials and Methods and Figure S4 ) . Indeed , the anti-pY510 antibody recognized phosphorylated DrafWT , but not DrafY510F , following incubation with Src64Bact ( Figure 2E ) . Consistent with the in vitro kinase assay data , pY510 levels were increased in Src64Bact-transfected S2 cells ( Figure 2F ) , suggesting Src64B can phosphorylate endogenous Draf on Y510 in vivo . To investigate whether phosphorylation of Y510 in Draf correlates with Draf activation in vivo , we examined the phosphorylation status of Draf immediately following its activation in S2 cells . It has previously been shown that S2 cells respond to insulin stimulation by activating Draf [40] . Immediately following insulin stimulation , we detected tyrosine phosphorylation , albeit low , in transfected DrafWT , but not DrafY510F ( Figure 2G ) . It has previously been shown for C-Raf that activated Raf , which resides in the membrane and is tyrosine phosphorylated , constitutes only a minority of total Raf proteins [13 , 14] . This may explain the difficulty to detect tyrosine phosphorylation of Raf proteins . Taken together , these results suggest that Y510 becomes phosphorylated upon Draf activation . To explore whether Y510 is important for Draf activation , we measured the kinase activity of purified Draf proteins with different amino acid substitutions at Y510 . An acidic substitution can mimic the effects of tyrosine phosphorylation by its negative charge [17 , 21 , 41–43] . We found that full-length Draf proteins with different Y510 substitutions exhibited different kinase activities when purified from bacteria . Whereas DrafY510F and DrafWT exhibited no and barely detectable kinase activities , respectively , DrafY510E , however , exhibited dramatically higher kinase activities ( Figure 3A and 3B ) . Similar results were found when full-length DrafWT , DrafY510F , and DrafY510E proteins were expressed in and immunoprecipitated from S2 cells ( Figure 3C ) . Thus , acidic substitution of Y510 results in activation of full-length Draf . To confirm that Y510E mimics Src64B phosphorylation , we subjected DrafWT and DrafY510E immunoprecipitated from S2 cells to added Src64Bact . We found that Src64Bact greatly stimulated the kinase activity of DrafWT but hardly affected that of DrafY510E ( Figure 3D ) . To investigate the activities of Draf with different Y510 substitutions in vivo in a developmental context , we expressed these molecules in early Drosophila embryos by mRNA injection . mRNA injection into early embryos has been used as a robust assay for functions of expressed proteins , including Draf [8 , 44] . Based on the ability of injected Draf mRNA to rescue the mutant phenotypes of Draf or torso null embryos ( see Materials and Methods for mutant embryo production ) , the basal and inducible activities of different Draf proteins expressed can be quantitatively assessed [8 , 44] . Wild-type Drosophila embryos develop distinct cuticle patterns with eight abdominal ventral denticle bands and a prominent “tail” ( Filzkörper; Figure 3E i , arrowhead ) . The posterior structures ( A8 and Filzkörper ) are specified by the Torso RTK signal transduction pathway , mediated by the Ras1/Draf signaling cassette [33] , and torso or Draf null embryos fail to develop these posterior structures ( Figure 3E ii ) . We injected mRNA encoding each version of full-length Draf into torso embryos , in which the endogenous Draf remains inactive , and found that DrafY510E restored the posterior structures in 9 . 5% of injected torso embryos , whereas no rescue was found for DrafY510F and DrafWT ( Figure 3E iii and 3F ) . This suggests that DrafY510E has higher basal activity ( p < 0 . 001 ) , which is consistent with the in vitro kinase assay data . When injected into Draf embryos , which are devoid of the endogenous Draf , DrafY510E exhibited the highest rescuing activity , followed by DrafWT and DrafY510F ( Figure 3F ) . These results demonstrate that acidic substitution of Y510 results in Draf activation in vivo and a conserved substitution to a non-phosphorylatable residue renders Draf recalcitrant to activation . To further confirm the mRNA injection results above , we generated transgenic flies carrying the three versions of full-length Draf transgenes under the control of the Gal4-inducible promoter ( UAS; [45] ) . First , we tested the ability of each Draf transgene to rescue the lethality of flies hemizygous for the hypomorphic allele DrafC110 [7] . We found that , when expressed at low levels by the basal activity of hsp70-Gal4 , DrafY510E and DrafWT rescued 99% and 80% , respectively , of the DrafC110 hemizygotes , whereas DrafY510F had very low ability to do so ( Figure 3G ) . Upon heat-shock induction , DrafY510E , but not DrafWT , caused lethality ( unpublished data ) . Second , we used a maternal Gal4 driver ( see Materials and Methods ) to express the Draf transgenes in a torso null background and examined the cuticle structures of the resulting embryos . We found that expressing DrafWT had no effect on torso−/− embryos ( Figure 3H i ) , and expressing DrafY510F worsened the torso−/− phenotypes ( Figure 3H ii ) , suggesting that DrafY510F may have dominant-negative effects . In contrast , expressing DrafY510E had dramatic effects , resulting in rescuing torso−/− embryos even to full viability , such that 68% of these embryos ( n = 86/127 ) hatched to crawling larvae that subsequently developed to morphologically normal adults ( unpublished data ) . Although the rest of the embryos failed to hatch , they exhibited wild-type posterior structures , with normal-appearing A8 and the Filzkörper ( Figure 3H iii , arrowhead ) . Finally , we expressed these Draf transgenes in the developing eye and found that indeed DrafY510E behaved like an activated and DrafY510F a dominant-negative form of Draf ( see below ) . These results strongly suggest that DrafY510E possesses elevated levels of kinase activity that are sufficient to overcome the lack of an upstream receptor Torso . We next explored the molecular mechanisms for the involvement of Y510 phosphorylation in Draf activation . We first tested whether Y510 modification is important for the enzymatic activity of the Draf kinase by mutating Y510 in Draf-C , consisting of the kinase domain only . Surprisingly , in contrast to the full-length Draf ( see Figure 3A and 3C ) , different Draf-C variants immunoprecipitated from S2 cells exhibited comparable in vitro kinase activities and kinetics , and the levels were similar to the activity of full-length DrafY510E ( Figure 4A ) . Similar results were found for the Draf-C variants purified from bacteria ( Figure S5 ) . These results indicate that Y510 is not directly involved in the enzymatic reaction of the kinase or its substrate recognition . Y510 and Y538 of Draf are equivalent to Y537 and Y565 of human B-Raf , respectively . Based on the crystal structure of B-Raf [28] , Y565 is partially buried into the kinase domain ( Figure S6 ) , which may explain why Src64B cannot phosphorylate Y538 in Draf . As a conserved amino acid , Y538 may be important for the structure of Draf . Indeed , we found that the Y538F mutation completely abolished the kinase activity of Draf-C in S2 cells ( Figure S7 ) . In contrast , Y537 is exposed on the surface of B-Raf kinase domain ( Figure S6 ) , and mutating its equivalent in Draf had no effect on the kinase activity of Draf-C ( Figures 4A and S7 ) . These results are consistent with the idea that Y510 of Draf is accessible to modification , plays a regulatory role , and yet may not be critical for maintaining the structure of the kinase domain . Since the full-length Draf differs from Draf-C by the presence of the inhibitory N-terminal regulatory region , the different kinase activities exhibited by full-length and Draf-C proteins with the same Y510 substitutions ( cf . Figures 3A–3C and 4A ) suggest that Y510 may normally mediate the inhibitory association between the N-terminal regulatory domain and C-terminal kinase domain of Draf , and that mutating Y510 to a charged residue may disrupt Draf N- C-fragment interaction , resulting in an open configuration and exposed kinase domain . To test this idea , we investigated the ability of Draf-N to bind to different versions of Draf-C by coimmunoprecipitation . It has been shown that overexpressing C-Raf N-terminal fragment inhibits separately expressed C-Raf activity by physical interaction with its C-terminal kinase domain [17] . Consistent with this report , we found that separately expressed Draf-N was indeed able to bind to Draf-C ( Figure 4B; lane 3 ) . As predicted , this interaction is impaired by the Y510E mutation ( Figure 4B; lane 4 ) , suggesting that the full-length DrafY510E has reduced self-inhibition and higher basal activity as observed . Thus , phosphorylation of Y510 by Src64B may play an important regulatory role in Draf activation by relieving the autoinhibition of full-length Draf imposed by its own N-terminal regulatory domains . To test whether endogenous Src64B is generally required for Draf activation in vivo , we analyzed the phenotypes of Src64B mutants in multiple biological processes that require RTK-Draf signaling . In Drosophila , well-characterized RTKs include Torso , epidermal growth factor receptor ( EGFR ) , and Sevenless [46] . We have shown that Src64BΔ17 mutant flies possess reduced Draf kinase activity ( see Figure 1B ) , which can be attributed to lack of tyrosine phosphorylation of Draf by Src64B ( see Figure 1C ) . However , although Draf kinase activity is reduced in Src64BΔ17 mutants , these flies nonetheless do not exhibit any overt phenotypes that can be attributed to lack of Draf activation . To investigate the importance of Src64B in Draf activation in vivo , we examined the role of Src64B in genetic backgrounds in which Draf signaling is reduced . As with the genetic screen in which Src64B was identified [30] , such genetic backgrounds may be more sensitive to a reduction in Src64B activity . To this end , we generated double-mutant combinations between Src64BΔ17 and mutant alleles of torso , Draf , and Egfr , and examined the phenotypic consequences in a few RTK-mediated developmental processes in which the requirement for Draf has been well defined . Although we were unable to detect any phenotypic defects in the Torso system ( unpublished data ) , we were able to show that Src64B mutation compromised signaling by EGFR and Sevenless ( see below ) . The different outcomes of these genetic tests possibly reflect a different threshold requirement for Draf activation and/or functional redundancy among Src proteins ( see Discussion ) . During oogenesis , EGFR-Draf signaling is required for specifying the dorsal anterior cell fates in the follicular epithelium of the egg chamber [10] . Reductions in EGFR or Draf gene activities cause ventralization of the egg chamber , resulting in fusion or missing of the pair of dorsal appendages of the eggshell as well as a reduction in the expression of the EGFR target gene kekkon ( kek ) ( Figure 5A and 5B ) [47 , 48] . Egfr heterozygotes ( Egfr/+ ) do not exhibit any discernable abnormalities and were indistinguishable from wild type in eggshell morphology ( unpublished data ) or kek expression ( Figure 5B ) . Src64BΔ17 homozygous females lay fewer and smaller eggs [30 , 36] , presumably due to a disruption in the cytoplasmic transfer from the nurse cells to the oocyte during oogenesis . These eggs , however , showed normal spacing between the pair of dorsal appendages ( unpublished data; also see [36] , indicating that EGFR signaling was normal in Src64BΔ17 egg chambers . In contrast , females heterozygous for Egfr and homozygous for Src64BΔ17 ( Egfr/+; Src64BΔ17 ) laid eggs that were ventralized in 100% of them ( n > 500; Figure 5A ) , suggesting a deficiency in EGFR signaling . We next examined the expression of the EGFR target gene kek . In wild-type as well as in Src64BΔ17 homozygous or Egfr heterozygous flies , kek is expressed in a gradient with the highest levels in the dorsal anterior region of the follicle cell layer overlying the oocyte nucleus ( Figure 5B , left; unpublished data ) [47] . Consistent with the ventralized phenotype of Egfr/+; Src64BΔ17 animals , kek expression in the dorsal anterior region of stage 10 egg chamber was undetectable ( Figure 5B , middle ) . Conversely , expression of a Src64Bact transgene , which encodes a constitutively active form of Src64B [31] , resulted in expansion of kek expression domain to the ventral region of the follicle layer ( Figure 5B , right ) . Thus , Src64B is required for the expression of the EGFR target gene kek and patterning the dorsal appendages of the eggshell . We next investigated whether Src64B also plays a role in mediating signaling from the RTK Sevenless in photoreceptor differentiation during eye development ( Figure 5C ) . It has been shown that Sevenless ( Sev ) signaling is required for specifying the R7 photoreceptor cell fate [49] . A loss of Sev or a reduction in Ras1 or Draf function results in the loss of R7 , and overactivation of these molecules leads to supernumerary R7 phenotype [49] . Src64BΔ17 homozygotes have normal eyes and all the ommatidia in eye sections were of normal morphology and were indistinguishable from those of wild type ( n > 600 ommatidia; unpublished data ) . Normal morphology and the presence of R7 were also found in DrafSu2 homozygotes ( unpublished data; also see [50] . However , DrafSu2; Src64BΔ17 double homozygotes exhibit slightly smaller and rougher eyes; their eye sections revealed unequal spacing , and portions of the ommatidia are missing R7 ( Figure 5C , right ) . It has previously been shown that expression of Src64Bact can induce ectopic R7 photoreceptor formation [31] and ectopic expression of the Torso target gene tll [30] , and expression of a dominant-negative Src64B results in loss of R7 cells [31] . Thus , Src64B is required for Sev RTK signaling . Moreover , we investigated the effects of expression Draf variants on eye development . When expressed by the eye-specific drive GMR-Gal4 , DrafWT and DrafY510E caused rough eye phenotypes , whereas DrafY510F led to a much-reduced eye size ( Figure 5D , top row ) . To understand these eye phenotypes with regard to Draf activity , we expressed the Draf variants in the background of Ras1V12 expression . Ras1V12 overactivates Draf and causes rough eyes [51] . We found that DrafWT partially suppressed , DrafY510E enhanced , and DrafY510F was epistatic to the rough eye phenotype due to Ras1V12 expression ( Figure 5D , bottom ) . These results are consistent with the interpretation that DrafWT was slightly and DrafY510F strongly dominant-negative , whereas DrafY510E had elevated kinase activity . We further found that expression of Src64Bact in the wing imaginal disc causes extra vein formation ( unpublished data ) and activation of Rolled/ERK ( Figure S8 ) . Taken all together , these results strongly suggest that Src64B functions in multiple developmental processes that require RTK signaling , consistent with a role in Draf activation in vivo .
In light of our results , we propose the following general model to account for Raf activation by phosphorylation ( Figure 6 ) . In the inactive state of Raf , Y510 may participate in physical interaction between the N-terminal regulatory domains and the C-terminal kinase domain . Binding to Ras by the N-terminus will transiently dissociate it from the C-terminus , forming to an “open” conformation and exposing Y510 . Subsequent phosphorylation of Y510 will prevent a reassociation of the N- and C- termini , stabilizing the “open” conformation of Raf . This model is supported by results of our mutagenesis studies . Mutations in Y510 in the context of Raf C-terminus have minimal effects on Raf kinase activity . In contrast , mutating Y510 to an acidic residue ( glutamate ) in full-length Raf could result in a static electrical hindrance similar to the effects of phosphorylation , thereby preventing the formation of the inactive , “closed” Raf configuration and leading to Raf activation . Phosphorylation as a means to interfere with interaction between protein domains has been documented for Raf and other proteins [17 , 21 , 41–43] . It has previously been shown that expressing activated Src in cultured mammalian cells leads to C-Raf activation and a concomitant phosphorylation on Y340 and/or Y341 [15 , 21–23] , and that purified Src can directly phosphorylate Y341 in vitro [52] . Y340 and Y341 immediately follow two serine residues ( S338 and S339 ) . Together , these residues constitute the N-region that appears important for C-Raf regulation [6 , 11] . In Draf , the positions equivalent to Y340/341 of C-Raf are occupied by two glutamate residues . The sequence SSEE in the N-region of Draf is thus more similar to the arrangement in B-Raf , which has SSDD occupying these positions . Since it has been reported that Src does not activate B-Raf [15] , the sequence similarity between Draf and B-Raf has raised the issue of whether Draf can be phosphorylated or activated by Drosophila Src at all . However , results from our genetic and biochemical analyses of Src64B and Draf activation in Drosophila contrast the conclusion concerning B-Raf activation and indicates that Src64B may directly activate Draf by phosphorylating Y510 , a conserved tyrosine residue located in the kinase domain of all Raf proteins . Since Y510 is conserved in all Raf proteins , but not in other kinases other than Ksr ( whose kinase domain is mostly similar to Raf in sequence ) , we propose that the equivalent of Y510 in Raf kinase domain may serve as a key residue mediating Raf phosphorylation and activation by Src . Genetic studies of Src64B suggest that different biological processes may require different threshold levels of Src64B function . Mutagenic studies intended to isolate loss-of-function alleles of Src64B have so far resulted in a few weak alleles that do not affect viability [36] . However , animals with reduced Src64B function , such as Src64BΔ17 homozygotes , are partially female sterile and are associated with defective ovarian ring canal morphogenesis [36] . We detected different degrees of disruption in RTK signaling in Src64B mutant flies only in conditions in which the RTK pathway is compromised , such as in combination with viable Draf mutants or Egfr heterozygotes . Ovarian ring canal morphogenesis probably requires the highest amount of Src64B and becomes the first process to fail when Src64B function is reduced . Interestingly , we also observed different threshold requirement for Src64B in different RTK signaling pathways . Among the best characterized biological processes that require RTK/Draf signaling , the dorsoventral polarity of the egg chamber appears the most sensitive; we found a 100% penetrant ventralization phenotype in Egfr/+; Src64BΔ17 flies . DrafSu2; Src64BΔ17 double homozygous flies also exhibit a 100% eggshell ventralization , and DrafSu2 homozygotes alone exhibit a certain degree of eggshell ventralization ( unpublished data ) . The SEV pathway appears less sensitive to reductions in Src64B/Draf activity , such that approximately 22% of the DrafSu2; Src64BΔ17 double homozygous flies are missing R7 . The least-affected RTK pathway is the Torso pathway , which utilizes the same Ras1/Draf signaling cassette to specify the embryonic terminal structures [33] . However , in none of the above mutant combinations did we detect defects in the Torso pathway ( unpublished data ) . Thus , genetic studies based on a partial loss-of-function Src64B allele , Src64BΔ17 , suggest a differential threshold requirement for Src64B function in the following biological processes: ring canal formation > egg chamber dorsoventral polarity > R7 specification > embryonic termini formation . Since DrafY510F poorly rescued the terminal structure in torso null embryos , tyrosine phosphorylation on Y510 could be conferred by protein kinases other than Src64B . It is possible that Src64B is functionally redundant with other cytosolic tyrosine kinases such as Src42A and Tec29A . This may explain the subtle phenotypes of Src64B mutant alleles . Src42A is another Src homolog in the fly genome . Outside of the Src family , Tec kinases are mostly homologous to Src kinases . In Drosophila , Tec29A has been identified that functions downstream from Src64B during oogenesis [36 , 53] . Moreover , Tec29A is also a potential Draf activator [30] . It is thus an interesting scenario that these three tyrosine kinases may function redundantly in phosphorylating and activating Draf . However , it has been shown that the kinase activity of Src42A is not crucial for mediating RTK signaling , as overexpression of either a wild-type or kinase-dead form of Src42A equally induces hyperactivation of RTK signaling [54] . It has recently been shown that a kinase-independent scaffolding function of Src42A regulates Draf by a novel CNK-dependent derepression mechanism , and Src64B does not share such a function [55] . Thus , Src64B and Src42A may have overlapping as well as distinct functions , and further investigation is required to determine how these tyrosine kinases are involved in Draf signaling .
The following strong or null alleles were used in this study: torsoXR1 [56] , Ras1ΔC40B [35] , Draf11−29 [57] , DrafC110 , DrafSu2 [50] , Src64BΔ17 , and Src64BPI [36] . Egfrf2 , sev-RasV12 ( on a TM3 balancer chromosome ) , Nanos-Gal4 ( on X ) , GMR-Gal4 ( on the second chromosome ) , and hsp70-Gal4 are from the Bloomington Drosophila Stock Center . Transgenic lines carrying hsp70-Src64Bact–expressing heat-shock–inducible , activated Src64B and Draf were previously described [31] . The dominant female sterile ( DFS ) technique [58] was used to generate germline clone ( GLC ) embryos that lack the maternal product of Ras1 and Draf . Embryos lacking the maternal torso product were produced by torsoXR1 homozygous females . Standard techniques were used to produce transgenic flies carrying different UAS-Draf variants in the pUAST vector [45] . His-tagged Draf and Draf-C ( a . a . 368–739 ) constructs were made by ligating Draf cDNA fragments to pQE vectors ( Qiagen ) . Mutations of Y510 and Y538 were introduced by PCR and verified by sequencing . Draf-N ( a . a . 1–367 ) and Draf-C were subcloned into the HA-pUAST and Flag-pUAST vectors . GST-tagged Src64Bact was made by ligating a Src64B cDNA fragment encoding a . a . 1–540 to the pGEX-KG vector . Non-tagged Src64Bact and full-length Draf variants were cloned into the pUAST vector . GST-tagged Dsor and the kinase-dead Draf ( K497M ) were generous gifts from E . Hafen and L . Ambrosio , respectively . V5-Draf and V5-Src64Bact were made by ligating PCR-amplified cDNA fragments to the pMT-V5 vector ( Invitrogen ) . Antibodies specific for phospho-MEK1/2 , MEK1/2 , and phosphotyrosine ( pY102 ) were from Cell Signaling Technology and were used at 1:1 , 000 dilution . Anti-Draf serum ( gift from D . Morrison ) , anti-Src64B ( gift from J . Cooper ) , mouse anti-V5 ( Invitrogen ) , mouse anti-Flag ( Sigma ) , and rat anti-HA ( Roche ) were used at 1:1 , 000 . Rabbit anti-V5 ( QED ) and goat anti-HA ( QED ) were used to immunoprecipitate tagged proteins . Rabbit polyclonal antibody against phospho-Y510 of Draf ( Y510 ) was produced by immunizing rabbit with the phospho-peptide CEGSSLpYKHVHVS , which represents the amino acid sequence around Draf Y510 ( underlined ) . Antibody production and affinity purification were carried out by Proteintech Group . To determine the effects of heat-shock expression of Src64Bact on embryogenesis and Torso signaling , 0–1-h-old embryos carrying one copy of hsp70-Src64Bact were collected and were allowed to continue development for an additional hour at 25 °C . They were then subjected to a brief heat shock at 34 °C for 5 min in water bath . Heat-shocked embryos were allowed to develop for 20 min . at 25 °C and then fixed , or allowed to develop for 24 h for cuticle examination . To examine the effects of Src64Bact during oogenesis , females containing one copy of hsp70-Src64Bact were heat shocked at 34 °C for 10 min in water bath , allowed to recover for 10 min at room temperature , and then dissected to fix the ovary . Detection of tll and kek mRNA expression by in situ hybridization was carried out as previous reported [8 , 47] . Drosophila Schneider L2 ( S2 ) cells were cultured at 25 °C in Drosophila Serum-Free Media ( SFM; Invitrogen ) supplemented with 10% Fetal Bovine Serum ( FBS; Invitrogen ) and 0 . 5× Antibiotic-Antimycotic ( Invitrogen ) . Cells were cultured at 2 . 5 × 106/ml prior to transfection . Transfection was performed with Cellfectin ( Invitrogen ) according to the manufacturer's instructions . An actin-Gal4 plasmid was used to induce expression of pUAST transgenes . To induce expression of genes cloned in pMT-V5 vector , Cu2SO4 ( Sigma ) was added to the medium at the final concentration of 0 . 5 mM 16 h after transfection , and cells were harvested 24 h after induction . S2 cells were harvested in the cell lysis buffer ( from Cell Signaling Technology ) . A total of 100 μl of embryos ( 0–4 h after egg laying ) of different genotypes were homogenized in 100 μl of 2× kinase buffer ( from Cell Signaling Technology ) . The lysate was cleared by centrifugation for 10 min at 4 °C . To immunoprecipitate Draf , 2 μl of anti-Draf antibody and 10 μl of Protein A-Sepharose ( Sigma ) were incubated with 100 μl of embryo lysates overnight at 4 °C . To measure Draf kinase activity , 50 μl of cell-free embryo extracts or immunoprecipitates were mixed with 2 μg of GST-Dsor and 200 μM ATP for 40 min at 30 °C . To deplete Draf from the lysate , 1 μl of anti-Draf antibody and 10 μl of Protein A-Sepharose ( Sigma ) were added to 100 μl of lysate and the mixture was incubated for 2 h at 4 °C in the presence of 1× protease inhibitor cocktail ( Sigma ) prior to centrifugation . To assess the effects of Src64B , Draf immunoprecipitates were incubated with 2 μg of GST-Srcact for 60 min at 30 °C in the presence of 200 μM ATP in 50 μl of kinase buffer . The reaction mixture was centrifuged , and the Draf immunoprecipitates were washed three times and were then subjected to in vitro Draf kinase assay . Draf kinase activity was detected as Dsor phosphorylation by anti-pMEK following SDS-PAGE . His-Draf and His-Draf-C variants , GST-Src64Bact , and GST-Dsor were purified from Escherichia coli BL-21 by standard affinity purification . To purify His-Draf variants ( the bulk of which was insoluble ) , E . coli cells were grown at 28 °C , protein expression was induced by IPTG when cell density reached an optical density ( O . D . ) of 0 . 8 , and cells were harvested 30 min following induction . Four liters of clear cell lysate ( containing a low concentration of soluble His-Draf ) was incubated with 2 ml of Ni-NTA agarose overnight at 4 °C in the presence of ß-mercaptoethanol . Soluble His-Draf protein was eluted from Ni-NTA agarose pellet with 8 ml of elution buffer , dialyzed , and further concentrated by using BIOMAX centrifugal filter ( Millipore ) to a 30-μl final volume . Substrate proteins were dephosphorylated by alkaline phosphatase ( Promega ) prior to kinase assays . Kinase assays for bacterial proteins were carried out by mixing 1 μg of each of the kinase and substrate into 50 μl of kinase buffer in the presence of 1 mM ATP at 30 °C for 6 h ( for Dsor as substrate ) or 15 h ( for Draf as substrate ) . mRNA microinjection was performed as previously described [8] . Draf mRNA was synthesized using the mMESSAGE mMACHINE T7 Kit ( Ambion ) and was injected from the posterior at 1 μg/μl into syncytial blastoderm-stage embryos of desired genotypes . The effects of expressed Draf protein on Torso signaling were analyzed with injected torso and Draf germline clone embryos . Protein sequence alignment of Raf family members were generated by the ClustalW program ( http://www . ebi . ac . uk/clustalw/index . html ) .
The FlyBase ( http://flybase . bio . indiana . edu ) accession numbers for the genes mentioned in this paper are as follows: Draf ( FBgn0003079 ) , Ras ( FBgn0003205 ) , Src64B ( FBgn0003501 ) , and Torso ( FBgn0003733 ) . The GenBank ( http://www . ncbi . nlm . nih . gov/Genbank ) accession numbers are as follows: Anopheles gambiae raf homolog ( EAA13186 ) , A-Raf ( TVHUAF ) , B-Raf ( TVHUBF ) , Btk29A ( AAF52631 ) , CG8789 ( AAF49129 ) C-Raf ( TVHUF6 ) , CTR1 ( AAA32779 ) , Draf ( AAF45774 ) , Egfr ( AAM70919 ) , ksr ( AAF52021 ) , LIMK1 ( AAF48176 ) , lin-45 ( AAA28142 ) , PhKgamma ( AAG22343 ) , and Src64B ( AAF47922 ) . The Protein Data Bank ( http://www . rcsb . org/pdb ) code of the B-Raf structure is 1UWH . | Receptor tyrosine kinase ( RTK ) /Ras signaling pathways control many different biological processes during metazoan development . Mutations that disrupt this signaling pathway cause many human diseases , including cancer . The proto-oncoprotein Raf functions downstream of Ras in transducing signals from RTK . Activating mutations in both Ras and Raf have been linked to many types of human cancers . Despite the importance of these oncoproteins in tumorigenesis , the molecular mechanisms of Raf activation remains unresolved . Here , using a genetic screen in Drosophila , we show that the Src homolog Src64B is an activator of Drosophila Raf ( Draf ) . Src64B phosphorylates tyrosine Y510 , in the Draf kinase domain and will activate a full-length Draf , but not a truncated Draf that contains only its kinase domain , suggesting that Y510 phosphorylation may relieve the autoinhibition of full-length Draf . Since Y510 is conserved among all the members of the Raf protein family , its phosphorylation may serve as a mechanism of Raf regulation in general . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods",
"Supporting",
"Information"
] | [
"cell",
"biology"
] | 2008 | Raf Activation Is Regulated by Tyrosine 510 Phosphorylation in Drosophila |
Sexual dimorphism in body weight , fat distribution , and metabolic disease has been attributed largely to differential effects of male and female gonadal hormones . Here , we report that the number of X chromosomes within cells also contributes to these sex differences . We employed a unique mouse model , known as the “four core genotypes , ” to distinguish between effects of gonadal sex ( testes or ovaries ) and sex chromosomes ( XX or XY ) . With this model , we produced gonadal male and female mice carrying XX or XY sex chromosome complements . Mice were gonadectomized to remove the acute effects of gonadal hormones and to uncover effects of sex chromosome complement on obesity . Mice with XX sex chromosomes ( relative to XY ) , regardless of their type of gonad , had up to 2-fold increased adiposity and greater food intake during daylight hours , when mice are normally inactive . Mice with two X chromosomes also had accelerated weight gain on a high fat diet and developed fatty liver and elevated lipid and insulin levels . Further genetic studies with mice carrying XO and XXY chromosome complements revealed that the differences between XX and XY mice are attributable to dosage of the X chromosome , rather than effects of the Y chromosome . A subset of genes that escape X chromosome inactivation exhibited higher expression levels in adipose tissue and liver of XX compared to XY mice , and may contribute to the sex differences in obesity . Overall , our study is the first to identify sex chromosome complement , a factor distinguishing all male and female cells , as a cause of sex differences in obesity and metabolism .
Obesity represents a risk factor for many types of metabolic disease , including diabetes , coronary heart disease , osteoarthritis , and even cancer . The study of rare mutations in humans and induced mutations in mouse models has identified numerous genetic factors that influence energy balance [1] , [2] , [3] . Less is known , however , about common genetic factors that may contribute to moderate differences in body fat storage among individuals in a population [4] . In humans and many other mammals , differences exist between males and females in the amounts and anatomical distribution of fat storage [5] , [6] , [7] , [8] , [9] . In general , males tend to have more visceral fat while females have more lower body and subcutaneous fat [10] . The two sexes also differ in the levels of adipose tissue-derived hormones leptin and adiponectin [11] , [12] , [13] , and in the response of fat store depletion to caloric restriction [14] . These differences may contribute to differences between men and women in susceptibility to metabolic disease . The genetic origins of sex differences in fat tissue accumulation are not well understood . Most studies have focused on the role of gonadal hormones ( reviewed in [15] , [16] . It is well established that the reduction in levels of estrogens , progestins , and androgens occurring at menopause in women increases central fat accumulation and risk for diabetes , cardiovascular diseases and other disorders [17] . Further evidence that estrogens play an important role in fat metabolism comes from mouse studies . For example , both male and female mice lacking estrogen receptor α have increased white adipose tissue mass and insulin resistance [18] . In men , the accumulation of excess abdominal adipose tissue is associated with low levels of gonadal androgens [19] . Hyperandrogenism is also associated with increased abdominal obesity in women with polycystic ovarian syndrome [20] . Androgen receptor-deficient male mice develop late onset obesity , particularly affecting visceral fat [21] , [22] . In addition , the administration of dihydrotestosterone suppresses the development of subcutaneous adipose tissue in wild-type but not androgen receptor-deficient mice [22] . Thus , gonadally derived hormones from both males and females influence body fat , albeit in distinct ways . Although gonadal hormones are a key determinant of sexual dimorphism in body fat and metabolism , it is notable that even prior to the differentiation of the gonads , human and mouse male embryos are larger than female embryos , suggesting that non-gonadal factors also contribute [23] , [24] . In addition to gonadal hormones , an additional fundamental genetic difference exists within every cell in the body of females compared to males ( reviewed in [25] , [26] , [27] , [28] . This is the presence in female cells of two X chromosomes , and in male cells of an X and a Y chromosome . The Y chromosome , and specifically the Sry gene located there , initiates differentiation of the testes . Mice that have a Y chromosome from which Sry has been deleted develop ovaries rather than testes . Conversely , an Sry transgene inserted onto autosome is sufficient to convert XX female mice to gonadal males [29] . Inactivation of one X chromosome in each non-germline XX cell greatly reduces the sex difference in level of expression of X genes that is predicted based on the number of copies of X genes [30] . However , a finite set of genes on both mouse and human X chromosomes escape inactivation , and would therefore be expected to exhibit higher expression levels in XX compared to XY cells [31] , [32] , [33] , [34] . Recent studies indicate that genes escaping X chromosome inactivation exhibit elevated expression in metabolic tissues from XX compared to XO mice [35] , and could potentially contribute to sex differences in metabolic phenotypes . In the present study , we utilize the Four Core Genotypes ( FCG ) mouse model to distinguish between the effects of gonadal sex ( testes or ovaries ) and sex chromosomes ( XX or XY ) on adiposity and related metabolic traits [25] , [26] , [27] , [36] . The FCG model allows the generation of gonadal male and female mice carrying either XX or XY sex chromosome complements . Specifically , the FCG Y chromosome sustained a mutation deleting the Sry gene ( yielding the “Y minus” chromosome , Y− ) , which is complemented in some mice by an Sry transgene located on an autosome . Mice having both the Y− chromosome and the Sry transgene will develop normally as fertile gonadal male mice . If these mice are bred to a normal female ( XX ) , four types of progeny are produced: female mice with ovaries and XY or XX sex chromosomes ( XYF and XXF , respectively ) , and male mice with testes and XY or XX sex chromosomes ( XYM and XXM , respectively ) . If differences in a trait of interest occur between the gonadal male mice ( XYM and XXM ) and gonadal female mice ( XYF and XXF ) , it is most likely related to differences in gonadal hormones , although the groups also potentially differ because of possible effects of Sry on non-gonadal tissues . By contrast , differences between XX and XY mice suggest a sex chromosome effect , likely directly caused by the difference in number of X or Y chromosomes . In our study , FCG mice were gonadectomized as adults to remove the acute sex differences resulting from gonadal hormones , and thereby uncover the contribution of sex chromosome complement . We found that gonadectomized XX mice of both gonadal sexes have two-fold increased adiposity compared to XY mice of either gonadal sex . Further genetic studies with mice carrying XO and XXY chromosome complements revealed that the difference is attributable to dosage of the X chromosome , rather than effects of the Y chromosome . These results demonstrate a fundamental difference in adiposity and metabolism conferred by genes on the sex chromosomes , and specifically implicate X chromosome genes as the direct cause of these differences . These results further suggest that X chromosome genes whose expression levels are influenced by dosage or parental imprinting are candidates for metabolic disease differences in men and women .
To determine whether sex chromosome effects contribute to sex differences in body weight and fat mass in adulthood , we examined these traits in C57BL/6 FCG mice ( XX gonadal females , XX gonadal males , XY gonadal females , and XY gonadal males ) . Mice were maintained on a standard chow diet with low fat content ( 5% by weight ) . At the time of weaning at postnatal day 21 , the four groups of FCG mice did not differ in body weight ( Figure 1A ) . By 45 days of age , gonadal males of either sex chromosome complement were approximately 20% heavier than gonadal females . At 75 days of age the gonadal males were 25% ( XX background ) or 28% ( XY background ) heavier than corresponding gonadal females ( Figure 1B , time 0 ) . Importantly , however , in these gonadally intact mice there was also a significant influence of sex chromosomes on body weight . At 75 days of age , XX mice were heavier than XY mice by 6 . 3% ( XX>XY gonadal males ) and 8 . 8% ( XX>XY gonadal females ) ( p<0 . 0001 ) ( Figure 1B , time 0 ) . Differences observed between male and female gonadally intact FCG mice can be attributed to either activational effects of gonadal hormones ( reversible effects caused by sex differences in on-going action of gonadal hormones ) or organizational effects ( long-lasting or permanent gonadal hormone effects exerted at an earlier stage of development ) . To distinguish between these alternatives , mice were gonadectomized at 75 days of age to remove activational effects of gonadal hormones . At the time of gonadectomy , male XY and XX mice had significantly higher body weight than female XX and XY mice , although XX mice of either gonadal type weighed more than XY mice , as described above ( Figure 1B ) . In the 4 weeks following gonadectomy ( GDX ) , the body weights of all genotypes converged , and differences between mice that were originally gonadal males and females disappeared ( Figure 1B ) . By 7 weeks , there emerged significant differences based on sex chromosome complement , with XX mice weighing more than XY mice ( p<0 . 000005; Figure 1 ) . At 10 months after GDX , the XX mice weighed 24% more than XY mice ( p<0 . 0001 , Figure 1A ) . In addition , XX gonadal females continued to weigh more than XX gonadal males despite the absence of gonadal secretions for 10 months ( Figure 1A , p<0 . 01 for XX female vs . XX male mice ) , suggesting an interaction between XX sex chromosome complement and long-acting ( organizational ) gonadal hormone effects ( interaction p<0 . 05 ) . Thus , the male-female difference in number of X chromosomes influences body weight in the opposite direction to the male-female difference in gonadal hormones . The increased body weight in XX compared to XY mice reflects a near doubling of the absolute fat mass as ascertained by NMR analysis of whole mice , with 88% higher fat mass in XX compared to XY mice ( Figure 1C ) . When expressed as a percent of body weight , XX mice had 50% higher proportional fat mass than XY mice ( p<0 . 00001 , Figure 1C ) . This dramatic difference in fat mass between XX and XY mice is particularly striking considering that the mice were fed a standard mouse chow diet with very low fat content . XX mice also had slightly higher lean body mass than XY mice ( Figure 1C ) . The increased total body adiposity of XX compared to XY mice was reflected in isolated fat pad mass ( Figure 1D; p<0 . 0005 for absolute fat pad mass , p<0 . 005 for mass relative to kidney; kidney weight did not differ among genotypes ) . Fat mass , percent lean mass , and fat pad mass all exhibited significant sex chromosome effects , and also significant interactions between sex chromosome and gonadal sex ( indicated in Figure 1A , 1C and 1D by ‘Int’ ) . In parallel with the increased adiposity , plasma leptin levels were elevated 2–3-fold in XX compared to XY mice ( p<0 . 00005 ) ( Figure 1E ) . Plasma leptin was also higher in females ( p<0 . 05 ) , but only in XX mice ( interaction p<0 . 05 ) . This suggests that long lasting gonadal effects , as well as genetic factors conferred by sex chromosome complement , directly or indirectly influenced leptin levels . Leptin mRNA levels in adipose tissue mirrored plasma leptin levels , with highest levels in XX mice ( p<0 . 0005 vs . XY mice ) , and significantly higher levels in mice that previously had ovaries rather than testes ( p<0 . 01 ) ( Figure 1E ) . To identify metabolic differences that could contribute to the increased adiposity of XX mice , we measured food intake , physical activity , and energy expenditure parameters while mice were housed in metabolic cages . We performed these studies at two ages: ( 1 ) at 4 weeks following gonadectomy , at which time the body weights for all four genotypes were similar and measurements were not complicated by differences in body weight or composition , and ( 2 ) at 10 months after gonadectomy , after body weight differences in XX vs . XY mice were pronounced ( see Figure 1B ) . At 4 weeks post-GDX , we detected a difference among the genotypes in food intake patterns monitored continuously throughout the circadian cycle . During the dark period when mice typically consume 70% of total calories [37] , gonadal female mice of both XX and XY chromosome complements consumed more than gonadal males ( p<0 . 05; Figure 2A ) . Since these measurements were made only 4 weeks after GDX , this may reflect lingering effects of gonadal secretions . However , during the daytime , food intake was significantly higher in XX females and males compared to XY mice ( p<0 . 01; Figure 2A and 2C ) . Since this difference occurred at an age when no differences exist in body weight , the increased daytime food intake is likely to contribute to subsequent divergence of body weight between XX and XY mice . At 10 months after GDX , the average absolute food intake for all genotypes was reduced compared to values at 4 weeks post-GDX , but no significant differences in food intake were observed among the four genotypes ( Figure 2B ) . Using indirect calorimetry , we detected significant sex chromosome effects on respiratory quotient ( RQ ) , a measure of the relative reliance on carbohydrate ( RQ = 1 ) and fat substrates ( RQ = 0 . 7 ) as metabolic fuel . At four weeks post-GDX , all mice exhibited the expected diurnal variation in RQ , with highest values in the dark phase . Notably , however , XX mice maintained a significantly higher RQ than XY mice during the light phase ( Figure 2D ) , which may be related to the differential feeding pattern in XX mice ( Figure 2A ) . In addition , compared to XY mice , XX mice exhibited a smaller amplitude change in RQ from dark to light periods ( Δ RQ ) , suggesting reduced flexibility in fuel switching ( Figure 2D ) . By contrast , at 10 months post-GDX , after XX mice had accumulated nearly twice as much adipose tissue as XY mice , the pattern of fuel utilization had changed . At this point , the XX mice had lower RQ than XY mice during the dark phase ( Figure 2E ) , indicating increased fat utilization in the fed state , possibly an adaptive change in response to the excess fat storage . Besides food intake and RQ , other energy balance parameters did not differ significantly among the four genotypes . These include oxygen consumption ( which was assessed per mouse , per lean body mass [38] , and via linear regression [39] to account for contributions of both lean and fat mass in energy metabolism ) , thermogenic gene expression , and physical activity in the horizontal and vertical planes ( Figure S1 ) . Thus , the key differences in energy metabolism between XX and XY mice were increased daytime food intake and reduced flexibility in RQ in XX mice . Both of these were apparent prior to the divergence in body weight . Despite the greater adiposity , XX mice did not exhibit substantially impaired glucose homeostasis . At four weeks after GDX , fasting glucose levels were higher in gonadal female than gonadal male mice ( p<0 . 0001 ) , and slightly higher in XX than XY mice ( p<0 . 05 ) ; there were no differences in fasting insulin levels among the genotypes ( Figure S2A ) . At ten months after GDX when XX mice had considerably greater adiposity , glucose and insulin levels were similar among the four genotypes , and no differences were revealed by glucose tolerance test ( Figure S2A ) . The ability to maintain glucose homeostasis despite excess fat storage in the XX mice may be related to adaptive changes in metabolism in these mice . For example , at ten months after GDX when XX mice had substantially higher fat mass , they exhibited increased expression of fatty acid oxidation genes encoding acyl CoA oxidase ( Aox1 ) and carnitine palmitoyl transferase ( Cpt1 ) in both muscle and liver ( Figure S2B ) . Increased fatty acid oxidation may reduce the extent of lipid accumulation in liver and skeletal muscle , and prevent impaired glucose homeostasis . As described above , on a chow diet containing minimal fat , XX mice accumulate excess adipose tissue without impaired glucose homeostasis . Metabolic dysregulation in human obesity typically occurs in the presence of a more stressful nutritional environment . We hypothesized that a combination of sex chromosome complement and a high fat diet may make XX mice more vulnerable to metabolic dysregulation than XY mice . To test this , we placed FCG on a high fat , simple carbohydrate diet that promotes weight gain [40] . Mice were gonadectomized at 75 days of age , continued on a chow diet for 4 weeks , and then fed the high fat diet for 16 weeks . As shown in Figure 3A , the mice of all four genotypes had similar body weight at the beginning of the high fat diet . However , within just 3 days of beginning the high fat diet , the XX and XY mice diverged , with significantly higher body weight in XX gonadal males and females than in the corresponding XY mice ( p<0 . 005 ) . XX mice continued to gain weight at an accelerated pace throughout most of the 16 weeks , and weighed about 15% more than XY mice at the end of the diet ( p<0 . 000005 ) . The enhanced weight gain on the high fat diet appeared to obscure the male-female difference in XX mice that was observed on the chow diet ( Figure 1B ) . NMR assessment of body composition showed that after 16 weeks on the high fat diet , XX mice had higher absolute fat mass than XY mice ( p<0 . 005 ) , but the increase in fat mass was not significant when expressed as a proportion of body weight ( Figure 3B ) . Nevertheless , the increased fat mass was reflected in elevated plasma leptin levels in XX compared to XY mice ( p<0 . 000001 ) ; leptin levels were also significantly higher in female vs . male mice ( p<0 . 01; interaction of sex by sex chromosome complement p<0 . 05 ) ( Figure 3C ) . Absolute lean mass was also increased in XX mice ( p<0 . 00005 ) , but not when expressed as a proportion of body weight ( Figure 3B ) . Thus , it appears that the greater increase in body weight observed in XX compared to XY mice on a high fat diet is attributable to increases in both fat and lean mass , and that XX mice exhibit increased absolute fat mass and circulating leptin levels . The analysis of tissue weights of mice after 16 weeks on the high fat diet revealed sex chromosome effects on the liver and adipose tissue depots . Absolute kidney weight did not differ among the four genotypes despite differences in body weight ( Figure 3D ) , and was used to normalize the weights of other tissues . Inguinal subcutaneous fat pads weighed more in XX compared to XY mice ( Figure 3D; absolute weight , p<0 . 0005; normalized to kidney , p<0 . 001 ) . We also detected a sex chromosome by gonadal sex interaction in inguinal fat pad weight when normalized to kidney weight ( p = 0 . 006 ) , suggesting a role for organizational hormone action in combination with XX or XY status in determining subcutaneous fat pad expansion on a high fat diet . Unexpectedly , the gonadal fat depot showed the opposite pattern , with slightly higher values in XY mice expressed both as absolute weight ( p<0 . 05 ) and normalized weight ( p<0 . 05 ) ( Figure 3D ) . These results indicate that distinct genetic and hormonal factors may influence the expansion of the gonadal and inguinal fat depots on a high fat diet . The high fat diet elicited formation of a fatty liver specifically in XX mice . XX mice exhibited a significant increase in liver weight , an abundance of lipid droplets , and increased triglyceride content ( p<0 . 0005 , XX vs . XY mice ) ( Figure 4A , 4B ) . The XX mice also exhibited evidence of reduced insulin sensitivity , as fasting insulin levels and HOMA were elevated 2-fold compared to XY mice in the presence of similar glucose levels ( Figure 4C ) . These metabolic disturbances were not associated with increased circulating triglyceride or free fatty acid levels in XX mice , which instead differed between gonadal males and females ( Figure 4D ) . This suggests that triglyceride and fatty acid levels are influenced by organizational hormone effects rather than sex chromosome complement , and are not likely an underlying factor in the development of the fatty liver in XX vs . XY mice . Gene expression in liver of mice fed the high fat diet showed enhanced expression of lipogenic factors , including the transcription factor peroxisome proliferator-activated receptor γ , and the triglyceride biosynthetic enzyme diacylglycerol acyltransferase 1 ( Figure 4E ) . Hepatic expression of genes encoding proteins involved in fatty acid uptake ( Cd36 ) , fatty acid synthesis ( fatty acid synthase ) , and fatty acid desaturation ( stearoyl CoA desaturase ) were not significantly different among the four genotypes ( data not shown ) . Despite the increased triglyceride accumulation , fatty acid oxidation gene expression was also elevated in XX compared to XY liver ( Figure 4F; p<0 . 005 ) . This pattern was also observed in XX mice fed the chow diet ( Figure S2B ) , and may represent an adaptive or compensatory response that prevents even more pronounced fat storage in XX mice . In contrast to liver , Aox1 and Cpt1b mRNA levels in muscle correlated with gonadal sex rather than sex chromosome complement ( Figure 4G ) . Metabolic gene expression is clearly under complex control , with sex chromosomes and gonadal sex effects having differing roles in specific tissues and conditions . Overall , our results reveal that the XX chromosome complement led to accelerated weight gain and less desirable metabolic profile than XY mice in response to a high fat diet . XX mice differ genetically from XY mice in both the dose of the X chromosome and in the absence of a Y chromosome . We analyzed body weight and fat mass in mouse strains with abnormal Y chromosomes that allow the dissection of effects of X and Y chromosome number . As described below , our results indicate that the XX vs . XY difference is caused by genes on the X chromosome and not the Y chromosome . We took advantage of mice carrying an unusual Y chromosome , Y* , that undergoes abnormal recombination with the X chromosome , producing mice with aberrant numbers of X or Y chromosomes [25] , [41] . Thus , by breeding XY* fathers , we obtain progeny with the following genotypes: XX , XXY* ( similar to XXY ) , XY* ( similar to XY ) , and XY*X ( similar to XO+an extra pseudoautosomal region , PAR ) ( see Table S1 ) . After gonadectomy at day 75 , mice with two X chromosomes ( XX and XXY ) had higher body weight ( p<0 . 000001 ) and fat mass ( p<0 . 0005 ) than mice with one X chromosome ( XY and XO+PAR ) ( Figure 5A , 5B ) . The presence of the Y chromosome appeared to have no effect . We conclude that the inherent genetic difference conferred by presence of two X chromosomes is responsible for the effects on body weight and adiposity . A potential mechanism underlying the observed effect of two X chromosomes on adiposity is the presence of a higher dose of X chromosome genes in XX compared to XY cells . Although X inactivation prevents most X genes from being expressed at higher levels in females , it is well established that a proportion of X chromosome genes in both mouse and human escape inactivation [31] , [32] , [33] , [34] . If genes that escape X chromosome inactivation are expressed at higher levels in metabolic tissues of XX than XY mice , they may contribute to the differences that we have observed between XX and XY mice . We evaluated the expression levels in adipose tissue depots and liver of the FCG mice for protein-coding genes that have been shown to escape X inactivation in an interspecific female mouse cell line , or are candidate “escapees” from X-inactivation because of higher expression in XX vs . XO mice , or XX vs . XY mice ( listed in Figure 6A ) [34] , [35] , [42] , [43] . We found that 8 of 11 genes in our panel exhibited higher expression levels in XX compared to XY mouse adipose tissues ( gonadal and/or inguinal ) ( Figure 6A ) . These include four genes that are established X escapees in both mouse and human ( Eif2s3x , Kdm6a , Ddx3x , Kdm5c ) , and these genes also show higher expression in gonadectomized XX liver as well as adipose tissue ( Figure 6A–6D ) . Another gene that is also known to escape inactivation in mouse and human , Mid1 , exhibited a unique expression pattern , with significantly lower expression in XX compared to XY inguinal fat and liver . The mechanism for this reduced expression in XX tissues is unclear , but nevertheless constitutes a difference that is determined by sex chromosome complement . Only a handful of genes ( Ddx3x , Uba1 , Mid1 ) showed significant differences in expression levels between gonadal female and male mice , which may reflect long-lasting effects of gonadal hormones on expression levels of these genes . These results reveal that a subset of genes escaping X inactivation are expressed at elevated levels in metabolic tissues of XX compared to XY mice . These genes represent candidates for future studies to identify the mechanism by which increased X chromosome dosage affects adiposity and metabolism .
Sexual dimorphism occurs in many fundamental metabolic processes , which likely influence the development of metabolic diseases . Understanding the sex-specific factors and pathways that promote or mitigate disease may lead to a better understanding of disease pathogenesis and useful interventions . The present results illustrate the complex interplay between several major classes of sex-specific factors that cause sexual dimorphism in obesity , and highlight the utility of the FCG model for investigating such interactions . For the first time , we demonstrate that sex chromosome complement , independent of gonadal sex , has substantial effects on body weight and adiposity in adult mice on a chow diet , and on the rate of weight gain in mice fed a high fat diet . We found that the increased adiposity observed in XX mice is attributable to the presence of two X chromosomes rather than to the lack of a Y chromosome . These results focus attention of future studies on a specific set of X chromosome genes that exhibit altered expression in metabolic tissues of XX compared to XY animals because of escape from X chromosome inactivation or sex chromosome-specific imprinting . The role of sex hormones in the determination of body weight and adiposity has been documented in many studies in humans and rodent models . For example , gonadally intact male mice have higher body weight , and exhibit more pronounced diet-induced weight gain , than females; this sex difference is reversed partially or completely by ovariectomy of female mice [44] . In humans , the loss of estrogens with menopause is associated with deposition of visceral body fat , and this effect can be ameliorated to some extent by hormone replacement therapy [45] , [46] , . Modulating testosterone levels also affects adipose tissue storage in healthy young men , with testosterone levels inversely correlated with adipose tissue mass [49] . Thus , it is clear that gonadal hormones play a strong role in determining sex differences in adiposity in mice and humans . However , few models have allowed the interrogation of potential genetic effects underlying sex differences independent of gonadal hormones . In our characterization of the FCG mice , body weight and/or metabolic traits were influenced by all three of the major classes of sex-biasing factors: activational ( acute ) hormonal effects , long-lasting ( organizational ) hormonal effects , and sex chromosome effects [25] , [28] . Several traits were influenced by interactions between two or more of these factors . At 75 days of age , gonadal males weighed 25–28% more than gonadal females , irrespective of their sex chromosome complement , suggesting that the sex difference is caused by gonadal secretions . That conclusion was confirmed because the sex difference disappeared by one month after gonadectomy . However , further analysis of the FCG model revealed that attributing sex differences in body weight solely to gonadal hormones would be a significant oversimplification . Prior to gonadectomy , XX mice weighed 6–9% more than XY mice , in both gonadal males and females . The XX vs . XY difference became dramatically larger after gonadectomy , with XX mice having up to 2-fold greater adiposity than XY mice . Layered on top of this was an effect of Sry ( likely mediated by long-lasting effects of the original gonadal hormones ) , as without their gonads , gonadal female XX mice lacking Sry had higher body weight , fat pad mass , and plasma leptin levels than gonadal male XX mice possessing Sry . The results indicate that although sex chromosome effects act in both intact and gonadectomized mice , gonadal hormones blunt the influence of sex chromosome complement , suggesting that the hormones may have different effects depending on the chromosomal sex of cells . Thus , understanding how males and females differ from one another is not simply a matter of studying an apparently dominant factor that causes the sex difference , but requires disentangling the interactive effects of several sex-biasing factors . The increased body weight of XX mice was preceded by increased food intake compared to XY mice; interestingly , this was restricted to the light portion of the diurnal cycle ( see Figure 2A ) . After differences in adiposity were established between XX and XY mice , food intake was not distinguishable , but leptin levels were elevated in XX mice , suggesting relative leptin resistance in the XX mice . Since there were no detectable compensatory changes in energy expenditure or physical activity in XX mice , this increase in food intake likely contributes to the increased body weight . The increased consumption of carbohydrates during the light period was reflected in slightly elevated RQ during the same period . This difference was evident even before the GDX XX mice had increased body weight . A trend toward increased food intake during the light period continued after the XX mice were substantially heavier ( at 10 months post-GDX ) , although it was no longer statistically significant . A recent study has shown that mouse food intake during the light period of the circadian cycle leads to greater weight gain than equivalent intake during the dark period , when mice typically consume the majority of their calories [50] . Many other studies have provided evidence that energy balance is tightly integrated with the circadian clock and that disruption of this cycle has detrimental effects on many aspects of metabolism [51] , [52] . Thus , a focus of future studies in the FCG model will be the investigation of whether sex chromosome complement influences regulation of the circadian clock and/or networks for food consumption and satiety . Sex chromosome complement had a key effect on the response to a high fat diet , with XX mice having an almost immediate divergence in weight gain from XY mice . An interesting finding was the greater expansion in the subcutaneous fat depot in XX mice , and greater increase in the gonadal fat depot in XY mice . It has been shown that women store a greater percentage of dietary fatty acids in subcutaneous adipose tissue than men [53] . Our observations in mice raise the possibility that sex chromosome complement may be a factor in determining the propensity to store fat in various anatomical depots . The high fat diet also led to the development of more pronounced metabolic dysregulation in XX mice , particularly fatty liver . Non-alcoholic fatty liver disease affects up to one-third of American adults , usually in association with obesity and insulin resistance [54] , [55] . The occurrence of fatty liver disease and its progression to cirrhosis , end-stage liver disease and hepatocellular carcinoma are influenced by many factors , including genetic factors . Our studies reveal that XX sex chromosome complement is one genetic factor that promotes development of fatty liver in mice . It is likely that the fatty liver in XX mice fed the high fat diet was influenced by risk factors such as increased adiposity and hyperinsulinemia . Interestingly , however , fatty liver did not parallel circulating triglyceride and fatty acid levels , which were more influenced by gonadal sex ( likely influenced by organizational effects of gonadal hormones ) rather than sex chromosome complement . In future studies , it will be interesting to determine whether sex chromosome complement also influences the propensity for progression of steatotic livers to cirrhosis , the basis of which is currently not understood . The sex chromosome effects reported here indicate that inherent sex differences in expression of X chromosome genes , stemming from the difference in number or parental imprint of X genes in XX vs . XY mice , contribute to sex differences in adiposity and metabolic disease . The sex chromosome effects are unlike typical linkage of genes to phenotype , because they are not caused by differences in the genetic sequence of the X chromosome , which was identical in all mice studied . Because X-inactivation effectively reduces the inherent bias toward higher expression of X genes in XX mice relative to XY mice , prime candidates for the genes causing the adiposity are those that escape X inactivation , or those that receive a parental imprint , leading to differential expression in XX compared to XY mice [28] . A significant proportion of X chromosome genes ( 15–25% ) are thought to escape X chromosome inactivation in humans [56] , and most of the genes escaping X inactivation in mice also escape in humans [34] . We tested expression levels of candidate genes that are known to escape inactivation in both mouse and human ( Eif2s3x , Kdm6a , Kdm5c , Ddx3x ) or have a Y paralogue with some evidence for higher expression in XX than XY mice and humans ( Usp9x , Uba1 ) [31] , [34] , [35] , [42] , [43] , [57] . Each of these genes was expressed at higher levels in XX than XY gonadal fat in gonadectomized mice , providing evidence that these genes escape inactivation in a metabolic tissue . Thus , these genes are candidates for those causing the XX–XY differences in physiology and adiposity reported here . Alternatively , differential expression of X chromosomes escapee genes could occur secondarily to differences in adiposity between XX and XY mice , in which case they may be downstream players in the observed metabolic differences . In addition to sex chromosome genes , autosomal genes that are differentially expressed in response to X chromosome gene dosage may contribute to the observed metabolic differences between XX and XY mice . It is known , for example , that the dysregulation of genes involved in mitochondrial metabolism and protein translation occurs in tissues of XX compared to XO mice , but the metabolic consequences are not known [35] . A reasonable question is whether these studies in the mouse have relevance to obesity in humans . Unusual numbers of sex chromosomes in human conditions such as Klinefelter ( XXY ) and Turner ( XO ) syndromes are associated with metabolic disease and/or adiposity [58] , [59] , [60] , [61] . However , in these diseases , endocrine abnormalities may contribute and are difficult to distinguish from the sex chromosome effects . The utility of our model is that it is genetically tractable in a way that human studies are not . Since fundamental genetic and metabolic processes are shared between mice and humans , we propose that the identification of X-linked genes that have a large impact on obesity in the mouse could lead to the discovery of novel mechanisms that impact obesity in humans . The increasing longevity of the human population means that the hypogonadal period may extend for up to half of a persons' lifetime , and the inherent genetic sex differences uncovered here may have important ramifications . Furthermore , since the gene content of the X chromosome is conserved in mouse and human , and several of the same genes escape inactivation in both species , there is hope that characterizing the action of X gene ( s ) in mouse will advance our understanding of human metabolic disease .
All animals were handled in strict accordance with good animal practice as defined by the relevant national and/or local animal welfare bodies , and all animal work was approved by the appropriate committee . All experiments in this paper were carried out with UCLA IACUC approval . “Four core genotypes” ( FCG ) mice were used . “Male” denotes a mouse with testes , and “female” denotes a mouse with ovaries . In these mice , the testis-determining Sry gene is deleted from the Y chromosome and inserted as a transgene onto an autosome [27] , [36] . Thus , gonadal type is no longer controlled by sex chromosome complement ( XX vs . XY ) , and the effect of sex chromosome complement on traits can be studied independent of the gonadal type of the mouse . For the present study , the FCG model was transferred to a C57BL/6J ( B6 ) genetic background by backcrossing male MF1 XY−Sry ( Y− chromosome denotes deletion of Sry; Sry denotes presence of the autosomal Sry transgene ) to B6 XX females for 13–14 generations . Four groups of mice are generated , XX and XY gonadal males ( XXM and XYM , carrying the Sry transgene ) , and XX and XY gonadal females ( XXF and XYF , without Sry ) . In all the FCG mice the Y− chromosome derives from strain 129 . Advantages and caveats in the use of FCG mice have been discussed [25] , [27] . Gonadectomy was performed at 75 days of age . Under isoflurane anesthesia , mice were given a subcutaneous injection of carprofen and the gonads were removed . Using aseptic procedures , gonads were exposed , clamped , ligated , and excised . Successful gonadectomy was confirmed at the time of euthanasia . Although no gonadal hormones are present in GDX mice , sex steroid hormones ( e . g . , androgens or estrogens produced de novo in adrenal , adipose tissue , or brain ) are probably present in mice after GDX . In one study , we compared mice born of XY* fathers , which have an aberrant Y chromosome that recombines abnormally with the X chromosome . The XY* males from strain B6Ei . LT-Y*/EiJ from the Jackson Laboratories were crossed with B6/J females for 2–3 generations , so that the mice were a mixture of C57BL/6J and C57BL/EiJ strains . In all case littermates were compared , so that the percentage of the two B6 parental strains was comparable across groups . We studied four different types of progeny of XY*: XX , XXY* , XY* , and XY*X . These mice are roughly similar to XX , XXY , XY , and XO+an extra pseudoautosomal region , respectively ( see Table S1 ) [41] . Gonadal males and females were housed in separate cages and maintained at 23°C with a 12∶12 light∶dark cycle . For studies using chow fed mice , mice were fed Purina 5001 chow diet ( approximately 5% fat , PMI Nutrition International , St . Louis , MO ) throughout their lifetime . For high fat diet treatment , mice were gonadectomized at 75 days of age and 4 weeks later were switched from chow to a high fat diet containing 35% fat , 33% carbohydrate ( Diet F3282 , Bio-Serve , Frenchtown , New Jersey ) . Fresh diet was added to cages twice per week . Animal studies were performed under approval of the UCLA Institutional Animal Care and Use Committee . DNA was extracted from tails using Chelex resin ( Bio-Rad , Hercules , CA ) . The genotype of mice was determined by PCR based on the presence or absence of Sry and of the X/Y chromosome paralogues Jarid1d/Jarid1c [41] . Ear fibroblasts from offspring of XY* mice were cultured and metaphase spreads were used to determine the sex chromosome status based on karyotype [41] . FCG and XY* mice were weighed on postnatal days 21 , 45 and day 75 and then gonadectomized ( GDX ) on day 75 . After GDX mice were weighed at weekly intervals . At various ages , body composition was determined with a Mouse Minispec apparatus ( Bruker Woodlands , TX ) with Echo Medical Systems ( Houston , TX ) software . This apparatus uses NMR spectroscopy for fat and lean mass measurements with coefficients of variation of <3% [62] . Correlation between NMR and gravimetric measurements is better than 0 . 99 . Eight calibrated Oxymax metabolic cages ( Columbus Instruments ) were used to detect numerous variables related to energy balance: food and water intake , horizontal and vertical physical activity , heat production , oxygen consumption , CO2 production , energy expenditure , and respiratory quotient ( RQ ) . The room housing the metabolic cages was kept very quiet to avoid stress or other interference with the activity of the mice . Mice were housed individually in the Oxymax metabolic cages from midday Friday to midday Monday , during which parameters were monitored dynamically at 20 min intervals . Mice had free access to water and food resented from a food hopper attached to a scale . Data for 3 full nights and 2 full days were analyzed . Baseline glucose and insulin levels were determined after a 4 . 5-hour fast ( 8:00AM–12:30PM ) . Glucose tolerance tests were performed after similar fast by injecting mice intraperitoneally with glucose ( 2 mg/g body weight ) and determining glucose levels ( using Lifescan OneTouch glucose meter ) at 15 , 30 , 60 and 180 minutes after injection [63] . Liver , quadriceps skeletal muscle , gonadal fat , inguinal fat and brown fat were dissected out rapidly , flash frozen in liquid nitrogen , and stored at −80°C . RNA was isolated from tissues using Trizol ( Invitrogen , Carlsbad , CA , USA ) and treated with RNase-free DNase ( Promega , Madison , USA ) to remove possible genomic DNA contamination . First-strand cDNA synthesis was generated by reverse transcription with SuperScript III RNase H-RT ( Invitrogen ) . Quantitative real time PCR ( n = 7–8 per genotype ) was performed on an ABI 7300 Sequence Detection system ( Applied Biosystems , Foster City , CA , USA ) using the SensiMixPlusSYBR Green & Fluorescein Master Mix Kit ( Quantace , USA ) . Two or three control genes were amplified as normalization controls: beta-2 microglobulin , TATA box-binding protein ( TBP ) , and BC022960 . Primer sequences for all genes assessed are listed in Table S2 . Cycling conditions were: 95°C for 10 min; 40 cycles of 95°C for 15 sec , 60°C for 30 sec and 72°C for 30 sec . assay contained a standard curve for the target gene and control genes with 4 serial dilution points of control cDNA: 50 ng , 10 ng , 2 ng and 0 . 4 ng . Dissociation curves were examined to eliminate the possibility of genomic DNA contamination . Groups were compared using two-way ANOVA ( NCSS 2001; Number Cruncher Statistical Systems , Kaysville , UT , USA ) with main factors of sex ( gonadal male vs . gonadal female , same as Sry present vs . absent ) and sex chromosome complement ( XX vs . XY ) . Sometimes a three-way repeated measures ANOVA was also applied with between factors of sex and sex chromosome complement , and within factors of gonadal status ( before vs . after GDX ) or age . Statistical analyses ( main effects of each of the two factors , or interaction of the two ) are presented if they were statistically significant , but usually not if they were not significant ( p>0 . 05 ) . Multiple regression analyses of energy metabolism data was performed with Stata Data Analysis and Statistical Software ( StataCorp LP , College Station , TX ) . | Differences exist between men and women in the development of obesity and related metabolic diseases such as type 2 diabetes and cardiovascular disease . Previous studies have focused on the sex-biasing role of hormones produced by male and female gonads , but these cannot account fully for the sex differences in metabolism . We discovered that removal of the gonads uncovers an important genetic determinant of sex differences in obesity—the presence of XX or XY sex chromosomes . We used a novel mouse model to tease apart the effects of male and female gonads from the effects of XX or XY chromosomes . Mice with XX sex chromosomes ( relative to XY ) , regardless of their type of gonad , had increased body fat and ate more food during the sleep period . Mice with two X chromosomes also had accelerated weight gain , fatty liver , and hyperinsulinemia on a high fat diet . The higher expression levels of a subset of genes on the X chromosome that escape inactivation may influence adiposity and metabolic disease . The effect of X chromosome genes is present throughout life , but may become particularly significant with increases in longevity and extension of the period spent with reduced gonadal hormone levels . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"animal",
"models",
"medicine",
"animal",
"genetics",
"model",
"organisms",
"genetics",
"genetics",
"and",
"genomics",
"biology",
"mouse",
"metabolic",
"disorders"
] | 2012 | The Number of X Chromosomes Causes Sex Differences in Adiposity in Mice |
The regulation of alternative splicing involves interactions between RNA-binding proteins and pre-mRNA positions close to the splice sites . T-cell intracellular antigen 1 ( TIA1 ) and TIA1-like 1 ( TIAL1 ) locally enhance exon inclusion by recruiting U1 snRNP to 5′ splice sites . However , effects of TIA proteins on splicing of distal exons have not yet been explored . We used UV-crosslinking and immunoprecipitation ( iCLIP ) to find that TIA1 and TIAL1 bind at the same positions on human RNAs . Binding downstream of 5′ splice sites was used to predict the effects of TIA proteins in enhancing inclusion of proximal exons and silencing inclusion of distal exons . The predictions were validated in an unbiased manner using splice-junction microarrays , RT-PCR , and minigene constructs , which showed that TIA proteins maintain splicing fidelity and regulate alternative splicing by binding exclusively downstream of 5′ splice sites . Surprisingly , TIA binding at 5′ splice sites silenced distal cassette and variable-length exons without binding in proximity to the regulated alternative 3′ splice sites . Using transcriptome-wide high-resolution mapping of TIA-RNA interactions we evaluated the distal splicing effects of TIA proteins . These data are consistent with a model where TIA proteins shorten the time available for definition of an alternative exon by enhancing recognition of the preceding 5′ splice site . Thus , our findings indicate that changes in splicing kinetics could mediate the distal regulation of alternative splicing .
Pre-mRNA splicing is catalysed by small nuclear ribonucleoprotein particles ( snRNP ) that recognise the splice sites on pre-mRNA and remove the introns with great precision . U1 and U2 snRNPs recognise the core motifs present at the 5′ and 3′ splice sites , respectively [1] . These core splice site motifs , however , contain only about half of the information required to define exon/intron boundaries [2] . Additional sequence elements can recruit regulatory RNA-binding proteins either to enhance or silence splice site recognition depending on their position relative to the splice sites [3] , [4] . T-cells intracellular antigen 1 ( TIA1 ) and TIA1-like1 ( TIAL1 , also known as TIAR ) are closely related RNA-binding proteins . They have three RNA recognition motifs ( RRMs ) and a carboxyl-terminal glutamine-rich region [5] , [6] . RRM2 is the major domain binding to uridine-rich sequences , RRM3 is thought to bind to RNA with no specificity , and RRM1 has no detectable RNA binding affinity in vitro [7] . Instead , RRM1 and the C-terminus interact with U1 snRNP to enhance its recruitment to the 5′ splice site of alternative exons [8]–[11] . TIA1 and TIAL1 are involved in multiple aspects of RNA metabolism . They are present in both the cytoplasm and the nucleus and shuttle between these two compartments in a manner that requires the RRM2 and RRM3 domains [12] , [13] . In the nucleus , TIA1 and TIAL1 regulate alternative splicing by binding to U-rich sequences adjacent to the 5′ splice site and recruiting U1-C to promote exon inclusion [8] , [10] , [11] , [14] , [15] . They also regulate the splicing of their own mRNAs , and the resulting two major isoforms have different splicing activity [16] , [17] . In the cytoplasm , TIA1 and TIAL1 function as translational silencers by binding to the 3′ untranslated region ( 3′ UTR ) of mRNAs [18] , [19] . They were also implicated in stress-induced translational silencing in stress granules [12] , [20] . In addition , TIA1 and TIAL1 were shown to promote apoptosis [21] , and depletion of both proteins promotes cell proliferation [22] . The role of cis-regulatory RNA motifs located close to alternative exons has been widely investigated , but recent studies suggest that distal regulatory motifs might also play an important role [4] , [23] , [24] . For instance , Nova1 and Nova2 proteins can silence inclusion of an alternative exon when binding downstream of the preceding exon [4] , [24] . In contrast , Nova proteins enhance inclusion when binding directly downstream of an alternative exon [4] , [24] . This suggested that the local and distal effects of Nova binding downstream of a 5′ splice site are reciprocal [25] . Since the function of TIA proteins in recruiting U1 snRNP to 5′ splice site is well characterised , these proteins offered a unique opportunity for a comprehensive study of the distal splicing effects of changes in 5′ splice site recognition . Ultraviolet ( UV ) -crosslinking and immunoprecipitation ( CLIP ) was first developed to identify RNA sites bound by the splicing regulators Nova1 and Nova2 in brain tissue [26] . The traditional CLIP cDNA library preparation protocol suffers from a potential loss of cDNAs due to truncation immediately before the “crosslink site , ” where at least one amino acid remains covalently attached after proteinase K digestion [27] . Therefore , we used a modified cDNA library preparation protocol that was recently developed ( iCLIP ) , which identifies truncated cDNAs by introducing the second adapter to cDNAs after reverse transcription [28] . In addition , iCLIP introduces a random DNA sequence ( barcode ) to cDNAs during reverse transcription to differentiate between unique cDNA products and PCR duplicates . Since the first nucleotide of resulting cDNA sequences most likely locates directly downstream of the crosslink site , iCLIP enables quantitative and high resolution analysis of protein crosslinking to the target RNAs [28] . Here , we used iCLIP to identify the RNA crosslink sites of TIA proteins . iCLIP showed a high density of TIA crosslinking in 3′ UTRs of mRNAs and in non-coding RNAs ( ncRNAs ) . Intronic TIA binding clusters were restricted to positions immediately downstream of 5′ splice sites . TIA binding at the 5′ splice site of an alternative exon and/or the preceding exon predicted its dual splicing effects . TIA binding enhanced inclusion of proximal upstream alternative exons and usage of upstream alternative splice sites but silenced distal downstream alternative exons if these lacked direct TIA binding . Interestingly , TIA proteins also regulated distal alternative 3′ splice sites , suggesting that by enhancing 5′ splice site recognition , they can indirectly silence downstream alternative exons .
iCLIP was used to identify the crosslink sites of TIA1 and TIAL1 in HeLa cells in a transcriptome-wide manner . Briefly , cells were UV-irradiated , lysed , and RNA was digested with RNase I to a size of approximately 40–100 nucleotides ( nt ) . The proteins were immunoprecipitated using TIA1 or TIAL1-specific antibodies , and the protein-RNA complexes were subjected to 5′ labelling using 32P-γ-ATP for visualisation after SDS-PAGE separation . The specificity of each antibody was determined by overexpression of TIA1 or TIAL1 in HeLa cells ( Figures 1A , 1B , S1A and S1B ) . TIA1 and TIAL1 antibodies each detected RNA-protein complexes of correct size only from UV-crosslinked cells and only if an antibody was used for immunoprecipitation . Furthermore , the signal decreased in TIA1/TIAL1 double knockdown ( KD ) cells and increased when the corresponding protein was overexpressed ( Figure 1A and 1B ) . There was a slight cross-reactivity of the TIA1 antibody to the TIAL1 protein ( Figures 1A and S1A , when TIAL1 was overexpressed ) . However , during immunoprecipitation , the TIA1 antibody mainly recognised the TIA1 protein , as no increase was seen when TIAL1 was overexpressed ( Figure S1B ) . To assess the RNA sequence specificity of TIA1 or TIAL1 without the use of antibodies , we also developed crosslinking and affinity purification ( iCLAP ) , a method to purify Strep/His double-tagged TIA1 and TIAL1 proteins using stringent affinity purification . This method circumvents any cross-reactivity of antibody that would identify the same crosslink sites for both proteins . TIA1 and TIAL1 with the Strep/His tag on the N- or C-terminus were overexpressed in HeLa cells . After UV crosslinking and RNase I digestion , the protein-RNA complexes were first purified with magnetic streptavidin bead before ligation to the 3′ RNA adaptor . Cobalt beads were then used to further purify the protein-RNA complexes under denaturing conditions ( 8 M urea , Figure S1E ) . iCLAP detected protein-RNA complexes only if cells were transfected with an appropriate construct , and no signal was detected in vector-transfected or non-crosslinked cells ( Figure S1F ) . In summary , the analysis of radioactive protein-RNA complexes indicated that both iCLIP and iCLAP isolated specific protein-RNA complexes without contamination from other co-purified proteins or RNAs . To amplify the co-purified RNAs , these were dephosphorylated and ligated to the 3′ RNA adaptor on beads during immunoprecipitation ( iCLIP ) or affinity purification ( iCLAP ) . After SDS-PAGE and nitrocellulose transfer , the protein-RNA complexes of 70–150 kDa were excised from the membrane ( Figure S1C ) and subjected to proteinase K digestion . The RNA was reverse transcribed with a primer complementary to the 3′ RNA adaptor , which contained a second half complementary to the 5′ Solexa sequencing primer separated by a BamHI digestion site . The cDNA was then self-circularised , digested with BamHI , giving a product with corresponding adaptors at both ends , and amplified by PCR ( Figure S1D ) . PCR products were sequenced using single-end 44 nt reads on the Illumina GA2 system . Three independent replicate iCLIP experiments and one iCLAP experiment were performed for both TIA1 and TIAL1 ( Table S1 ) . In total , 18 . 4 million iCLIP sequences were generated , 74% of which aligned to the human genome by allowing only single genomic hits and one nucleotide mismatch ( Table S2 ) . Unique cDNAs were identified based on random barcodes , and the crosslink site was mapped to the first nucleotide preceding the start of the cDNAs ( Figure 1C ) . Together , iCLIP produced 869 , 782 unique cDNA reads for TIA1 and 2 , 966 , 801 unique cDNA reads for TIAL1 ( Table S2 ) . The iCLIP no-antibody controls , performed in parallel with two of the iCLIP experiments , did not generate detectable PCR products . When submitted for sequencing , they generated 1 , 074 and 7 , 798 unique cDNAs mapping to the human genome ( Table S2 ) . Since TIA1 or TIAL1 iCLIP generated 100-fold more cDNAs than controls , we estimated that over 99% of cDNAs from the iCLIP experiment represent RNA sites specifically crosslinked to TIA1 or TIAL1 . The random barcode introduced into iCLIP cDNAs allowed us to analyse the distribution of TIA1 and TIAL1 on human RNAs in a quantitative and reproducible manner ( Figure S2 ) . Only 1 . 7% of cDNAs mapped in antisense orientation to annotated genes , confirming the high strand specificity of iCLIP . Only 10% of cDNAs mapped to intergenic regions ( Figure 2A ) . The highest cDNA density was seen in 3′ UTRs and ncRNAs , which together contained 22% of all cDNAs ( Figure 2A and 2B ) . 2 , 277 ncRNAs and 8 , 602 3′ UTRs had a higher cDNA density than the whole-genome average , and the cDNA enrichment correlated between TIA1 and TIAL1 iCLIP ( Pearson correlation coefficient r = 0 . 95 and r = 0 . 90 , respectively; Figure S3G and S3H ) . The ncRNA and 3′ UTR sites with the highest cDNA counts mapped to highly expressed RNAs such as tRNAs and histone mRNAs ( Figure S4B ) . Interestingly , cDNA enrichment in 3′ UTRs was 5-fold higher than in the coding sequence ( Figure 2B ) , in agreement with past findings that TIA proteins bind 3′ UTR to regulate translation [18] , [29]–[31] . Fifty-eight percent of TIA1 and 60% of TIAL1 cDNAs mapped to introns ( Figure 2A ) . 67 , 002 introns had a cDNA density higher than the whole-genome average , and the cDNA enrichment correlated between TIA1 and TIAL1 iCLIP ( r = 0 . 81; Figure S3I ) . The cDNA density in introns was on average 18-fold lower than in 3′ UTRs and ncRNAs ( Figures 2B and S3I ) . Past studies have shown that TIA1 and TIAL1 regulate alternative splicing of exon 6 of FAS mRNA [14] , [17] , [32] , [33] . Both TIA1 and TIAL1 crosslinked to previously characterised intronic binding sites in FAS pre-mRNA ( Figure S5A ) . Past studies suggested that TIA1 and TIAL1 have different RNA binding specificities [7] , [34] . However , the two proteins can regulate alternative splicing of the same exons [17] . We therefore analysed the in vivo RNA specificity of the two proteins using our iCLIP data . As a control , the iCLIP positions were randomised within the co-expressed genomic regions . The 21 nt sequence surrounding the crosslink sites was compared to randomised positions to identify the pentamers enriched at the TIA1 and TIAL1 crosslink sites . Pentamer enrichment in TIA1 and TIAL1 iCLIP data was highly correlated , and UUUUA and AUUUU were the two most common pentamers ( r = 0 . 99; Figures 2C and S3A , Table S3 ) . Comparing replicate iCLIP experiments of either protein verified the high reproducibility of the observed sequence specificity ( Figure S3C and S3D ) . Similarly , iCLAP experiments with both proteins were also highly correlated and were enriched for the same pentamers as iCLIP , independently supporting the determined sequence specificity of both proteins ( Figure S3B ) . These results demonstrated that TIA1 and TIAL1 share the same in vivo RNA binding specificity . Due to the high stringency of purification of protein-RNA complexes , iCLIP purified RNA sites that directly interact with TIA proteins . However , it is possible that some of these sites represent transient and low-affinity TIA-RNA interactions . To specifically analyse the high-affinity RNA binding sites , we determined clusters of TIA1 or TIAL1 crosslink sites with a maximum spacing of 15 nt containing a significant cDNA count when compared to randomised positions ( FDR <0 . 05 ) . This identified 12 , 048 TIA1 and 34 , 058 TIAL1 crosslink clusters . Uridine represented 82% of TIA1 and 75% of TIAL1 clustered crosslink sites ( i . e . , crosslink sites that located within these clusters ) and was also the most common nucleotide at all positions up to 10 nt away from these crosslink sites ( Figure S3E ) . This agreed with the past studies showing that TIA proteins bind to uridine-rich motifs [8] , [14] , [17] , [19] . To compare the overlap between binding of TIA1 and TIAL1 to the same sites , we analysed the proportion of crosslink clusters identified by both proteins . Eighty-three percent ( 10 , 021 / 12 , 048 ) of TIA1 crosslink clusters overlapped with a TIAL1 crosslink site , and 59% ( 20 , 047 / 34 , 058 ) of TIAL1 crosslink clusters overlapped with a TIA1 crosslink site . The overlap depended on the number of cDNAs that defined a cluster . Ninety-nine percent of the crosslink clusters that were defined by five or more TIA1 cDNAs overlapped with a TIAL1 crosslink site ( Figure 2D ) . We also assessed the distances between clustered TIA1 and TIAL1 crosslink sites . TIAL1 crosslink sites overlapped with TIA1 crosslink sites 15-fold more common than with randomised TIA1 iCLIP positions ( Figure S3F ) . These analyses indicated that the binding sites of TIA1 and TIAL1 largely overlap . Since both proteins showed a redundant binding behaviour , the iCLIP data of TIA1 and TIAL1 were merged to increase the reliability of cluster definition , which depends on the number of unique cDNA sequences . 46 , 970 crosslink clusters were identified in the joint TIA1/TIAL1 data with a maximum spacing of 15 nt ( FDR<0 . 05 ) . The cDNA counts within these crosslink clusters were highly correlated between TIA1 and TIAL1 data , indicating that they had similar affinity to their common RNA binding sites ( r = 0 . 85; Figure 2E ) . Furthermore , the cDNA counts were correlated even at the single-nucleotide level , as evident in the 3′ UTR of MYC mRNA , which is a functionally validated translational target of the TIA proteins ( r = 0 . 73; Figures 2F and S5B ) [30] , [35] . Taken together , iCLIP data suggested that TIA1 and TIAL1 have similar affinity for their common RNA binding sites . In order to identify the candidate positions where TIA proteins bind to regulate splicing , we compared the distribution of TIA1/TIAL1 iCLIP clustered crosslink sites close to constitutive and alternative cassette exons ( Figure 3A ) . Thirty-fold enrichment was seen at positions 10–28 nt downstream of exon/intron boundaries compared to the last 20 nt of both type of exons . Surprisingly , the constitutive and alternative exons showed a similar extent of crosslinking in this region . Approximately 5% of both types of exons contained a TIA crosslink cluster in this region . Since TIA proteins were previously described as splicing enhancers [8] , [14] , we hypothesised that TIA crosslink clusters could be used to predict proximal enhanced exons ( Figure 3B ) . Therefore , a regulatory logic was defined where TIA crosslink clusters 10–28 nt downstream of an alternative exon ( region α in Figure 3B ) predicted enhanced exon inclusion . This predicted 1 , 620 alternative cassette exons as enhanced by TIA proteins . Furthermore , since binding of Nova proteins downstream of the preceding exon could silence distal alternative exons [4] , we defined a second regulatory logic for distal silencing . TIA crosslink clusters 10–28 nt downstream of the preceding exon ( region β in Figure 3B ) predicted silenced alternative exon inclusion if TIA crosslink clusters were absent in the region α . This predicted 1 , 962 alternative cassette exons as silenced by TIA proteins . To assess the iCLIP predictions in an unbiased way , the splicing changes in TIA1/TIAL1 KD HeLa cells were analysed using a high-resolution splice-junction microarray . Microarray data were analysed with the ASPIRE 3 software [28] . The microarray detected splicing changes in 1 , 213 cassette exons ( |ΔIrank|≥1 ) , 46 of which were further assessed using reverse transcription and PCR ( RT-PCR ) and capillary electrophoresis ( Text S1 ) . RNA was isolated from cells treated with three different siRNA oligonucleotide pairs targeting TIA1 and TIAL1 , and with control siRNA oligonucleotides ( for knockdown efficiency , see Figure S6 ) . Primer pairs generated a PCR product for 40 of the 46 tested splicing events , with 30 detecting two or more splicing isoforms . Among these , RT-PCR validated 86 . 7% ( 26/30 ) of the splicing changes , confirming the high accuracy of the microarray data ( Figure S9A , Table S4 ) . We assessed the accuracy of iCLIP predictions by comparing them to the splicing changes identified by the microarray . Exons were divided into subsets according to the confidence of the detected splicing change ( ΔIrank ) , and the number of exons correctly predicted by iCLIP was identified in each subset . iCLIP predicted approximately 5% of false positives in control exons ( |ΔIrank|<0 . 1; Figure 3C ) . However , iCLIP predicted a significantly higher number of exons among those with a detectable splicing change ( p<0 . 05 , Fisher's Exact Test; Figure 3C ) . For these exons , iCLIP correctly predicted the direction of splicing change for 105 of 123 exons ( 85% ) , out of which 87 were enhanced ( ΔIrank ≥2 ) and 18 were silenced ( ΔIrank ≤−2 ) . For example , a TIA crosslink cluster downstream of exon 23a in NF1 pre-mRNA located to a previously described functional TIA binding site [36] , and it correctly predicted the enhancing effect of the TIA proteins ( Figure 3D ) . In contrast , a TIA crosslink cluster downstream of the exon preceding the alternative exon in LRRFIP2 pre-mRNA correctly indicated the TIA-dependent silencing ( Figure 3E ) . To assess whether TIA proteins bind at additional positions to regulate splicing , we comprehensively analysed the positions of TIA crosslinking in target RNAs with respect to the observed splicing changes ( Figure 4A ) . Each clustered crosslink site in an individual RNA was considered as one crosslink event . The RNA map showed an increase in the number of crosslink events downstream of the exons , confirming that the predictive code included the primary positions where TIA proteins regulate splicing . There was a decrease in the number of crosslink events downstream of the silenced exons , combined with a significant increase downstream the preceding exon ( Figure 4A ) . Similarly , mutually exclusive exons such as in FYN pre-mRNA displayed TIA crosslinking downstream of the enhanced exon but not in the vicinity of the silenced exon ( Figure S8A ) . Since silenced exons had a reduced proximal TIA binding compared to control exons , TIA binding at the preceding exon was the most likely cause for the silencing effect . The RNA map demonstrated a significant increase in the number of crosslink events downstream of the enhanced exons . To further validate the function of TIA binding at this position , we constructed a reporter minigene containing the alternative exon 5 and the flanking introns and exons of OGT pre-mRNA ( Figure 4B ) . In the iCLIP data , the only TIA crosslink cluster present in this region was located downstream of the 5′ splice site . Overexpression of either TIA1 or TIAL1 in HeLa cells increased exon inclusion , whereas TIA KD HeLa cells decreased exon inclusion ( Figure 4B ) . Overexpression of either TIA1 or TIAL1 in the KD cells was able to restore exon inclusion . This confirmed that both TIA1 and TIAL1 could enhance inclusion of the exon . To directly test whether the RNA sequence underlying the TIA crosslink cluster is necessary for the ability of TIA proteins to enhance exon inclusion , the 40 nt of intronic sequence downstream of the exon were replaced with a sequence from CDC25C pre-mRNA , which did not contain any TIA crosslink sites ( Figure 4B ) . Splicing of the mutant minigene did not change in response to the increased or decreased TIA protein levels ( Figure 4B ) . Thus , the TIA crosslink clusters located downstream of enhanced exons identified the RNA sites necessary to mediate the enhancing effect of TIA proteins . Extensive TIA crosslinking downstream of constitutive exons suggested that in addition to regulating alternative splicing , TIA proteins might also play a role in maintaining splicing fidelity . Microarray analysis detected increased retention of 143 introns and decreased retention of 102 introns in TIA KD cells . We tested 18 of these introns using real-time PCR with a 94% validation rate ( Figure S7B , Table S5 ) . An example of an intron retained in KD cells is shown in PIAA2 pre-mRNA , which contains TIA crosslink sites downstream of the 5′ splice site ( Figure S7A ) . The introns that were inefficiently spliced in KD cells had a significantly increased number of crosslink events downstream of the 5′ splice sites compared to control introns , indicating that TIA1 negatively regulates intron retention ( Figure 4C ) . This is consistent with a past study using the msl-2 reporter minigene , which showed that TIA1 binding to the uridine-rich track prevents intron retention [37] . Our results indicate that maintaining splicing fidelity at 5′ splice sites of constitutive exons is a widespread function of TIA proteins . In addition to regulating splicing of cassette exons , TIA proteins can also regulate the usage of alternative 5′ splice sites [38] . We therefore hypothesised that TIA crosslink clusters could be used to predict regulation of variable-length exons ( Figure 5A ) . Since the variable regions are often very short , precise identification of binding sites is crucial . To test whether the resolution of iCLIP was sufficient to resolve dual regulation of alternative 5′ splice sites , a regulatory logic was defined where TIA crosslink clusters 10–28 nt downstream of the intron-distal 5′ splice site ( position α in Figure 5A ) predicted silenced variable exons ( i . e . , increased usage of the intron-distal 5′ splice site ) . Conversely , TIA crosslink clusters 10–28 nt downstream of the intron-proximal 5′ splice site ( position β in Figure 5A ) predicted enhanced variable exons if TIA crosslink clusters were absent downstream of the intron-distal site . This logic predicted TIA-dependent silencing and enhancing for 84 and 172 variable-length exons , respectively . The microarray detected a splicing change in 213 variable-length exons in KD cells ( |ΔIrank|≥1 ) , 147 of which were a result of alternative 5′ splice site use . The accuracy of microarray data was assessed by analysis of seven alternative 5′ splice sites with a 100% validation rate ( Figures 5D , 5E , and S9B; Table S4 ) . iCLIP predicted approximately 3% of false positives in either direction among the control exons ( |ΔIrank|<0 . 1; Figure 5B ) . However , iCLIP predicted a significantly higher number of true positives among the exons that had splicing change in KD cells ( p<0 . 05 , Fisher's Exact Test; Figure 5B ) . Among these exons , iCLIP correctly predicted 18 out of 19 exons ( 95% ) , of which 8 were enhanced exons ( ΔIrank ≥2 ) and 10 were silenced ( ΔIrank ≤−2 ) . For example , a crosslink cluster downstream of the intron-distal alternative 5′ splice site was associated with silencing of the variable portion of exon 11 in CLIP4 pre-mRNA ( Figure 5D ) . In contrast , a crosslink cluster downstream of the intron-proximal alternative 5′ splice site was associated with enhanced inclusion of the variable portion of exon 33 in CHD9 pre-mRNA ( Figure 5E ) . To further assess the predictive value of iCLIP independently of the microarray , the nine alternative 5′ splice sites with the highest iCLIP cDNA counts at predictive positions were analysed by RT-PCR . RT-PCR detected alternative isoforms for five of these exons , and all of these showed a splicing change in KD cells ( Figure S9D ) . iCLIP correctly predicted the direction of splicing change for all of these exons ( Figures 5C ) . Furthermore , we comprehensively analysed the positions of TIA crosslink events in target RNAs with respect to the observed splicing change ( Figure 6A ) . In agreement with the predictive regulatory logic , there was a significant increase in the number of crosslink events downstream of the enhanced 5′ splice sites , but not at any other positions in the RNA map . To further verify TIA regulation of an alternative 5′ splice site , we constructed a reporter minigene containing the variable exon 33 and downstream intron and exon from CHD9 pre-mRNA ( Figure 6B ) . The only TIA crosslink cluster present in this region located downstream of the intron-proximal 5′ splice site ( Figure 5E ) . Overexpression of either TIA1 or TIAL1 significantly increased inclusion of the variable portion of the exon , whereas knockdown of the TIA proteins showed the opposite effect , which could again be rescued by TIA overexpression ( Figure 6B ) . Replacing 40 nt of intronic sequence downstream of the intron-proximal 5′ splice site with a sequence from CDC25C pre-mRNA rendered the minigene unresponsive to the changing TIA protein levels ( Figure 6B ) . This confirmed that the TIA crosslink clusters identified the RNA sites that mediated the effect of TIA proteins at the alternative 5′ splice sites . The microarray also detected a splicing change in 84 alternative 3′ splice sites , four of which were analysed by RT-PCR with a 75% ( 3/4 ) validation rate ( Figure S9C , Table S4 ) . The RNA map of TIA binding showed no enrichment of TIA crosslink events at the regulated alternative 3′ splice sites ( Figure 6C ) . Instead , there was an increase in TIA crosslink event 10–28 nt downstream of the preceding 5′ splice site if TIA silenced inclusion of the variable portion of the exon , as shown for C3orf23 pre-mRNA ( Figures 6C and S8B ) . Although the enrichment did not reach the significance level , these observations indicated that TIA proteins might silence the intron-proximal alternative 3′ splice sites by binding at the preceding 5′ splice site . In order to test this hypothesis , we constructed a reporter minigene containing variable-length exon 4 and the upstream intron and exon from C3orf23 pre-mRNA ( Figure 6D ) . The only TIA crosslink cluster present located downstream of the 5′ splice site ( Figure S8B ) . Since the intron has a size of more than 6 kb , we fused its first and last 600 nt to form the intron in the minigene . Overexpression of TIA1 or TIAL1 protein significantly decreased the inclusion of the variable part of the exon , whereas TIA knockdown significantly increased the inclusion ( Figure 6D ) . In order to verify functionality of this binding site , the 83 nt downstream of the 5′ splice site were replaced with the corresponding region from constitutive exon 2 of GAPDH pre-mRNA ( Figure 6D ) . This resulted in a loss of TIA regulation . Surprisingly , the mutation abolished inclusion of the variable portion of the exon , possibly due to the enhanced TIA-independent recognition of the GAPDH 5′ splice site . Taken together , the minigene reporter analysis confirmed that TIA binding downstream of the 5′ splice site can affect the usage of distal alternative 3′ splice sites .
TIA crosslink clusters were identified downstream of 5% of alternative and constitutive exons , suggesting that TIA proteins play a widespread role in 5′ splice site recognition . However , in spite of this widespread TIA binding , we were able to use consistent rules to predict the effects of TIA binding on splicing of regulated alternative exons . Assessing TIA crosslinking at the 5′ splice sites of both the alternative and the preceding exon enabled us to distinguish between silenced and enhanced exons . Splice-junction microarray analyses and minigene experiments validated the accuracy of these predictions on a genome-wide scale and at the level of individual regulated exons , supporting the finding that the position of TIA binding on pre-mRNA determines its dual splicing effects . The ability of iCLIP in predicting the dual TIA splicing effects is particularly noteworthy in the case of variable-length exons , since the distance between alternative splice sites is often very short , as shown in the example of CLIP4 pre-mRNA ( Figure 1C ) . The ability of iCLIP to directly identify crosslink sites and thereby resolve TIA binding at such proximal sites was crucial for the accuracy in separating silenced from enhanced variable-length exons . In the present study , we have evaluated both the local and distal splicing effects of TIA binding at 5′ splice sites . Locally , TIA proteins can regulate alternative 5′ splice sites in a manner consistent with the splice site competition model [41] . TIA proteins enhanced the closest upstream 5′ splice site , leading to a concomitant decrease in usage of the competing 5′ splice site ( Figure 7A ) . Similarly , TIA binding downstream of a cassette exon acts by promoting the usage of its 5′ splice site ( Figure 7B ) . A past study evaluating splicing intermediates showed that in cases of Nova binding downstream of enhanced exons , the downstream intron is removed prior to the upstream one [4] . This result is consistent with a splice site competition model , indicating that Nova and TIA proteins can enhance cassette exons by promoting the splicing pathway that uses the 5′ splice site of the alternative exon , with concomitant decrease in the exon skipping pathway that uses the 5′ splice site of the preceding exon ( Figure 7B ) . We also found that TIA binding can lead to distal splicing silencing effects . This supports the hypothesis of indirect silencing action , where an RNA-binding protein could cause a distal negative splicing effect by its local enhancing function , initially proposed to explain the same observation in the Nova RNA map [4] , [25] . A recent study of PTB-RNA interactions observed an opposite scenario where the local silencing function of PTB causes a distal positive splicing effect [23] . These findings point to the observation that a local splicing function of RNA-binding proteins generally leads to reciprocal distal splicing effects . The study analysing the distal effects of PTB proposed a model where competition between the constitutive and the alternative 5′ splice site was responsible for these distal effects [23] . To gain further insights into the distal regulation of alternative splicing , we analysed the effect of TIA binding on distal alternative splice sites . Surprisingly , we found that TIA proteins regulate usage of alternative 3′ splice sites without binding directly at the 3′ splice sites . This distal effect does not involve a competition between the constitutive and the alternative 5′ splice sites . Instead , TIA proteins regulate the distal alternative 3′ splice sites by modulating recognition of the upstream 5′ splice site ( Figure 7C ) . This result was also supported by the minigene experiment , which showed that a constitutive non-TIA dependent 5′ splice site promotes skipping of the variable portion of the distal exon . This suggests that regulation of a 5′ splice site recognition can affect splicing of downstream alternative 3′ splice sites even if it doesn't compete with another 5′ splice site . This effect could also contribute to the splicing regulation of distal cassette exons ( Figure 7D ) . Unlike Nova , FOX2 , and PTB proteins [4] , [23] , [24] , [42] , which bind at positions close to either 3′ or 5′ splice sites , TIA RNA maps did not identify significant binding at 3′ splice sites or within the alternative exons . Instead , TIA binding was enriched only downstream of the exons , where TIA proteins were reported to recruit U1 snRNP to the 5′ splice sites [9]–[11] . It is clear that the uridine-rich motifs downstream of exons are not the only determinant of TIA binding , since these motifs are present also in the polypyrimidine tracts upstream of exons . It is therefore possible that TIA binds to RNA cooperatively in complex with U1 snRNP , which would ensure TIA binding only downstream of exons . Interestingly , TIA crosslinking was equally common downstream of constitutive and alternative exons . Past studies found that uridine tracts are among the most enriched motifs downstream of constitutive exons , but the function of TIA binding to these motifs was not validated [15] , [43] . We found that TIA binding downstream of constitutive exons often promotes efficient splicing of the corresponding intron . Interestingly , this function is shared with the yeast orthologue Nam8p , suggesting that it represents a primary evolutionary function of the TIA proteins [44]–[46] . We find that TIA proteins can cause a distal splicing effect by regulating recognition of a constitutive 5′ splice site , even if this site does not compete with an alternative 5′ splice site ( Figure 7C ) . Several models of splicing regulation could account for this effect . TIA proteins might change the conformation of the U1 snRNP complex in a way that promotes its pairing with the intron-distal 3′ splice site . Alternatively , binding of TIA proteins could lead to a change in the long-range RNA-RNA interactions , or a change in interactions with other RNA-binding proteins that bind at a distal site . Finally , the result could also be explained in light of the splicing kinetics model . Kinetic parameters of splicing were shown previously to influence the splice site choice [41] . By enhancing U1 snRNP recruitment to the 5′ splice site , TIA proteins might allow the splicing reaction to proceed faster , thereby shortening the time available for definition of the downstream alternative exon . In contrast , the slower splicing kinetics in knockdown cells could allow additional time for trans-acting factors , such as SR proteins , to recognise exonic elements that prevent skipping of the alternative exon ( Figure 7C , D ) [47] . Such effects of splicing kinetics might be related to the transcriptional effects on splicing , which act partly by changing the ability of SR proteins to define the alternative exons [48] . Taken together , this study of TIA proteins highlights the importance of identifying the positions of protein-RNA interactions with high precision in order to reveal the full complexity of splicing regulation .
HeLa cells were irradiated with UV light . Upon cell lysis , RNA was partially fragmented using RNase I . For iCLIP , TIA1 or TIAL1 were immunoprecipitated with protein G Dynabeads ( Invitrogen ) conjugated to goat-anti TIA1 ( Santa Cruz , C-20 ) or goat-anti TIAL1 ( Santa Cruz , C-18 ) antibody . For iCLAP , the Strep/His-tagged proteins were affinity purified using Streptavidin and Cobalt beads . RNA was ligated at 3′ ends to an RNA adapter and radioactively labelled on beads . After gel electrophoresis and nitrocellulose membrane transfer , protein-RNA complexes were visualised by autoradiogram . RNA was recovered by proteinase K digestion and reverse transcribed using primers with adapter regions separated by a BamHI restriction site and a barcode region at their 5′ end . cDNA was size-purified , circularised , annealed to an oligonucleotide complementary to the restriction site , and digested with BamHI . Linearised cDNA was then PCR-amplified using primers complementary to the adapter regions and subjected to high-throughput sequencing using Illumina GA2 . A more detailed description is available in Text S1 . The sequences corresponding to the individual experiment were identified by their defined barcode , the random barcodes were registered , and the barcodes were removed before mapping the sequences to the human genome sequence ( version Hg18/NCBI36 ) , allowing one mismatch using Bowtie version 0 . 10 . 1 ( command line: -a -m 1 -v 1 ) . The randomisation was done within co-transcribed regions that were expected to have the same expression levels . For instance , a single gene contains abundant exonic and non-abundant intronic RNA , and intronic RNA contains non-coding RNA genes that are usually highly abundant . We have randomised positions within each individual intron , excluding the non-coding RNA genes . Each non-coding RNA gene was randomised separately . Exons were grouped into one single region for randomisation . However , as evident in Figure 2B , the 3′ UTR contains a higher cDNA enrichment than the coding sequence; therefore , we randomised positions in each UTR region separately from the coding sequence . All annotations were based on the version Hg18/NCBI36 of the human genome sequence . Each cDNA was considered independent when randomising the positions . Only crosslink positions ( but not the cDNA count ) were compared between the different datasets . Crosslink positions in the first dataset define the position of 0 when analysing the positions in the second dataset . For sequence analysis of iCLIP crosslink sites , the position of the crosslinking site was extended 10 nt in both directions . The z score for pentamer enrichment at the 21 nt region surrounding the crosslink sites was then calculated relative to randomised genomic positions . A more detailed description is available in Text S1 . This followed the same statistical approach as the analysis of CLIP sequence clusters [42] with a few modifications as described in [28] . Rather than combining crosslink sites into larger clusters , the crosslink sites within the clusters were kept individually in order to preserve the nucleotide resolution of the data . A more detailed description is available in Text S1 . Three different oligonucleotides were used together with a scrambled control ( Invitrogen , 12935-112 ) . The following siRNA oligonucleotides were used: siTIA1-1: 5′-GCAAGUUCCUGCAUAUGGAAUGUAU-3′ siTIA1-2: 5′-AGAAUAUCAGAUGCCCGAUGGUAA-3′ siTIA1-3: 5′-GGCAACAGGAAAGUCUAAGGGAUAU-3′ siTIAL1-1: 5′-CGGAUAUGGUUGGCAAGUUACCAA-3′ siTIAL1-2: 5′-CCGAACCAAUUGGGCCACUCGUAAA-3′ siTIAL1-3: 5′-GCGUCUGGGUUAACAGAUCAGCUUA-3′ 5 nM of each siRNA ( siTIA1 and siTIAL1 ) were transfected using Lipofectamine iMax ( Invitrogen ) according to manufacturer's instructions . Cells were transfected again 2 d after the first transfection and were harvested 2 d after the second transfection for protein and RNA analyses . The protein concentration was determined using Lowry's Assay ( Bio-RAD ) . Goat anti-TIA1 antibody ( 1∶1000 ) ( Santa Cruz , C-20 ) and anti-TIAL1 antibody ( 1∶1000 ) ( Santa Cruz , C-18 ) were used to detect TIA1 and TIAL1 protein . Rabbit anti-GAPDH ( 1∶5000 ) ( Cell signalling ) was used for loading control . For overexpression , mouse anti-Strep tag antibody ( 1∶1000 ) ( Qiagen ) was used to detect the tagged protein . Donkey anti-goat HRP , goat anti-rabbit HRP , and goat anti-mouse HRP ( Invitrogen ) were used as secondary antibodies . The membrane was visualised using ECL kit ( Amersham ) . A total of six samples were used , three from the siRNA control and three from KD3 . The high-resolution splice-junction microarrays were produced by Affymetrix , monitoring 260 , 488 exon-exon junctions ( each with eight probes ) and 315 , 137 exons ( each with 10 probes ) . cDNA samples were prepared using the GeneChip WT cDNA Synthesis and Amplification Kit ( Affymetrix ) . Analysis of microarray data was done using version 3 of ASPIRE ( Analysis of SPlicing Isoform Reciprocity ) . ASPIRE predicts splicing changes from reciprocal sets of microarray probes that recognise either inclusion or skipping of an alternative exon . In version 3 the background detection levels are experimentally determined for each probe , allowing background subtraction in a probe-specific manner [28] . Description of RT-PCR validation is available in Text S1 . RNA maps were produced by assessing the positioning of clustered crosslink sites at the exon/intron boundaries of alternative and flanking exons . For each exon , the positions between 20 nt of exonic and 60 nt of intronic sequence for each of the splice sites were analysed . When introns or exons were shorter than two times the length of the analysed area , analysis was restricted to region up the middle of the intron and exon . To draw the RNA map , the percentage of exons containing a crosslink site at the corresponding position is drawn . The alternative exons with the flanking constitutive exons were amplified by PCR with genomic DNA from HeLa cells and cloned into pcDNA3 vectors . The 40 nt downstream of the exon-intron boundaries where TIA binding site are present were mutated to sequences with no observed TIA binding sites . In the case of the enhanced cassette exon ( OGT1 ) and the alternative 5′ splice sites ( CHD9 ) , the 40 nt were replaced with sequences downstream of a silenced cassette exon ( CDC25C ) . In the case of the alternative 3′ splice sites ( C3orf23 ) , the 83 nt were replaced with sequences downstream of GAPDH exon 2 . The intron of C3orf23 was too long , so only 600 nt from the exon-intron boundaries on either side were cloned . The constructs were transfected together with GFP , TIA1 , or TIAL1-pcDNA3/Step-His vectors using polyfect ( Qiagen ) . The siRNA for TIA1 ( KD3 ) and negative control were transfected at the same time using Lipofectamine iMax ( Invitrogen ) . The cells were collected 2 d later and the splicing effects were assessed by RT-PCR using a T7 forward primer ( 5′-TAATACGACTCACTATAGGG-3′ ) and gene-specific reverse primers ( Table S4 ) . | Studies of splicing regulation have generally focused on RNA elements located close to alternative exons . Recently , it has been suggested that splicing of alternative exons can also be regulated by distal regulatory sites , but the underlying mechanism is not clear . The TIA proteins are key splicing regulators that enhance the recognition of 5′ splice sites , and their distal effects have remained unexplored so far . Here , we use a new method to map the positions of TIA-RNA interactions with high resolution on a transcriptome-wide scale . The identified binding positions successfully predict the local enhancing and distal silencing effects of TIA proteins . In particular , we show that TIA proteins can regulate distal alternative 3′ splice sites by binding at the 5′ splice site of the preceding exon . This result suggests that alternative splicing is affected by the timing of alternative exon definition relative to the recognition of the preceding 5′ splice site . These findings highlight the importance of analysing distal regulatory sites in order to fully understand the regulation of alternative splicing . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"molecular",
"biology/rna-protein",
"interactions",
"computational",
"biology/sequence",
"motif",
"analysis",
"computational",
"biology/alternative",
"splicing",
"molecular",
"biology/rna",
"splicing",
"biochemistry/bioinformatics",
"biochemistry/transcription",
"and",
"translation"
... | 2010 | iCLIP Predicts the Dual Splicing Effects of TIA-RNA Interactions |
Dengue is a mosquito-borne disease that is common in many tropical and subtropical areas . Dengue infections can occur at any age and time in the lifespan , including during pregnancy . Few large scale studies have been conducted to determine the risk of preterm birth ( PTB ) and low birthweight ( LBW ) for infants born to women who had symptomatic dengue infection during pregnancy . This study is a retrospective cohort study using medical records from 1992–2010 from pregnant women who attended a public regional referral hospital in western French Guiana . Exposed pregnancies were those with laboratory confirmed cases of dengue fever during pregnancy . Each of the 86 exposed infants was matched to the three unexposed births that immediately followed them to form a stratum . Conditional logistic regression was used to analyze these matched strata . Three groups were examined: all infants regardless of gestational age , only infants> = 17 weeks of gestational age and their strata , and only infants> = 22 weeks of age and their strata . Odds ratios were adjusted ( aOR ) for maternal age , maternal ethnicity , maternal gravidity , interpregnancy interval and maternal anemia . There was an increased risk of PTB among women with symptomatic dengue; ( aOR all infants: 3 . 34 ( 1 . 13 , 9 . 89 ) , aOR 17 weeks: 1 . 89 ( 0 . 61 , 5 . 87 ) , aOR 22 weeks: 1 . 41 ( 0 . 39 , 5 . 20 ) ) but this risk was only statistically significant when all infants were examined ( p value = 0 . 03 ) . Adjusted results for LBW were similar , with an increased risk in the exposed group ( aOR All infants: 2 . 23 ( 1 . 01 , 4 . 90 ) , aOR 17 weeks: 1 . 67 ( 0 . 71 , 3 . 93 ) , aOR 22 weeks: 1 . 43 ( 0 . 56 , 3 . 70 ) ) which was only statistically significant when all infants were examined ( p value = 0 . 05 ) . Symptomatic dengue infection during pregnancy may increase the risk of PTB and LBW for infants . More research is needed to confirm these results and to examine the role of dengue fever in miscarriage .
Dengue is a mosquito-transmitted viral infection that is common in most tropical and sub-tropical areas . Approximately 390 million dengue virus infections occur each year , and about 500 , 000 of these require hospitalization [1] , [2] . Concern regarding women who are pregnant becoming infected with dengue has been heightened in recent years due to an increase in adolescent and adult infections [3] , [4] . Currently , it is unclear if dengue infection in a pregnant woman results in serious health consequences for the mother or the child . Previous research has suggested higher proportions of preterm birth and low birth weight in infants born to mothers who had dengue during pregnancy [5]–[7] . However , many of the previous studies had small sample sizes , poor comparison groups , or other methodological problems [6] , [8] . The poor birth outcomes of preterm birth and low birthweight are associated with increased morbidity and mortality . By some estimates , 60% of all neonatal deaths occur to infants who are born preterm or low birthweight [9] , [10] . Preterm and low birthweight infants are also more likely to suffer from long term health consequences that can place a significant burden on hospitals , education systems , and the individual families of these infants [11] . This study calculates the risk of preterm birth or low birthweight infants in women who had symptomatic dengue infection during pregnancy , using a retrospective cohort study in French Guiana .
As this study collected personal information from French citizens , the requirements and recommendations of the Commission Nationale de l'Informatique et des Libertés were followed . The Tulane University Institutional Review Board approved this study as an exempt study for which informed consent was not sought from subjects . Informed consent was not sought for this study as all information was taken from the existing medical record , and data were analyzed anonymously . This study used medical archive data from the Franck Joly Hospital in St . Laurent du Maroni , French Guiana . French Guiana is an overseas department of France , with endemic and epidemic transmission of all four serotypes of dengue , as well as a high fertility rate [12]–[14] . Medical data from pregnant women who delivered at Frank Joly hospital during the years 1992–2010 were used in this study . Study subjects were defined as belonging to either exposed or unexposed groups . Exposed subjects were pregnant women who had a laboratory confirmed case of symptomatic dengue fever during pregnancy and who subsequently delivered their infants at Franck Joly Hospital . Many of the exposed subjects were included in previous case series describing the clinical effects of dengue fever in pregnant women [15]–[17] . Unexposed subjects were defined as pregnant women who did not have signs or symptoms of dengue during pregnancy , or who received a negative dengue test result if they were febrile . Unexposed subjects were identified either by using the delivery number of the infant ( for fetuses≥22 weeks as determined by ultrasound or≥500 grams ) , or by using the time of miscarriage of the fetus ( for infants <22 weeks of gestation and <500 grams ) . The three deliveries or miscarriages immediately following the birth of an exposed infant were used as the unexposed matches for the exposed infants . During the time period of this study there were different practices for recording miscarriages . After January 1st 1997 , miscarriages of≥17 weeks that occurred at the hospital were recorded in the hospital log books , which previously had only contained information on births≥22 weeks of gestational age . Due to differences in recording miscarriages over time , a sensitivity analysis was conducted . Three levels of inclusion were examined: all fetuses and infants regardless of gestational age and their matches , only fetuses and infants≥17 weeks of gestational age and their matches , and only fetuses and infants≥22 weeks of age and their matches . For the purpose of this study , a confirmed diagnosis of dengue was a symptomatic case of dengue fever accompanied by a positive test result from one of the following methods: IgM detection by ELISA , viral RNA detection via PCR , a positive NS-1 viral antigen test , or a positive viral culture . Test confirmation was conducted by the Pasteur Institute of Guiana , the national reference laboratory for arbovirus infections in French Guiana . The outcome definitions used in this study were as follows: a preterm birth was one <37 weeks of gestational age , including miscarriages . An infant with low birth weight was one born weighing <2 , 500 grams , irrespective of gestational age . Definitions of live birth , stillbirth , and miscarriage used in this study were based on the French definitions [18] . A stillbirth was defined as the birth of a dead infant who weighed≥500 grams or was≥22 weeks of gestational age . A miscarriage was defined as the birth of a dead fetus that was <22 weeks of gestational age and weighed <500 grams . None of the miscarriages included in this study were deliberately terminated pregnancies . Gestational age in this study was determined by ultrasound . Ultrasound information was missing in 5 cases , and gestational age was determined by the date of the last menstrual period ( LMP ) in two cases , by clinician estimate in two cases , and by both in one case . Information relating to dengue virus infection , gestational age , and birthweight was abstracted from patient medical files and the Obstetrics and Gynecology ward log books as well as information used to adjust for potential confounders included maternal age , maternal ethnicity , maternal gravidity , interpregnancy interval , and maternal anemia . Adjustment variables were chosen based on a review of the literature and on the information available in the medical archives . Ethnic information was classified by the investigators , based on information contained in the log books and in the patients' medical file . This information was included as previous studies have indicated that infants with African heritage are at increased risks of poor birth outcomes compared to infants of European heritage [19]–[21] . Maternal ethnicity was dichotomized into mothers of African heritage versus mothers of other ethnic backgrounds . Maternal gravidity was collected as a continuous variable and categorized into four groups ( 1 , 2–3 , 4–5 , >5 ) . Interpregnancy interval , the time between a prior pregnancy and the current pregnancy , was dichotomized into a short interpregnancy interval ( ≤18 months between deliveries ) versus a longer interpregnancy interval , or no interpregnancy interval due to being primigravid . Maternal age was categorized into four groups ( ≤20 , 20–29 , 30–35 , ≥36 ) with the reference group consisting of mothers 20–29 years of age . All data were abstracted from the medical records by an obstetrician trained in abstraction , using a standardized manual of procedures . As almost all infants had complete records , complete case analysis was used in all models . Statistical analysis for this study included univariate , bivariate , and multivariable investigations . Analyses included descriptive statistical measures for individual variables and tests of association for each variable with the outcomes under investigation . Conditional logistic regression was used to model multivariate associations , with each exposed infant and the three unexposed infants following it modeled as a matched stratum . For the outcome of preterm birth , only dengue infections occurring until the 37th week of gestation were considered . All statistical analyses were done using SAS version 9 . 2 ( Cary , North Carolina ) .
A total of 86 exposed infant records were eligible for use in this study and were included in analysis . A total of 281 unexposed infant records were identified as matches to the exposed births by delivery number or time of birth . Out of the 281 records of unexposed infants , , 258 were able to be located in the medical archives , and were matched to the exposed infants . One unexposed infant had information missing for several variables , and was not included in multivariate models . Maternal socio-demographic characteristics in the total study sample reflected the larger population of St . Laurent du Maroni , although differences between the exposed and unexposed groups were seen [13] , [14] , [22] . The exposed mothers were more likely to be of non-African heritage as compared to the unexposed mothers ( Table 1 ) . Exposed mothers were also more likely to be anemic and to require a cesarean delivery ( Table 1 ) . The exposed and unexposed mothers had similar age distributions , with a mean age of 26 . 6 years . Out of the 344 infants included in this sample , 10 . 5% were born preterm , 13 . 4% were low birthweight , and 3 . 8% were stillbirths . Stillbirths were more common among exposed pregnancies than among unexposed pregnancies , regardless of the inclusion or exclusion of miscarriages in the subject population ( Table 2 ) . Of the 53 total fetuses or infants who had one or more poor birth outcomes , 54 . 7% were both preterm and low birthweight , 32 . 1% were born low birthweight but were not preterm , and 13 . 2% were preterm but not low birthweight . All of these poor birth outcomes were more common among the exposed fetuses and infants ( Table 2 ) . Among the 86 dengue-exposed pregnancies , 53 . 5% of dengue infections occurred in the third trimester , 34 . 9% in the second trimester , and 11 . 3% in the first trimester ( Table 3 ) . The median gestational age at dengue onset was 29 . 5 weeks , with a range of 7 to 40 weeks of gestational age , and 69 dengue infections before 37 weeks gestation ( 80 . 3% ) . In dengue-exposed pregnancies , mothers were noted as being febrile at the time of delivery in 25 . 9% percent of cases , and had threatened preterm labor attributed to dengue in 13 cases ( Table 3 ) . Dengue tests were ordered for 37 newborns , with positive test results in 5 ( 13 . 5% ) . In the majority of both maternal and congenital confirmatory testing , IgM tests were used , followed by NS-1 tests ( Table 3 ) . Unadjusted odds ratios for preterm birth resulted in point estimates that showed an increased risk of preterm birth for women who had symptomatic dengue infections during pregnancy . Odds ratio point estimates ranged from 1 . 92 for all infants regardless of gestational age to 1 . 28 for models including infants≥22 weeks and their matches ( Table 4 ) . However , none of the unadjusted estimates had significant confidence intervals ( p values ranged from 0 . 10 to 0 . 61 ) ( Table 4 ) . In adjusted models , point estimates ranged from 3 . 34 for all infants regardless of gestational age to 1 . 41 for models restricted to infants≥22 weeks of gestational age and their matched strata . The adjusted odds ratio including all infants regardless of gestational age was significant ( aOR = 3 . 34 ( 1 . 13 , 9 . 89 ) ) ( p value = 0 . 03 ) ( Table 5 ) . Unadjusted odds ratios for low birthweight births showed point estimates that indicated increased risk for infants whose mothers had symptomatic dengue infection during pregnancy . Unadjusted point estimates ranged from 2 . 06 for all infants regardless of gestational age to 1 . 62 for models restricted to infants≥22 weeks of gestational age and their strata . Only the odds ratio including all infants regardless of gestational age reached statistical significance in unadjusted models ( p value = 0 . 04 ) ( Table 5 ) . After adjustment , point estimates once again showed an increased risk of low birthweight . Adjusted point estimates ranged from 2 . 23 for all infants regardless of gestational age to 1 . 43 for models restricted to infants≥22 weeks of gestational age and their matches . However , only the estimate for all infants regardless of gestational age was significant ( aOR = 2 . 23 ( 1 . 01 , 4 . 90 ) ) ( p value = 0 . 047 ) ( Table 5 ) .
This study found increases in the risk of preterm birth and low birthweight for infants whose mothers had symptomatic dengue during pregnancy . To our knowledge , the present study is the largest and most epidemiologically sophisticated analysis using individual level data to examine the relationship between dengue fever during pregnancy and poor birth outcomes . This study is also among the first to have examined both the magnitude and variation in the risk of poor birth outcomes while adjusting associations for confounders . Previous research on dengue during pregnancy has generally been in the form of case reports and case series , which sometimes have suggested higher numbers of preterm and low birthweight infants among women with dengue during pregnancy [23]–[27] and sometimes have not [28]–[31] . There have also been studies that have utilized ecologic data to investigate the effect of dengue during pregnancy . A recently published study conducted in Cayenne , French Guiana found an increased risk of preterm birth when dengue transmission was occurring locally during the first trimester of pregnancy , but no significant associations were seen between local dengue transmission and low birthweight infants [32] . In the present study we detected increased risks for low birthweight as well as preterm birth in infants whose mothers were infected with dengue . The infants in this study who had low birthweight were also likely to be preterm ( Pearsons r = 0 . 68 ) , suggesting that these infants were low birthweight due to a shorter duration of gestation , rather than impaired fetal growth in utero [33] , [34] . In the present study , the largest odds ratios and those that had significant confidence intervals were obtained when the outcomes of interest were examined using the entire population of infants . However , the group containing all infants is also subject to diagnostic bias . Miscarriages <17 weeks of gestational age are more likely to be reported among women who were already at the hospital with dengue symptoms when they miscarried . This diagnostic bias is expected to decrease as gestational age increases and as the miscarriages approach the gestational age of 22 weeks , the age at which mandatory reporting of fetal death begins . The differences in results between these categories suggest that the models including all infants and the models including infants≥17 weeks of gestational age were influenced by the inclusion of the 4 miscarriages at <22 weeks of gestation in the dengue-exposed group . In a recent study by Tan et al . , women who had miscarriages at <22 weeks of gestation were more likely to have a positive NS-1 or IgM test for dengue [26] , suggesting that it is also possible that the different results obtained in this study reflect actual differences in risk between miscarriage and preterm birth in women who had dengue infection during pregnancy . Symptomatic dengue infection results in a number of physiologic changes , some of which might result in the initiation of early labor . The immune response to dengue could promote preterm birth by inducing placental inflammation and trophoblast apoptosis , production of inflammatory cytokines and chemokines , or fever [35] . Previous research has shown that some of the cytokines and chemokines released during dengue fever , including Il-6 , IL-8 and IL-18 , are also seen in preterm delivery [21] , [36]–[40] . It is also possible that the presence of fever in response to dengue infection could promote early labor , although the evidence linking fetal loss to febrile episodes is mixed [41] , [42] . Several mechanisms have been proposed to explain elevated maternal temperature and fetal loss , including heat shock protein interaction causing damage to the placenta or fetus , and stimulation of uterine contractions [43]–[47] . Interestingly , while the overall study population resembled that of St . Laurent du Maroni , the exposed and unexposed groups differed in regard to ethnicity . Women of non-African descent were overrepresented among women who had dengue during pregnancy ( 82 . 35% of the exposed ) . This may indicate that particular populations are more likely to suffer from symptomatic dengue during pregnancy , or may simply indicate a greater willingness or ability of these women to seek medical treatment [48] , [49] . While the French medical system provides universal care , there are transportation barriers to accessing care in this part of French Guiana . The data used in this study were limited by several constraints . All information used in this study was abstracted from the existing medical record in the archives of the Franck Joly Hospital and was limited by the accuracy and completeness of the medical records . Most importantly , this study was limited by the number of dengue fever cases in the medical record , leading to a lower than desired sample size of exposed pregnancies . While there were few missing data among the variables used in this study , there were other possible confounders that were not examined due to complete absence of the data , in particular , maternal education or socioeconomic status and maternal housing . The time period of this study ( 1992–2010 ) encompasses a period with great changes in dengue diagnosis methods . It is difficult to predict how this may have impacted the ascertainment of exposure over time , as all of the diagnosis methods that were used to determine exposure ( IgM , viral RNA , NS-1 viral antigen test , and positive viral culture ) have benefits and detriments [50] , [51] . It is also the case that obstetrical practice has changed during this time period , and changes in the management of complications may have had an impact on the frequency of the outcomes of interest . This study considered infants as exposed only if their mothers had a symptomatic dengue infection during pregnancy . A large proportion of all dengue infections are asymptomatic , with as many as 90% of all dengue infections occurring without symptoms [52]–[55] . Due to the retrospective nature of this study , we were not able to examine the effect of asymptomatic dengue infection on poor birth outcomes . As testing for dengue was only done if clinically indicated , it is possible that the infants of women with asymptomatic dengue were mistakenly included in the unexposed group . The inclusion of asymptomatic dengue infections in the unexposed group would have either had no effect on our odds ratios , or would have biased our odds ratios towards the null , depending on whether misclassified infants experienced poor birth outcomes . Despite these limitations , the findings of the present study have great clinical significance for areas with dengue transmission . If dengue infection during pregnancy increases the risk of preterm birth and low birthweight by 40% , then implementation of mosquito avoidance measures during pregnancy should help to lower the risk . The Centers for Disease Control already recommends that pregnant women stay indoors during peak mosquito activity , wear protective clothing , and use insect repellant on clothing and sparingly on skin [56]–[58] . Additional modifications , including screen installation and removal of standing water can be used to reduce the transmission of dengue from Aedes mosquitoes to pregnant women in areas where dengue is endemic [59] . Findings from this study also suggest several possible areas of future research . A larger study examining the outcomes of preterm birth and low birthweight is necessary in order to confirm the results of the present study and to allow for more precise confidence intervals . Additional research on the effects of asymptomatic dengue infection on poor birth outcomes , as well as more research on the possible biologic mechanisms linking preterm labor and dengue , are needed i to clarify the relationship between dengue and poor birth outcomes . | Previous studies have reported that dengue fever during pregnancy may be related to preterm birth and low birthweight among infants . However , few studies have used an appropriate control group to compare the risk of these outcomes for infants whose mothers had dengue fever to infants whose mothers did not . We designed this study to provide information on the amount of risk ( odds ratios ) and the stability of this risk ( confidence intervals ) of being born preterm or with low birthweight to a mother with documented dengue infection during the pregnancy . In this study there was an increased risk among pregnant women with symptomatic dengue to deliver infants who are preterm or low birthweight , but both the amount of risk and the stability of this risk were affected by the inclusion or exclusion of miscarriages ( infants born before 22 weeks of gestational age ) This suggests that women who are pregnant should take extra precautions to avoid dengue infections during pregnancy , since it may cause an early delivery , or the birth of a small infant . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"plant",
"science",
"pediatric",
"epidemiology",
"infectious",
"disease",
"epidemiology",
"epidemiology",
"plant",
"pathology",
"biology",
"and",
"life",
"sciences"
] | 2014 | Symptomatic Dengue Infection during Pregnancy and Infant Outcomes: A Retrospective Cohort Study |
Genome-wide association studies ( GWAS ) have identified >300 loci associated with measures of adiposity including body mass index ( BMI ) and waist-to-hip ratio ( adjusted for BMI , WHRadjBMI ) , but few have been identified through screening of the African ancestry genomes . We performed large scale meta-analyses and replications in up to 52 , 895 individuals for BMI and up to 23 , 095 individuals for WHRadjBMI from the African Ancestry Anthropometry Genetics Consortium ( AAAGC ) using 1000 Genomes phase 1 imputed GWAS to improve coverage of both common and low frequency variants in the low linkage disequilibrium African ancestry genomes . In the sex-combined analyses , we identified one novel locus ( TCF7L2/HABP2 ) for WHRadjBMI and eight previously established loci at P < 5×10−8: seven for BMI , and one for WHRadjBMI in African ancestry individuals . An additional novel locus ( SPRYD7/DLEU2 ) was identified for WHRadjBMI when combined with European GWAS . In the sex-stratified analyses , we identified three novel loci for BMI ( INTS10/LPL and MLC1 in men , IRX4/IRX2 in women ) and four for WHRadjBMI ( SSX2IP , CASC8 , PDE3B and ZDHHC1/HSD11B2 in women ) in individuals of African ancestry or both African and European ancestry . For four of the novel variants , the minor allele frequency was low ( <5% ) . In the trans-ethnic fine mapping of 47 BMI loci and 27 WHRadjBMI loci that were locus-wide significant ( P < 0 . 05 adjusted for effective number of variants per locus ) from the African ancestry sex-combined and sex-stratified analyses , 26 BMI loci and 17 WHRadjBMI loci contained ≤ 20 variants in the credible sets that jointly account for 99% posterior probability of driving the associations . The lead variants in 13 of these loci had a high probability of being causal . As compared to our previous HapMap imputed GWAS for BMI and WHRadjBMI including up to 71 , 412 and 27 , 350 African ancestry individuals , respectively , our results suggest that 1000 Genomes imputation showed modest improvement in identifying GWAS loci including low frequency variants . Trans-ethnic meta-analyses further improved fine mapping of putative causal variants in loci shared between the African and European ancestry populations .
Obesity is a worldwide public health epidemic , with current US estimates of 37 . 9% obese and 7 . 7% morbidly obese adults [1] . Disparities in obesity rates , as well as rates of comorbidities and mortality , are evident across sex and racial/ethnic groups . Estimates from NHANES for 2013–2014 [1] show that obesity is more prevalent among African Americans ( 48 . 5% ) than among non-Hispanic Whites ( 37 . 1% ) . In addition , obesity rates are higher among African American women ( 57 . 2% ) than among African American men ( 38 . 2% ) . For comparison , the obesity rates in non-Hispanic Whites were 38 . 7% and 35 . 4% , respectively , for women and men . Genome-wide association studies ( GWAS ) in diverse populations have identified > 300 loci associated with measures of adiposity including body mass index ( BMI ) and waist-to-hip ratio ( adjusted for BMI , WHRadjBMI ) in populations of European [2–9] , African [10–12] , and East Asian ancestry [13–15] . The majority of associated variants are common ( MAF >5% ) with small effect size , and jointly explain only a fraction of the phenotypic variances [7–8] . It has long been hypothesized that low frequency ( MAF = 0 . 5–5% ) and rare ( MAF < 0 . 5% ) variants may also contribute to variability in complex traits . However , these variants are not well captured in previous GWAS imputed to the HapMap reference panel [16–17] . The availability of higher density reference panels such as the 1000 Genomes Project ( 38M variants in 1092 individuals from phase 1 ) [18] has demonstrated improved imputation quality in European populations particularly for low frequency variants ( aggregate R2 ~0 . 6 for MAF = 0 . 5% ) . However its impact is less clear for non-European populations [19] . We took this opportunity to use higher density imputation to reevaluate our previous GWAS for associations with anthropometric traits in individuals of African ancestry ( AA ) including African Americans and Africans . The African Ancestry Anthropometry Genetics Consortium ( AAAGC ) previously identified seven genome-wide significant loci for BMI in up to 71 , 412 AA individuals , and an additional locus when combined with European ancestry ( EA ) data from the Genetic Investigation of ANthropometric Traits ( GIANT ) consortium using GWAS imputed to the HapMap Phase 2 reference panel [11] . No genome-wide significant loci were identified for WHRadjBMI in a GWAS of up to 27 , 350 AA individuals [12] . The low yield of discovery in AA studies is likely due to their relatively smaller sample sizes in comparison to EA studies [7–8] , as well as their lower degree of linkage disequilibrium ( LD ) and thus poorer imputation quality . Here , we extended our previous work in the AAAGC to perform meta-analyses and replication of GWAS imputed to the 1000 Genomes reference panel in up to 52 , 895 AA individuals for BMI and up to 23 , 095 AA individuals for WHRadjBMI . We aimed to 1 ) discover novel variants , 2 ) fine map established loci , and 3 ) evaluate the coverage and contribution of low frequency variants in genetic associations in AA populations .
We conducted sex-combined and sex-stratified meta-analyses of GWAS summary statistics across 17 studies for BMI ( N = 42 , 752 ) and 10 studies for WHRadjBMI ( N = 20 , 384 ) in AA individuals in stage 1 discovery ( S1 and S2 Tables , S1 Fig ) . Missing genotypes in individual studies were imputed to the 1000 Genomes Project cosmopolitan reference panel ( Phase I Integrated Release Version 3 , March 2012 ) [18] using MaCH/minimac [20] or SHAPEIT2/IMPUTEv2 [21–22] ( S3 Table ) . Among all variants with MAF ≥ 0 . 1% in the largest Women’s Health Initiative ( WHI ) study , the average info score was 0 . 81 and 90 . 5% had imputation info score ≥ 0 . 3 ( S4 Table ) . Genomic control corrections were applied to each study and after meta-analysis ( λ = 1 . 07 for BMI , 1 . 01 for WHRadjBMI ) ( S3 Table , S2–S5 Figs ) . Association results for ~18M variants for BMI and ~21M variants for WHRadjBMI were subsequently interrogated further . From stage 1 meta-analyses , variants associated with BMI ( 3 , 241 in all , 1 , 498 in men , 2 , 922 in women ) and WHRadjBMI ( 2 , 496 in all , 1 , 408 in men , 2 , 827 in women ) at P < 1×10−4 were carried forward for replication in AA and EA . Stage 2 included 10 , 143 AA ( 2 , 458 men and 7 , 685 women ) for BMI and 2 , 711 AA ( 981 men and 1 , 730 women ) for WHRadjBMI analyses . Stage 3 included 322 , 154 EA ( 152 , 893 men and 171 , 977 women ) for BMI and 210 , 086 EA ( 104 , 079 men and 116 , 742 women ) for WHRadjBMI analyses by imputing HapMap summary statistics results [7–8] to 1000 Genomes [23] ( S1 Fig ) . Meta-analyses were performed to combine either sex-combined or sex-specific results from AA ( stages 1+2 , N ≤ 57 , 895 for BMI , ≤ 23 , 095 for WHRadjBMI in sex-combined analyses ) and both AA and EA ( stages 1+2+3 , N ≤ 380 , 049 for BMI , ≤ 233 , 181 for WHRadjBMI in sex-combined analyses , S6–S9 Figs ) . Variants that reached genome-wide statistical significance ( P < 5×10−8 ) were assessed for generalization of associations with BMI to children in two additional AA cohorts ( N = 7 , 222 ) . Among the locus-wide significant established loci ( 44 for BMI given two of 45 lead regional variants were identical in two loci , and 21 for WHRadjBMI ) , and novel loci ( three for BMI and six for WHRadjBMI ) derived from the sex-combined and sex-stratified analyses , we performed fine mapping to localize putative causal variants . We constructed 99% credible sets containing variants that jointly account for 99% posterior probability of driving the association in a locus using the corresponding sex-combined or sex-stratified meta-analysis results from AA , EA and combined ancestry ( S13 Table ) . A smaller number of variants in a credible set represent a higher resolution of fine mapping and we considered a credible set containing ≤ 20 variants as “tractable’ for follow up . The credible sets in the EA analyses were generally smaller than those in the AA given their larger sample size . As compared to the EA analyses , the number of tractable loci in the meta-analyses of AA and EA increased from 23 to 26 for BMI , and from 14 to 17 for WHRadjBMI . Among these 43 tractable loci , the lead variants in the combined ancestry analyses had posterior probability ≥ 0 . 95 in six BMI loci ( SEC16B , TLR4 , STXBP6 , NLRC3 , FTO and MC4R ) and seven WHRadjBMI loci ( DCST2 , PPARG , ADAMTS9 , SNX10 , KLF13 , CMIP and PEMT ) ( S13 Table ) . Functional characterization of variants within the tractable credible sets revealed two loci contain nonsynonymous variants ( ADCY3: rs11676272 S107P; SH2B1: rs7498665 T484A from the ATP2A1 locus ) , but they had low posterior probability to drive the respective associations ( 0 . 02 and 0 . 15 , respectively ) ( S14 Table ) . On the other hand , the ADCY3 non-coding variants rs10182181 and rs6752378 had higher posterior probability ( 0 . 26–0 . 72 ) and are cis-eQTLs of ADCY3 and nearby genes . Several BMI loci including MTCH2 , MAP2K5 , NLRC3 and ATP2A1 , and WHRadjBMI loci including TBX15-WARS2 and FAM13A , also contained cis-eQTL variants regulating nearby gene expression in subcutaneous and/or visceral adipose tissue ( S14 Table ) .
In our large-scale meta-analyses of GWAS in up to 52 , 895 and 23 , 095 individuals of African ancestry for BMI and WHRadjBMI , respectively , we identified three novel ( IRX4/IRX2 , INTS10/LPL and MLC1 ) and seven established ( SEC16B , TMEM18 , GNPDA2 , GALNT10 , KLHL32 , FTO and MC4R ) BMI loci , as well as three novel ( TCF7L2/HABP2 , SSX2IP and PDE3B ) and one established ( ADAMTS9-AS2 ) WHRadjBMI loci in either sex-combined or sex-stratified analyses . By employing a recently developed method [23] to impute European GWAS summary statistics to the denser 1000 Genomes reference panel , followed by meta-analyses of both African and European ancestry individuals , we also identified three additional novel loci ( SPRYD7/DLEU2 , CASC8 and ZDHHC1/ HSD11B2 ) for WHRadjBMI . While all lead variants from established loci are common ( MAF ≥ 5% ) , four of the nine lead variants from novel loci were low frequency ( 0 . 5% ≤ MAF < 5% ) . In addition , the lead variants from established loci including TMEM18 and ADAMTS9-AS2 were absent in HapMap . Overall , these results suggest the deeper genome coverage and/or improved imputation quality using 1000 Genomes , and complemented with additional sex-stratified analyses , facilitate the discovery of novel loci and identification of variants with stronger effects in established loci . Among the novel sex-specific BMI loci ( IRX4/IRX2 , INTS10/LPL and MLC1 ) , we did not identify any putative coding variants or regulatory regions underlying our association signals . Additionally , no associations have been reported with other metabolic traits in these novel BMI-associated signals . The first lead variant rs112778462 is located between the IRX4 and IRX2 genes which are members of the Iroquois homeobox gene family . IRX2 expression has been associated with deposition of fat in the subcutaneous abdominal adipose tissue but no sex difference was observed [29–30] . Irx4 knock out mice demonstrated cardiomyopathy with compensated increased Irx2 expression [31] . The second lead variant rs149352150 is located between the INTS10 and LPL genes . LPL encoded lipoprotein lipase is expressed in several tissues including adipose to mediate triglyceride hydrolysis and lipoprotein uptake . The serum LPL mass [32] and LPL activity and fat cell size of adipose tissues at gluteus and thigh [33] have been reported to be higher in women than in men . Previous GWAS demonstrated association of LPL with triglycerides and HDL cholesterol [34–35] . However , the reported lead variant rs12678919 was not in strong LD with rs149352150 ( r2 = 0 . 005 in AFR and 0 . 006 in EUR ) . The third lead variant rs56330886 is located in a gene-rich region on chromosome 22q13 including MLC1 . No biological candidates are identified in this region , therefore further analyses may be needed to explain the causative mechanism for this association signal . Among the novel WHRadjBMI loci , rs116718588 is located between TCF7L2 and HABP2 . TCF7L2 is the most significant type 2 diabetes locus in African Americans [36] and other populations [37] . However , rs116718588 was not in LD ( r2 < 0 . 01 in AFR ) with the reported type 2 diabetes associated variants . The second lead variant rs2472591 is located near SPRYD7 , DLEU2 and TRIM13 . This locus was associated with height in previous GWAS [6] , but rs2472591 was not associated with height in our study ( P > 0 . 05 ) , suggesting different variants in this locus regulate different measures of body size . In addition , a surrogate of rs2472591 , rs790943 , is a cis-eQTL for TRIM13 [26] suggesting it may be the target gene . TRIM13 encodes an E3 ubiquitin-protein ligase involved in endoplasmic reticulum-associated degradation . The third lead variant rs140858719 is located between SSX2IP and LPAR3 . LPAR3 is a plausible candidate as it encodes a receptor for lysophosphatidic acid ( LPA ) . The autotaxin/LPA pathway mediates diverse biological actions including activation of preadipocyte proliferation [38] , suppression of brown adipose differentiation [39] , and promotion of systematic inflammation [40] which lead to increased risk for cardiometabolic diseases including obesity and insulin resistance [41–42] . LPA receptor 1 which is highly expressed in adipocytes and the gut primarily mediates these effects [43] . It has also been reported that LPA , via LPA1 and LPA3 receptors , mediated leukocytes recruitment and pro-inflammatory chemokine secretion during inflammation [44] . The fourth lead variant rs185693786 is located at intron 2 of PDE3B . The association signal spanned a large genomic region and harbors GWAS loci for adiponectin and height . Phosphodiesterase 3B is critical for mediating insulin/IGF-1 inhibition of cAMP signaling in adipocytes , liver , hypothalamus and pancreatic β cells [45] . Pde3b-knockout mice exhibited multiple alterations in regulation of lipolysis , lipogenesis , and insulin secretion , as well as signs of peripheral insulin resistance [46] . PDE3B expression has been reported to be higher in microvascular endothelial cell culture derived from skeletal muscles from male rats than in female rats [47] . The fifth lead variant rs6499129 is located intergenic between ZDHHC1 and HSD11B2 . HSD11B2 encodes 11β-hydroxysteroid dehydrogenase type 2 which converts the active glucocorticoids to inactive metabolites . HSD2 activity was elevated in severe obesity and negatively associated with insulin sensitivity [48] . HSD2 expression is higher in omental than abdominal subcutaneous adipose tissue which may contribute to adipocyte hypertrophy and visceral obesity [49] . The sixth lead variant rs378854 is located at the long non-coding RNA CASC8 . Associations of variants at CASC8 have been reported for various cancers [50–52] but no association was reported for cardiometabolic traits . In our SNP and locus transferability analyses , a moderate number of EA-derived BMI and WHRadjBMI associated variants shared the same trait-raising alleles and displayed nominally significant associations in AA individuals , similar to previous findings [11–12] . While the BMI variants were similar in terms of their effect sizes and frequencies of trait-raising alleles between EA and AA populations , there were more discrepancies for WHRadjBMI variants . In addition , a substantial proportion of lead regional variants in AA were not in strong LD with EA lead variants , suggesting AA populations either have different association signals or the results may be spurious . Taken together , only <30% of EA loci were associated with BMI and WHRadjBMI in AA . Trans-ethnic fine mapping improved resolution to refine putative causal variant ( s ) in some loci as compared to using EA studies alone . In the meta-analyses of AA and EA GWAS , four BMI loci ( SEC16B , STXBP6 , FTO and MC4R ) and six WHRadjBMI loci ( PPARG , ADAMTS9 , SNX10 , KLF13 , CMIP and PEMT ) only contained one variant in the 99% credible sets . Among 16 BMI and 3 WHRadjBMI loci that were examined in both the previous trans-ethnic meta-analysis studies using HapMap imputation [7–8] and the present study , the number of variants and the interval of credible sets were either the same or lower in the present study for 13 and 15 loci , respectively . The majority of credible variants are non-coding in those sets containing ≤ 20 variants . Several of them located at the MTCH2 , MAP2K5 , NLRC3 , ATP2A1 , TBX15-WARS2 and FAM13A loci are cis-eQTL variants regulating nearby gene expression in subcutaneous and/or visceral adipose tissue , suggesting the putative causal variants may have a regulatory role instead of directly altering protein structure and function . Despite the low posterior probabilities , the coding changes of credible variants at ADCY3 and SH2B1 suggest that they may be the causal genes in the respective loci modulating BMI . Further studies are warranted to delineate putative causal variants including functional annotation in trans-ethnic fine mapping efforts [53] . Our large-scale GWAS meta-analyses in African ancestry individuals imputed to the 1000 Genomes reference panel , complemented by imputation of European GWAS using summary statistics and additional sex-stratified analyses , boosts the study power and improves resolution , leading to the identification of nine novel loci and fine mapping 37 loci with tractable credible sets . We observed significant associations for variants with MAF ≥ 0 . 5% , but rare variants were unlikely to be detected due to limited power and poor imputation quality . Large scale sequencing studies are needed to evaluate the contribution of rare variants in modulating complex traits such as BMI and WHR . Given the substantially larger sample size in European than in African ancestry samples , the trans-ethnic fine mapping results are largely driven by variants showing strong associations in Europeans . Future trans-ethnic studies including additional non-European populations will further improve the fine mapping effort .
We used a three-stage design to evaluate genetic associations with BMI and WHRadjBMI in sex-combined and sex-stratified samples ( S1 Fig ) . Stage 1 included GWAS meta-analyses in AA individuals and stage 2 included replication of top associations from stage 1 . Stage 3 included meta-analysis of top associations from stages 1 and 2 AA studies and EA meta-analysis results . In the discovery stage 1 of AAAGC , 17 GWAS of up to 42 , 752 AA individuals ( 16 , 559 men and 26 , 193 women; 41 , 696 African Americans and 1 , 056 Africans ) were included for the BMI analyses . A total of 10 GWAS of up to 20 , 384 AA individuals ( 4 , 783 men and 15 , 601 women; all African Americans ) were included for the WHRadjBMI analyses . For variants with P < 1×10−4 in either the sex-combined or the sex-stratified meta-analyses , stage 2 replication was performed in additional AA individuals from AAAGC ( N = 10 , 143 for BMI , N = 2 , 711 for WHRadjBMI ) , followed by meta-analysis with EA individuals from the GIANT consortium ( 322 , 154 for BMI , 210 , 086 for WHRadjBMI ) . Variants that reached genome-wide significance ( P < 5×10−8 ) were assessed for associations with BMI in two cohorts of children ( N = 7 , 222 ) . All AA participants in these studies provided written informed consent for the research , and approval for the study was obtained from the ethics review boards at all participating institutions . Detailed descriptions of each participating study and measurement and collection of height , weight , waist and hip circumferences are provided in S1 Text , S1 and S2 Tables . Genotyping in each study was performed with Illumina or Affymetrix genome-wide SNP arrays . Pre-phasing and imputation of missing genotypes in each study was performed using MaCH/ minimac [20] or SHAPEIT2/IMPUTEv2 [21–22] using the 1000 Genomes Project cosmopolitan reference panel ( Phase I Integrated Release Version 3 , March 2012 ) [18] . The details of the array , genotyping and imputation quality-control procedures and sample exclusions for each study are listed in S3 Table . In general , samples reflecting duplicates , low call rates , gender mismatch , or population outliers were excluded . Variants were excluded by the following criteria: call rate < 0 . 95 , minor allele count ( MAC ) ≤ 6 , Hardy-Weinberg Equilibrium ( HWE ) P < 1×10−4 , imputation quality score < 0 . 3 for minimac or < 0 . 4 for IMPUTE , or absolute allele frequency difference > 0 . 3 compared with expected allele frequency ( calculated as 1000 Genomes frequency of AFR × 0 . 8 + EUR × 0 . 2 ) . We evaluated the performance of 1000 Genomes imputation using the largest study , the Women’s Health Initiative ( WHI ) ( N = 8 , 054 ) . A total of 25 . 1 million variants with MAF ≥ 0 . 1% were imputed to the 1000 Genomes reference panel . Of these , 98 . 1% ( 8 . 8 million ) common variants , 95 . 4% ( 9 . 3 million ) low frequency variants ( 0 . 5% ≤ MAF < 5% ) , and 72 . 5% ( 4 . 6 million ) rare variants ( 0 . 1% ≤ MAF < 0 . 5% ) were well imputed with IMPUTE info scores ≥ 0 . 3 ( S4 Table ) . Notably , these frequencies are slightly lower than those obtained by imputation using 1000 Genomes phase 1 interim reference panel in Europeans [54] . However , 72 . 6% , 95 . 5% and 99 . 5% of the common , low frequency and rare variants , respectively , from the 1000 Genomes reference panel were not present in the HapMap and therefore demonstrate deeper coverage of the genome , particularly for the low frequency and rare variants . At all stages , genome-wide association analyses were performed by each of the participating studies . BMI was regressed on age , age squared , principal components and study site ( if needed ) to obtain residuals , separately by sex and case-control status , if needed . WHR was regressed on age , age squared , principal components , BMI and study site to obtain residuals , separately by sex and case-control status . Principal components were included to adjust for admixture proportion and population structure within each study . Residuals were inverse-normally transformed to obtain a standard normal distribution with mean of zero and standard deviation of one . For studies with unrelated subjects , each variant was tested assuming an additive genetic model with each trait by regressing the transformed residuals on the number of copies of the variant effect allele . The analyses were stratified by sex and case-control status ( if needed ) . For studies that included related individuals , family based association tests were conducted that took into consideration the genetic relationships among the individuals . Sex stratified , case-control stratified and combined analyses were performed . Association results with extreme values ( absolute beta coefficient or standard error ≥ 10 ) , primarily due to small sample sizes and/or low minor allele count , were excluded for meta-analysis . The latest summary statistics of sex-combined and sex-stratified meta-analyses of BMI and WHRadjBMI imputed to the HapMap reference panel in EA from the Genetic Investigation of ANthropometric Traits ( GIANT ) consortium were obtained from http://www . broadinstitute . org/collaboration/giant/index . php/GIANT_consortium_data_files [7–8] . These association summary statistics were used to impute z-scores of unobserved variants at the 1000 Genomes Project EUR reference panel ( Phase I Integrated Release Version 3 ) using the ImpG program [23] . In brief , palindromic variants ( AT/CG ) and variants with allele mismatch with the reference were removed from the data . Using the ImpG-Summary method , the z-score of an unobserved variant was calculated as a linear combination of observed z-scores weighted by the variance-covariance matrix between variants induced by LD within a 1 Mb window from the reference haplotypes . The sample size of each unobserved variant was also interpolated from the sample sizes of observed variants using the same weighting method for z-score as Ni=∑t=1t=T|wi , t|∑ |wi , t|Nt . Here , t = 1 , 2 , … . , T , where T is the number of observed variants , wi , t is the element of the covariance matrix Σi , t for the unobserved variant i and the observed variant t within window . The performance of imputation was assessed by r2pred , with similar characteristics as the standard imputation accuracy metric r2hat [20] . Results of variants with r2pred ≥ 0 . 6 were used in subsequent analyses . In the discovery stage 1 , association results were combined across studies in sex-combined and sex-stratified samples using inverse-variance weighted fixed-effect meta-analysis implemented in the program METAL [55] . The study-specific λ values of association ranged from 0 . 97 to 1 . 05 for BMI , and 0 . 98 to 1 . 05 for WHRadjBMI ( S3 Table ) . Genomic control correction [56] was applied to each study before meta-analysis , and to the overall results after meta-analysis ( λ = 1 . 07 for BMI , 1 . 01 for WHRadjBMI ) . Variants with results generated from < 50% of the total sample size for each trait were excluded . After filtering , the numbers of variants reported in the meta-analyses were 17 , 972 , 087 for BMI , and 20 , 502 , 658 for WHRadjBMI . Variants with P < 1×10−4 in stage 1 sex-combined or sex-stratified meta-analyses were carried forward for replication in additional AA individuals ( stage 2 ) and EA individuals ( stage 3 ) . For each of the replication AA studies , trait transformation and association were performed as in stage 1 and results were meta-analyzed using the inverse-variance method in METAL . For the replication study in EA , HapMap imputed summary statistics of each trait from the GIANT consortium were used to impute z-scores of unobserved variants at the 1000 Genomes . In stages 1 and 2 , meta-analysis results of AA studies were combined using the inverse-variance weighted method . In all stages including both AA and EA studies , meta-analysis results expressed as signed z-scores were combined using the fixed effect sample size weighted method in METAL due to the lack of beta and standard error estimates from the ImpG program [23] . Evidence of heterogeneity of allelic effects between males and females , within and across stages were assessed by the I2 statistic in METAL . Genome-wide significance was declared at P < 5×10−8 from each of the sex-combined and sex-stratified meta-analysis including AA and/or combined AA and EA individuals . Difference in effects between men and women was assessed using Cochran’s Q test and nominal Phet < 0 . 05 declared as significant . A lead variant in a locus was defined as the most significant variant within a 1 Mb region . A novel locus was defined as a lead variant with distance > 500 kb from any established lead variants reported in previous studies . By convention , a locus was named by the closest gene ( s ) to the lead variant . For the genome-wide significant loci identified in sex-combined and sex-stratified analyses in AA ( stages 1+2 ) , we used the program GCTA [57–58] to select the top independent associated variants from summary statistics of the meta-analyses . This method uses the LD correlations between variants estimated from a reference sample to perform an approximate conditional association analysis . We used 8 , 054 unrelated individuals of African ancestry from the WHI cohort with ~15 . 7M variants available as the reference sample for LD estimation . To select the top independent variants in the discovery and replication meta-analysis results , we first selected all variants that had P < 5×10−8 and conducted analysis conditioning on the selected variants to search for the top variants iteratively via a stepwise model to select the independent variants from this list . Then we proceeded to condition the rest of the variants that had P > 5×10−8 on the list of independent variants in the same fashion until no variant had conditional P that passed the significance level P < 5×10−8 . Finally , all the selected variants were fitted jointly in the model for effect size estimation . We also tested if the genome-wide significant variants identified from sex-combined GWAS in AA and the locus-wide significant variants identified from sex-combined and sex-specific locus transferability studies in AA were independent from nearby established loci identified from EA studies [7–8] . First , the published lead variants from EA studies were used to search for all surrogate variants that were in high LD ( r2>0 . 8 in 1000 Genomes Project EUR population ) . Second , these variants were pruned to select only variants in low LD in AA ( r2<0 . 3 in the 1000 Genomes Project AFR population ) to avoid collinearity in conditional analysis . Third , association analysis was conducted on the AA significant variants conditioned on the selected EA lead and surrogate variants , using the program GCTA and estimated LD correlation from the WHI cohort . For genome-wide significant loci , an AA derived association signal is considered as independent from the established EA signals when the difference in–logP <3 and difference in effect size < 1 standard error after conditional analysis . For locus-wide significant loci , given the lower level of significance , independence is only considered as difference in effect size < 1 standard error after conditional analysis . We investigated the transferability of EA BMI and WHR associated variants and loci in AA individuals from stage 1 sex-combined and sex-stratified meta-analyses . First , we tested for replication of lead variants previously reported to be associated with BMI ( 176 variants from 170 loci ) and WHRadjBMI ( 84 variants from 65 loci ) at genome-wide significance in sex-combined and sex-stratified analyses from the GIANT consortium studies [7–9] . We defined SNP transferability as an EA lead variant sharing the same trait-raising allele at nominal P < 0 . 05 in AA individuals . To account for differences in local LD structure across populations , we also interrogated the flanking 0 . 1cM regions of the lead variants to search for the best variants with the smallest association P in AA individuals . Locus-wide significance was declared as Plocus < 0 . 05 by Bonferroni correction for the effective number of tests within a locus , estimated using the Li and Ji approach [59] . We compared the credible set intervals of established loci that showed locus-wide significance ( Plocus < 0 . 05 ) in the sex-combined or sex-specific analyses from this study in summary statistics datasets including the 1000 Genomes imputed results from GIANT , AAAGC and meta-analysis of GIANT and AAAGC . In each dataset , a candidate region is defined as the flanking 0 . 1cM region of the lead variant reported by the GIANT consortium . Under the assumption of one causal variant in a region of M variants , the posterior probability of a variant j with association statistics Z driving the association , P ( Cj|Z ) , was calculated using the formula P ( Cj|Z ) =exp ( 12zj2 ) ∑j=1Mexp ( 12zj2 ) . A 99% credible set was constructed by ranking all variants by their posterior probability , followed by adding variants until the credible set has a cumulative posterior probability > 0 . 99 [53] . Given our sample sizes in the discovery and replication stages in our African ancestry populations , we have >80% power to detect variants explaining 0 . 08% variance for BMI that corresponds to effect sizes of 0 . 09 and 0 . 20 SD units for MAF of 0 . 05 and 0 . 01 , respectively . For WHRadjBMI , we have >80% power to detect variants explaining 0 . 18% variance that corresponds to effect sizes of 0 . 14 and 0 . 30 SD units for MAF of 0 . 05 and 0 . 01 , respectively . | Genome-wide association studies ( GWAS ) have identified >300 genetic regions that influence body size and shape as measured by body mass index ( BMI ) and waist-to-hip ratio ( WHR ) , respectively , but few have been identified in populations of African ancestry . We conducted large scale high coverage GWAS and replication of these traits in 52 , 895 and 23 , 095 individuals of African ancestry , respectively , followed by additional replication in European populations . We identified 10 genome-wide significant loci in all individuals , and an additional seven loci by analyzing men and women separately . We combined African and European ancestry GWAS and were able to narrow down 43 out of 74 African ancestry associated genetic regions to contain small number of putative causal variants . Our results highlight the improvement of applying high density genome coverage and combining multiple ancestries in the identification and refinement of location of genetic regions associated with adiposity traits . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"genome-wide",
"association",
"studies",
"africans",
"alleles",
"ethnicities",
"mathematics",
"statistics",
"(mathematics)",
"genome",
"analysis",
"trait",
"locus",
"analysis",
"research",
"and",
"analysis",
"methods",
"genomic",
"signal",
"processing",
"mathematical",
"a... | 2017 | Discovery and fine-mapping of adiposity loci using high density imputation of genome-wide association studies in individuals of African ancestry: African Ancestry Anthropometry Genetics Consortium |
In present-day eukaryotes , the cell division cycle is controlled by a complex network of interacting proteins , including members of the cyclin and cyclin-dependent protein kinase ( Cdk ) families , and the Anaphase Promoting Complex ( APC ) . Successful progression through the cell cycle depends on precise , temporally ordered regulation of the functions of these proteins . In light of this complexity , it is surprising that in fission yeast , a minimal Cdk network consisting of a single cyclin-Cdk fusion protein can control DNA synthesis and mitosis in a manner that is indistinguishable from wild type . To improve our understanding of the cell cycle regulatory network , we built and analysed a mathematical model of the molecular interactions controlling the G1/S and G2/M transitions in these minimal cells . The model accounts for all observed properties of yeast strains operating with the fusion protein . Importantly , coupling the model’s predictions with experimental analysis of alternative minimal cells , we uncover an explanation for the unexpected fact that elimination of inhibitory phosphorylation of Cdk is benign in these strains while it strongly affects normal cells . Furthermore , in the strain without inhibitory phosphorylation of the fusion protein , the distribution of cell size at division is unusually broad , an observation that is accounted for by stochastic simulations of the model . Our approach provides novel insights into the organization and quantitative regulation of wild type cell cycle progression . In particular , it leads us to propose a new mechanistic model for the phenomenon of mitotic catastrophe , relying on a combination of unregulated , multi-cyclin-dependent Cdk activities .
S and M are triggered by the phosphorylation of specific cellular proteins by a family of protein kinases , called cyclin-dependent kinases ( Cdk’s ) [2] . The activity of a Cdk depends on obligatory association with a regulatory subunit of the cyclin family , and a variety of Cdk:cyclin complexes are responsible for initiating DNA replication and mitosis in present-day eukaryotes . These observations naturally led to the “qualitative model” of cell cycle control , in which the temporal alternation of S and M is a consequence of alternating oscillations of at least two different Cdk:cyclin complexes , SPF ( S-phase promoting factor ) and MPF ( M-phase promoting factor ) , with different substrate specificities [3] . This qualitative model might be true for cell cycle control in higher eukaryotes , but it is difficult to reconcile with the fact that a single Cdk1:cyclin B complex can drive an ordered sequence of S and M phases in fission yeast [4 , 5] . ( In fission yeast , Cdk1 is encoded by the cdc2 gene and its only essential partner , a B-type cyclin , is encoded by cdc13 . ) The observation that Cdc2:Cdc13 alone is sufficient to orchestrate the fission yeast cell cycle led to a “quantitative model” of cell cycle control [4 , 6] , in which low Cdc2 activity is sufficient to trigger DNA replication while high activity blocks re-replication and brings about mitosis . Strong experimental support for the quantitative model was provided by the demonstration that a Cdc13-Cdc2 fusion protein ( called Cdc13-L-Cdc2 , L for “linker” ) can by itself drive cell cycle progression in a manner that is indistinguishable from wild type fission yeast cells [7] . In the minimal strain ( genotype: cdc13-L-cdc2 Δcdc13 Δcdc2 ) , the genomic copies of cdc13 and cdc2 have been deleted , so that cells cannot make normal Cdc2:Cdc13 heterodimers and therefore rely solely on the fusion protein for MPF activity . In addition , because these cells lack Cdc2 monomers , they should not be able to make heterodimers of Cdc2 with G1- or S-specific cyclins ( Cig1 , Cig2 and Puc1 , encoded by cig1 , cig2 , and puc1 , respectively ) . Nevertheless , to prevent the formation of potential alternative complexes between the Cdc2 moiety of the fusion protein and these cyclins ( e . g . Cdc13-L-Cdc2:Cig2 ) , which may contribute to temporal ordering of the cell cycle , the three other fission yeast cell cycle cyclin genes were deleted ( Δcig1 Δcig2 Δpuc1 , referred to as ΔCCP ) . Strikingly , cdc13-L-cdc2 Δcdc13 Δcdc2 ΔCCP cells progress through S and M in perfectly wild type fashion , indicating that the fusion protein Cdc13-L-Cdc2 has both SPF and MPF activities . We will refer to cell cycle control in this strain as the “Minimal Cdk Network” , and we will use MCN to denote the genotype of these cells ( i . e . , cdc13-L-cdc2 Δcdc13 Δcdc2 ΔCCP ) . Both wild type and MCN cell cycles are characterized by a very short G1 phase ( from the end of M phase to the onset of S phase ) and a long G2 phase ( from the end of S phase to the onset of mitosis ) ; and both types of cells divide at a length of 14–16 μm ( Table 1 ) . The long duration of G2 results from inhibition of Cdc2:Cdc13 activity by the Wee1 and Mik1 kinases [8 , 9] . However , loss-of-function of these inhibitory kinases has very different consequences in wild type and MCN cells . In wild type cells , inactivation of Wee1 ( using a temperature-sensitive mutation , wee1–50ts ) advances cells into mitosis , shortening the time spent in G2 and reducing cell size at division by almost 50% [10] . Furthermore , simultaneous inactivation of both Wee1 and Mik1 ( wee1–50ts Δmik1 double mutant ) or mutation of their target sites in Cdc2 into non-phosphorylable residues ( T14A and Y15F , referred to as cdc2AF ) drives otherwise wild type cells into mitosis before they have completed DNA synthesis , a lethal situation called “mitotic catastrophe” [9 , 11 , 12] . In contrast , MCN cells devoid of Cdc2 inhibitory phosphorylation ( i . e . , MCN Δwee1 Δmik1 and MCN-AF ) are perfectly viable , and their average size at division is comparable to that in wild type cells ( Table 1 ) . These unusual and intriguing properties of MCN cells have prompted us to develop a mathematical model of the minimal Cdk network that addresses , in particular , the central question of why Cdk1 inhibitory phosphorylation is mostly essential in wild type cells but dispensable in MCN cells . If , as experiments suggest , the fundamental timing of S and M phases in the fission yeast cell cycle is a quantitative property of the temporally varying activity of Cdk1 , then the phenotypic properties of wild type and MCN cells cannot be accurately explained in terms of qualitative reasoning about genetic effects . Rather , coupling experimental results with a quantitative , computational model is necessary to capture the consequences of genetic manipulations in the context of wild type and MCN genetic backgrounds . By creating a mathematical model of the interactions of the fusion protein with Wee1 and other Cdk-regulatory proteins , we provide a consistent , quantitative understanding of how such a minimal control system can direct perfectly normal fission yeast cell cycles . Furthermore , confronting the model’s predictions with the phenotype of alternative minimal yeast cells , we propose an explanation for the paradoxical effects of abrogating inhibitory phosphorylation of Cdk1 in wild type and MCN backgrounds , thereby revisiting the mechanistic origin of mitotic catastrophe . Thus , not only does our model behave identically to the MCN strains described by Coudreuse & Nurse [7] , but it also has important implications for wild type cell cycle control , shedding new light on the functional interactions between Cdk and its inhibitor Rum1 , on the roles of CCP-dependent Cdk activity in regulating the timing of mitosis , and on the effects of molecular noise on cell cycle robustness .
The ‘Minimal Cdk Network’ presented in Fig . 1 was converted into a set of kinetic equations ( S1 and S2 Tables ) based on the assumptions described in the Methods section ( Mathematical modelling ) . These non-linear ordinary differential equations were solved numerically for carefully chosen kinetic parameter values ( S3 Table ) to generate the time evolution of the variables of the minimal network . In Fig . 2A ( left panel ) , we plot the time evolution of some representative variables in our mathematical model of MCN cells: the active forms of MPF and Wee1 , the total levels of Rum1 ( Rum1T ) and Cdc13-L-Cdc2 ( FPT ) , the activity of the Anaphase Promoting Complex ( APC:Slp1 ) and cell mass as a proxy for cell size . Between successive cell divisions , we can distinguish three stages of MPF activity . During the initial , brief stage immediately after cell division , MPF activity is close to zero as a result of Rum1-dependent inhibition and degradation of MPF . Low MPF activity allows for the re-accumulation of replication licensing factors ( Cdc18 and Cdt1 ) at the replication origins in the yeast genome [13] . During the following intermediate stage , the total amount of fusion protein rises quickly , but MPF activity rises slowly as a consequence of Wee1-mediated conversion of MPF to its phosphorylated , inhibited form , MPFP . Finally , a transient stage of high MPF activity is observed during mitosis , as MPFP is abruptly converted to MPF by the Cdc25 phosphatase . Note that inhibitory phosphorylation of Cdc2 is the rate-limiting process for cell cycle progression in MCN cells , as it is also in wild type cells . In both experiments and simulations , loss of Rum1 from the minimal Cdk network ( strain cdc13-L-cdc2 Δcdc13 Δcdc2 ΔCCP Δrum1 = MCN Δrum1 ) introduces no obvious changes in the temporal pattern of MPF activity and cell size ( compare left and right panels of Fig . 2A with Fig . 2B; see also Table 1 Row 7 and Table 2 Row 2 ) , because Rum1 is not rate-limiting for cell cycle progression in this background . MPF activity is still characterized by three stages with low , intermediate and high activities . The low activity of MPF in early G1 phase in this simulation is a consequence solely of APC:Slp1-dependent degradation of the fusion protein . Removal of Wee1 from the minimal Cdk network ( MCN Δwee1 Δmik1 cells ) is simulated in Fig . 2C , left panel . Although the lack of inhibitory phosphorylation changes the temporal dynamics of MPF activity , it does not significantly affect cell size at division ( Fig . 2C right panel , and Table 1 Row 8 ) , as observed experimentally [7] . However , the duration of the low MPF activity state ( corresponding to G1 ) is extended , as reflected by the accumulation of cells with a 1C DNA content ( Fig . 2C , right panel inset ) , due to the persistence of high levels of Rum1 , whose degradation has become the rate-limiting step in cell cycle progression . To understand the unusual phenotype of MCN Δwee1 Δmik1 cells , as compared to Δwee1 Δmik1 in a wild type background ( see below ) , let us return to a consideration of MCN cells . In the presence of Wee1 , the fusion protein in our model accumulates first in a phosphorylated form , MPFP , which is not efficiently inhibited by Rum1 . We assume that MPFP binds only weakly to Rum1 , limiting the effect of persistent stoichiometric inhibition of MPFP by Rum1 but allowing phosphorylation of Rum1 by MPFP . Hence , despite its lower intrinsic activity compared to MPF , MPFP can promote early phosphorylation and subsequent degradation of Rum1 . Contrariwise , in the absence of Wee1 , the fusion protein accumulates in its unphosphorylated form , which is strongly inhibited by tight binding to Rum1 . MPF-dependent mono-phosphorylation of Rum1 is therefore slow and Rum1P is rapidly dephosphorylated by a phosphatase . Because MPF preferentially re-associates with non-phosphorylated Rum1 , the rate of multi-phosphorylation of Rum1 and its subsequent degradation is strongly reduced , leading to the maintenance of high levels of Rum1 . Therefore , the mutant cells ( MCN Δwee1 Δmik1 ) must first accumulate enough fusion protein to titrate out non-phosphorylated Rum1; only then will there be some free MPF available to catalyse the second phosphorylation of Rum1P ( followed by subsequent degradation of Rum1P2 ) , allowing entry into the next regime of MPF activity . ( See Mathematical modelling in the Methods section for further details concerning the assumptions we have made about the binding of Rum1 to MPF and to MPFP and the rates of phosphorylation of Rum1 by the two forms of MPF . ) The persistence of significant levels of active Rum1 in the MCN Δwee1 Δmik1 cells provides a mechanistic explanation for the observed viability and extended G1 phase in this genetic background ( see [7] and Fig . 2C right panel ) . If our understanding of the role of Rum1 in these cells is correct , then rum1 should be an essential gene in MCN Δwee1 Δmik1 cells , a prediction confirmed by our observation that deletion of rum1 in the MCN Δwee1 Δmik1 background is lethal ( Table 1 Row 13; see also Experimental procedures section ) . Our model also predicts that the G2/M transition in this strain is brought about by abrupt Rum1 degradation rather than by Cdc2 dephosphorylation ( as is the case in wild type and MCN cells ) . In a wild type genetic background , Δwee1 Δmik1 cells enter M phase prematurely and undergo unconditional “mitotic catastrophe” , i . e . they divide before they have completed DNA replication [14] . Hence , newborn cells do not receive complete copies of the genome and eventually die . In the MCN genetic background , Δwee1 Δmik1 cells avoid mitotic catastrophe because they have a long gap ( ~50 min ) between the onset of DNA synthesis and entry into mitosis resulting from MPF inhibition by Rum1 . However , MCN Δwee1 Δmik1 cells lack an active S phase checkpoint and hence remain subject to conditional mitotic catastrophe; i . e . if DNA synthesis is challenged by drugs such as hydroxyurea , these cells enter mitosis with incompletely replicated DNA and die [7] . Importantly , applying our mathematical approach to the particularities of these MCN-derived cells provides a novel opportunity to study the causes of mitotic catastrophe in fission yeast cells devoid of Cdc2 inhibitory phosphorylation , as will be described in a later paragraph . It is instructive to analyse our model of MCN cell cycles on one-parameter bifurcation diagrams [15 , 16] . A one-parameter bifurcation diagram plots the stable and unstable attractors ( steady states and oscillations ) of a dynamical system as functions of a “control parameter” . In the case of yeast cell cycle regulation , it is sensible to choose cell mass ( m ) as the control parameter , because cell growth ( increase in m ) is a major driving force for cell cycle progression in lower eukaryotes . In our model , m increases slowly and exponentially , so one may think of the bifurcation parameter as eμt , where μ is the specific growth rate of the cell in the culture medium . Hence , progression in time also corresponds to movement from left to right along the horizontal axis of the one-parameter bifurcation diagram . By defining distinct MPF-thresholds for initiation of S and M phases ( θS = 0 . 01 and θM = 0 . 2 ) , MPF activity can be divided into three regimes ( Fig . 2A , middle panel ) , which can be associated with G1 , S/G2 and M activities . For MCN cells , the bifurcation diagram shows three distinct and partially overlapping steady states of MPF activity ( Low , Intermediate and High ) within these regimes . The low and intermediate steady states are stable , while the high steady states are unstable and surrounded by stable limit cycle oscillations . The low MPF activity steady state is defined by Rum1-dependent inhibition of MPF and degradation of Cdc13-L-Cdc2 by APC . The intermediate MPF activity state is stabilized by Wee1-dependent inhibition of MPF . The high activity state is destabilized by the negative feedback loop between MPF and APC:Slp1 . Wherever a stable steady state ends with increasing cell mass ( i . e . , increasing time ) , the control network must jump to the next stable state , which corresponds to a cell cycle transition . For our choice of parameter values ( S3 Table ) , the transition from L to I is smooth ( not associated with a bifurcation ) , but the two steady states are still qualitatively different: in the L state , MPF is inhibited by high levels of Rum1; in the I state , MPF activity is downregulated by Wee1 . In contrast , the transition from I to H is abrupt and defined by a SNIC bifurcation ( “saddle-node on an invariant circle” [15] ) , corresponding to the transition from G2 phase ( intermediate activity of MPF ) to M phase ( high activity of MPF ) . By overlaying the time course of a cell cycle simulation ( MPF activity and mass from Fig . 2A , left panel ) on the bifurcation diagram , we plot how the “cell cycle states” in the model change during progression through a MCN cycle ( green curve in the middle panel of Fig . 2A ) . To describe this cycle , we start where the green curve rises above θM = 0 . 2 , the threshold for MPF initiation of mitosis . MPF activity increases rapidly , but it cannot settle on the H steady state because H is unstable . Shortly after the cell enters M phase , it exits mitosis as activated APC:Slp1 marks the fusion protein for degradation . When MPF activity drops below θM , the cell divides and m is reset to m/2 ( 1 . 00 to 0 . 50 ) . ( For simplicity , we assume that the threshold for exit from mitosis is identical to the threshold for entry into mitosis . ) The newborn cell enters the domain of attraction of the stable I steady state , but its trajectory on the way to this attractor carries it to a lower level of MPF activity ( 0 . 0035 ) . It is necessary for MPF to drop to such low activity in order to allow the relicensing of the DNA replication origins . Subsequently , as the cell grows and MPF activity increases again ( it is being attracted to the I steady state ) , the green curve crosses θS = 0 . 01 , the threshold for MPF-triggered initiation of DNA synthesis . At this point , the young cell leaves the G1 phase of the cell cycle ( unreplicated chromosomes ) and enters S phase ( replicating chromosomes ) . Note that the MCN cell never visits the L steady state of the control network , which corresponds to a stable G1-like phase . G1 only represents a short , transient phase of the cell cycle in fission yeast because the previously described G1/S size control is cryptic [17] . In addition , the cell trajectory ( green curve ) on the bifurcation diagram never gets very close to the I stable steady state , due to the slow turn-over of the fusion protein when APC is inactive . Therefore cell growth carries the cell past the SNIC bifurcation into the domain of attraction of the stable limit cycle oscillations , before it can reach the I attractor . Finally , MPF activity undergoes an explosive rise , which carries it across θM , where we started this tour of the MCN cell cycle . For Rum1-depleted MCN cells , the Wee1/Mik1-mediated activity state ( I ) is extended to a broader range , including the low MPF activity regime at very small cell size ( Fig . 2B , middle panel ) . However , since the cell cycle trajectory of MCN rum1+ cells never approaches the stable L state , rum1 deletion has no discernible effect on cell cycle progression ( compare middle panels of Figs . 2A and 2B ) , as observed ( Table 1 Row 7 ) . Because they lack the Rum1-dependent L state , Δrum1 cells cannot be arrested in G1 with mating pheromones [18] . Depletion of Wee1 and Mik1 from MCN cells eliminates the stable I state from the bifurcation diagram ( Fig . 2C , middle panel ) , as expected . However , the Rum1-dependent L state now becomes extended over the MPF threshold for DNA replication ( θS ) into the S/G2 activity regime and terminates at a SNIC bifurcation point at m ≈ 0 . 7 , which is nearly the same size as the SNIC bifurcation in the background strain ( MCN ) . Hence , MCN Δwee1 Δmik1 cells divide at about the same size as MCN cells ( see Tables 1 and 2 ) . Compared to the initial MCN strain , the stable limit cycles in the absence of Wee1 and Mik1 have a smaller amplitude , but this is of little practical effect because cell division carries the cell cycle trajectory into the domain of attraction of the stable L state . Roughly half of the trajectory for the growing newborn cell lies close to the L attractor , consistent with the fact that these cells have an extended G1 phase ( Fig . 2C right panel , Table 1 Row 8 and Table 2 Row 3 ) . Note that DNA synthesis is initiated when the cell cycle trajectory ( the green curve in the middle panel of Fig . 2C ) crosses θS , while the control system is still in the domain of attraction of the stable L steady state . For this reason , there is a significant time delay ( ~50 min ) between the onset of DNA synthesis ( when the green curve crosses θS ) and entry into mitosis ( when the green curve crosses θM ) , which prevents these cells from undergoing an unconditional mitotic catastrophe , despite the fact that they lack an effective S phase checkpoint . Coudreuse & Nurse [7] observed that lack of inhibitory Cdc2 phosphorylation in MCN cells surprisingly preserves both viability and average cell size at division , in contradiction to the properties of wild type fission yeast cells ( Table 1 ) . Inactivation ( or deletion ) of Wee1 in otherwise wild type cells reduces cell size at division by ~50% , and complete elimination of inhibitory Cdc2 phosphorylation ( wee1ts mik1Δ , wee1ts cdc25op , or cdc2AF ) strongly affects cell viability , with a high incidence of death by unconditional mitotic catastrophe . Why do similar mutations in the MCN background have such different effects ? Interestingly , Coudreuse & Nurse [7] also demonstrated that MCN cells can exhibit “conditional” mitotic catastrophe after “G1 reset” . To do this experiment , they introduced the Shokat mutation [19] , cdc2as , into the fusion protein in the MCN strain , thereby making MPF activity responsive to the ATP analog NmPP1 . The protocol of a “G1 reset” experiment is: ( 1 ) arrest cells in G2 with a low dose of inhibitor ( 1 μM NmPP1 ) so that cells complete DNA synthesis but do not enter mitosis; ( 2 ) transfer cells to a high dose of inhibitor ( 10 μM NmPP1 ) to reduce MPF activity to very low level , thereby re-setting cells to G1 without an intervening mitosis and re-licensing origins of DNA replication; ( 3 ) transfer cells to inhibitor-free medium to allow rapid rise in MPF activity . Under these conditions , cells initiate a new round of DNA synthesis , but also rapidly enter into mitosis and show a “cut” phenotype , indicating that they are dividing with incompletely replicated chromosomes . To simulate a “G1-reset” experiment with the minimal Cdk network model , we multiply MPF activity by a factor of 1/ ( 1+NmPP1 ) , where NmPP1 is the concentration of inhibitor in units of its IC50 ( concentration that inhibits 50% of the kinase activity ) . In Fig . 3 , we allow a simulated cell to enter S phase at NmPP1 = 0 , then set NmPP1 = 1 for 100 min to block the cell in G2 and allow it to accumulate an excess of fusion protein , then set NmPP1 = 20 for 70 min to reset the cell to G1 , and finally set NmPP1 = 0 to activate the pool of fusion protein . As can be seen in panel B of the figure , MPF activity increases abruptly after release , triggering a new round of DNA replication at t = 320 min . At t ≈ 325 min , Wee1 tries unsuccessfully to inhibit the fusion protein , and at t = 336 min MPF activity surpasses θM , triggering premature entry into mitosis , almost exactly as observed by Coudreuse & Nurse [7] . In this context , conditional mitotic catastrophe happens at twice the normal division size ( m ≈ 2 , Fig . 3 ) . Since the control system is far away from the SNIC bifurcation point ( m ≈ 0 . 7 ) , the vertical rise in MPF activity is abrupt and it crosses the thresholds for S and M phases within less than 20 min ( Fig . 3 ) . This simulation , modifying the normal dynamic changes in Cdk activity , demonstrates that cells governed by the minimal Cdk network are fundamentally capable of entering mitotic catastrophe , suggesting that specific properties of the Cdk control network in MCN Δwee1 Δmik1 cells allow for their viable progression through the cell cycle . The obvious difference between wild type and MCN cells is that G1 and S cyclins ( Cig1 , Cig2 and Puc1 , collectively called CCP ) are absent in the MCN strain . In fact , even when present , these cyclins are less likely to form complexes with Cdc2 , as demonstrated by Coudreuse & Nurse ( see Table 1: in the cdc13-L-cdc2 Δcdc13 Δcdc2 background , there are no significant differences between CCP+ and ΔCCP cells ) . CCP-dependent Cdc2 activity may therefore account for the major differences between wild type and MCN cells in the absence of Cdc2 inhibitory phosphorylation . To test this hypothesis , we supplemented our model of MCN cells with a “generic” CCP-dependent Cdc2 activity , reflecting the situation in cdc13-L-cdc2 Δcdc13 cdc2+ CCP+ cells . These additional cyclins account for additional sources of Cdc2 activity throughout the cell cycle , but their temporal patterns are unknown , except for Cig2 [20] . Therefore , in the extended model , this generic Cdc2 activity is represented simply by a parameter , CCP = 1 , and its only effect is to promote the phosphorylation and degradation of Rum1 [5 , 21] . As a result , the peak of Rum1 accumulation is lowered ( Fig . 4A , left panel ) and the range of the stable L steady state is restricted ( Fig . 4A , middle panel ) compared to MCN cells ( Fig . 2A ) . However , the cell cycle properties of these two strains are nearly identical in the model ( Table 2 Rows 1 and 5 ) , as the cell cycle trajectory remains far from the L steady state . We have confirmed these predictions by examining the phenotype of the corresponding cdc13-L-cdc2 Δcdc13 cdc2+ CCP+ strain ( Fig . 4A right panel and Table 1 Row 10 ) . The only discrepancy is that the observed cell size at division is slightly smaller than MCN cells , which could be caused by a small effect of CCPs on mitotic control . On the other hand , the effect of CCP-dependent Cdc2 activity on cells operating with a non-phosphorylable fusion protein is quite different , as seen in the cdc13-L-cdc2AF Δcdc13 cdc2+ CCP+ strain ( Fig . 4B ) . We purposefully consider non-phosphorylable fusion protein rather than Δwee1 Δmik1 so that only the fusion protein bypasses inhibitory phosphorylation . The simulation clearly suggests the emergence of a “wee” phenotype with reduced cell size at division and extended G1 phase . The I stable steady state disappears from the bifurcation diagram ( Fig . 4B , middle panel ) , similar to the situation in MCN Δwee1 Δmik1 cells ( Fig . 2C , middle panel ) . The mutant cells ( cdc13-L-cdc2AF Δcdc13 cdc2+ CCP+ ) cycle between a stable L and an unstable H state , replicating their DNA near the L steady state . Notably , although extended , the duration of G1 phase is shorter than in Fig . 2C , as the generic Cdc2:CCP background activity helps the fusion protein to eliminate Rum1 ( see also Table 2 Row 6 ) . Consequently , cell size at the SNIC bifurcation in Fig . 4B , middle panel , is about half the value of the SNIC bifurcation point in Fig . 2C , middle panel , so these mutant cells divide at a smaller size than MCN cells deleted for wee1 and mik1 . These cells ( CCP = 1 ) avoid unconditional mitotic catastrophe because their chromosomes get fully replicated during the 63 min interval when MPF rises from θS to θM ( see Table 2 ) . Consistent with these predictions , we find that cdc13-L-cdc2AF Δcdc13 cdc2+ CCP+ cells are viable and small with a slightly extended G1 ( Fig . 4B , right panel and Table 1 Row 11 ) . As previously mentioned , replacing cdc2+ by cdc2AF in a wild type genetic background causes mitotic catastrophe ( Table 1 ) . In the context of our model , we identify such cdc2AF cells ( in a wild type genetic background ) with the strain cdc13-L-cdc2AF Δcdc13 cdc2AF CCP+ ( Table 2 Row 8 ) . In contrast to the situation in cdc13-L-cdc2AF Δcdc13 cdc2+ CCP+ strain ( CCP = 1 ) , Cdc2AF:CCP complexes in these MCN-derived cells cannot be inhibited by Wee1 and Mik1 [22]; hence , we set CCP = 2 to model these cells . In the first case ( CCP = 1 ) , cells are small and viable ( Fig . 4B ) , whereas in the second case ( CCP = 2 ) , cells are small and inviable ( Fig . 4C ) as they cannot properly re-license the DNA replication origins after cell division ( green curve above θS in right panel of Fig . 4C ) . In addition , in the absence of Wee1- and Mik1-dependent phosphorylation of Cdc2 , the S phase checkpoint cannot delay MPF activation , and these cells enter into unconditional mitotic catastrophe at the next mitosis . Our analysis of the MCN network in the absence of Cdk1 inhibitory phosphorylation therefore suggests that mitotic catastrophe in cdc2AF cells in an otherwise wild type background results from unregulated Cdc2 activity mediated by both Cdc13 and CCP cyclins . To test this prediction , we compared a wee1–50ts Δmik1 strain , which is not viable at the restrictive temperature of 36 . 5°C , with a wee1–50ts Δmik1 ΔCCP strain ( Fig . 5 ) . At permissive temperature , both populations of cells were slightly shorter at division than ΔCCP cells; we also observed a significantly longer G1 in wee1–50ts Δmik1 ΔCCP cells . Strikingly , the absence of G1/S cyclins significantly rescued the effects of loss of Wee1 and Mik1 at restrictive temperature ( 5h ) , with a lower incidence of “cut” cells . These results support the conclusions from the model , suggesting that the viability of MCN-AF and MCN Δwee1 Δmik1 cells is not due to an intrinsically lower activity of the fusion system even in the absence of Cdc2 inhibitory phosphorylation . Importantly , they provide compelling evidence that similar to the situation in MCN cells , mitotic catastrophe in a wild type background results from the combination of non-regulated Cdc2 activity in association with Cdc13 and the G1/S cyclins . MCN cells , MCN-AF cells , and MCN Δwee1 Δmik1 cells exhibit the same mean cell size at division ( see Tables 1 and 2 ) , but the size distribution is considerably broader in the absence of inhibitory phosphorylation of Cdc2 [7] . To explore this property , we created a stochastic version of the model and studied the effects of Cdc2 phosphorylation on the robustness of minimal Cdk oscillations towards molecular noise . We introduced both intrinsic and parametric noise into the model . Intrinsic noise , i . e . molecular fluctuations due to random reaction events , was implemented by using Gillespie’s stochastic simulation algorithm [23] . The reaction steps and propensities of the stochastic model are listed in S4 Table . The reaction rate constants used in the stochastic model are identical to the values used in the deterministic model ( see S3 Table ) . Molecular concentrations in the deterministic model were converted into numbers of molecules by multiplying concentrations by system size ( Ω = 1000 ) , giving protein numbers in the 100–1000 molecule range . Parametric noise was attributed to variations in total protein levels from one cell to another , due perhaps to differences in the associated rates of transcription [24 , 25] . Cell size at division is influenced most sensitively by the expression of the fusion protein and of Rum1 ( simulations not shown ) , so we only considered parametric noise for these proteins . Using both intrinsic and parametric noise , we illustrate the expected variability in cell cycle progression in MCN , MCN Δrum1 , and MCN-AF cells in Fig . 6 . MCN Δwee1 Δmik1 cells behave similarly to MCN-AF cells , as expected ( simulations not shown ) . Note that if we consider smaller number of molecules , i . e . Ω = 500 or Ω = 200 , the cells are still viable . However , the cell size distribution is enlarged in each case and does not correspond to experimental observations from Coudreuse and Nurse [7] . Previous experimental and theoretical studies of cell cycle control have stressed the role of positive feedback loops in promoting robust oscillations in Cdk:cyclin activity [26–31] . Hence , it is no surprise that eliminating Cdc2 phosphorylation from the MCN strain , which eliminates the positive feedback loops at the G2/M transition , results in lower amplitude MPF oscillations and greater sensitivity to molecular noise ( compare Fig . 6A and B to Fig . 6C ) . The similarities between these results and the experimental data ( Fig . 6 right panels ) , showing increased variability in cell size at division in the MCN-AF cells [7] , suggest that molecular noise is an integral part of the cell cycle regulatory circuit . With this framework for modelling intrinsic and parametric noise in the minimal Cdk network , we then explored the robustness of our deterministic models of MCN-derived cells . In Figs . 7 and 8 we have repeated the deterministic simulations in Figs . 2 and 4 , respectively , in the stochastic setting . For five of the six genetic backgrounds , the deterministic model predicts that cells are viable: at cell division , MPF activity drops below θS long enough for a newborn cell to re-license its replication origins , then rises above θS to initiate DNA replication , then ( after a sufficiently long time period to complete DNA replication ) MPF rises above θM to initiate mitosis , and finally drops below θM to initiate mitotic exit and cell division . In each of these five cases , the stochastic simulations show the same global course of events ( Figs . 7 and 8 ) . Therefore , although stochastic fluctuations are expected to introduce considerable variability in cell cycle progression , the minimal Cdk network is sufficiently robust to maintain viability in each of these five mutant strains . In the sixth case , cdc13-L-cdc2AF CCP = 2 ( Fig . 4C ) , the deterministic model predicts that cells do not re-license their replication origins after cell division and die . The stochastic simulations of these cells show that they in fact encounter multiple problems ( Fig . 8C ) . In some cases , MPF activity does not drop low enough for origin re-licensing , a major source of mitotic catastrophe . But , whether or not origins are re-licensed , MPF activity rises so rapidly that these cells are likely to show a cut phenotype , as suggested by the “G1 reset” experiments in [7] . This supports our previous conclusions on the role of unregulated CCP-dependent MPF activity in promoting unconditional mitotic catastrophe and suggests that this phenomenon can be the result of different cellular events ( i . e . lack of origin relicensing or fast rise in Cdk activity ) .
Despite the apparent complexity of cell cycle regulation in eukaryotic cells , a minimal Cdk control network , consisting of an autonomous monomolecular cyclin-Cdk fusion protein , is sufficient to drive normal progression through the entire fission yeast cell cycle [7] . Here , we propose a computational model for this control network ( Fig . 1 ) based on the Cdc13-L-Cdc2 fusion protein ( referred to as MPF ) and on the notion of “quantitative” control of DNA synthesis and mitosis: MPF initiates DNA replication when its activity exceeds a low threshold ( θS ) , and the same MPF activity initiates mitosis when it exceeds a high threshold ( θM > θS ) . Our model conforms to previous experimental studies supporting the idea that quantitative–rather than qualitative–changes in Cdk:cyclin heterodimer activity orchestrate the sequence of cell cycle events [1 , 4 , 6 , 7 , 32] . Model simulations recapitulate all published phenotypes of mutant fission yeast strains based on the MCN genetic background ( cdc13-L-cdc2 Δcdc13 Δcdc2 ΔCCP ) . In particular , our model accounts for the unexpected phenotypes of MCN cells lacking Cdc2-inhibitory phosphorylation ( i . e , MCN cells deleted for wee1 and mik1 , or MCN cells carrying the cdc13-L-cdc2AF fusion cassette ) . To explore the roles of Cdc2 activity associated with alternative cyclins ( Cig1 , Cig2 and Puc1; collectively referred to as CCP ) on the timing of mitosis , we modified the model slightly and compared the model’s predictions with a set of new minimal strains we generated . Cdc2:CCP complexes were assumed to contribute a background Cdc2 kinase activity in the model equations , specifically targeting Rum1 for degradation . Without this background activity , MCN-AF cells ( cdc13-L-cdc2AF Δcdc13 Δcdc2 ΔCCP ) divide at wild type size , but with this background activity cdc13-L-cdc2AF Δcdc13 cdc2+ CCP+ cells are “wee” ( i . e . divide smaller than wild type cells ) . This CCP-dependent Cdc2 activity may therefore be at least partly responsible for the smaller size at division of cells with reduced Cdc2 inhibitory phosphorylation in an otherwise wild type background . This led us to propose a new explanation for the mitotic catastrophe that characterizes wild type cells entirely devoid of Cdc2 inhibitory phosphorylation . Our results–that the differences between cdc2AF cells and MCN-AF cells are consequences of unregulated CCP-dependent Cdc2 activity–suggest that the mitotic catastrophe observed in cdc2AF cells may come from excessive accumulation of Cdc2AF:CCP complexes in addition to Cdc2AF:Cdc13 , a hypothesis that we have experimentally validated ( Fig . 5 ) . In the MCN background , Cdc13-L-Cdc2AF is not sufficient to induce mitotic catastrophe , because the lack of inhibitory phosphorylation is counteracted by an elevated Rum1-dependent inhibition of the fusion protein . Our data suggest that this is also the case in a wild type background , and therefore , the Cdc2AF:Cdc13 complex represents only a necessary but not sufficient condition for mitotic catastrophe; entry into mitosis before completion of S phase also involves lack of inhibitory phosphorylation on Cdc2:CCP complexes , which down-regulate the levels of Rum1 . This creates a catastrophic situation because replication licensing becomes compromised while entry into mitosis is advanced . Our model provides a mechanistic explanation for different physiological consequences of lack of inhibitory Cdc2 phosphorylation in wild type and MCN backgrounds ( Fig . 9 ) . When the Cdc2 inhibitory phosphorylation network is intact ( Wee1+Mik1 = 100% ) , it determines the critical size for the G2/M transition in both genetic backgrounds , and the S phase size control is cryptic [17] . Subsequently , reduction in Wee1+Mik1 activity leads to a decrease in cell size at division and in cell size at the G1/S transition . In a wild type background ( with CCP activities ) , decreasing Wee1+Mik1 activity causes a decline in cell size at division down to 40% of normal . However , for Wee1+Mik1 < 10% , Cdc2 activity does not drop low enough for efficient licensing of DNA replication origins , and therefore cells divide with catastrophic consequences . In contrast , in MCN cells , as Wee1+Mik1 is reduced , the size at division falls to a minimum of 67% of normal , which is reached at Wee1+Mik1 = 20–30% . Further reduction of Wee1+Mik1 activity leads to an increase in the critical size at the G1/S transition caused by persistence of Rum1 and efficient Rum1-dependent inhibition of the unphosphorylated fusion protein , accompanied by a corresponding increase in cell size at division . Therefore , lack of CCP-dependent activities in the MCN background provides cells with the opportunity to switch from the traditional size control at G2/M to an S phase size control mechanism , thereby avoiding mitotic catastrophe . Stochastic simulations of the model show that MCN-AF and MCN Δwee1 Δmik1 cells are more sensitive to molecular noise than cells of the parental MCN strain ( compare Fig . 6C with Fig . 6A ) . These results support the idea that positive feedback loops at the G2/M transition are critical to generate robust oscillations of MPF . When the positive feedback loops are abrogated ( by the cdc2AF allele or by deletion of the inhibitory kinases ) , populations of MCN cells show a broader distribution of cell size at division both experimentally and in simulations . In the presence of both Cdc13- and CCP-dependent Cdc2AF activities , our model suggests that molecular noise induces a range of phenotypes that are consistent with the lethality of Cdc2AF in a wild type background , from cells that do not re-license their replication origins to mitotic catastrophe resulting from too fast rise in Cdk activity ( Fig . 8C ) . The model proposed here is based on quantitative regulation of the cell cycle , where increasing activity of a single Cdk:cyclin complex drives orderly progression through the successive phases of the DNA replication-division cycle . Our computational view of the quantitative model is supported by experimental evidence from mutant fission yeast cells that rely on a single cyclin-Cdk fusion protein to drive the cell cycle . This situation contrasts with wild type yeast cells ( and cells in higher eukaryotes ) , where different Cdk activities are thought to drive specific cell cycle events . Such “qualitative” regulation of the cell cycle has been modeled in earlier publications [31 , 33–41] . Although quantitative regulation of the cell cycle may appear to be an unrepresentative property of an unnatural strain of yeast cells , the absence of phenotype in the MCN strain suggests that primeval eukaryotic cells may have controlled their cycle of DNA replication and mitosis using similarly simple mechanisms based on a single protein kinase activity . Later in the evolution of eukaryotes , additional components of the control network may have been introduced to improve its fitness , for instance by increasing robustness through redundancy . Nonetheless , even in higher eukaryotes , simplified versions of the cell cycle control system , with some of this redundancy removed , can still function reliably [42 , 43] .
To develop a mathematical model of the minimal Cdk network , we assume that the fusion protein is regulated similarly to the Cdc2:Cdc13 heterodimeric complex in wild type cells , which we have previously modeled [44–47] . That is , we assume that the activity of Cdc13-L-Cdc2 is controlled by 1 ) inhibitory phosphorylation of the Cdc2 moiety , 2 ) proteolytic degradation of the fusion protein mediated by the cyclin destruction box , and 3 ) binding of the stoichiometric Cdk inhibitor Rum1 ( Fig . 1 ) . For simplicity , although the Cdc13-L-Cdc2 fusion protein and the normal Cdc2:Cdc13 heterodimeric complex have SPF as well as MPF activities , we shall refer to both as MPF ( M-phase Promoting Factor ) , because only Cdc13-dependent Cdc2 activity can bring about M phase in fission yeast [4] . Inhibitory phosphorylation of MPF by Wee1 and Mik1 ( on T14 and Y15 of Cdc2 ) results in the accumulation of a less active form of MPF , labeled here as MPFP ( Fig . 1 , S/G2 module ) . We assume that the activity of MPFP is 5% of MPF activity [48 , 49] . For our study , the distinction between Wee1 and Mik1 is not necessary , and we therefore refer to them together as Wee1 ( hence , Δwee1 in our simulations is equivalent to Δwee1 Δmik1 in the experiments ) . MPF inhibitory phosphorylations are antagonized by a protein phosphatase , Cdc25 , that dephosphorylates T14P and Y15P of Cdc2 and thereby releases MPF activity [12] . Importantly , feedback loops are built into the system , as the activities of Wee1 and Cdc25 are themselves regulated by MPF ( and to a lesser extent by MPFP ) . As indicated in Fig . 1 ( S/G2 module ) , Wee1 is phosphorylated and inactivated by MPF , whereas Cdc25 is phosphorylated and activated by MPF , thus establishing a double-negative feedback loop ( Wee1⊣ MPF⊣ Wee1 ) and a positive feedback loop ( Cdc25 → MPF → Cdc25 ) . These two feedback loops create a bistable switch that is responsible for abrupt MPF activation at the G2/M transition [50] . At the end of M phase , Cdc13 is ubiquitinated by the Anaphase Promoting Complex ( APC , also known as Cyclosome ) , which tags the fusion protein of the minimal module for rapid proteasomal degradation ( Fig . 1 , M module ) . Initial degradation of the fusion protein is mediated by the APC in conjunction with Slp1 ( the fission yeast orthologue of Cdc20 ) [51] . This system introduces a negative feedback loop , as MPF promotes its own degradation by activating APC:Slp1 , a feature that is common to all eukaryotes . Both theoretical arguments [52] and experimental evidence [53 , 54] suggest that this negative feedback loop is time-delayed , but the underlying molecular mechanism of this delay is unclear . For this study , we generate a time delay by inserting an intermediary enzyme ( IE ) between MPF and APC:Slp1 , as in earlier models [50] ( see Eq . ( 4 ) in S2 Table ) . We use Goldbeter-Koshland kinetics [55] to describe the ultrasensitive activation and inactivation of enzymes controlling the phosphorylation of MPF ( Wee1 and Cdc25 ) and the degradation of Cdc13 ( IE and APC ) . ( See equations ( 3 ) , ( 4 ) , ( 8 ) and ( 9 ) in S2 Table . ) In G1 phase , MPF activity is kept low by a stoichiometric Cdk-inhibitor , Rum1 , which binds to Cdc13-L-Cdc2 and blocks the catalytic activity of the fusion protein ( Fig . 1 , G1 module ) . MPF activity is also kept low by APC-dependent degradation of the fusion protein , in conjunction with Ste9 ( fission yeast orthologue of Cdh1 , also known as Srw1 ) [18] . Since Rum1 and Ste9 are both active during G1 phase of the fission yeast cell cycle , we lump them together as G1 Cdk-inhibitory activity ( in Fig . 1 ) , attributing G1-specific Cdc13 degradation to Rum1 [21] . Not only does the Cdk-inhibitor Rum1 bind to Cdc13-L-Cdc2 and quench its activity [56] , but Rum1 is also phosphorylated by active Cdc2 on multiple sites , targeting it for rapid ubiquitin-dependent degradation [57] . Therefore , Rum1 is both an inhibitor and a substrate of Cdc13-L-Cdc2 . Elsewhere , we have argued [16] that this relationship between Rum1 and MPF is characterized by a particular network motif ( SIMM = Substrate Inhibitor Multiply Modified ) that generates an abrupt G1/S transition in eukaryotic cells . In Fig . 1 ( G1 module ) , we implement the SIMM motif for distributive , two-step phosphorylation of Rum1 by MPF . In wild type cells , G1 phase is short , presumably because the CCP cyclins ( Cig1 , Cig2 and Puc1 ) , in combination with Cdc2 , effectively phosphorylate Rum1 and Ste9 , leading to their rapid degradation and inactivation , respectively . These “starter kinase” activities induce the SIMM motif to undergo an irreversible transition from G1 into S phase . In MCN cells , which lack CCP-dependent Cdc2 activity , the role of starter kinase must be played by MPFP ( i . e . , the tyrosine-phosphorylated form of Cdc2-L-Cdc13 ) . This conclusion suggests that MPFP is not effectively inhibited by Rum1 and that MPFP , despite its feeble kinase activity ( compared to MPF ) , is able to phosphorylate Rum1 and mark it for degradation . To put all these ideas together in a consistent fashion , we have made a number of assumptions in writing the differential equations in S2 Table and choosing the parameter values in S3 Table . In Eqs . ( 5 ) – ( 7 ) in S2 Table , we have implemented a SIMM motif for the distributive , two-step phosphorylation of Rum1 by MPF . In this scheme , the first phosphorylation of Rum1 by MPF is described by a Michaelis-Menten mechanism with tight binding of Rum1 to the kinase [57 , 58] and slow phosphorylation of the enzyme-bound substrate ( MPF:Rum1 ➔ MPF + Rum1P ) . In this case , the Michaelis constant of the enzyme ( Cdc13-L-Cdc2 ) is small , Km = ( kDISS + kIRUM1 ) /kASS ≈ 0 . 02 << [Rum1]total ≈ 0 . 4 ( in G1 phase ) , and the turnover number of the enzyme-substrate complex , kIRUM1 = 2 min−1 , is small compared to the dephosphorylation of Rum1P , kARUM1 = 35 min−1 . Hence , most of the enzyme ( MPF ) is tied up in the enzyme-substrate complex . In this context , Rum1 reduces the availability of Cdc2 for phosphorylating other substrates ( i . e . , Rum1 is a stoichiometric inhibitor of Cdc2 ) . After the first phosphorylation , Rum1P can be either dephosphorylated by a phosphatase ( Rum1P ➔ Rum1 ) or undergo a second phosphorylation by MPF ( Rum1P ➔ Rum1P2 ) , followed by rapid degradation ( Rum1P2 ➔ degradation ) . According to the SIMM concept , the phosphorylation of Rum1P by MPF has a large Michaelis constant ( Km2 >> [Rum1]total ) and a large turnover number ( k2 >> kdissociation ) . Therefore , the phosphorylation of Rum1P can be described by second-order mass-action kinetics ( kDRUM1P = 250 CU−1 min−1 , where CU = concentration unit for MPF and Rum1 ) , neglecting the concentration of enzyme-substrate complexes . Regarding the phosphorylation of Rum1 by MPFP , we assume that both the first and the second phosphorylation steps are governed by second-order mass-action kinetics . Even though MPFP is a less active kinase than MPF , it must be more efficient at phosphorylating Rum1 , because the phenotype of the MCN strain suggests that MPFP is an effective starter kinase . How can this be ? As a possible explanation ( in the absence of any experimental evidence one way or another ) , we suggest that the phosphorylation of Rum1 by Cdc2 kinase in the enzyme-substrate complex is a two-step process . When Rum1 first binds to Cdc2 , the kinase is unable to phosphorylate Rum1 . The complex must undergo a conformational transition before the kinase can do its job . Hence , the turnover number for Rum1 phosphorylation by MPF can be written k2 = f∙kp , where f is the fraction of Rum1:MPF complexes in the phosphorylable form and kp is the probability per unit time that MPF carries out the phosphorylation of Rum1 in this form . According to S3 Table , k2 ≡ kIRUM1 = 2 min−1 . The Michaelis constant for this reaction is Km ≈ f∙kp/kASS = 0 . 02 CU , because kDISS is very small ( according to S3 Table ) . For MPFP-catalyzed phosphorylation of Rum1 , the turnover number is k2’ = f ’∙kp’ ≈ α∙kp , where α = 0 . 05 = fractional activity of MPFP compared to MPF and we have assumed that f ’ ≈ 1 for the Rum1:MPFP complex . The Michaelis constant for the MPFP-catalyzed reaction is Km’ ≈ α∙kp/kASS’ , assuming that kDISS is also very small for the Rum1:MPFP complex . Assuming that Km’ >> 1 CU , we assure that the first phosphorylation of Rum1 by MPFP follows second-order mass-action kinetics , with rate constant = k2’/Km’ = kASS’ ≡ kI2RUM1 = 50 CU−1 min−1 . All these conditions can be satisfied if f << α/25 = 0 . 002 . In addition , we need a mechanism for coupling the minimal Cdk network to cell growth . In previous models [44 , 46] , we have implemented this idea by letting the effective concentration of Cdk:cyclin complexes increase with cell size . The idea behind this assumption is that the number of cyclin molecules increases steadily as cells grow , as is true for most cellular proteins , and that Cdk:cyclin complexes then move into the nucleus where their concentration at local defined sites of action increases in proportion to cell size . Note that changes in the total amount of Cdc13-L-Cdc2 throughout the cell cycle do not simply reflect cell mass increase due to the cell cycle-dependent regulation of Cdc13-L-Cdc2 amount by APC and Rum1 . For simplicity , we assume that cell size increases exponentially and is divided in half at cell division ( when MPF activity drops below 0 . 2 , in the arbitrary units adopted by the model ) . Finally , each phosphorylation reaction in the model is reversed by a dephosphorylation step catalyzed by a phosphatase ( Fig . 1 ) , and we assume that these phosphatases have constant activities . This assumption is clearly an oversimplification , because some of these phosphatases are known to be cell cycle regulated [59] . All simulations were performed by means of the software packages XPPAUTO ( http://www . math . pitt . edu/~bard/xpp/xpp . html ) and MATLAB . Stochastic simulations . At birth , each cell is assigned a value for kSMPF and VSRUM1 from a uniform distribution centered on the deterministic values given in S3 Table . More precisely , kSMPF = ( 1 + σSMPF*r ) *0 . 05 , where σSMPF is a constant ( 0 < σSMPF < 1 ) , r is a uniform random deviate on [−1 , 1] , and kSMPF = 0 . 05 is the deterministic value . VSRUM1 is computed from a similar equation , with σSRUM1 determining the cell-to-cell variations in the rate of synthesis of Rum1 . To estimate σSMPF and σSRUM1 , we performed a series of stochastic simulations for different values of these parameters , in each case calculating the distribution of cell size at division for MCN and MCN-AF cells . Comparing these simulated distributions to the observed distributions in Fig . 5C of Coudreuse & Nurse [7] , we conclude that σSMPF = 0 . 1 and σSRUM1 = 0 . 25 . These simulations suggest that production of Rum1 is considerably more variable from cell to cell than production of the fusion protein . Strains and growth conditions . Standard methods for fission yeast manipulation were used [60 , 61] . Strains described in this study are listed in S5 Table . Experiments were carried out in minimal medium plus supplements ( EMM4S ) at 32°C , except when otherwise indicated . The different Cdk fusion modules and gene deletions are as previously described [7] . The deletions of the G1/S cyclins in Fig . 5 are full deletions of the ORFs of the genes using antibiotic resistance cassettes ( see S5 Table ) . The synthetic lethality of the MCN-AF cells in combination with deletion of the Cdc2 inhibitor Rum1 was determined through genetic crosses . MCN-AF cells were crossed with Δrum1 cells . Given that either Δcdc2 or Δcdc13 in the absence of the fusion protein results in lethality and that the AF mutation gives rise to a generally higher frequency of non-germinating spores , we could not assess the Mendelian segregation of alleles . Instead , we identified clones deleted for cdc13 ( Kanamycin resistance ) after tetrad dissection of the cross . These cells must carry the AF fusion protein to be viable , which was verified by PCR . Those strains were then screened by PCR for the presence of the rum1 deletion . None of the 61 clones tested carried both the AF and Δrum1 alterations . Next , to address the possibility that the MCN-AF Δrum1 strain grows poorly and shows a higher incidence of cell death , random spore analysis of the above cross was performed in the presence of Phloxin B , which is retained in unhealthy and dead cells . Phloxin B-stained small red clones which were deleted for cdc13 and contained the AF fusion were genotyped by PCR for Δrum1 . None of the 48 colonies tested were positive for the rum1 deletion . Finally , we isolated diploid cells carrying at least one copy of the AF fusion cassette and heterozygous for Δrum1 . Following sporulation of this strain , we identified 45 haploid clones containing the cdc13 deletion and AF fusion , and none of them were deleted for rum1 . In total , we have tested 154 haploid strains carrying Δcdc13 and the AF fusion , and none of these contained the rum1 deletion . As we had no difficulty generating a strain with both the normal fusion protein and Δrum1 ( which are approximately 0 . 8 Mb apart ) , we conclude that the AF fusion cassette and rum1 deletion are synthetically lethal . Microscopy . For cell size measurement and live cell imaging , cells were stained with Blankophor ( MP Biohemicals ) . Images were then acquired with Visiview ( Visitron Systems GmbH ) using a Nikon Eclipse Ti epifluorescence microscope and a Hamamatsu Orca Flash 4 . 0 camera . Cell size was determined with ImageJ ( National Institutes of Health ) using the Pointpicker plug-in . For DAPI + Blankophor staining , cells were heat-fixed on microscope slides , stained with a DAPI/Blankophor solution ( 1:4 ) and imaged with Visiview ( Visitron Systems GmbH ) using a Zeiss Axio Observer microscope equipped with a Hamamatsu Orca Flash 4 . 0 camera . Flow cytometry . DNA content analysis within a population of cells was performed by flow cytometry using 70% ethanol-fixed and propidium-iodide-stained cells ( 2mg/ml PI in 50 mM sodium citrate after treatment with RNAse A ) and a BD FACSCalibur or Accuri C6 flow cytometer . | The eukaryotic cell division cycle is driven by fluctuating activities of cyclin-dependent kinases ( Cdk ) , which are activated and inactivated by several mechanisms , including cyclin synthesis and degradation . Although the cell cycle is driven by many different Cdk-cyclin complexes in present-day eukaryotes , experiments with fission yeast demonstrate that a single Cdk-cyclin complex is sufficient to order the events of the cell cycle . Surprisingly , a Cdk-inhibitory mechanism working through tyrosine phosphorylation of the kinase subunit , which is essential for modern fission yeast , becomes dispensable in the Minimal Cdk Network ( MCN ) . By developing both deterministic and stochastic models of the MCN , we show that a different inhibitory mechanism based on a stoichiometric Cdk inhibitor ( called Rum1 ) can compensate for the lack of inhibitory Cdk phosphorylation in the MCN . We also demonstrate that this compensation mechanism is suppressed in wild-type fission yeast cells by the other Cdk-cyclin complexes , which down-regulate the level of Rum1 . These predictions of computational modelling are supported by our experimental data . Our work provides new insights into the interplay between the structure of the control network and the physiology of the cell cycle . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Cell Cycle Control by a Minimal Cdk Network |
It is generally believed that associative memory in the brain depends on multistable synaptic dynamics , which enable the synapses to maintain their value for extended periods of time . However , multistable dynamics are not restricted to synapses . In particular , the dynamics of some genetic regulatory networks are multistable , raising the possibility that even single cells , in the absence of a nervous system , are capable of learning associations . Here we study a standard genetic regulatory network model with bistable elements and stochastic dynamics . We demonstrate that such a genetic regulatory network model is capable of learning multiple , general , overlapping associations . The capacity of the network , defined as the number of associations that can be simultaneously stored and retrieved , is proportional to the square root of the number of bistable elements in the genetic regulatory network . Moreover , we compute the capacity of a clonal population of cells , such as in a colony of bacteria or a tissue , to store associations . We show that even if the cells do not interact , the capacity of the population to store associations substantially exceeds that of a single cell and is proportional to the number of bistable elements . Thus , we show that even single cells are endowed with the computational power to learn associations , a power that is substantially enhanced when these cells form a population .
Almost all animals can associate neutral stimuli and stimuli of ecological significance [1] . An extensively studied example is eye-blink conditioning ( Figure 1 ) [2] , [3] . Naïve rabbits respond to an airpuff to the cornea ( Unconditioned Stimulus , US ) with eyelid closure ( Unconditioned Response , UR ) . By contrast , a weak auditory or visual stimulus ( Conditioned Stimulus , CS ) does not elicit such an overt response . Repeated pairing of the CS and the US forms a cognitive association between the CS and the US such that the trained animal responds to the CS with eyelid closure , a response known as Conditioned Response ( CR ) . Two important characteristics of associative learning are ( 1 ) specificity and ( 2 ) generality . The CR does not reflect a general arousal . Rather , the animal learns to respond specifically to the CS . The generality is reflected by the fact that a large family of potential stimuli can serve as a CS if paired with the US . Neuronal networks are particularly adapted to performing this association and in the last few decades there has been considerable progress in understanding the ways in which experience-based changes in synapses in the nervous system underlie this associative learning process [4] , [5] . Neural network models for associative memory , which explain how both specificity and generality are maintained , are typically based on three elements: ( 1 ) Synapses are the physical loci of the memory; ( 2 ) synaptic plasticity underlies memory encoding; ( 3 ) neural network dynamics , in which the activities of neurons depend on the synaptic efficacies , underlie the retrieval of the learned memories in response to the CS . Genetic regulatory networks ( GRN ) describe the interaction of genes in the cell through their RNA and protein products [6] , [7] , [8] . Previous studies have pointed out the similarity between the dynamics of GRNs and the dynamics of neural networks [9] . For example , GRNs , like neural networks , can implement logic-like circuits , where the concentration of a protein ( high or low ) corresponds to the binary state of the gate [10] , [11] , [12] . These findings prompted us to evaluate the capacity of GRNs to learn associations . Considering associative learning in animals , the US is typically a stimulus of biological significance , such as food or a noxious stimulus that elicits a response ( UR ) in the naïve animal , either in the form of muscle activation or gland secretion . The GRN correlate of a pain-inducing stimulus is stress . Stressful conditions such as heat , extreme pH , or toxic chemicals often result in a substantial change in the expression level of many different proteins in the cell . For example , Escherichia coli ( E . coli ) bacteria respond to a variety of stress conditions by a general stress response mechanism in which the master regulator controls the expression of many genes [13] . These stressful conditions can be regarded as a US and the resultant change in the expression level of the proteins can be regarded as a UR . By contrast , other stimuli may result in a narrow or absence of a response of the cell and in that sense can be referred to as potential CS . Learning in this framework would correspond to the formation of an association between these potential CS and US such that following the repeated pairing of the CS and US , the presentation of the CS would elicit a UR-like response ( CR ) . The responsiveness of the GRNs to different stimuli has been shown to change over time in response to evolutionary pressure in a manner that resembles associative learning [14] , [15] . These changes take place on time scales that are substantially longer than the lifetime of a single cell and in contrast to associative learning in animals , entail modifications of the genome through mutations . On a shorter timescale , there is some evidence that the single-celled Paramecium can learn to associate a CS with a US within its lifetime [16] . However , these findings have been disputed [17] and the question of whether Paramecia can learn associations and the characteristics of this learning await further experimental validation . The capacity of GRNs to learn associations in shorter , non-evolutionary time-scales has also been studied theoretically using GRN models . Learning in these models is restricted to a small subset of predefined stimuli [18] , [19] , [20] , [21] and thus the computational capabilities of these GRN models are limited compared to neural network models . Here we show that a GRN based on bistable elements and stochastic transitions can learn associations while retaining both specificity and generality . We further compute the capacity of the network and show that the number of different learned associations that the network can simultaneously retain is proportional to the square root of the number of bistable elements . Moreover , this capacity is substantially enhanced when considering a clonal population of GRNs . These results imply that even bacteria are endowed with the capacity to learn multiple associations .
In the previous section , we demonstrated that a GRN can learn to associate a CS with a US ( Figure 3 ) . However , this learning is limited , as it is specific to a single , predefined CS . This GAM can be trivially generalized to enable the learning of several different associations by postulating that the GAM is characterized by a number of memory elements , each associated with a single CS . However , this generalized GAM is still limited in its ability to learn associations because only those predefined CS can be learned . This limitation contrasts sharply with neural network models , which are capable of learning general associations . In this section we generalize the model presented in Figure 3A and show that similar to neural network models , GRNs are also endowed with the capacity to learn a large number of arbitrary , overlapping CS . Consider the network described Figure 4A . In contrast to the single-pathway model ( Figure 3A ) , in which a CS induces the expression of a single protein C , in the generalized model we assume that the CS are complex stimuli that activate N different receptors , , . Each receptor is associated with a single pseudo-synapse . The dynamics of each of the pseudo-synapses follow the same equations as in the single-pathway model ( not shown in Figure 4A , see Eq . ( 4 ) in the Materials and Methods ) . The last component of our generalized GAM is the readout scheme . We assume that similar to the single-pathway model , the UR and the CR manifest in the generalized model as the production of a response protein R . We assume two independent promoters that regulate the expression of R . The response to the presentation of the US is described by ( Eq . ( 3 ) ) and the response to the presentation of the CS is regulated by the cooperative binding of and , where different pairs of and independently regulate R ( Eq . ( 5 ) in the Materials and Methods and Text S1 in Supplementary Information ) . For simplicity , we assume in our analysis that the patterns of expression of the proteins that define the stimuli are random and independent . In this case , the statistics of the stimuli are fully determined by the sparseness of the stimuli , the probability that is in its high expression level , . To gain insights into the ability of the generalized GAM to learn multiple associations , we consider a naïve GAM , in which the values of the pseudo-synapses are random ( Figure 4B , bottom , t = 0 ) . The responses of the GAM to five different stimuli , denoted by A , B , C , D and E , presented to the GAM at times t = 0 , 1 , 2 , 3 and 4 h , respectively , are relatively small . This is due to the random , and hence relatively small overlap between the pattern of activation of pseudo-synapses ( color coded ) and the pattern of activation of the receptors of the five stimuli ( is denoted by an open blue rectangle in Figure 4B ) . In response to the pairing of C , B and A with the US ( at times t = 5 , 6 and 7 h , respectively ) , the expression levels of some of the become more similar to that of the in A , B and C , respectively . As a result , the GAM responds more vigorously to the presentation of A , B and C ( at times t = 8 , 9 and 10 h , respectively ) but not to the presentation of D or E ( at times t = 11 and 12 h , respectively ) . However , as a result of a repeated association of pattern E with the US ( at times t = 13 , 14 and 15 h ) , the GAM vigorously responds to the presentation of pattern E ( at time t = 17 h ) but not to pattern D ( at time t = 16 h ) . This example demonstrates that a GAM can selectively learn to associate several arbitrary CS patterns with a US . A careful analysis of Figure 4B reveals that after learning , the magnitude of the responses to the three learned CS is not equal . The response to stimulus C ( t = 10 h ) is smaller than the response to stimulus B ( t = 9 h ) and the response to B is smaller than the response to A ( t = 8 h ) . This difference reflects the fact that the order of association affects the magnitude of the response to a CS . This is because learning a new pattern may change the expression level of a pseudo-synapse that participates in the encoding of an older pattern . For example , consider pseudo-synapse 4 in Figure 4B . In response to the presentation of stimulus C ( at time t = 5 h ) , the state of the pseudo-synapse has changed to the high expression level , in line with the expression level of in CS C . However , the association of the US with A ( at time t = 7 h ) has reverted the state of the pseudo-synapse to the low expression level , decreasing the overall response to the CS C . In other words , the association with the CS A has overwritten the information stored in pseudo-synapse 4 concerning the CS C . More generally , because of the overwriting of memories by more recent memories , the magnitude of response to a CS is expected to decrease with the number of subsequent CSs . After the encoding of a large number of patterns , the response to an ‘old’ CS is expected to diminish to an extent where it is no longer distinguishable from the response to non-learned stimuli . In this case the CS is said to have been extinguished ( a more precise definition of “distinguishable” appears below ) . By contrast to the diminishing of the response to a pattern following the overwriting by other patterns , the repeated co-occurrence of the same pattern with the US ( at times t = 13 , 14 and 15 h ) augments the strength of association of that pattern with the US , as demonstrated by the response to pattern E at time t = 17 h . The magnitude of the order effect depends on two probabilities: the probability p that the co-occurrence of U and a high level of expression of would induce a transition from to in the corresponding pseudo-synapse and the probability q that the co-occurrence of U and a low level of expression of the corresponding would induce a transition from to in the corresponding pseudo-synapse . The probabilities p and q are determined by the two rates of the US-induced transitions and the duration of co-occurrence of the US and CS , T ( assuming that the rates of all other transitions are negligible , see above ) such that and , where and are the low-to-high and high-to-low transition rates , respectively . The larger the transition rates and the longer the duration , the larger the transition probabilities are . If all pseudo-synapses are determined by the most recent CS and the pattern of expression level of the different pseudo-synapses corresponds to the pattern of activation of the receptors in that CS . As a result , the response to the most recent CS is substantially larger than the response to a non-learned stimulus . However , this comes at a price . The most recent CS overwrites the memory trace of all previously encoded CS and therefore the responses to all these ‘older’ CS are indistinguishable from the responses to the non-learned stimuli . Thus , if , the GAM cannot store more than a single association . The smaller the values of p and q ( e . g . , due to smaller US-induced transition rates ) , the fewer pseudo-synapses change in the process of learning a CS , allowing the GAM to maintain information about previously-learned CSs . However , the transition probabilities should not be too small because the smaller these probabilities are , the weaker is the encoding . If these probabilities are too small , the response of the GAM even to the most recently stored GAM is too small to be distinguishable from non-learned stimuli . Therefore , in order for the GAM to be able store a large number of CS , the values of the US-induced transition rates should be sufficiently large to allow for a sufficiently large response to the learned-CS but sufficiently small to minimize the overwriting of old memories by new memories . To better understand the requirement that the response to a CS needs to be distinguishable from the response to non-learned stimuli , consider again Figure 4B . The responses of the GAM to the presentations of the non-learned stimuli A-E at times 0–5 h , respectively , are not identical . These differences are due to the fact that there is stochasticity in the response , resulting from stochasticity in the dynamics of the pseudo-synapses and in the realization of the different CS . Therefore , a memory of a CS is said to be maintained if the distribution of the responses of the GAM to the CS is distinguishable from the distribution of responses to the non-learned stimuli . This notion becomes exact in the next section . How many CS can be stored in a GAM ? Addressing this question using the full dynamical equations ( Figure 4 ) requires extensive simulations that are beyond the scope of this paper . Therefore , we use a binary approximation ( see Materials and Methods ) . The quality of the binary approximation is demonstrated in Figure S1 in the Supporting Information . As described in the previous section , responses to non-learned stimuli depend on the overlap of the pattern of activation of the stimuli with the pattern of activation of the pseudo-synapses . Because both the stimuli and the dynamics of the pseudo-synapses are stochastic , this response is a stochastic variable . The distribution of the responses to non-learned stimuli ( see Eq . ( 14 ) in the Materials and Methods ) is depicted in Figure 5A ( blue ) . The response of the GAM to learned CS is also a stochastic variable . The distribution of responses to the most recently learned CS is depicted in Figure 5A ( black; Eq . ( 13 ) and in the Materials and Methods ) . This distribution is well-separated from the distribution of responses to the non-learned stimuli . Therefore , recently-learned stimuli are distinguished from non-learned stimuli using a simple threshold mechanism ( e . g . , the dashed line in Figure 4B ) . The probability of an error depends on the overlap between the two distributions . If the overlap is small , the GAM almost always responds to the most recently learned CS and almost never responds to non-learned stimuli . On the other hand , a large overlap would result in a large number of errors , false positives or misses , depending on the choice of threshold . The difference between the means of the two distributions ( black and blue ) depends on the transition probabilities . The higher the probabilities , the larger the difference is . Therefore , the higher the transition rates are , the easier it is to distinguish between the most recently learned CS and the non-learned stimuli . The distribution of responses to the presentation of the second-most recently learned CS ( darkest gray ) is also to the right of the distribution of responses to non-learned stimuli ( blue ) . Nevertheless , it is shifted to the left relative to the distribution of responses to the most recently learned CS ( black ) . As a result , the overlap of this distribution with the distribution of responses to the non-learned stimuli is larger . The reason for this shift is that as noted in Figure 4B , the newer CS ‘overwrites’ the memory of the older CS , resulting in a decreased overlap between the CS and the pseudo-synapses . The degree of overwriting , manifested as a shift to the left of the distribution of responses to the second-most recently learned CS relative to the most recently learned CS , depends on the US-induced transition rates . The smaller the transition rates , the smaller the overwriting is and therefore the smaller the shift to the left of the distribution . More generally , the distributions of responses to a CS shift to the left with the ‘age’ of the CS . This is depicted in Figure 5A using grayscale . While the distribution of the several most-recently learned CS is well-separated from the distribution of responses to non-learned stimuli ( blue in Figure 5A ) , the distributions of responses to ‘older’ CS and non-learned stimuli largely overlap , indicating that ‘older’ CS are ‘forgotten’ . More formally , the ability of the GAM to distinguish between a learned CS and a non-learned stimulus depends on the signal to noise ratio ( SNR ) , which is defined as the ratio of the difference in mean responses to the two classes of stimuli , divided by the square root of the sum of variances of the two distributions ( Eq . ( 14 ) in the Materials and Methods ) . In general , the larger the SNR , the fewer errors when distinguishing between learned and non-learned stimuli . The SNR , as a function of the ‘age’ of the CS is depicted in Figure 5B: the newer the CS , the larger the SNR . The SNR of the CS ( where the numbering of patterns is reversed such that corresponds to the most recent stimulus ) is given by Eq . ( 14 ) in the Materials and Methods section . The capacity of the GAM can thus be defined as the ‘oldest’ CS such that the corresponding SNR is larger than 1 . In other words , the capacity of the GAM is defined as the largest value of such that . The capacity of the GAM depends on the US-induced transition rates , which determine the transition probabilities . As discussed above , if these rates are high , forgetting is fast . On the other hand , if these rates are too low the GAM cannot reliably retrieve even the most recent CS . The capacity of the GAM is maximal when the US-induced transition rates are intermediate , balancing between these two requirements . The capacity of the GAM as a function of the number of pseudo-synapses ( N ) is depicted in Figure 5C ( blue ) . The larger N , the larger is the capacity of the GAM . In the Materials and Methods section we show that in the limit of , if the US-induced transition probabilities are optimal , the capacity of the GAM is proportional to the square root of the number of pseudo-synapses , ( Eq . ( 20 ) ; red line in Figure 5C ) . This result is similar to the memory capacity of models of neural networks with binary synapses [47] , [48] . However , the learning rule proposed here , even in the binary approximation , differs from the Hebbian synaptic plasticity rule used in neural network models [47] , [48] . In the previous sections we studied the ability of a single GRN to learn associations . However in nature , GRNs often do not reside in isolation but in populations comprising of a large number of individual cells of the same type , e . g . , as in a colony of bacteria or in a tissue , all exposed to the same external conditions . This raises an interesting question: is the capacity of a population of GAMs to store associations larger than that of a single GAM ? The answer is trivially positive if we allow the different GAMs to communicate and form a recurrent network with specialized connections between individual GAMs , similar to neurons in neuronal networks . However , here we ask a different question: is the capacity of a population of non-interacting GAMs to store and retrieve memories different from that of the single GAM ? We consider a population of generalized GAMs as in Figure 4A . All GAMs are identical , exposed to the same sequence of stimuli but differ in their internal stochasticity . In other words , the noise associated with the dynamics of the pseudo-synapses ( Eq . ( 2 ) ) in the different GAMs is assumed to be independent . The population response in our model is assumed to be simply the accumulated response of all individual GAMs . In order to understand why the capacity of a population of identical GAMs to store memories may be larger than the capacity of a single GAM , we note that a CS of a particular ‘age’ can be retrieved if the overlap between the distributions of responses to the learned and non-learned stimuli is sufficiently small . This overlap is sensitive to the variances of the two distributions ( width of the curves in Figure 5A ) . The larger the variance , the larger is the overlap . Two sources contribute to this variance in the responses . First , there is stochasticity in the realization of CS and non-learned stimuli . Second , there is stochasticity in the encoding process . While the first type of stochasticity is external and thus shared by all GAMs in the population , the second type of stochasticity is independent for each GAM . As a result , when considering the cumulative response of a large population of GAMs , all other parameters being equal , the variance in the distribution of responses is considerably smaller ( Eq . ( 23 ) in the Materials and Methods ) . In Figure 6A we plot the distributions of responses to CS of different ‘ages’ ( gray , color-coded ) and non-learned stimuli ( blue ) . Similar to the case of a single GAM , the capacity of a population of GAMs depends on the US-induced transition rates . However , because the variance in the responses in the case of the population is considerably smaller than the variance in the case of a single GAM , the US-induced transition probabilities that maximize the capacity of the population are considerably smaller than those that maximize the capacity of a single GAM . In Figure 6B we plot the SNR as a function of the ‘age’ of the CS ( solid blue line ) . Compared to the SNR of a single GAM ( dashed blue line , identical to Figure 5B ) , the SNR of the response of the population of GAMs is larger than 1 for much ‘older’ CS . The capacity of the population of GAMs as a function of the number of pseudo-synapses ( N ) is depicted in Figure 6C ( solid blue line ) . The larger N is , the larger is the capacity of the GAMs . More quantitative analysis reveals that for an appropriate choice of parameters , the number of different CS that a large population of GAMs can store is proportional to the number of pseudo-synapses ( Eq . ( 31 ) ; solid red line ) , compared to a capacity that is only proportional to the square root of the number of pseudo-synapses in the case of a single GAM ( dashed blue line , identical to Figure 5C ) .
In this paper , we explored the ability of a general GRN to encode associations . We showed that a GRN that is endowed with bistable elements and stochastic dynamics is capable of storing and retrieving multiple arbitrary and overlapping associations . The capacity of a single GRN in our model , defined as the number of stored associations , is proportional to the square root of the number of bistable elements . This result is reminiscent of Hopfield-like models with bounded synapses , in which the capacity is proportional to the square root of the number of synapses [47] , [48] , [49] . Remarkably , in a large population of GRNs , as is in a colony of bacteria or in a tissue , this capacity is substantially higher and is proportional to the number of bistable elements . Despite the similarities between the GAM and the Hopfield model , there are two important differences that are noteworthy . First , the capacity of a single GAM may be limited by the presence of readout noise ( e . g . , in the dynamics of R ) . However , this readout noise is not expected to substantially affect the capacity of a population of GAMs because of averaging . Second , the number of neurons available in neuronal networks is much larger than the number of bistable elements in GRNs . Altogether , our model predicts that if the number of bistable elements in the GRN does not exceed several tens , it will be difficult to store more than one or two memories in a single GAM . Therefore , the storage of multiple memories is likely to require a population of GAMs . The key elements in our model are the bistability and the stochasticity of the dynamics of the GRN . Importantly , bistability and stochasticity are not restricted to the transcriptional machinery . Rather , they are found in various cellular processes , including post-transcriptional regulation ( e . g . , by non-coding RNA [50] , [51] ) or post-translational regulation ( e . g . phosphorylation and degradation regulation [36] , [52] , [53] , [54] ) . We modeled associative memory that is based on the interaction of proteins through the transcriptional machinery because these dynamics are better characterized and are more accessible experimentally than other cellular alternatives . Moreover , the GAM is not restricted to a particular organism . The parameters used in the simulations presented in this paper are biologically plausible for bacteria . However , because the basic elements of the GAM , namely , bistability and stochasticity , are widespread in GRNs of all cells , the potential for associative learning without a nervous system exists for virtually all cell types , including single-celled eukaryotes and plants . Furthermore , this work suggests that even in animals that possess a nervous system , learning that is independent of this nervous system is also possible . In particular , it could be interesting to consider GAM in the immune system , which has evolved to learn to respond to novel pathogens . Bearing this in mind , we believe that in view of the recent developments of experimental methods that quantitatively measure the expression level of proteins , bacteria , in particular the well characterized E . coli , are the ideal substrate to study the associative learning in GRNs . Each of the components of the GAM module ( Figure 3A ) , namely inducible elements , bistable switches and AND gates , have been established in the E . coli transcription network and therefore a synthetic implementation is achievable [55] , [56] . Beyond synthetic implementation , the complexity of the genetic networks suggests that GAM-like modules may exist . A first step in searching for GAMs in known networks should be the identification of plausible candidates for the US , UR and CS . In animals , the US is a stimulus that causes an overt response prior to learning , the UR . Typically the US is a stimulus of biological significance , such as food or a noxious stimulus and the UR is an ecologically-relevant overt response , often in the form of muscle activation . For example , in the eye-blink conditioning experiment ( Figure 1 ) the US is an air puff and the UR is an eye blink that protects the eye from the puff . An important point to consider when searching for associative learning in bacteria is ecological significance . Our model for associative learning , similar to models of associative learning in neuronal networks , does not incorporate any ecological information about the stimuli . However in animals , it is known that the ability to form an association depends on the ecological relevance of the CS to the US . For example , the association of the taste of a certain food ( CS ) with the symptoms caused by a toxic or spoiled food ( US ) , known as taste aversion , is easily-formed after a small number of repetitions . By contrast , it is substantially more difficult to form an association of a tone with the same US [57] . It is generally believed that this difference results from the fact that typically , taste is more informative about the chemical composition of substances than auditory signals . Therefore , taste-aversion but not tone-aversion has evolved as a specific learning mechanism aimed at preventing the consumption of poisonous substances . Drawing an analogy to associative learning in bacteria , we propose to utilize ecologically-relevant CS rather than arbitrary CS when searching for associative learning in bacteria . In our model , the strength of association increases with the number of repetitions due to the stochasticity in the encoding process . Such dependence of the strength of association on the number of repetitions is also observed in classical conditioning experiments in animals [58] . Therefore , experiments involving a large number of co-occurrences of the CS and US are more likely to reveal associative learning in GRNs or populations of GRNs . Note that standard experiments studying responses of bacteria are typically short and do not involve repetitions in the presentation of stimuli to the same population of bacteria . Therefore , associative learning in such experiments may have been overlooked . Moreover , we have shown that the learning capacity of the population of bacteria is higher than that of a single GRN . Therefore , the experimental search for associative learning in bacteria should be done at the population level . More specifically in bacteria , the presence of foreign bacteria is a signal of potential stress . For example , many bacteria produce antibiotics that are harmful to other strains [59] . Other bacteria are sensitive to these damaging antibiotics and respond to their presence by activating a pre-wired stress response , such as the multiple antibiotics response ( MAR ) [60] . We thus suggest that the R gene in our scheme corresponds to one of the outputs of MAR response , e . g . the micF gene [61] . Note that similar to the blink in the classic eye-blink conditioning that protects the eye from the air puff ( Figure 1 ) , the activation of micF prevents the entry of the antibiotics into the cell . Thus , the antibiotics can be considered as a US whereas the stress response can be considered as a UR . However , the production of harmful antibiotics is not present in all bacteria species . Therefore , learning to distinguish between harmful and benign strains of bacteria is of potential great ecological significance because it may allow the bacteria to respond faster . Thus , the presence of foreign bacteria could correspond to the CS in our framework . Indeed , bacteria are able to detect secondary metabolites that are produced by other strains [62] . In that line , we suggest as a candidate for the M protein in the model the MarA gene . MarA is known to positively autoregulate itself , and thus has the potential to be bistable . In addition , the promoter of that gene contains multiple binding sites for transcription factors , allowing for complex regulation of the gene expression including the realization of AND gates . Experimentally , the UR can be measured using a fluorescent-based reporter that is regulated by a promoter of a stress response gene . The CS in this framework should be stimuli that can be sensed by the bacteria but do not elicit the stress response . These include a change in the concentration of different molecules that does not activate the stress response . Repeated exposure to such conditions can be controlled using a chemostat [63] , which can maintain selected growth conditions at a constant level while changing others . Finally , the benefit of the stress response at the population level can also be found in the induction of the MAR response , as it triggers the activation of genes that inactivate toxic compounds . The benefit of this “pooled response” for the population comes from the decrease in the concentration of the toxic compound [64] . Whether or not associative learning exists in GRNs on a time-scale much shorter than required for evolution is an open question . However , whether considering bacteria that can predict a stress condition or human digestive cells that can predict food intake , associative learning in single and populations of cells seems to have an evolutionary advantage . In view of the computational capabilities of GRNs demonstrated in this paper , we believe that future careful investigations will reveal the existence of associative learning in single and populations of cells .
In this section we provide the rate equations that describe the dynamics of the generalized model ( Figure 4A ) . The kinetic reactions that underlie these dynamics and the derivation of the rate equations from the kinetic reactions are described in Text S1 in the Supporting Information . The single pathway equations correspond to the generalized model with . The equations that describe the dynamics of the pseudo-synapses are given by: ( 4 ) where . The nonlinear positive feedback term , , is described by where the parameters are kinetic parameters , and n is the Hill coefficient , corresponding to the cooperativity of binding . The second term in Eq . ( 4 ) denotes the protein degradation , where is a parameter . The third term in Eq . ( 4 ) describe the effect of and U , where are kinetic parameters . The last term in Eq . ( 4 ) models the stochasticity of the dynamics , and we assume that are independent white noise such that , where is Kronecker's delta function such that if and otherwise and is a parameter . The reactive equation that describes the dynamics of R is given by: ( 5 ) where is the degradation rate of R , and where and are kinetic parameters . In this section we compute the capacity of the GAM to learn associations . To that goal , we consider a binary approximation of the dynamics of the pseudo-synapses . Because the dynamics of M spend most of the time near the attractors of the deterministic dynamics , Eq . ( 1 ) , it can be approximated using a two-state Markov chain , where each state corresponds to one attractor of the deterministic dynamics . We further assume that the US and CS are presented in discrete “trials” composed of a fixed period of time . Therefore , the response of M to the presentation of the CS and US can be approximated by: ( 6 ) where and such that and denote epochs in which and , respectively and and denote epochs in which and , respectively . The variables m and m′ denote the states of the pseudo-synapse before and after the presentation of the external cues and their values; 0 or 1 denote epochs in which and , respectively . The steady state response to the presentation of a pattern is: ( 7 ) The selectivity of the response in Eq . ( 7 ) depends on the value of the sum of . In response to the presentation of a CS that was learned CSs ago , ( 8 ) where ( 9 ) and ( see Text S1 in Supplementary Information for a more detailed derivation ) . Dissociating a learned pattern from non-learned patterns ( which we denote as ) is possible only if is significantly different from . The difficulty in dissociating learned and non-learned patterns lies in the fact that the responses to the two types of patterns are stochastic variables that depend on the stochasticity in the realization of the learned and non-learned stimuli as well as the stochasticity in the learning . Therefore , we consider the distribution of responses to the learned and non-learned stimuli . To compute the distribution of , note that in response to the presentation of a sequence of CS , changes in the state of the pseudo-synapses follow a Markov chain such that ( 10 ) From Eq . ( 10 ) it follows that at the stationary distribution , ( 11 ) where . Using Eq . ( 11 ) , and the fact that and , a straightforward calculation yields that the mean and variance of are given by: ( 12 ) where ( 13 ) Note that for large N is the sum of a large number of independent and identically distributed random variables and therefore according to the central limit theorem is normally distributed . In order to compute the capacity of the GAM , we define the difference between the mean responses to learned and non-learned stimuli as the signal and the square root of the sum of the variances of the responses to the learned and non-learned stimuli as the noise . In the limit of large N , the ability of a binary classifier to discriminate between the learned and non-learned stimuli depends on the SNR . If the SNR is large , it is possible to achieve a high detection rate while maintaining a low level of false positives . A low SNR implies that the two stimuli are indistinguishable . Therefore , we define the capacity of the GAM to be the oldest memory such that the SNR is larger than 1 ( see also [47] , [48] for a similar approach in models of neural networks ) . Formally , the signal-to-noise-ratio for a pattern presented patterns ago is given by: ( 14 ) where ( 15 ) We compute the capacity in the limit of large N and consider the effect of the scaling of p and q with N on the capacity of the GAM . If the values of p and q are very different then the pseudo-synapses will saturate . Therefore , we consider the same scaling of p and q , . The signal in Eq . ( 13 ) depends on the product of two terms , that depends on and a prefactor , that is independent of . It is easy to see that the prefactor , . Similarly , it is easy to see that . Therefore , . Because , a necessary condition for the SNR to be larger than 1 is . The term decays exponentially fast with . However , because , as long as , . Therefore , for , as long as , . Thus , for , the capacity of the GAM is , which is maximal for . In other words , assuming that , the capacity of the GAM is . To gain insights into this result , we consider the optimal choice of and ( which minimizes the variance ) , in which and . In this case , in the limit of large N , ( 16 ) and Eq . ( 15 ) becomes ( 17 ) Thus , Eq . ( 14 ) becomes: ( 18 ) The requirement that yields: ( 19 ) where we used the fact that for , and therefore . In order to find the values of and that maximize the capacity of the GAM , we compute the zeros of the partial derivatives of Eq . ( 19 ) with respect to and , resulting in and . Thus , the capacity of the GAM is ( 20 ) For , the capacity is . Note that the capacity increases as the value of f deviates from . In this section we compute the capacity of a large population composed of Z identical GAMs . The population response to the presentation of is given by ( up to constant shift and scaling ) : ( 21 ) where is the pseudo synapse ( ) in the GAM ( ) ( compare to Eq . ( 9 ) ) . Similar to the analysis of the capacity of a single GAM , we compute the mean and variance of . The computation of mean of in the case of the population is similar to that computation for the case of a single GAM , yielding ( 22 ) Note that is independent of the size of the population: since all GAMs are identical , their contribution , on average , is equal . Computing the variance of Eq . ( 21 ) results yields: ( 23 ) In order to evaluate Eq . ( 23 ) , we consider the dynamics of a single pseudo-synapse . Similar to Eq . ( 3 ) , ( 24 ) where and are Bernoulli variables with parameters p and q , respectively . Using induction , it is easy to prove that the value of in response to the learning of an infinite sequence of CS is given by: ( 25 ) where and and and denote the values of the Bernoulli variables and , respectively , during the encoding of the CS x patterns ago . Using Eq . ( 24 ) , it can be shown that: ( 26 ) where . Substituting Eq . ( 26 ) in Eq . ( 23 ) and assuming that yields: ( 27 ) Note that in the case of a single network , , only the first term contributes , yielding Eq . ( 16 ) . The capacity of the population of GAMs is defined as the oldest memory such that the SNR is larger than 1 , where the signal and noise terms in Eq . ( 14 ) are given by ( 28 ) And ( 29 ) In the limit of ( large number of GAMs ) the contribution of the first term in Eq . ( 26 ) to the variance vanishes and the capacity depends on the second term . For a general value of , and this term dominates , resulting in . Therefore , the population capacity in this case is , similar to that of a single GAM . However , if this term vanishes and is dominated by , resulting in . Similar to the case of a single GAM , we compute the capacity in the limit of large N and consider the effect of the scaling of p and q with N on the capacity of the population of GAMs . If the values of p and q are very different then the pseudo-synapses will saturate . Therefore , we consider the same scaling of p and q , . The signal in Eq . ( 27 ) is the same as that of a single GAM ( Eq . ( 13 ) ) , therefore the prefactor in Eq . ( 13 ) is . Similarly , it is easy to see that . Therefore , . Because , a necessary condition for the SNR to be larger than 1 is . The term decays exponentially fast with . However , because , as long as , . Therefore , for , as long as , . Thus , for , the capacity of the GAM is , which is maximal for . In other words , assuming that , the capacity of the GAM is . In particular , assuming that , substituting Eqs . ( 28 ) and Eq . ( 29 ) in Eq . ( 14 ) yields ( 30 ) A straightforward calculation reveals that the capacity is maximal when , resulting in capacity: ( 31 ) In our simulations , we used the following parameters: For the generalized model ( Figure 3 ) we used: All other parameters were the same as the single pathway model ( Figure 3 ) . The derivation of the parameters from the reaction kinetic constants is provided in the Text S1 in the Supporting Information . The reaction kinetic constants that were used are provided in Table S1 in the Supporting Information . Simulations in Figures 3 and 4 were carried out using Euler method for numerical integration with step sizes and 0 . 5 min , respectively . | It has been known since the pioneering studies of Ivan Petrovich Pavlov that changes in the nervous system enable animals to associate neutral stimuli with stimuli of ecological significance . The goal of this paper is to study whether genetic regulatory networks that govern the production and degradation of proteins in all living cells are capable of a similar associative learning . We show that a standard model of a genetic regulatory network is capable of learning multiple overlapping associations , similar to a neural network . These results demonstrate that even bacteria that are devoid of a nervous system can learn associations . Moreover , as cells often reside in large clonal populations , as in a colony of bacteria or in tissue , we consider the ability of a large population of identical cells to learn associations . We show that even if the cells do not interact , the computational capabilities of the population far exceed those of the single cell . This result is a first demonstration of “wisdom of crowds” in clonal populations of cells . Finally , we provide specific guidelines for the experimental detection of associative learning in populations of bacteria , a phenomenon that may have been overlooked in standard experiments . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"systems",
"biology",
"neural",
"networks",
"computational",
"neuroscience",
"regulatory",
"networks",
"biology",
"computational",
"biology",
"neuroscience"
] | 2013 | Stochasticity, Bistability and the Wisdom of Crowds: A Model for Associative Learning in Genetic Regulatory Networks |
Although vitamin D deficiency is a common feature among patients presenting with active tuberculosis , the full scope of vitamin D action during Mycobacterium tuberculosis ( Mtb ) infection is poorly understood . As macrophages are the primary site of Mtb infection and are sites of vitamin D signaling , we have used these cells to understand the molecular mechanisms underlying modulation of the immune response by the hormonal form of vitamin D , 1 , 25-dihydroxyvitamin D ( 1 , 25D ) . We found that the virulent Mtb strain H37Rv elicits a broad host transcriptional response . Transcriptome profiling also revealed that the profile of target genes regulated by 1 , 25D is substantially altered by infection , and that 1 , 25D generally boosts infection-stimulated cytokine/chemokine responses . We further focused on the role of 1 , 25D- and infection-induced interleukin 1β ( IL-1β ) expression in response to infection . 1 , 25D enhanced IL-1β expression via a direct transcriptional mechanism . Secretion of IL-1β from infected cells required the NLRP3/caspase-1 inflammasome . The impact of IL-1β production was investigated in a novel model wherein infected macrophages were co-cultured with primary human small airway epithelial cells . Co-culture significantly prolonged survival of infected macrophages , and 1 , 25D/infection-induced IL-1β secretion from macrophages reduced mycobacterial burden by stimulating the anti-mycobacterial capacity of co-cultured lung epithelial cells . These effects were independent of 1 , 25D-stimulated autophagy in macrophages but dependent upon epithelial IL1R1 signaling and IL-1β-driven epithelial production of the antimicrobial peptide DEFB4/HBD2 . These data provide evidence that the anti-microbial actions of vitamin D extend beyond the macrophage by modulating paracrine signaling , reinforcing its role in innate immune regulation in humans .
Mycobacterium tuberculosis ( Mtb ) infects ∼2 billion people [1] and active tuberculosis ( TB ) represents the leading cause of death from a curable disease [2] . The typical Mtb life-cycle involves entry into the host via inhalation ( exposure ) , survival and replication of bacteria in the lungs and associated lymph nodes where they resist host elimination ( infection ) , and ultimately promotion of host immunopathology resulting in aerosol transmission via coughing to the next host ( disease ) . While current treatments aim to control active disease or prevent progression from infection to disease , an attractive point of intervention would be at the transition from exposure to infection by enhancing innate pathogen control at the time of exposure . Innate immunity to Mtb infection is critical for determining disease outcome , and it has long been recognized that during TB outbreak investigations , only 30–50% of those with an equivalent exposure develop a productive infection , as demonstrated by a tuberculin skin-test conversion [3] . The primary site of Mtb infection is alveolar macrophages , which lie within the alveolar space , adjacent to the epithelial lining [4] . Lung epithelial cells are generally not considered part of the innate immune responses to Mtb infection . However , they represent both a physical and an immunological barrier to infection by contributing to the maintenance of mucosal integrity , promoting phagocytosis , and producing several antimicrobial peptides , vanguards of innate immune responses to infection [4] , [5] . For example , antimicrobial peptide ( AMP ) DEFB4/HBD2 is expressed in upper airway epithelial cells , is inducible by interleukin-1β ( IL-1β ) , and has been detected in bronchoalveolar lavage fluid from normal healthy humans [6] . It has been shown to have direct antimicrobial activity against Mtb and drug-resistant Mtb [7] . IL-1β is critical for host resistance to Mtb infection , demonstrated by the substantially reduced survival of IL-1β−/− or IL1R−/− mice after infection [8] , [9] , [10] , [11] . Caspase-1 cleaves a precursor of IL-1β to generate its active form . Catalytically active caspase-1 binds to the apoptosis-associated speck-like protein containing a caspase recruitment domain ( ASC ) subunit of inflammasomes , multiprotein complexes containing pattern recognition receptors ( PRRs ) , which detect infection by pathogens or cellular stress . A number of PRRs have been implicated in the detection of intracellular pathogens , and the resulting IL-1β secretory response . For example , AIM2 is a cytoplasmic sensor for double stranded DNA ( dsDNA ) and has recently been implicated as a component of the caspase-1 inflammasome in cells infected with viral and intracellular bacterial pathogens [12] , [13] , [14] , [15] , [16] . Similarly , NOD2 is a cytoplasmic sensor for muramyl dipeptide [17] , which stimulates NF-κB signaling to induce IL-1β expression [18] , plays an important role in immunity against Mtb [19] , and associates with the NALP1-containing inflammasome [20] . Lastly , the NLRP3 pattern recognition receptor is activated by a wide range of signals [21] . It is of particular importance for IL-1β secretion from macrophages after infection with Mycobacteria marinum ( M . marinum ) [22] and Mtb [23] . Given the critical role of innate immunity in initiating an effective response to Mtb infection , we determined in detail the host macrophage transcriptomic and cytokine responses to virulent H37Rv infection and modulation of this response by the hormonal form of vitamin D , 1 , 25-dihydroxyvitamin D ( 1 , 25D ) . Historically , there is a correlation between vitamin D deficiency and TB susceptibility [24] , [25] . Epidemiologic studies have documented a higher occurrence of active TB during winter months when sunlight exposure is reduced [26] and vitamin D deficiency has been identified as a common feature of patients with active TB [27] , [28] , [29] . However , the mechanisms underlying vitamin D signaling and control of Mtb infection are not well understood . 1 , 25D synthesis is induced in macrophages and dendritic cells upon exposure to pathogen , and the 1 , 25D-bound vitamin D receptor ( VDR ) directly induces transcription of genes encoding AMPs DEFB4/HBD2 and cathelicidin antimicrobial peptide ( CAMP ) [30] , [31] , [32] , [33] , [34] , [35] , both of which have demonstrated anti-mycobacterial activity [36] , [37] . Despite this , the direct effects of 1 , 25D on Mtb bacterial burden in infected macrophages have been modest [38] , [39] , posing the question of whether these findings alone could account for the clinical and epidemiological observations , if they are indeed causally associated . Here , we show that one of the primary macrophage responses to Mtb infection is a broad cytokine/chemokine response , which is generally enhanced by 1 , 25D . Importantly , 1 , 25D directly stimulates IL1B gene transcription , a critical component of the macrophage response to Mtb infection [40] . As this mechanism of IL1B regulation was not conserved in mice , we developed a co-culture system between macrophages and primary small airway lung epithelial cells to model the effects of elevated IL-1β secretion . In these co-culture experiments , 1 , 25D potentiated IL-1β signaling from macrophages resulting in the secretion of DEFB4 from primary lung epithelial cells . Taken together these results suggest that the effects of 1 , 25D extend beyond the macrophage and involve the modulation of paracrine signaling to enhance the innate immune responses to Mtb infection .
In order to understand the host macrophage transcriptional response to Mtb infection , we performed expression profiling studies in PMA-differentiated human THP-1 macrophage cells . Cells were infected with virulent Mtb strain H37Rv ( I ) or left uninfected ( NI ) , and treated with vehicle ( DMSO ) or 100 nM 1 , 25D ( +D ) for 24 hours . 1 , 25D treatment of Mtb-infected macrophages produced broad changes in mRNA profiles ( Table S1 ) , in which expression of 328 genes was altered at least 5-fold by either infection or 1 , 25D ( Figure 1A , Table S2 ) . A heat map of highly induced transcripts identified three major groups of genes: those that were regulated by 1 , 25D in uninfected cells , but not in infected cells ( group 1 ) , those that were regulated in the same direction in infected cells treated with vehicle or 1 , 25D ( group 2 ) , and those that were regulated in infected cells in either the vehicle or 1 , 25D treated condition , but not both ( group 3 ) . From this it is clear that about half of all 1 , 25D target genes in uninfected macrophages ( NI+D ) are not regulated in infected cells , as they do not belong to group 1 . Infection resulted in broad changes in transcription , which substantially changed the cohort of genes regulated by 1 , 25D , indicated by group 3 , and the columns that change in intensity in group 2 ( Figure 1A ) . To understand the dominant effects of 1 , 25D in Mtb-infected macrophages , the 328 genes regulated more than 5-fold by infection were filtered to retain those whose expression was altered at least a further 1 . 5-fold by 1 , 25D ( Figure 1B ) . Functional clustering of the resulting 94 genes revealed that the largest classification of these encoded cytokines ( Fig . 1B ) , and Ingenuity IPA network clustering strongly implicated a role in inflammatory responses , immune cell trafficking , and signaling ( Figure 1C ) . IPA Pathway clustering of these revealed the strongest effects were on secreted factors , many of which were further induced by 1 , 25D ( Figure 1D ) . To confirm the transcriptional changes at the translational levels , the supernatants from the cells used for microarray were subjected to Milliplex Human Cytokine Assay for a wide range of cytokines and chemokines analysis . We found that , in agreement with microarray data , 1 , 25D significantly enhanced secretion of IL-1β , CCL3/MIP1α , CCL4 , CCL8 , TNFα , IL-8/CXCL8 , and CCL20 from macrophages infected with Mtb ( Table S3 ) . Considering the critical role of IL-1β in immunity to Mtb infection [8] , [9] , [41] , we next investigated the mechanisms by which 1 , 25D increased the expression and secretion of IL-1β in Mtb-infected macrophages . Consistent with the microarray data , RT/qPCR analysis showed that 1 , 25D up-regulated the expression of IL1B transcripts in uninfected and Mtb-infected macrophages ( Figure 2A ) . Essentially identical results were obtained with two independent cultures of primary human macrophages ( Figure 2B ) . Although the levels of pro-IL-1β were elevated in both 1 , 25D-treated macrophages as well as Mtb-infected macrophages treated with 1 , 25D , mature IL-1β was only detected in extracts of Mtb-infected cells treated with 1 , 25D ( Figure 2C ) . Results of western blotting were consistent with analysis of IL-1β released from THP-1 cells or primary human macrophages , where secretion was observed only in supernatants of infected cells , and significantly elevated upon treatment with 1 , 25D ( Figures 2D , E ) . Previous in silico studies [42] identified a promoter-proximal sequence corresponding to a consensus vitamin D response element ( VDRE ) in the IL1B gene ( Fig . S1 ) . To further investigate the mechanisms of IL1B gene expression by 1 , 25D , we evaluated the VDR binding to this element by chromatin immunoprecipitation ( ChIP ) assay . VDR binding was 1 , 25D-dependent , and significantly enhanced by infection ( Figure 2F ) . 1 , 25D-dependent binding of the VDR may be due in part to its elevated expression , which was stimulated by 1 , 25D ( Figure S2 ) , consistent with the VDR being a 1 , 25D target gene [43] . A similar profile was observed with RNA polymerase II ( polII ) binding at the IL1B transcription start site ( TSS ) ( Figure 2G ) . Collectively , these experiments show the 1 , 25D-bound VDR stimulates transcription of the IL1B gene . To determine the extent of conservation of this mechanism , we performed a comparison of the IL1B VDRE loci across mammalian species using the UCSC genome browser [44] . The VDRE and the surrounding regions are well conserved in the genomes of non-human primates , but not in the mouse , rabbit , or guinea pig ( Figure S3 ) , which are commonly used to model Mtb infection in vivo [45] . Indeed , while 1 , 25D significantly enhanced cyp24 expression in mouse macrophages , no regulation of il1b was seen ( Figure S4 ) , in contrast to the 6–8-fold induction of IL1B expression seen in uninfected THP-1 cells or primary human macrophages ( Figures 2A , B ) indicating that mice would not serve as a viable in vivo model to understand this effect . Pro-IL-1β can be cleaved into its mature form by caspase-1 [46] . Western blots of lysates from infected macrophages revealed that neither 1 , 25D nor infection altered expression of caspase-1 in its pro-form ( Figure S5A ) . Caspase-1 enzymatic activity was significantly higher in cytosolic lysates from Mtb-infected cells treated with 1 , 25D ( Figure S5B ) . Catalytically active caspase-1 is a component of inflammasomes , of which ASC is a core component . We found that while the 23 kD form of ASC was predominant in THP-1 cells , the 20 kD splice variant [47] , [48] coimmunoprecipitated with caspase-1 , an association only seen in infected cells and found to be slightly higher in infected cells treated with 1 , 25D , consistent with elevated cytosolic caspase-1 activity under these conditions ( Figure S5C ) . Inflammasomes also contain pattern recognition receptors ( PRR ) that detect infection or cellular stress [46] , [49] . To understand which inflammasome PRR , or combination of PRRs , was responsible for the 1 , 25D-driven secretion of IL-1β , we investigated the potential roles of NOD2 , NLRP3 , and AIM2 in regulating IL-1β maturation in infected cells . Each of these has been previously implicated in detection of Mtb or other intracellular pathogens and in stimulating inflammasome-mediated IL-1β maturation [22] , [50] , [51] . Notably , expression of AIM2 , a cytoplasmic sensor for dsDNA [13] , [15] , was induced ∼30-fold in Mtb-infected macrophages by RT/qPCR and western blotting ( Figure 3A ) , suggesting that the AIM2 inflammasome may contribute to IL-1β cleavage and secretion . Expression of AIM2 , NOD2 or NLRP3 was strongly reduced by siRNA-mediated knockdown ( Figure S6 ) , along with positive controls IL1B and CASP1 , in Mtb-infected macrophages . None of the knockdowns of PRRs had any effect on expression of pro-IL-1β protein or on expression of caspase-1 ( Figure 3B ) . Depletion of NOD2 or AIM2 expression had no significant effect on IL-1β secretion ( Figure 3C ) . In contrast , knockdown of NLRP3 essentially abolished IL-1β secretion from Mtb-infected macrophages ( Figure 3C ) . These findings suggest that induced AIM2 expression is not primarily responsible for driving IL-1β maturation , but are consistent with other studies showing that IL-1β production during Mtb infection is largely controlled by NLRP3 [22] , [23] . In contrast , they are not consistent with the finding that NOD2 function is of primary importance [51] . Upon infection , alveolar macrophages initially phagocytose Mtb . Importantly , alveolar macrophages are also in direct contact with the respiratory epithelial surface [4] . Although the immunological contribution of lung epithelia is well studied in other contexts [52] , it is poorly characterized during the course of Mtb infection . IL-1β induces expression of genes encoding AMPs in epithelial cells through its capacity to stimulate the activity of transcription factor , NF-κB , leading to secretion of antimicrobial proteins [53] , [54] . Thus , to understand the potential crosstalk between alveolar macrophages and the human respiratory epithelial surface , we conducted a series of experiments using an in vitro co-culture system between macrophages and human primary cultures of non-polarized small airway epithelial cells ( SAECs ) from multiple donors . In cultures of SAECs alone , recombinant IL-1β induced expression of the gene encoding DEFB4 , while 1 , 25D had no significant effect ( Figure 4A ) . Conversely , induction of CAMP gene expression was largely 1 , 25D-dependent ( Figure 4A ) . Media supernatants from these treated SAECs were assayed for DEFB4 and CAMP secretion by ELISA . DEFB4 was detected under control conditions , and was found to be elevated when SAECs were treated with rIL-1β , but not 1 , 25D ( Figure 4B ) . In contrast , the cleaved C-terminal of CAMP , LL-37 , was not detected above background signal under any of the conditions ( Figure 4B ) . In comparison , neither DEFB4 nor LL-37 was detected in media supernatants from THP-1 cells infected with Mtb using this assay ( data not shown ) . SAECs were then cultured with conditioned media from control or H37Rv-infected macrophages to test for effects of 1 , 25D and secreted factors on AMP expression . In addition , as IL-1β has been shown to induce NF-κB signaling through IL1R1 [54] , we investigated the expression of NF-κB target gene DEFB4 [53] , [54] in lung epithelial cells . Media from infected macrophages induced expression of DEFB4 ( Figure 4C ) in SAECs . Importantly , pre-incubation of the transferred media with IL-1β-neutralizing antibody abolished DEFB4 gene induction . Given that DEFB4 expression levels are driven by IL-1β , the combined effects of 1 , 25D and infection on DEFB4 transcription are consistent with levels of IL-1β secretion induced in 1 , 25D-treated , infected macrophages ( Figure 2D , E ) . As expected , CAMP expression in SAECs was dependent on 1 , 25D ( Figure 4D ) . Next , to model the consequences of the real-time interaction between infected macrophages and the alveolar epithelia during the initial round of infection , we established a transwell co-culture system between Mtb-infected macrophages and SAECs ( Figure 5A ) . Macrophages were infected with Mtb for 4 hours and washed extensively to eliminate any remaining extracellular mycobacteria . A transwell bucket containing co-cultured lung epithelial cells was then added to the tissue culture plate containing the infected macrophages . The two cell populations were separated by a 0 . 4 µm filter to allow for the exchange of secreted proteins but prevent migration of mycobacteria . Note that in control experiments no mycobacteria were detected in the transwell bucket 4 days after infection ( data not shown ) . Co-culture of macrophages infected at an multiplicity of infection ( MOI ) of 5 with SAECs dramatically extended macrophage cell survival at three days post-infection , as measured by cytoplasmic lactate dehydrogenase ( LDH ) release ( Figure 5B ) , a marker of plasma-membrane compromise and necrosis . After 3 days of infection , the relative amount of LDH release was comparable to that seen 24 hours after infection under macrophage-only conditions . 1 , 25D treatment had no effect on the survival of infected cells cultured in the absence or presence of SAECs . When macrophages infected at an MOI of 10 were co-cultured with SAECs , LDH release was slightly higher , but a similar protective effect of SAECs was observed ( Figure 5B ) . The protective benefit of the SAECs in co-culture was also clear by the relative amount of adherent cells remaining as visualized by phase-contrast microscopy ( Figure 5C ) ; most of the macrophages co-cultured with SAECs were still adherent , whereas infected macrophages cultured in the absence of SAECs had detached from the plate . To determine any effects of co-culturing with and without 1 , 25D on mycobacterial burden , colony forming unit ( CFU ) assays were performed with cells co-cultured as above . We determined changes in total Mtb burden after 72 hours of infection . The addition of SAECs resulted in a halving of mycobacterial burden at this time point , and addition of 1 , 25D to the co-culture system produced a further significant reduction in mycobacteria ( Figure 6A ) . To confirm the contribution of epithelial signaling by infection- and 1 , 25D-induced IL-1β towards the reduction in mycobacterial burden , we knocked down IL1R1 receptor expression in SAECs 36 hours prior to their co-culture . Control experiments showed that expression of epithelial IL1R1 was reduced for at least 72 hours after siRNA-mediated knockdown ( Figure 6B ) . Mycobacterial burden was sharply elevated 72 hours after infection in the absence or presence of 1 , 25D when macrophages were co-cultured with IL1R1-depleted SAECs . In contrast , co-culture of SAECs transfected with control siRNAs eliminated net mycobacterial growth in the presence of 1 , 25D ( Figure 6C ) , consistent with experiments described above . To determine if epithelial secretion of DEFB4 was responsible for the increased control of mycobacterial proliferation , we transfected SAECs with either siRNA against IL1R1 or DEFB4 transcripts 36 hours prior to their co-culture with infected macrophages . Reduced expression of DEFB4 was verified by qPCR in samples collected 72 hours after the initiation of their co-culture ( Figure 6D ) . CFU assays were performed at 72 hours after infection and demonstrated that siRNA-mediated knockdown of DEFB4 expression permitted levels of bacterial proliferation similar to what was observed in knockdown of IL1R1 ( Figure 6E ) . Taken together , these data reveal that IL-1β secreted from infected macrophages drives a paracrine signaling cascade which contributes to control of mycobacterial burden in our culture system . The contribution of 1 , 25D to reducing mycobacterial burden in macrophages arises at least in part from its capacity to enhance autophagy in infected macrophages in a CAMP-dependent manner [38] , [55] , a critical mechanism for control of intracellular pathogens . To understand if epithelial CAMP or DEFB4 secretion was helping to control bacterial burden by enhancing autophagy in macrophages , even if CAMP was undetected in the media supernatant in control experiments , we probed for colocalization of Mtb and LC3 protein , a marker of autophagosomes , in ( 1 , 25D-treated ) infected macrophages 3 days after infection . Using bright-field fluorescence microscopy , we visualized Mtb and the presence of any colocalized LC3 ( Figure 7A ) . Quantification of the number of times in which mycobacteria were found in LC3-containing membrane structures in confocal imaging revealed that the frequency of colocalization increased when infected macrophages were treated with 1 , 25D; however , the presence of SAECs in co-culture had no impact on the degree of colocalization , both in the absence and presence of 1 , 25D ( Figure 7B ) . Finally , analysis of 3-dimensional stacks of confocal images taken from 1 , 25D-treated conditions confirmed that the signal from Mtb colocalized with LC3-containing structures ( Figure 7C ) . Taken together , these data reveal that the decrease in CFU observed in co-culture experiments , which is dependent on paracrine IL-1β signaling and the reciprocal secretion of DEFB4 from epithelial cells ( Figure 8 ) , is not a consequence of an increase in autophagy in macrophages .
In this study we have presented the first large scale microarray profile to determine the host macrophage transcriptomic responses to Mtb infection and the effect of 1 , 25D on those responses . Analysis of this data demonstrated that infection markedly changed the profile of genes regulated by 1 , 25D . ChIP studies also revealed that infection enhances DNA binding by the VDR . These data suggest infection induces large-scale changes in chromatin that modify the availability of VDREs for binding by the VDR throughout the genome . Additionally , it is clear from this work that studies of 1 , 25D target genes in uninfected macrophages are not an accurate depiction of its contribution to host responses in the face of an active infection . A potential limitation of this study was our initial use of THP-1 cells for microarray analysis of target genes . As this is cell line is derived from a monocytic leukemia , not all of the genes identified as being regulated may be found under similar experimental conditions using primary human alveolar macrophages . Pathways analysis of genes differentially regulated by 1 , 25D in infected THP-1 cells revealed that the dominant function of 1 , 25D in the context of the innate immune response to Mtb is the up-regulation of specific components of the broad cytokine response induced by infection . Previous computational analysis identified VDREs proximal to the IL1B , CCL3 , CCL4 , CCL8 genes , but not IL-8 or TNF-a [42] . Probing 1 , 25D modulation of these cytokines in vivo is complicated by our results showing that these VDREs are not conserved in mouse , rabbit , or guinea pig ( this paper and Verway et al , manuscript in preparation ) , the model organisms typically used to study Mtb infection . Despite the fact that IL-1β signaling also is critical for innate immune responses to Mtb in these animal models , it would appear that they would not be appropriate for in vivo modeling of the modulatory effects of vitamin D on the early innate immune responses to infection . 1 , 25D markedly enhanced mRNA levels and secretion of IL-1β in Mtb-infected macrophages . Our data reveal that 1 , 25D is acting primarily at the level of IL1B gene transcription without affecting the levels of inflammasome , as substantial levels of pro-IL-1β were seen in uninfected macrophages after 1 , 25D treatment , whereas secretion required infection . Furthermore , of the cytokines whose secretion was elevated by 1 , 25D treatment after infection ( CCL3 , CCL4 , CCL8 , IL-8 , and TNF-α ) , treatment did not induce their secretion from uninfected cells . This would suggest that 1 , 25D is acting to prime and potentiate inflammatory innate immune responses , without inducing any unwanted inflammation in a resting state . IL-1β and IL1R1 are essential for survival in mouse models of Mtb infection [8] , [9] , [10] , [11] . IL-1β is also of critical importance during the early stages of infection as shown by in vitro infections using macrophages from these knockout mice [11] . Additionally , it has been demonstrated that Mtb expresses zmp1 , a metalloproteinase that prevents phagolysosomal maturation by inhibiting inflammasome-mediated IL-1β cleavage , a mechanism of virulence that suppresses the host response [40] . siRNA-mediated knockdowns of various inflammasome sensors demonstrated that the NLRP3 inflammasome is responsible for IL-1β secretion under all conditions , in agreement with previously published data looking at mycobacterial infection in mouse and human macrophages [22] , [23] . While mouse genetic models have revealed that IL-1β signaling plays a key role in Mtb resistance , nlrp3−/− and casp1−/− mice showed no deficiency in the production of IL-1β , control of bacterial burden , or survival [23] . Even though ASC was important in this model , this may be because ASC-null mice are deficient in antigen presentation and DC trafficking due to a loss of Dock2 expression [56] . It is therefore unclear at this time whether the NLRP3 inflammasome is required in vivo for control of Mtb infection . Previous studies on the role of IL-1β in the innate immune response to infection have focused on its capacity to control infection by autocrine signaling from infected macrophages . While epithelial cells are known to have important immunologic function in other diseases and contexts [4] , [57] , [58] , they are usually not considered to be an important part of the innate immune response to Mtb infection . Primary upper respiratory epithelial cells express DEFB4 in response to IL-1β stimulation in vitro [6] . Our data demonstrate that the IL-1β secreted from infected macrophages has the capacity to elicit an antimycobacterial response from small airway epithelial cells . Importantly , siRNA-mediated knockdown of epithelial IL1R1 or DEFB4 expression significantly negated the control of Mtb growth in co-cultured macrophages . Modeling of these findings in vivo will be necessary to fully understand the extent of the contribution of epithelial cells to the innate immune response in this context . Doing so in primates would pose a major undertaking , and it would be difficult to assess the contribution of the reciprocal signaling cascade between alveolar macrophages and epithelial cells in such a model , as it would likely occur in the acute response to a very low dose exposure . Macrophage-epithelial cell co-culture controls mycobacterial burden but did not induce autophagy , and epithelial knockdown of IL1R1 expression negated any protective benefit . Taken together , our data suggest that paracrine secretion of DEFB4 from epithelial cells provides a more substantial level of protection against infection than DEFB4 production by macrophages . Previous studies have provided evidence for a reduction in bacterial burden when Mtb-infected macrophages are treated with 1 , 25D in vitro [38] , [39] . We find that this effect is limited , and that bacterial burden is elevated 3 days after infection with virulent Mtb , representing a failure to control the infection . From this study it is clear that more complete models of the innate immune response are required to understand the full effect of 1 , 25D . Correlations between vitamin D deficiency and a higher incidence of TB have repeatedly been observed [27] , [28] , [29] . Clinical trials did not reveal a substantial benefit of vitamin D in the treatment of active disease [59] , [60] , although a recent study showed that vitamin D supplementation accelerated the resolution of inflammatory responses during treatment for Mtb disease [61] . Our results would suggest that the optimal point of treatment would be vitamin D supplementation to bolster innate responses and prevent infection . Such a program would be economically viable considering the negligible cost of vitamin D supplementation as compared to antibiotic chemotherapy , and a useful prophylactic measure against drug-resistant tuberculosis . Our mechanistic data supports the idea that serum levels of vitamin D may be causally important for defense against Mtb exposure , but clinical trials would be required to understand this .
Intraperitoneal macrophages were acquired from C57BL/6 mice as outlined in an animal use protocol approved by McGill University ( Permit #2010-5860 ) according to Canadian Council on Animal Care guidelines . De-identified human peripheral blood was purchased from Research Blood Components ( Boston , MA ) . Following informed written consent , blood was collected by venipuncture from healthy adult volunteers , recruited by Research Blood Components . Protocols for the collection of whole blood for research purposes were approved by New England Institutional Review Board . THP-1 cells ( TIB-202 , ATCC ) were cultured in RPMI-1640 ( Wisent ) with 10% FBS . SAEC cells were acquired following informed consent , permission , and ethical approval by Lonza , and were cultured in SAGM ( Lonza ) , as directed by manufacturer . These cells were isolated by a proprietary method from the 1 mm bronchiole of the lung , which includes alveoli . H37Rv was cultured to mid-log phase in a rolling incubator at 37°C in Middlebrook 7H9 ( Difco ) with . 05% Tween-80 and 10% ADC enrichment ( BD Biosciences ) . 1×106 THP-1 cells were terminally differentiated by 2×10−8M PMA for 24 hours in RPMI with 10% charcoal stripped FBS , inducing cell cycle arrest . H37Rv cultures were centrifuged and pellets were resuspended in RPMI-1640 with 10% charcoal stripped FBS and . 05% Tween-80 and clumping was disrupted by repeated passage through a 27-gauge needle . Media was removed from THP-1 cells and replaced with media containing Mtb in the indicated multiplicity of infection ( MOI ) for 4 hours . THP-1 cells were then vigorously washed three times with RPMI to remove extracellular bacteria , followed by incubation in RPMI with 10% charcoal stripped FBS containing either vehicle ( DMSO ) or 1 , 25D at a final concentration of 10−7 M . RNA extraction was performed with TRIzol/chloroform ( Invitrogen ) as per manufacturers' instructions . RT was performed with iScript cDNA Synthesis Kit ( Bio-Rad ) and qPCR was performed with SsoFast Eva Green with low ROX ( Bio-Rad ) on an Eco qPCR cycler ( Illumina ) , normalizing expression to β-actin and 18S . Primers used are listed in Table S4 . Human Gene 1 . 0 ST arrays ( Affymetrix ) were used to measure samples from two independent experiments , each performed in triplicate . Microarray data presented is from one triplicate set , which is representative of the other . Protein extracts from THP-1 cells were prepared in lysis buffer ( 10 mM Tris , pH 7 . 5 , 150 mM NaCl , 1% Triton X-100 , 1 mM phenylmethylsulfonyl fluoride , 0 . 2 mM sodium orthovanadate , 0 . 5% Nonidet P-40 ) and processed for Western blotting and separated on Tris-HEPES-SDS gradient protein gels ( Pierce ) using standard transfer and blotting protocols . Immunoprecipitation was performed as described in [62] , using 2 µg of either rabbit serum IgG ( sc-2027 ) or α-caspase-1 p10 ( sc-515 , Santa Cruz ) antibody for each sample . Antibodies: α-IL-1β ( MAB601 , R&D Systems ) , α-AIM2 ( ab93015 , abcam ) , α-caspase-1 p20 ( sc-1780 , Santa Cruz ) , α-actin ( sc-1616 , Santa Cruz ) , α-VDR ( sc-1008 , Santa Cruz ) , α-ASC ( sc-271054 , Santa Cruz ) , murine serum IgG ( Sigma ) . Recombinant IL-1β was purchased from Millipore . ELISAs for DEFB4 ( Abnova ) and LL-37 ( Hycult Biotech ) were performed in accordance with the manufacturer's instructions . At the indicated time points , tissue culture plates were centrifuged to pellet any liberated mycobacteria and non-adherent cells . Media was aspirated and macrophages were lysed with water for 5 min , after which an equal volume of 2x 7H9 with . 1% Tween-80 was added . Samples were vigorously resuspended and plated in serial dilution on Middlebrook 7H10 ( Difco ) with 10% OADC enrichment ( BD Biosciences ) . Following informed consent , blood was drawn from healthy adults . Buffy coats were isolated from whole blood by Ficoll density-gradient centrifugation using Ficoll-Paque Premium ( GE Healthcare ) , and thrice resuspended in PBS and centrifuged to remove platelets . Monocytes were purified from the total leukocyte population by sorting of adherent cells after 2 hours , by washing the culture plates twice with RPMI-1640 with 20% human serum . Cells were allowed to differentiate into naïve macrophages over 6 days in RPMI-1640 with 20% human serum before infection , with media changes every two days throughout . Intraperitoneal macrophages were elicited from C57BL/6 mice ( Jackson ) by intraperitoneal injection of 2 ml of 3% thioglycolate solution . Three days later , macrophages were collected by lavage of the peritoneal cavity . Macrophages were purified using CD11b Microbeads ( Miltenyi Biotech ) and allowed to adhere for 24 hours before infection . One day after PMA-induced differentiation , siRNAs targeting CASP1 , AIM2 , IL1B , NLRP3 , NOD2 , and the non-targeting scrambled control ( NC1 ) ( Integrated DNA Technologies ) were transfected into THP-1 cells using Transductin ( Integrated DNA Technologies ) in 10% Q-serum , according to manufacturer's instructions . After 4 hours , media was replaced with RPMI with 10% charcoal stripped FBS . After another 48 hours , cells were infected with H37Rv in accordance with the above protocol . THP-1 cells were not infected or infected with H37Rv followed by treatment with DMSO or 1 , 25D at a final concentration of 10−7 M for 24 hours . Media was filter sterilized by passage through 0 . 20 µm filters . Media was then incubated with anti-IL-1β or non-specific murine IgG for 30 minutes before transfer to epithelial cell cultures . Cultures were harvested for RNA after 24 hours . Epithelial cells were seeded in transwell buckets with 0 . 4 µm pores ( Corning ) and cultured in RPMI with 10% charcoal stripped FBS 48 hours before infection of the THP-1 cells . After the THP-1 cells were infected for 4 hours at a MOI of 5 and washed , SAECs were placed in co-culture with H37Rv-infected THP-1 cells by transferring the buckets containing the epithelial cells to the plates containing the infected macrophages , and then treated with DMSO or 1 , 25D ( 10−7 M ) . At indicated time points , plates were centrifuged and both cell populations were lysed with water and pooled for CFU assay , performed as described above . Following 4% paraformaldehyde fixation over 10 minutes , cells on coverslips were washed with 100 mM glycine and then PBS . Cells were then permeabilized with 0 . 1% Triton X-100 for 5 minutes . Cells were incubated in PBS-0 . 2% BSA during 5 minutes . Primary antibody against LC3 ( Novus Biological , dilution 1/300 ) was incubated with coverslips for 1 hour at 37°C in a humidified chamber in phosphate-buffered saline with 1% BSA . Following 3 washes , cells were incubated with the secondary antibody anti-rabbit Alexa-647 ( Life Technologies , dilution 1/1000 ) and with the anti-Mtb antibody coupled to FITC ( Abcam ab20962 , dilution 1/100 ) for 45 minutes at RT in the dark . Slides were mounted in Vectashield containing DAPI ( Vector ) and observed with a Zeiss Axiovert X100 bright field microscope or a Zeiss LSM510 X100 confocal microscope . Images were acquired with LSM510 software . Stacks of confocal images , 3D reconstitution , and quantification of percent of colocalization were performed with Imaris 7 . 4 and the figures were assembled with Photoshop ( Adobe ) . Total caspase-1 activity was assayed using ICE/Caspase-1 Colorimetric Protease Assay Kit ( Millipore ) as per manufacturers' instructions . Cell lysates were incubated with YVAD-pNA and read in a spectrophotometer at 405 nM . Media from THP-1 cells was collected 24 hours after treatment and sterilized by passage through . 20 µm filters . Levels of secreted cytokines and chemokines were assayed by Milliplex Human Cytokine Panels 1 , 2 and 3 ( Millipore ) and read on a BioPlex ( Bio-Rad ) . Infected and uninfected THP-1 cells were collected after 24 hours of treatment . Cells were fixed for 20 minutes with 1% formaldehyde , and washed with 1 . 25 µM glycine . Following cell and nuclei lysis , chromatin was sonicated for 75 cycles of 10 s ON/20 s OFF on a Bioruptor Sonicator ( Diagenode ) . IPs were then performed with either normal rabbit IgG , 6 µg of anti-VDR ( Santa Cruz sc13133 ) or 6 µg of anti-RNA Polymerase II ( Abcam anti-polIICTD #ab5131 ) . Primers used for region amplification are listed in Table S4 . Quantification of immunoprecipitated material was performed by qPCR and normalized for input DNA . Student's t-test , ANOVA , or a two-tailed Fisher's exact test was performed where indicated using GraphPad software . For microarray samples , Flexarray v1 . 6 software was used to normalize overall chip signal using the Affymetrix Power Tools ( APT ) Robust Multi-Array Average ( RMA ) algorithm . The EB ( Wright and Simon ) algorithm was used for statistical analysis to calculate fold transcript expression and significance between of each experimental condition relative to the uninfected , untreated condition ( NI ) . Cell-free media supernatant was collected from infected macrophages with and without SAECs in co-culture . Cytotoxicity Detection Kit ( LDH ) from Roche was used in accordance with manufacturer's instructions . IL-1β: NM_000576/NP_000567 , Caspase-1: NM_001223/NP_001214 , NLRP3: NM_001079821/NP_001073289 , AIM2: NM_004833/NP_004824 , NOD2: NM_022162/NP_071445 , DEFB4: NM_004942/NP_004933 , CATH: NM_004345/NP_004336 | In 2010 there were ∼9 million cases of tuberculosis and 1 . 4 million deaths , representing the second largest cause of death worldwide and the leading cause of death from a curable disease . M . tuberculosis ( Mtb ) replicates within cells of the immune system called macrophages over an approximate 72 hour period , ultimately inducing cell death . Notably , macrophages are sites of vitamin D signaling , and there is broad evidence that vitamin D modulates macrophage responses to Mtb . Elevated levels of TB have long been associated with vitamin D deficiency , strongly suggesting that vitamin D supplementation may be of therapeutic benefit . In this study we profile the host macrophage response to Mtb infection through the use of high-throughput genomics techniques . From this we have discovered that the dominant function of vitamin D is the modulation of the levels of specific cytokines , mediators of immune cell to cell signaling . Of particular interest was the increase in IL-1β signaling , which we show to be directly regulated by vitamin D . We also show that this increase in IL-1β is critical for driving a signaling cascade between macrophages and lung epithelial cells leading to epithelial antimicrobial peptide production that helps to contain Mtb infection in our model culture system . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"bacterial",
"diseases",
"infectious",
"diseases",
"mycobacterium",
"immune",
"cells",
"cytokines",
"monocytes",
"tuberculosis",
"immunology",
"biology",
"immune",
"response",
"immune",
"system"
] | 2013 | Vitamin D Induces Interleukin-1β Expression: Paracrine Macrophage Epithelial Signaling Controls M. tuberculosis Infection |
Existing methods to ascertain small sets of markers for the identification of human population structure require prior knowledge of individual ancestry . Based on Principal Components Analysis ( PCA ) , and recent results in theoretical computer science , we present a novel algorithm that , applied on genomewide data , selects small subsets of SNPs ( PCA-correlated SNPs ) to reproduce the structure found by PCA on the complete dataset , without use of ancestry information . Evaluating our method on a previously described dataset ( 10 , 805 SNPs , 11 populations ) , we demonstrate that a very small set of PCA-correlated SNPs can be effectively employed to assign individuals to particular continents or populations , using a simple clustering algorithm . We validate our methods on the HapMap populations and achieve perfect intercontinental differentiation with 14 PCA-correlated SNPs . The Chinese and Japanese populations can be easily differentiated using less than 100 PCA-correlated SNPs ascertained after evaluating 1 . 7 million SNPs from HapMap . We show that , in general , structure informative SNPs are not portable across geographic regions . However , we manage to identify a general set of 50 PCA-correlated SNPs that effectively assigns individuals to one of nine different populations . Compared to analysis with the measure of informativeness , our methods , although unsupervised , achieved similar results . We proceed to demonstrate that our algorithm can be effectively used for the analysis of admixed populations without having to trace the origin of individuals . Analyzing a Puerto Rican dataset ( 192 individuals , 7 , 257 SNPs ) , we show that PCA-correlated SNPs can be used to successfully predict structure and ancestry proportions . We subsequently validate these SNPs for structure identification in an independent Puerto Rican dataset . The algorithm that we introduce runs in seconds and can be easily applied on large genome-wide datasets , facilitating the identification of population substructure , stratification assessment in multi-stage whole-genome association studies , and the study of demographic history in human populations .
Genetic structure among and within human populations reflects ancient and recent historical events , migrations , bottlenecks , and admixture , and carries the signatures of random drift and natural selection . The complex interplay among these forces results in patterns that could be used as tools in diverse areas of genetics . In population genetics , uncovering population structure can be used to trace the histories of the populations under study [1] . In medical genetics , identifying population substructure and assigning individuals to subpopulations is a crucial step in properly conducting association studies to unravel the genetic basis of complex disease . With data from large-scale association studies becoming increasingly available , it has become apparent that population substructure resulting from recent admixture or biased sampling can increase the number of false-positive results or mask true correlations [2–5] . Detection of and correction for stratification in a given dataset is a problem that has been discussed at length in recent literature [6–13] . One of the prevailing methods for identifying population structure is a model-based algorithm implemented in the program STRUCTURE [14 , 15] . STRUCTURE has been shown to effectively assign individuals to clusters [16–18] . However , anticipating data from thousands of individuals and thousands of markers , this algorithm will become impractical due to its intensive computational cost [13 , 19 , 20] . At the same time , it is sensitive to the choice of prior distributions of model parameters and relies heavily on explicit assumptions about the data that may not always hold , making the method unstable when these assumptions are violated [19 , 21] . Recently , Principal Components Analysis ( PCA ) , a classical nonparametric linear dimensionality reduction technique , is regaining favor for uncovering population structure . PCA can be used to extract the fundamental structure of a dataset without the need for any modeling of the data; see [22] and references therein for a detailed discussion . It is computationally efficient and can handle genome-wide data for thousands of individuals . PCA was first used in population genetics by Cavalli-Sforza to infer axes of human variation [23] . It has recently been shown to be a powerful tool for the identification of population structure and the correction of stratification in the setting of association studies [13 , 20] . Coupled with a clustering tool , it can also be used for inferring population clusters and assigning individuals to subpopulations [19] . Identifying a set of markers that could effectively be used for inference of population structure will reduce genotyping costs and will potentially provide insight to genetic loci that have undergone selective pressures . Several approaches have been used to this end [24–29] . For instance , informative markers have been traditionally selected to maximize δ , the absolute difference in allele frequency between two ancestral populations , or Wright's FST [24 , 25 , 30–32] . The cutoff value for δ is highly subjective; its statistical properties are not well defined and it can only be used for two source populations at a time [26 , 28] . On the other hand , it is not clear how FST can be applied to the selection of informative markers for admixed populations , when the parental contributions are unequal [26 , 28] . Informativeness for assignment ( In ) , as defined by Rosenberg et al . [26] , is an FST-correlated measure that computes the mutual information on allele frequencies . In all cases in which an allele frequency–based metric is used , knowledge of individual membership to a studied population is a prerequisite . When studying admixed populations , it may be difficult to define or sample the ancestral source populations . The origin of the study individuals may also be unknown in other situations , e . g . , studies involving large samples of blood donors . We have developed a novel algorithm to identify a subset of SNP markers that capture major axes of genetic variation in a genotypic dataset without use of any prior information about individual ancestry or membership in a population . Our approach is a greedy deterministic variant of a Monte-Carlo algorithm that has provable performance guarantees [33–35] . Here , we describe the theory of our method and its derivation from PCA , and evaluate it extensively on a previously described dataset of 255 individuals from 11 populations typed for 10 , 805 autosomal SNPs [36] . First , we infer the structure of the dataset using PCA followed by a standard clustering algorithm ( k-means ) . We show that this two-step approach almost always achieves 100% accuracy for assigning individuals to their true clusters . We then use our method to identify subsets of SNPs that can extract the same structure and evaluate the performance of these SNP panels for clustering individuals in their respective populations both within and across continents . We compare the efficiency of these panels to SNPs chosen based on the measure of informativeness for assignment ( In ) [26] as well as randomly chosen SNPs . We then validate our results both by splitting the studied individuals in training and test sets and by using the selected SNPs for clustering individuals from the HapMap database . Analyzing genotypes for approximately 1 . 7 million SNPs that have been made available through the HapMap project [37 , 38] , we identify a set of SNPs that differentiate the Japanese and Chinese populations . We demonstrate that our algorithm for selecting structure informative SNPs always converges to the results of applying PCA and the clustering algorithm on the full dataset , while achieving almost 99% genotyping savings . Furthermore , using data from nine indigenous populations , we manage to ascertain a global panel of PCA-correlated SNPs that accurately assigns individuals to their population of origin . Finally , analyzing two independent Puerto Rican datasets , we demonstrate the applicability of our method for the selection of structure informative markers , when admixed populations are analyzed .
We will first develop the theoretical underpinnings of our method and explain its connections to PCA . PCA is a linear dimensionality reduction technique that seeks to identify a small number of “dimensions” or “components” that capture most of the relevant structure in the data . In genetics , given a large number of genetic markers ( e . g . , thousands of SNPs ) for a large number of individuals , PCA and the Singular Value Decomposition ( SVD ) have been used in order to infer population structure . We note here that SVD is the fundamental algorithmic and mathematical component of PCA; indeed , PCA is equivalent to computing the SVD of a distance matrix representing the data . Consider a SNP data matrix A whose m rows correspond to m individuals and whose n columns correspond to n SNPs . Let m ≤ n , which is almost invariably the case in genetics data . The elements of this matrix may be encoded as +1 or 0 or −1 , denoting ( respectively ) whether an individual is homozygotic with respect to the first allele , heterozygotic or homozygotic with respect to the second allele [22] . ( See Methods for more details on our encoding of the data . ) The SVD of this m × n matrix A returns m pairwise orthonormal vectors ui , n pairwise orthonormal vectors vi , and m nonnegative singular values σi . The matrix A may be written as a sum of outer products ( rank-one components ) as For SNP data matrices A of the above form , the left singular vectors ( the ui's ) are associated with the columns ( SNPs ) of A—indeed , they are linear combinations of the columns of A—and may be called eigenSNPs [39] . A common strategy is to perform dimensionality reduction by keeping a small number ( e . g . , two or three ) of eigenSNPs and then perform further data analysis ( e . g . , clustering or classification ) by representing all individuals with respect to the selected eigenSNPs . Since eigenSNPs are mathematical abstractions and do not correspond to actual SNPs , a natural question arises: is it possible to identify a small subset of actual SNPs ( i . e . , columns of the original data matrix ) such that the top few singular vectors of the matrix containing only the chosen SNPs are strongly correlated with the top few singular vectors of the original SNP matrix ? Drineas et al . [33–35] prove that the top few singular vectors of any matrix A may be well approximated by the top few singular vectors of a matrix consisting of a much smaller number of judiciously chosen columns of A . Here , we apply these recent results from the theoretical computer science literature to the problem of SNP selection for structure identification . The selected SNPs will be chosen to correlate with the top principal components , and thus we will call them PCA-correlated SNPs . Assume that there are k principal components and thus k eigenSNPs of interest; the choice for k will be addressed below . Following [35] , we seek the columns of the original matrix that mostly lie in the subspace spanned by the top k eigenSNPs . Notice that we shall seek columns ( SNPs ) that are simultaneously correlated with all top k eigenSNPs , and not with each of them individually . Surprisingly , the SVD immediately suggests such SNPs . By manipulating the expression of Equation 1 , we see that the j-th column ( SNP ) of the full SNP data matrix A ( denoted by Aj ) may be expressed as Here , is the j-th element of the i-th right singular vector . Thus , the j-th column of A is a linear combination of all left singular vectors and corresponding singular values , and the are the coefficients of this linear combination . Instead of using all m left singular vectors and singular values , we can express Aj as a linear combination of only the top k left singular vectors and corresponding singular values; some loss of information is now inevitable: We will pick columns of A that have large coefficients , i = 1 . . . k . In particular , we shall order the columns of A with respect to the scores Drineas et al . [34 , 35] proved that if a small number of columns is picked in independent identical random trials , where in each trial the j-th column of A is picked with probability proportional to pj , then the top k left singular vectors of the selected columns are very close to the top k left singular vectors of the original matrix . While the probabilistic nature of their approach is important for the formal mathematical proof , a greedy variant that picks the columns corresponding to the largest pj's should also work well for many practical datasets . This greedy variant is implemented here , and does work well for SNP data . We should note that even for the largest matrices that we experimented with ( i . e . , the HapMap data ) the computation of the scores pj takes less than a few seconds on a conventional computer . In order to compute the scores pj of Equation 4 , we must know how many principal components to keep; that is , we must know the rank parameter k . This is equivalent to determining the number of significant principal components in the data , which is a challenging research topic in the data analysis and numerical analysis literature; see [40] for a review . In order to determine whether the i-th principal component is significant , we will compare the data matrix corresponding to the i-th and all smaller principal components to a randomly generated matrix with the same elements . If the former matrix does not have significantly more structure than the latter one , then we conclude that the i-th principal component is not significant . Intuitively , a principal component is significant if and only if it has more structure than a random matrix , which has no useful structure . See Methods for a detailed description of this procedure , which is strongly motivated by recent algorithmic developments in random matrix theory [41 , 42] . We evaluated our methods extensively on a previously described dataset of 11 populations from around the world [36] . Only data from autosomal chromosomes was included in our analysis ( 10 , 805 SNPs ) . To demonstrate the resolution that could be achieved by our algorithms , we analyzed the entire dataset as well as subsets of the data consisting of populations within a single continent . Our PCA-based algorithm does not operate on matrices with missing entries . The procedure we followed for the handling of missing entries resulted in different numbers of SNPs being analyzed for the population group each time under focus . ( See Methods for details on encoding and handling of missing entries . ) After selecting a subset of either PCA-correlated SNPs or high-In SNPs , we employed the procedure described above to determine the number of significant principal components for the selected subset , and we used these components in the subsequent analysis . We first examined if our algorithms could be used to select a small subset of SNPs that cluster individuals in broad continental regions . The studied populations can be assigned to four different continents: Africa ( Mbuti , Mende [East African] , Burunge [West African] , and African Americans ) , Europe ( European Americans and Spanish ) , Asia ( Mala [South Indian] , East Asian , and South Altaian ) , and America ( Nahua and Quechua ) . A total of 9 , 419 SNPs were included in our analysis . As discussed later in this report , PCA will recognize much finer resolution than broad intercontinental clustering . So , for this particular experiment we manually set the number of principal components for further analysis to three . The rationale behind this choice is that the first principal component captures 37 . 4% of the variance in the data , the second an additional 7 . 5% , the third an additional 3 . 1% , whereas the contribution of the fourth one drops below 1 . 5% . ( The experimental results would be essentially the same even if four principal components were kept . ) These top three eigenvectors were used to cluster the data in four clusters using the k-means algorithm . We then compared individual assignment to a cluster to actual membership to a continent and found that PCA and k-means achieved perfect clustering of individuals to different continents using all available SNPs ( Figure 1 ) . Next , we reproduced this result using only a small subset of SNPs . Calculating the scores described in Equation 4 , we selected ten to 200 PCA-correlated SNPs and repeated PCA and k-means clustering using only this small subset of SNPs . Figure 2 shows the scores and positions of the top 30 PCA-correlated SNPs throughout the genome as well as their genotype frequency patterns across the four continents . As shown in Figure 1 , these 30 PCA-correlated SNPs achieve close to perfect clustering of all studied subjects to their respective continents of origin ( correlation coefficient between predicted and true membership to a continental cluster is 0 . 99 ) . We compared the efficiency of the PCA-correlated SNPs that we selected to that of a set of SNPs selected using the informativeness for assignment measure ( In ) [26] . In was estimated for all available SNPs , using four geographically distinct population groups ( its calculation requires knowledge of predefined populations ) . Again , the top ten to 200 highest-ranking In SNPs were picked and PCA and k-means was run in order to assign individuals to four continental clusters using only these subsets of SNPs ( Figure 1 ) . High In SNPs also perform very well . However , in this case , even though PCA-correlated SNPs have been selected in an entirely unsupervised manner , they perform better than high-In SNPs ( Figures 1 and 2 ) . When choosing high-ranking In SNPs , about 60 SNPs are needed in order to achieve accurate clustering to four continents . Interestingly , there was a 53% overlap between the top PCA-correlated SNPs and SNPs ranking high for In . We repeated the aforementioned procedure using sets of randomly selected SNPs . Experiments were repeated 30 times , each time selecting ten to 200 random SNPs . The average correlation coefficient of individual membership using these random SNPs to membership using all available SNPs is shown in Figure 1 . As expected , random SNPs perform far worse than carefully selected SNPs for the inference of population structure . In order to validate our method , we split the studied individuals into a training set ( 50% ) and a test set ( 50% ) . This time , we first used both our method and the In measure to select structure informative SNPs in the test set and then applied this panel of SNPs to assign individuals to continental regions in the training set . Experiments were repeated with 50 random splits and the average correlation coefficients between predicted and true membership over all runs are shown in Figure 3 . Analysis of the test set produced essentially the same results as in the training set , with PCA-correlated SNPs doing slightly better than In SNPs . We expect that our results would improve with larger training sets . Finally , we examined the value of the SNP panels that we selected for clustering individuals to different continental regions by testing their performance for assigning individuals from the four HapMap populations ( Yoruba in Ibadan , Nigeria; Utah residents with ancestry from northern and western Europe ( CEPH ) ; Han Chinese in Beijing , China; and Japanese in Tokyo , Japan ) to their true continent of origin ( Figure 3 ) . Both PCA-correlated SNPs and high-In SNPs perform exceptionally well and as few as 14 PCA-correlated SNPs or 20 high-In SNPs are enough to accurately cluster all samples to three distinct clusters . However , this task seems to be much easier than clustering 11 different populations to different continents as we did before , and as few as 40–45 random SNPs also suffice to accurately assign individuals to their continents of origin . We next tested the efficiency of our methods for detecting population structure in finer detail . To this end , we studied populations that originated from the same geographic region separately , and repeated the empirical analysis described earlier in this report . Three of the populations that we studied are indigenous Africans . We tried to define a subset of SNPs that could be used in order to accurately cluster individuals to each of these populations ( Figure 1 ) . Our analysis showed the existence of five significant principal components and these were used to perform k-means clustering using 8 , 928 SNPs . This achieved almost 100% accuracy ( correlation coefficient 0 . 97 ) . Analyzing 20 PCA-correlated or high-In SNPs is enough to replicate the results of PCA and k-means on the full dataset . On the other hand , as many as 200 random SNPs are needed for the correlation coefficient between true and predicted membership to reach 0 . 95 . The overlap between SNPs selected using our method and In is 34% . Interestingly , PCA-correlated SNPs are in high linkage disequilibrium ( LD ) with In SNPs; see Table 1 for more details . Adding the admixed ancestry population of African Americans to this group decreases our ability to perfectly cluster individuals in a distinct population of origin using k-means ( Figure 1 ) . Again , five principal components were identified as significant , this time for the analysis of 9 , 193 SNPs . While each of the autochthonous African populations is still accurately clustered , African Americans do not all fall in a single cluster . African Americans are not one homogeneous population and overlap exists with one of the other populations that were available for study in this continent . In fact , 19 out of the 42 studied individuals are assigned , as could perhaps be expected , to the western African cluster of the Burunge population . We note that we found no clear overlap between the African American and the Caucasian samples that were analyzed here ( unpublished data ) . Again , great genotyping savings seem possible , with only 20 PCA-correlated or 50 high-In SNPs needed to converge to the performance of PCA and k-means clustering using all available SNPs . Once more , we observed high overlap and LD between PCA-correlated and high-In SNPs ( Table 1 ) . We next studied two European populations: a Spanish sample and a broadly defined Caucasian sample ( Figure 1 ) . The dataset of 9 , 668 SNPs was reduced to two principal components . Two Spanish subjects were clear outliers ( unpublished data ) and were removed from this analysis . The correlation coefficient between the actual membership of each individual to one of the two samples and the predicted membership using PCA on all SNPs and k-means is 0 . 9 . Analysis with small subsets of informative markers ( ten high-In or 20 PCA-correlated SNPs ) quickly converges to the results of analyzing the full dataset . The European American sample represents cell lines curated at the Coriell Insitute and although the degree of Spanish admixture in this sample is not known to us , it does not seem to be significant according to our findings . PCA-correlated SNPs and high-In SNPs have an overlap of 27 . 5% and are also in LD , although slightly less than what we observed in other geographic regions ( Table 1 ) . For the Asian and American populations , 9 , 707 and 8 , 202 SNPs , respectively , were analyzed ( Figure 1 ) . Three principal components were retained in Asia and two in America . Again , the top PCA-correlated and the highest-In SNPs perform exceptionally well for the inference of population structure in these continents . For the Asian populations , as few as ten PCA-correlated SNPs are required for almost perfect assignment , whereas a greater number of high-In SNPs is required ( about 80 SNPs for convergence to 1 ) . Moving to America , ten carefully chosen SNPs can distinguish between the Quechua and the Nahua . It should be noted that the 76 . 5% overlap between PCA-correlated and high-In SNPs in this continent , is the highest that we observed in this study ( in Asia the corresponding overlap was 49% ) ; see Table 1 for details . Both our method and the In metric do not suffer from selecting a large number of redundant SNPs in this dataset ( Table 2 ) . For the top 200 PCA-correlated SNPs within each geographic region , we calculated r2 between all pairs . Out of the thousands of possible pairs very few are actually in high LD . The same is true for the top 200 In SNPs selected to cluster populations within continents . We finally explored the feasibility of accurately clustering two populations of related ancestry , the Han Chinese and the Japanese , using data available from the HapMap database [37 , 38] . We downloaded all available data ( release 21–1r; 3 , 776 , 850 SNPs ) and excluded from the analysis SNPs that were fixed in both populations ( 1 , 356 , 867 SNPs ) and had more than one missing entry in both samples . This left us with 1 , 662 , 041 SNPs genotyped for 45 Chinese and 45 Japanese samples . We ran PCA and k-means on all available SNPs using the two principal components detected as significant . As a result , one Japanese individual was misclassified; this individual is a clear outlier ( Figure 4 ) . We then selected a subset of SNPs that could be used to infer this population structure , using both our method and the In measure . We found that with 50 PCA-correlated SNPs only two Japanese and one Chinese are misclassified ( correlation coefficient 0 . 97 ) , while with 150 PCA-correlated SNPs we are able to exactly reproduce the results of the analysis on the complete dataset of roughly 1 . 7 million SNPs . High-In SNPs are quite efficient as well . However , when choosing between 50 and 300 high-In SNPs , our algorithm for determining the number of principal components detects a very large number of significant principal components ( more than 40 ) . This causes the artifact of Figure 4C ( a drop in the performance of high-In SNPs ) . Manually fixing the number of principal components that we use for k-means clustering to two corrects this artifact , as the dotted line shows in Figure 4C . Such artifacts do not seem to arise with our method of choosing PCA-correlated SNPs in our empirical evaluation . We investigated whether SNPs that were selected for assigning individuals to clusters in one continent would be useful in another continent or for intercontinental differentiation . In an effort to answer this question , we tested the panels that we selected in each of the four continental regions that we studied ( both using the PCA-correlated measure and In ) for population clustering in another continent . Results were very poor and it seems that SNPs chosen for ancestry inference in one continent are in general no better than random SNPs and not transferable to other continents . The fact that the average overlap over selected SNPs across the different continental regions is about 2% underlines these results . In Figure S1 , representative results are shown for testing the portability of SNPs chosen in Europe and Africa in order to infer structure in the remaining three continents for which data were available . In a similar fashion , it seems that neither PCA-correlated SNPs nor high-In SNPs for intercontinental clustering can resolve the population structure within continents any better than randomly chosen SNPs ( Figure S1 ) . Differentiation among the Asian populations studied here is an exception and the correlation coefficient between actual and predicted membership exceeds 0 . 9 when studying 60 or more PCA-correlated SNPs that were ascertained for intercontinental clustering . In this case , the overlap between PCA-correlated SNPs selected for differentiation within the three Asian populations that we studied and intercontinental clustering is somewhat high ( 9% ) . Next , we explored the possibility of ascertaining a general SNP panel that could be used for ancestry inference and the study of population structure around the world . Results are shown in Figure 5 . We excluded admixed populations from this analysis ( African Americans and Caucasians ) , and studied nine indigenous populations for which data were available ( Mbuti , Mende , Burunge , Spanish , Mala , East Asian , South Altaian , Nahua , and Quechua ) . We ran PCA on 9 , 160 SNPs and k-means clustering on the ten detected significant eigenvectors , and managed to successfully assign each individual to their country of origin . Notice that the 3-D plot presented in Figure 5 is somewhat deceiving , as our method picked ten principal components as significant . If visualization in a 10-D space were possible , further differentiation of the studied populations would become apparent . Investigating the possibility of identifying a small set of SNPs to reproduce this structure , we selected sets of ten to 400 PCA-correlated and high-In SNPs and repeated the analysis . Surprisingly , using only 50 PCA-correlated SNPs , we were able to correctly assign all individuals to one of the nine studied populations . On the other hand , in this case , high-In SNPs do not seem to do any better than randomly selected SNPs ( Figure 5 ) . In order to test how the set of PCA-correlated SNPs is modified each time an additional population is added to the analysis , we studied incrementally distinct subsets of the data and compared the SNPs selected as structure informative in each subset to the panel of 50 SNPs that are sufficient for accurate clustering of the individuals to nine different populations ( see experiment described above ) . The first subset of populations that we analyzed consisted of genotypes for one population from each continent ( East African , Spanish , East Asian , and Nahua ) and in each round one additional population was added randomly to the analysis . Results are shown in Table 3 . Clearly , there is significant overlap between the panels of informative SNPs at each stage , which increases as more populations are added . Finally , we investigated the applicability and efficiency of our method to select structure informative SNPs in admixed populations . To this end , we studied two independent samples of Puerto Ricans . The first dataset ( Puerto Rican A ) is a sample of 192 Puerto Ricans [43] genotyped for approximately 100 , 000 SNPs , 7 , 259 of which overlapped with our worldwide dataset and were included in our analysis . The second Puerto Rican dataset ( Puerto Rican B ) , has been described in [36] and constitutes a sample of 19 individuals , genotyped for the same markers as the rest of the worldwide samples that we analyzed so far . It is well known that Puerto Ricans are genetically complex and composed of various proportions of Native American , African , and European genetic origins . We first investigated the Puerto Rican A dataset , and explored the ancestry of the 192 individuals across the African–European and the African–European/American axis ( Figure 6 ) . PCA performed on the Puerto Rican sample alone , revealed two significant principal components . We then added data on Europeans ( Spanish and Caucasians ) , West Africans ( Burunge ) , and Native Americans ( Quechua , Nahua ) to this analysis . Four principal components were found significant . We calculated the variance of the 192 Puerto Ricans across the two axes of ancestry . We observed that the sample variance across the African–European axis was six times larger than the variance across the African–European/American axis , which indicated that our sample had very little interindividual variation in Native American ancestry . Given this observation and for simplicity of exposition we only analyzed variation across the African–European axis . As is clear in Figure 6 , the Puerto Ricans that we studied lie virtually on a straight line between Africans and Europeans and are much closer to Europeans than Africans . We interpreted individuals from the Puerto Rican sample as a combination of European and African ancestry , with the proportion of admixture being equal to the distance of each individual from the centroid of the ancestral population ( ancestry coefficient , see Methods for details ) . After evaluating the ancestry of all 192 individuals using all available SNPs , we attempted to accurately predict it using a subset of PCA-correlated SNPs , selected on the Puerto Rican data only . No information from ancestral populations was necessary . As shown in Figure 6 , we accurately predict the ancestry coefficient of each individual . For example , the Pearson correlation coefficient between true and predicted ancestry is higher than 0 . 8 using 30 PCA-correlated SNPs and higher than 0 . 9 using 100 PCA-correlated SNPs . Finally , we cross-validated these findings by applying the panel of PCA-correlated SNPs that we selected on the Puerto Rican A dataset to infer individual ancestry in the Puerto Rican B dataset . As shown in Figure 6 , the SNP panel that we selected in Puerto Rican A performs equally well on Puerto Rican B . Notice , however , that this time random SNPs do better than before , perhaps due to the small number of individuals in this sample ( 19 individuals ) .
Geographic ancestry can be inferred from genotypic data [16 , 44–47] . The Bayesian approach implemented by Pritchard et al . [14] and PCA have been the two main tools of choice for identification of population structure and subdivision [18 , 36 , 48] . Recent studies have demonstrated that PCA is a fast , easy-to-implement method with great power for analysis of the very large datasets that are increasingly becoming available [13 , 19 , 20 , 22] . Extending recent algorithmic work that provably extracts matrix columns that correlate well with the dominant subspaces identified by PCA , we have developed a method to ascertain a small subset of SNPs that explicitly capture the structure of a population as identified by PCA . The population structure informative SNPs that are selected by our algorithm are also ancestry informative , and we show that they can be effectively used to assign individuals to different continents or populations . Achieving , in most cases that were studied here , 99% genotyping savings , these panels of SNPs can be used to reduce considerably the number of markers needed for ancestry inference . This is desirable in a variety of different research scenarios , such as association mapping ( where unrecognized population stratification may lead to spurious associations with disease ) , forensics , conservation studies , and population genetics [10 , 30 , 49–52] . For instance , we believe that our method is useful for investigators conducting association studies in two-stage designs and seeking to replicate the stratification assessment they are doing with the first stage ( e . g . , from a genome-wide dense SNP screen ) to a lower throughput method . For these types of applications , our method would be able to faithfully replicate this assessment with minimal additional markers in the second stage . However , a precondition for replication is that the second population contains the same substructure as the first population . Our algorithm is simple and computationally fast ( less than one minute for the largest runs presented here ) and thus allows the analysis of very large genome-wide datasets with thousands of individuals . Perhaps the most important advantage of our method for selecting PCA-correlated SNPs is that it is nonparametric and does not rely on any assumptions or modeling of the data . We simply detect SNPs that are correlated with the subspace spanned by the top few eigenvectors after determining the number of significant principal components . All other methods in the literature that are used to identify ancestry informative markers either rely on a specific model or are frequency based and demand prior knowledge of the origin of individuals [24–29] . Situations may exist where the use of prior information about the studied populations is desirable and we are currently working towards extensions of our approach to such settings . It should be noted that the final number of SNPs needed to describe population structure is not directly provided by the method and can only be estimated through empirical evaluation of a specific dataset . Also , in applying k-means clustering , we manually fixed the number of clusters to the number of populations in the data . It should be made clear , however , that the identification of the number of clusters in the data is not necessary for the implementation of our method; k-means clustering is only used here for demonstrating the efficiency of our approach . PCA-correlated SNPs are computed and can be used independently of k-means clustering . We chose k-means simply because it is a well-known and widely applicable clustering algorithm , which has numerous efficient implementations . One might experiment with different ( and perhaps more accurate ) clustering algorithms [53] , such as hierarchical clustering , spectral clustering , k-median approaches , etc . We were unable to compare the SNPs we selected as ancestry informative using our algorithm to published lists of ancestry informative markers [54–60] because the overlap between these lists and the SNPs that were available to us was either extremely small or different populations were analyzed . However , we have compared the efficiency of our method to selecting ancestry informative markers using the popular measure of In [26] . Rosenberg et al . [26] have previously compared this metric to other frequency-based measures that have been used for the selection of population differentiating markers ( FST , δ , etc . ) , and concluded that it was well correlated with other measures , equally efficient , and in some cases easier to use . SNPs that were selected using our PCA-correlated measure achieved comparable performance to high ranking In SNPs for recovering population structure in the datasets we studied . Interestingly , there is considerable overlap between the SNPs selected by the two different algorithms . It seems that very often , our method selects either the same SNPs or SNPs that are in high LD with those selected using the In measure . This is not necessarily surprising , since our approach is ranking markers based on how well they are recreating the fundamental structure in the data , and high-In SNPs are those that are also most likely to be associated with major clusters in the genotypic data independent of their location [14] . Dissecting substructure in admixed populations is a central challenge in association studies , especially for common complex disorders [3 , 5 , 11 , 47 , 61] , and our approach may prove to be particularly important in such settings . If allele frequency–based measures are to be used for the identification of a small number of structure informative markers , assumptions need to be made about the origin of the parental source populations and the extent to which they have each contributed to the admixed population . In some cases , ancestral populations may require complex sampling or may even no longer exist [28 , 62 , 63] . With the method that we propose here , there is no need to trace the origins of an admixed population in order to define markers that accurately capture the substructure of the population , and our ongoing work is exploring the applicability of our methods on large samples of admixed populations . As we have shown here , analyzing two independent Puerto Rican datasets , PCA-correlated SNPs can be successfully used to reproduce the structure of admixed populations and predict the ancestry proportions of the studied individuals . Interestingly , we found that interindividual variation across the Native American axis in the Puerto Rican samples that we studied was very low , perhaps depicting the fact that admixture with Native Americans occurred very long ago , and was random over several generations . Our findings demonstrate that to a large extent , SNPs identified as structure informative in one geographic region are not portable for the analysis of populations in a different geographic region , suggesting that the forces that shaped population structure in each geographic region have influenced different parts of the genome . However , analyzing jointly nine populations from around the world and 9 , 160 SNPs , we showed that using 50 PCA-correlated SNPs we can assign the studied individuals with 100% accuracy to their population of origin . SNPs with high-In rankings did not perform any better than random SNPs in this particular setting . One reason underlying the success of our approach may be the fact that it has been explicitly designed to converge to the results of PCA , whereas , to the best of our knowledge , this argument does not necessarily apply to In . Nevertheless , In does work well in most cases . Even though our results suggest that our method is powerful enough to be used for the identification of a universal panel of SNPs for the analysis of different populations from around the world , we also showed that each time a new population is added to the analysis , the panel of SNPs needed for population differentiation is modified . So , it should be made clear that we only studied a few representative populations from each continent and much more detailed studies are needed in order to test a universal structure informative SNP panel . This is also true for each of the continental regions that we discussed . We believe that many more population samples should be analyzed in order to accurately define a set of SNPs that could be used to reproduce fine-resolution population structure in a given geographic region . We have not dealt with the effect of local LD on the results of our algorithm and PCA in general . We showed that given the worldwide dataset that we analyzed here , structure informative SNPs picked by our method are not redundant for the most part in terms of LD . However , as SNP scans become denser , local LD will become a prominent feature of a dataset and we are currently working to see how this affects PCA . At the same time , since our method is not allele frequency based , it is possible that we are able to pick up global correlations among SNPs and haplotype patterns , and more research is necessary to clarify the relationship between the output of PCA and LD . In summary , we have developed a fast and simple algorithm for the selection of SNPs that uncover the structure of populations without knowing a priori the origin of individuals . After extracting meaningful dimensions from a dataset using PCA , we pick small sets of markers that retain the information carried in the full dataset . We believe that PCA-based algorithms will prove to be an invaluable tool for geneticists in a world of complex and ever-increasing genome-wide data .
The first dataset we used has been described in detail previously [36] . Briefly , we studied here 274 individuals from 12 populations ( 20 Mbuti , 20 Mende , 22 Burunge , 42 African Americans , 42 Caucasians , 20 Spanish , 11 Mala , 20 East Asians , 20 South Altaians , 20 Nahua , 20 Quechua , and 19 Puerto Ricans ) . Three of these populations are admixed ( Caucasians , African Americans , and Puerto Ricans ) . All individuals were typed using the 10K Affymetrix array . We also analyzed data available from the HapMap database on four populations ( Yoruba , CEPH , Han Chinese , and Japanese; release 21–1r ) . Finally , we studied a dataset of 192 self-described Puerto Ricans collected in New York and Puerto Rico as part of an asthma association study [43] . This sample has been genotyped using the 100K Affymetrix chip but we only analyzed here genotypes for the 7 , 259 SNPs that overlapped with the 10K array . We transformed the raw data to numeric values , without any loss of information , in order to apply SVD . Our data on a population X consist of m subjects; for each subject n , biallelic SNPs have been assayed . Thus , we are given a table TX , consisting of m rows and n columns . Each entry in the table is a pair of bases , ordered alphabetically . We transform this initial data table to an integer matrix AX , which consists of m rows , one for each subject and n columns , one for each SNP . Each entry of AX will be −1 , 0 , +1 , or empty . Let B1 and B2 be the bases that appear in the j-th SNP ( in alphabetical order ) . If the genotypic information for the j-th SNP of the i-th individual is B1B1 the ( i , j ) -th entry of AX is set to +1; else if it is B1B2 the ( i , j ) -th entry of AX is set to 0; else if it is B2B2 the ( i , j ) -th entry of AX is set to −1 . In order to handle missing data without rejecting too many SNPs that may contain important structural information , we first removed all SNPs with more than 10% missing entries . ( This was done independently for each experiment that we ran . ) This results in an average of roughly 2% of missing entries in each SNP . We subsequently filled in the missing entries using a least-squares regression-based technique from [64] . This technique fills in the missing data by using all available information from similar SNPs in the matrix . Since we ran this technique on groups of populations and not on each population individually , this filling in of the missing entries would tend to make SNPs more uniform across the different populations , instead of introducing artifactual biases . A more conservative approach would be to randomly fill in the missing entries with −1 , 0 , or +1 with probabilities respecting Hardy–Weinberg equilibrium . This approach returned similar results in most cases . However , we chose to employ the “best-guess” approach in order to preserve as far as possible the properties of each studied SNP . For the HapMap data on the Han Chinese and Japanese individuals , given the abundance of SNPs , we set the threshold to one missing entry per SNP , which was filled in as described above . Given the filled-in data matrix A , we applied SVD on A in order to compute its singular vectors and values . We would like to note that , from a mathematical perspective , our procedure is exactly equivalent to applying PCA on the covariance matrix AAT , which is an m × m matrix measuring the angular distance between all pairs of individuals . ( From a mathematical perspective , SVD enjoys very strong optimality properties , see [65 , 66] for details . ) After determining the number of significant principal components , k-means clustering was applied on low-dimensional data in order to split the individuals to their respective populations . For concreteness , consider the SNP data matrix that emerges from the HapMap Han Chinese and Japanese populations , where the data matrix A has 90 rows and approximately 1 . 7 million columns . Two significant principal components were identified , and we denote the corresponding two left singular vectors by u1 and u2 ( eigenSNPs ) . Recall that these are 90-D vectors ( each vector has 90 entries , each corresponding to one individual in the Han Chinese–Japanese dataset ) . Plotting the 90 individuals with respect to u1 and u2 , i . e . , if is the i-th entry in u1 and is the i-th entry in u2 , the coordinates of the i-th individual are ( , ) , results in Figure 4 . Clearly , the two populations are two separate clusters , with the exception of one individual who is roughly in the middle . Running k-means on these 2-D coordinates results to an almost perfect clustering . We summarize the algorithm for selecting PCA-correlated SNPs . Let k be the number of significant principal components ( see below ) . Return the columns ( SNPs ) of A that correspond to the top r pj's . An implementation of our method is posted at http://www . cs . rpi . edu/~drinep/PCASNPS . Informativeness for assignment ( In ) was computed using the algorithm described previously in [26] . In order to compare two clusterings , we simply compute the correlation coefficient ( normalized inner product ) of the cluster indicator vectors . This is effectively the Pearson correlation coefficient without the mean centering; recall that our vectors are zero–one vectors . For example , in the HapMap Han Chinese and Japanese experiment described above , given a total of 90 individuals , the ground truth cluster indicator vector for the Han Chinese population is a vector whose first 45 entries are set to one and the remaining entries are set to zero . After running k-means , two clusters emerge: the one corresponding to the Han Chinese population has the first 45 entries set to one , as well as the 73rd entry , whereas all remaining entries are set to zero . The correlation coefficient between the two indicator vectors ( and thus the respective clusters ) is 0 . 99 . Correlation coefficients range between zero and one; we report the average correlation coefficient of ground truth clusters and the clusters that emerge after running PCA and k-means on the selected sets of SNPs . We outline our analysis of the Puerto Rican dataset A ( 192 individuals ) . In Figure 6 , we calculated the centroids of the European ( Spanish and Caucasians ) , West African ( Burunge ) , and Native American ( Quechua and Nahua ) populations . ( Four principal components were identified as significant for the joint data , hence we worked on a 4-D space; Figure 6 plots the three most significant dimensions . ) We now defined two perpendicular axes of variance: one joining the centroids of the European and West African populations , and the other projecting the centroid of the Native American population on the European–West African axis . We subsequently computed the coordinates of each Puerto-Rican individual on the coordinate system defined by these new axes . We interpreted the resulting coordinates as ancestry information for each individual across the two axes . A simple variance analysis showed that the variance across the European–West African axis is dramatically larger than the variance across the other axis . Hence , we focused our analysis on predicting the relative location of each Puerto Rican subject with respect to the centroids of the European and West African populations using a small subset of PCA-correlated SNPs . We computed the Pearson correlation coefficient between our prediction and the “ground truth” value that was computed using all available SNPs and reported the results in Figure 6 . We would like to emphasize that the PCA-correlated SNPs were selecting by looking only at the Puerto Rican dataset A , with no information from the European or West African populations . In order to determine whether the k-th principal component of an m × n data matrix A is significant , we will compare the part of A—denoted by Am − k—that corresponds to the k-th and smaller principal components to a random matrix Ãm − k that emerges by randomly permuting all the entries of Am − k . ( We actually repeat this process ten times and average the results . It should be noted that the variance of this process is very small; for our data , it was orders of magnitude smaller than the mean . Thus , a small number of repetitions suffices . ) We now compute the top singular value of Am − k and the top singular value of Ãm − k . This is a standard metric that compares the structure of a given matrix with a random matrix . If the ratio of the top singular value of Am − k over the top singular value of Ãm − k is more than 115% , then we call the i-th principal component significant; otherwise we discard it . Essentially , we retain a principal component if it has 15% more structure than a random one with the same entries . The 15% value was chosen after extensive experimentation on the available data and performed well in all test cases . The method is computationally fast and runs in a few minutes even on the largest dataset we analyzed here ( HapMap Han Chinese and Japanese populations ) . Theoretically , it scales linearly with the number of individuals and number of SNPs . Two special cases exist . The minimal number of principal components that we keep is at least two . In all populations—except for the combination of Europeans and Spanish—at least two principal components are returned by the aforementioned algorithm as well . However , in order for our PCA-correlated SNPs algorithm to identify the appropriate correlations if exactly one principal component is kept ( in which case the associated subspace is just a line ) , we need some normalization of the original data ( e . g . , mean centering ) . To avoid this unnecessary complication , we always keep at least two principal components , which fixes this issue by embedding the data in—at least—the Euclidean plane . The other special case is when too many principal components ( e . g . , more than 80% of all principal components ) are selected by the above algorithm . In this case , we simply skip dimensionality reduction and directly cluster the original data . This never appears when using all SNPs , but may appear when a small number of SNPs is selected from a very large dataset ( e . g . , ten out of 10 , 000 SNPs ) . In Figure S2 , we show that as more PCA-correlated SNPs are picked , we can approximately identify the number of principal components that were significant in the original dataset . Finally , we should mention that the aforementioned test could potentially be replaced by the test proposed by Patterson et al . [20] . The two tests are actually very similar in spirit . They both draw their motivation from the theoretical analysis of the eigenvalues of a matrix whose entries are drawn independently from some distribution with bounded variance . Our test is heavily influenced from the seminal paper of Füredi and Komlós [41] , who proved that the eigenvalues of a matrix with the above properties satisfy certain bounds . This provides an elegant way to test for structure in a matrix [42] . Similarly , [20] is influenced by analogous statistical results; we feel that [41 , 42] require fewer assumptions and thus might be more generally applicable . | Genetic markers can be used to infer population structure , a task that remains a central challenge in many areas of genetics such as population genetics , and the search for susceptibility genes for common disorders . In such settings , it is often desirable to reduce the number of markers needed for structure identification . Existing methods to identify structure informative markers demand prior knowledge of the membership of the studied individuals to predefined populations . In this paper , based on the properties of a powerful dimensionality reduction technique ( Principal Components Analysis ) , we develop a novel algorithm that does not depend on any prior assumptions and can be used to identify a small set of structure informative markers . Our method is very fast even when applied to datasets of hundreds of individuals and millions of markers . We evaluate this method on a large dataset of 11 populations from around the world , as well as data from the HapMap project . We show that , in most cases , we can achieve 99% genotyping savings while at the same time recovering the structure of the studied populations . Finally , we show that our algorithm can also be successfully applied for the identification of structure informative markers when studying populations of complex ancestry . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"homo",
"(human)",
"genetics",
"and",
"genomics",
"computer",
"science"
] | 2007 | PCA-Correlated SNPs for Structure Identification in Worldwide Human Populations |
Ribonuclease P ( RNase P ) is an essential enzyme required for 5′-maturation of tRNA . While an RNA-free , protein-based form of RNase P exists in eukaryotes , the ribonucleoprotein ( RNP ) form is found in all domains of life . The catalytic component of the RNP is an RNA known as RNase P RNA ( RPR ) . Eukaryotic RPR genes are typically transcribed by RNA polymerase III ( pol III ) . Here we showed that the RPR gene in Drosophila , which is annotated in the intron of a pol II-transcribed protein-coding gene , lacks signals for transcription by pol III . Using reporter gene constructs that include the RPR-coding intron from Drosophila , we found that the intron contains all the sequences necessary for production of mature RPR but is dependent on the promoter of the recipient gene for expression . We also demonstrated that the intron-coded RPR copurifies with RNase P and is required for its activity . Analysis of RPR genes in various animal genomes revealed a striking divide in the animal kingdom that separates insects and crustaceans into a single group in which RPR genes lack signals for independent transcription and are embedded in different protein-coding genes . Our findings provide evidence for a genetic event that occurred approximately 500 million years ago in the arthropod lineage , which switched the control of the transcription of RPR from pol III to pol II .
RNase P catalyzes the essential removal of the 5′ leader sequence from precursor tRNAs ( pre-tRNAs ) [1]–[5] . With the exception of some protein-only variants in eukaryotes [6] , [7] , RNase P is a ribonucleoprotein ( RNP ) complex that consists of a catalytic RNA ( RNase P RNA , RPR ) and as many as ten protein cofactors ( RNase P proteins , RPPs ) in eukaryotes , up to five protein cofactors in archaea , and just one in bacteria [1] , [2] . Conserved sequences and structural elements ( including the active site ) in all RPRs are suggestive of a shared evolutionary ancestry . By contrast , homology among RPPs is restricted to those of archaea and eukaryotes . Biochemical characterization of bacterial RNase P has provided insights into how a single protein cofactor aids RNA catalysis by enhancing affinity for metal ions and substrate recognition [8] , [9] . Comparisons of bacterial RNase P to its multi-subunit archaeal and eukaryotic counterparts provide an opportunity to examine whether structural and functional attributes of the RPR have been appropriated by additional protein cofactors . Of additional interest is understanding the role of these RPPs in regulating the function of RNase P during development and in response to environmental cues . In our efforts to develop Drosophila RNase P as a multicellular eukaryotic experimental model , we examined the transcription of RPR , and our work has unexpectedly shed some light on the evolution of this ancient ribozyme . Eukaryotic RPRs that have been analyzed to date , ranging from yeast to human , are transcribed by pol III [2] , [10]–[13] . The RPR gene in all Drosophila species examined [14]–[16] has been annotated in the last intron of ATPsynC/CG1746 [17] , the pol II-transcribed gene that encodes subunit C of the F0 complex that is part of the mitochondrial ATP synthase [15] . We showed that the RPR locus within this gene does indeed produce , in a splicing-independent fashion , a functional RPR . In subsequent analysis of genomic databases , we found that such embedding of the RPR gene within a pol II-transcribed gene is also a characteristic in all insects and crustaceans examined . This common feature within a major group of arthropods suggests that the change from pol III to pol II transcription of RPR occurred approximately 500 million years ago [18] .
Each of the twelve species of Drosophila for which genome sequence is available has a single copy of the RPR gene . In all cases , the RPR gene is inserted in the last intron of a recipient gene , ATPsynC/CG1746 , with both genes arranged in the same 5′ to 3′ orientation ( Fig . 1A ) . The RPR sequence is conserved , as are the ATPsynC exons and UTRs , but the other introns of the gene are not conserved ( Fig . 1B ) . In keeping with a functional role , RPR-derived RNAs accumulate at higher levels ( 3- to 5-fold in polyA+ samples and 20-fold in total RNA samples ) than those corresponding to the preceding intron ( Fig . 1C ) . Like its recipient gene , RPR is expressed throughout development in D . melanogaster and in multiple tissues ( Fig . 1C and S1 Fig . ) . Although the expression data suggest that RPR-derived RNAs are expressed , we could not identify an RPR promoter by sequence analysis . The flanking sequences required for transcription by pol III , which are found in known eukaryotic RPR genes [10]–[12] , [19] , are absent in the vicinity of the Drosophila ATPsynC-RPR genes ( S2A Fig . ) . The Drosophila RPR genes also lack internal pol III recognition sequences that are characteristic of tRNA genes ( S2A Fig . ) [16] , [20] . Analysis of data from genome-wide chromatin immunoprecipitation ( ChIP ) assays in D . melanogaster shows binding of pol II in the 5′ region of the ATPsynC-RPR locus ( Fig . 1D ) [20] , but ChIP studies mapping pol III binding in Drosophila do not identify a pol III target in the vicinity of the RPR genes [16] , [20] . Together , these findings show that Drosophila RPR is expressed , but sequence analysis did not identify a pol III promoter that could drive its independent expression . The insertion of the RPR gene , which apparently lacks an independent promoter , into the last intron of ATPsynC in Drosophila suggests that RPR may be transcribed from the recipient gene promoter . To test this idea experimentally , we generated a reporter gene with the RPR-coding intron from D . virilis inserted between two red fluorescent protein ( RFP ) exons ( Fig . 2A ) . The reporter gene was tested in D . melanogaster S2 cells in which D . virilis RPR can be distinguished from the endogenous D . melanogaster RPR by size and sequence differences . Transfected S2 cells expressed RFP from the reporter gene when driven by an Actin 5C ( Act5C ) promoter ( pol II ) [21] . RFP expression indicated that all the cis-elements required for correct splicing of the intron were present in the construct . We analyzed RNA products from the reporter gene using RT-PCR and northern analysis . As expected from RFP expression , the mature RFP mRNA was expressed ( R2 in Fig . 2B ) . D . virilis RPR was also expressed , demonstrating that the intron contained all the sequences necessary for production of mature RPR ( R2 in Fig . 2C ) . The D . virilis RPR co-purified with RNase P activity ( S3 Fig . ) , indicating that it assembled with endogenous RPPs to form a functional holoenzyme . D . virilis RPR was also expressed from a reporter gene with a UAS-Hsp70 promoter [22] , showing that the production of RPR is not dependent on the identity of the pol II promoter ( S4 Fig . ) . In a reporter gene lacking a promoter sequence ( Fig . 2A ) , no RPR was detected by northern analysis ( R1 in Fig . 2C ) . This finding ruled out the possibility that the RPR gene was transcribed by a cryptic promoter in the intron that we could not identify by sequence analysis . Importantly , the failure to produce RPR also showed that transcription of RPR was dependent on the pol II promoter of the recipient gene . To assess if splicing is required for processing of RPR from the intron , we designed splicing-deficient reporter genes and analyzed the RNA products using RT-PCR and northern analysis . We tested two reporter genes , one with a 5′ splice-site mutation and another with both 5′ and 3′ splice-site mutations ( R3 and R4 in Fig . 2A ) . These mutations effectively blocked splicing as only the pre-mRNA for RFP was detected in the cells and no RFP expression was observed ( Fig . 2B ) . In contrast , mature RPR accumulated ( Fig . 2C ) , indicating that splicing is not required to process RPR from the primary transcript . The embedded Drosophila RPR is the only RPR copy in the genome suggesting that it fulfills the essential function as the ribozyme component of RNase P . We examined the association of the RPR with Drosophila RNase P to verify its functional role in the enzyme . The holoenzyme was partially purified from D . melanogaster S2 tissue-culture cells using sequential , ion-exchange chromatography on DEAE-Sepharose ( anionic ) followed by SP-Sepharose ( cationic ) . The presence of RNase P activity in fractions from each matrix was detected using a pre-tRNA processing assay ( Fig . 3A ) [23] . Peak activity from both matrices was found in fractions eluted with 300 to 500 mM NaCl . D . melanogaster RNase P cleaved pre-tRNAGly to yield two products identical in size to those generated by the Escherichia coli enzyme , which was used as a reference standard ( Fig . 3A ) . The mature tRNA resulting from cleavage by D . melanogaster RNase P had a 5′ phosphate on G+1 , an end group expected from RNase P catalysis ( Fig . 3B ) . This inference was based on finding guanosine-3′ , 5′-bisphosphate ( pGp ) in a thin-layer chromatogram of the products from RNase T2 digestion of D . melanogaster RNase P-generated mature tRNAGly . RPR present in the SP-Sepharose fractions was then detected using reverse-transcription and PCR ( RT-PCR ) . The enrichment of RPR in fractions that also showed RNase P activity is consistent with its co-purification with the holoenzyme ( Fig . 3A ) . To test if this co-purified RPR is required for RNase P activity , we designed an antisense RNA oligonucleotide ( α-RPR-j7/2 ) that is complementary to a predicted single-stranded region that is part of the RPR active site ( Fig . 3C ) . Incubation with α-RPR-j7/2 inhibited RNase P activity in a concentration-dependent fashion ( Fig . 3D ) . In contrast , another oligonucleotide with the same nucleotide composition as α-RPR-j7/2 but a scrambled sequence ( sc-RPR-j7/2 ) was ineffective at inhibiting activity even at the highest concentration tested . Together , these results confirm that the intronic RPR encodes the RNA component of D . melanogaster RNase P and is required for its activity . To determine if the insertion of RPR in a recipient gene is unique to the Drosophila genus or more widespread in the animal kingdom , we analyzed RPR genes in the genomes of additional animals . All newly identified genes were verified to encode RPRs by their resemblance to typical eukaryotic RPRs in secondary structures and location of conserved nucleotides , including those essential for catalysis ( S5 Fig . ) [13] . Strikingly , we found a divide that classifies animals into two groups— ( i ) insects and crustaceans that have embedded RPR genes lacking signals for pol III transcription ( Fig . 4 , Fig . 5 , S2A Fig . and S6A Fig . ) , and ( ii ) other animals that have typical signals for pol III-dependent transcription ( Fig . 4 , Fig . 5 , and S2B Fig . ) . We draw these conclusions from an examination of species in the four subphyla of extant arthropods [Hexapoda ( Insecta and Entognatha ) , Crustacea , Myriapoda , Chelicerata] and some non-arthropods that had not been previously examined . Within the Hexapoda and Crustacea , we examined species in eight orders of insects and three orders of crustaceans . All these RPR genes lack signals required for pol III transcription ( S2A Fig . ) . In 26 out of 27 insect species , the RPR gene is present in an annotated pol II-dependent recipient gene and oriented in the same 5′ to 3′ direction ( Fig . 4 ) . The one exception is Pediculus humanus ( human body louse ) where RPR is in a poorly annotated region . Nevertheless , it is likely that the P . humanus RPR is part of a recipient gene because it lacks signals for pol III transcription . In the case of Tribolium castaneum ( red flour beetle ) and Heliconius melpomene ( Postman butterfly ) , there are two copies of RPR in the same recipient gene ( Fig . 4 and S7B Fig . ) . The two RPR copies are present in tandem within the same intron in T . castaneum , while they are present in two different introns of the same gene in H . melpomene . We were unable to examine species in Entognatha , the other Hexapod class , because there is no genomic sequence available . In the five crustaceans that we examined , there are two or more RPR genes in a given species and all lack signals for pol III transcription ( Fig . 4 and S6 Fig . ) . For example , there are ten RPR-like genes in Daphnia pulex , which is consistent with the extensive gene duplications that have occurred in its genome ( S6A Fig . ) [24] . At least one D . pulex RPR gene is expressed [24] and may be a functional gene ( Fig . 4 ) . Finding an inserted type of RPR gene in insects and crustaceans is consistent with their close evolutionary relationship [25]–[28] ( Fig . 5 ) . Within the Myriapoda and Chelicerata , we examined one myriapod ( centipede ) and four chelicerates ( spider , tick , scorpion , and mite ) . All species have an RPR gene with typical signals for pol III-dependent transcription ( S2B Fig . ) . The same was found for five non-arthropod animal species we examined [two molluscs ( snail and oyster ) , two annelids ( polychaete worm and leech ) , and a sponge] in which RPR had not been previously analyzed ( S2B Fig . ) . RPR genes in all these non-insect and non-crustacean species are present in intergenic regions , except for the centipede Strigamia maritima , where the gene is found in an intron in the opposite orientation to the recipient gene . These genes have typical signals for pol III transcription ( Fig . 4 and S2B Fig . ) . These arthropods ( Myriapoda and Chelicerata ) and all other animals examined to date have what has been considered a typical RPR that is transcribed by pol III ( Fig . 4 and Fig . 5 ) . Although the initial insertion of RPR into a recipient gene in the arthropod lineage appears to have been a single event , RPR moved again multiple times after this event as shown by its association with several different recipient genes ( Fig . 5 ) . In the eight orders of insects we examined , five different recipient genes were identified ( Fig . 5 ) . RPR recipient genes were also different within an order; for example , RPR is present in Regulator of chromosome condensation 1 ( Rcc1 ) in mosquitoes , but it is in ATPsynC in the other species of Diptera . Using the recipient gene as an indicator , ATPsynC appears to be the oldest recipient gene for RPR in the insects , as it is the common recipient gene in species belonging to the most divergent orders—the highly derived Diptera and the basal Ephemeroptera and Odonata ( Fig . 4 and Fig . 5 ) . Moreover , in D . melanogaster , Ephemera danica ( mayfly ) and Ladona fulva ( dragonfly ) , RPR resides in the same intron providing further support for ATPsynC being the original recipient site in insects ( Fig . 4 , Fig . 5 , and S7A Fig . ) . Another common recipient gene for RPR is eukaryotic initiation factor 4B ( eIF-4B ) . RPR is present in eIF-4B in seven species belonging to three orders—Lepidoptera ( moths and butterflies ) , Coleoptera ( beetles ) and Hemiptera ( true bugs , including aphids ) ( Fig . 4 , Fig . 5 , and S7B Fig . ) . In five of the seven species , the insertion of RPR is in the same intron in eIF-4B . Although there is no significant conservation of its sequence , the intron can be identified based on the conserved amino acid sequence of the flanking exons ( S7B Fig . ) . Presence of the RPR in the same intron is consistent with a common ancestor for these orders , but this is not supported by a well-established insect phylogeny [25] . An alternative explanation is that these were independent events and examples of recipient-site convergence . This idea is supported by the case of the Asian citrus psyllid ( Diaphorina citri ) and the bull-headed dung beetle ( Onthophagus taurus ) where RPR is in different eIF-4B introns ( S7B Fig . ) , reflecting independent insertions of RPR into eIF-4B likely due to a bias for this recipient gene . In D . melanogaster , the homologs of recipient genes in other insects and crustaceans are all expressed throughout development and in multiple tissues , with ATPsynC being one of the most highly expressed genes ( S1 Fig . ) . This observation supports the idea that the expression pattern and level of expression may constrain possible recipient genes , so that only those genes with ubiquitous and high expression are suitable sites for insertion of RPR ( S1 Fig . ) . In Tribolium castaneum , there are two RPR genes embedded in tandem in the myosin binding subunit/protein phosphatase 1 regulatory subunit 12B-like gene ( Mbs/PPP1R12B ) . Although Mbs shows a low level of expression relative to the other recipient genes , the two copies of RPR may compensate for this ( Fig . 4 and S1 Fig . ) . Analyzing more insect genomes and transcriptomes will provide information about genomic contexts suitable for functional insertion of RPR and may reveal common features of recipient sites . RNase MRP has roles in mitochondrial DNA replication , nucleolar rRNA processing , and mRNA turnover , and is present only in eukaryotes . It is an RNP that shares eight protein subunits with RNase P [29] . Furthermore , the RNA subunit of RNase MRP ( MRP RNA ) resembles RPR and appears to have derived from a common ancestor by a gene duplication event early in eukaryotic evolution [17] , [30] . The two RNAs , albeit similar in secondary structure , have distinctive features that enable their unambiguous identification . Given our unexpected findings of a transcriptional switch for the RPR in insects and crustaceans , we conducted a survey of MRP RNA genes in 26 insect species ( in addition to Drosophila [14] , [31] ) . These newly identified genes encode bona fide MRP RNAs , as judged by secondary structures and the location of various previously established signature motifs; for example , a five-nucleotide “GARAR” consensus in L8 ( the terminal loop which caps the P8 helix; [17] ) is present in all of them . In all 26 cases , we found signals for pol III transcription ( S8 Fig . ) [14] , [31] . Therefore , MRP RNA genes , in contrast to RPR genes , appear to have maintained pol III regulation throughout the animal kingdom , including insects and crustaceans .
In Drosophila species , the RPR gene is embedded in the last intron of the ATPsynC gene . We found splicing was not required to produce mature RPR using an experimental reporter system . In the native context , RPR could either be generated from the spliced-out intron or from the primary transcript , with additional processing required to trim sequences beyond the mature RPR termini . Certain classes of micro RNAs ( miRNAs ) [33] and intron-derived small nucleolar RNAs ( snoRNAs ) [34] , [35] also require processing to generate their mature 5′ and 3′ termini . The intronic miRNAs , which also do not require splicing when assessed using reporters [33] , are processed to their mature lengths by Drosha and Pasha/DGCR8 [36] . It is unlikely these endonucleases trim Drosophila RPR , because their recognition sequences are absent in the regions flanking the mature RPR . In the case of intron-derived snoRNAs , examples of both splicing-dependent and splicing-independent processing are found , wherein nucleolytic trimming guides the maturation of the snoRNA termini following the assembly of snoRNP proteins [34] , [35] . Like snoRNP proteins aiding the processing of the snoRNAs , RPPs could play a role in the maturation of the intronic RPR , but details of the assembly of the RPPs on the intronic RPR remain to be investigated . To further understand the biogenesis of the intronic RPR , it will be important to identify the nucleases that act on the RPR ends to produce the mature form . We presume these enzymes were already present in the founder animal for processing other non-coding RNAs , and were co-opted to generate the mature RPR from the recipient gene transcript . If so , identifying the enzymes acting on RPR may also provide general information on the biogenesis of some other non-coding RNAs . It has been reported that some other non-coding RNAs show differences in their transcriptional control and are transcribed by pol II in some organisms and pol III in others [37] , [38] . The significance , if any , for the different mechanisms is unclear . One of the possible effects of a change in the transcriptional control of RPR is altered RNA activity ( for example , from differences in modification ) . Testing this idea using RPR produced in vivo by pol II or pol III in a pre-tRNA processing assay will provide a tractable experimental model for determining whether the transcriptional shift between pol II and pol III has functional consequences for a non-coding RNA . Based on sequence analysis , it has been hypothesized that RPR gene gave rise to the MRP RNA gene in eukaryotes , presumably through gene duplication followed by neofunctionalization of the new gene copy [17] , [30] ( Fig . 6A ) . While MRP RNA is under pol III regulation in all animals that we and others have examined , RPR has undergone a second genetic event that inserted it into a recipient gene in crustaceans and insects ( Fig . 6A ) . Current data indicate that this genetic change , which caused embedding of RPR within the arthropod lineage , occurred approximately 500 million years ago in an ancestor of the insects and crustaceans , an estimate that is placed prior to the emergence of the insects at approximately 479 million years ago [18] . The species of crustaceans we examined are examples of the so-called true crustaceans ( Vericrustacea ) [26] , which are closely related to the insects ( Hexapoda ) ; both are members of the epic Pancrustacea clade ( Fig . 6B ) [26]–[28] . The other major group of Pancrustaceans is the Oligostraca , that includes the seed shrimp , oar-feet , fish lice , and tongue worms , for which there is currently no genomic sequence . If Oligostraca species have an embedded RPR , this would support an earlier origin—in an ancestor of all pancrustaceans ( Fig . 6B ) . As more genomes become available , we will be able to refine when a pol II-regulated RPR first occurred and test whether it was indeed a single event in arthropod evolution . Generating an embedded RPR could have involved DNA- or RNA-mediated duplication and subsequent loss of any associated signals for pol III transcription [39] . It is also possible , given the catalytic function of RPR , that the insertion resulted from reverse splicing , similar to at least one route hypothesized for the spread of self-splicing group I and group II introns [40] , although this activity has yet to be demonstrated for RPR . Regardless of the mechanism , the insertion caused a change in the regulation of RPR so that it became dependent on pol II transcription . This is not the case for MRP RNA ( S8 Fig . ) , which shows the switch in transcriptional regulation occurred uniquely to RPR . As a first step towards determining any consequence of the change in RPR transcription , genome editing could be used to engineer , for example , D . melanogaster to have only a pol III-dependent RPR gene . Such a strategy would allow determination of the phenotypic consequences of reverting to the ancestral regulation of RPR . Following the initial event that caused embedding in a recipient gene , RPR moved again multiple times into different recipient genes ( Fig . 4 , Fig . 5 , and S7B Fig . ) . Insertion does not appear to be random because RPR inserted independently into the same gene more than once . In cases where RPR is present in two copies , such as T . castaneum ( Fig . 4 ) and H . melpomene ( S7B Fig . ) , both are present in the same recipient gene either in the same or two different introns , which is suggestive of local duplications . The crustaceans that we examined all had multiple RPR copies , however , these were associated with different recipient genes ( S6A Fig . ) . While we have not identified a ‘signature’ of an insertion site , it appears that in all instances a pol II-regulated RPR has been retained and no case of a pol III-regulated RPR was found . Our studies have shed some light on the evolution of RPR , a legacy of the RNA world and the first true trans-acting ribozyme discovered , and suggest that RPR transcription and subsequent processing entails the use of a different mechanism in a large group of animals . Although it is not known if this mode of biogenesis has functional consequences , our findings add to the variations in RNase P , an essential housekeeping enzyme , already noted for the diversity in its subunit composition [3] , [5] .
D . melanogaster S2 cells [41] were grown in Schneider insect medium ( Sigma ) with 10% ( v/v ) fetal bovine serum . DNA transfections were performed using Effectene ( Qiagen ) . Cells were harvested 30 h post transfection and total RNA was isolated using Trizol ( Invitrogen ) . The intronic region containing RPR was amplified from D . virilis genomic DNA using PCR and the following primers: forward primer virilis intron , 5′-CTGCTTCATCTACAAGGTTCGTATTGGTTACC-3′ and reverse primer virilis intron , 5′-CCGATGAACTTCACCTGTTGTATTGGTTGTC-3′ . A DsRed ORF ( from pP ( RedH-Stinger ) [42] was used as a template to generate two exons , separated at nucleotide 323 , which creates a match to the consensus Drosophila splice junction [43] . The exons were generated using PCR and the following primer pairs: Exon I forward primer RFP , 5′-TCCGATATCATGGCCTCCTCC-3′ and Exon I reverse primer RFP , 5′-GGTAACCAATACGAACCTTGTAGATGAAGCAG-3′; Exon II forward primer RFP , 5′-GACAACCAATACAACAGGTGAAGTTCATCGG-3′ and Exon II reverse primer RFP , 5′ -ACCTCTAGACTACAGGAACAGGTGGTG -3′ . The intron and the two exons were combined using overlapping PCR [44] and cloned into pPACPL , which contains the Act5C promoter [21] . DsRed exons were also generated with splice mutations . Splice mutations were created by site-directed mutagenesis with the following primer pairs: 5′ splice mutant forward , 5′-CTGCTTCATCTACAAGATTCGTATTGGTTACC-3′ and 5′ splice mutant reverse , 5′-GGTAACCAATACGAATCTTGTAGATGAAGCAG-3′; 3′ splice mutant forward , 5′-CAATGACAACCAATACAACCTGTGAAGTTCATCGGCGTGAACT-3′ and 3′ splice mutant reverse , 5′-AGTTCACGCCGATGAACTTCACAGGTTGTATTGGTTGTCATTG-3′ . In addition , the fragment containing the DsRed exons with the D . virilis intron was cloned into pUAST [22] to generate a reporter gene under control of the UAS promoter . This reporter was expressed using a Gal4 gene ( pGaTB ) [22] cloned into the pPACPL vector using the following primers , Gal4 forward , 5′-TCCGATATCATGAAGCTACTGTCTTCTATC-3′; Gal4 reverse , 5′-AAATCTAGATTACTCTTTTTTTGGGTTTGGTGGGGTATCTTC-3′ . cDNAs were prepared using an oligo dT primer ( for mRNAs ) or gene specific primers ( for RPRs ) by reverse transcription using an Omniscript RT kit ( Qiagen ) . cDNAs were amplified with Taq DNA polymerase ( NEB ) using the recommended conditions and the following primer pairs: Forward DmelRPR , 5′-AGTCAGTTGCAAACTAGCATC-3′ and Reverse Dmel RPR , 5′- AGTCAGTCACAGATTAGTCTGAATTG-3′; Forward GFP , 5′-TAAGATATCATGGTGAGCAAGGG-3′ and Reverse GFP , 5′- ACCTCTAGATTACTTGTACAGCTCGTCC-3′; Forward Oda , 5′-GTCCTTCGGTAGAGCGACAT-3′ and Reverse Oda , 5′- GCACCATCTCGACTTCGTCT-3′ . D . melanogaster and D . virilis RPRs were detected using full-length anti-sense RNA probes labeled with [α-32P]-ATP in an in vitro transcription reaction . The DNA templates were generated from PCR-mediated amplification of the genomic DNA using the following primers ( for both species ) : Forward primer-genomic , 5′-AGTCAGTTGCAAACTAGCATCTG-3′ and Reverse primer-genomic , 5′-TCACTATAGGAGTCAGTCACAGATTAGTCTG-3′ . A T7 RNA polymerase promoter was introduced to the above PCR product using a second round of PCR using the same forward primer and the following reverse primer: 5′-GAGAATTCTAATACGACTCACTATAGGAGTCAGTCACAG-3′ . D . virilis RPR was also detected using the following DNA oligo , 5′- CCGCGACACACAATCACCTCTCGGCTTTTGTATGTTGTTACAGCAAC-3′ . U6 RNA was detected using the DNA oligo , 5′- GCAGGGGCCATGCTAATCTTCTCTGTATCG-3′ . Both DNA oligos were 5′-labeled using [γ-32P]-ATP and T4 polynucleotide kinase . Eight micrograms of total RNA isolated from transfected cells was separated on a 7 . 5% ( w/v ) polyacrylamide gel containing 8 M urea , transferred to a nylon membrane ( Hybond N+ , GE Healthcare ) and analyzed by northern hybridization . After pre-hybridization in the same hybridization buffer , RNA probes were hybridized in hybridization buffer ( 5X SSC , 1% ( w/v ) SDS , 5X Denhardt's solution , 200 µg/ml of sheared salmon sperm DNA ) for 16 h at 65°C . DNA oligo probes were hybridized in QuikHyb buffer ( GE Healthcare ) for 16 h at 55°C . Membranes were washed with 2X SSC with 0 . 1% ( w/v ) SDS at 10°C below hybridization temperature . The binding of labeled probes to their complementary target RNAs was detected using phosphorimaging . D . melanogaster S2 cells were collected by centrifugation and washed once with phosphate-buffered saline . Packed cells ( 100 µL ) were lysed in 400 µL of lysis buffer [15 mM HEPES ( pH 7 . 9 ) , 3 mM MgCl2 , 50 mM NaCl , 1 mM dithiothreitol , 0 . 2 mM phenylmethylsulfonyl fluoride , 0 . 1% ( v/v ) Tween-20 , 10% ( v/v ) glycerol , 0 . 2 U/μL of Ribolock RNase Inhibitor ( Thermo Scientific ) ] . Cells were homogenized using a type A glass Dounce homogenizer ( Wheaton ) on ice and debris was removed by centrifugation at 2 , 500 g for 10 min . The crude lysate was mixed with 100 µL of diethylaminoethyl ( DEAE ) -Sepharose resin ( GE Healthcare ) , which had been pre-equilibrated with lysis buffer at 4°C for 30 min . The resin was collected by centrifugation ( 2 , 500 g for 5 min ) and washed twice , each with 1 mL of lysis buffer to remove weakly bound constituents . Fractions were eluted stepwise with increasing NaCl concentration ( from 50 mM to 1 M ) in lysis buffer , and tested for RNase P activity ( as described below ) . Fractions with detectable activity were pooled and dialyzed twice , each with 500 volumes of lysis buffer ( without NaCl and RNase inhibitor ) for 2 h at 4°C . The dialysate was then mixed with 100 µL sulfopropyl ( SP ) -Sepharose resin ( GE Healthcare ) , washed with 1 ml of lysis buffer , and bound constituents eluted with increasing NaCl concentration ( as described above for the DEAE-Sepharose purification ) . Four μL of partially purified Drosophila RNase P fractions were assayed in a 20-μL reaction containing 10 mM HEPES ( pH 7 . 9 ) , 10 mM magnesium acetate , 200 mM ammonium acetate , 0 . 1% ( v/v ) Nonidet P-40 and 250 nM of in vitro transcribed pre-tRNAGly ( tobacco chloroplast; without 3′-CCA ) , a trace amount of which had been internally labeled with [α-32P]-GTP ( 28 ) . The reactions were incubated at 28°C for 10 min , and then terminated with 10 µL of 2X urea loading dye [8 M urea , 15 mM EDTA , 0 . 025% ( w/v ) xylene cyanol , 0 . 025% ( w/v ) bromophenol blue , 20% ( v/v ) phenol] . The products were separated on an 8% ( w/v ) polyacrylamide ( 19∶1 ) gel containing 8 M urea , and detected using phosphorimaging . Oligo-inhibition assays were performed as described earlier for bacterial and archaeal RNase P [23] , [45] , [46] . For these experiments , the RNA oligo ( final concentration 300 , 400 or 500 µM ) was pre-incubated with 4 µL partially purified RNase P in assay buffer for 5 min at 28°C . After addition of substrate ( pre-tRNAGly ) in assay buffer , the reaction was incubated for 15 min at 28°C , and then terminated and characterized as described above . We performed sequence analysis using genomic data from multiple sources—i5K Pilot Project ( Baylor College of Medicine , Human Genome Sequencing Center ) , NCBI Genome , Ensembl Genomes , VectorBase , Penaeus Genome Database ( PAGE ) , BeeBase , wFleaBase , FlyBase , DOE Joint Genome Institute ( Table S1 ) , using the Infernal package ( release 1 . 1 ) [47] . The secondary structures for newly discovered RPRs were drawn using a ClustalW2-aided multiple sequence alignment of the RNA sequences and Mfold [48] . The results of our search , which yielded 32 new RPRs , were independently validated by the fact that 9 of these 32 RPRs were identified using a different bioinformatics approach [49] . These putative RPRs were manually analyzed for size , the presence of conserved nucleotides ( identity and location ) and the length of P1 helix ( which help define the 5′ and 3′ termini ) . The RPRs we identified diverged in the length and sequence from prototypes and were not identified using previous methods . Incorporating this information into covariance model-based searches will improve future RPR searches . Our results should also serve as a cautionary note for excluding putative RPR genes because they lack pol III promoters and terminators . MRP RNAs were identified and analyzed using an approach similar to that employed for RPRs . The cladograms in Fig . 5 were generated using the NCBI taxonomy browser , Dendroscope [50] , and current literature on arthropod phylogeny [25] , [26] . The VISTA conservation graphs in Fig . 1B and S7A Fig . were generated using mVISTA [51] . For analysis of RPR expression during D . melanogaster development and in various tissues , data derived from the analysis of total RNA by tiling arrays were examined [52] . These data were obtained from the modENCODE consortium [53] as wig files and viewed using the Integrative Genomics Viewer [54] ( S2 Table ) . For D . virilis and D . pseudoobscura , data from polyA-selected samples analyzed by RNA-seq were used ( sam files ) . The SAMtools program was used to index and sort the RNA-seq reads [55] and the Integrative Genomics Viewer was used to visualize the reads . In both the total RNA and polyA-selected samples , reads corresponding to the RPR-containing intron are higher than those corresponding to the preceding intron in ATPsynC . The presence of RPR in the polyA+ sample may have resulted from incomplete removal of highly expressed RNAs , as has been observed in RNA-seq analyses for other non-polyA+ RNAs [56] . The reads for the polyA+ samples are as follows; D . virilis: RPR , 2435 reads/353 bp and preceding intron , 978 reads/683 bp; D . pseudoobscura , RPR 2478 reads/322bp and preceding intron 1980 reads/804 bp . For T . castaneum , RNA-seq data as raw reads ( SRR1048514 and SRR1161702 in SRA format; refer to Table S1 for details ) were downloaded from NCBI GEO , and ERR161589 as FASTA was downloaded from EMBL ENA . The SRR files were converted to FASTQ format using SRAtoolkit [57] . These reads were mapped to T . castaneum genome ( Genome version Tcas3 , from EMBL indexed using Bowtie2 [58] ) and TopHat [59] . The mapped reads were then analyzed using Cufflinks [60] with annotated transcripts ( Tcas3 . 22 . gtf ) . The locus of T . castaneum recipient and housekeeping genes were identified in the version of the genome mentioned above , using TBLASTN ( NCBI-BLAST+ ) ( refer to Table S3 for loci ) [61] . SAMtools was also used to extract genomic sequences flanking the identified RPRs and these sequences were aligned using ClustalW2 [62] . For each species , a consensus proximal sequence element ( PSE ) and TATA box was generated using an alignment of the U6 and 7SK RNA promoters . This consensus was used to search for candidate pol III promoter elements proximal to a given newly identified RPR sequence . | The processing of the 5′ end of nascent tRNAs is catalyzed by ribonuclease P ( RNase P ) , an essential enzyme . In the ribonucleoprotein form of this enzyme , the RNase P RNA ( RPR ) functions as a ribozyme aided by protein cofactors . All previously examined eukaryotic RPR genes are transcribed from their own promoters by RNA pol III . In contrast , the Drosophila RPR gene is embedded in an intron of a recipient gene . We have shown that the embedded sequence , the only copy of RPR in the genome , is transcribed by pol II from the promoter of its recipient gene and encodes the functional RPR . Analysis of other animal genomes revealed that an embedded RPR is also present in the genomes of other insects and crustaceans . This feature provides evidence that the mode of transcription of RPR changed as the result of insertion into a recipient gene approximately 500 million years ago . This new , inserted type of RPR must first have appeared in the arthropod lineage in a common ancestor of insects and crustaceans . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biochemistry",
"rna",
"enzymes",
"biology",
"and",
"life",
"sciences",
"molecular",
"evolution",
"enzymology",
"evolutionary",
"biology",
"ribozymes"
] | 2015 | Transcriptional Control of an Essential Ribozyme in Drosophila Reveals an Ancient Evolutionary Divide in Animals |
Early diagnosis of dengue can assist patient triage and management and prevent unnecessary treatments and interventions . Commercially available assays that detect the dengue virus protein NS1 in the plasma/serum of patients offers the possibility of early and rapid diagnosis . The sensitivity and specificity of the Pan-E Dengue Early ELISA and the Platelia™ Dengue NS1 Ag assays were compared against a reference diagnosis in 1385 patients in 6 countries in Asia and the Americas . Platelia was more sensitive ( 66% ) than Pan-E ( 52% ) in confirmed dengue cases . Sensitivity varied by geographic region , with both assays generally being more sensitive in patients from SE Asia than the Americas . Both kits were more sensitive for specimens collected within the first few days of illness onset relative to later time points . Pan-E and Platelia were both 100% specific in febrile patients without evidence of acute dengue . In patients with other confirmed diagnoses and healthy blood donors , Platelia was more specific ( 100% ) than Pan-E ( 90% ) . For Platelia , when either the NS1 test or the IgM test on the acute sample was positive , the sensitivity versus the reference result was 82% in samples collected in the first four days of fever . NS1 sensitivity was not associated to disease severity ( DF or DHF ) in the Platelia test , whereas a trend for higher sensitivity in DHF cases was seen in the Pan-E test ( however combined with lower overall sensitivity ) . Collectively , this multi-country study suggests that the best performing NS1 assay ( Platelia ) had moderate sensitivity ( median 64% , range 34–76% ) and high specificity ( 100% ) for the diagnosis of dengue . The poor sensitivity of the evaluated assays in some geographical regions suggests further assessments are needed . The combination of NS1 and IgM detection in samples collected in the first few days of fever increased the overall dengue diagnostic sensitivity .
Dengue is the most important mosquito-borne viral disease of humans and an enormous public health burden in affected countries . An estimated 50–100 million dengue cases occur annually , including 250 , 000–500 , 000 cases of severe illness and around 25 , 000 deaths . Approximately 2 . 5 billion people live in dengue endemic countries and the illness is reported in Southeast Asia , Western Pacific , the Americas , Africa and Mediterranean regions [1]–[3] . Dengue viruses ( DENVs ) , of which there are four serotypes , cause a variable spectrum of disease that ranges from an undifferentiated fever to dengue fever ( DF ) through to more severe syndromes called dengue haemorrhagic fever ( DHF ) and dengue shock syndrome ( DSS ) . DHF/DSS is a vasculopathy characterized by capillary leakage and haematological dysregulation . There are no licensed vaccines or specific antiviral therapies for dengue , and patient management relies on good supportive care . Early , sensitive and specific diagnosis of dengue can assist in patient triage and for those who require it , early supportive management . In principle , early diagnosis could also facilitate timely public health interventions , e . g . vector control targeted at the households of index cases . Existing approaches to dengue diagnosis rely primarily on detection of DENV-reactive IgM; in more specialised settings this is augmented with detection of DENV RNA using home made RT-PCR or rarely , virus isolation [4] , [5] . Whilst generally robust , a limitation of IgM-based diagnostic approaches is poor sensitivity in the first few days of illness and in some settings , serological cross-reactivity with other Flaviviruses [4] , [5] . Recently , the diagnostic accuracy of commercial diagnostic assays that detect the DENV NS1 protein in plasma/serum samples have been described [6]–[13] . NS1 is a 55kDa glycoprotein secreted by DENV infected cells “in vitro” and “in vivo” . Whilst the role of NS1 in DENV biology is not well understood , high plasma NS1 concentrations early in illness have been associated with more severe disease [14] , [15] . The targeting of NS1 in diagnostic assays potentially offers the opportunity for an early , specific diagnosis of DENV infection since it can be detected prior to the appearance of measurable DENV-reactive IgM [8] . Whilst NS1 is a promising diagnostic target , the assessment of currently available NS1 assays across a breadth of patient populations , viral serotypes and lineages is important in evaluating where and when these assays [16] may fit into the laboratory diagnosis of dengue . At the end of 2006 , the Dengue Scientific Working Group under the leadership of the World Health Organization Special Programme for Research and Training in Tropical Diseases ( WHO/TDR ) established priorities for dengue research aimed at improving dengue treatment , prevention and control . The evaluation of new diagnostics were included among these priorities [17] , [18] . To this end , the purpose of the current study was to assess the sensitivity and specificity of two commercial NS1 assays in six countries .
The DENCO project was a multi-centre prospective observational study of dengue in Southeast Asia ( Malaysia , Thailand , The Philippines and Vietnam ) and the Americas ( Nicaragua and Venezuela ) . The study sites at which patients were enrolled were: Department of Paediatrics , Faculty of Medicine , University of Malaya , Kuala Lumpur , Malaysia; Queen Sirikit National Institute of Child Health , Bangkok , Thailand; San Lazaro Hospital , Manila , The Philippines; Hospital for Tropical Disease , Ho Chi Minh City , Viet Nam , Children's Hospital #1 , Ho Chi Minh City , Viet Nam; Children's Hospital #2 , Ho Chi Minh City , Viet Nam; Children's Hospital Manuel Jesus de Rivera , Managua , Nicaragua; Research Centre Jose W . Torrealba , University des Andes , Trujillo and Hospital Central , Maracay , Venezuela . Following written informed consent by the study participant , or a parent/guardian in the case of children , patients above 6 months of age with clinically suspected dengue and fever for less than 7 days were enrolled in the study . At 5 centres out-patients were recruited as well as in-patients . Patients were followed daily by trained study physicians using standardised case report forms ( CRFs ) describing clinical , laboratory , diagnostic and management information in detail . Ethical approval was obtained from the Ethics Review Committee of WHO and each institution involved . All patients in these studies were assessed daily by a study physician and had serial haematocrit and platelet estimations performed , as well as appropriate sampling for diagnostic serology and virology . Two plasma or sera samples were collected from each patient , one at day of the enrolment and the second 7–14 days after fever onset . Dengue diagnosis was confirmed by either of the following methods: virus isolation in Aedes albopictus cell line ( C6/36 ) , by RT/PCR detection as previously described and IgM ( MAC-ELISA ) , IgG ( GAC-ELISA or Inhibition ELISA Method , EIM ) and total antibody seroconversion ( by Hemagglutination Inhibition assay ) following the standard procedures at each study site [19]–[30] . The Hemagglutination Inhibition assay was standardized following WHO criteria and WHO recommended cut-off values were utilized [29] . As previously described , RT/PCR methods used here have sensitivity figures from 90 to 100% [20]–[23] . Other investigations and clinical management were at the discretion of the attending physicians . After discharge each patient was classified using the former WHO criteria for DF , DHF and DSS [30] . From November 2007 to January 2008 , we prospectively tested acute plasma ( or serum ) samples from children and adults enrolled in these studies . Between August 2006 and May 2007 a total of 2259 patients were recruited to the DENCO study at the 11 participating hospitals . NS1 detection was attempted using at least one of the two NS1 tests in 1821 patients . From amongst the 1821 patients , there were 1385 with laboratory-confirmed dengue and 45 with no laboratory evidence of acute or recent dengue . A further 391 had either indeterminate laboratory results or suggestive serology; results from these cases were not included in the analysis . The flow-chart in Figure 1 summarises the numbers and geography of enrolment and the classification of patients according to the results of reference diagnostic tests including demographic information . Serological and virological dengue diagnostics were performed in each participating country according to local protocols , with support provided by WHO designated laboratories as necessary ( for participating laboratories see Table 1 ) [19]–[30] . The definitions employed at each site for “confirmed dengue case” are described in Table 2 . For NS1 sensitivity analysis , patients with laboratory confirmation of dengue by serological or virological means were the reference population . For an assessment of NS1 specificity , patients in whom there was no evidence of acute or recent dengue ( defined as serologically and virologically negative and in whom there were a minimum of 2 plasma or serum samples tested with the second collected ≥7 days after fever onset and >2 days after the first sample ) were studied . As an additional assessment of specificity , two sera panels ( one prepared at the Institute of Tropical Medicine “Pedro Kouri” in Cuba and the other at the Mahidol University , Bangkok , Thailand ) from healthy individuals and from non-dengue patients were employed . Pan-E Dengue Early ELISA from Panbio ( Brisbane , Australia ) , ( Kit Pan-E ) and Platelia Dengue NS1 AG from Bio-Rad ( Marnes-la-Coquette , France ) , ( kit Platelia ) were evaluated . Both kits are based on a sandwich format microplate enzyme immunoassay for the detection of DENV NS1 employing a peroxidase-labelled murine monoclonal antibodies as probes . Samples were tested for NS1 detection following the manufacturer's recommendations . Sera were classified as NS1 positive , negative and equivocal according to the manufacturer's instructions . For the purposes of analysis , equivocal samples were excluded from the analysis . Data were double-entered and checked at two established data-entry facilities in Guatemala ( Center for Health Studies , Universidad del Valle de Guatemala ) and Thailand ( WHO/TDR Clinical Data Management Collaborating Center , Faculty of Allied Health Sciences , Thammasat University , Thailand ) and the two datasets were subsequently merged . Data analysis was performed at the Section of Clinical Tropical Medicine at the University of Heidelberg , Germany , using STATA versions 9 . 2 and 10 , ( STATA corporation , College Park , Texas ) .
The diagnostic sensitivity of kits Pan-E and Platelia assays was evaluated in 854 and 1284 serum samples respectively ( Figure 1 ) from patients with a laboratory confirmed dengue diagnosis . Kit Pan-E it could not be performed in all available samples for logistical reasons relating to assay availability at some sites . The sensitivity of the kit Pan-E ranged from 24% in The Philippines to 72% in Vietnam ( overall sensitivity rate of 52% ) . The sensitivity of the kit Platelia ranged from 34% in Nicaragua to 76% in Thailand ( overall sensitivity rate of 66% ) ( Figure 2A ) . Compared to RT/PCR results , sensitivity of kit Pan-E ranged from 29–79% ( overall sensitivity rate of 67%; 95% CI 63–71% ) and the sensitivity of kit Platelia from 36–88% ( overall sensitivity rate of 77%; 95% CI 74–79% ) ( Figure 2B ) . The sensitivity of both kits Pan-E and Platelia was influenced by the patient's duration of illness prior to test sample collection . . In Asian patients , kits Pan-E and Platelia were more sensitive in test samples collected early in the disease phase than at later time points ( Figure 3A ) . The analysis was limited to days with more than 40 observations total which is why for Latin America only a narrow range of days can be shown ( Figure 3B ) and due to small sample size and large confidence intervals no trend is visible . A higher sensitivity of both NS1 detection assays were observed in Asian patients than in Latin-American patients at the first four days of illness ( Figure 3B ) . The sensitivity of each NS1 assay was considered in the context of the infecting serotype . Table 3 shows the sensitivity of kit Pan-E and Platelia assays according to DENV serotype as determined by RT-PCR or virus isolation . In our mainly hospital-based patient samples from 2006/2007 DENV-1 was most prevalent in Asia and DENV-2 most prevalent in Latin America ( Table 4 ) . For each of the four DENV serotypes kit Platelia had a greater sensitivity except for DENV-2 , where the sensitivity was the same in both kits . In kit Platelia , sensitivity for DENV-2 was statistically significantly lower than for the other three serotypes pooled ( DENV-2: 63%; 95% CI 57–69% versus 84%; 95% CI 82–88% for DENV-1 , 3 and 4 ) . The greater prevalence of DENV-2 in Latin American patients compared with Asian patients may help explain the lower sensitivity of both kit Pan-E and Platelia assays in Latin America ( Figure 3B ) . Detection of DENV-reactive IgM by MAC ELISA is the most commonly used approach to making a presumptive diagnosis of acute or recent dengue in endemic countries . Table 5 summarises NS1 sensitivity ( kit Platelia assay only ) in the context of IgM status and day of illness in confirmed dengue patients . The average sensitivity of NS1 testing in the first 7 days of sample collection was 65% ( 95%CI 62–69% ) in acute samples where the IgM result was negative and 66% ( 95%CI 62–70% ) when the acute test sample was IgM positive . Sensitivity figures increased to 74% and 70% if only samples collected in the first four days of illness were considered . Taking an algorithmic approach , when either the NS1 test or the IgM test on the acute sample was positive , the sensitivity for a presumptive ( IgM ) or definitive ( NS1 ) diagnosis versus the reference result was 74% ( 95%CI 69–78 ) in samples collected at days 5 to 7 . These figure increased to 82% ( 95%CI 79–84 ) in samples collected in the first four days of fever . These results suggested a combination of either IgM testing or NS1 testing ( with kit Platelia ) was sufficient to allow a presumptive ( IgM ) or definitive ( NS1 ) diagnosis on an average of 82% of dengue cases enrolled in this study when acute early samples are tested . A similar analysis was performed with data obtained in the evaluation of kit Pan-E . Sensitivity figures of 66% ( 95%CI 60–72 ) in samples collected at days 5 to 7 and 71% in samples collected in the first four days of fever ( 95%CI 67–75 ) were obtained ( Table S1 ) . The sensitivity of each NS1 assay was considered in the context of disease severity and geographical region ( Table 6 ) . Cases were classified according the former WHO criteria for DF and DHF/DSS [27] . Sensitivity of kit Pan-E ranged from 29% ( 95%CI 12–46 ) in DF to 60% ( 95%CI 39–82 ) in DHF cases from Latin-American countries and from 50% ( 95%CI 43–57 ) in DF to 62% ( 95%CI 57–67 ) in DHF/DSS cases from Asia ( overall sensitivity 47% in DF and 62% in DHF/DSS cases ) . The sensitivity of kit Platelia ranged from 41% ( 95%CI 28–55 ) in DF and 68% ( 95%CI 47–89 ) in DHF/DSS cases from Latin-American countries and 70% ( 95%CI 66–75 ) in DF and 68% ( 95%CI 64–72 ) in DHF/DSS cases from Asia ( overall sensitivity of 68% for both DF and DHF/DSS total cases ) . Kit Pan-E showed higher figures of NS1 positive tests in severe cases , which are borderline statistically significant for Asia . Kit Platelia with overall higher sensitivity figures did not show a statistically significant association with disease severity . The diagnostic specificity of kits Pan-E and Platelia assays was evaluated in 36 and 45 samples respectively from patients with no virological or serological laboratory evidence of acute or recent dengue . Both kits were negative in all these samples , which translates into a specificity of 100% . Since the number of patients with no evidence of acute or recent dengue was relatively small ( n = 45 ) in this study , efforts were made to assess the specificity of dengue NS1 assays in patients with other confirmed infectious diseases whose transmission geographically overlaps with dengue , in healthy blood donors , and in blood donors with a serological history of DENV exposure . For the specificity analysis , a total of 304 sera were tested at two study sites ( Cuba and Thailand ) . The specificity of kit Platelia was 100% in both sites whilst the kit Pan-E was 89% ( Table 7 ) . The lower specificity of kit Pan-E was in part due to false positive results in patients with Japanese encephalitis , Yellow Fever and acute Influenza .
Dengue is increasing in incidence globally and therefore accurate and efficient diagnostic tests are more important than ever for clinical care , surveillance support , pathogenesis studies and vaccine research . Diagnosis is also important for case confirmation , to differentiate dengue from other diseases such as leptospirosis , rubella , and other flavivirus infections , and for the clinical management and evaluation of patients with severe disease [16] , [31] . The multicentre study described here assessed the diagnostic accuracy of two commercially available NS1 diagnostic tests . Two main findings were observed here: a ) NS1 detection was overall only modestly sensitive for dengue diagnosis , with sensitivity highest in patients who presented early in their illness and b ) a combined NS1 and IgM detection increased the overall sensitivity of dengue diagnostic . The global dengue research agenda includes evaluating the validity , role and accessibility of available and new diagnostics of importance to reducing disease severity and case fatality [32] . Recognizing the importance of early diagnosis and taking advantage of the platform of the multicentre DENCO project , two commercial available NS1 detection ELISA kits ( Pan-E Early Dengue , Panbio Ltd and Platelia™ Dengue NS1 Ag , Bio-Rad ) , named here as kits Pan-E and Platelia , were evaluated in terms of sensitivity and specificity . Overall and within country sensitivity figures were higher for kit Platelia than kit Pan-E . With the exception of Nicaragua and The Philippines , sensitivity figures of kit Platelia varied from 64% to 76% while the sensitivity of kit Pan-E varied from 36% to 72% . Depending on the diagnostic method used for comparison , different figures of sensitivity of NS1 detection have been reported by others [12] , [33] , [34] . Kumarasamy et al . , obtained an overall sensitivity of 93% using Platelia™Dengue NS1 Ag oscillating from 68% ( in samples where the virus was isolated ) to 90% in paired sera serologically confirmed as dengue [11] , [35] . In the present study , relatively higher levels of sensitivity were observed in samples collected in the first four days of fever when samples from Asian patients were studied ( interpretation limited for Latin America because of small sample size per day of illness ) . . Sensitivity was also higher in Asian patients compared with patients from Nicaragua and Venezuela . The small number of samples from Nicaraguan and Venezuelan patients ( including a lower proportion of DHF/DSS cases ) as well as the serotypes circulating could partially explain these observations ( a high proportion of serotype 2 was found in Nicaraguan samples ) . The influence of duration of illness at the time of sample collection has been highlighted by others [6] , [8] , [10] . Figures of 93–100% sensitivity were obtained in samples collected at days 3 to 5 of fever [8] while others have reported figures higher than 85% in samples from day 1 to 3 in the Platelia assay [6] , [11] . NS1 protein has been detected concomitant with viremia and coincident with the febrile stage [8] . In the present study , the highest sensitivity was obtained in RT-PCR positive samples . Sensitivity of kit Platelia in RT-PCR positive samples was 71% to 88% in Asian countries and 66% in Venezuela , but much lower in Nicaraguan samples ( 36% ) . Samples from this country were retested in a different laboratory by both NS1 detection kits but similar sensitivity results were observed ( data not showed ) . The basis for low sensitivity in Nicaraguan samples remains unclear and will require further studies – but may partly be explained by the high proportion of serotype 2 in Nicaragua , which in both assays was associated with lower sensitivity . Indeed , as 94% ( N = 32 ) of the serotypes recovered from Nicaragua were serotype 2 , we cannot determine an estimate of sensitivity for the remaining 6% ( N = 2 ) . Sensitivity varied by infecting serotype for each kit . The sensitivity of kit Pan-E was highest for DENV-1 infection ( 77% ) and significantly lower for DENV-2 ( 60% ) , DENV-3 ( 57% ) and DENV-4 ( 52% ) . The sensitivity of kit Platelia was also highest for DENV-1 infection ( 83% ) and lowest for DENV-2 ( 60% ) . Consistent with DENV-1 infection being associated with high levels of NS1 detection , Xu et al . , 2006 , reported a sensitivity of 82% in an “in house” ELISA for the detection of NS1 protein of DENV-1 [36] . Similar results for the same serotype were reported by Alcon et al . , 2002 [8] . The basis for different sensitivities for different serotypes requires further investigation . Potentially , this reflects different levels of avidity of the test mAbs for the relevant epitope ( s ) in NS1 from different serotypes , and potentially , different lineages from the same serotype . Also , this could potentially be related to the different sensitivities of the reference RT/PCR methods employed for dengue diagnosis . Alternatively , this might reflect different overall magnitudes of virus burden in patients with different serotypes . A relationship between NS1 detection and viraemia levels has been established previously [12] , [15] . Since high early viraemia levels have also been linked to increased disease severity , it is plausible that NS1 tests are more sensitive in the first few days of illness in patients at risk of developing severe complications in their illness compared to patients with a more benign disease evolution . However , in our study , no association between NS1 detection and disease severity ( indicated by classification of DF or DHF/DSS ) was observed . Furthermore a regression analysis on NS1 positivity for DHF/DSS vs . DF ( or severe vs . mild ) and adjusted for serotype and for country was done and there was no effect seen ( data not shown ) . The specificity of NS1 tests could not accurately be estimated in the DENCO patient population as only a small number of cases had no serological or virological evidence of acute or recent dengue . Nonetheless , in patients who met our criteria for “not dengue” , the specificity of both NS1 test kits was very high ( 100% ) . To provide further insights into specificity , two sera panels from patients with other confirmed diagnosis and healthy individuals were tested . Kit Platelia showed the higher specificity ( 100% ) . Similar specificity values has been previously reported by others [10] , [11] , [15] , [37] . The inclusion into the evaluating panel of samples from patients with acute Yellow fever and Japanese encephalitis virus infections suggest that no cross reaction among flaviviruses is observed with kit Platelia , however a larger number of samples collected from acute flavivirus infected patients need to be studied . The dengue serotype , duration of illness prior to sample collection , and the presence of immunocomplexes ( NS1-IgG ) in previous dengue immune individuals could explain the low sensitivity observed in the Nicaraguan and The Philippines samples [38] . In the case of Nicaragua , DENV-2 was present in the 94% of the samples where the virus was identified by virus isolation and RT/PCR suggesting that this was the predominant serotype . The generally poor sensitivity for DENV-2 ( 60% ) observed for both assays suggests this partially explains the low sensitivity in Nicaraguan samples [12] . In The Philippines , a conjunction of factors such as to the duration of illness prior to sampling and the high level of individuals with a secondary infection could partially explain the low sensitivity since high sensitivity was observed in RT-PCR positive samples ( 83% ) . One of the limitations of our study is that it is heavily biased towards Asian patients and viruses , with 93% of the total samples coming from this region . The strengths of our study were that it was multicentre , prospective and encompassed a broad range of DENV serotypes and clinical presentations . It is important to mention that no proficiency panel study on positive or negative samples was performed prior to evaluating the tested samples allowing us to have more comparable reference methods among participant laboratories . However protocols employed at each site , have been extensively evaluated previously [19]–[30] . In addition , the laboratories participants ( including some WHO collaborating centres ) are the reference centres for dengue diagnosis and laboratory surveillance in their respective countries and have participated in previous regional and international proficiency testing ( [39] , [40] This study confirms and extends the findings of others in relation to the use of NS1 detections assays for the early diagnosis of dengue [6]–[12] . Although we could not study NS1 sensitivity and specificity in primary and secondary cases , in a small subset of samples classified as primary or secondary cases , a higher percentage of diagnose ( 90% over 80 . 6% ) was obtained in the former ( Vazquez S , manuscript in preparation ) . In summary , we found the kit Platelia to be more sensitive and specific than kit Pan-E , with the sensitivity of both assays highest in the first few days of illness . Furthermore , we found that NS1 testing combined with IgM testing on the same test sample could yield a presumptive ( IgM ) or definitive ( NS1 ) diagnose in as many as 82% of confirmed dengue cases using samples collected in the first four days of fever . As IgM detection is widely used for making a presumptive dengue diagnosis and in epidemiological surveillance , the use of a combined diagnostic algorithm including NS1 and IgM detection in samples collected in the first days of fever could provide clinically useful information to assist patient triage , management and outbreak response . | Dengue is the most important mosquito-borne viral disease of humans and an enormous public health burden in affected countries . Early , sensitive and specific diagnosis of dengue is needed for appropriate patient management as well as for early epidemic detection . Commercially available assays that detect the dengue virus protein NS1 in the plasma/serum of patients offer the possibility of early and rapid diagnosis . Here we evaluated two commercially available ELISA kits for NS1 detection ( Pan-E Dengue Early ELISA and the Platelia™ Dengue NS1 Ag ) . Results were compared against a reference diagnosis in 1385 patients in 6 countries in Asia and the Americas . Collectively , this multi-country study suggests that the best performing NS1 assay ( Platelia ) had moderate sensitivity ( median 64% , range 34–76% ) and high specificity ( 100% ) for the diagnosis of dengue . The combination of NS1 and IgM detection in samples collected in the first few days of fever increased the overall dengue diagnostic sensitivity . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"virology/diagnosis",
"infectious",
"diseases/neglected",
"tropical",
"diseases",
"virology/emerging",
"viral",
"diseases",
"infectious",
"diseases/viral",
"infections",
"infectious",
"diseases/tropical",
"and",
"travel-associated",
"diseases"
] | 2010 | Multi-Country Evaluation of the Sensitivity and Specificity of Two Commercially-Available NS1 ELISA Assays for Dengue Diagnosis |
Nucleoproteins ( NPs ) encapsidate the Phlebovirus genomic ( - ) RNA . Upon recombinant expression , NPs tend to form heterogeneous oligomers impeding characterization of the encapsidation process through crystallographic studies . To overcome this problem , we set up a standard protocol in which production under both non-denaturing and denaturing/refolding conditions can be investigated and compared . The protocol was applied for three phlebovirus NPs , allowing an optimized production strategy for each of them . Remarkably , the Rift Valley fever virus NP was purified as a trimer under native conditions and yielded protein crystals whereas the refolded version could be purified as a dimer . Yields of trimeric Toscana virus NP were higher from denaturing than from native condition and lead to crystals . The production of Sandfly Fever Sicilian virus NP failed in both protocols . The comparative protocols described here should help in rationally choosing between denaturing or non-denaturing conditions , which would finally result in the most appropriate and relevant oligomerized protein species . The structure of the Rift Valley fever virus NP has been recently published using a refolded monomeric protein and we believe that the process we devised will contribute to shed light in the genome encapsidation process , a key stage in the viral life cycle .
The genus Phlebovirus belongs to the Bunyaviridae family that includes four others genera , namely Hantavirus , Nairovirus , Orthobunyavirus , and Tospovirus . Phleboviruses have a worldwide distribution and are associated with a wide variety of arthropods ( sandflies , mosquitoes , ticks ) . Rift Valley fever virus ( RVFV ) , the prototype species of the genus , is endemic in Africa , where it is zoonotic , infecting mainly sheep , and causing severe disease with high rates death through abortion in sheep . During these outbreaks , RVFV can pass to human either directly via abortion products or via mosquito transmission , leading occasionally to potentially fatal meningoencephalitis and/or haemorrhagic fevers . Recent outbreaks occurred in the horn of Africa [1]–[2] and the virus also spread into the Arabic peninsula [3] . Outbreaks are directly correlated to rainfalls in these regions and thus , climate and vegetation data may be used to predict areas and periods at risk [4] . In the Mediterranean basin phleboviruses other than RVFV are well established , and seroprevalence can in some regions reach 20% in man [5] . Phleboviruses are mainly represented by Sandfly Fever Sicilian virus ( SFSV ) , Sandfly Fever Naples virus ( SFNV ) , Toscana virus ( TOSV ) and viruses more or less distantly related to SFNV and SFSV [6] , [7] , [8] . These phleboviruses are transmitted by phlebotomine flies in regions where the latter circulate . SFN- and SFS-like viruses can cause mild febrile illnesses ( sometimes paucisymptomatic ) which are likely to be largely underestimated due to the lack of diagnosis , little awareness among health professionals . Toscana virus is in the top three viral etiologies of aseptic meningitis in Italy , Spain and France [9]–[10] . In this respect , phleboviruses are representative of Bunyaviridae . Among the viral world , this family was the major responsible for emerging infectious diseases ( EIDs ) events between 1940 and 2004 ( figure 1 , data extracted from [11] ) , even more represented than the Flaviviridae . Therefore , efforts should be brought not only in diagnosis but also in the understanding of the viral cycle to emphasize antiviral research against these viruses . Phleboviruses , and in general members of the family Bunyaviridae , are enveloped spherical viruses of icosahedral geometry [12]–[13] with an 80–120 nm long diameter . Their genome is made of three segments [14] . Among these segments , two are single negative stranded ( M and L ) whereas the third one ( S ) can adopt an ambisense strategy of expression [15] . The L fragment is encoding the L protein that is carrying the RNA dependent RNA polymerase activity involved in the primary and secondary transcription generating mRNA and replicative intermediates , respectively . In complement , L is also presumably carrying the cap snatching activity required in the viral mRNA capping [16] . The M segment is coding for the glycoprotein precursor that is cleaved by host proteases in two structural domains GC and GN [17] . The S fragment can code for the non structural protein NSs as well as the nucleoprotein NP protein . The nucleoprotein is a 245 amino acids protein ( for Rift Valley fever virus ) that can bind to genomic RNA and replicative intermediates to form ribonucleoproteic complexes ( RNP ) of circular appearance [18]–[19] . The NP oligomerization and its RNA binding properties have not been extensively described until recently . NPs can dimerize with the involvement of their N-terminal domain and this NP-NP interaction does not depend on the presence of RNA [20]–[21] . To date , no physiological multimeric stage over dimerization has been clearly identified . Biophysical and structural studies of the NP alone would therefore provide insights into RNP formation process . For such studies , it would be beneficial to purify homogeneous preparations of monomeric nucleoprotein , or at least , NP assemblies of tractable , defined multimerization status . Many previous studies reported the expression and purification of nucleoproteins for various applications . The NP protein of RVFV can be expressed in insect cells using the recombinant baculovirus technology , but the protein forms a high molecular weight RNP complex , as shown by size exclusion chromatography [22] . The NP proteins have already been produced in E . coli for ELISA experiments . The proteins were purified under denaturing conditions [23] , with a large N-terminal non cleavable tag [24] , or the purification procedure was stopped after the first affinity purification step [25] . Only recently , RVFV NP was purified by refolding the recombinant protein while the natively produced protein was considered as heterogeneous . The refolded protein lead to the first crystal structure determination of the phlebovirus NP [26] . In solution , the refolded protein behaved as a monomer and the NP crystallized as a dimer that was thought to occur naturally . Nevertheless , in this study , the role and the presence of the dimer in the NP oligomeric form observed by electron microscopy remained unclear . In order to further understand the structural properties of phlebovirus NP , it is necessary to set up a process that would lead to the production of protein oligomers pure and homogeneous in size , as previously performed for the rabies NP [27] . Additionally , the process would include a tag removal to improve crystallization . To that aim , and based on the existing results , we decided to evaluate two strategies relying on bacterial recombinant expression for the production of several phlebovirus NP protein suitable for structural studies . These strategies already met success in large scale structural genomics projects . Firstly , the screening of N-terminal tags can drastically improve soluble expression [28]–[29] . Secondly , when proteins are reluctant to soluble expression , they can be expressed as inclusion bodies ( IB ) before being refolded in non-denaturing conditions [30]–[31] . In this study , both strategies will be performed in parallel , even if proteins can be expressed in the soluble fraction , in order to provide comparative data suitable to design optimized production protocols for NP proteins . In addition to the production of NPs suitable for structural studies , these data may highlight trends in the larger field of recombinant protein expression .
cDNA corresponding to the three Nucleoproteins ( NP ) of Rift Valley fever virus ( RVFV , strain Smithburn DQ380157 . 1 ) , Sandfly Fever Sicilian virus ( SFSV , strain J04418 . 1 ) and Toscana virus ( TOSV , strain AR2005 ) were amplified using two Polymerase Chain Reactions ( PCR ) . A first amplification was performed using i ) a forward primer carrying the coding sequence of the Tobacco Etch Virus ( TEV ) protease cleavage site followed by the 21 nucleotides of the NP sequences ( 5′ GAAAACCTGTACTTCCAGGGT-21 nt 3′ ) and ii ) a reverse primer carrying attB2 sequence for cloning by recombination , two stop codons and the 21 nucleotides long reverse complement sequence of the NP ( 5′ GGGGACCACTTTGTACAAGAAAGCTGGGTC TTATTA -21 nt reverse complement 3′ ) . A second PCR was done on the first PCR product using the same reverse primer and a universal forward primer carrying the attB1 sequence as well as two additional nucleotides ( TA ) for the coding frame , and a sequence hybridizing the TEV protease cleavage site ( 5′ GGGGACAAGTTTGTACAAAAAAGCAGGCT TA GAAAACCTGTACTTCCAGGGT 3′ ) . This second PCR products were cloned into the pDonR201 plasmid by recombination ( Gateway , Invitrogen ) . The resulting entry clones were sequenced . The entry clones were then used as templates to clone the NP into two expression plasmids ( see figure 2 ) : pDest17 ( Invitrogen ) , that allows the expression of the NP in fusion with a N-terminal Hexahistidine ( 6His ) tag , removable with the inserted TEV protease cleavage site and pETG20A ( kindly provided by Dr A . Geerlof ) that allows the expression of the NP in fusion with a removable Thioredoxin-Hexahistidine ( TRX-6His ) tag . The resulting expression plasmids were transformed in C41 ( DE3 ) E . coli strain ( Avidis SA ) carrying the pRARE plasmid ( Novagen ) . For each construct , one liter of Terrific Broth ( Athena Enzymes ) containing 100 mg/l of ampicillin and 34 mg/l of chloramphenicol was inoculated with 30 ml of an overnight pre-culture . The bacteria were grown at 37°C up to OD600 nm reached 0 . 8 . Recombinant protein expression was then induced by adding 0 . 5 mM isopropyl β-D-1-thiogalactopyranoside ( IPTG ) and the culture temperature was dropped to 17°C for 16 hours . Cells were harvested by centrifugation at 4 000 g , 10 minutes . Cell pellets were then resuspended in 50 mM Tris buffer , 300 mM NaCl , 10 mM imidazole , 0 . 1% Triton , and 5% glycerol ( pH 8 . 0 ) . Lysozyme ( 0 . 25 mg/ml ) , phenylmethylsulfonyl fluoride ( 1 mM ) , DNase I ( 2 µg/ml ) , and EDTA-free protease cocktail ( Roche ) were added before performing a sonication step . The lysates were centrifuged at 12 000 g for 45 minutes . For both 6His and TRX-6His constructs , the supernatants were collected for the purification procedure in non-denaturing conditions , whereas for the 6His constructs , the pellets were used for the purification process in denaturing conditions , as described in figure 2 . The recombinant soluble proteins were purified from the supernatant previously recovered using the Akta Xpress fast purification liquid chromatography system ( GE Healthcare ) as follows . The first purification step ( immobilized metal affinity chromatography ) was performed on a 5-ml His prep column ( GE Healthcare ) . The clarified bacterial lysates were loaded at 5 ml/min . The columns were then washed with a washing buffer ( 50 mM Tris , 300 mM NaCl and 50 mM imidazole pH 8 . 0 ) and the proteins were eluted with 50 mM Tris , 300 mM NaCl and 500 mM imidazole pH 8 . 0 . The elution fraction ( 15 ml ) was then dialyzed against a buffer compatible with the TEV protease activity ( Hepes 10 mM , NaCl 300 mM , pH 7 . 5 ) . 6His tagged TEV protease mutant selected for optimized expression [32] was added to the protein samples in a 1/20 ( w/w ) ratio and cleavage was performed 16 hours à 4°C . The tag removal efficacy was evaluated on a Coomassie blue stained SDS Page gel by ImageJ software and was calculated on the band surface and intensity of the full lengths TRX-6His fusions: [100- ( IACSAC/IBCSBC ) x100]% where I is the intensity of the band , S is the surface , “AC” is “After Cleavage” and “BC” is “Before Cleavage” . After cleavage , the solution was loaded on a 5-ml His prep column and the cleaved proteins were collected in the flow through , whereas 6His-TEV protease , TRX-6His tag and not cleaved fusions were retained onto the column . The cleaved proteins were further purified by a Size Exclusion Chromatography ( SEC ) in 10 mM Hepes , 300 mM NaCl , pH 7 . 5 using a 16/60 Superdex 75 ( GE Healthcare ) . The SEC was calibrated using the LMW and HMW calibration kits ( GE Healthcare ) in order to convert elution volumes in Molecular Weights . The purity of the samples was checked on Coomassie blue stained SDS Page gels . The fraction of the low molecular weight oligomer of NP ( LMWNP ) amenable to protein crystallization was defined as: ( PALMWNP/PAALLx100 ) % where PA is the “peak area” of the SEC chromatogram at OD280 nm . The identity of the recombinant proteins were confirmed by matrix-assisted laser desorption ionization-time of flight mass ( MALDI-TOF , Bruker Autoflex , Bruker Daltonics , ) spectrometry after trypsin digestion . Finally , the oligomerization was also assessed on Coomassie blue stained SDS Page gels , with an upstream treatment with glutaraldehyde , as previously described [20] . Briefly , different collected fractions were incubated 30 minutes at room temperature with glutaraldehyde at a 0 . 05% concentration . The samples were then denatured 5 minutes at 95°C in SDS Page sample buffer ( 100 mM Tris pH 6 . 8 , 25% glycerol , 10% sodium dodecyl sulfate , 5 mM β-mercaptoethanol , bromophenol blue ) prior electrophoresis on SDS Page gels . The pellets obtained from the centrifuged cell lysates were washed twice to purify the inclusion bodies . The pellets were resuspended in a first washing buffer ( 50 mM Tris , 25 mM Imidazole , 300 mM NaCl , 1 M urea , 0 . 1% Triton , pH 8 ) and centrifuged at 12 000 g during 30 minutes . The supernatant was removed and the new pellets were resuspended in a second washing buffer ( 50 mM Tris , 25 mM Imidazole , 300 mM NaCl , 1 M urea , pH 8 ) . After a second centrifugation step ( 12 000 g , 30 minutes ) , inclusion bodies recovered from the pellets were solubilized in a denaturing buffer ( 50 mM Tris , 300 mM NaCl , 25 mM Imidazole , 8 M Guanidium , pH 8 ) . The purity of the recombinant proteins was assessed on Coomassie blue stained SDS Page gels and their quantity was evaluated at OD280 nm . Denaturing proteins were concentrated up to about 10 mg/ml and diluted 1/20 ( v/v ) in a refolding buffer ( 50 mM Sodium acetate , 100 mM KCl , 10 mM β-mercaptoethanol , pH 4 . 5 ) at 4°C overnight . The refolding volume did not exceed 60 ml ( 30 mg of the protein ) in order to have protein quantities and volumes compatible with the downstream procedure . The observed aggregates after refolding were removed by centrifugation ( 10 000 g , 15 minutes ) and the solution was further clarified using filtration on 0 . 22 µm filters . A preliminary refolding efficacy was calculated as follow: ( 100×[NP]Sol/[NP]BR ) % , where “[NP]Sol” is the NP concentration in the soluble fraction after refolding and [NP]BR is the NP concentration before refolding ( 0 . 5 mg/ml ) . The refolded proteins were then concentrated on Amicon Ultra 10 K ( Milipore ) up to 5 ml before being loaded on a 16/60 Superdex 75 ( GE Healthcare ) equilibrated with 10 mM Hepes , 300 mM NaCl , pH 7 . 5 for SEC purification and homogeneity analysis , as described previously . Crystallization trials were initiated with the RVFV NP purified under non-denaturing conditions and TOSV NP purified from IB using a nano-drop dispenser ( Honeybee; Genomic Solutions ) in 96-well sitting drop plates ( Greiner Bio One ) . Three commercial crystallization kits were tested at 20°C: Structure Screen combination , Stura footprints ( Molecular Dimensions Limited ) , and Nextal SM1 ( Qiagen ) . For each condition , three drops were done: 300 , 200 or 100 nl were added to 100 nl of the crystallization solution .
The ORFs encoding NP proteins of RVFV , SFSV and TOSV were cloned in pDONR201 before being re-introduced in two plasmids for expression as a N-terminal tag fusion , as described in figure 2 . Based on their small size and ability to be purified by Immobilized Metal Affinity Chromatography ( IMAC ) , only 6His ( 3 . 3 kDa ) and TRX-6His ( 14 . 6 kDa ) tags were selected for the tag screening although other tags such as 6His/MBP ( Maltose Binding Protein ) , or 6His/GST ( Glutathion S-Transferase ) could be available and compatible with the cloning procedure[33]–[34] . Nevertheless , the latter tags were not tested because they are much larger than 6His or TRX-6His and might interfere in the oligomerization process by steric hindrance . When fused to a removable 6His tag , the NP proteins can follow two procedures . If the protein is expressed in the soluble fraction , it can be purified under non-denaturing conditions . The 6His tagged fusions can also be expressed as inclusion bodies for subsequent refolding . By contrast , the TRX-6His was used only for soluble expression for two reasons . Firstly , the TRX tag is expected to improve protein solubility . Secondly , if the TRX-6His tag needs to be refolded in a condition that isn't compatible with recombinant protein optimum , and vice versa , the presence of the tag would be deleterious for refolding efficacy . Protein expression was performed in only one culture condition that was chosen on the conclusions of a previous report [35] . Briefly , the two main culture parameters having an impact on soluble expression are the use of rare tRNA co-expressing strains to improve expression yields , and post-induction cultures at low temperatures to promote solubility . E . coli strain carrying a pRARE plasmid ( Novagen ) and bacterial growth at 17°C were thus selected . For each NP fused with an Nterminus 6His tag , a part of the recombinant protein can be expressed in its soluble form , as shown by the amount of protein quantified from the soluble fraction after IMAC purification ( figure 3 , data summarized in table1 ) . Among the 3 NPs , RVFV NP was the only one to be soluble at yields above 1 mg/L culture . This latter yield was arbitrarily defined as the threshold compatible with downstream crystallogenesis experiments . Therefore , among the 6His tagged fused proteins , only RVFV NP was further purified and characterized . When compared to the amount of the insoluble fraction ( data in table 1 ) , the soluble fraction of NPs corresponds to 1 . 5% of the overall expression ( 3 . 5 mg in the soluble fraction compared to 240 mg in the inclusion bodies for RVFV NP ) , or less for SFSV and TOSV NPs . By contrast , when fused to the solubilizing TRX-6His tag , the NPs are 7 to 20-fold more soluble , providing thus enough material for tag cleavage . Interestingly , the solubility trend observed with the 6His fusions ( RVFV>TOSV>SFSV ) is conserved with the TRX-6His tag . The removal of the TRX-6His tag was then assessed based on two criteria . First , the remaining full length TRX-6His-NP can be quantified and compared to the quantities of the full length constructs before cleavage . Using this calculation , the best cleavage efficacy was observed for TOSV NP , for which no residual full length fusion protein was observed after the TEV protease cleavage , as shown in figure 4 . RVFV NP was also efficiently recovered since only 13% of the full length was reluctant to TEV protease cleavage . In contrast with to these two NPs , more than 90% of the full length TRX-6His-SFSV NP remained uncleaved after the incubation with TEV protease , resulting in about 10% cleavage yield . Following the cleavage , the protein solution is loaded again on a Nickel immobilized column . Theoretically , this step binds 6His tagged proteins ( TRX-6His , full length fusions , and TEV protease ) and separates them from the untagged protein ( cleaved protein of interest ) going through the column . Practically , most of the cleaved TOSV NP was found in the flow through fraction whereas the 6His tagged proteins and a small amount of cleaved TOSV NP were trapped on the column and released during elution ( figure 4 ) . The purification of RVFV NP was not as efficient as for TOSV NP . Indeed , cross contaminations of full length and cleaved protein can be found in both flow through and elution pools . However , the cleaved RVFV NP represents about 90% of the flow through and is therefore amenable to SEC . For SFSV NP , the cleaved protein was observed at the elution of the IMAC with the TRX-6His-NP fusion . Moreover , only non cleaved NP went through the nickel immobilized column . It was thus concluded that the purification of SFSV NP failed with the TRX-6His construct and the process was aborted . In summary , among the six fusion proteins tested from the soluble fraction , three constructs yielded amounts and homogeneity criteria , as shown in table 1: 6His-RVFV NP , TRX-6His-RVFV NP and TRX-6His-TOSV NP . 6His-RVFV NP showed a degradation pattern after the IMAC purification , resulting in four sub-products observed from 26 to 30 kDa ( figure 3 ) , whereas the expected size of the NP with the 6His Tag is 31 kDa . The oligomerization of the sub-products was then analyzed by SEC ( figure 5 , panel A ) . Based on the calibration curve , the main part of the 6His-RVFV NP eluted at 94 kDa that could correspond to a trimer of 6His-NP ( theoretical MW: 93 kDa ) . A larger oligomer over 300 kDa was also observed . The oligomerization of the NP is independent to the protein degradation since the four cleaved products are almost equally distributed along the chromatogram ( figure 5 , panel A ) . Therefore , although the criteria in homogeneity and quantity were reached ( see Table 1 ) , crystallization trials were not launched because of protein degradation . SEC was also performed for the NP of both RVFV and TOSV after release of the TRX-6His tag . RVFV NP mainly eluted at 93 kDa ( figure 5 , panel B ) following the same trend as that of 6His-RVFV NP . In order to compare the oligomerization state of the RVFV NP in this protocol to a previous study [20] , several fractions from 300 to 30 kDa were treated with or without glutaraldehyde and analyzed on Coomassie blue stained SDS Page ( figure 6 ) . The cross-linking with glutaraldehyde resulted in three major populations ( monomers ( 1NP ) , dimers ( 2NP ) and timers ( 3NP ) ) as well as to a lower extent tetramers and high molecular complexes on the top of the gel . The amounts of monomers , dimers , trimers and tetramers remained almost unchanged along the chromatogram . By contrast , crosslinked HMW complexes could be observed in the fractions that were collected at low elution volumes . Unlike the corresponding 6His tagged NP , the protein was pure and not degradated as judged by Coomassie Blue stained SDS Page ( Figure 6 ) . Fractions corresponding to the major peak were pooled and the so-called LMW NP gathered 33% of the injected NP . The behavior of TOSV NP after the TRX-6His tag cleavage was different: most of the protein eluted at 193 kDa and the LMW NP corresponds to about 5% , leading to less than 1 mg of protein ( table 1 ) . Nevertheless , the protein eluting at 193 kDa was pooled and concentrated for crystallogenesis experiments . More than 200 mg of each of the 6His-NP constructs were expressed as inclusions bodies ( table 1 ) . Since the recombinant proteins were highly expressed , two standard washes of the inclusion bodies were sufficient to recover NPs that are more than 90% pure , as shown in the total fraction of the refolded NPs ( figure 7 ) . Therefore , neither additional purification step nor optimized washes were needed before refolding . The theoretical isoelectric point ( pI ) for RVFV , SFSV and TOSV are respectively 9 . 8 , 10 . 1 and 9 . 9 . In order to refold 6His-NPs at a pH distant to the protein pI , it was decided to refold at a low pH and sodium acetate at pH 4 . 5 was selected . A first analysis of the protein refolding efficacy was performed by comparing the total amount of 6His-NP to be refolded ( e . g . 30 mg for each NP ) with the soluble and filtrated fraction after overnight dilution in the refolding buffer . Quantitative data ( table 1 ) were obtained by comparing OD280 nm in the total and soluble fractions . Qualitative data ( protein purity , and degradation ) were evaluated using Coomassie blue stained SDS Page ( figure 7 ) . More than 25% of RVFV and TOSV 6His-NP were recovered in the soluble fraction whereas the corresponding SFSV construct was completely insoluble after refolding . The oligomerization states of RVFV 6His-NP after refolding was different to the ones observed for the soluble corresponding NPs . Most of the protein eluted at 134 kDa ( 4 , 3 molecules of 6His-NP ) during the SEC . A minor peak at 56 kDa ( 1 , 8 6His-NP ) was also observed ( figure 5 ) . This latter population corresponds to 16% of the injected protein , leading to less than 1 mg of protein from the 30 mg used for refolding . The chromatogram of the refolded TOSV 6His-NP shows a homogeneous protein with peaks at 97 kDa ( 3 molecules of TOSV 6His-NP ) and 33 kDa ( 1 TOSV 6His-NP ) . The NP eluting at 97 kDa corresponds to 63% of the injected protein ( 4 mg ) . The two protein pools corresponding to the peaks 97 and 33 kDa were separately used for crystallogenesis trials . RVFV NP and TOSV NPs obtained from the non-denaturing production pipeline after TRX-6His removal , as well as the two pools ( 97 kDa and 33 kDa ) of the refolded 6His-TOSV NP were finally concentrated in the SEC buffer up to 6 . 6 mg/ml , 6 . 3 mg/ml , 7 . 4 mg/ml and 3 mg/ml respectively . From the commercial kits , several crystal hits were obtained for RVFV NP ( figure 8 , panel A ) , and one condition lead to sea urchin crystals of the 97 kDa MW TOSV NP ( figure 8 , panel B ) .
The standard pipeline investigating conditions under both denaturing and non-denaturing conditions was proven to be efficient and could be applied to any recombinant protein . With no need of further refinement , two out of the three phlebovirus NPs were produced and purified in suitable amount and quality for crystallogenesis . Trimeric forms of RVFV and TOSV NPs yielded crystals . This result is a starting point for structural studies aiming at the elucidation of the RNA encapsidation mechanism , a targetable step for antiviral research [40] . | Phleboviruses have a worldwide distribution and are usually represented by their prototype Rift Valley fever virus that can have a great impact on health and economy in Africa . The genome of phleboviruses is a segmented negative strand RNA that is encapsidated by the nucleoprotein . The structure of the monomeric nucleoprotein has been recently published but it's not sufficient to decipher a convincing mechanism for the nucleoprotein oligomerization . In order to understand this key step in the virus life cycle , the purification of oligomers homogeneous in size would be a key step to launch structural studies . To that aim , a procedure relying on recombinant protein production in both denaturing and non-denaturing conditions was applied to three phlebovirus nucleoproteins . Although the best production pipeline differs for each protein , pure and homogeneous solutions of Rift Valley fever virus and Toscana virus nucleoproteins were successfully obtained . Both proteins , behaving as apparent trimers in solution , lead to protein crystallization , a starting point to understand the genome encapsidation through structural studies . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"biochemistry",
"molecular",
"biology/rna-protein",
"interactions",
"biophysics/protein",
"folding",
"virology/virion",
"structure,",
"assembly,",
"and",
"egress",
"biotechnology/protein",
"chemistry",
"and",
"proteomics"
] | 2011 | Comparative Production Analysis of Three Phlebovirus Nucleoproteins under Denaturing or Non-Denaturing Conditions for Crystallographic Studies |
Molecular research in cancer is one of the largest areas of bioinformatic investigation , but it remains a challenge to understand biomolecular mechanisms in cancer-related pathways from high-throughput genomic data . This includes the Nuclear-factor-kappa-B ( NFκB ) pathway , which is central to the inflammatory response and cell proliferation in prostate cancer development and progression . Despite close scrutiny and a deep understanding of many of its members’ biomolecular activities , the current list of pathway members and a systems-level understanding of their interactions remains incomplete . Here , we provide the first steps toward computational reconstruction of interaction mechanisms of the NFκB pathway in prostate cancer . We identified novel roles for ATF3 , CXCL2 , DUSP5 , JUNB , NEDD9 , SELE , TRIB1 , and ZFP36 in this pathway , in addition to new mechanistic interactions between these genes and 10 known NFκB pathway members . A newly predicted interaction between NEDD9 and ZFP36 in particular was validated by co-immunoprecipitation , as was NEDD9's potential biological role in prostate cancer cell growth regulation . We combined 651 gene expression datasets with 1 . 4M gene product interactions to predict the inclusion of 40 additional genes in the pathway . Molecular mechanisms of interaction among pathway members were inferred using recent advances in Bayesian data integration to simultaneously provide information specific to biological contexts and individual biomolecular activities , resulting in a total of 112 interactions in the fully reconstructed NFκB pathway: 13 ( 11% ) previously known , 29 ( 26% ) supported by existing literature , and 70 ( 63% ) novel . This method is generalizable to other tissue types , cancers , and organisms , and this new information about the NFκB pathway will allow us to further understand prostate cancer and to develop more effective prevention and treatment strategies .
Proteins in the nuclear-factor-kappa-B ( NFκB ) complex belong to a family of transcription factors ( NFκB1/p105 , NFκB2/p100 , RELA/p65 , RELB , REL/c-REL ) that regulate expression of genes involved in immune and inflammatory responses , cell growth , differentiation , and apoptosis . While these proteins are highly pleiotropic , their activation is context-specific [1] . The activation of NFκB protects against infection and stress , which is regulated by inhibitors of NFκB ( IκB ) proteins that keep NFκB inactive by binding to its protein complex , resulting in the phosphorylation of the IκBs by the IκB kinase ( IKK ) complex . Previous reports have shown NFκB to play an essential role in cancer by regulating the expression of genes involved in cell growth and proliferation , apoptosis , angiogenesis , and metastasis [2–5] . While the biomolecular activities and activation of the NFκB proteins have been studied previously [6 , 7] , the NFκB pathway still remains incomplete . Prostate cancer cells in particular have been reported to have constitutive NFκB activity due to increased activity of the IκB kinase complex , which can lead to cell growth and proliferation , while apoptosis is inhibited in prostate cancer cells [3 , 7–12] . Genome-wide methods , such as GWAS and expression studies , have linked a variety of NFκB-associated pathways to prostate cancer progression , including inflammatory processes ( CXCL12 , IL4 , IL6 , IL6ST , PTGS2 , STAT3 , and TNF ) [13] , cellular differentiation ( LEPR , CRY1 , RNASEL , IL4 , and ARVCF ) [14] , and cell cycle regulation ( FoxM1 , SPP1 ) [15] . Within NFκB itself , p100 and p105 can mediate interaction with NFκB subunits that can also function as IκB proteins , and stimuli including cytokines , TLR signaling , and cellular stress can all activate or contribute to misregulation of the pathway [7] . Along with other inflammatory genes , signaling between NFκB and its regulators during inflammation [6 , 16–18] and cancer [7 , 19] has been the subject of close study , but neither the full repertoire of molecular players nor their mechanisms of interaction have been fully specified . It is now possible to predict detailed , mechanistic interactions and pathway components using large-scale computational data integration [20 , 21] . This entails , for example , combining physical interaction and gene expression data with combinatorial and integrative approaches [22 , 23] . These methods have been previously used to predict a molecular signature of indolent prostate cancer [23] and biomarkers of metastatic breast cancer [22] . However , these efforts failed to take advantage of high-throughput experimental results from biological databases , which represent substantial resources for translational and bioinformatic research in clinical biomarker discovery and computational inference of biomolecular mechanism . In this study we address this challenge and provide the first steps toward computational recovery of mechanistic pathway components specific to the NFκB pathway as perturbed in prostate cancer ( Fig 1 ) . This was done by taking advantage of high-throughput experimental results from heterogeneous databases and training a model for specific biological contexts and specific to the NFκB pathway in prostate cancer . Here , we leveraged recent advances in Bayesian data integration [24] to simultaneously provide information specific to biological contexts and individual biomolecular mechanisms and applied this method to predict a novel NFκB pathway during its activity in cell death , inflammation , adhesion and differentiation as perturbed in prostate cancer . We integrated 651 gene expression datasets and 1 . 4M gene interactions in a context-specific manner using prior knowledge from known NFκB pathways . Focusing on genes differentially expressed in lethal prostate cancer versus indolent , we extracted a high-confidence pathway around such genes which are highly functionally related with the NFκB complex to predict a novel NFκB pathway specific to prostate cancer ( Fig 2 ) . Our predicted NFκB pathway suggested 8 novel genes which were found to be highly down-regulated in lethal prostate cancer and highly functionally related to NFκB , namely ATF3 , CXCL2 , DUSP5 , JUNB , NEDD9 , SELE , TRIB1 , and ZFP36 ( Table 1 ) . Notable genes in the predicted pathway included ATF3 , JUNB , KLF6 , NR4A2 , ZFP36 , DUSP5 and NEDD9 , as well as STAT3 and IRF1 as novel upstream regulators , and SELE , CXCL1 and CXCL2 as novel downstream targets of NFκB in prostate cancer . Connected by 112 predicted mechanistic interactions [13 ( 11% ) previously known , 29 ( 26% ) supported by existing literature , and 70 ( 63% ) novel predictions ( S17 Table ) ] , these genes represent a promising and novel NFκB pathway as disturbed in human prostate cancer .
The predicted NFκB pathway specific to prostate cancer consisted of 50 genes connected by 112 biomolecular mechanisms ( Fig 2 ) : of these mechanisms , 13 ( 11% ) were previously known , 29 ( 26% ) were supported by existing literature , and 70 ( 63% ) were novel ( S17 Table ) . In this pathway we show 10 known NFκB pathway genes ( NFκB1 , NFκB2 , REL , RELA , RELB , IκB-α , IκB-ε , IKK-α , IKK-β , and IKK-γ; S16 Table ) , along with 8 novel genes that we found to be significantly down-regulated in lethal versus indolent prostate cancer in publically available databases [25 , 26] and that were highly functionally associated with NFκB in multiple biological contexts ( Tables 1 and S12 ) . Additionally , we recovered genes that previously have been reported to be associated with NFκB or prostate cancer ( TNF-α/TNFAIP3 , STAT3 , MAP3K8 , NR4A2/NR3C4 , BCL2 , and IL18; S16 and S17 Tables ) , as well predicted novel upstream regulator genes ( ATF3 , JUNB , KLF6 , NR4A2 , ZFP36 , DUSP5 and NEDD9 , and IRF1 ) and downstream target genes of NFκB ( SELE , CXCL1 and CXCL2; S16 Table ) which may be involved in the NFκB pathway for development and progression of lethal prostate cancer as detailed below . As a first step , we tested whether the genes newly predicted to this pathway were enriched for pathways , diseases , or biological processes from the Gene Ontology , KEGG , or the Pathway Interaction Database associated with tumorigenesis or prostate cancer [27] ( S13A Table ) . Interestingly , this analysis revealed that many of the highly enriched biological processes and molecular functions from Gene Ontology were related to inflammation and innate immunity ( S13B Table ) : processes in which chemokines and cytokines play an important role [28] , of which CCL20 , CCL3 , CCL8 , CXCL1 , CXCL2 , and CXCL3 occurred in our pathway ( Fig 2 ) . Additionally , we observed strong functional enrichments in the extracellular space , a major component in cancer development and progression [29] ) for AREG [30 , 31] , SELE [32] , LIF [33] , and several chemokines and cytokines [34–38] , as well as high disease enrichment ( NCI Cancer Gene Index ) in 47 different cancer associations from all major tissues ( S13B Table ) , indicating that these genes are not only involved in prostate cancer , but also in a variety of other cancer types . The NFκB complex consists of 5 proteins [NFκB1 ( p105 ) , NFκB2 ( p100 ) , RELA ( p65 ) , RELB and REL ( c-Rel ) ] and , upon activation , provides a powerful defense mechanism against infection and stress; regulation of the complex in managed in part by families of NFκB inhibitor genes ( IκB ) and kinases ( IKK ) [4 , 7] . Here , our predictions suggested that the IκB genes [NFκBIA ( IκB-α ) and NFκBIE ( IκB-ε ) ] directly bind to NFκB and regulate ( inhibit ) NFκB upstream to maintain an inactive state , while IKK kinases ( CHUK/IKK-α , IκBKB/IKK-β , and IκBKG/IKK-γ ) phosphorylate NFκB for downstream activation , which is in line with previous reports [7 , 39] . We not only recovered these NFκB complex genes , their inhibitors , and their correct biomolecular mechanisms , but also identified 8 additional genes as significantly down-regulated in lethal prostate cancer and highly functionally associated with NFκB in multiple biological contexts ( Table 1 ) . 40 additional novel genes were suggested to constitute a novel NFκB pathway in prostate cancer . Within this pathway , we predicted 112 interactions’ biomolecular mechanisms , out of which we could verify 29 ( 26% ) based on previous studies , while 70 ( 63% ) were novel . Along these 70 novel interactions , 18 gene pairs were reported in other literature as co-regulated without an explicit mechanism of interaction ( S17 Table ) . In particular , our results predicted BCL2 and several inflammatory chemokines to be novel downstream targets of NFκB , including the anti-apoptotic protein BCL2A1 , the chemokine ( C-X-C motif ) ligand 1 ( CXCL1 ) , and ( C-C motif ) ligand 8 ( CCL8 ) . This conclusion is based on a predicted direct binding and downstream regulation of BCL2A1 by the NFκB complex ( in particular the REL , RELB , and NFκBIE subunits ) , a predicted downstream regulation of CCL8 by NFκB2 , and a direct binding of CXCL1 with NFκB2 and RELB ( Fig 2 ) . This is in line with previous findings that BCL2 expression is dependent upon REL and RELA [40] to promote resistance to programmed cell death and important pro-survival functions [7 , 41] , while BCL2L1 ( BCL-XL ) , another anti-apoptotic protein , was observed to be upregulated by NFκB as a critical link between inflammation and cancer [4] and tumor progression [41] . In addition , previous studies showed that IKK-NFκB signaling pathways may lead to downstream upregulation in expression of certain tumor-promoting cytokines and survival genes , including BCL2 and inflammatory chemokines [4 , 42] ( as predicted in this study ) . We were able to further confirm reports that the NEMO-dependent NFκB pathway regulates the expression of many proinflammatory genes , including CCL8 , CXCL2 , CCL2 , SELE , and several interleukins [43] . Specifically , these reports are complementary with our predictions that inflammatory chemokines directly interact with each other ( e . g . CXCL1 and CCL20 , CCL8 and CXCL2 , CXCL2 and CCL3 ) , while CXCL2 was predicted to directly bind or regulate SELE , and IL18R1 was predicted to co-regulate NFκB jointly with CXCL2 in a feedback loop ( Fig 2 ) . In this predicted feedback loop we found that IL18R1 , an interleukin receptor binding to the IL18 gene , regulates REL and RELB and directly binds to NFκB1 , which can be supported by previous findings [44] . In addition to downstream targets of NFκB , we also recovered important upstream regulators for cancer development and progression , including STAT3 , MAP3K8 , and TNF . Our predictions suggested that STAT3 ( signal transducer and activator of transcription ) and the bone morphogenetic protein BMP2 concordantly influence NFκB in prostate cancer by predicted direct interaction , which complements previous studies showing that BMP2 induces apoptosis with modulation of STAT3 [43] . Additionally , STAT3 was predicted to regulate the transcription factor and tumor suppressor gene IRF1 ( interferon regulatory factor 1 ) , as previously confirmed [45] , which was predicted to regulate NFκB ( Fig 2 ) . Based on this prediction we suggest that STAT3 , BMP2 , and IRF1 concordantly regulate NFκB activation upstream in prostate cancer , as it was shown that mechanisms that underline the oncogenic functions of NFκB are likely to require additional transcription factors such as STAT3 , which can function cooperatively with NFκB , and are likely to help to drive NFκB-dependent tumorigenesis [7 , 46] . The oncogene MAP3K8 was correctly predicted as an upstream activator of NFκB and activator of both the MAP kinase and JNK kinase pathways , which leads to an activation of downstream genes such as c-Jun , and JUNB , an AP1 transcription factor and oncogene [47 , 48] . This is based on a predicted phosphorylation of the nuclear receptor NR4A2 gene , a family member of AR ( NR3C4 ) , and regulation of JUNB and DUSP5 , all predicted genes to act as novel upstream regulators of NFκB ( see above ) , suggesting that MAP3K8 would be another important upstream regulator of NFκB in prostate cancer . Another gene that was predicted to directly interact with the JNK kinase pathway , in particular JUNB [49] , was the tumor necrosis factor alpha induced protein TNFAIP3 ( A20 ) , a known inhibitor of NFκB activation [50] . In this case , we did not observe a direct regulation of NFκB by TNFAIP3 , but rather an indirect interaction , as we predicted that TNFAIP3 physically interacts with the chemokine CXCL1 [50–52] , which is a predicted downstream target of NFκB ( see above ) . Our prediction did not reveal a direct upstream nor downstream effect of TNFAIP3 on NFκB , which may be the result of its functional role in negative feedback loops [7] . We predicted several genes to act in prostate cancer as novel upstream regulators of NFκB or novel downstream regulatory targets of NFκB . These included 8 genes differentially expressed in lethal prostate cancer and highly functionally related to NFκB ( Table 1 ) along with additional promising candidates such as CXCL1 , KLF6 , and IRF1 ( S16 Table ) . Among these genes , our predictions highlight ATF3 , JUNB , KLF6 , NR4A2 , ZFP36 , DUSP5 , NEDD9 , STAT3 and IRF1 as novel and promising upstream regulators of NFκB in prostate cancer , while SELE and the chemokines , including CXCL1 and CXCL2 , act as novel downstream targets of NFκB in prostate cancer . In particular , we predicted the nuclear receptor NR4A2 and the activating transcription factor 3 ( ATF3 ) as novel upstream regulators of NFκB in prostate cancer . This was based on the observation that ATF3 was predicted to be highly functionally related to NFκB during regulation of cell cycle and cytokine metabolic process ( Table 1 ) and to indirectly bind to NFκB via NR4A2 after phosphorylation by MAP3K8 ( Fig 2 ) ; notably , previous findings only observed ATF3 as a co-repressor with NFκB in prostate cancer [53] . The tumor suppressor gene and transcription factor KLF6 ( Kruppel-like factor 6 ) is another gene that we predicted to directly regulate not only ATF3 ( as previously suggested [54] ) but also NFκB inhibitor α , and which we therefore suggest as another important upstream regulator of the NFκB cascade in prostate cancer . Another gene predicted to act as an upstream inhibitor of NFκB via NR4A2 was the transcription factor and proto-oncogene JUNB ( Fig 2 ) , which appeared to be highly functionally related to NFκB in multiple biological contexts , including the positive regulation of NFκB transcription factor activity ( Table 1 ) . However , instead of a direct interaction with NFκB as observed in previous studies [55 , 56] , we predicted an indirect inhibition of NFκB via NR4A2 as a mechanism in prostate cancer suppression . These novel predicted genes that regulate NFκB via NR4A2 suggest a key role of this nuclear receptor within the NFκB pathway . NR4A2 is a family member of AR ( NR3C4 ) , which is known to be activated downstream of the MAPK pathway in cancer [57] and directly interacts with NFκB ( specifically the REL subunit ) , as correctly predicted for NR4A2 [58] . Along with these genes ( ATF3 , JUNB , and NR4A2 ) that we predicted to be regulated by the MAP kinase MAP3K8 , a known oncogene involved in prostate cancer growth [48] , we additionally suggest the dual specificity phosphatase 5 ( DUSP5 ) as another upstream regulator of NFκB in prostate cancer: DUSP5 was not only predicted to be highly functionally related to NFκB in the positive regulation of NFκB transcription factor activity ( Table 1 ) , but also correctly predicted to be regulated by MAP3K8 ( Fig 2 ) [59] . We correctly predicted and confirmed the NFκB inhibitors ε ( NFκBIE ) [60] and α ( NFκBIA ) [61 , 62] as upstream regulators of NFκB ( see above ) and also predicted NEDD9 ( neural precursor cell expressed , developmentally down-regulated 9 ) as another upstream regulator of NFκB that acts by directly binding to an NFκB inhibitor , NFκBIA ( Fig 2 ) . Additionally , NEDD9 was predicted to interact directly with the zinc finger protein 36 homolog ( ZFP36 ) , a tumor suppressor gene that negatively regulates NFκB [63 , 64] . This is complementary with our predictions , suggesting that after being regulated by KLF6 , ZFP36 directly binds to NEDD9 ( Fig 2 ) , thus acting as another novel upstream inhibitor of NFκB with a role in the amelioration of prostate cancer . In addition to these novel upstream regulators of NFκB in prostate cancer , we also predicted new downstream targets , including several cytokines and a selectin . Our prediction of the chemokine ( C-X-C motif ) ligand 1 and 2 ( CXCL1 , CXCL2 ) as direct and indirect downstream targets of NFκB in prostate cancer can be supported by previous finding in different contexts [7 , 51 , 65 , 66] . The selectin E gene ( SELE ) was predicted to be downstream regulated by NFκB via such chemokines ( CCL8 , CXCL2 , CXCL3 ) ( Fig 2 ) , while previous studies observed that it activates the PI3K/NFκB pathway in colon cancer [67] . However , SELE is found in cytokine-stimulated endothelial cells and is thought to be responsible for the accumulation of blood leukocytes at sites of inflammation [68] , supporting our confident predicted relationship between SELE and cytokines in this pathway . As CXCL1 and CXCL2 were predicted as downstream targets of NFκB in prostate cancer , we suggest SELE as another important downstream target in this process . Additionally , one of these predicted downstream chemokines , CXCL1 , was predicted to directly bind to the Human Tribbles homolog 1 ( TRIB1 ) , a gene that is reportedly involved in the regulation of NFκB and MAP kinases [69] . This report agrees with our prediction and suggests that TRIB1 could be posttranslationally modified by IKBκB , an NFκB inhibitor , providing an indirect effect of NFκB in prostate cancer . We inferred NFκB pathway components in prostate cancer using information from 860 total datasets . 651 of these were gene expression studies , of which 18 were included specifically due to profiling prostate cancer tissues . We additionally incorporated 225 interaction networks ( protein-protein , regulatory , and genetic interactions ) together comprising 1 . 4M interactions . These data were unified into a predicted set of pathway-specific interactions using a Bayesian framework to model the probability of each dataset providing accurate results relevant to disease pathways in prostate cancer [70 , 71] . This procedure automatically down-weights noisy datasets and those not relevant in a particular context , ultimately providing a single model within which many different types of interaction mechanisms can be captured . One context-specific network [cell death , cell differentiation , cell cycle , cell proliferation , cell migration , and NFκB regulation ( S3 Table ) ] was produced for each interaction mechanism in this study , using the independent subset of data in each case ( see Methods ) . From such predicted functional relationship networks specific for interaction mechanisms , we extracted genes highly confidently related with a set of predefined query genes [NFκB , IκB , and 8 down-regulated genes ( Tables 1 and S12 ) ] , which were integrated into one NFκB pathway as outlined in Fig 1 and illustrated in Fig 2 ( see Methods for addition details ) . To identify interactors within the NFκB pathway in each context , we extracted the subnetworks most confidently associated with the NFκB1 gene , i . e . its nearest neighbors , which resulted in 66 genes in total ( S5 Table ) . Among these genes , most were highly confidently associated with NFκB in multiple contexts ( S5 Table ) . For example , CCL20 , a cytokine regulated by other inflammatory cytokines ( e . g . TNF , INF , or IL-10 ) [72] was highly associated with NFκB in vasculature development , cell migration , positive regulation of NFκB TF activity , and regulation of cell motion , while the transcription factor and proto-oncogene JUNB showed strong association with NFκB in the context of cell cycle and cell motion regulation , as well as positive regulation of NFκB transcription factor activity , which is in line with previous findings [55 , 56] . Next , we analyzed which of these 66 genes showed a significant change in gene expression ( at a significance level of 5% after FDR correction ) between lethal and indolent prostate cancer ( see Methods ) [25 , 26] , which resulted in a set of 8 genes: cyclic AMP-dependent transcription factor ( ATF3 ) , chemokine ( C-X-C motif ) ligand 2 ( CXCL2 ) , dual specificity protein phosphatase 5 ( DUSP5 ) , transcription factor jun-B ( JUNB ) , enhancer of filamentation 1 ( NEDD9 ) , e-selectin ( SELE ) , tribbles homolog 1 ( TRIB1 ) , and zinc finger protein 36 homolog ( ZFP36; Table 1 ) . All of these genes were down-regulated in patients who had disease that relapsed after a prostatectomy , which could be the result of negative feedback loops in lethal prostate cancer that turn off important cancer regulators , such as ZFP36 , DUSP5 , and ATF3 [7] . Surprisingly , none of the NFκB genes were significantly differentially expressed ( NFκB1: FDR = 0 . 69 , NFκB2: FDR = 0 . 19 , REL: 0 . 60 , RELB: FDR = 0 . 71 , RELA: FDR = 0 . 91 ) ( S7 Table ) , which could be a result of their constitutive activation , negative feedback loops , or the presence/absence of cancer regulator genes that determine whether it promotes cancer to develop metastatic disease [7] . The eight genes found to be significantly down-regulated in prostate cancer ( Table 1 ) were further explored in a meta-analysis based on the Gene Expression Atlas ( GXA ) [73 , 74] . This database of meta-analysis is based on summary statistics over a curated subset of ArrayExpress Archive , servicing queries for condition-specific gene expression patterns as well as broader exploratory searches for biologically interesting genes/samples . Additionally , this meta-analysis revealed that a subset of the eight genes ( ATF3 , CXCL2 , JUNB , and ZFP36 ) were significantly up-regulated in normal ( non-disease ) prostate tissue ( S16 Table ) , further supporting their role as high-confidence regulators of NFκB in prostate cancer . As outlined above , we produced one context-specific network for each interaction mechanism based on our interaction ontology ( Fig 4A ) and using the independent subset of data in each case ( see Methods ) . To take the ontology of interaction mechanisms into account , we applied a multi-labeled hierarchical classification formulation enabling us to infer one mechanism-specific network for each interaction type while keeping conserved and non-conserved gene pairs in child-parent relationships in the interaction ontology ( see Methods ) . To identify interactors with NFκB in each interaction mechanism , we extracted the subnetworks most confidently associated with the NFκB1 gene from each mechanism-specific network , which we then integrated into one NFκB pathway ( see Methods and Fig 2 ) . The resulting pathway consisted of 50 genes in total [including all NFκB complex genes and its inhibitors , as well as the 8 significantly upregulated genes as derived above ( Table 1 ) ] connected by 112 non-redundant interactions from 7 different biomolecular mechanisms ( Fig 4A ) . To validate our computational reconstruction of novel interactions , we chose to assay the potential interaction between NEDD9 and ZFP36 by co-immunoprecipitation ( based on the availability of antibodies and their relative expression levels in the cell model system; Fig 5 ) . NEDD9 was confirmed by western blot during immunoprecipitation by anti-ZFP36 antibody ( Fig 5A ) , supporting the association of NEDD9 with the protein complex identified by anti-ZFP36 . NEDD9 was also suggested to play a role specifically in prostate cancer cell proliferation . After successful knockdown of NEDD9 by siRNA in the LAPC4 prostate cancer cell line ( Fig 5B ) , proliferation was significantly inhibited relative to control ( Fig 5C ) . These preliminary validation studies thus support our computational reconstructions and the predicted new roles and interactions of at least these two genes newly characterized in the NFκB pathway in prostate cancer .
In this study we provide the first steps toward computational recovery of mechanistic pathway components specific to the NFκB pathway as perturbed in prostate cancer . We used a Bayesian data integration model to simultaneously provide information specific to biological contexts and individual biomolecular mechanisms for predicting a novel NFκB pathway during its activity in prostate-related biological contexts , including cell death , inflammation , adhesion and differentiation . Our predicted NFκB pathway ( Fig 2 ) revealed 8 genes highly down-regulated in lethal prostate cancer and highly functionally related to NFκB ( Table 1 ) , including novel upstream regulators ( ATF3 , JUNB , KLF6 , NR4A2 , ZFP36 , DUSP5 , NEDD9 , STAT3 and IRF1 ) and novel downstream targets ( SELE , CXCL1 and CXCL2 ) of NFκB in prostate cancer . The identification of disease- and tissue-specific pathways remains a challenging problem—one which we addressed here in the context of a prostate cancer specific NFκB pathway . Historically , automated pathway reconstruction has required extensive expert knowledge and manual curation . Although there exist several pathway collections and databases ( e . g . BioCarta , KEGG [85] , Reactome [86] , NCI Pathway Interaction Database [87] ) , most focus on pathways that are gene-specific ( e . g . the NFκB signaling pathway from BioCarta or Cell Signaling Technology [6 , 16–18] ) rather than disease- or tissue-specific [such as the prostate cancer pathway from KEGG ( hsa05215 ) [88–91]] . The construction of novel pathways from a set of genes or the inclusion of novel genes within existing pathways is often based on literature curation [86 , 87] , predictive computational models [92–94] , or lab experiments [6 , 7 , 18] . In contrast , the association of promising candidate genes with diseases has been widely studied in mutation analyses [95 , 96] and genome-wide-association studies [97 , 98] , but also in predictive models for disease gene prioritization [99–101] and tissue types [102 , 103] . Relatively few studies have predicted novel pathways or networks for specific diseases or tissues [104 , 105] , with a more common trend being reporting disease-specific dysregulation in specific pathways of interest [106–109] . Here , we address this challenge and associate NFκB , which is known to be involved in prostate cancer [7] and other diseases [110–112] , with a novel predicted pathway activated during prostate cancer . Although our validation recovered genes known to be associated with NFκB and prostate cancer ( S8–S11 Tables ) and the majority of our predictions were co-expressed in lethal prostate cancer ( S6 and S7 Figs , S15 Table ) , we did not recover all known genes , e . g . WNT16 or TP53 . The WNT16 gene is known to be regulated by NFκB after DNA damage and subsequently activates the canonical Wnt program in prostate tumor cells [113] . The tumor suppressor gene TP53 regulates NFκB in inflammation and cancer more generally [7 , 114–116] . However , these two genes were not included in our novel NFκB in prostate cancer pathway because their predicted associations with the pathway were not sufficiently confident ( below our predefined threshold of 0 . 9 ) . The low-confidence scores between WNT16 , TP53 and NFκB were based on the integrated datasets , as they showed a lower co-expression in the integrated gene expression datasets after regularization in the Bayesian model than the high-confidence genes from the predicted pathway . Therefore , the quality of our predictions depends mainly on the underlying data ( e . g . integrated data , gold standard , context-specific gene sets ) , which is highly influenced by the disease and tissue of interest . While there will be a large amount of disease- and tissue-specific data available for well-studied diseases ( e . g . prostate and breast cancer ) , there is often less data accessible for diseases in which the relevant tissue is difficult to access ( e . g . brain- and neurodevelopmental disorders ) , diseases which are less intensively studied ( e . g . rare monogenic diseases ) , or complex diseases that involve multiple tissues or phenotypes ( e . g . diabetes and autism ) . Having defined a novel set of biomolecular activities representative of NFκB activation in prostate cancer , two logical next steps would be to 1 ) experimentally validate their molecular mechanisms of action and 2 ) evaluate gene sets derived from our extended pathway as potential clinical biomarkers for prostate cancer risk ( e . g . in conjunction with criteria such as Gleason 6 , low volume PSA < 10 ) . The bulk of newly predicted pathway components represent physical protein-protein interactions: stable co-complexing or transient interactions such as post-translational modifications ( e . g . phosphorylation ) . These should be assayed by extending our preliminary co-immunoprecipitations with additional targets and antibodies , using complex-targeted techniques such as TAP-tagging , and ( when possible ) specifically assessing protein state by phosphoantibody or mass spectrometry targeting . Transcriptional regulatory predictions , especially those downstream of NFκB itself ( e . g . BIRC3 , TICAM1 ) , can be more easily assessed by qPCR readout in the presence of knockdown or other perturbations . Each of these targeted experimental readouts could then be re-incorporated into a refined prediction model to further extend or increase confidence in NFκB pathway components . More importantly , molecular epidemiological data are needed to link these genes’ activities ( post-transcriptionally or post-translationally ) to prostate cancer severity and outcome . Many patients are not destined to progress to higher grade and potentially lethal disease , and they can thus avoid surgery or radiotherapy if their low risk is detected early by molecular or other biomarkers . Purely expression-based biomarkers of low prostate cancer progression risk have yet to be identified; a more detailed mechanistic perspective ( as provided by our predictions ) may lead to targeted transcript assays or to sets of informative gene products ( e . g . phosphoproteins ) . Additionally , more nuanced molecular predictors might identify patients with high risk of micrometastatic disease at time of surgery or radiotherapy , who would then be in need of systemic adjuvant therapy to prevent relapse and death . Data assessing the transcriptional and post-translational states of genes in the extended NFκB pathway along with clinical outcome should therefore be collected . The tightly linked pathway components predicted here thus represent one new step along the route to more effective molecular therapies and diagnostics in prostate cancer .
We integrated high-throughput and heterogeneous functional genomic data ( see below ) using a naïve Bayesian approach with regularization [70 , 71] . Briefly , as implemented in the Sleipnir library , the process first performs a maximum likelihood count to reconstruct the joint probability distribution for each dataset between its discretized data values and the gold standard of known present and absent functional relationships . Regularization was performed by mixing this joint distribution with a uniform distribution using weights proportional to the normalized mutual information shared between the dataset and all other datasets to be integrated; for details , see [70] . This parameter regularization ensures that datasets that contain unique information are upweighted , while datasets that contain common information are downweighted to prevent “overconfidence” due to the naïve Bayes independence assumption . We trained one classifier for each biological context and each interaction mechanism individually , using the corresponding gold standard ( see below ) as the underlying ground truth in the training and testing process . We incorporated 633 baseline microarray expression datasets from the NCBI Gene Expression Omnibus repository ( GEO ) [118] as identified in [70] . These comprised 14 , 617 individual conditions , to which we further added 18 human gene expression datasets identified by ARepA [119] as containing the phrase “prostate tumor” or “prostate cancer” in their metadata annotations ( S1 Table ) . All data acquisition , processing , and normalization were performed using ARepA’s default parameters , specifically 1 ) RMA normalization using the R/affy package [120] , 2 ) co-expression using z-score normalized Pearson correlation [70 , 71] , and 3 ) gene identifier harmonization using BridgeDB [121] . We computed a normalized correlation measure for each gene pair in each dataset to assess a similarity score as co-expression for all gene pairs [70 , 71 , 119] . In addition to gene expression assays , we collected 225 non-microarray datasets from the protein interaction databases BioGRID [122] , IntAct [123] , STRING [124] , Prosite [125] , Domine [126] , Transfac [127] , and ORegAnno [128] , which collectively contained 1 , 351 , 782 pairwise gene interactions ( S6 Table ) derived from 878 datasets . To generate a gold standard specific to the NFκB pathway , we manually chose 30 pathways from the PathwayCommons database [129] ( S2 Table ) that 1 ) contained the NFκB1 gene , 2 ) were non-redundant , and 3 ) contained at most 200 genes . This collection of known NFκB1 pathways was converted into a set of 57 , 533 related ( positive or related ) gene pairs , to which the same quantity of random ( negative or not related ) gene pairs was added to generate both positive and negative gold standards for use in the data integration process described below . We performed two screens for NEDD9 activity and ZFP36 interaction: co-immunoprecipitation and siRNA knockdown in a prostate cancer cell line . LAPC4 cells were received from Dr . Robert Reiter , University of California , Los Angeles . These were maintained in RPMI 1640 at 37°C , 5% CO2 , and 100% relative humidity and supplemented with 10% FBS and 100 IU of penicillin and streptomycin ( 100 μg/ml ) . For NEDD9 / ZFP36 interaction testing , LAPC4 cells were lysed by RIPA buffer and protein concentration was measured by protein BCA assay ( Bio-Rad ) . Cell lysis of 500 μg was applied for each immuno-precipitation for NEDD9 and ZFP36 . Rabbit IgG was included as a control and the results were analyzed by western blot with NEDD9 ( Fisher Scientific ) . For knockdown , we read out cell proliferation as a phenotype using LAPC4 cells cultured until ~80% confluence and then transfected with siRNAs ( Origene ) using lipofectamine 2000 . NEDD9 siRNA probes 1 through 3 were purchased as catalog #SR303132 with sequences CCCAAGAACAAGAGGUAUAUCAGGT , GGCCUUAUAUGACAAUGUCCCAGAG , and CAACAGAAGCUCUAUCAAGUGCCAA , respectively . Knockdown efficiency was detected by western blot at 3 days after transfection . For cell proliferation , cells were split into 96-well plate with a confluence of ~40% after siRNA transfection for 24 hours . The cell proliferation assay was carried out at different days after splitting using the WST-1 assay ( Roche ) with the detection of the absorption at a wavelength of 450 nm ( following manufacture instructions ) . Each experiment was performed in triplicate . | In molecular research in cancer it remains challenging to uncover biomolecular mechanisms in cancer-related pathways from high-throughput genomic data , including the Nuclear-factor-kappa-B ( NFκB ) pathway . Despite close scrutiny and a deep understanding of many of the NFκB pathway members’ biomolecular activities , the current list of pathway members and a systems-level understanding of their interactions remains incomplete . In this study , we provide the first steps toward computational reconstruction of interaction mechanisms of the NFκB pathway in prostate cancer . We identified novel roles for 8 genes in this pathway and new mechanistic interactions between these genes and 10 known pathway members . We combined 651 gene expression datasets with 1 . 4M interactions to predict the inclusion of 40 additional genes in the pathway . Molecular mechanisms of interaction were inferred using recent advances in Bayesian data integration to simultaneously provide information specific to biological contexts and individual biomolecular activities , resulting in 112 interactions in the fully reconstructed NFκB pathway . This method is generalizable , and this new information about the NFκB pathway will allow us to further understand prostate cancer . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"urology",
"medicine",
"and",
"health",
"sciences",
"genetic",
"networks",
"protein",
"interactions",
"gene",
"regulation",
"regulatory",
"proteins",
"protein",
"interaction",
"networks",
"genitourinary",
"tract",
"tumors",
"cancers",
"and",
"neoplasms",
"dna-binding",
... | 2016 | Computational Reconstruction of NFκB Pathway Interaction Mechanisms during Prostate Cancer |
Mycobacterium tuberculosis ( Mtb ) requires the ESX1 specialized protein secretion system for virulence , for triggering cytosolic immune surveillance pathways , and for priming an optimal CD8+ T cell response . This suggests that ESX1 might act primarily by destabilizing the phagosomal membrane that surrounds the bacterium . However , identifying the primary function of the ESX1 system has been difficult because deletion of any substrate inhibits the secretion of all known substrates , thereby abolishing all ESX1 activity . Here we demonstrate that the ESX1 substrate EspA forms a disulfide bonded homodimer after secretion . By disrupting EspA disulfide bond formation , we have dissociated virulence from other known ESX1-mediated activities . Inhibition of EspA disulfide bond formation does not inhibit ESX1 secretion , ESX1-dependent stimulation of the cytosolic pattern receptors in the infected macrophage or the ability of Mtb to prime an adaptive immune response to ESX1 substrates . However , blocking EspA disulfide bond formation severely attenuates the ability of Mtb to survive and cause disease in mice . Strikingly , we show that inhibition of EspA disulfide bond formation also significantly compromises the stability of the mycobacterial cell wall , as does deletion of the ESX1 locus or individual components of the ESX1 system . Thus , we demonstrate that EspA is a major determinant of ESX1-mediated virulence independent of its function in ESX1 secretion . We propose that ESX1 and EspA play central roles in the virulence of Mtb in vivo because they alter the integrity of the mycobacterial cell wall .
Mycobacterium tuberculosis ( Mtb ) is a devastating pathogen that causes epidemic disease and latently infects much of the world's population . However , the molecular details of its pathogenesis are poorly understood . Many lines of evidence underscore the importance of an alternative protein secretion system , ESX1 , to Mtb survival in the macrophage and virulence in animals . The primary attenuating deletion in the vaccine strain , Mycobacterium bovis BCG is the loss of nine genes from the ESX1 locus [1]–[4] . Deletion of the ESX1 locus from virulent Mtb significantly attenuates the bacterium for growth in macrophages and animals [5]–[6] . ESX1 has been implicated in the ability of the bacterium to trigger macrophage production of IFN-β [7]–[8] , activate the inflammasome [9] , modulate macrophage cytokine production and signaling [5] , and escape from the phagolysosome [10]–[11] . The ESX1 substrate proteins are also important targets of the adaptive immune response and are recognized by both CD4+ and CD8+ T cells in a majority of infected individuals [12] . The primary function of ESX1 activity in mediating virulence is unknown , however . There are data demonstrating that ESX1 is required for Mtb to damage the host cell membranes but it is less clear whether this is a direct function of the ESX1 locus . Mtb induces IFN-β production during macrophage infection by activation of the cytosolic pattern receptors [8] , [13] . ESX1 dependent escape from phagolysosomes [10]–[11] , [14] could similarly result from ESX1-mediated membrane damage . This has been hypothesized to be the direct effect of one of the ESX1 substrates , EsxA ( Esat6 ) which has been found to be capable of forming pores in a variety of membrane systems [2] , [15]–[16] . The pore-forming function of EsxA is controversial , however , in part because the ESX1 locus and EsxA are highly conserved in nonpathogenic gram positive organisms that lack obvious pore-forming ability [17]–[18] . In non-pathogenic organisms , ESX1 function has been associated with intrinsic bacterial processes including conjugative DNA transfer [19] and phage susceptibility [20] although the molecular basis for this is unclear . Interestingly , in pathogenic mycobacteria , loss of ESX1 function has also been associated with changes in colony morphology . Both M . bovis BCG and H37Ra , which are spontaneous mutants of virulent mycobacteria that were attenuated through loss of ESX1 function [21]–[23] , were initially isolated from populations of virulent organisms because of changes in their colony morphology . When BCG was complemented with a wildtype copy of the ESX1 locus , the colony morphology reverted to that of virulent Mtb [3] . These data have suggested that ESX1 activity modifies Mtb cell wall composition although the basis for these observations is also unclear . None of the ESX1 substrates has predicted cell wall modifying activity . In addition to EsxA , four other substrates of the ESX1 locus have been reported in Mtb . EsxB ( Cfp10 ) heterodimerizes with EsxA and appears to direct its secretion [24]–[25] . ESX1 secretes a transcriptional regulator , EspR ( Rv3849 ) [26] and two proteins of unknown function , EspA ( Rv3616c ) [27] and EspB ( Rv3881 ) [28]–[29] . Identifying a unique function for any ESX1 substrate has been complicated by the fact that EsxA , EsxB , EspA and EspB require each other for secretion [27]–[28] . Thus , it has not been possible to use a loss-of-function approach to define the distinct activities of the individual substrate proteins . In this study , we designed a novel strategy to determine whether the EspA has an independent role in virulence beyond its role in codependent secretion , using a structure-function approach to examine determinants of EspA's post-secretory activity . We demonstrate that EspA forms disulfide bonded homodimers after secretion and that abrogation of EspA disulfide bond formation does not alter protein secretion , the ability of Mtb to trigger the IFN-β response , or to stimulate robust CD4+ and CD8+ T cell responses . However , blocking EspA disulfide bond formation significantly attenuates the virulence of Mtb in animals and this attenuation correlates with a loss of cell wall integrity . Taken together , these data suggest that ESX1 is required for Mtb to survive and cause disease in animals in part because the full activity of at least one of its substrates , EspA , is required to maintain the structural integrity of the mycobacterial cell wall .
We sought to define unique functions of EspA that are independent of its role in the secretion of other ESX1 substrates . To do this we reasoned that EspA might participate in unique protein-protein interactions after secretion that could be targeted to disrupt EspA's post-secretory function . Indeed , when we analyzed culture filtrates from wildtype Mtb using SDS-PAGE in the absence of reducing agent , secreted EspA predominantly migrated with an apparent molecular mass of 80 kDa though smaller forms were detected ( Figure 1A ) . Upon reduction , these forms of EspA resolved to a single species with an apparent molecular weight of 38 kDa , close to the predicted molecular weight of the monomer . As EspA contains a single cysteine at position 138 , we hypothesized that after secretion , EspA dimerizes either with itself or with another protein via intermolecular disulfide bond formation . Of note , small amounts of the higher molecular weight forms of EspA were also detectable in the cell pellets ( Figure 1A ) . We hypothesize that these represent secreted EspA that remains associated with the mycobacterial cell wall or perhaps is retained in the functional periplasmic space of the bacterium [30] . To identify the proteins that were disulfide-bonded to EspA , we affinity purified complexes associated with a C-terminally-tagged EspA allele , which we previously showed fully complements an EspA deletion mutant [27] . As a negative control we evaluated a strain carrying a deletion of the espA gene complemented with an empty vector in parallel . When affinity-purified proteins were resolved by SDS-PAGE and visualized by Coomassie staining , we identified bands specific to the EspA-his6 expressing strain only at 38kDa and 80 kDa . Western blot analysis indicated that both bands contained EspA ( Figure 1B ) . Using tandem mass spectrometry , we identified only multiple unique peptides from EspA in both bands ( Figure 1B , Table S1 ) . A nonspecific 60 kDa band isolated from both strains was identified as Mtb GroEl1 , a protein that contains a naturally occurring polyhistidine motif [31] and thus , would be expected to copurify . These data suggested that the 80 kDa species represents a homodimer of EspA which is covalently linked via an intermolecular disulfide bond . To further test the model that EspA homodimerizes , we co-expressed EspA tagged with a FLAG epitope and EspA tagged with a Myc epitope in M . smegmatis . When heterologously expressed in M . smegmatis , EspA is found in both the 38 kDa and 80 kDa forms that are observed in M . tuberculosis ( data not shown ) . As predicted , when EspA-Myc was affinity purified with an anti-Myc antibody from bacteria expressing EspA-Myc and EspA-FLAG , both the Myc- and FLAG- tagged forms of the protein were isolated , but EspA-FLAG was not isolated from the control strain which did not express EspA-Myc ( Figure 1C ) . Taken together , these data demonstrate that EspA homodimerizes and that a subset of these homodimers are covalently linked through intermolecular disulfide bond formation . Further analysis of secreted EspA suggested that the intermediate forms of EspA that migrate between the EspA dimer and monomer ( Figure 1A ) represent cleavage products of the EspA dimer ( Figures S1A–C ) . Because disulfide bond formation occurs rarely in the reducing cytosolic environment [32] , we reasoned that disulfide bond formation in the EspA dimer occurs after secretion and could , therefore , be targeted to disrupt EspA function but not interfere with ESX1 secretion . To test this prediction , we mutated the unique cysteine in EspA , at position 138 , to alanine ( espAC138A ) . We find that EspA is significantly more abundant when expressed in the context of the other genes in its operon , espC and espD ( data not shown ) . To test the effect of the espAC138A mutation , we therefore generated an unmarked deletion of espACD and complemented this mutant with the wildtype espACD genes under the control of a tetracycline inducible promoter [33] , a similar construct expressing espAC138ACD or an empty vector as a negative control . We confirmed that when EspA was expressed in RvΔespACD::pACD EspA was secreted and formed the same high molecular weight complexes that are observed in the culture filtrates of wildtype Mtb ( Figure 2A ) . In contrast , EspAC138A was secreted but did not form the SDS-stable dimer . Thus , mutation of the sole cysteine in EspA inhibits disulfide-bonding of the EspA dimer but does not inhibit EspA secretion . To comprehensively determine whether espAC138A alters Mtb protein secretion we used quantitative tandem mass spectrometry to analyze the culture filtrate proteins of RvΔespACD::pVector , RvΔespACD::pACD and RvΔespACD::pAC138ACD . To determine relative protein abundance , we made use of the fact that , using appropriate data acquisition parameters , the number of peptide spectra observed from a given protein directly reflects its overall abundance . Thus , we could estimate the relative abundance of each protein by quantifying the protein's spectral counts [34]–[35] . For robust quantitation , we focused on the 150 most abundant culture filtrate proteins , each of which was quantifiable by 75 or more spectra ( Figure 2B and Table S2 ) . To validate the method , we assessed how the presence or absence of the espACD affected Mtb protein secretion . As previously shown [27] , [36] , we found that optimal EsxA and EsxB secretion requires the presence of the espACD operon ( Figure 2C ) . By proteomic analysis , EsxA and EsxB secretion was ∼20 fold less efficient in the absence of espACD than presence of wildtype genes; however , EsxA and EsxB could still be identified in the culture filtrates of this strain ( Figure 2B ) . By quantitative western blot analysis , we estimated that there was ∼100 fold less EsxA in the culture filtrates of Mtb lacking espACD , consistent with the proteomic data but suggesting that the quantitative dynamic range of the proteomic method is compressed . Interestingly , secreted isoforms of EsxA are found in the culture filtrates of the espACD deletion mutant ( ) although we and others have not found them in culture filtrates from ESX1 mutants lacking core components of the ESX1 apparatus such as the FtsK-like ATPases , Rv3870 and Rv3871 ( A . Garces and T . Ramsdel , unpublished data , and as previously shown in [6] ) , suggesting that optimal EsxA and EsxB secretion requires espACD but that residual EsxA and EsxB secretion occurs in the absence of these genes . We then assessed the effect of inhibiting EspA disulfide bond formation on Mtb protein secretion . Loss of EspA disulfide bond formation did not substantially alter the global protein secretion profile of Mtb ( Figure 2B & Table S2 ) . Importantly , inhibition of EspA disulfide bond formation did not affect EsxA and EsxB secretion , which we confirmed by western blot analysis ( Figure 2C ) . In Mtb expressing espAC138ACD , the proteomic analysis suggested that EspA secretion was intact though somewhat reduced relative to wildtype . EspC was also identified by mass spectrometry in the culture filtrates , as has been predicted by recently published studies in M . marinum [37] , and the total secretion of EspC was not altered in bacteria expressing espAC138ACD . Thus , the proteomic data indicate that inhibition of EspA disulfide bond formation does not globally alter protein secretion in Mtb and is not required for ESX1 secretion of EsxA and EsxB . We hypothesized that inhibition of EspA disulfide bond formation would allow us to specifically identify aspects of ESX1 mediated virulence that require EspA function and dissociate them from those that require ESX1 secretion of EsxA and EsxB . To do this , we assessed the effect of the espAC138A mutation on the virulence of Mtb . In a SCID mouse model of infection , which was chosen in order to assess the virulence of the Mtb strains independent of the effects of adaptive immunity , animals infected with wildtype Mtb succumbed to infection after roughly 35 days ( Figure 3A ) . The espACD deletion mutant was significantly attenuated for virulence; mice infected with this strain survived for an average of 127 days . Wildtype espACD significantly but incompletely complemented the deletion mutant for virulence . It is possible that the failure to fully complement the virulence defect is due to the fact that the espACD genes were ectopically expressed from an episomal vector under the control of an inducible promoter . Unlike the ESX1 locus , the espACD genes have been shown to be under the control of multiple regulators including EspR [26] and PhoPR [23] . Thus , it is not surprising that ectopic expression of the locus via an inducible promoter does not fully recapitulate the appropriate amount and timing of secretion during infection of an animal . However , in comparison to the strain expressing wild type espACD , the strain of Mtb expressing espAC138ACD was significantly attenuated ( Figure 3A ) . Mice infected with the strain expressing espAC138ACD survived 95 days on average , about 35 days longer than mice infected with bacteria expressing the wildtype espACD . We obtained very similar survival times in mice infected in parallel with a three fold dilution of each innocula , demonstrating that the differences in survival reflect marked differences in the virulence of the infecting strains rather than small differences in infecting doses ( Figure 3A ) . We also found that virulence depended critically on C138 of EspA in immunocompetent mice . In both C57Bl/6 and C3H/HeSnJ mice , whose MHC haplotype also allows us to simultaneously measure EsxA and EsxB specific T cell responses as described below , the espACD mutant complemented with espAC138ACD was attenuated to nearly the same extent as the deletion mutant complemented with an empty vector while complementation with wildtype espACD largely restored Mtb growth in lungs and spleen ( Figures 3B–D ) . Loss of ESX1 has also been shown to attenuate Mtb for growth in macrophages [6] . We therefore assessed the ability of Mtb expressing espAC138ACD to survive in murine bone marrow derived macrophages . Like the ESX1 deletion mutant , Mtb lacking espACD or expressing espAC138ACD were attenuated for survival in macrophages ( Figure 3E ) . Thus , we find that inhibition of EspA disulfide bond formation significantly attenuates the virulence of Mtb in animals and in macrophages despite apparently normal secretion of EsxA and EsxB . Strains expressing mutant EspA could be attenuated because they elicit different host responses or host damage . Because EsxA has been postulated to disrupt host cell membranes , we sought to determine whether Mtb expressing espAC138ACD retain the ability of perturb host cell membranes . To test this we took advantage the fact that ESX1 is required for the rapid induction of IFN-β transcription upon M . tuberculosis infection [7] . We have shown that maximal IFN-β expression depends on activation of the NOD2 pathway which is triggered by bacterial peptidoglycan in the host cell cytosol [13] . We therefore assessed the ability of Mtb expressing wildtype or mutant EspA to induce secretion of IFN-β after macrophage infection . As previously shown , wildtype Mtb activates IFN-β expression and secretion in an ESX1 , esxA and espA dependent fashion Figure 4A–B ) while loss of these virulence determinants did not affect induction of TNF-α ( Figure 4C ) . Complementation of the espACD deletion mutant with espAC138ACD restored the ability of the cells to activate IFN-β production to the same extent as complementation with the wildtype genes . Thus , inhibition of EspA disulfide bond formation does not perturb the bacterium's ability to activate the cytosolic pattern receptors . We extended these observations by assessing whether the espAC138A mutant's ability to stimulate the IFN-β response correlated with its ability to prime a CD8+ T cell response . EsxB is an important CD8+ T cell antigen in both mice and humans [38]–[39] . The path by which Mtb antigens reach the class I MHC processing pathway has not been well established . However , we have previously shown that ESX1 secretion is required in order to prime a CD8+ T cell response to EsxB [40] . We reasoned that the ESX1 substrates might strongly induce CD8+ T cell responses because they can gain access to the host cell cytosol and thus are readily sampled by the cytosolic class I MHC processing and presentation pathway . Consequently , we assessed the EsxB-specific CD8+ T cell response elicited by Mtb expressing espAC138ACD . We found a robust CD8+ T cell response to EsxB in the spleens and lungs of animals infected with RvΔespACD::pAC138ACD ( Figures 4D and 4G ) . These findings are consistent with the data showing this strain is also capable of secreting EsxB and inducing IFN-β production . As anticipated from previously published results [4] , the CD4+ T cell response to EsxA is abrogated in the absence of espACD ( Figure 4E ) . However , it is intact in animals infected with espAC138ACD ( Figure 4E ) , providing evidence that EsxA is secreted in vivo as well as in vitro in the absence of EspA disulfide bond formation . T-cells from mice infected with the various mycobacterial mutants produced similar amounts of IFN-γ in response to mycobacterial whole cell lysate , indicating that the global T-cell response to Mtb was not affected by EspA disulfide bond formation ( Figure 4F ) . Spontaneous loss of ESX1 function during the laboratory evolution of both M . bovis BCG and H37Ra was associated with marked changes in colony morphology [21]–[22] . Complementation of BCG with a wildtype copy of the ESX1 genes resulted in colonies that again appeared similar to colonies of virulent Mtb [3] . More recent expression studies have also indicated that the espACD genes are highly transcriptionally regulated by cell wall stress [41]–[43] , suggesting a link between these ESX1 substrates and cell wall structure . Based on these observations , we hypothesized that inhibition of EspA disulfide bond formation might alter the virulence of Mtb because it compromises the integrity of the cell wall . Colony morphology is a subjective measure of cell wall structure and we have found it difficult to reproducibly and quantitatively score for ESX1 associated changes in colony morphology . Therefore , we sought more objective assays to assess cell wall integrity in our mutants . We found no evidence that loss of ESX1 or espACD altered bacterial resistance to reactive oxygen or nitrogen species ( data not shown ) . However , we found that Mtb strains lacking the ESX1 locus , an FtsK-family ATPase in the ESX1 locus ( Rv3871 ) , or espACD were significantly more susceptible than wild type to a direct cell wall stress , SDS treatment ( Figures 5A and 5B ) . Deletion of the ESX1 locus had a quantitatively greater effect on cell wall integrity than loss of espACD . The cell wall defect could be complemented by introduction of the wildtype genes ( Figures 5A and 5B ) . We then assessed whether EspA disulfide bond formation was required for EspA's contribution to the cell wall integrity of Mtb . Strikingly , we found that bacteria expressing espAC138ACD show a similar susceptibility to SDS-induced stress as strains lacking the espACD locus entirely ( Figure 5B ) . We tested other cell wall stressors and found that Mtb lacking ESX1 , espACD or Mtb expressing espAC138ACD were also susceptible to other detergent stresses including n-dodecyl beta-D-maltoside and TritonX-100 ( Figure 5C and Figure S3 ) . Thus , ESX1 activity was required for the functional integrity of the mycobacterial cell wall and this effect requires EspA disulfide bond formation .
The ESX1 secretion system is critically required for the virulence of Mtb yet little is understood about its mechanism of action . One hypothesis is that EsxA is the primary mediator of ESX1-associated virulence , acting as a pore-forming molecule that allows the bacterium access to the host cell cytosol [11] , [16] , [44] . Alternatively , the ESX1 system might function like a type IV secretion system , secreting effector proteins directly into the host cell cytosol [45] . In both of these models , the ESX1-dependent stimulation of cytosolic immune pathways and CD8+ T cell responses has been used as evidence that the ESX1 system targets host cell membranes . These models of ESX1 function do not address the experimental observations that suggest that the ESX1 locus affects the composition of the mycobacterial cell wall . In this work , we have dissociated ESX1 secretion and the effects of the ESX1 apparatus on the innate and adaptive immune systems from ESX1 dependent cell wall effects and virulence . Disruption of EspA disulfide bond formation does not perturb ESX1 secretion or ESX1 dependent interactions between Mtb and the innate and adaptive immune systems . It does , however , alter the functional integrity of the mycobacterial cell wall and dramatically attenuate the bacterium for virulence in vivo . These data suggest that ESX1 is required for Mtb to survive and cause disease in vivo at least in part because of its effects on the cell wall . Perturbation of the cell wall structure may attenuate the organism for growth in vivo because it broadly disrupts the bacterial interface with the host cell , undermining specific virulence functions , or because the organism is more susceptible to host antimicrobial defenses . The most parsimonious explanation for our findings is that EspA acts directly on the mycobacterial cell wall . We and others have shown that EspA is secreted in standard mycobacterial growth media , which includes low concentrations of nonionic detergent . However , we have found that this protein remains associated with the cell wall when Mtb was grown in the absence of detergent ( data not shown ) , in keeping with recently published microscopy data demonstrating that several ESX1 substrates are associated with the mycobacterial capsule in minimally disturbed bacterial cultures [46] . Thus , EspA could reasonably be engaged in modifying the cell wall and perhaps the capsule more specifically . For example , EspA may be directly required for the transport of cell wall components or may regulate the activity of other cell wall acting proteins . Alternatively , EspA could also have indirect effects that alter gene expression , although there is little evidence of ESX1 dependent changes in transcription [47] . We show that Mtb strains lacking ESX1 or EspA function have a marked defect in cell wall integrity as measured by detergent susceptibility . However , we have found that ESX1 function does not affect other measures of cell wall permeability or structure such as susceptibility to the hydrophobic antibiotic , rifampin , the cell wall acting antibiotics , isoniazid and meropenem , or lysozyme ( data not shown ) . Our findings are consistent with studies of other cell wall mutants which have found that different mutants in cell wall biosynthesis have variable defects in permeability and susceptibility assays . In some cases , susceptibility can be easily predicted by gene function . For example , disrupting ponA , which acts on peptidoglycan , causes hypersusceptibility to lysozyme [48] . In many cases , however , the link between the genetic lesion and susceptibility to different cell wall stressors is not obvious [48]–[50] , reflecting our limited insight into cell wall assembly in Mtb . In the case of ESX1 , further biochemical analysis will be required to determine the specific cell wall defect caused by loss of activity . Our model does not exclude the possibility that other ESX1 substrates , such as EsxA , have a direct activity on the macrophage as we find that activation of the host cytosolic surveillance systems occurs independently of EspA disulfide bond formation but requires ESX1 activity . The espACD deletion mutant is less virulent in SCID mice than Mtb lacking EspA disulfide bond formation , suggesting that isolated EsxA secretion may make an independent contribution to virulence in animals . However , EsxA , like the rest of the ESX1 locus , is highly conserved in both pathogenic and nonpathogenic gram positive bacteria [18] , suggesting that this protein has an important biologic function in the bacterium that is a prerequisite for the virulence of Mtb but that it does not directly mediate virulence . Indeed , the data presented here suggest that the primary target of the ESX1 system is the bacterial cell wall .
Mtb and M . smegmatis strains were maintained as previously published [27] , [51] . The EsxA deletion mutant and Rv3871 transposon mutant have been previously described [6] . For analysis of protein expression and secretion , bacteria from cultures normalized to the same growth phase were washed and resuspended in designated medium at an O . D . ∼0 . 3 for 72 hours at 37°C . Where indicated , bacteria were cultured in N salt media ( 100 mM Bis/Tris HCl , 5 mM KCL , 7 . 5 mM ( NH4 ) 2SO4 , 0 . 5 mM K2HSO4 , 1 mM KH2PO4 , 10 mM MgCl2 , 38 mM glycerol , pH 7 . 0 ) . N salt media is a minimal medium that allows titration of the divalent cation concentration which we have historically used when collecting samples for proteomic analysis [27] . Cell pellets and culture filtrates were collected and processed as described previously [27] except that culture filtrates were concentrated by precipitation with 10% trichloroacetic acid unless otherwise noted . The genes encoding espACD were deleted from wildtype H37Rv through homologous recombination using a suicide vector approach . Deletion was confirmed by PCR analysis . As described in detail in Text S1 , espACD was amplified from H37Rv genomic DNA and the espAC138A mutation was introduced via PCR mutagenesis and cloning of an internal gene fragment . The PCR products were recombined into a Gateway donor vector ( Invitrogen , Carlsbad , CA ) and transferred to an episomal expression vector ( pTET ) that was constructed to express genes under the control of a tetracycline inducible mycobacterial promoter [33] . All constructs were confirmed by sequencing . The constructs or an empty vector were transformed into RvΔespACD to generate the designated strains . Rv3871 was amplified ( Forward primer: GGCTAAGAAGGAGATATACATATGACTGCTGAACCGGAAGTA; Reverse primer: CTTGTCGTCGTCGTCCTTGTAGTCACCGGCGCTTGGGGGTGC ) and the PCR product was similarly recombined into pTET . This construct or an empty vector was transformed into the Rv3871 transposon insertion mutant . Gene expression was induced from the tetracycline inducible promoter with 100 ng/ml of anhydrotetracycline ( AT ) ( Spectrum Chemicals , Gardena , CA ) for 24 hours prior to beginning culture filtrate collections . Samples were analyzed via SDS-PAGE and western blotting as previously published [27] . Where noted , samples were reduced with 10 mM dithiothreitol ( DTT ) for 30 minutes at 37°C prior to gel electrophoresis . Antibodies to EsxA , EsxB and EspA were published previously [27] . The antibody to poly-histidine ( #NB600-318 ) , which was used to detect GroEl1 , was obtained from Novus Biologicals ( Littleton , CO ) as were antibodies to the Myc- and FLAG- epitopes . Antibodies were used according to the manufacturer's directions . In addition , where indicated , relevant gel slices were excised and analyzed by tandem mass spectrometry ( MS/MS ) as using published methods [52]–[54] and as described in Text S1 . Affinity purification of EspA ( his6 ) from RvΔespA::pEspA ( 6his ) and EspA-FLAG and EspA-myc from M . smegmatis were performed as described in Text S1 . BALB/c-SCID , C57Bl/6 and C3H/HeSnJ mice were purchased from Jackson Laboratory ( Bar Harbor , ME ) . 24 hours prior to infection , mycobacterial strains were cultured overnight in media containing 100 ng/ml AT and mice were started on chow containing 2000 ppm tetracycline ( Research Diets , New Brunswick , NJ ) . Mice were maintained on tet-chow through the course of the experiment . Mice were infected by intravenous tail vein injection and doses were confirmed by plating the innocula . At the indicated times , 4 mice/group were sacrificed and bacterial organ burden was determined by plating for CFU . Organs from C57Bl/6 mice were plated on medium in the presence and absence of hygromycin to assess for loss of the episomal plasmid over the course of the experiment . No significant vector loss was detected . Mice with organ burdens that differed by more than 5 fold from other animals in the group were considered missed injections and these data were discarded . CD8+ and CD4+ T cell responses were assessed as previously published [40] . In order to ensure that the infected mice had equivalent bacterial burdens at the time of T cell analysis , the infecting doses of RvΔespACD::pVector and RvΔespACD::pAC138ACD were ten fold higher than that of RvΔespACD::pACD or H37Rv . Bacterial strains were prepared and induced as described for murine infections . Murine bone marrow derived macrophages were prepared from C57Bl/6 mice according to previously published protocols [55] . After 7 days of culture , differentiated macrophages were frozen for future use . For infections , bone marrow derived macrophages were thawed and plated at a density of 2 . 5×104 cells per well of a 96 well tissue culture treated plate and allowed to adhere overnight . Monolayers were washed , and infected with the indicated strains at an MOI of 10 to produce a final infection of roughly 1 bacterium/macrophage . Bacteria were spun onto the macrophage monolayer and infection was allowed to proceed for 3 hours . Monolayers were washed three times and fresh medium was added containing 100 ng/ml AT . At the indicated times after infection , monolayers were lysed with PBS-0 . 1%TritonX-100 and bacteria in the well were enumerated by plating serial dilutions . For cytokine assays , RAW-264 . 7 macrophages were infected with the indicated strains at an MOI of 1 bacterium/macrophage as previously described [56] . At the indicated times , culture filtrates were removed and IFN-β secretion was assessed by ELISA for IFN-β ( R&D Systems , Minneapolis , MN ) . In addition , RNA was isolated from infected macrophages as previously described [56] . 2 µg of RNA was transcribed into DNA using random hexamers with Superscript III reverse transcriptase ( Invitrogen , Carlsbad , CA ) . Quantitative PCR assays were performed with TaqMan Gene Expression IFN-β , TNF-α and GAPDH assays ( Applied Biosystems , Foster City , CA ) . For these assays , standard curves were generated using serial dilutions of pooled cDNA from macrophages 4 hours after LPS stimulation . Mtb strains were grown to early log phase ( ∼0 . 2 O . D . at 600nm ) in Sauton's medium supplemented with 0 . 05% Tween-80 . Strains complemented with tetracycline inducible constructs , pEmpty , pACD , pAC138ACD and pRv3871 , were cultured overnight in Sauton's containing 100 ng/ml AT . Cells were pelleted , washed once and resuspended at a density of 1 . 2×108 cells/ml in 7H9 medium containing the indicated concentrations of SDS , DDM ( n-dodecyl β-D-maltoside ) and Triton-X-100 . DDM and Triton-X-100 were purchased from Sigma-Aldrich ( St . Louis , MO ) . In studies of strains expressing the tetracycline inducible constructs the medium also contained AT at 100 ng/ml . Bacteria were incubated in SDS for 6h at 37°C with shaking , washed twice and then plated for CFU on 7H10 agar plates containing 10% OADC . For susceptibility to DDM and Triton-X-100 , bacteria were incubated overnight with detergent on a shaker at 37°C , washed twice and resuspended in 7H9 media supplemented with 10% OADC and 0 . 05% Tween-80 . Cultures were serially diluted ( 10-fold ) onto 96 wells plate and their viability determined by adding 20 µl of 10× Alamar Blue dye ( AbD Serotec , Raleigh , NC ) . After incubating at 37°C for 2 days , cells were fixed for 1 h with 2% paraformaldehyde and absorbance measured at 570 and 600 nm on a Versamax microplate reader using Softmax Pro version 5 . 3 ( Molecular Devices , CA ) . Proteomics data analysis was performed as described in Text S1 according to published methods . Statistical analyses and graphing were otherwise performed with GraphPad Prism . All animal experimentation was conducted following the National Institutes of Health guidelines for housing and care of laboratory animals and performed in accordance with Institutional regulations after protocol review and approval by the Harvard Medical Area Standing Committee on Animals . | From studies of BCG , the tuberculosis vaccine , we know that Mycobacterium tuberculosis requires a specialized protein secretion system , ESX1 , to cause disease in people . ESX1 is required for Mtb to co-opt the host cells in which the bacterium resides and it is thought that this explains its central role in virulence . However , other data suggests that ESX1 serves an important role in the bacterium itself , altering the organism's cell wall . It has been difficult to determine the relative significance of these ESX1-associated functions , however , because deletion of any piece of the apparatus completely abolishes all ESX1 activities . Here we use a simple approach to pinpoint the functionally significant target of one of the proteins secreted by ESX1 , EspA . We mutate EspA such that the ESX1 system still secretes its substrates but the bacterium no longer causes disease . The attenuated EspA mutant has defects in its cell wall but not in its interactions with host cells in vitro . We propose that the ESX1 system and the proteins it secretes are important for Mtb to survive and cause disease in people because they act to ensure the integrity of the bacterial cell wall . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"infectious",
"diseases/bacterial",
"infections",
"microbiology"
] | 2010 | EspA Acts as a Critical Mediator of ESX1-Dependent Virulence in Mycobacterium tuberculosis by Affecting Bacterial Cell Wall Integrity |
Human Cytomegalovirus ( HCMV ) is a widespread pathogen , infection with which can cause severe disease for immunocompromised individuals . The complex changes wrought on the host’s immune system during both productive and latent HCMV infection are well known . Infected cells are masked and manipulated and uninfected immune cells are also affected; peripheral blood mononuclear cell ( PBMC ) proliferation is reduced and cytokine profiles altered . Levels increase of the anti-inflammatory cytokine IL-10 , which may be important for the establishment of HCMV infections and is required for the development of high viral titres by murine cytomegalovirus . The mechanisms by which HCMV affects T cell IL-10 secretion are not understood . We show here that treatment of PBMC with purified pUL11 induces IL-10 producing T cells as a result of pUL11 binding to the CD45 phosphatase on T cells . IL-10 production induced by HCMV infection is also in part mediated by pUL11 . Supernatants from pUL11 treated cells have anti-inflammatory effects on untreated PBMC . Considering the mechanism , CD45 can be a positive or negative regulator of TCR signalling , depending on its expression level , and we show that pUL11 also has concentration dependent activating or inhibitory effects on T cell proliferation and on the kinase function of the CD45 substrate Lck . pUL11 is therefore the first example of a viral protein that can target CD45 to induce T cells with anti-inflammatory properties . It is also the first HCMV protein shown to induce T cell IL-10 secretion . Understanding the mechanisms by which pUL11-induced changes in signal strength influence T cell development and function may provide the basis for the development of novel antiviral treatments and therapies against immune pathologies .
Human Cytomegalovirus ( HCMV ) is a ubiquitous human pathogen with a high seroprevalence of between 45 and 100% worldwide [1] . While mostly asymptomatic in healthy individuals , infection with HCMV in immunocompromised individuals can cause severe disease or death . Congenital HCMV infection , for example , results in permanent disability in an estimated number of approximately 8 , 000 children per year in the US and 1 , 100 in France [2] . HCMV primary infection or reactivation from latency is also a major problem for both stem cell and solid organ transplant recipients , as it can cause clinical disease and also have indirect effects on survival , including increasing the likelihood of the occurrence of secondary bacterial , fungal and viral infections due to CMV-mediated myelosuppression [3–5] . The intricate and complex changes wrought on its host’s immune system during all stages of HCMV infection are well documented [6] . Infected cells are masked from recognition and have their functions manipulated to the benefit of the virus . Uninfected immune cells are also affected; a generalized myelosuppression has been described , the ability of peripheral blood mononuclear cells ( PBMC ) to proliferate in response to stimuli is reduced and the cytokine profile is altered [5 , 7–9] . The mechanisms by which uninfected cells are manipulated are not yet fully understood but appear to involve both direct contact with infected cells and also the actions of secreted factors [10 , 11] . Levels of the anti-inflammatory cytokine IL-10 increase during both productive and latent stages of infection , which may be important for the establishment and maintenance of a stable HCMV infection . IL-10 levels are positively associated with the incidence and duration of viraemia in HCMV positive transplant recipients [8 , 9 , 12–15] . Interestingly , it has also been shown that changes in serum IL-10 expression due to polymorphisms in the IL-10 gene positively correlate with the risks of HCMV infection and disease in transplant recipients [16 , 17] . IL-10 inhibits the function of antigen presenting cells and the production of proinflammatory cytokines , resulting in impaired CD4 T cell responses [18 , 19] . This has been suggested to contribute to the immunosuppressive effects seen during acute HCMV infection [8 , 9] . Many different cell types can produce IL-10 , although CD4 T cells seem to be the most important source in vivo [19] . HCMV itself contains a gene ( UL111A ) that encodes a viral IL-10 homologue with several similar functions to the cellular protein and that is able to induce cellular IL-10 production from monocytes and dendritic cells ( DC ) [20–22] . In murine cytomegalovirus ( MCMV ) infections , IL-10 production from T cells is also important for the establishment of high viral titres [23 , 24] . Although IL-10 producing T cells specific for both lytic and latent HCMV proteins have been identified in vivo , the mechanisms by which HCMV affects T cell IL-10 secretion during lytic infection are not yet understood in detail [25–27] . It is , however , known that ligation of the lymphocyte phosphatase CD45 can induce T cell IL-10 production [28–30] . The CD45 phosphatase regulates TCR signal strength via its control over the activity of the Src family kinase Lck and , thereby , influences diverse aspects of T cell function [31] . Changes in cytokine production have been linked to CD45 ligation and altered TCR signalling . Both a monoclonal antibody ( mAb ) directed against the R0 and RB isoforms of CD45 and the lectin galectin-1 , which binds to CD45 , have been shown to modulate the cytokine secretion profile of treated cells upon CD45 ligation , including by the induction of IL-10 [29 , 30 , 32] . As we have previously shown that the extracellular domain of the HCMV glycoprotein pUL11 interacts with CD45 in trans and perturbs T cell signalling and functions , we investigated the effects of pUL11 treatment on IL-10 secretion by primary T cells and considered the underlying changes to T cell signalling [33] . In this study , we show that treatment of PBMC with pUL11 induces IL-10 producing T cells and that supernatants of pUL11 treated cells have anti-inflammatory effects on untreated PBMC . CD45 can act as both a positive and negative regulator of T cell receptor ( TCR ) signal strength , depending on its expression level , and we show that pUL11 can also have both activating and inhibitory effects on T cell proliferation and on the kinase function of the CD45 substrate Lck .
The HCMV glycoprotein pUL11 interacts with the tyrosine phosphatase CD45 , resulting in inhibited TCR signalling and with a range of potential consequences for T cell function [33] . Changes in T cell function resulting from CD45 ligation by other ligands include the induction of IL-10 secretion in cells with activated TCR signalling [29 , 30 , 32] . To determine whether pUL11 has a similar effect , we measured IL-10 secretion by pUL11 treated PBMC . The UL11 protein is a transmembrane protein expressed on the cell surface of HCMV infected cells [34 , 35] . To investigate the effects of pUL11 , fusions of the extracellular domain of pUL11 with the Fc domain of human IgG1 ( UL11Fc ) were used [33] ( S1 Fig ) . This enabled us to investigate the effects of pUL11 in the absence of the complex mixture of other proteins produced during HCMV infection , some of which may also influence cytokine production . The sequence of pUL11 varies between different HCMV strains . To confirm that any effects seen were not limited to the protein from one strain of HCMV , we used fusion proteins of pUL11 derived from both the TB40 and Merlin strains of HCMV . A similar Fc fusion with pUL6 , a related protein from the same HCMV gene family , and the Fc domain alone were used as negative controls . PBMC were incubated in the presence or absence of pUL11 or control proteins in combination with the anti-CD3 antibody ( OKT3 ) for two days to induce TCR signalling . Supernatants were harvested and released IL-10 was measured by ELISA ( Fig 1A ) . Increased secretion of IL-10 was observed in the presence of pUL11 from either HCMV strain plus anti-CD3 and was not induced by the control proteins . Titration of pUL11 revealed that IL-10 production did not show a linear correlation with the concentration of pUL11 used; intermediate levels of pUL11 between 12 . 5nM and 50nM , depending on the strain of virus from which pUL11 was derived , were most effective at inducing IL-10 secretion . The difference in optimum concentration of pUL11 between the viral strains may reflect donor variations or small quantitative differences in the ability of the two proteins to induce IL-10 secretion . The difference in IL-10 secretion level between the pUL11 proteins from the two strains is due to donor variation in IL-10 secretion capacity; neither strain was consistently able to induce higher levels of IL-10 secretion than the other . To confirm that the effects of pUL11 on IL-10 secretion depended on its interaction with CD45 , we used a mAb directed against CD45 ( AICD45 . 2 ) . This antibody had previously been shown to inhibit the interaction of pUL11 with CD45 [33] . In the presence of this antibody , the production of IL-10 was abrogated in a dose-dependent manner by increasing antibody concentrations ( Fig 1B ) . IL-10 secretion was also not induced by pUL11 treatment in the absence of TCR stimulation ( Fig 1B ) . Several different cell types can secrete IL-10 [36] . To determine whether pUL11 treatment induces T cell IL-10 secretion , we enriched T cells from PBMC by negative selection to 98% purity ( S2 Fig ) . After stimulation using plate-bound anti-CD3 and soluble anti-CD28 to induce TCR signalling and co-stimulation , these T cells upregulated IL-10 secretion in the presence of pUL11 ( Fig 1C ) . Titration of pUL11 again resulted in a peak of IL-10 production at intermediate concentrations of pUL11; the curve is slightly shifted with respect to that seen for PBMC , both in terms of peak IL-10 production and of optimal pUL11 concentration , which may reflect differences in the intensity of TCR stimulation between the two experimental settings and variations between donors . The induction of IL-10 by the UL11 protein is concentration dependent ( Fig 1A ) . To determine whether the concentration of pUL11 expressed on the surface of HCMV infected cells is suitable for the induction of IL-10 from PBMC in coculture , retinal pigment epithelial cells ( RPE ) were infected with either the low passage parental Merlin strain of HCMV ( referred to here as HCMV wt ) , or HCMV ΔUL11 , from which the UL11 open reading frame ( ORF ) has been deleted [35] . Both viruses encode green fluorescent protein ( GFP ) , allowing the approximate determination of infected cell number by flow cytometry . A mouse mAb directed against pUL11 was used to measure the percentage of cells within the culture with pUL11 surface expression by flow cytometry ( S3 Fig ) . After two days of infection , cells from parallel cultures with similar infection rates of HCMV wt and HCMV ΔUL11 ( 50–80% infected , 17–26% expressing pUL11 over all three experiments , 0% , 2% and 6% differences in HCMV wt and HCMV ΔUL11 infection rates , with the higher infection rate for HCMV ΔUL11 ) were directly mixed with anti-CD3 treated PBMC at ratios of 1:2 , 1:5 and 1:10 . After two days of incubation , secreted IL-10 was measured by ELISA . The experiment was repeated three times using PBMC from three different donors ( Fig 2 ) . Proximity to HCMV infected cells induced the secretion of IL-10 in anti-CD3 stimulated cells . This secretion was reduced in the absence of pUL11 expression and also required anti-CD3 stimulation . Due to variations in IL-10 production between donors , IL-10 concentrations for each donor were normalized to the level induced by incubation of stimulated PBMC with uninfected RPE cells at a ratio of 1:5 . A significant difference in fold increase of IL-10 induction could be seen between the cultures containing HCMV wt and those containing HCMV ΔUL11 infected RPE cells ( p = 0 . 0005 ) . It should be noted that IL-10 secretion was not completely abrogated in the absence of pUL11 , indicating the presence of further viral factors capable of inducing IL-10 expression . The induction of IL-10 by pUL11 treatment could be a result of increased cytokine production by pre-existing IL-10 producing cells , or of the induction/expansion of new IL-10 secreting cells . To distinguish between these two possibilities we first measured the kinetics of IL-10 induction upon pUL11 treatment by ELISA ( Fig 3A ) . The pUL11 induced IL-10 production required 3–4 days to reach its maximum . As the half-life of IL-10 is short ( 60 minutes [37] ) , the delay in reaching high IL-10 levels may not be due to the slow accumulation of the cytokine , but rather to a process involving proliferation or conversion of cells to produce IL-10 . To investigate the impact of pUL11 on IL-10 producing cells directly , flow cytometry was used ( Fig 3B ) . The proportion of CD4 T cells producing IL-10 increased following three days of treatment with pUL11 from both TB40 and Merlin strains of the virus , more so than following stimulation with the anti-CD3 antibody alone or in the presence of the control protein , indicating that new IL-10 producer cells are induced . Under the conditions used in this experiment , no general expansion of the CD4 T cell compartment was seen; percentages of CD3+CD4+ cells remained between 49% ( UL11Fc treated cells ) and 54% ( Fc control treated cells ) ( Fig 3B ) . IL-10 is a potent anti-inflammatory cytokine that limits excess immune system activation . Phytohemagglutinin ( PHA ) is a lectin that induces the production of pro-inflammatory cytokines including IFNγ by PBMC [38] . The action of IL-10 on PBMC includes the inhibition of IFNγ production [36 , 39] . Treatment of PBMC with pUL11 induces IL-10 , but may also affect the secretion of other immunoregulatory cellular factors . For this reason we investigated whether or not conditioned medium from pUL11 treated anti-CD3 stimulated cells has an overall anti-inflammatory effect by measuring the inhibition of PHA-induced IFNγ production by PBMC . While the addition of conditioned medium from control anti-CD3 stimulated cells enhanced the effects of PHA , medium from pUL11 treated cells had a significant inhibitory effect ( Fig 4 ) . The main target of the CD45 phosphatase in T cells is the Src kinase p56Lck , which initiates TCR signalling by activating the signal-transducing immunoreceptor tyrosine activation motifs ( ITAMs ) in subunits of the TCR/CD3 complex [31 , 40 , 41] . The kinase activity of Lck is regulated by the phosphorylation of two of its tyrosine residues , one positive and one negative , both of which can be dephosphorylated by CD45 . Lck can , therefore , be either activated or inhibited , resulting in hyper- or hyposensitivity to TCR stimulation , depending on the availability of the CD45 phosphatase activity [42] . It has been shown , in mice , that titrating CD45 expression levels has a non-linear effect on T cell proliferation as a result , with reductions in CD45 expression levels down to approximately 30% of wild type resulting in enhanced signalling and proliferation , and larger reductions below this level inhibiting T cell signalling and proliferation [42] . If the interaction of pUL11 with CD45 results in a concentration dependent reduction of the CD45 phosphatase function , titrating the concentration of pUL11 in the experiment would be expected to have similar effects to titrating CD45 expression . Treating PBMC with varying concentrations of pUL11 in the presence of TCR stimulation did result in biphasic effects on proliferation , with lower concentrations of pUL11 enhancing proliferation and higher concentrations having an inhibitory effect , for pUL11 from both the TB40 and the Merlin strains of HCMV ( Fig 5A and 5B ) . Interestingly , at the pUL11 concentrations shown to induce IL-10+ CD4 T cells ( 50nM ( Fig 3B ) ) , no increased overall proliferation was seen , again indicating that IL-10 induction is not due to a general T cell expansion . The non-linear concentration dependent effects of CD45 on T cell proliferation have been shown to be due to its actions on the Lck kinase [42] . Phosphorylation at tyrosine 505 of Lck induces an intermolecular interaction , “closing” and inactivating Lck , and , thereby , preventing proliferation in response to TCR stimulation [43 , 44] . Phosphorylation of Lck at tyrosine 394 , however , is necessary to allow substrates access to the kinase domain of Lck and , in excess , leads to hyperresponsiveness characterized by uncontrolled T cell proliferation [45–47] . Relatively low amounts of active CD45 are sufficient to activate Lck by dephosphorylating tyrosine 505 , but even a slight loss of CD45 activity impairs its ability to restrict the kinase activity of Lck by dephosphorylating tyrosine 394 , meaning that minor decreases in CD45 activity result in hypersensitive T cell signalling and enhanced proliferation [42] . To explore whether the pUL11-induced proliferation changes in PBMC are linked to a modulation of the available CD45 phosphatase activity and , therefore , the ability of CD45 to dephosphorylate its substrate Lck , we measured changes in the two regulatory phosphotyrosines of Lck . PBMC cultures were treated with differing concentrations of pUL11 for three days and then Lck phosphorylation in CD4 T cells was measured by phosphoflow cytometry ( Fig 6A ) . The experiment was repeated using PBMC from three different donors , with similar results obtained in all three cases . The combined data from all three experiments indicated significant increases in the degree of phosphorylation of both regulatory tyrosines of Lck between pUL11 treated and control treated cells ( Fig 6B ) . Treatment with the control Fc domain resulted in a slight reduction of phosphorylation compared to that seen in cells treated with anti-CD3 alone , possibly due to non-specific steric hindrance of anti-CD3 binding . The UL11 protein affected Lck Y505 and Y394 differently . Whereas only treatment with the highest concentration of pUL11 ( 200nM ) resulted in increased Y505 phosphorylation , increased Y394 phosphorylation could be seen at all concentrations of pUL11 used . These results mirror the effects on Lck phosphorylation of titrating CD45 expression and suggest that pUL11 treatment modulates the available phosphatase activity of CD45 [42] . High concentrations of pUL11 inhibit CD45 phosphatase activity and lead to increased levels of phosphorylated Y505 , thereby promoting the adoption of a closed , inactive conformation of Lck . The increased levels of phosphorylated Y394 would result in higher numbers of active Lck kinase molecules , an effect that would predominate at low concentrations of pUL11 . IL-10 secretion in T cells is known to be influenced by changes in TCR signal strength , such as are induced by alterations in Lck activity [18 , 40 , 48] . The induction of IL-10 secretion after treatment with pUL11 requires TCR stimulation , provided here by the anti-CD3 antibody OKT3 , and is also highly dependent on the concentration of pUL11 used ( Fig 1 ) . Altered signalling states could be generated in the T cells by TCR stimulation in the presence of varying concentrations of pUL11 . We therefore examined the effects of pUL11 treatment on the early events of T cell signalling and activation in more detail . As the intensity of signalling effects in primary T cells strongly varies between donors , we used the Jurkat T cell line for these experiments . Although Jurkat T cells differ from primary T cells in their functional responses , the initial signalling events downstream of the TCR in Jurkat cells are similar to those of primary cells , making this cell line a useful model to address this question [49] . The treatment of anti-TCR stimulated Jurkat T cells with high concentrations of pUL11 results in a generally decreased tyrosine phosphorylation [50] ( Fig 7A ) . While artificial , this experimental setting allows us to determine whether pUL11 is , in principle , able to influence T cell function by affecting the TCR signalling pathway . In resting cells , elements of the TCR signalling pathway are in dynamic equilibrium , resulting from the combined activities of kinases and phosphatases . Following TCR stimulation , the balance shifts and the ITAMs of the CD3 γ , δ and TCR ζ-subunits are increasingly phosphorylated by activated Lck . The subunit with the largest number of ITAMs is the TCR ζ-chain and this is generally considered the most important for signal transduction [51] . The absence of Lck results in reduced ζ-chain Y142 phosphorylation [52 , 53] . In pUL11 treated cells TCR stimulation leads to reduced phosphorylation of Y142 of TCR ζ ( Fig 7B ) in comparison to control cells . This is consistent with a perturbed Lck function in pUL11 treated cells . Following TCR ζ-chain phosphorylation by Lck , the ζ-chain associated protein kinase of 70kDa ( ZAP-70 ) is recruited and phosphorylated , again by Lck [54] . Treatment with pUL11 resulted in reduced phosphorylation of ZAP-70 Y319 upon stimulation ( Fig 7C ) , which is again consistent with reduced Lck activity [53] . ZAP-70 phosphorylation is essential for the subsequent signalling step; the first with a direct link to alterations in T cell function [40] . The main substrates of ZAP-70 are the scaffold-forming adaptor proteins Linker for activation of T cells ( LAT ) and leukocyte phosphoprotein of 76kDa SLP-76 [55] . Which signalling proteins assemble upon LAT is determined by LAT’s phosphorylation state and then influences the degree to which the downstream branches of the signalling pathway are activated , controlling e . g . T cell proliferation and differentiation [41] . Treatment of Jurkat T cells with pUL11 resulted in decreased LAT Y171 phosphorylation ( Fig 7D ) . Phospholipase C ( PLC ) γ is activated upon interacting with LAT [56] . TCR induced PLCγ Y783 phosphorylation is also reduced by pUL11 treatment ( Fig 7D ) . These results indicate that pUL11 can impair TCR signalling and is likely to be able to induce alterations in T cell function by this route .
The HCMV glycoprotein pUL11 interacts with the lymphocyte surface phosphatase CD45 [33] . CD45 plays important roles in many immune system functions; in the absence of CD45 , both humans and mice suffer from a severe combined immunodeficiency ( SCID ) phenotype . It has been known for many years that signalling through the T cell receptor is completely abrogated in the absence of CD45 , B cell receptor signaling is severely disrupted and the functions of other cell types are also affected [31] . It is , however , also clear that CD45 can have far wider effects on T cell function than simply acting as an on/off switch [42] . The level of CD45 expression or activity controls TCR signal strength , which influences a diverse range of T cell features [40] . Although no specific cellular CD45 ligand has been identified , lectins that interact with CD45 , among other cell surface glycoproteins , have been described . Galectin-1 and Macrophage Galactose-type Lectin ( MGL ) for example are able to inhibit CD45 activity and to manipulate T cell function [29 , 57] . Interestingly , the interaction of galectins with CD45 induces changes to T cells that are reminiscent of those seen following treatment with some CD45-specific monoclonal antibodies [28 , 30] . These mAbs have been shown to reduce rejection of transplants and also to have anti-inflammatory effects [28 , 58–61] . Both the galectin and mAb CD45 ligands induce a suppressive phenotype in T cells , characterized by changes in cytokine secretion including increased production of the anti-inflammatory cytokine IL-10 . We have shown that pUL11 treatment , as has been seen for other CD45 ligands and CD45 antibodies , can affect T cell function ( Fig 8 ) [28–30] . We demonstrated that T cells are a source of IL-10 following pUL11 treatment and that the enhanced IL-10 secretion is both dependent on the interaction of pUL11 with CD45 in the prescence of TCR signaling and associated with an increased proportion of IL-10 producing CD4 T cells . Our results , therefore , support the view that CD45 is a controller of T helper cell decisions , as well as a regulator of activation thresholds and that these properties can be manipulated by the binding of CD45 ligands . We were also able to demonstrate that pUL11 potentiated induction of IL-10 is part of the repertoire of HCMV infected cells; IL-10 induction was markedly reduced in a deletion mutant lacking pUL11 , but not completely abrogated . It is not unusual for HCMV to use more than one mechanism to control important functions; interestingly , HCMV encodes a viral IL-10 homologue , the product of the UL111A gene , which diminishes macrophage , dendritic cell and T cell responses in vitro and has recently been shown to induce the production of cellular IL-10 in monocytes and DCs , but not T cells [21 , 22 , 62–65] . The residual IL-10 induced by ΔUL11 HCMV infected cells may therefore be , at least partly , due to the activity of this protein . Conditioned medium from cells treated with pUL11 has an anti-inflammatory effect on PHA stimulated PBMC , shown by significant reductions in IFNγ production . IL-10 is a powerful inhibitor of both monocyte dependent and independent T cell IFNγ production making this a likely mechanism for the observed effects of pUL11 [39 , 66] . IL-10 has a broad range of anti-inflammatory functions; it acts on antigen presenting cells to reduce surface expression of stimulatory and co-stimulatory molecules and inflammatory cytokine secretion and also directly on CD4 T cells [19 , 36] . T cell proliferation is reduced leading to suppression of Th1 functions in particular . The production of IL-10 is necessary for the host , in order to reduce the immune pathology otherwise associated with infections , but an excess also reduces the effectiveness of immune control [36] . Increased levels of IL-10 have been described in HCMV infected transplant recipients , with larger increases associated with viraemia and disease [8 , 9 , 12 , 15] . The induction of high levels of IL-10 during acute HCMV infections has been speculated to power other immunosuppressive properties of HCMV [8 , 9] . IL-10 production has also been shown to be important for murine cytomegalovirus ( MCMV ) infections; by reducing anti-viral responses , establishment of infection is permitted and immune pathology is inhibited [63 , 67] . IL-10 derived from CD4 T cells in particular suppresses antiviral responses to MCMV , and contributes to the development of high viral titres [23 , 24] . In the salivary glands , where the virus establishes chronic infection lasting for several months , clearance is ultimately mediated by CD4 T cells . Virus induced CD4 T cell IL-10 production is pivotal to the establishment of this chronic infection , due to its inhibitory effects on CD4 T cell function [68] . This mechanism may have relevance to other mucosal sites of infection , shown to be important for cytomegalovirus transmission [69] . The mechanisms leading to IL-10 secretion have been less well studied in T cells than in monocytes , but appear to be affected by changes in TCR signal strength , such as could result from altered CD45 activity [18 , 49] . CD45 acts via its substrate the Src family kinase Lck , which plays an essential role in TCR signal transduction by phosphorylating key signalling intermediates . Lck can exist in several phosphorylation and activation states that modulate TCR signal strength [40] . As do all Src family kinases , Lck contains two regulatory tyrosine residues . The inhibitory residue , Y505 in Lck , maintains an intermolecular interaction when phosphorylated , holding Lck in a closed , inactive state [47] . Dephosphorylation of Y505 by CD45 allows Lck to open , forming a “primed” state . Insufficient CD45 activity means that Lck cannot be optimally primed and T cells are less able to transduce incoming TCR signals . CD45 is the only phosphatase that can remove the inhibitory phosphate group from Y505 of Lck and it is for this reason that TCR signalling is completely prevented in the absence of CD45 [70 , 71] . Titrating CD45 activity does not , however , have a simple linear effect on T cell activation [42] . The second regulatory tyrosine of Lck , Y394 , is in a loop adjacent to the kinase domain and must be phosphorylated for the kinase to be functional [47] . Excess Y394 phosphorylation results in increased Lck kinase activity , meaning that T cells are hyperreactive to stimulation via the TCR [45] . The dephosphorylation of these two residues by CD45 are not , however , equally favoured [42] . While small amounts of CD45 ( 3% of normal levels in mice ) can be sufficient to activate Lck by dephosphorylating Y505 , close to the full wild type concentration is required to prevent excess Lck activity by dephosphorylating Y394 [42] . A small loss of CD45 activity therefore results in T cells that are hypersensitive to TCR stimulation . We measured the effects of pUL11 on the phosphorylation of Lck in primary T cells . With the highest concentration of pUL11 , the inhibitory Y505 residue showed increased phosphorylation . This increase in Y505 phosphorylation was largely lost at intermediate and low concentrations of pUL11 . The activatory Y394 residue , however , showed comparatively increased phosphorylation at all concentrations of pUL11 tested . Similar results have been reported in mice; at CD45 levels of between 10 and 60% of wild type , Y394 phosphorylation was increased 1 . 4 to 2 . 1 fold , resulting in enhanced T cell signalling capacity despite some inhibition of Y505 dephosphorylation . At lower CD45 expression levels , the disruption in pY505 dephosphorylation became functionally more important and T cell activation was inhibited [42] . We determined how varying the concentration of pUL11 affects the threshold for TCR signalling using PBMC proliferation as a readout . The results mirrored what is known for CD45 titration; whereas high concentrations of pUL11 , which would correspond to low amounts of CD45 activity , inhibited T cell proliferation , when low pUL11 concentrations were used proliferation was markedly increased . We have previously shown that pUL11 treatment of the Jurkat CD4 T cell line can reduce the rapid induction of tyrosine phosphorylation seen upon stimulation via the TCR [33] . Here we considered the effects of pUL11 treatment on individual proteins immediately downstream of the TCR . The first event following CD3 stimulation , phosphorylation of ITAM motifs in the TCR ζ-chain , was partially inhibited in the presence of pUL11 . Phosphorylation of tyrosine 142 of the ζ-chain is exclusively mediated by Lck and its reduction shows the effects of reduced Lck function following pUL11 treatment [52 , 53] . Disruption in ζ-chain ITAM phosphorylation is likely to reduce the recruitment of ZAP-70 , the next protein in the pathway and also an Lck substrate [53 , 72] . Phosphorylation of ZAP-70 at Y319 was also reduced upon pUL11 treatment . These effects are then passed down the pathway , meaning that phosphorylation of the scaffolding nodes with control over differentiation and cell fate is also affected [55 , 73] . Tyrosine 171 of LAT is a key residue in the formation of the signalling complex [74 , 75] . We showed that phosphorylation of this site is inhibited upon pUL11 treatment . PLCγ binding to LAT and its phosphorylation is essential for its activation [76] . Again , the phosphorylation of PLCγ at Y783 was reduced in the presence of pUL11 [56] . We have shown here that the regulation of TCR signal strength and key signalling proteins is affected by pUL11 treatment and it seems plausible that this can underlie the pUL11 induced changes in T cell function . IL-10 induction in T cells , while not yet completely understood , is sensitive to the strength of the input signal and appears to require several pivotal signalling events downstream of the TCR [18 , 48 , 77–79] . This is consistent with the observed induction of IL-10 secretion at intermediate pUL11 concentrations corresponding to increased TCR signal strength . The reason that IL-10 induction is not seen when the signal strength is highest is less clear . It is , however , well established that increased TCR signal strength can have inhibitory effects on a range of functions , including cytokine production , via alterations in downstream signalling events [40 , 80 , 81] . pUL11 can induce changes in phosphorylation of Lck . It must , however , be borne in mind that although Lck is considered to be the primary substrate of CD45 , it is not the only one . pUL11 treatment may also affect other signalling networks with the potential to influence T cell proliferation and IL-10 production . The Src kinase Fyn is also under the control of CD45 and is important in the regulation of TCR signalling strength and its activity [82 , 83] . CD45 has also been shown to be a negative regulator of Jak/Stat signalling , independently of its effects on Lck [84] . Nevertheless , the TCR signalling pathway appears to be pivotal to the pUL11 induced effects shown here , as in the absence of TCR stimulation no effects of pUL11 treatment were seen . While pUL11 is able to potentiate IL-10 induction in vitro , the situation in vivo is likely to be far more complex . The effects of pUL11 appear to be highly concentration dependent; its effects in vivo may therefore vary between microenvironments . CD4 T cells are more sensitive to changes in CD45 activity than CD8 T cells are; CD8 T cells from mice expressing only 3% of wild type levels of CD45 retained full cytotoxic function , whereas CD4 T cell helper functions were still impaired in mice expressing 30% of wild type CD45 [42] . pUL11 may therefore be more likely to affect CD4 T cell function in vivo . It may also be for this reason that HCMV specific CD8 T cell secretion of IFNγ was not affected by the presence of pUL11 on infected cells [35] , in contrast to the results presented here for IL-10 secretion , which is preferentially produced by CD4 T cells [19] . The UL11 protein seems likely to play a role in vivo; antibodies to pUL11 have been found in patient sera , implying that the protein is expressed at detectable levels [85] . The protein is variable between strains , but so far no clinical isolates of the virus in which pUL11 is absent or nonfunctional have been confirmed [86] . In a first attempt to investigate whether pUL11 variants from different HCMV strains might differ with regard to their ability to induce IL10 secretion and T-cell proliferation we tested the two pUL11 variants from the TB40 and Merlin strains . However , both pUL11 variants appeared to induce similar effects in our experimental system . The effects of variations in UL11 sequence on protein function and disease outcome will therefore have to be the subject of future more extensive studies . The interaction of pUL11 with CD45 is the first example of a viral protein targeting CD45 to induce T cells with anti-inflammatory properties . It is also the first HCMV protein shown to induce IL-10 secretion in T cells . Understanding the mechanisms by which changes in signal strength can influence T cell development and function will contribute to the development of therapies against immune pathologies and may also provide the basis for antiviral treatments .
Human blood cells were provided by voluntary blood donors in the Institute of Transfusion Medicine , Hannover Medical School . All materials and data were analysed anonymously . The use of human blood cells was approved by the ethics committee of Hannover Medical School . The Jurkat E6-1 cell line was cultured in RPMI1640 medium supplemented with 10% FCS , 1% penicillin/streptomycin , and 2mM L-glutamine . PBMC were flushed from leukocyte filters used to prepare white blood cell-depleted erythrocytes from healthy voluntary blood donors for transfusion and separated using Biocoll ( Merck Millipore , Darmstadt , Germany ) density gradient centrifugation . Cells were frozen and stored in liquid nitrogen until usage . PBMCs were cultured in the same medium as Jurkat cells , with the addition of 1mM sodium pyruvate . T cells were isolated from PBMCs using a magnetic negative selection enrichment kit ( BD IMagT Human T Lymphocyte Enrichment Set-DM ( Becton Dickinson , Heidelberg , Germany ) ) and used directly after preparation . PBMC ( or primary T cells ) from at least two different donors were used for each experiment . RPE cells were cultured in Dulbecco's Modified Eagle Medium: Nutrient Mixture F-12 ( DMEM/F12 , ThermoFisher Scientific , Darmstadt ) supplemented with 10% FCS , 1% penicillin/streptomycin , 2mM L-glutamine . The two viruses were used in this study were both kindly provided by Martin Messerle: HCMV wt was produced using the bacterial artificial chromosome ( BAC ) -cloned genome of HCMV Merlin-UL128LTB40 , which contains a frameshift in RL13 and , after UL122 , an IRES followed by the gene encoding GFP [87 , 88] . HCMV ΔUL11 is based on the same BAC , but with a deletion of the UL11 ORF and containing the ORF encoding gaussia luciferase in its place ( for construction see[35] , HCMV HM11DL mutant ) . HCMV wt and HCMV ΔUL11 are based on the Merlin strain of HCMV . When propagated in RPE cells , this is a cell associated virus . HCMV wt or HCMV ΔUL11 infected cell cultures were therefore maintained by adding uninfected RPE cells to infected RPE at a ratio of 1:1 or 2:1 after the infected cells reached 100% cytopathic effect as determined by microscopy . Fc fusion proteins were generated using the predicted extracellular domains of pUL11 from the TB40 and Merlin strains of HCMV and from the predicted extracellular domain of pUL6 from the TB40 strain of HCMV . UL11Fc ( TB40 ) , UL11Fc ( Merlin ) and UL6Fc consist of the predicted extracellular regions of the respective proteins fused to the Fc domain of human IgG1 [33] . The Fc control protein consists only of the Fc domain . All three proteins were expressed from retrovirally transduced 293T cells ( UL11Fc ( TB40 ) and Fc control protein ) or from adenovirally transduced RPE cells ( UL11Fc ( Merlin ) and UL6Fc ) and purified by protein A affinity chromatography as described previously [33 , 35] . Statistical analysis was performed using GraphPad Prism 5 . Two-way ANOVA was used to analyse differences between groups due to the presence of the UL11 protein . p-values are shown as: *<0 . 05 , **<0 . 01 , ***<0 . 001 . 2 . 5 x 105 Jurkat T cells were incubated with 800 nM UL11Fc or the Fc control protein in 100 μl medium for 30 min at 37°C . Cells were stimulated with 100 μl of anti-Jurkat-TCR antibody C305 hybridoma supernatant [89] ( kindly provided by B . Schraven , Magdeburg ) for the indicated times at 37°C . The reaction was stopped by adding 1 ml ice-cold PBS and cells were immediately centrifuged at 4°C . The cell pellet was lysed with lysis buffer ( 1% NP-40 , 1% N-dodecyl-β-D-maltoside , 50 mM Tris-HCl pH 7 . 4 , 150 mM NaCl , 10 mM EDTA , 2 mM sodium vanadate and protease inhibitor cocktail ( Calbiochem , Merck Millipore , Darmstadt , Germany ) and incubated on ice for 20 min . The samples were frozen and stored until usage at -20°C . The lysate was centrifuged for 10 min at 12 , 000 rpm and proteins were separated using SDS-PAGE and transferred to a nitrocellulose membrane . Phosphorylated proteins were detected with phospho-specific antibodies; anti-phospho-tyrosine ( 4G10 ) ( Merck Millipore , Darmstadt , Germany ) anti-TCR ζ-chain p-142 ( BD , Heidelberg , Germany ) , anti-ZAP-70p319 ( Merck Millipore , Darmstadt , Germany ) , anti-LATp-171 ( Cell Signaling Technology , Leiden , Netherlands ) , and anti-PLCγp-783 ( Cell Signaling Technology , Leiden , Netherlands ) . Control antibodies to determine equal loading of the gel were anti-GAPDH ( Cell Signaling Technology ) , anti-TCR ζ-chain ( Sigma Aldrich , Taufkirchen , Germany ) , anti-ZAP-70 ( Merck Millipore ) and anti-actin ( Cell Signaling Technology ) . Secondary antibodies were IR Dye 800CW Goat anti-mouse IgG , IR Dye 680LT Goat anti-mouse IgG , IR Dye 680LT Goat anti-rabbit IgG and IR Dye 800 CW Goat anti-rabbit IgG ( Licor ) . Imaging and quantification were performed using a Licor Odyssey system with Odyssey Application Software version 3 . 0 . To measure proliferation of PBMCs in response to pUL11 , UL11Fc or Fc control fusion proteins were adsorbed at indicated concentrations together with anti-CD3 ( OKT3 , 1 μg; purified from hybridoma supernatant ) onto Nunc Maxi-Sorb 96-well plates ( Sigma Aldrich ) for 30 min at 37°C . The plate was washed three times and 1x105 PBMCs per well were incubated in 100 μl of culture medium . After 48 h , 0 . 4 mCi [3H]-thymidine ( Amersham Biosciences , Braunschweig , Germany ) was added . 24h later the cells were harvested and incorporated [3H]-thymidine was measured in a beta-counter ( Perkin Elmer , Rodgau , Germany ) . To measure IL-10 secretion , PBMCs or primary T cells were stimulated with plate-bound anti-CD3 ( OKT3 , 1μg; purified from hybridoma supernatant ) with or without soluble co-stimulatory anti-CD28 ( 2 μg/ml ) together with plate bound UL11Fc or the Fc control protein as in PBMC proliferation experiments . The amount of IL-10 secretion was determined after stimulation by ELISA ( IL-10 ELISA MAX Standard set ( BioLegend , Fell , Germany ) ) according to the manufacturer’s instructions . Supernatants were always pooled from triplicate wells of a 96-well plate and used for duplicate ELISA measurements ( technical replicates ) . PBMCs were incubated in Nunc Maxi-Sorb 96 well plates coated with anti-CD3 ( OKT3 1 μg; purified from hybridoma supernatant ) and UL11Fc or Fc control protein ( 50 and 100 nM ) for four days . Supernatant ( conditioned medium ) was taken at day four , immediately frozen and stored at -20°C until use . Freshly isolated PBMCs were incubated together with 50 μl of the conditioned medium as decribed above and stimulated with PHA ( 1μg/ml ) ( eBioscience , Frankfurt am Main , Germany ) for two days in a total volume of 200 μl . On day two , IFNγ secretion was measured by ELISA ( human IFNγ ELISA MAX Standard ( BioLegend ) ) according to the manufacturer’s instructions . Fresh uninfected RPE cells were added to HCMV wt or HCMV ΔUL11 infected cell cultures as described above . After 48h , to provide an indication of the infection rate of the cells for experiments , GFP expression was measured using flow cytometry . Expression of pUL11 at the cell surface was detected by flow cytometry following staining with a mouse monoclonal antibody specific for pUL11 from the Merlin strain of HCMV [35] , kindly provided by Martin Messerle , followed by a PE-coupled goat anti-mouse secondary antibody ( Immunotools , Friesoythe , Germany ) . Infected cultures were used for experiments when the infection rate of the two cultures did not differ by more than 6% , with the HCMV ΔUL11 infected culture having the highest infection rate , and when the infection rate was at least 50% . Sixty-thousand PBMC were mixed with the appropriate number of RPE cells to result in ratios of RPE:PBMC of 1:2 , 1:5 and 1:10 and incubated in a anti-CD3 coated 96 well plate . After 2 days , supernatant was harvested and the concentration of IL-10 measured by ELISA as described above . To determine the phenotypes of T cell subsets after UL11Fc treatment 1x105 PBMCs were stimulated and incubated with UL11Fc or the Fc control protein as above . After 24 , 48 , or 72 h the cells were labelled to detect surface expression of CD3 and CD4 ( using anti-human CD3-APC eFluor 480 Clone SK7 and anti-human CD4-APC Clone SK3 , ( eBioscience ) ) . Intracellular IL-10 expression was detected ( using anti-IL-10-PE , clone JES3-19F1 , BioLegend ) following 2 h of treatment with Brefeldin A ( 5μg/ml ) ( BioLegend ) treatment and fixation and permeabilisation using Fixation/Permeabilization concentrate and diluent ( eBioscience ) according to the manufacturer’s instructions . Live/dead cells were distinguished by exclusion of the Aqua fluorescent reactive dye ( ThermoFisher scientific , Darmstadt ) . Measurements were made on a BD LSRII flow cytometer . To detect phosphorylated Lck , PBMCs were incubated with anti-CD3 ( OKT3 ) together with UL11Fc or the Fc control protein , as in proliferation experiments , but for three days . Cells were harvested and handled on ice . Live cells were detected by exclusion of dye ( Zombie NIR Fixable Viability kit , BioLegend ) . p-Lck was stained intracellularly and measured with specific antibodies for Lck-p505 ( BD Phosflow , anti-human Lck ( pY505 ) , BD Heidelberg , Germany ) and for Lck-p394 ( BD Phosflow . anti-Src ( pY418 ) , BD ) after fixation and permeabilization of the cells using Fixation buffer and Permeabilization Wash buffer ( BioLegend ) according to the manufacturer’s recommendation . Measurements were made on a Beckman Coulter FC 500 Analyzer flow cytometer . To inhibit IL-10 secretion 1x105 PBMCs were pretreated with the AICD45 . 2 anti-CD45 antibody ( kindly provided by R . Schwinzer ) that has been shown to prevent the interaction of UL11 with CD45 [33] for 30 min at 37°C . Pretreated PBMCs were then incubated in 96-well plates coated with anti-CD3 ( OKT3 1 mg; purified from hybridoma supernatant ) and UL11Fc or Fc control ( 50 nM ) for 2 days . After incubation secreted IL-10 was measured by ELISA . | Human cytomegalovirus ( HCMV ) infects from 45% to 100% of people worldwide , depending on local socio-economic factors . Although usually harmless in healthy individuals , infection with HCMV can cause severe disease in people with weakened or immature immune systems such as transplant recipients and newborns . The establishment and maintenance of life-long infections by HCMV are greatly aided by its ability to modulate the host’s immune system during both active and latent infection; infected cells are masked and both infected and uninfected immune cells have their functions manipulated . One effect of HCMV infection is the induction of the cytokine IL-10 , a secreted protein that suppresses many antiviral responses . Here , we identify a viral protein , pUL11 , which can induce IL-10 expression by T cells and reduce the production of mediators of inflammation . pUL11 interacts with CD45 , an immune system regulator that controls the sensitivity of T cells and has been linked to IL-10 production . We show that pUL11 can likewise affect T cell responses to stimuli , depending on its concentration , and suggest that this underlies its functions . pUL11 is the first viral protein known with this mechanism and further understanding of its effects may lead to the development of novel antiviral therapies and also help in the treatment of immune system disorders . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"blood",
"cells",
"phosphorylation",
"t",
"helper",
"cells",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"enzyme-linked",
"immunoassays",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"immunology",
"microbiology",
"cytomegalovirus",
"infection... | 2017 | The human cytomegalovirus glycoprotein pUL11 acts via CD45 to induce T cell IL-10 secretion |
Discovering the mechanisms by which proteins aggregate into fibrils is an essential first step in understanding the molecular level processes underlying neurodegenerative diseases such as Alzheimer’s and Parkinson's . The goal of this work is to provide insights into the structural changes that characterize the kinetic pathways by which amyloid-β peptides convert from monomers to oligomers to fibrils . By applying discontinuous molecular dynamics simulations to PRIME20 , a force field designed to capture the chemical and physical aspects of protein aggregation , we have been able to trace out the entire aggregation process for a system containing 8 Aβ17–42 peptides . We uncovered two fibrillization mechanisms that govern the structural conversion of Aβ17–42 peptides from disordered oligomers into protofilaments . The first mechanism is monomeric conversion templated by a U-shape oligomeric nucleus into U-shape protofilament . The second mechanism involves a long-lived and on-pathway metastable oligomer with S-shape chains , having a C-terminal turn , en route to the final U-shape protofilament . Oligomers with this C-terminal turn have been regarded in recent experiments as a major contributing element to cell toxicity in Alzheimer’s disease . The internal structures of the U-shape protofilaments from our PRIME20/DMD simulation agree well with those from solid state NMR experiments . The approach presented here offers a simple molecular-level framework to describe protein aggregation in general and to visualize the kinetic evolution of a putative toxic element in Alzheimer’s disease in particular .
The aggregation of amyloid β protein ( Aβ ) , the likely cause of Alzheimer's disease , is widely studied via experiment and computational efforts . [1 , 2] The end product of the Aβ aggregation process is a fibril whose structure depends strongly on the environment and has diverse polymorphic features , although U-shape ( β strand—turn- β strand motif ) β-sheets’ protofilaments are a consistent theme . [2–7] One of the important goals in the current research is to understand the kinetic mechanism of fibril formation together with the ultimate goal for identifying the toxic species , which are now thought to be early-stage soluble oligomers , and also clarifying their structural and kinetic characters . [8–11] A number of candidates for those toxic oligomers have been suggested , including the paranuclei , pentamers and hexamers of Aβ42 peptides , which are observed in vitro , and β-rich structures with exposed hydrophobic residues which are thought to form in vivo in the vicinity of bi-lipid membranes . [8 , 11–16] Several candidate toxic oligomers appear to have a generic structural character such as a bend in the C-terminal near residues G37 and G38 . [8 , 14 , 17] In fact Pande and coworkers have shown that designing a turn into the C-terminal by mutation enhances the stability of the oligomers which , in their experiment , appear to be off-pathway . [17] In addition Smith and coworkers observed S-shape monomers ( from K16 to A42 ) containing a C-terminal turn within disc-shape pentamers and found these oligomers to be toxic . [8] Teplow and coworkers also detected toxicity in Aβ oligomers containing a C-terminal turn designed by mutation . [14] Despite some advances recently in our knowledge on the fibrillization process and the identity of toxic species , detailed molecular-level descriptions of the structural conversion of Aβ monomers to early stage oligomers to potentially toxic oligomers to protofilaments are not yet available . Knowledge of the oligomerization and structural conversion of Aβ peptides to proto-fibrils at the atomic scale would allow us to ascertain how the toxic species emerge and how they achieve meta-stability . The focus of this paper is Aβ17–42 , a 26-residue C-terminal fragment of Aβ42 , the peptide whose aggregation is most strongly linked to Alzheimer’s disease . Aβ17–42 is produced from the cleavage of amyloid precursor protein by α- and γ-secretases and is observed in amyloid plaques which are composed of amyloid fibrils . [18] It has been suggested that the Aβ17–42 structures form U-shape protofilaments similar those for Aβ40 or Aβ42 , which is supported by computational stability study . [19] Since Aβ17–42 is a key fragment of Aβ42 , the formation of its U-shape protofilament is likely to be very similar to that of its longer parent Aβ42 . Justification for this idea is that Aβ17–42 contains the two hydrophobic stretches that dominate the aggregation and fibrillization of Aβ42 as well as the turn region . In addition , it appears that the N-terminal Aβ residues 1–10 or 1–16 do not participate in the rigid portions of the U-shape protofilament in synthetic fibrils observed by the groups of Tycko , Riek and Bertini , [3 , 4 , 20 , 21] although they do participate in the tubular shape protofilaments observed by Zhang et al . and Miller et al . [5 , 22 , 23] serving as arms that form a sheath surrounding the hollow core structures . They also participate in the ordered fibril structures derived from Alzheimer’s brain tissue . [24] Aβ42 is a more toxic peptide than Aβ40 or Aβ17–42 , which means that the two C-terminal residues and the rather flexible N-terminal residues likely play an important role in amyloidogenesis and toxicity , as is supported by mutagenesis and bioinformatics studies . [25–27] Since it would be extremely difficult to simulate spontaneous fibril formation of full length Aβ42 given current computational constraints , we focus here on the role of the C-terminal residues and their turns . Restricting our attention to Aβ17–42 also makes it easier to watch the spontaneous U-shape conformation form without the encumbrances that would occur in the presence of the highly flexible N-terminal residues . Molecular in-silico description of spontaneous fibril formation by Aβ40 , Aβ42 and even Aβ17–42 based on all-atom or coarse-grained models is still extremely challenging due to our inability to capture the multi-scale nature of the force-field , the very long time scales for fibrillation ( much longer than that for protein folding ) and the variety of polymorphic structures observed with different backbone orientations , protofilament conformations , and protofilament stacking arrangements . [6] All atom simulations examining the kinetic stability of Aβ17–42 peptides in preformed stacked fibrillar structures and annular oligomeric structures characteristic of ion-channels have been conducted . [19 , 28–30] Computational studies of the spontaneous oligomerization of Aβ40 and Aβ42 using coarse-grained models have been performed . [31 , 32] Hybrid combinations of all-atom and coarse-grained simulations of Aβ40 , Aβ42 and Aβ17–42 have been performed on 2 and 3-peptide systems . [33 , 34] Fibril elongation by monomer addition to a preformed Aβ17–42 fibrillar structure has been simulated for very long times ( ~1 . 3ms ) by a hybrid resolution molecular dynamics . [35] The extensive model and kinetic network analysis reveals atomistic details of a monomer participating in the dock-and-lock mechanism[36] , thereby explaining unidirectional fibril growth . This work shows that addition and reorganization near a preformed structure containing as little as one monomer needs extensive simulation time . All-atom simulations with Aβ42 dimer and inhibitors have been performed to understand the inhibitory mechanism for oligomerization . [37] However , the entire fibrillation pathways starting from randomly denatured structures progressing through the formation of oligomeric intermediates and leading to formation of the U-shape fibril structures , to be consistent with those from experiments , are not accessible yet . In this study , we apply discontinuous molecular dynamics ( DMD ) simulations in conjunction with the PRIME20 force field[38–40] to simulate fibrillation of systems containing 8 Aβ17–42 peptides initially starting from their randomly disordered conformations . Fibril structures with U-shape β-sheets are constructed successfully in our DMD simulations , which then provide structural information during the entire kinetic process of aggregation . Among the many simulations performed at various temperatures , we focus on on-pathway trajectories which provide excellent fibril structures consistent with those suggested in the experiments . [8 , 20] Two different mechanisms for structural conversion from randomly disordered conformations to protofilaments emerge from different pathways: ( 1 ) one-by-one monomeric conversion to a fibrillar structure , and ( 2 ) slow conversion through a meta-stable oligomer with “S”-shape conformations containing a C-terminal outward turn to a final fibrillar structure . The “S”-shape conformations are postulated to be a key component of toxic oligomers , based on results of our simulations and others’ experiments . [8 , 14 , 17]
We performed preliminary simulations with different numbers of Aβ17–42 peptide chains ( NC ) i . e . NC = 1 , 2 , 4 , 5 , 6 , 8 , 10 , 12 . Representative structures for NC = 1 , 2 , 4 , 5 , 6 are shown in S1 Fig of Supporting Information ( SI ) . For NC = 1 , a disordered monomer is observed ( S1A Fig ) . For NC = 2 , two separated disordered monomers are seen in most simulations and partially ordered dimers are rarely observed ( S1B Fig ) . For NC = 4 , a tetramer is easily formed but β-helix structures are observed in most simulations ( S1C Fig ) , which means oligomerization occurs but structural conversion toward U-shape conformation is not accessible yet . For NC = 5 , we observe both β-helix and U-shape conformations ( S1D Fig ) . For NC = 6 , we observe U-shape conformations more frequently ( S1E Fig ) . Hence NC = 5 or NC = 6 can be considered to be the critical nucleus size for conformational conversion for Aβ17–42 peptide under present simulation conditions . For NC = 8 , we also observe U-shape conformations , but for NC = 10 and 12 , we observe partial U-shape conformations which means a longer simulation time or more thermal fluctuations are needed to convert toward highly ordered protofilaments . Hence NC = 8 is the best system size for studying both structural conversion from disordered structures and growth mechanisms toward protofilaments within present accessible simulations . The detailed results on the various structures and analysis of the system size dependence of Aβ17–42 peptides will be presented in a future paper . DMD/ PRIME20 simulations were performed on an 8-chain system of Aβ17–42 peptides starting from a random configuration . The original PRIME20 force field[38 , 40] was augmented to include the parallel preference constraints for hydrogen bond angles , an enhanced salt-bridge interaction ( εKE = 0 . 4εHB ) between K28 and D23 residues where εHB is the hydrogen bonding energy between NH and C = O , and double well potentials for all of the side-chain pair interactions . These modifications to PRIME20 significantly reduce the complexity associated with sampling the energy landscape , which contains a variety of polymorphic conformations as has been shown in computational studies and in experiments . [4–6 , 19 , 41–43] We simulated at reduced temperatures , T* = ( kBT/εHB ) in the range from 0 . 19 to 0 . 205 , which is near the “fibrillization temperature” , the temperature above which fibrils cease to form spontaneously; above this temperature the peptides equilibrate as random monomers and small disordered oligomers without β-sheet content . Each simulation is started at high temperature T* = 0 . 5 with different random seeds from eight separated denatured monomers without any inter-peptide contacts in a periodic box ( L3 = ( 160Å ) 3 corresponding to 1 . 7mM ) to ensure a randomly disordered initial configuration . The system is slowly cooled from T* = 0 . 50 to final temperature over the first 8 billion collisions ( t* = 788 ) and thereafter remains at constant reduced temperature . Fig 1A–1C shows the total interaction energy versus time for ten long ( 668 billion collisions or t* = t/σ ( kBT/m ) 1/2 ≈ 61 , 000 ) independent simulations at T* = 0 . 20 , where σ and m are the united N-H sphere diameter and mass respectively . The 3rd ( green ) , 5th ( red ) and 10th ( blue ) trajectories have the lowest total interaction energy and hence are most thermodynamically stable according to the PRIME20 force field . Fig 2 shows these three configurations that are nicely-formed 8-chain fibrillar ( protofilament ) structures . The other seven structures are shown in S2 Fig; they are all partially ordered with high β-sheet content . Interestingly the structures are quite varied even though the same simulation conditions were applied in each run . The structure in Fig 2A and 2B ( from 3rd run ) shows all eight peptides adopting a bent shape with in-register β-sheets but the loop has a triangular shape rather than a U-shape . Two of the chains ( silver and gray ) in Fig 2A are antiparallel to the other six chains , which means that the turning points between the two β-sheets are mismatched . S3A and S3B Fig shows the location of the glycines on these structures . While the only glycine residue participating in the U-turn in S3C–S3F Fig is G25 ( blue ) , both G25 ( blue ) and G33 ( red ) participate in the U-turn in S3A and S3B Fig; they facilitate the formation of the two turns that appear on the anti-parallel β-sheets ( cyan and gray in S3A and S3B Fig ) , encouraging the protofilament to form a triangular shape . This structure is very similar to a structure suggested by the Wetzel group in their early studies of Aβ with the slight differences in the location of the turns . [44 , 45] The structures resulting from the fifth and the tenth runs shown in Fig 2C–2F are highly organized U-shape β-sheet structures that have hydrogen bonds between neighboring chains in adjacent strands within the sheet . This structure has the U-shape characteristic of those found via experiment by Lührs et al and Petkova et al . [20 , 21] Fig 3A–3C shows ribbon and ball-and-stick snapshots of the structures ( including the positions of the side chains for each of the eight chains in the structure ) for the 10th run , enabling the visualization of the turn region , the salt-bridge interaction , and the residues which interact hydrophobically inside the U-shape β-sheet . Fig 3D–3K show the positions and identities of the side chains in each of the eight chains from Fig 3A–3C . Although the D23 ( red ) and K28 ( cyan ) spheres in Fig 3D–3H are inside the U-shape β-sheet , the K28 ( cyan ) spheres in Fig 3I and 3J are outside , indicating that salt-bridges are formed for the first five chains but absent for the last three chains . The structure for the 5th run in Fig 2C and 2D has fewer salt-bridges; there are only two D23 and K28 pairs ( silver and orange chains ) inside the U-shape β-sheet as shown in S4 Fig . Hence we see the D23-K28 salt bridge is not easily found kinetically in our simulations even though the D23-K28 pair interaction is enhanced . This is not surprising in the light of the observation of weaker D23-K28 coupling in quiescent conditioned fibrils than in agitated conditioned fibrils ( presuming here that agitation enhances the approach to fibrillization ) . [43] For the 10th run , five chains in Fig 3D–3H show hydrophobic interaction between F19 ( purple ) , I32 ( green ) , L34 ( pink ) residues inside the U which is consistent with the Tycko group experimental model[20] and other recent experiments by Smith and coworkers[8] . The structure of the turn region from V24 to N27 for the highly fibrillized chains in Fig 3D–3H and the pattern of the hydrophobic side chains on the inside of the U-shape β-sheet are consistent with the Tycko model[20] and the Ma-Nussinov model[46] , but is slightly different from the Lührs model[21] , which has the turn running from S26 to A30 . An unexpected feature in the Fig 2C–2F structure is the existence of a second turn near the C-terminal; this has been suggested in some experiments as being characteristic of toxic oligomers and has been observed in simulations . [8 , 14 , 17 , 31] This tendency to turn is likely enhanced in our simulations due to our omission of the N-terminal , Aβ1–16 , which frees up the hydrophobic L17 residue to interact with C-terminal I41 and A42 residues . We traced out the time evolution of the structure for the best organized fibrils—the 5th run and the 10th run structures in Fig 2C–2F . The trajectories are presented as movie files in S1 and S2 Videos . Fig 4 shows nine snapshots from the 5th run whose final structure is shown in Fig 2C–2D taken at t* = ( A ) 5 , ( B ) 1244 , ( C ) 2608 , ( D ) 3656 , ( E ) 4233 , ( F ) 5442 , ( G ) 6086 , ( H ) 10454 , ( I ) 11063 after slowly cooling a configuration of random coils from T* = 0 . 50 to T* = 0 . 20 over the course of the first 8 billion collisions ( t* = 788 ) . At first we just observe random coil structures ( Fig 4A ) . Early snapshots including those in the slow cooling stage are shown in S5A–S5B Fig . Small disordered oligomers then start to form ( Fig 4B ) and these merge with monomers into one large oligomer by t* = 2608 ( Fig 4C ) which has some β-sheet character but no U-shape β-sheets . By t* = 3656 , two peptides ( red & blue chains ) have re-arranged to form U-shape β-sheets ( Fig 4D ) and by t* = 4233 the tan chain joins the U-shape and the C-terminal residues curve inward ( Fig 4E ) . The gray , yellow and silver chains sequentially join the U-shape at t* = 4233 , 5442 and 6086 ( Fig 4F and 4G and 4H ) . Finally the orange chain joins the U-shape and we observe seven peptides forming fibril-like structure by t* = 11063 ( Fig 4I ) . The green chain remains flexible till the end of our simulations . Thus it is apparent that the small U-shape β-sheets in Fig 4D serve as a nucleus driving further fibril growth by conformationally changing the attached monomers on a pre-existing proto-filament . This one-by-one structural conversion process is relatively fast once the U-shape nucleus forms . Fig 5 shows nine snapshots for the 10th run whose final structure is shown in Fig 2E and 2F taken at t* = ( A ) 5 , ( B ) 2772 , ( C ) 6623 , ( D ) 19100 , ( E ) 20142 , ( F ) 23060 , ( G ) 26371 , ( H ) 28120 , ( I ) 34039 . The starting configuration ( Fig 5A ) is a random distribution of random coils . Early snapshots are shown in S5C–S5E Fig . Small disordered oligomers are observed ( S5E Fig ) . By t* = 2772 one large disordered oligomer has formed ( Fig 5B ) . By t* = 6623 , two peptides ( yellow & tan chains ) form in-register β-sheets with a partially-attached orange chain ( Fig 5C ) . Interestingly , the C-terminal residues curve outward , forming “S-shape” β-sheets . This oligomer with its partial S-shape is meta-stable and remains for a very long time till t* = 19100 ( Fig 5D ) . The orange and gray chains join the S-shape sequentially at t* = 20142 ( Fig 5E ) and t* = 23060 ( Fig 5F ) , respectively . Thereafter the C-terminal residues begin to change and curve inward forming a U-shape at t* = 26371 ( Fig 5G ) . The silver and blue chains join the U-shape by t* = 28120 ( Fig 5H ) . Finally we observe a nice protofilament-like fibril at t* = 34039 ( Fig 5I ) . The evolution of this fibrillar structure from the disordered oligomer that precedes it is especially interesting . Fig 1C ( blue line ) confirms that this meta-stable oligomer undergoes structural conversion between t* = 26 , 000 and 34 , 000 where the potential energy undergoes a relatively rapid decrease after having been constant for a long time ( t* = 10 , 000 and 24 , 000 ) . The rapid change in total interaction energy in Fig 1C corresponds to the structural conversion of C-terminal residues from S-shape to U-shape and consecutive monomeric conversion upon joining the U-shape . Having identified the equilibrated U-shape protofilament together with the long-lived meta-stable oligomer with S-shape conformations , we estimate the relative stability of the various structures that occur in the simulations . We introduce Pmax ( max population ) which is defined as the population of configurations , within each Δt* = 5 , 000 interval along the trajectory , whose total interaction energy is between Emax—4εHB and Emax + 4εHB , where Emax is the total interaction energy of a configuration with the maximum population . Hence Pmax tells us how often the most populated structures emerged in a given time interval Δt* . It becomes a free-energy-like quantity if we transform it by—kBT log ( Pmax ) . Plots of Pmax versus reduced time are shown in Fig 1D . For the 3rd ( green circle ) and the 5th ( red square ) trajectories , we observe the monotonic increase and saturation in Pmax as the nice fibril structures are reached . However the 10th trajectory ( blue diamond ) shows a sub-maximal peak in the population between t* = 7 , 500 and 17 , 500 which indicates the emergence of meta-stable structures with S-shape conformations in Fig 5C ( t* = 6623 ) and Fig 5D ( t* = 19100 ) . Rapid change in Pmax is also observed between t* = 27 , 500 and t* = 32 , 500 . The changes in this free-energy-like quantity are one of the indications for the meta-stability of the oligomers with S-shape conformations . The fact that the oligomer with S-shape chains has such a long lifetime is consistent with the experimental observation that designing in such a conformation via mutation stabilizes Aβ oligomers . [17] The outward turn of the C-terminal residues was also suggested by Smith and coworkers as being characteristic of the toxic Aβ42 oligomers that they observe at low temperature , which converts to U-shape fibrils by increasing temperature . [8] Here we demonstrate at the molecular level that the S shape undergoes a structural conversion to a U shape and hence a nice fibril structure . One difference between our results and the experimental findings of Smith and coworkers is that their S-shape monomer exists within a disc-shape pentamer , not a partial β-sheet structure as we observe here . One possibility is that the fibrillation pathway that Smith and coworkers observed might have included formation of the S-shape β-sheet as an intermediate step between the disc-shape pentamer and the protofilament . This seems to be plausible to us given that they had to raise the temperature to get to the protofilament state and this could have imparted sufficient kinetic energy to transform the disc-shape pentamer to the lower energy S-shape β-sheet . We analyzed our data to further explore the possible meta-stability of intermediate oligomers with S-shape conformations by performing all-atom simulations for these configurations and final fibrillar structures . All-atom PDBs were generated based on snapshots of Fig 4C–4I and Fig 5C–5I and all-atom molecular dynamics with AMBER/ff99SB force field and explicit solvent TIP3P water were performed for 10ns at 298K . Twelve independent simulations for each initial structure among the fourteen different snapshots were run with different initial random seeds to measure observables related to stability such as system energy , binding energy ( intermolecular energy ) , and RMSF ( backbone atomic positional fluctuations ) , which are shown in S6 Fig . While the highly ordered structures ( G , H , I ) for the 5th run have lower energy ( S6A Fig ) , the meta-stable intermediates ( C , D ) and final fibrillar structure ( I ) for the 10th run show similar energy levels ( S6B Fig ) . However the highly ordered structures ( I ) have strong binding energy ( S6C–S6D Fig ) for both runs . Interestingly snapshot E ( t* = 20142 ) from the 10th run , which is just before rapid structural conversion and is corresponding to high energy in Fig 1C and less β-strand content ( Fig 5 caption ) , also has high system energy ( S6B Fig ) , high binding energy ( S6D Fig ) and high RMSF ( S6F Fig ) . This observation indicates that the intermediates ( C and D ) with S-chain conformations are meta-stable , consistent with our explanations for the results Figs 1D and 5 . We used RMSF instead of RMSD since the latter shows too much fluctuation in our atomistic simulations , losing discrimination power . The RMSF sheds light on the meta-stability of snapshots C , D and the low stability of E for the 10th run . Hence we reconfirmed our observation of the meta-stability of intermediate oligomers with S-shape conformations by all-atom molecular dynamics . Another interesting observation in our PRIME20/DMD simulations has to do with the pattern of side-chain interactions between one side of the U and the other . Fig 6 shows four snapshots for the 5th run at four different times ( t* = 34032 , 34059 , 34096 , 34310 ) selected from the time period t* = 33500 to 42736 shown in S3 Video . In Fig 6A and 6C , the N-terminal β-sheet and the C-terminal β-sheet are slightly tilted with respect to each other so as to form inter-molecular side-chain interactions , similar to those observed in experiments . [20 , 21] However in Fig 6B and 6D , the N-terminal β-sheet and the C-terminal β-sheet are not tilted , and form intra-molecular side-chain interactions . The surprising result here is that inter-chain interactions which stabilize the U-shape β-sheet structure constantly change to intra-chain interactions and vice versa over the course of our simulations as is clearly shown in S3 Video . It is interesting that we see two polymorphic conformations simultaneously in fibril formation . This observation may be due to finite size effects associated with having only 8 chains; in a larger system one fixed conformation may be stabilized as the structure grows . Although we focused our discussion of structural conversion on two trajectories at a specific temperature T* = 0 . 20 among many independent runs , we actually performed DMD simulations at six other temperatures T* = 0 . 19 , 0 . 195 , 0 . 198 , 0 . 20 , 0 . 202 and 0 . 205 . S7 and S8 Figs show final structures at T* = 0 . 198 and 0 . 202 respectively . We observed a triangular shape in S7G Fig , U-shape in S8H Fig , S-shape in S8F Fig , and many partial U-shape or β–helix structures . However we observe only random monomers and disordered small oligomers at T* = 0 . 205 , which means that it is above the fibrillation temperature , and only partially ordered structures at T* = 0 . 19 and 0 . 195 . The structural and temporal features of our simulations were analyzed using two different methods . In the first method , the β-strand content for each residue was calculated during three time windows in the simulation . The analysis was applied to the 3rd , 5th and 10th runs , all of which led to fibril-like structures . To identify the β-strand content for each residue , we calculated the dihedral angles ( φ and ψ ) and used the STRIDE program[47] which identifies the secondary structure of each residue over the simulation trajectory . Trajectories were collected and averaged over three time windows: very early stage ( t* = 2153~6727 ) in Fig 7A , middle stage ( 19648~24266 ) in Fig 7B and late stage ( 38120~42736 ) stages in Fig 7C , to gauge how the β-strands develop over time . The two hydrophobic regions ( V18-V24 and A30-V36 ) develop β-strands starting from the earliest stage and continue to have high β-strand content through the late stage . During the middle stage where the 10th run is still a meta-stable oligomer with S-shape chains , the β–strand content for the 10th run ( blue line ) is not fully developed yet . At late stage , the 5th and 10th runs clearly show turn regions ( V24-S26 and G37-G38 ) . The 3rd run shows another turn region near G33 which is what makes for the triangular shape as shown in Fig 2A and 2B . Those portions of the chain that easily transform to β-strands play an important role in fibril formation because they readily form β-sheets whose protruding side-chains provide opportunities for stacking interactions leading to a cross-β spine . In the second analysis method , the solvent accessible surface area ( SASA ) of each residue is measured during the late time window ( t* = 38120~42736 ) shown in Fig 7D . These values are estimated using the STRIDE program which we have further tailored to accommodate the four-sphere PRIME geometry . The zigzag pattern that is observed along the sequence is an expected consequence of the alternating side-chain patterns on the β-strands . Roughly speaking , the N-termini in the 5th and 10th runs are more exposed than the C-terminal , which is consistent with other coarse-grained simulations . [34] This indicates that the hydrophobic residues in C-terminal drive themselves to be buried as is known to occur in oligomers or fibril structures . [4 , 8] As discussed in the previous section , the 10th run ( blue line ) shows results that are the most consistent with experiments , having small SASA values at D23 , K28 , F19 , I32 and L34 , which are characterized by an internal salt-bridge and hydrophobic interactions . [8 , 20] To check if our SASA results were an artifact of using the PRIME20 reduced 4-sphere protein representation , we generated all-atom based PDB files by applying the MODELLER program to our PRIME model trajectories and ran the STRIDE program again . The SASA pattern for the all atom structures was similar to the 4-sphere result . Analysis of the data presented in Fig 7 and that of the snapshots presented in Figs 4 and 5 gives us a molecular-level picture of how disordered oligomers undergo structural conversion towards fibril structure . At an early stage , disordered oligomers are formed with exposed N-terminal and buried C-terminal sites , probably due to having more hydrophobic residues at the C-terminal . The two aggregation-prone regions ( V18-V24 , and A30-V36 ) , which are strongly hydrophobic , start to develop β-strands and form β-sheets early . These β-sheets then rearrange themselves so that a U-shape nucleus or a meta-stable oligomer with S-shape can form . The meta-stable oligomer eventually changes to a U-shape after which structural conversion to a protofilament structure occurs through monomeric addition templated by the U-shape nucleus . The preceding results describing energies , snapshots of structures , kinetic pathways and structural analyses are from 10 independent runs of 668 billion collisions ( t*≈ 61 , 000 ) . We performed another 100 independent runs more of 468 billion collisions ( t*≈ 43 , 000 ) to get better statistics on the various conformations and to validate the generality of our suggested kinetic pathways . Besides the disordered , partially-ordered oligomers with U-shape or S-shape and mixed fibrillar structures containing more than two different fibril-like conformations as shown in S8C and S8H Fig , we observe nine fibrillar structures ( S9 Fig ) , four fibril-like structures with the full S-shape conformations ( S10 Fig ) and three triangular fibril-like structures ( S11 Fig ) . Among the nine fibrillar structures , we observe three trajectories along which there is a structural conversion from S-shape to U-shape ( S12 Fig ) and six trajectories in which there is U-shape nucleus formation and monomeric conversion process . We observed fewer ordered fibril structures and even fewer S-shape to U-shape pathways to reach the nice U-shape fibrillar structure than we expected . We see that some S-shape conformations further order without structural conversion to U-shape so that fibril-like structures with full S-shape conformations remain at the end of the simulation ( S10 Fig ) , which means some S-shape conformations might be off-pathway . Actually simple structural conversion from S-shape to U-shape events and formation of partial U-shape are observed more frequently , but not all U-shape nuclei succeed at driving the system to form nice fibrillar structures . Hence many disordered or partially ordered oligomers are trapped and do not undergo further ordering . At present we do not know if this trapping and freezing of disordered structures is intrinsic to protein aggregation processes or is an artifact due to weaknesses of the coarse-grained model , since our force field has less detailed atomic movements and fluctuation preventing us from optimizing the fibril structures in atomic scale . We expect that our model’s ability to model on-pathway processes would be improved if we enhanced the stability of the U-shape nucleus by assigning known atomistic constraints such as the steric zipper interface by atomistic van der Waals interactions . Although we have not achieved unequivocal results for the fibrillar structures and kinetic pathways , we believe that our simulations provide a major leap forward in our ability to simulate the fibrillization process compared to other coarse-grained and all-atom simulations . In summary , by performing 100 independent runs , we reconfirm clearly two distinct kinetic on-pathways toward U-shape protofilaments .
The powerful combination of a four-sphere-per-residue protein model , PRIME20 , and discontinuous molecular dynamics significantly facilitated the tracking of the aggregation process of peptide chains ( Aβ17–42 ) that are longer than the 6 to 10 amino-acid peptides that have been considered usually in the past . Aβ17–42 is a good stand-in for its longer parent proteins Aβ1–42 and 1–40 , because it contains the two hydrophobic stretches that dominate the aggregation and fibrillization of Aβ1–42 as well as the turn region . We consider this to be an important progress in the realistic modeling of Aβ oligomerization and fibrillization . Insights into the kinetic pathways and structural features for toxic oligomers that form spontaneously could be beneficial for the design of antibodies or small molecule inhibitors . Our main focus has been to capture molecular-level insight on how random Aβ17–42 monomers turn into fibrillar structures through structural conversion via oligomers . We observe two different pathways during the structural change from disordered oligomers to ordered protofilament . The first pathway is one-by-one monomeric conversion templated by a U-shape nucleus; this is a fast process once the nucleus is formed . Although the monomeric conversion takes place within the oligomer , the nucleation and monomeric conversion share a common theme with other known mechanisms of fibril formation such as nucleated polymerization and the dock-lock mechanism by monomer addition . [48 , 49] The other pathway goes through a meta-stable oligomer with S-shape conformations due to the turn opportunities presented by the two flexible glycine residues ( G37 , G38 ) ; this adds considerably to the time it takes to convert to a U-shape nucleus and hence a protofilament . Experimental studies showed that designing a turn into the C-terminal region by mutation enhances the stability of the more toxic oligomers which are off-pathway species that are indeed detected in experiments . [14 , 17] Although the disc-shape pentamers containing S-shape monomers ( from L17 to A42 ) observed by Smith and coworkers were considered by them to be on-pathway , the fact that their pentamer could only be converted to a fibril by raising the temperature suggests us that it was off-pathway . [8] Here we observe a meta-stable oligomer , which is very long lived but nevertheless on-pathway without changing temperature and which shares the S-shape that may be a characteristic of toxic oligomers as experiments suggested . [8] Teplow and coworkers also examined wild type Aβ42 oligomers and found that toxicity peaked at intermediate times . [14] Although they did not mention any evidence or possibility for a C-terminal turn , we speculate that this could be explained by the existence of a meta-stable on-pathway C-terminal turn at intermediate stage which caused toxicity but that this eventually converted to a less-toxic fibrillar structure . The simulation temperatures of T* = 0 . 19 to 0 . 205 , slightly below the fibrillization temperature were chosen to give us the best possible opportunity to watch the peptides evolve toward the lowest energy state . While this does introduce some artificiality by , in effect , smoothing the energy landscape , it has a number of advantages . The high entropic fluctuation that occurs in the reduced temperature range 0 . 19 ≤ T* ≤ 0 . 205 helps disordered oligomers to form in-register parallel β-sheets without getting trapped in meta-stable disordered states , as we have mentioned in a previous paper . Another advantage of simulating at a high temperature just below the transition temperature is that this condition slows down oligomerization as much as possible . It does this by preventing the trapping in large amorphous oligomeric states that usually accompanies rapid hydrophobic collapse . Instead it allows smaller oligomers to form alongside of the free monomers at an early simulation stage , making it easier for the oligomers to convert to ordered structures . A similar retardation of oligomerization could be achieved at lower protein concentration but this would have slowed down our simulations considerably . We do not have a satisfactory way at this time to relate the reduced temperature in our model to the real temperature . This is because the hydrogen bond energy which is used to scale the temperature is a potential of mean force due to the surrounding water rather than a direct interaction . While we were able to construct fibrillized protofilaments successfully with our 8 chain systems after fairly long simulations runs , we also see a number of other very diverse structures including a triangle-shape β-sheets , β-helices and oligomers with β-strands arranged in a disordered fashion . This broad ensemble of structures may be a consequence of having a highly rugged energy landscape , which causes a variety of amorphous aggregates or polymorphic fibril conformations even in experiments , or it may partly depend on our having a coarse-grained force field like PRIME20 where the side-chain sizes and interactions are imprecise and not optimized for zipping up via van der Waals interactions between atoms . Nevertheless the impreciseness of the coarse-grained model is beneficial because it allows us to overcome the large energy barriers to achieving a fibrillar structure within our current capacity of computation .
We employ our new intermediate-resolution force field PRIME20 [38–40] in discontinuous molecular dynamics [50] simulations to study the aggregation of the Aβ17–42 peptide . PRIME20 is an extension of PRIME ( Protein Intermediate-Resolution Model ) and is designed to be applicable to all twenty amino acid residues . [38 , 51–53] The adequacy and efficiency of PRIME20 has been proven by applying to short peptide systems such as Aβ16–22 , fragments of the prion proteins , the designed sequences of Lopez de la Paz et al , and the tau fragment ( VQIVYK ) . [39 , 40 , 54 , 55] Aβ17–42 is modeled using the PRIME20 4-sphere-per-residue representation ( backbone united atoms NH , CαH , and CO , and a single sphere side chain ) . The masses for the backbone united atom spheres are CαH ( 0 . 866 ) , NH ( 0 . 999 ) , CO ( 1 . 863 ) , and for the side-chain united atom spheres are R ( 6 . 728 ) , N ( 3 . 862 ) , D ( 3 . 860 ) , Q ( 4 . 795 ) , E ( 4 . 793 ) , H ( 5 . 394 ) , K ( 4 . 865 ) , S ( 2 . 064 ) , T ( 2 . 997 ) , A ( 1 . 000 ) , I ( 3 . 799 ) , L ( 3 . 799 ) , M ( 4 . 998 ) , F ( 6 . 061 ) , Y ( 7 . 126 ) , V ( 2 . 866 ) in mass units of CH3 ( 15amu = 1 . 0 ) . Each amino acid has a different set of geometric parameters including hard-sphere diameters and pair-interaction ranges which are given in our previous paper . The distances from the side-chain spheres to Cα , NH , and CO united atoms are carefully designed to ensure that all amino acids remain in an L-isomer form during DMD simulations . We presented four Supporting Information Tables . The geometry distances for 20 amino acids are given in S1 Table , the minimum non-bonded distances between a side-chain sphere and other neighboring united backbone spheres called the squeeze parameters are given in S2 Table , the inner well diameters for double well potential are given S3 Table , and the pairwise interactions for double well potential are given in S4 Table . The outer well diameters are not included since they are given in the supplemental table 3 of the reference [38] . In this simulation , we added two biases , parallel preference constraints and an enhanced D23-K28 salt-bridge interaction , because we were aiming to simulate the formation of parallel in-register U-shape protofilament which are experimentally observed for Aβ structure . [20 , 21] In order to reduce the potential complexity among polymorphic backbone orientation within β-sheets[6 , 42] , we use the “parallel preference” set of distance cutoffs for backbone hydrogen bonds to enhance formation of parallel in-register β-sheets . [39] The parallel preference hydrogen-bond distance constraints between i-th donor residue and j-th acceptor residue are: Ni-Cαj ( 5 . 10Å ) , Ni-Nj+1 ( 4 . 54Å ) , Cj-Cαi ( 4 . 96Å ) , Cj-Ci-1 ( 4 . 58Å ) , the same as used on our simulations of the tau fragment [39]; it is slightly changed from the non-biased cutoff distances for directional hydrogen bonds used in our earlier papers: Ni-Cαj ( 5 . 00Å ) , Ni-Nj+1 ( 4 . 74Å ) , Cj-Cαi ( 4 . 86Å ) , Cj-Ci-1 ( 4 . 83Å ) . [40 , 55] These changes are obtained by measuring the distributions for these four distances in 620 NMR PDBs and decomposing them into distributions for parallel and anti-parallel β-sheets . This slight bias in our simulations , only 2–5% variation in cutoff distances , helps encourage the formation of parallel , as opposed to anti-parallel , pairs of β-strands . In our previous simulations of the tau fragment ( VQIVYK ) , the implementation of parallel preference constraints converted the final configuration from β-sheets with a random mixture of parallel and anti-parallel β-strands in the original H-bond constraints to nearly perfect parallel β-sheets as had been seen experimentally . [39] Hence the parallel preference directional hydrogen bonds can assist in promoting in-register β-sheets and suppressing mixed β-sheets with parallel and anti-parallel pairs of strands . This is a reasonable approximation since the difference between the formation energy of a parallel β-sheet and an anti-parallel β-sheet in all atom simulations and coarse-grained models is very small . This slight bias greatly enhances the possibility to form perfect Aβ fibril structures with parallel in-register U-shape β-sheets . Although we obtain nice results for Aβ17–42 peptides and tau fragments ( VQIVYK ) systems , this bias is obviously not applicable to other systems such as Aβ16–22 peptides which are known to form highly anti-parallel β-sheet . [40] If we apply parallel preference constraints to Aβ16–22 peptides , we observe mixed β-sheets having almost half parallel and half anti-parallel pairs of strands . We also use an enhanced salt-bridge interaction between K28 and D23 residues . This salt-bridge is critical to fibril formation in Aβ . Forming and burying the salt-bridge inside a protofilament is believed to generate a high energy barrier and hence is a rate-limiting step . [33 , 56] Thus the enhancing the possibility that a salt bridge will form in experiments , for example by making a Lactam bond between D23 and K28 , can significantly increase the fibrillation rate . [57] It is interesting to note that the salt-bridge is not as highly populated in quiescent experiments as it is in agitated fibrillation experiment . [43] Hence enhancing the possibility of salt bridge formation in simulation is quite reasonable . It helps the disordered oligomers to move easily on the free energy surface toward the expected U-shape protofilament structure; essentially reducing the ruggedness of the energy landscape . We increase the pair interaction value from its original value in PRIME20 of 0 . 136εHB to 0 . 4εHB , which is almost twice the strength of the strong hydrophobic side-chain interaction ( εF-F = 0 . 205εHB ) . For comparison , we present the results from 10 more independent simulations with non-enhanced salt-bridge interactions ( εKD = 0 . 136εHB ) performed for 468 billion collisions . The final structures are shown in S13 Fig . Partial U-shape structures are observed in S13A and S13G Fig but they are mixed fibrillar structures consisting of more than two different fibril-like conformations . S13A Fig contains triangular shape and U-shape conformations . S13G Fig contains two U-shape conformations oriented in different directions . The average numbers of salt-bridge interactions per structure during the last 50 billion collisions are 0 . 13±0 . 24 for intramolecular and 0 . 16±0 . 29 for intermolecular interactions , while simulations with the enhanced salt-bridge interaction give numbers of 1 . 7±1 . 4 for intramolecular and 2 . 2±2 . 0 for intermolecular interactions at the same time and the same temperature . Therefore we can see that the number of salt-bridge interactions is greatly increased when using the enhanced salt-bridge interaction condition . Roughly speaking , S13 Fig looks similar to the other disordered structures , such as S2 Fig , but the nice fibrillar structures are far less accessible since the probability of salt-bridge interactions is very low . While these two biases clearly enhance the likelihood of fibrillization and allow us to skip some of the meta-stable structures along the way , they also could alter the oligomerization mechanism . However we think the important mechanisms or structural insights for oligomerization and fibril formation are being captured . Actually these nice well-organized fibrillar structures could not be obtained spontaneously without those two biases . And if we could not obtain any nice fibrillar structures ( which we know do form ) , we would not know whether our intermediate structures in simulations are real oligomers observable in experiments or artifacts of simulations ( trapped meta-stable oligomers in highly rugged energy landscape seen in the most simulations ) . So the biased simulations may look ad-hoc , but they are necessary if we want to examine spontaneous fibrillation given our present computational limitations . In addition to the above two biases ( compared to the original PRIME20 ) mentioned above , we add another improvement to the PRIME20 model . Instead of the standard single well potential we use double well potentials for every pair interaction between side-chain spheres . In our previous studies [39 , 40] interactions between side-chain spheres were modeled using a single well potential . The well diameters were evaluated from the pairwise distance distributions between the side-chain spheres over 711 PDBs . Each pairwise interaction was included in the pairwise distribution when over half of the distances between heavy atoms on different side-chains were less than 5 . 5Å cut-off ( called 5 . 5Å heavy atom criteria ) , as is explained in detail in the previous work . [38] In this paper we use a double well potential; the previous single well is replaced by two wells—an inner deep well and an outer shallow well . We determine the inner deep well diameter using a 4 . 5Å heavy atom criteria and the outer shallow well diameter using a 5 . 5Å heavy atom criteria . The depth of the inner deep well is taken to be 1 . 3ε ( ij ) and the depth of the outer shallow well depth is taken to be 0 . 7ε ( ij ) where ε ( ij ) are the well depths ( pair interaction strength ) from the original 19 parameter PRIME20 force field [38] , which were estimated by the perceptron learning algorithm . We observed very similar results in Aβ16–22 simulations for the single well potentials and the double well potentials but felt that a more detailed force field might be necessary for longer chain systems like the Aβ17–42 peptide . | Understanding the mechanisms of protein folding and aggregation is of fundamental importance in elucidating the biological function of proteins and their complex . Many advances have been made in our ability to describe protein folding based both on ideas from biophysics and improvements in supercomputing power , yet realistic simulations of the entire kinetic process of protein aggregation including fibril formation still remain challenging tasks in biophysics and computational biology . This work describes a breakthrough in our ability to simulate the aggregation of proteins on a molecular level and the emergence of the toxic species responsible for the cause of neuro-degenerative diseases such as Alzheimer’s disease . Based on this work , one can now trace the entire aggregation process starting from disordered monomers to meta-stable oligomers to protofilament and then amyloid fibril . This is a significant advance over the current state of the art in both biophysics and computational biology in uncovering the fundamental mechanisms behind the amyloid fibril formation for aggregation-prone proteins . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2015 | Structural Conversion of Aβ17–42 Peptides from Disordered Oligomers to U-Shape Protofilaments via Multiple Kinetic Pathways |
We investigated changes in the spatial distribution of schistosomiasis in Mali following a decade of donor-funded control and a further 12 years without control . National pre-intervention cross-sectional schistosomiasis surveys were conducted in Mali in 1984–1989 ( in communities ) and again in 2004–2006 ( in schools ) . Bayesian geostatistical models were built separately for each time period and on the datasets combined across time periods . In the former , data from one period were used to predict prevalence of schistosome infections for the other period , and in the latter , the models were used to determine whether spatial autocorrelation and covariate effects were consistent across periods . Schistosoma haematobium prevalence was 25 . 7% in 1984–1989 and 38 . 3% in 2004–2006; S . mansoni prevalence was 7 . 4% in 1984–1989 and 6 . 7% in 2004–2006 ( note the models showed no significant difference in mean prevalence of either infection between time periods ) . Prevalence of both infections showed a focal spatial pattern and negative associations with distance from perennial waterbodies , which was consistent across time periods . Spatial models developed using 1984–1989 data were able to predict the distributions of both schistosome species in 2004–2006 ( area under the receiver operating characteristic curve was typically >0 . 7 ) and vice versa . A decade after the apparently successful conclusion of a donor-funded schistosomiasis control programme from 1982–1992 , national prevalence of schistosomiasis had rebounded to pre-intervention levels . Clusters of schistosome infections occurred in generally the same areas accross time periods , although the precise locations varied . To achieve long-term control , it is essential to plan for sustainability of ongoing interventions , including stengthening endemic country health systems .
Mali was one of the first countries in sub-Saharan Africa to initiate a national schistosomiasis control programme . Control efforts started regionally in 1978 in Dogon Country ( region of Mopti ) after the construction of small dams for growing vegetables , and became a national programme in 1982 . During the first 10 years , the programme was run by the Malian Ministry of Health in partnership with the World Health Organization and the German Technical Cooperation ( Deutsche Gesellschaft für Technische Zusammenarbeit , GTZ ) [1] . Parasitological surveys followed by mass treatment of the entire population in target areas were conducted by a central team from Bamako . Additionally , in selected areas , identification of infected individuals and case treatment was implemented . The control programme was intensively focused on two major endemic areas: Office du Niger ( irrigation area ) and in the area around Bandiagara in the Plateau Dogon ( small dams area ) . Initial evaluation ( 1–3 years after intervention ) showed reductions in both prevalence of infection and prevalence of heavy-intensity infections ( >50 eggs/10 ml urine for Schistosoma haematobium and >100 eggs/gram stool for S . mansoni ) . For S . haematobium , prevalence of infection was reduced from 58 . 9 to 26 . 8% and that of heavy infections from 18 . 4 to 3 . 8% , whereas for S . mansoni , prevalence of infection was only reduced from 49 . 0 to 48 . 1% and that of heavy infections from 10 . 6 to 8 . 9% [2] . Estimated impact of the intervention varied by intervention approach , ecological zone and time to follow up ( 1–3 years ) . GTZ support for the programme ceased in 1992 , with the government taking over financial responsibility . However , lack of resources led to control activities being considerably reduced and the implications of this for infection levels were not assessed in the immediate post treatment period . From 1998 , a new , decentralised control programme was approved by the Ministry of Health but , due to lacking continuous financial support from the government , many planned activities were not implemented . In 2004 , a new initiative to recommence national control activities was established with support from the Schistosomiasis Control Initiative ( SCI; http://www . schisto . org ) . Again the main intervention strategy was mass treatment with praziquantel , with a particular focus on treating school-age children [3] . The potential of using risk mapping to describe the spatial patterns of infections is now well-established , and has been demonstrated for a range of diseases including malaria [4] , [5] , schistosomiasis [6] , Loa loa filariasis [7] and lymphatic filariasis [8] . The combination of geographical information systems ( GIS ) , remote sensing and geostatistics has led to an increase in the understanding of the spatial epidemiology of infectious diseases , the prediction of occurrence , and the targeting of large-scale control programmes . For example , Bayesian geostatistical modelling is being used increasingly to predict spatial patterns of human schistosomiasis in Africa [9] , [10] , [11] , [12] , [13] . Much of this work to date has used data from a single geographical area at a single point in time to develop predictions for similar locations . Preliminary work has investigated the spatial extent to which risk models can be reliably extrapolated [14] but it remains unclear the extent to which models based on data from one area can be extrapolated temporally . This is particularly important in determining whether control programmes can be spatially targeted on the basis of historic data , or whether it is necessary to conduct new surveys ( which are expensive and time consuming ) to define the spatial distribution of disease . This issue is especially relevant in the context of the dramatic up-scaling of disease control interventions and the need for survey data to target suites of alternative interventions . In this paper , we use unique data on schistosome infections , available from two nationwide surveys conducted in Mali , the first undertaken during the 1980s prior to the implementation of the GTZ-supported national control programme and the second between 2004–2006 , 12 years after this programme had ceased and prior to implementation of the SCI-supported programme . We aim to determine whether the overall prevalence and spatial distribution of schistosomiasis in Mali is different in 2004–2006 compared to the 1980s and to determine whether the spatial distribution , including covariate relationships with environmental variables and parameters that describe the spatial dependence structure ( i . e . clustering ) , have changed in Mali over the last two decades .
A nationwide survey was carried out between May 1984 and May 1989 prior to implementation of the GTZ-supported programme ( see Traoré et al . [15] for further details ) . In brief , villages were selected using a three-stage sampling approach: two to three districts were randomly selected in each province , then three to five arrondissements ( sub-districts ) were randomly selected in each district , and five villages were randomly selected in each arrondissement . In each village , individuals were randomly selected to provide urine ( 200 individuals ) and stool samples ( 150 individuals ) . For each individual , a single urine slide ( for diagnosis of S . haematobium infection by filtration method ) , and two Kato-Katz slides prepared from a single faecal sample ( for diagnosis of S . mansoni ) were examined microscopically using standard methods . While egg counts were done , only data on the number tested and proportion positive ( i . e . with one or more eggs ) in a given location were available for the current study . Longitude and latitude co-ordinates of each village were identified during the current study from a national village GIS database ( http://www . who . int/health_mapping/tools/healthmapper/en/ ) ; of the 323 villages surveyed we were able to geo-reference 300 villages , from which data were available on 52 , 104 individuals . A more recent nationwide survey was conducted in 194 schools ( including 15 , 051 school-aged children ) between December 2004 and May 2006 . Ethical approval for these surveys was obtained from St . Mary's Hospital Research Ethics Committee UK and the National Public Health Research Institute's ( INRSP ) scientific committee in Mali . All data collection activities were carefully explained to , and oral consent was obtained from traditional authorities in the village ( the village head and the elders ) , the schoolmaster , the representative of the pupils' parents and the local health authorities . Child participants were given an explanation of the data collection activities and were free not to participate if they so chose . Written consent was not obtained and oral consent was not specifically documented because the survey was considered by the UK and Malian ethical committees as part of the monitoring and evaluation of routine health activities carried out by the Malian Ministry of Health's national schistosomiasis control programme . Survey protocols ( available on request ) instructed survey teams to select 30 boys and 30 girls per school using systematic random sampling . Schools were selected to maximise geographical coverage of the study area; all parts of Mali excluding the northern desert and far eastern regions , where transmission is known not to occur [16] , were included in the survey . This was done in a GIS ( ArcView 9 . 2 , ESRI , Redlands , CA ) by overlaying a 1 decimal degree squared grid over the country . The locations of communities in Mali were obtained from the aforementioned national village database . Communities were selected using simple random selection from each grid cell and , if more than one school was present in a town or village , a school was sampled on arrival using simple random selection . The selected children were assembled and asked to provide a urine and stool sample . For each child , a single urine slide and two Kato-Katz slides prepared from a single faecal sample were examined microscopically as described above . Numbers of eggs of S . haematobium and S . mansoni in each child's sample were recorded on paper forms , in addition to the geographic location of the school ( determined using a hand-held global positioning system ) . All school and individual data were transferred to a Microsoft Access database . For the current study , numbers tested and positive ( defined as one or more eggs for each species of schistosome ) were calculated for each survey location . School or community-level raw prevalence was then plotted in the GIS . Electronic data for land surface temperature ( LST ) and normalised difference vegetation index ( NDVI ) were obtained from the National Oceanographic and Atmospheric Administration's ( NOAA ) Advanced Very High Radiometer ( AVHRR; see Hay et al . [17] for details on these datasets ) and the location of large perennial waterbodies was obtained from the Food and Agriculture Organization of the United Nations ( FAO-GIS ) . Values for LST , NDVI and distance to the nearest perennial water body ( DPWB ) were calculated in the GIS for each survey location . Multivariable logistic regression models were developed for each species of schistosome and each of the two survey periods in a frequentist statistical software package ( Stata version 10 . 1 , Stata corporation , College Station , TX ) . Prelimary results were similar for each species of schistosome and each study period . A quadratic association between LST and prevalence was assessed and was found to be significant and DPWB was also significantly and negatively associated with prevalence . NDVI was not found to be significantly associated with prevalence in the preliminary multivariable models and was excluded from further analysis . Therefore , it was decided to enter LST ( in quadratic form ) and DPWB as covariates into the final spatial models . Bayesian geostatistical models , developed in WinBUGS 1 . 4 ( Medical Research Council , Cambridge , UK and Imperial College London , UK ) , were identically structured for each species of schistosome and each study period . Statistical notation is presented in Text S1 . Three chains of the models were run consecutively . A burn-in of 1 , 000 iterations was allowed , followed by 10 , 000 iterations where values for the intercept and coefficients were stored . Diagnostic tests for convergence of the stored variables were undertaken , including visual examination of history and density plots of the three chains . Convergence was successfully achieved after 10 , 000 iterations in each model and the posterior distributions of model parameters were combined across the three chains and summarized using descriptive statistics . Geostatistical prediction across Mali was done in WinBUGS using the spatial . unipred command [18] . To compare predictions accross time periods , the 1984–1989 model was used to predict infection prevalence at the 2004–2006 survey locations and vice versa , for both S . haematobium and S . mansoni . The predicted prevalence was compared to the observed prevalence , dichotomised at 50 , 20 , 10 and 0% ( to assess predictive performance relative to different observed prevalence thresholds , including the World Health Organisation-recommended thresholds for annual and biannual mass chemotherapy of 50% and 10% respectively ) . The diagnostic test evaluation statistic , area under the curve ( AUC ) of the receiver operating characteristic , was used for the comparison . An AUC value of >0 . 7 was taken to indicate acceptable predictive performance [19] . A stationary model is one where the parameters that define the spatial dependence structure are the same for the two time periods and a non-stationary model is one where the parameters are different ( note we refer to stationarity across time periods , not different parts of the study area ) . Models were developed using the combined datasets , including with different intercepts for each time period and: 1 ) different coefficients , spatial dependence parameters and random effects ( i . e . assuming separate sub-models for each time period ) ; 2 ) the same coefficients but different spatial dependence parameters and random effects ( i . e . allowing the sub-models to have common covariate effects ) ; 3 ) the same coefficients and spatial dependence parameters but different random effects ( i . e . allowing common covariate effects and stationary spatial dependence structures , but separate predicted risk surfaces ) ; and 4 ) the same coefficients , spatial dependence parameters and random effects ( i . e . a single model giving an overall predicted risk surface across the two time periods ) . Models 1 and 2 were non-stationary models and models 3 and 4 were stationary models . Statistical notation is presented in Text S2 . The best-fitting model ( of 1–4 ) was selected using the deviance information criterion ( DIC ) . An additional comparison of the spatial distribution of schistosomiasis accross time periods was done by subtracting predicted prevalence from the best-fitting S . haematobium and S . mansoni models in 2004–2006 from predicted prevalence in 1984–1989 .
The Bayesian geostatistical models for each time period are presented in Table 1 . Note that the odds ratios are on the same scale for each variable , which were standardised to have a mean of zero and standard deviation of one . DPWB was significantly and negatively associated with each outcome , with very similar odds ratios for all four models . The quadratic term for LST was not significant in any of the models , where significance is defined by a 95% posterior interval that excludes one ( note , outputs of Bayesian models are distributions termed posterior distributions that describe the probability associated with each of a range of plausible values for the variable being estimated ) . Phi ( ) , which indicates the rate of decay of spatial correlation ( with a bigger indicative of smaller clusters ) varied from 1 . 68 to 9 . 02 for S . haematobium and S . mansoni in 2004–2006 . S . haematobium clusters were , therefore , generally larger than S . mansoni clusters . For both types of infection , the sill was lower in 1984–1989 than in 2004–2006 , indicating a stronger tendency towards spatial clustering in the latter time period . Models developed on 1984–1989 and 2004–2006 data were generally able to discriminate infection prevalence for the other dataset to an acceptable level ( Table 2 ) . For S . haematobium , models tended to perform better when discriminating at lower prevalence thresholds ( present versus absent , <10% versus ≥10% ) , while for S . mansoni , models tended to perform better at high prevalence thresholds ( <50% versus ≥50% ) . The only comparison that gave an AUC <0 . 7 , the acceptability criterion , was for prediction of S . mansoni presence ( prevalence >0% ) in 1984–1989 . The deviance information criterion for models 1–4 , for S . haematobium and S . mansoni , are presented in Table 3 . For S . haematobium , the model with the lowest DIC ( indicating the model with the best compromise between model fit and parsimony ) was model 2 ( Table 4 ) , with common covariate effects but a non-stationary spatial dependence structure across time periods . For S . mansoni , the model with the lowest DIC was model 3 ( Table 5 ) , with common covariate effects and a stationary spatial dependence structure across time periods . As for the period-specific models , prevalence of both infections was negatively associated with increasing DPWB and was not significantly associated with LST . In the non-stationary model for S . haematobium ( Table 4 ) , the sill was lower for 1984–1989 than for 2004–2006 , again indicating greater clustering in the latter time period , and the rates of decay of spatial correlation , phi , were similar for the two time periods . The overlapping 95% posterior interval limits for the 1984–1989 and 2004–2006 intercepts in both the S . haematobium and S . mansoni models suggest that overall ( mean ) prevalence was not significantly different across time periods for either species of schistosome . Spatial predictions ( showing the mean of the posterior distributions for predicted prevalence ) based on the best model for each type of schistosome infection are presented in Figures 3 and 4 . In 2004–2006 , S . haematobium occurred in large clusters in a mid-latitudinal band from western to central Mali and low predicted prevalence was apparent in both southern and northern latitudinal bands ( Figure 3B ) . In 1984–1989 ( Figure 3A ) , the pattern was similar but more fragmented . The prediction maps for S . mansoni ( Figure 4 ) were remarkably similar to each other , with infection limited to small high-prevalence clusters in central and southwestern regions , althought the clusters occurred in slightly different locations . Comparative maps show predicted prevalence in 1984–1989 subtracted from predicted prevalence in 2004–2006 , using the best-fitting models ( Figure 5 ) . Most areas of both maps had an estimated difference of <10% in predicted prevalence between the two periods . However , there were some areas on both maps that had an estimated difference of >20% in predicted prevalence; for S . haematobium , higher predicted prevalence in 2004–2006 mainly occurred in central and western regions and lower predicted prevalence was mainly along the Niger river and in southwestern regions; for S . mansoni , differences coincided with the locations of the small high-prevalence foci in central and southwestern regions because the precise location of these clusters varied somewhat between the study periods .
Despite differences in survey design and study population between the time periods , this study demonstrated remarkable similarities in the spatial distribution of prevalence of infection with S . haematobium and S . mansoni in Mali between 1984–1989 and 2004–2006 . While clusters of infection occurred in generally the same area of the country , the precise location did vary slightly between the two time periods . Nonetheless , our analysis of predictive performance of models across time periods suggests it may be possible , in the first instance , to use historical data to predict contemporary distributions at national scales ( assuming a stable climate and an absence of new , large water resource development projects , both of which should be investigated ) . It is perhaps not surprising that the statistical associations between prevalence and DPWB did not vary between the study periods as the essential biology of schistosome infections is unlikely to have changed , but it is interesting that the spatial dependence structure was different ( i . e . non-stationary ) for S . haematobium between the time periods . Possible reasons for non-stationary spatial variation of S . haematobium can be broadly categorised into those related to the different sampling strategies used , and those related to changing epidemiology between the two study periods . Regarding the sampling strategies , the data were based on different sample locations , collected for different purposes and from different populations . The data from 1984–1989 were collected from the general population including adults , whilst the 2004–2006 data were from school-aged children . Age-stratified prevalence and intensity of S . haematobium infections in Mali have been reported [15] but individual or location-specific , age-stratified prevalence data were not available in the current study , which can be seen as its major limitation . However , previous analyses ( including an analysis of the same 1984–1989 dataset used in this report ) have shown that , while prevalence in school-aged children is generally higher than in the adult population , there is a consistent relationship between the prevalence in the two populations such that prevalence in one can be used to predict prevalence in the other [15] , [20] . The overall prevalence of S . haematobium in 1984–1989 , 25 . 7% , corresponds to an age-adjusted prevalence of approximately 36% in children aged 7–14 years [15] , which is very similar to the prevalence in school-aged children ( 38 . 3% ) in 2004–2006 . The 1984–1989 surveys had a less uniform geographical distribution than the 2004–2006 surveys , which is not surprising given that the 1984–1989 surveys were not explicitly designed with subsequent spatial analysis in mind , whereas uniform geographical coverage was an aim of the survey design for the 2004–2006 study to facilitate spatial analysis . Investigation of the impact of different sampling strategies on observed spatial correlation is an area of future research . Factors potentially related to changing epidemiology include desertification , urban growth and rural-urban migration [21] , [22] , changing demographic and socioeconomic characteristics of the population , long-term impacts of interventions on transmission and implementation of water resource development projects such as irrigation schemes , large dams and reservoirs [23] , [24] , [25] . These factors can influence not only stationarity of spatial variation but any differences observed in the location of spatial disease clusters . The earlier , GTZ-supported control programme focussed on specific , perceived high-risk areas of the country , with treatment coverage highest in Bandiagara , Office du Niger , Baguinéda and Sélingué . It might be suggested that spatial variation in changes in prevalence ( Figure 5 ) could relate to uneven geographical coverage of the intervention , but the main intervention areas do not correspond consistently to those where prevalence was lower in 2004–06 than 1984–89 . In addition to the limitation of different survey designs between periods , we were not able to compare spatial variation in intensity of infection between time periods because location-specific mean egg counts were not available from the 1984–1989 surveys . Maps of intensity would be useful for determining any changes in transmission across the periods . Examination of a single urine slide or single stool sample as a diagnostic approach results in sub-optimal sensitivity and this will also have affected the accuracy of our maps . We also did not incorporate anisotropy ( where the spatial correlation structure varies by direction ) or non-stationary spatial variation between different parts of the country , within each time period; these are future potential refinements of the models . We should also point out that the model predictions are distributions and here we have only presented the posterior mean . Examination of the full posterior distribution of predicted prevalence enables assessment of uncertainties arising from sampling and measurement error ( including in the model covariates ) . We have recently described how an understading of these uncertainties can assist decion making in schistosomiasis control programme planning [9] . Our results show that , while there were differences in the raw data , the overall prevalence of neither S . haematobium nor S . mansoni was significantly different between the time periods , despite ten years of donor-funded schistosomiasis control throughout the 1980s and early 1990s . The most likely explanation is that , in the absence of ongoing exposure reduction measures , re-infection with schistosomes following chemotherapy inevitably occurred . In endemic settings this is often apparent within 24 months [26] , [27] . Rates of infection and re-infection are generally similar among different age groups , although older people typically reacquire schistosome infection at slower rates than younger people [28] . Problems of re-infection were acknowledged by the managers of the 1980s control programme and this was reflected in the goal to reduce morbidity associated with infection in the treated communities ( which was successfully demonstrated in some areas [29] ) rather than transmission . The result was a predictable failure of the national programme to have a lasting impact on the burden of schistosomiasis in subsequent generations of Malians . One of the most important conclusions arising from the current work is that it is essential to develop a sustainability strategy to ensure ongoing benefits from the current national control programme . Recognising this fact , SCI has developed a sustainability plan which is outlined in Fenwick et al . [30] . Briefly , sustainability is based on initially using annual mass chemotherapy in areas with prevalence ≥50% , or biannual mass chemotherapy where prevalence is ≥10% and <50% , to rapidly reduce prevalence and intensity of infection . Then , when prevalence reaches <10% ( after up to four rounds of treatment , depending on levels of transmission ) , the Malian government plans to make treatments available in health facilities , carry out regular surveys and target treatment in schools if the prevalence rises above 10% . Sustainability also depends on developing the Malian health system and integrating schistosomiasis control with routine health care delivery [31] . Improved water sanitation and health education could be promoted for sustainable control [32] , snail control could be revisited and schistosomiasis vaccines might also have a future role [33] . The maps presented here can be used to target what are likely to be more limited national resources in the longer term to the highest-risk areas , where they will have the greatest impact on infection , morbidity , and ( hopefully ) transmission . The current move towards integration of control of neglected tropical diseases means that the government may have the opportunity to implement a cost effective control programme encompassing schistosomiasis , soil transmitted helminth infections , lymphatic filariasis , river blindness and trachoma . It is clear that a commitment from the Malian government and international donors for substantial resources is required long into the future , or alternative strategies need to be found , if control of schistosomiasis transmission in Mali is to be achieved . | Geostatistical maps are increasingly being used to plan neglected tropical disease control programmes . We investigated the spatial distribution of schistosomiasis in Mali prior to implementation of national donor-funded mass chemotherapy programmes using data from 1984–1989 and 2004–2006 . The 2004–2006 dataset was collected after 10 years of schistosomiasis control followed by 12 years of no control . We found that national prevalence of Schistosoma haematobium and S . mansoni was not significantly different in 2004–2006 compared to 1984–1989 and that the spatial distribution of both infections was similar in both time periods , to the extent that models built on data from one time period could accurately predict the spatial distribution of prevalence of infection in the other time period . This has two main implications: that historic data can be used , in the first instance , to plan contemporary control programmes due to the stability of the spatial distribution of schistosomiasis; and that a decade of donor-funded mass distribution of praziquantel has had no discernable impact on the burden of schistosomiasis in subsequent generations of Malians , probably due to rapid reinfection . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"infectious",
"diseases/neglected",
"tropical",
"diseases",
"public",
"health",
"and",
"epidemiology/epidemiology",
"infectious",
"diseases/helminth",
"infections",
"public",
"health",
"and",
"epidemiology/infectious",
"diseases",
"infectious",
"diseases/epidemiology",
"and",
"... | 2009 | A Comparative Study of the Spatial Distribution of Schistosomiasis in Mali in 1984–1989 and 2004–2006 |
Adaptor protein ( AP ) complexes regulate clathrin-coated vesicle assembly , protein cargo sorting , and vesicular trafficking between organelles in eukaryotic cells . Because disruption of the various subunits of the AP complexes is embryonic lethal in the majority of cases , characterization of their function in vivo is still lacking . Here , we describe the first mutation in the human AP1S1 gene , encoding the small subunit σ1A of the AP-1 complex . This founder splice mutation , which leads to a premature stop codon , was found in four families with a unique syndrome characterized by mental retardation , enteropathy , deafness , peripheral neuropathy , ichthyosis , and keratodermia ( MEDNIK ) . To validate the pathogenic effect of the mutation , we knocked down Ap1s1 expression in zebrafish using selective antisens morpholino oligonucleotides ( AMO ) . The knockdown phenotype consisted of perturbation in skin formation , reduced pigmentation , and severe motility deficits due to impaired neural network development . Both neural and skin defects were rescued by co-injection of AMO with wild-type ( WT ) human AP1S1 mRNA , but not by co-injecting the truncated form of AP1S1 , consistent with a loss-of-function effect of this mutation . Together , these results confirm AP1S1 as the gene responsible for MEDNIK syndrome and demonstrate a critical role of AP1S1 in development of the skin and spinal cord .
Protein trafficking between organelles in eukaryotic cells is mainly mediated by clathrin-coated vesicles and their assembly requires adaptor protein ( AP ) complexes [1] , [2] . The AP complexes also determine protein cargo selection for transport between the trans-Golgi network ( TGN ) , endosomes , lysosomes and the plasma membrane [3] , [4] and clathrin is important in establishing the basolateral domain [5] . Four ubiquitous AP complexes ( AP 1–4 ) have been characterized and each of them is composed of four subunits . The large subunits ( α , γ , δ or ε and β1–4 ) mediate binding to the target membrane and clathrin recruitment . The small subunit σ is part of the AP complex core and has been suggested to contribute to the stabilization of the complex , in conjunction with the medium subunit μ , which is primarily involved in protein cargo sorting [3]–[6] . Although the molecular understanding of the role of AP complexes in vesicular transport is progressing rapidly , the evidence for their role in vivo and in disease is more limited [4]–[7] . Knockdown or knockout of various AP-complex subunits has been attempted in different animal models , including the mouse γ and μ subunits and C . elegans σ subunits of AP-1A [4]–[7] . However , these are all embryonic lethal , further emphasizing the importance of these complexes for appropriate development . So far , a few but severe genetic disorders caused by mutations in genes encoding AP complex components have been described in humans . One of the most studied involves a mutation in the β3A subunit of AP-3 which underlies the Hermansky–Pudlak syndrome 2 ( HPS-2 ) [8] . This syndrome is characterized by oculocutaneous albinism , bleeding diathesis with absence of platelet dense bodies and abnormal depositions of ceroid lipofuscin in various organs . Mutated AP3B3A is believed to cause abnormal formation of intracellular vesicles from the trans-Golgi network or late endosomes , and probably mistrafficking of lysosomal proteins [7] , [8] . Recently , three mutations in AP1S2 , encoding the σ1B isoform of AP-1 , have been associated with X-linked mental retardation [9] . As AP-1 is associated with synaptophysin and the vesicular acetylcholine transporter , it was suggested that these mutations cause abnormal synaptic development and function . Erythrokeratodermia variabilis ( EKV ) is an autosomal dominant disease characterized by erythematous lesions and hyperkeratosis caused by mutations in two epidermally expressed connexin genes , GJB3 ( Cx31 ) and GJB4 ( Cx30 . 3 ) [10] , [11] . Because a significant proportion of EKV families do not have mutations in GJB3 and GJB4 , additional EKV genes remain to be identified [10] . We previously described the identification a new locus on chromosome 7q22 for an atypical form of EKV , in families with EKV lesions , as well as lamellar and erythrodermic ichthyosis ( Figure S1 ) [12] . In addition to the skin lesions , affected individuals from these families exhibit severe psychomotor retardation , peripheral neuropathy , and sensorineural hearing loss , together with elevated very-long-chain fatty acids and severe congenital diarrhea ( Table S1 ) . Given the similarities with the more recently described CEDNIK syndrome [13] , we used the related acronym MEDNIK for mental retardation , enteropathy , deafness , neuropathy , ichthyosis , and keratodermia to designate this unique syndrome . These MEDNIK families live in a relatively isolated population descended from a limited number of ancestors , and the gene responsible for this autosomal recessive syndrome was mapped by identifying a common homozygous region [12] . In this study we present a novel splice mutation in human AP1S1 , a ubiquitously-expressed gene encoding the small subunit σ1A of AP-1 , in four families with MEDNIK syndrome from the Quebec population . This founder mutation is predicted to cause the skipping of exon 3 , leading to a premature stop codon at the beginning of exon 4 . To further validate the AP1S1 mutation , we knocked down native Ap1s1 using antisense morpholino oligonucleotides ( AMOs ) in the developing zebrafish and examined the ability of wild-type ( WT ) and mutated human mRNA to rescue the developmental phenotype . Overall , our results confirm that mutation of the AP1S1 gene causes MEDNIK syndrome and suggest a critical implication for the AP1S1 gene in development of the skin and spinal cord .
The region harbouring the causative gene for MEDNIK syndrome , previously named Erythrokeratodermia Variabilis type 3 ( EKV3 ) , was recently mapped to a 6 . 8 Mb segment of chromosome 7p using a genome-wide single nucleotide polymorphisms ( SNP ) panel in 3 families originating from the Bas-St-Laurent region in the province of Quebec ( Canada ) , sharing a common ancestor at the 10th or 11th generation [12] . We genotyped a fourth pedigree , which enabled us to reduce the critical region to 5 . 3 Mb between markers D7S2539 and D7S518 ( data not shown ) . Among the candidate genes mapping to that interval , GJE1 ( encoding a connexin ) and CLDN15 ( encoding a claudin ) were sequenced but no mutation was found . Recently , a mutation in a SNARE protein ( SNAP29 ) was associated with cerebral dysgenesis , neuropathy , ichthyosis and palmoplantar keratoderma ( CEDNIK ) [13] . Since clinical manifestations of CEDNIK show striking similarities to the MEDNIK syndrome described here , we hypothesized that a mutation in AP1S1 , a functionally related gene mapping to the candidate interval , may cause the disease . By sequencing the gene , we identified a mutation in the acceptor splice site ( A to G ) of exon 3 in all individuals with MEDNIK ( IVS2-2A>G ) . This splice mutation is predicted to cause skipping of exon 3 , leading to a premature stop codon at the beginning of exon 4 ( Figure 1D ) . All parents and an unaffected sibling were heterozygous for this mutation ( Figures 1B and 1C ) . This mutation was not observed in 180 CEPH controls . In order to confirm the loss of exon 3 , RT-PCR analyses were performed on mRNA isolated from fibroblasts using primers located in exons 2 and 4 . As expected , a single band was observed in the controls . In contrast , two bands were detected in the carriers and patients ( Figure 1C ) . Direct sequencing confirmed that the lower band corresponded to an mRNA isoform lacking exon 3 . The higher band from the affected individuals corresponded to another RNA isoform , in which a cryptic splice acceptor site located 9 bp downstream of the start of the third exon was used . The resulting in frame protein is thus predicted to lack only three amino acids ( Figure 1D ) . The full-length AP1S1 mRNA species was not detected in these individuals . A semi-quantitative RT-PCR was performed on RNA isolated from mutation carriers and controls fibroblasts . Whereas heterozygous carriers had wild-type mRNA levels ranging form 40 to 75% of the expected value , the relative expression levels of both mutant isoforms was very low in affected individuals , corresponding to less than 10 % of the expected amount of RNA ( Figure 1C ) . Western blot analysis of skin proteins showed faint expression of the AP1S1 protein in affected individuals , suggesting partial expression of the isoform lacking three amino acids ( Figure S1C ) . The histological analysis of the skin revealed an epidermal hyperplasia accompanied by hypergranulosis and compact hyperkeratosis ( Figure S1B ) . To validate whether the AP1S1 mutation found in MEDNIK patients alters the biological function of this gene , we first knocked down Ap1s1 in zebrafish by inhibiting mRNA translation using an AMO [14] targeting its start codon ( Figure 1D ) . The morphological deficits of 48 hours post-fertilization ( hpf ) knocked down ( KD ) larvae ( n = 68/91 ) are summarized in Figure 2 , as the treatment was embryonic lethal at later stages . The 48 hpf Ap1s1 KD larvae were well formed but smaller in size compared to WT , and had reduced pigmentation ( Figures 2A and 2D ) . In addition , the KD larvae revealed prominent changes in the skin organization which were most visible in the fins ( Figures 2B and 2E ) . In contrast to the well-defined , fan-like , ray structure of the WT caudal fin , the fin of the Ap1s1 KD larvae was disorganized with rounded-up cells conferring a rough outline . Immature WT larvae did not show abnormal morphology of the skin and fin , suggesting that this phenotype is specific to the morpholino treatment rather than a general developmental retardation . The specificity of the AMO effect was confirmed by using Ap1s1 Western blotting and immunolabelling in wholemount larvae . With both methods we observed a decrease in the intensity of the Ap1s1-specific labeling in the Ap1s1 KD larvae compared to the WT ( Figure 2D , inset , Figures 2F and 2C ) . Also , larvae injected with a control AMO ( 5 mispaired bases ) did not show significant differences compared to the WT ( n = 26/26 , Figure S2D ) . Finally , in order to mimic the splice mutation found in individuals with MEDNIK , we designed a morpholino targeting the Ap1s1 intron 2 acceptor splice site ( Figure 1D , Figure S2G , n = 32/58 ) . In this latter experiment , we found the same abnormal skin and fin morphology as observed by using AMO targeting the Ap1s1 start codon , although the phenotype was less penetrant . To determine if an increase in cell death underlies the skin phenotype in the KD embryo , we stained these larvae with the vital dye acridine orange [15] . We did not observe a difference compared with control ( not shown ) , suggesting that the skin and fin disorganization was not due to an initial outgrowth followed by tissue degradation . We further tested whether the skin malformation was due to a problem in early epidermal patterning by using immunolabeling for p63 , a marker of basal keratinocyte nuclei [16] . Despite the prominent changes in the size and the shape of the tail , p63-positive keratinocytes were present both in WT ( Figure 3A ) and KD larvae ( Figure 3B ) . To look for a change in the population of proliferating cells , we performed immunolabelling with the phosphorylated-histone-H3 ( PH3 ) antibody to visualize cells undergoing histone modification during mitosis , which did not reveal any obvious difference between the KD and control larvae ( not illustrated ) . Similar results were obtained with co-immunostaining against p63 and PH3 , suggesting unaffected proliferation level of basal keratinocytes population in the KD larvae ( not illustrated ) . To further investigate whether the keratinocytes in the KD larvae exhibit specific abnormalities , we immunolabeled WT and KD larvae for laminin ( Figures 3C and 3D ) and for cadherin ( Figures 3E and 3F ) . Laminins , in particular laminin 5 , are synthesized by keratinocytes and are their main anchor to the basement membrane [17] , while cadherins are localized to the keratinocyte cell membrane and are essential in maintaining cell-cell adhesion [18] . In the WT , laminin was detected at the outer edges of the fin ( Figure 3C ) while in the KD larvae ( Figure 3D ) the detected laminin appeared diffuse , with an abnormal localization . Furthermore , in the KD larva , cadherin immunolabeling was less obvious at the cell membrane of keratinocytes doubly-labeled with cadherin ( green ) and p63 ( orange ) ( Figure 3F ) In contrast , the localization of cytokeratin , a major cytoskeletal protein expressed exclusively in epithelial cells [19] , [20] seemed to be preserved in KD larvae ( Figures 3G and 3H ) . At 48 hpf WT larvae normally respond to touch by swimming , which is characterized by alternating tail movements with a beat frequency of about 30 Hz ( Figure 4A ) [21] , [22] . In contrast , Ap1s1 KD larvae reacted to touch by tail coils ( Figure 4F ) , an embryonic motility pattern that usually disappears around 24 hpf [21] . Since the KD larvae exhibited severe motor impairment , we further investigated the spinal cord neural organization . An anti-acetylated tubulin staining revealed a reduction in axonal processes in the spinal cord of Ap1s1 KD larvae ( Figure 4G ) compared to the WT ( Figure 4B ) . To quantify the number of newly born neural cells , wholemount 48 hpf larvae were labeled using anti-HU , as this RNA binding protein is found in neuronal cells leaving the mitotic cycle [23] . The number of newly born neurons in KD larvae ( Figure 4H , n = 3 , 41±3 ) significantly decreased to 51% of control , WT , levels ( Figure 4C , n = 3 , 81±9 , p<0 . 001 ) . We also quantified the progenitor population in the spinal cord using an anti-PH3 , but we did not find a significant change between Ap1s1 KD and control larvae groups ( n = 6 each , not illustrated ) , nor did we observe significant cell death upon staining with acridine orange . To study which population of neurons was specifically affected , we labeled interneurons and motoneurons by using anti-Pax2 , which labels a large subset of early differentiating interneurons [24] and anti-HB9 , a homeobox gene necessary for motoneuron differentiation [25] . Interestingly , whereas the number of motoneurons was unchanged ( Figures 4D and 4I; n = 3 each , p = 0 . 42 ) , we observed a 46 % reduction in the number of interneurons in Ap1s1 KD larvae compared to the WT ( Figures 4E and 4J; n = 3 each , WT 28±1 . 5 , KD 13±0 , p<0 . 001 ) . This behavioral and spinal phenotype was specific to the morpholino treatment and not just a reflection of general developmental retardation , as reflected by the sparing of motoneurons and loss of interneurons , which is not observed during normal development . All larvae co-injected with human wild type human AP1S1 mRNA and Ap1s1 AMO exhibited restoration of the skin organization , pigmentation ( Figures 2G–I ) , as well as swimming behavior ( n = 35/35 fish ) . Conversely , larvae co-injected with human AP1S1-exon3 mRNA and Ap1s1 AMO showed skin and motor deficits similar to those observed in Ap1s1 KD larvae , suggesting a loss of function of this truncated form of the protein ( Figure S2F , n = 24/24 ) . However , co-injection of the human alternative mutant AP1S1-9bp mRNA together with the AMO rescued the phenotype ( Figure S2E , n = 19/19 fish ) , suggesting that this protein isoform lacking 3 amino acids remains functional . Larvae injected with the mismatch morpholino oligonucleotide were similar both morphologically and behaviorally to the WT ( Figure S2D , n = 26/26 ) .
Little is known about the σ subunit role in AP complex formation and function in vivo . It is suggested that AP1S1 contributes to the AP complex core stabilization [6] , [26] . Furthermore , in AP-1 and AP-3 , the σ subunit is suggested to interact with “dileucine-based” recognition signal on cargo proteins , in combination with the γ or the δ subunit respectively . Therefore , this implicates the σ subunit in protein sorting as well [27] . However , attempts to interfere with AP1S1 function in vivo were not successful so far , as they resulted in embryonic lethality . Similar results were obtained by interfering with most of the other subunits of the AP-1 complex , further emphasizing its importance for appropriate development [4] , [7] . In this study , we knocked down Ap1s1 in zebrafish and were able to rescue the morphological and behavioral phenotypes observed in KD larvae by co-injecting WT human AP1S1 mRNA , which further support the specificity of the Ap1s1 knockdown . The remaining levels of Ap1s1 protein may explain viability in zebrafish , at least for the first 48 hours of development . However , because some of the AP complexes have overlapping function , compensation by other AP complexes cannot be excluded [6] , [28] . Nevertheless , since the Ap1s1 KD larvae exhibit severe deficits , neither residual levels of AP-1A and B , nor the activity of other AP complexes were sufficient for appropriate development of many cell types ( skin , pigment and neural ) . In this study , we demonstrate for the first time that disruption of an AP-1 subunit , more specifically the σ1A subunit , causes perturbation in epithelial cell development in vivo . The presence of p63 immuno positive basal cytokeratinocytes in the KD larva suggested that knocking down Ap1s1 did not interfere with early epidermal patterning . The skin phenotype was not accompanied by an increased cell death or in the level of proliferating basal keratinocytes . Carney et al . [29] observed an increase in proliferating basal keratinocytes in zebrafish mutants suffering from severe epithelial disintegration and suggested that this phenomenon is a secondary consequence of inflammation and consequent loss of epithelial integrity . The lack of increased proliferation in our study could be explained by the presence of sufficient residual laminin to provide some anchoring for the keratinocytes , allowing the maintenance of some epithelial properties . However these residual levels of laminin appeared insufficient for appropriate basement membrane development . Interestingly , zebrafish embryos carrying a mutation in the gene encoding for laminin 5 suffer from severe deficits in fin formation due to disruption in basement membrane integrity [30] . In Ap1s1 KD larvae , we also found an alteration in the localization of cadherin in basal keratinocytes , which was not accompanied by changes in cytokeratin localization , suggesting that this component of epithelial cells cytoskeleton remain unaffected by AP1S1 dysfunction . Interestingly , the nature of the specific adaptor complex that recognizes the cadherin dileucine sorting motif is unknown , although AP-1 is a candidate [18] . Based on these observations , we suggest that Ap1s1 knockdown resulted in failure to localize cadherin to the basolateral cell membrane which , together with an abnormal pattern of expression of laminin 5 , lead to a loss of epidermal layer integrity . The well-formed 48 hpf Ap1s1 KD larvae showed a severe behavioral phenotype . Instead of reacting to touch by swimming , the KD larvae coiled in a motility pattern distinctive of younger embryos . Consistent with this observation , detailed examination of the spinal cord revealed an abnormal development . The extent of axonal processes was diminished and the number of newly born neurons was reduced to half of the WT levels due mainly to a decrease in the interneuron population , but not in motoneurons . Interestingly , as observed in the skin , no change was seen in the levels of neuronal progenitors in the spinal cord . There is mounting evidence that AP complexes such as AP-2 and AP-3 are implicated in neural function [31] . For example , mice with knockout of the AP-3 μ3Β subunit are susceptible to epileptic seizures because of deficient GABAergic vesicle formation and function [32] . Also , mocha , one of the mouse models for Hermansky-Pudlak syndrome ( HPS ) in which the δ subunit of AP-3 is mutated , suffer from neurological disorders [33] . The loss of AP-3 in these mice affected spontaneous and evoked neurotransmitter release in hippocampal mossy fiber synapses [34] . AP-2 is implicated in selective endocytosis and recycling of synaptic vesicles and also of receptors and transporters from the plasma membrane of nerve terminals [31] , [35] . For example , internalization of α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid ( AMPA ) receptors by binding to AP-2 is essential for N-methyl-D-aspartic acid ( NMDA ) -induced long-term-depression in the hippocampus and therefore to synaptic plasticity [36] , [37] . In turn , little is known about AP-1 function in neurons , although it was reported to interact with synaptophysin , one of the most abundant proteins in synaptic vesicles [38] , as well as with vesicular acetylcholine transporter [39] . Moreover , AP-1 binds to the ubiquitous microtubule-associated motor protein KIF13A , a member of a protein family implicated in neuronal transport of membranous organelles , synaptic vesicles and proteins from the cell body to the axons and dendrites [40] , [41] . Mice with mutations in members of this protein family ( KIF1A , KIF1Bβ ) show reduced synaptic vesicles in the synaptic terminals and suffer from in sensory-motor deficits [42] . Also , mutations in KIF1Bβ cause Charcot-Marie-Tooth hereditary peripheral neuropathy type 2A in humans [42] . It is thus possible that AP1S1 , in addition to its possible implication in synaptic vesicles regulation and formation , could be implicated in their transport toward the neural processes . Although not much is known about the precise role of AP-1 in the developing central nervous system , we show here that the disruption of the AP-1 function is associated with substantial perturbation of a subset of spinal interneuron differentiation . Ap1s1 KD larvae exhibit abnormal development of neurons and skin cells , a phenotype that shows similarities to the clinical manifestations observed in individuals with MEDNIK . Based on the observation of reduced neurogenesis we have made in zebrafish , we speculate that MEDNIK syndrome in affected patients is caused by an impaired development of various neural networks , including the spinal cord ( ataxia and peripheral neuropathy ) and possibly the brain ( microcephaly and psychomotor retardation ) and inner ear ( sensorineural deafness ) . We also hypothesize that disruption of AP1S1 in humans may be associated with more extensive perturbation of organogenesis . Indeed , growth retardation , digestive tract malformations and dysfunction ( chronic diarrhea ) , and elevation of very long chain fatty acid observed in individuals with MEDNIK syndrome might reflect more widespread perturbation of vesicular transport and of epithelial cell development . One intriguing question is why the AP1S1-exon 3 mutation is not lethal in homozygous individuals with MEDNIK . Indeed , overexpression of human AP1S1-exon3 mRNA failed to rescue the phenotype observed in Ap1s1 KD larvae , suggesting a loss of function of this critical protein . However , co-injection of the AP1S1-9bp human mutant mRNA with AMO , the alternative RNA species detected in our MEDNIK patients , rescued the phenotype , suggesting that this alternative splicing results in a functional protein . The expression of that protein isoform in patients may thus explain their viability . The fact that the AP1S1-9bp mRNA is expressed at low levels ( less than 10 % of normal levels in fibroblasts ) could explain why it is not sufficient to sustain normal development and function and further highlight the important role of AP1S1 in normal development . Furthermore , the expression levels of the different AP1S1 isoforms may vary from one tissue to another , as well as between individuals , thereby contributing to the variability of the phenotype . Overall , these observations in zebrafish , in light of previous in vitro studies [31] , [34] , [43]–[45] , suggest that AP1S1 and AP-1 complex are most likely implicated in appropriate protein sorting and transport . Interference with these pathways could therefore result in perturbation of cellular organization and be detrimental for the development of specific cell subpopulations , as we observed respectively in the skin and the spinal cord of the Ap1s1 KD larvae . The results suggest avenues for both basic and clinical research , in order to better understand the mechanisms underlying MEDNIK and related neuro-cutaneous syndromes .
Seventeen individuals from four families including three affected children were ascertained and examined as described [12] . Genetic material of affected individuals and unaffected siblings and parents was isolated from blood lymphocytes at Le Service de Dermatologie du CHRGP de Rivière-du-Loup and Le Service de Génétique du CHUQ ( Hôpital St-François d'Assise ) . Fibroblast cell cultures were obtained from 3 mm punch biopsies from patients , relatives or healthy controls and were maintained in Dulbecco's Modified Eagle's medium ( DMEM ) supplemented with fetal calf serum 10% . The study was approved by the Institutional Review Board of the Hôpital St-François d'Assise and informed consent was obtained from all family members . Coding regions of AP1S1 were amplified by PCR from genomic DNA ( primer sequences are available upon request ) . Total RNA was extracted from cultured primary fibroblasts harvested from skin biopsy samples using standard protocols . cDNA was prepared using random hexamers and standard procedures , and a fragment from exon 2 to exon 4 of AP1S1 was amplified with the primers used for the Taqman exon 3 assay ( see below ) . All DNA templates were amplified using HotStar Taq polymerase ( Qiagen , Valencia , CA ) and standard conditions ( 95°C for 5 min; 40 cycles of 95°C for 30 sec , 60°C for 30 sec and 72°C for 30 sec; and 72°C for 10 min . ) . Amplicons were sequenced in both directions using the same primers than for PCR . Taqman assay was performed on cDNA ( obtained from fibroblast isolated RNA ) using the Taqman kit ( Applied Biosystems , Foster City , CA ) and according to the manufacturer's conditions . For the exon 2 assay , 300 nM of these PCR primers , AP1S1_exon2F , 5′-gagctcatgcaggttgtcct-3′; AP1S1TaqR , 5′-AGTTGAAGATGATGTCCAGCTC-3′ , and 200 nM of the probe , AP1S1TaqP_exon2 , 5′FAM-CCTGGAGTGGAGGGACCTCAA-TAMRA3′ , were used . For the exon 3 assay , 300 nM of these PCR primers , AP1S1TaqF , 5′-TGGAGGGACCTCAAAGTTGT-3′ and AP1S1TaqR , 5′-AGTTGAAGATGATGTCCAGCTC-3′ , and 200 nM of the probe , AP1S1TaqP , 5′FAM-CACACTGGAGCTGATCCACCGATAC-TAMRA3′ , were used . All primers were designed using NM_001283 as the reference sequence . As an expression control for use in quantification , universal 18S primers were included in the same reaction mixes . PCR conditions were: 95°C for 10 min , 45 cycles of 30 sec at 95°C , 30 sec at 56°C , and 30 sec at 72°C . Reactions were cycled on the 7900HT Real-time PCR instrument ( Applied Biosystems ) . Relative expression for each sample was evaluated by using the difference in the threshold cycle ( ΔCt ) value to achieve a similar level of fluorescence . 18S relative expression was used to normalize for the cDNA quantity of each sample . All values correspond to an average of three independent experiments . We designed primers ( AP1S1-5′-TAAGCGGATCCATGATGCGGTTCATGCTATTATTC , and AP1S1-3′-GTAAGCCTCGAGTCAGTGGGAAAAGGGGAAAGTGG ) to amplify the complete open reading frame of AP1S1-variant1 from a human brain cDNA library ( Marathon-ready , BD Biosciences Clontech ) , using Pfu Polymerase ( Stratagene ) . The same primers were used on patient's cDNA to get the mutated alleles , using Advantage 2 Polymerase ( Clontech ) . By using BamHI and XhoI restriction sites introduced into the primer sequences , the PCR products was directionally cloned into pCS2+ vector . All constructs were completely sequenced to confirm the mutations , as well as to exclude any other variants that could have been introduced during the PCR amplification . Capped sense mRNAs were synthesized from pCS2+ by using the mMESSAGE mMACHINE SP6 kit ( Ambion ) . Skin biopsies were also used to perform histological analysis . The samples were fixed in formalin 10% and embedded in paraffin . Sections of aproximately 5 µm were cut by using cryostat , and stained with haematoxylin and eosin . Experiments were performed on zebrafish ( Danio rerio ) larvae raised at 28 . 5°C according to previously established procedures [46] , and in compliance with Canada Council for Animal Care and institutional guidelines . To knockdown the function of the gene encoding for the σ1A subunit of AP-1 in zebrafish , which shares 91% identity with the human AP1S1 protein , an AMO ( Gene Tools ) was designed to target the initial codon of zebrafish Ap1s1 gene ( 5′-ACAGAAGCATAAAGCGCATCATTTC- 3′ ) , which differs in sequence from human AP1S1 . In addition , a second morpholino was designed to target the acceptor splice site ( intron 2 ) of the zebrafish Ap1s1 gene , 5′-GACTAGCATACCTACGTAAACACAC-3′ . All AMO preparation and injection procedures were according to previously described protocols [13] . The specificity of our AMO was verified by injection of a control , 5 base pairs mismatch morpholino oligonucleotide ( 5′-ACACAAGGATAAACCGCATGATATC- 3′ ) as well as by Western blotting as will be described below . After establishing the AMO phenotype ( 1 mM ) , rescue experiments were preformed in which both AMO ( 1 mM ) and human AP1S1 WT or mutated mRNA ( 110 ng ) were injected . Skin biopsies were obtained from normal individual , carrier and patients ( lesional and non-lesional skin ) . The samples were frozen in liquid nitrogen and homogenized in lysis buffer ( RIPA: Tris-HCl 50 mM , NaCl 150 mM , EDTA pH 8 . 0 , Triton 1% , Sodium deoxycholate 1% , SDS 0 . 1% , Protease inhibitors ( complete mini , Roche ) , Aprotinin 10 µg/ml , Leupeptine 10 µg/ml , phenylmethylsulphonyl fluoride ( PMSF ) 1 mM ) . The lysates were centrifuged at 12 000 g for 20 min at 4°C . To quantify gene knockdown , thirty 48 hpf WT and Ap1s1 KD larvae were dechorionated and anaesthetized in 0 . 2% MS-222 ( Sigma ) and then homogenized in lysis buffer ( 150 mM NaCl , 1% IGEPAL CA-630 , 50 mM Tris , pH 8 . 0 , 0 . 5% sodium deoxycholate , 0 . 1% SDS . The lysates were centrifuged 10 min at 2000 g at 4°C in complete protease inhibitor cocktail ( Roche ) . After the protein extraction , western blot protocols were the same for both human skin samples and zebrafish . The supernatants were removed and the proteins were quantified using DC protein Assay ( BIO-RAD ) with bovine serum albumin ( BSA ) as a standard . As a primary antibody , rabbit antisera DE/1 directed against Ap1s1 was used at a concentration of a 1∶5000 ( antibody obtained from Dr . Traub ) [47] . Horseradish peroxidase-conjugated donkey anti-rabbit IgG ( 1∶5000; Jackson Immunoresearch Laboratories Inc . ) was used as a secondary antibody . Visualization was performed by using Western Lightning Chemiluminescence Reagent Plus ( PerkinElmer ) . Hybridization of the same blot using anti-actin antibody was used to assess equal loading of the samples ( mouse monoclonal anti-actin 1∶5000 , Chemicon #MAB1501 ) . Briefly , all dechorionated larvae were collected , anesthetized in 0 . 2% MS-222 ( Sigma ) and fixed for two hours in 4% paraformaldehyde ( PFA ) at room temperature as previously described [46] . Samples were then washed in phosphate buffered saline ( PBS ) before dehydration in 100% methanol and kept at −80°C for later use . For cytokeratin labeling , larvae were stored in Dent's fixative at –20°C Primary and secondary antibody incubations were conducted overnight at 4°C in blocking solution . Then samples were washed in PBS-Tween and incubated overnight with Alexa 488 ( anti-rabbit ) or 568 ( anti-mouse ) antibodies ( Molecular Probes ) . After four washouts in PBS-tween , larvae were mounted on slides in glycerol 90% , for immunofluorescence imaging . Primary antibodies were used at the following dilutions: rabbit antisera DE/1 directed against Ap1s1 1∶200; monoclonal mouse anti-acetylated tubulin ( Sigma ) 1∶1000; monoclonal mouse anti-HB9 ( Developmental Studies Hybridoma Bank 81 . 5C10 ) 1∶200; polyclonal rabbit anti-Pax2 ( Covance PRB-276P ) 1∶100; polyclonal rabbit anti-phosphohistone H3 ( Ser10 ) ( Upstate 06 570 ) 1∶100; monoclonal mouse anti-HU ( Molecular Probes A21271 ) 1∶100; rabbit anti-laminin ( Sigma L9393 ) 1∶100; rabbit anti-pan cadherin ( Sigma C 3678 ) 1∶400; monoclonal mouse anti-p63 ( Santa Cruz sc-8431 ) 1∶100; monoclonal mouse anti-cytokeratin type II KS Pan 1-8 ( Progen Biotechnik 61006 ) 1∶10 . To verify for cell death in wholemount larva , we stained them using the vital dye Acridine Orange , as described previously [48] . The fluorescent images represent the maximum projection of a series of 2 µm optical sections obtained in whole mount larva using a laser confocal microscope ( Perkin Elmer Ultraview system mounted on a LEICA DM LFSA microscope with a 63X oil objective 1 . 25 NA ) and Metamorph software ( Universal Imaging Corp ) . Antibody-labeled cells ( HU , BH9 , PAX2 and PH3 ) were counted in equal length spinal cord segments ( 75 µm ) imaged at the 14th somite and cover the entire spinal cord volume . Statistical significance between Ap1s1 KD and WT larva groups was verified using Mann-Whitney rank sum test ( Sigmastat ) . Transmitted light images were digitized using a digital camera ( Axio Cam HRC , Zeiss ) mounted on a dissecting microscope ( Stem1 SV 11 , Zeiss ) and Axiovision 4 . 2 software . To document the response to touch of the 48 hpf larva high-speed video films were digitized ( 250 frames/sec ) using a Photron Fastcam PCI high-speed video camera mounted on a Zeiss dissection microscope . The captured films were analyzed off line to determine swim frequency . Representative images from these films were used to reconstruct the movements of Ap1s1 KD and WT larvae in Figure 4 . | We describe a novel genetic syndrome that we named MEDNIK , to designate a disease characterized by mental retardation , enteropathy , deafness , peripheral neuropathy , ichthyosis and keratodermia . This syndrome was found in four French-Canadian families with a common ancestor and is caused by a mutation in the AP1S1 gene . This gene encodes a subunit ( σ1A ) of an adaptor protein complex ( AP-1 ) involved in the organisation and transport of many other proteins within the cell . By using rapidly developing zebrafish embryos as a model , we observed that the loss of this gene resulted in broad defects , including skin malformation and severe motor deficits due to impairment of spinal cord development . By expressing the human AP1S1 gene instead of the zebrafish ap1s1 gene , we found that the normal human AP1S1 gene could rescue these developmental deficits but not the human AP1S1 gene bearing the disease-related mutation . Together , our results confirm AP1S1 as the gene responsible for MEDNIK syndrome and demonstrate a critical role of AP1S1 in the development of the skin and the spinal cord . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"dermatology/pediatric",
"skin",
"diseases,",
"including",
"genetic",
"diseases",
"neurological",
"disorders/spinal",
"disorders",
"cell",
"biology",
"neurological",
"disorders",
"genetics",
"and",
"genomics/genetics",
"of",
"disease",
"genetics",
"and",
"genomics/gene",
"f... | 2008 | Disruption of AP1S1, Causing a Novel Neurocutaneous Syndrome, Perturbs Development of the Skin and Spinal Cord |
Pseudogenes are usually considered to be completely neutral sequences whose evolution is shaped by random mutations and chance events . It is possible , however , for disrupted genes to generate products that are deleterious due either to the energetic costs of their transcription and translation or to the formation of toxic proteins . We found that after their initial formation , the youngest pseudogenes in Salmonella genomes have a very high likelihood of being removed by deletional processes and are eliminated too rapidly to be governed by a strictly neutral model of stochastic loss . Those few highly degraded pseudogenes that have persisted in Salmonella genomes correspond to genes with low expression levels and low connectivity in gene networks , such that their inactivation and any initial deleterious effects associated with their inactivation are buffered . Although pseudogenes have long been considered the paradigm of neutral evolution , the distribution of pseudogenes among Salmonella strains indicates that removal of many of these apparently functionless regions is attributable to positive selection .
One of the most distinctive features of bacterial genomes is their high coding densities , in which genic regions typically constitute more than 80% of the total genome [1] . This is in sharp contrast to many eukaryotes whose genomes contain vast stretches of non-coding DNA and a multitude of transposable and repetitive elements , with protein-coding regions often accounting for only 1% of the genome [2] . The paucity of non-coding regions in bacterial genomes has lead to the idea that pseudogenes would be exceedingly rare [3]; however , recent large-scale analyses have found that virtually all bacterial genomes contain disrupted and eroded genes that have full-length counterparts in other related genomes [4]–[8] . Pseudogenes are particularly prevalent in those bacterial species that have recently become associated with or dependent upon eukaryotic hosts [9]–[10] , and in the most extreme cases , pseudogenes can number in the 1 , 000 s and occupy over half of the genome [11]–[12] . The pseudogenes in bacterial genomes are continually created from ongoing mutational processes and are subject to degradation , and eventual removal , by the further accumulation of mutations . However , the most surprising aspect of bacterial pseudogenes is that their retention time appears to be extremely short . Even in comparison of very closely bacteria , there are very few pseudogenes that are shared among strains typed to the same bacterial species [6] , [7] , [13] . This observation indicates that bacterial pseudogenes , although often present in high numbers , are deleted at a relatively rapid rate . This feature is again in sharp contrast to eukaryotes , in which pseudogenes often persist over evolutionary timescales and may be shared by distantly related lineages , such as rodents and primates [14]–[17] . Due to the pervasive mutational bias towards deletions that has been observed across bacterial genomes [18]–[19] , the rapidly removal of pseudogenes could be caused by the random fixation of background mutations . Because pseudogenes have long been viewed as “a paradigm of neutral evolution” [20] , this is the favored hypothesis . Alternatively , pseudogenes could effect a cost and be eliminated from bacterial genomes by an adaptive process . For example , pseudogenes might be detrimental to the organism through energetic costs incurred by the continued transcription and translation of non-functional genes and/or through the production of proteins that are toxic to cells . In this study , we examine the formation , loss and phylogenetic distribution of disrupted genes in Salmonella . We focus on this bacterial genus because: ( 1 ) high-quality genomic sequences of several Salmonella strains have been determined , ( 2 ) Salmonella genomes , like those of most other pathogens , possess considerable numbers of pseudogenes [21]–[22] , ( 3 ) the population structure of Salmonella enterica is essentially clonal [23] , allowing the resolution of an unambiguous strain phylogeny , and ( 4 ) both experimental [24] and comparative [25]–[27] studies provide evidence of a strong deletional bias in Salmonella , such that genes that are not maintained by selection are rapidly inactivated and eliminated by mutational events . And it is against this background that we test the possibility that the removal of bacterial pseudogenes is adaptive .
Despite the similarity in overall genome sizes , the number of pseudogenes identified in the five Salmonella enterica subsp . enterica genomes can vary by an order of magnitude , ranging from 13 in S . enterica sv . Typhimurium to 147 in S . enterica sv . Gallinarum ( Figure 1 ) . The abundance of pseudogenes is not associated with divergence time or phylogenetic affiliations . In fact , the two most closely-related strains , S . enterica sv . Gallinarum and S . enterica sv . Enteritidis , represent nearly the observed extremes of pseudogene abundance ( 147 versus 21 ) . Interestingly , the abundance of pseudogenes in the Salmonella genomes is reflected in the genome-wide Ka/Ks ratio , which can serve as a proxy for measuring the level of genetic drift experienced by a lineage . For example , despite their independent origins , the three strains with the highest numbers of pseudogenes also have the highest genome-wide Ka/Ks ratios . This observation is consistent with the expectation that deleterious mutations , such as inactivation of functional genes , are more likely to reach fixation in populations under high levels of genetic drift . In addition to higher rates of pseudogene production , high levels of genetic drift may also result in slower rates of pseudogene removal ( assuming that the removal of pseudogenes is favored by positive selection ) . Consistent with this expectation , highly degraded pseudogenes with multiple inactivating mutations were only found in the three strains with high genome-wide Ka/Ks ratios . But despite the high correlation coefficient between genome-wide Ka/Ks ratios and the abundance of pseudogenes in these genomes ( r = 0 . 74 ) , the correlation does not reach statistical significance ( P = 0 . 15 ) , possibly because of the limited number of genomes examined or the use of a distant outgroup limits the resolution in Ka/Ks ratio estimation ( i . e . , the difference in high vs . low Ka/Ks groups is underestimated because most substitutions occurred on the branch leading to the outgroup ) . Consistent with previous studies in other bacterial genera [6] , [7] , [13] , most pseudogenes in Salmonella genomes are strain-specific . Of the 147 pseudogenes in the S . enterica sv . Gallinarum genome , only five are shared with its closest relative , S . enterica sv . Enteritidis . And of these five , only three share the same inactivating mutations and can be inferred as ancestral . The remaining two shared pseudogenes have different inactivating mutations and were inferred to result from independent events . To examine mutational processes responsible for creating pseudogenes , we characterized each pseudogene in these Salmonella genomes by the number and type of gene-inactivating mutations . The vast majority of Salmonella pseudogenes ( 346/378 ) have only a single inactivating mutation ( Figure 2 ) , and among these , short deletions that removed 20% or less of the original open-reading frame predominate ( 141/346 ) . We observed two cases of complete removal , including a 2 , 557-bp deletion containing a 1 , 224-bp gene and a 441-bp deletion encompassing a 189-bp gene . Because we required a pseudogene to be flanked by two conserved genes for its identification , any pseudogene that was removed by a deletion including a neighboring gene would not be recognized by this synteny-based approach . As expected from the mutational bias towards deletions in bacterial genomes , only a small fraction ( 17% ) of the 346 pseudogenes were produced by an insertion; and with the exception of two cases of transposon insertions , all insertions are <10 bp . All remaining cases are due to point mutations: in 131 cases , there is a premature stop codon that reduced the length of the open reading frame by more than 20% and in two cases , a point mutation altered the start codon ( one ATG to ATA , and one ATG to ATT ) . We identified 32 pseudogenes with more than one inactivating mutation . There is no significant difference in average gene length between pseudogenes containing multiple inactivating mutations and those with a single inactivating mutation . Detailed information of the 378 curated pseudogenes is presented in Table S1 . Because neutral sequences accrue mutations with time , the relative age of a pseudogene is reflected in its number of accumulated mutations . The preponderance of pseudogenes with a single inactivating mutation indicates that most pseudogenes in these genomes are very young ( Figure 2 ) . This is also supported by the fact that most pseudogenes are restricted to individual genomes , as expected if they are newly formed . Further attesting to the recency of most pseudogenes is that there is little or no sign of accelerated sequence divergence relative to their functional orthologs ( Figure 3 ) . Given the strong deletional bias in bacterial genomes , it is possible that the lack of old pseudogenes results from the rapid elimination of non-functional sequences by random fixation of mutations alone . If the jettisoning of pseudogenes is largely governed by a strictly neutral process , we expect that the probability of pseudogene removal to be independent to its age . Under this scenario , the age-class distribution is expected to decrease linearly in a log-normal plot ( e . g . , Figure 2 and Figure 3 ) . However , there is an overabundance of young pseudogenes relative to this expectation no matter which one of the three methods we used to assign age class ( phylogenetic distribution , number of inactivating mutations , and level of accelerated sequence divergence ) . This indicates that a non-neutral force is operating to remove young pseudogenes such that few remain in the genome long enough to accumulate multiple inactivating mutations or to exhibit accelerated sequence divergence rates ( as would be expected for non-functional regions that were released from selective constraints ) . Consistent with this hypothesis , we detected a strong negative correlation between the loss rate and the age of pseudogenes estimated by the number of inactivating mutations ( r = −0 . 99 , P = 0 . 013 ) . The scarcity of old pseudogenes is not likely to be an artifact of the methodologies used to identify pseudogenes ( or , as noted above , the methods used to assign pseudogene age ) : our synteny-based approach is capable of detecting highly degraded pseudogenes harboring more than 10 frameshift mutations [19] . Of the 378 identified pseudogenes among S . enterica genomes , 120 have functional orthologs in Escherichia coli str . K-12 substr . MG1655 for which protein-protein interaction data are available . The average numbers of interacting partners in the protein-protein interaction network ( i . e . , the connectivity ) for pseudogenes having different numbers of inactivating mutations revealed that highly degraded pseudogenes ( i . e . , those with three inactivating mutations ) have , on average , significantly fewer interacting partners that do newly formed pseudogenes ( i . e . , those with a single inactivating mutation ) ( Figure 4 ) .
A possible explanation for this violation of neutrality is that there is selection for minimizing the size of mutational targets [28]; and since bacteria often have larger effective population sizes than do eukaryotes [29] , it might be possible for selection to operate on mutations with extremely small effects ( e . g . , a 1-kb pseudogene accounts for only about 0 . 02% of a Salmonella genome ) . Unfortunately , this hypothesis cannot explain the observed pattern: If selection against inert DNA were the primary factor causing the removal of pseudogenes , we would expect to find fewer , but not necessarily a higher loss rate , for young pseudogenes . The age-distribution pattern predicted by this mutational-target model would , in fact , be indistinguishable from a strictly neutral model . To account for differences in loss rate among pseudogenes belonging to various age classes requires methods that can accurately determine the relative ages of the pseudogenes present in this bacterial clade . Because most mutations accumulate as a function of time , one method was to use the level of sequence degradation as an indicator of pseudogene age . Because the youngest pseudogenes , i . e . , those containing only a single inactivating mutation , have a higher probability of being expressed , their increased loss rate could result from the energetic costs of transcription and translation , which are known to shape the genome organization in prokaryotes [30]–[32] and eukaryotes [33] . Because bacterial cells are haploid , all mutations are effectively dominant since non-functional gene products cannot be masked by the corresponding functional allele as in diploid organisms . In that short indels ( i . e . , less than 10-bp ) that cause frameshift and pre-mature stop codons are two of the most common types of mutations observed among bacterial pseudogenes ( see Results and [6]–[7] ) , it is likely that the products from these altered open reading frames are disruptive to normal operation of cellular networks . This ‘toxic protein’ hypothesis is supported by our inference of protein-protein interactions: We find that those few pseudogenes that have persisted in Salmonella genomes ( i . e . , those that have accumulated multiple inactivating mutations ) correspond to genes with relatively few interacting partners ( Figure 4 ) . The low connectivity of these genes can perhaps serve to minimize the deleterious effects of their inactivation . In contrast , for genes with large numbers of interacting partners , alteration of the open reading frames would potentially impact many protein-protein interactions . As such , mutations that remove such pseudogenes would be highly favorable and quickly reach fixation in the population . The negative correlation observed between loss rate and the age of pseudogenes is consistent with this model . The premise that pseudogenes are removed because their encoded products are either energetically costly or toxic relies on the assumption that , after their initial disruption , these sequences are still being transcribed and translated . The majority of pseudogenes that we analyzed are newly arisen ( i . e . , have a single inactivating mutation in their coding regions ) , and since the mutational target of the regulatory portion of a gene is much smaller than the coding region , they are unlikely to harbor mutations that affect their expression . This is confirmed by the fact that most of these pseudogenes have nearly 100% sequence identity to their functional orthologs across the entire upstream intergenic region ( i . e . , from the end of the anchoring gene to the start codon ) . As originally observed in E . coli [34] , [35] , and recently shown to occur in other bacteria [36] , virtually all ( even antisense ) sequences in bacterial genomes are transcribed . Direct evidence of pseudogene expression is available for several strains of Salmonella . In a global analysis of Typhimurium gene expression using microarrays , Hautefort et al . [37] reported values for the relative expression of about 4 , 000 genes during host-cell infection . Some pseudogenes were up-regulated ( e . g . , putA , rffH ) , and others were down-regulated ( e . g . , dgoA ) , more than two-fold under the experimental conditions . An RNA-seq analysis of Typhi found that many pseudogenes were transcribed , albeit at highly reduced levels [38] . In this analysis , nine Typhi pseudogenes– both young and old – were still expressed high levels , but the overall reduction in pseudogene expression was taken to indicate that the majority of pseudogenes were no longer active [38] , possibly to ameliorate the deleterious effects that we detected . To determine if reduced expression fosters the maintenance of pseudogenes in bacterial genomes , we examined the codon adaptation index ( CAI , which is an indicator of overall expression levels over evolutionary timescales ) of genes in the difference age classes . Paralleling the effect shown in Figure 4 , the average CAI is significantly lower in older pseudogenes ( average CAI = 0 . 29 for age class 3 vs . 0 . 36 for age class 1; p = 0 . 003 , one-tailed unpaired t-test assuming unequal variance ) . These results are consistent with our expectation that selection acts to remove more highly expressed ( and connected ) genes once they become pseudogenized . Mutations in bacterial genomes are known to be highly biased toward deletions [18] , [19] . Therefore , it is not surprising to find that accumulation of deletions is the primary force responsible for the erosion of bacterial pseudogenes . However , only a small fraction of pseudogenes detected in Salmonella genomes were found to have lost more than 20% of their original length , despite the high sensitivity of our synteny-based method for pseudogene detection . Given the high incidence of kilobase-sized ( and larger ) deletions observed during Salmonella experimental evolution [24] , the main mechanism for the complete removal of pseudogenes is likely to be large deletions , most of which are large enough to remove neighboring genes and therefore cannot be detected using a local-synteny based approach . Our systematic characterization of multiple Salmonella genomes indicates that the evolution of bacterial pseudogenes is not strictly neutral such that newly formed pseudogenes have a higher likelihood of being removed . This deviation from the generally accepted view that pseudogenes represent completely neutral regions [20] is likely due to the fact that bacteria have haploid genomes and generally large effective population sizes , therefore increasing the exposure of mutations to selective forces . If pseudogenes are deleterious due either to the energetic costs of transcription and translation or to the dominant-negative effects of anomalous proteins , the high efficacy of selection in bacterial genomes is likely to have a role in their removal . This is consisitent with our finding that those Salmonella genomes with the lowest genome-wide Ka/Ks ratios denoting a relatively high efficacy of selection harbor the lowest numbers of pseudogenes . Because all bacterial groups , as well as those Archaea examined , display a mutational pattern that is biased towards deletions [18] , [19] , [33] and their haploid genomes would be more susceptible to dominant-negative effects that pseudogenes might impart , it is likely that the process of adaptive removal of pseudogenes is pervasive among prokaryotes . And given the evidence for selection on intron size in some eukaryotic genomes , presumably due to the energetic cost of transcription [39] , these effects need not be restricted to those cellular organisms with haploid genomes , and pseudogene degradation and removal may be found to be operating under similar principles in representatives from all domains of life .
We obtained the complete genome sequences of six Salmonella enterica strains from NCBI GenBank [40] , including S . enterica subsp . enterica serovar Enteritidis str . P125109 ( NC_011294 ) , S . enterica subsp . enterica serovar Gallinarum str . 287/91 ( NC_011274 ) , S . enterica subsp . enterica serovar Choleraesuis str . SC-B67 ( NC_006905 ) , S . enterica subsp . enterica serovar Typhimurium str . LT2 ( NC_003197 ) , S . enterica subsp . enterica serovar Typhi str . CT18 ( NC_003198 ) , and S . enterica subsp . arizonae serovar 62:z4 , z23:– ( NC_010067 ) as the outgroup . This set of genome sequences were selected because: ( 1 ) the low level of divergence allows for reliable sequence alignment and thus confident inference of gene inactivation events , ( 2 ) the phylogenetic relationship among the six strains allows for straightforward assignment of age-class for pseudogenes base on their phylogenetic distribution pattern , and ( 3 ) the sequencing was performed by high-coverage whole-genome shotgun sequencing with the Sanger method , which provides high accuracy in homopolymer regions . The last point was of particular importance because our preliminary analysis suggests that sequencing errors in homopolymer regions are a major factor that contributes to erroneous pseudogene annotations in several other S . enterica genome sequences . The difficulties involved in distinguishing true frameshift mutations from sequencing errors prohibit a more comprehensive taxon sampling . To identify orthologous gene shared among the six S . enterica genomes , we performed all-against-all NCBI-BLASTN [41] searches with an e-value cutoff of 1×10−15 for all annotated protein-coding genes . A set of custom Perl scripts written with Bioperl modules [42] were used for data parsing and processing . The BLASTN results were supplied as the input for OrthoMCL [43] to perform ortholog clustering . The algorithm is largely based on the popular criterion of reciprocal best hits between genomes and has been shown to perform well by a benchmarking study [44] . To infer the phylogenetic relationship among the six S . enterica strains , we aligned the nucleotide sequences of the 2 , 898 single-copy genes shared by all six strains using MUSCLE [45] with the default parameters . We used TREE-PUZZLE [46] to infer the distance matrix and the phylogenetic tree based on a concatenated alignment with 2 , 772 , 598 sites . The changes from default setting in TREE-PUZZLE include: ( 1 ) use exact parameter estimates , ( 2 ) estimate the nucleotide frequencies and transition/transversion ratio from the data set , ( 3 ) use a mixed model with one invariable and eight Gamma rates for rate heterogeneity , and ( 4 ) estimate the fraction of invariable sites and the Gamma distribution parameter from the data set . We calculated the genome-wide Ka/Ks ratio for each of the five ingroup strains to estimate the level of genetic drift experience by the lineage . This ratio is a good approximation for the level of genetic drift because it measures the efficacy of purifying selection in protein-coding region; an elevated level of genetic drift can result in increased incidence of slightly deleterious amino acid replacement , and thus , an increase in genome-wide Ka/Ks ratio . Although positive selection favoring certain amino acid changes can also increase Ka , such scenario is expected to be limited to particular genes and sites rather than driving changes throughout the entire genome [1] , [47] . For each of the 2 , 898 single-copy genes shared by all six strains , we performed multiple sequence alignment at amino acid level using MUSCLE [44] with the default parameters . The resulting protein alignments were converted into codon-based nucleotide alignment using PAL2NAL [48] . To account for possible base composition and codon usage bias in any of the genes examined , we applied the YN00 method [49] implemented in the PAML package [50] to estimate the substitution rates . For each of the five ingroup strains , we calculated Ka and Ks using the outgroup S . enterica subsp . arizonae as the reference . To avoid potential bias in Ka/Ks ratio estimation due to non-sufficient sequence divergence or saturation , we removed genes that have an estimated Ks of less than 0 . 1 or greater than 1 . 5 in any of the five pair-wise comparisons . The average Ka/Ks ratio calculated from the 2 , 290 remaining genes was used to represent the genome-wide estimate for each of the five ingroup strains . We utilized a synteny-based approach similar to that described previously [19] for pseudogene identification . Although this approach may underestimate the total number of pseudogenes in a genome due to the exclusion of pseudogenes that lack positional homologs in other closely related genomes ( which may have originated from horizontal transfer ) , the stringent requirement allows for confident inference of the gene inactivation events . To identify pseudogenes with positional homologs , each of the five S . enterica subsp . enterica strains was used as the query against every other genome . The outgroup S . enterica subsp . arizonae was not considered as a query because the ancestral state of any pseudogene identified in this genome cannot be established with our taxon sampling . For each pair of query and subject , we utilized single-copy genes shared by the two genomes as anchors to systematically examine the intergenic regions in the query genome . An intergenic region is flagged as containing a putative pseudogene if an annotated protein-coding gene was found in the syntenic region of other genomes . For each candidate region , we aligned the query genome to the subject genome using MUSCLE [45] with the default parameters . The two anchoring genes were included to improve the quality of alignment and to allow for examination of the entire intergenic region of the query genome . Possible gene-inactivating mutations , including insertions , deletions , pre-mature stop codons , and/or point mutations in the start codon were inferred based on the annotated gene in the subject genome . The results were manually inspected for the consistency regarding gene synteny and the identified mutations across different reference genomes . To ensure a high level of confidence when inferring gene-inactivating mutations , we required at least two positional homologs to establish the ancestral state of a pseudogene . During our curation process , the following types of false-positives were removed before the final analysis: ( 1 ) the entire open reading frame is intact in the query genome but not annotated as a gene , ( 2 ) the putative pseudogene may be explained as an annotation artifact ( e . g . , the region was annotated as a part of either anchoring genes in the query genome ) , ( 3 ) pre-mature stop codons are the only type of mutations and the protein lengths were reduced by less than 20% , ( 4 ) the reference gene is a transposase from a insertion sequence element or of viral-origin ( i . e . , likely a gene gain in the subject genome instead of a gene loss in the query genome ) , ( 5 ) the phylogenetic distribution of the reference gene suggests that a single gene gain event ( e . g . , horizontal gene transfer ) is the most likely explanation for the absence of corresponding gene in the query genome , and ( 6 ) the reference gene is a poorly conserved hypothetical protein and is shorter than 300 bp . In the rare cases where the identified mutations exhibit inconsistency across different reference genomes or indicate extensive sequence divergence , we extracted the syntenic region from all genomes to perform multiple sequence alignment and deduced the inactivation events based on the most parsimonious scenario . One special case involved a 873-bp inversion within the srfB pseudogene in the Gallinarum genome; we manually corrected the inversion before the multiple sequence alignment to infer other possible inactivation events . Furthermore , the exact boundaries of all indel events that affect the 5′- or 3′-end of a pseudogene were manually verified . To classify the curated pseudogenes into different age classes , we examined their phylogenetic distribution pattern to characterize the likely time point of gene inactivation events on the phylogeny . Because 302 out of the 378 curated pseudogenes are specific to one genome , we used two additional methods for age class assignments . In the first method , we categorized the pseudogenes based on the number of gene-inactivating mutations that have been accumulated . In the second method , we utilized an ortholog in the outgroup to quantify the level of accelerated sequence divergence in the pseudogene relative to its functional ortholog in another genome . The nucleotide sequence alignments were inferred using MUSCLE [45] with the default parameters and subsequently used as the input for TREE-PUZZLE [46] to calculate distance matrices . Due to the lack of appropriate orthologs in the outgroup , only 227 out of the 378 curated pseudogenes can be classified using this method . To infer the potential role of a Salmonella pseudogene in cellular fitness , we identified the orthologous gene in Escherichia coli str . K-12 substr . MG1655 ( NC_000913 ) , a related enteric strain on which extensive experimental and functional assays have been conducted [51] , [52] . For ortholog identification , we used the full-length gene from the closest reference genome as the query to perform NCBI-BLASTP [40] searches . To qualify as an ortholog , we required the BLASTP hit to satisfy all of the following conditions: ( 1 ) is the best hit among all of the protein-coding genes in the E . coli MG1655 genome , ( 2 ) has an BLASTP e-value of less than 1×10−15 , ( 3 ) the difference in gene length is no more than 20% of the shorter sequence , ( 4 ) the high scoring pairs ( HSPs ) account for at least 80% of the shorter gene , and ( 5 ) the fraction of conserved amino acid sites is at least 60% within HSPs . For pseudogenes that had a corresponding ortholog in the E . coli MG1655 genome , we extracted the protein-protein interaction information from the high quality combined dataset available from an integrated protein interaction database [53] , [54] to infer numbers of interacting partners . | Pseudogenes have traditionally been viewed as evolving in a strictly neutral manner . In bacteria , however , pseudogenes are deleted rapidly from genomes , suggesting that their presence is somehow deleterious . The distribution of pseudogenes among sequenced strains of Salmonella indicates that removal of many of these apparently functionless regions is attributable to their deleterious effects in cell fitness , suggesting that a sizeable fraction of pseudogenes are under selection . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genetics",
"and",
"genomics/microbial",
"evolution",
"and",
"genomics",
"evolutionary",
"biology/microbial",
"evolution",
"and",
"genomics",
"evolutionary",
"biology/evolutionary",
"and",
"comparative",
"genetics",
"evolutionary",
"biology/genomics",
"genetics",
"and",
"geno... | 2010 | The Extinction Dynamics of Bacterial Pseudogenes |
Alveolar echinococcosis ( AE ) is a life-threatening disease caused by larvae of the fox-tapeworm Echinococcus multilocularis . Crucial to AE pathology is continuous infiltrative growth of the parasite's metacestode stage , which is driven by a population of somatic stem cells , called germinative cells . Current anti-AE chemotherapy using benzimidazoles is ineffective in eliminating the germinative cell population , thus leading to remission of parasite growth upon therapy discontinuation . We herein describe the characterization of EmPlk1 , encoded by the gene emplk1 , which displays significant homologies to members of the Plk1 sub-family of Polo-like kinases that regulate mitosis in eukaryotic cells . We demonstrate germinative cell-specific expression of emplk1 by RT-PCR , transcriptomics , and in situ hybridization . We also show that EmPlk1 can induce germinal vesicle breakdown when heterologously expressed in Xenopus oocytes , indicating that it is an active kinase . This activity was significantly suppressed in presence of BI 2536 , a Plk1 inhibitor that has been tested in clinical trials against cancer . Addition of BI 2536 at concentrations as low as 20 nM significantly blocked the formation of metacestode vesicles from cultivated Echinococcus germinative cells . Furthermore , low concentrations of BI 2536 eliminated the germinative cell population from mature metacestode vesicles in vitro , yielding parasite tissue that was no longer capable of proliferation . We conclude that BI 2536 effectively inactivates E . multilocularis germinative cells in parasite larvae in vitro by direct inhibition of EmPlk1 , thus inducing mitotic arrest and germinative cell killing . Since germinative cells are decisive for parasite proliferation and metastasis formation within the host , BI 2536 and related compounds are very promising compounds to complement benzimidazoles in AE chemotherapy .
The metacestode larval stage of the fox-tapeworm E . multilocularis is the causative agent of alveolar echinococcosis ( AE ) , a life-threatening zoonosis prevalent in the Northern Hemisphere [1] , [2] . Infection of the mammalian intermediate host ( rodents , humans ) is initiated by oral uptake of ‘infectious eggs’ , which contain the oncosphere larva . Upon hatching from the egg in the host intestine , the oncosphere penetrates the intestinal epithelium and gains access to the host organs . Typically within the liver , the parasite then undergoes a developmental transition towards the metacestode stage which is entirely driven by parasite stem cells ( germinative cells ) that have been carried to the host by the oncosphere [2]–[4] . As an asexual multiplication stage , the metacestode tissue grows multivesicularly and infiltratively , like a malignant tumor , into the surrounding host tissue , eventually leading to organ failure [1]–[4] . In natural rodent infections , head regions of the future adult worm ( protoscoleces ) are formed from germinative cells of the cellular layer ( germinal layer; GL ) of the metacestode , and are subsequently taken up when the definitive host takes its prey [4] . In human infections , asexual parasite growth occurs similar to the situation in rodents , but protoscoleces are only formed in rare cases [1] . Human AE is very difficult to treat and only in ∼20% of cases the parasite mass can be removed by surgery [1] , [2] , [5] . In all other cases , chemotherapy is the only option of treatment and is currently mainly carried out using benzimidazoles , which target parasite β-tubulin [5] . However , β-tubulins between parasite and host are highly similar [6] , [7] , so that only parasitostatic concentrations of these drugs can be applied to prevent significant adverse side effects [1] , [5] , [8] . Hence , although the introduction of benzimidazole chemotherapy in the 1990s has significantly improved patient's prognosis , treatment has to be applied for years to decades , and in many cases even life-long [1] , [5] , [8] . Furthermore , a significant number of patients cannot tolerate benzimidazole treatment at all [1] , [5] . Several attempts to improve anti-AE chemotherapy by identifying novel anti-parasitic drugs are currently undertaken [8] , [9] but , so far , no reliable alternative to benzimidazole treatment is available . This underscores an urgent need for novel chemotherapeutic options against AE . A hallmark of both free-living and parasitic flatworms is the employment of a population of totipotent stem cells ( called ‘neoblasts’ in free-living species ) that decisively contribute to the enormous regenerative capacity and developmental plasticity of this group of organisms [4] , [10] . In Echinococcus , earlier ultrastructural studies also strongly suggested the presence of undifferentiated stem cells ( ‘germinative cells’ ) in parasite larvae and is has been hypothesized that they might contribute to parasite growth [11] , [12] . By establishing cultivation techniques for germinative cells , we later demonstrated their proliferative potential and showed that they can form mature metacestode vesicles in vitro [13] . Very recently , we also demonstrated that germinative cells are the only proliferative cells in parasite larvae , that they give rise to all differentiated cells , and that there are important differences between the stem cell populations of E . multilocularis to those of the related schistosomes , and to neoblasts of free-living flatworms [14] . Since the germinative cells are absolutely decisive for asexual multiplication of the E . multilocularis metacestode , they constitute one of the most important cell types for the development of chemotherapeutics to prevent parasite proliferation . Polo-like kinases ( PLKs ) are serine/threonine kinases ( STK ) that act as important regulators of cell-cycle progression in all eukaryotic lineages [15]–[17] . They are particularly important in the M-phase during which they regulate the assembly of the spindle apparatus and the activation of cyclin-dependent protein kinases ( CDC ) [15]–[17] . In humans , five PLKs are expressed of which Plk1-3 are very similar in structure . They comprise a conserved N-terminal STK domain , necessary for phosphorylation of downstream molecules , and two C-terminal Polo-box domains ( PBD ) , which govern protein-protein-interaction and subcellular localization [15]–[17] . The so far best investigated PLK is mammalian Plk1 , which is mainly expressed in late G2 and M phases and regulates both mitosis and meiosis [15] , [16] . Most importantly , Plk1 activates the dual-specific phosphatase Cdc25C , which dephosphorylates , and thus activates , the maturation promoting factor ( MPF ) , resulting in nuclear MPF translocation . Since Plk1 is highly expressed in proliferating cells , including many cancer cells , it has already been intensely validated as an anti-cancer drug target , and several inhibitors that specifically inhibit Plk1 activity are available [15] , [18] . In contrast to Plk1 , Plk2 and Plk3 are encoded by early response genes that are activated during serum stimulation of cells , and both proteins are involved in checkpoint-mediated cell cycle arrest [15] . Plk4 is a divergent member of the PLK family and differs from Plk1-3 in its domain composition . Plk4 shares little overall homology with Plk1-3 , and only contains one PBD [15] . Plk4 gene expression increases from late G1 to S phase , and the protein is known to be involved in centriole duplication , and most probably also plays a role in chromosome maturation and mitotic progression [15] . Recently , a fifth PLK was identified in mammals , Plk5 , which is localized to the nucleolus and whose expression is induced by stress conditions and DNA damage [15] . PLKs with structural and functional homologies to mammalian Plk1 and Plk4 have already been described in invertebrate model systems such as Drosophila melanogaster and Caenorhabditis elegans [17] . In free-living flatworms , functional studies on Polo-like kinases have not yet been carried out , although their transcripts have been specifically detected in neoblasts and reproductive organs of the planarian Schmidtea mediterranea [19] , [20] . In parasitic flatworms , PLKs have so far exclusively been investigated in the trematode Schistosoma mansoni [17] , [21] , [22] . In a first study , Long et al . [21] characterized SmPlk1 , which displays considerable homologies to mammalian Plk1 , and demonstrated expression of the respective gene in female vitelline cells and oocytes as well as in male spermatocytes , indicating a role of SmPlk1 in schistosome mitosis and/or meiosis . Interestingly , the PLK inhibitor BI 2536 , which was originally designed to inhibit human Plk1 [23] , induced dramatic alterations in schistosome gonads in vitro , which affected oogenesis and spermatogenesis at 100 nM concentrations [21] . Very recently , the same authors also characterized a Plk4-like PLK in S . mansoni , named SmSak , which was mostly expressed in schistosome female ovary and vitellarium , and which interacted with SmPlk1 [22] . In contrast to SmPlk1 , however , the activity of recombinantly expressed SmSak was not affected by Bi 2536 , indicating a high selectivity of this inhibitor for Plk1-like kinases [17] , [22] . The successful identification of SmPlk1 in proliferative cells of S . mansoni , and its inhibition by an available small molecule compound , prompted us to investigate the role of PLKs in asexual growth of E . multilocularis larvae . We herein describe the characterization of a Plk1-like Echinococcus kinase , EmPlk1 , and demonstrate its expression in germinative cells of the Echinococcus larval stages . We show that the activity of EmPlk1 , when heterologously expressed in the Xenopus oocyte system , can be seriously affected by available PLK inhibitors . Furthermore , using in vitro systems for the cultivation of parasite stem cells [13] , [24] and metacestode larvae [25] , [26] we show that the compound BI 2536 significantly inhibits parasite development already at concentrations as low as 20 nM . The potential of PLK inhibitors in anti AE chemotherapy is discussed .
In vivo propagation of parasite material was performed in mongolian jirds ( Meriones unguiculatus ) , which were raised and housed at the local animal facility of the Institute of Hygiene and Microbiology , University of Würzburg . This study was performed in strict accordance with German ( Deutsches Tierschutzgesetz , TierSchG , version from Dec-9-2010 ) and European ( European directive 2010/63/EU ) regulations on the protection of animals . The protocol was approved by the Ethics Committee of the Government of Lower Franconia ( Regierung von Unterfranken ) under permit number 55 . 2-2531 . 01-31/10 . PLK inhibitors BI 2536 [23] and BI 6727 ( Volasertib ) [27] were purchased from Axon Medchem ( Groningen , The Netherlands ) and Selleckchem . com ( München , Germany ) , respectively . Both inhibitors were dissolved in dimethyl sulfoxide ( DMSO ) from Sigma Aldrich ( D8418-50ML ) as 10 mM stock solutions and were stored at −80°C until use ( according to the manufacturer's instructions ) . All experiments were performed using the E . multilocularis isolates H95 ( cloning procedures; drug treatment ) [28] and GH09 ( whole mount in situ hybridization ) [28] . Mongolian jirds ( M . unguiculatus ) were used for in vivo propagation of the parasite by intraperitoneal passages as previously described [26] . Co-cultivation of metacestode vesicles with host cells was carried out essentially as previously described [26] . Axenic cultivation of metacestode vesicles was performed as described by Spiliotis et al . [25] . Echinococcus primary cells were isolated and cultivated under axenic conditions as described by Spiliotis et al . [13] . Conditioned medium ( A4 medium ) was prepared by seeding 1×106 rat Reuber hepatoma cells [26] together with 100 ml DMEM medium ( Life technologies ) in a culture flask , followed by 1 week incubation . Subsequently , the supernatant was removed and sterile filtrated ( A4 medium ) . Inhibitor tests on metacestode vesicles were performed under axenic culture conditions ( nitrogen atmosphere [13] ) in A4 medium as previously described by Hemer et al . [29] and Gelmedin et al . [30] , with up to ten vesicles per well of a 6-well ( 5 ml vol . per well ) culture plate . Primary cell cultures were set up essentially as described by Spiliotis et al . [24] and the amount of isolated cells was subsequently measured indirectly through densitometry . 1 Unit of primary cells was defined as the amount that yields an OD600 of 0 . 02 in Phosphate Buffered Saline ( PBS ) . 50 Units isolated primary cells ( ∼15 . 000 cells ) were then seeded in a 48-well plate with 1 . 3 ml A4-medium . Medium was changed every second day and fresh inhibitor was added . Inhibitors were used in final concentrations of 5 , 10 , 25 , 50 , and 100 nM . All experiments were performed with at least three technical and biological replicates . Linear regression analysis was carried out using MicrosoftExcel-2007 . Error bars in figures represent standard deviation . Differences were considered significant for p values below 0 . 05 ( indicated by ( * ) ) , for p between 0 . 001 and 0 . 01 ( ** ) , and for p<0 . 001 ( *** ) . For p>0 . 5 , differences were considered non-significant . Proliferation of stem cells was investigated by staining of newly synthesized DNA using the Click-iT EdU cell proliferation assay ( Invitrogen , C10337 ) . Metacestode vesicles from hepatocyte co-culture [26] were washed once with 1× PBS and carefully transferred into a 15 ml tube containing 3 ml A4-medium . A final concentration of 50 µM EdU ( Component A ) was added and the metacestode vesicles were incubated for 5 h at 37°C ( with gentle agitation every 45 min ) . For fixation , metacestode vesicles were transferred to a Petri-dish and carefully opened with a syringe tip to remove hydatid fluid . After intense washing with 1× PBS , fixative ( 4% paraformaldehyde ( PFA ) in 1× PBS ) was added and samples were incubated for 1 h at room temperature . The samples were washed again ( 1× PBS ) to remove the fixative , and then transferred to a 1 . 5 ml tube . The EdU labeling procedure was then carried out according to the manufacturer's instructions ( Invitrogen , C10337 ) , but with extended incubation times . Metacestode vesicles were finally analyzed by fluorescence microscopy . For long-term treatment ( 21 days ) with BI 2536 , metacestode vesicles were cultivated in A4-medium for 21 days with inhibitor ( medium and inhibitor change every second day ) , followed by three days recovery without treatment . As a control , medium with and identical amount of DMSO ( without inhibitor ) was used . Randomly chosen metacestode vesicles ( t = 4 ) were isolated and EdU stained as outlined above . Of each metacestode vesicle , three randomly chosen sections of the GL were analyzed by microscopy and EdU positive cells were counted . The number of EdU positive cells was calculated to cells per mm2 of the GL . For short-time inhibitor experiments , metacestode vesicles were treated in vitro for 24 , 48 and 72 hours with 50 nM or 100 nM BI 2536 , followed by 24 h regeneration without inhibitor . EdU staining and calculation of Edu positive cells per mm2 of GL were subsequently performed as described above . The E . multilocularis genome sequence assembly [28] was used as available in GeneDB under http://www . genedb . org/Homepage/Emultilocularis . Published sequences of human Plk1 ( GenBank accession number: P53350 ) and SmPlk1 ( AY747306 ) were employed in extensive BLASTP searches on the genome sequence . Amino acid sequence and domain predictions were carried out using UniProt ( http://www . uniprot . org/ ) and SMART ( http://smart . embl-heidelberg . de/ ) software . Based on the genomic sequence information , primers were designed to amplify the full EmPlk1 coding sequence . Total RNA was isolated from metacestode vesicles with TRIzol reagent and reverse transcribed into cDNA as previously described [30] . The cDNA then served as a template for gene amplification by PCR using primers AKO-053 ( 5′-GAC TTC TGC CCG GGT ATG GAT A-3′ ) and AKO-054 ( 5′-GGA AGA CGG CAA ACA TGT GAT-3′ ) . The PCR product was subsequently cloned into pJET 1 . 2 ( Thermo Scientific CloneJET PCR Cloning Kit ) and sequenced . According to the previously described protocol for heterologous expression of SmPlk1 in Xenopus oocytes [21] , mutated versions of EmPlk1 were produced which included a constitutively active form of the kinase ( T179D ) , a version that cannot be phosphorylated at T179 ( T179V ) , and kinase dead versions of wild-type EmPlk1 ( wtKD ) and T179D ( T179DKD ) . The kinase dead versions were generated by replacing the highly conserved active loop motif D163FG for D163SV . Kinase domain mutants were generated by mutagenesis PCR using the following primer combinations: T179D ( AKO-138 , 5′-GGT GAA ATG AAG AAG GAC TTA TGT GGG ACG CCA AAC TAT ATT GCT CC-3′ and AKO-139 , 5′-CCA CAT AAG TCC TTC TTC ATT TCA CCT TCT TTA GTA ATT CTA G-3′ ) , T179V ( AKO-140 , 5′-GGT GAA ATG AAG AAG GTA TTA TGT GGG ACG CCA AAC TAT ATT GCT CC-3′ and AKO-141 , 5′-CCA CAT AAT ACC TTC TTC ATT TCA CCT TCT TTA GTA ATT CTA G-3′ ) , and D163SV ( AKO-146 , 5′-GAC ATG ATT GTA AAG ATC GGG GAT TCG GTG TTG GCC TCT AGA ATT ACT AAA GAA GG-3′ and AKO-147 , 5′-GTA ATT CTA GAG GCC AAC ACC GAA TCC CCG ATC TTT ACA ATC ATG TCA TCA TTT AAA AAC AGA TTG GC-3′ ) . For mutagenesis , the emplk1 reading frame was cloned into the pBAD TOPO/Thio expression vector ( Life Technologies ) . Mutations were then introduced , employing the above listed primers , using the QuickChange Site-directed mutagenesis kit ( Agilent Technologies ) according to the manufacturer's instructions . All plasmid constructs were finally sequenced to verify wild-type sequences and the successful introduction of the desired mutation . Wild-type and mutant versions of the EmPlk1 reading frame were further cloned into the pSecTag2/Hygro A vector ( Life Technologies ) that contains a T7 promoter sequence . cRNA encoding EmPlk1 proteins was synthesized in vitro using the T7 mMessage mMachine Kit ( Ambion , USA ) and injected in stage VI Xenopus laevis oocytes according to [31] . Heterologous expression of EmPlk1 wild-type and mutant forms in Xenopus oocytes , in vitro treatment of oocytes with the inhibitors BI2536 and BI6727 , and germinal vesicle breakdown assays were conducted essentially as previously described by Long et al . [21] for SmPlk1 . As a positive control for GVBD , oocytes were stimulated with progesterone ( PG ) , the natural inducer . All experiments were carried out on samples composed of 20 oocytes originated from three different Xenopus females . Total RNA was isolated using TRIzol as previously described [32] from 2 , 5 , and 11 day old primary cell cultures , from dormant and pepsin/low pH activated protoscoleces [33] , and from mature metacestode vesicles . Isolated RNA was DNase treated ( RQ1 RNase-free DNase , Promega ) followed by phenol/chloroform extraction . RNA concentration was quantified by spectrophotometry and 750 ng of each stage were used for reverse transcription ( RT ) as previously described [32] . Intron-flanking , gene specific primers for emplk1 ( AKO-57; 5′-GAG CAT GTT CAG TGT GAT GG-3′ and AKO-60; 5′-CGA TCT ATC ATA TCG TAG GCG-3′ ) were used to estimate the semi-quantitative expression profile by PCR employing a protocol of 30 cycles of 30 sec at 94°C , 30 sec at 58°C , and 1 min at 72°C . The constitutively expressed gene elp [34] served as a control . Metacestode vesicles of isolate GHO9 with developing protoscoleces were used for in situ hybridization according to Koziol et al . [14] . A 1 . 3 kb fragment of the emplk1 open reading frame was amplified and subcloned into pDrive ( Qiagen ) using primers AKO-55 ( 5′-CTC TCA TGG AAC TGC ATA AGA G-3′ ) and AKO-60 ( 5′-CGA TCT ATC ATA TCG ATG GCG-3′ ) . Digoxygenin-labelled antisense and sense RNA probes were in vitro transcribed using the T7 or SP6 promoter of the linearized vector as previously described [14] . Detection was performed using anti-digoxygenin antibodies coupled to alkaline phosphatase , and NBT/BCIP as colorimetric substrates , as described [14] . The complete emplk1 cDNA sequence reported in this paper was submitted to the GenBank database and is available under accession number HG931729 .
By BLASTP genome mining of the available E . multilocularis genome sequence [28] ( http://www . genedb . org/Homepage/Emultilocularis ) using the full length sequences of human Plk1 [35] and S . mansoni SmPlk1 [21] as queries , one single locus encoding an Echinococcus Plk1 ortholog ( EmuJ_000471700 ) was identified on chromosome 3 . Due to its homologies ( see below ) , the respective gene was designated emplk1 ( E . multilocularis Polo-like kinase 1 ) encoding the protein EmPlk1 . The emplk1 gene spanned a genomic region of 3 . 174 bp and , like the SmPlk1 encoding gene of S . mansoni [21] , comprised 7 exons , separated by 6 introns . All introns displayed canonical GT-AG dinucleotide sequences at the 5′ splice donor and the 3′ splice acceptor sites . Informed by the emplk1 genomic sequence , primers were designed to PCR-amplify the entire emplk1 reading frame from metacestode cDNA preparations . The cDNA fragment was cloned and fully sequenced , which confirmed the exonic gene sequence as determined by the E . multilocularis genome project . The full length EmPlk1 reading frame comprised 1833 bp and coded for a protein of 610 amino acids , EmPlk1 , with a calculated molecular mass of 69 . 5 kDa . Analysis of the EmPlk1 protein sequence by SMART [36] identified an N-terminal STK catalytical domain ( between Y22 and F274 ) and a C-terminal protein binding domain , which included two Polo box domains ( Figure 1 ) . The EmPlk1 kinase domain displayed sequence motifs characteristic of all 11 typical sub-domains ( I–XI ) of protein kinases and included all residues previously determined to be invariable for protein kinases [37] at the respective positions ( Figure 1 ) . Particularly the sequence motifs HRDLKxxN ( sub-domain VI ) and GTPNYIAPE ( VIII ) are strong indicators for STK activity [37] . In human Plk1 , residue T210 was previously shown to be the major phosphorylation site of mitotic PLKs [15] and a respective threonine residue was also conserved in schistosome SmPlk1 ( T182 ) [21] . In EmPlk1 , this residue was also conserved ( T179; Figure 1 ) . As in SmPlk1 [21] and human Plk1 [35] , the ATP binding site of EmPlk1 contained a GxGGFAxC motif ( Figure 1 ) , which is a PLK-typical variation of the canonical GxGxxGxV motif found in the majority of protein kinases [37] . Overall , the EmPlk1 kinase domain displayed significant homologies to the kinase domains of well investigated PLKs such as S . mansoni SmPlk1 ( 68% identical residues ) , human Plk1 ( 62% ) , Drosophila Polo ( 60% ) , and Xenopus Plx1 ( 60% ) . Protein-protein interaction and cellular localization of PLKs is regulated by the C-terminally located protein binding domain [15] . As typical for Plk1-like PLKs , the EmPlk1 protein binding domain contains two Polo boxes ( Y378-T441 and W475-Y545 ) , which are separated from the kinase domain by a non-conserved linker region . In human Plk1 , three residues ( W414 , H538 , K540 ) have previously been shown to be essential for the binding to phospho-S/T binding motifs , and all three residues are also perfectly conserved in EmPlk1 ( W371 , H499 , K501; Figure 1 ) . Kothe et al . [38] previously identified several amino acid residues that are important for binding of the PLK inhibitor BI 2536 to human Plk1 . In particular , the absence of a bulky side chain at position 132 ( L132 in human Plk1 ) was an important specificity determinant that ensured optimal binding of BI 2536 to the Plk1 subfamily of PLKs [38] . In EmPlk1 , all these residues , including the leucine residue ( L101 ) , were conserved at the respective positions ( Figure 1 ) . Taken together , all above analyses clearly identified EmPlk1 as a member of the Plk1 subfamily of PLKs , indicated that the Echinococcus protein is most probably enzymatically active , and that the PLK inhibitor BI 2536 should be able to bind to the parasite-derived kinase . To determine whether emplk1 is expressed in Echinococcus larval stages relevant to the infection of the intermediate host , semi-quantitative RT-PCR experiments were carried out . As relevant parasite stages , we chose mature metacestode vesicles and protoscoleces before and after activation by pepsin/low pH ( mimics transition into the definitive host ) , to cover late stages of the infection . In our established primary cell cultivation system [13] , [24] , E . multilocularis germinative cells are capable of developing into metacestode vesicles similar to the oncosphere metacestode transition process [7] . Hence , in order to cover early stages of the infection , we also included primary cell cultures at different time points ( 2 , 5 , 11 days ) of development . Total RNA was isolated and , after cDNA preparation , emplk1 gene specific PCR was carried out on serial dilutions . As shown in Figure 2 , emplk1 transcripts could be clearly identified in all larval stages tested , and particularly prominent expression was observed in primary cell cultures , which typically contain large proportions of parasite stem cells [13] , [14] . During the E . multilocularis genome project , preliminary deep sequencing transcriptome data were generated for primary cell cultures ( 2 and 11 days old ) as well as for metacestode vesicles and activated/dormant protoscoleces [28] . When we analyzed these profiles we found particularly high expression in primary cell cultures after 2 days of cultivation , which was reduced in primary cells after 11 days , and basal in metacestode vesicles and protoscoleces ( Figure S1 ) . This verified the RT-PCR data mentioned above and , again , indicated possible stem cell specific expression of emplk1 since young primary cell cultures contain particularly high percentages of germinative cells , which steadily decline during development ( differentiation ) into mature vesicles [14] . We recently investigated cellular proliferation profiles in E . multilocularis development [14] and demonstrated that mitotically active parasite stem cells are distributed throughout the germinal layer , and are strongly accumulated in brood capsule and protoscolex buds . In late stage protoscoleces , stem cells are prominently located at the base of developing suckers , but are also present in the posterior body [14] . To investigate a possible stem cell specific expression of emplk1 , we therefore carried out in situ hybridization experiments using a recently established protocol that is applicable to metacestode vesicles [14] . As depicted in Figure 3 , prominent emplk1 signals were obtained for all regions of parasite larvae that typically contain large numbers of proliferating stem cells , such as early brood capsules ( Figure 3D ) , and developing protoscoleces ( Figure 3E , F ) . Furthermore , germinal layer cells with a distribution highly reminiscent of germinative stem cells stained positive for emplk1 ( Figure 3C ) . Taken together , these data strongly indicated that emplk1 is specifically expressed in parasite stem cells . For functional studies on the schistosome PLK SmPlk1 , the heterologous Xenopus oocyte expression system has previously been employed [21] , and in the present study we used this system to further characterize EmPlk1 . To this end , we first generated a mutant form of EmPlk1 in which T179 of the activation loop was replaced by phospho-mimetic aspartate ( T179D ) , thus yielding a constitutively active form of the enzyme ( similar to [21] ) . We also produced a mutant in which T179 was replaced by valine ( T179V ) , which would prevent phosphorylation , and thus activation , at T179 in Xenopus oocytes . Finally , we produced ‘kinase dead’ versions by replacing the D163FG motif of the active loop by D163SV in both the wild-type and the T179D background ( wtKD; T179DKD ) . In Xenopus oocytes , it has already been demonstrated that the injection of mRNA encoding activated forms of Plx1 ( the Xenopus Plk1-like kinase ) or SmPlk1 can induce meiosis resumption , which results in germinal-vesicle breakdown ( GVBD ) [21] . We therefore injected mRNAs encoding the wild-type and mutant forms of EmPlk1 into Xenopus oocytes and carried out GVBD assays . As shown in Table 1 , injection of ( non-activated ) wild-type EmPlk1 alone was ineffective in inducing GVBD . However , injection of the ( activated ) T179D mutant led to GVBD in 90% of oocytes , which is comparable to activities previously determined for SmPlk1 in this system [21] . The induction of GVBD by an activated mutant of SmPlk1 in Xenopus oocytes has previously been shown to involve increased phosphorylation of Xenopus Cdc25C [21] and , although we did not specifically test this for EmPlk1 , we assume that the Echinococcus enzyme is also able to directly activate the Xenopus downstream target . In the case of the kinase dead version of activated EmPlk1 ( T179DKD ) , on the other hand , only a basal rate ( 10% ) of GVBD was observed ( Table 1 ) , indicating that the kinase activity of EmPlk1 is responsible for meiosis resumption in Xenopus . The T179V version of EmPlk1 was also completely inactive in inducing Xenopus oocyte GVBD , indicating that phosphorylation of EmPlk1 at T179 is essential for enzymatic activity . It should be noted that BI 2536 was rather ineffective in inhibiting GVBD in response to the positive control progesterone ( Table 1 ) . This is in line with previous investigations showing that progesterone-induced GVBD of Xenopus oocytes occurs via several pathways , among which are the cAMP pathway ( involving protein kinase A ) and the mitogen-activated protein kinase ( MAPK ) pathway ( including Mos , an oocyte-specific MAPKKK ) [39] . Furthermore , meiosis entry of Xenopus oocytes in response to progesterone also depends on a balance between Cyclin B synthesis and the activity of Myt1 , a member of the Wee1 family of inhibitory kinases , in a Plx1- and MAPK-independent manner [40] . Hence , although Plx1 is required for the activation of Cdc25C in Xenopus oocytes [41] , leading to accelerated meiosis resumption , Plx1 inhibition can eventually not prevent GVBD . The fact that even high concentrations of BI 2536 do not significantly affect progesterone-induced GVBD in our experiments thus indicates that this compound is not generally cytotoxic to Xenopus oocytes and that the BI 2536 effects on EmPlk1 T179D-induced GVBD solely result from the specific inhibition of the parasite enzyme . Taken together , these experiments verified that EmPlk1 is an enzymatically active kinase that can stimulate meiosis resumption in the Xenopus oocyte system similarly to SmPlk1 . Having established that EmPlk1 can induce GVBD in Xenopus oocytes , we then tested whether an available inhibitor , BI 2536 , originally designed against human Plk1 [23] , is also able to affect the parasite protein . As shown in Table 1 , already at concentrations as low as 5 nM , the capability of EmPlk1 to induce GVBD in Xenopus oocytes started to diminish , and was completely abolished at concentrations of 50 nM or higher . Similar results were observed when we used BI 6727 , another Plk1-specific inhibitor with improved pharmacokinetic profile [27] , although in this case slightly higher concentrations were necessary to induce EmPlk1 inhibition ( data not shown ) . Taken together , these experiments indicated that EmPlk1 can be inhibited by BI 2536 and BI 6727 in a concentration dependent manner . Having established that the PLK inhibitor BI 2536 affects the enzymatic activity of EmPlk1 , we next investigated whether this compound could also inhibit the formation of metacestode vesicles from parasite stem cells . To this end , we set up Echinococcus primary cell cultures and measured the formation of mature metacestode vesicles in the presence of different concentrations of BI 2536 . As shown in Figure 4 , metacestode formation was already significantly reduced in the presence of 5 nM BI 2536 , while concentrations of 25 nM or higher almost completely prevented parasite development . We also tested BI 6727 in the primary cell culture system and , as BI 2536 , this inhibitor prevented metacestode vesicle formation in a concentration dependent manner , although at slightly higher concentrations than BI 2536 ( data not shown ) . Since the formation of metacestode vesicles in the primary cell system crucially depends on proliferating stem cells [14] , and since we had already shown that EmPlk1 is specifically expressed in this cell type ( see above ) , we therefore concluded that the inhibition of EmPlk1 by BI 2536 and BI 6727 either resulted in stem cell killing , or at least prevented stem cell proliferation , in the primary cell system . We then tested the effects of BI 2536 on mature metacestode vesicles , which is the actual target stage of chemotherapy in AE [1] , [5] . Interestingly , as shown in Figure S2 , even incubation for 21 days in the presence of 100 nM BI 2536 did not lead to structural disintegration or collapse of metacestode vesicles which , however , had slightly reduced sizes and were no longer capable of growing . This was similar to an approach we had recently undertaken concerning metacestode treatment with the ribonucleotide reductase inhibitor hydroxyurea ( HU ) , which is toxic to cells that undergo proliferation [14] . Although HU treatment specifically eliminated germinative cells in metacestode vesicles , these remained structurally intact for several weeks , indicating that this parasite stage is able to survive for long periods under conditions of slow cellular turnover [14] . We therefore tested whether BI 2536 treatment of metacestode vesicles might specifically eliminate the germinative cell population and carried out EdU incorporation experiments . To this end , metacestode vesicles were incubated for 21 days in presence of different concentrations of BI 2536 . After recovery for 3 days without inhibitor , proliferating cells were detected by EdU pulse labeling [14] . As shown in Figure 5 , control vesicles displayed a typical pattern of proliferating germinative cells within the germinal layer . In samples treated with 10 nM BI 2536 , however , the number of proliferating stem cells was reduced by ∼50% . A statistically significant reduction to ∼10% of normal numbers was observed after incubation with 25 nM BI 2536 and in the presence of higher inhibitor concentrations ( 50 , 100 nM ) , cell proliferation in the germinal layer was completely abolished ( Figure 5 ) . Like in our previous experiments using HU as an inhibitor [14] , no effects of BI 2536 treatment were observed on differentiated cells . As expected from the absence of proliferating cells , even after 3 weeks of further cultivation , these vesicles did not resume growth or proliferation capacity ( data not shown ) . Using vesicles after 21 day treatment with 50 or 100 nM BI 2536 as a source for parasite primary cell cultivation , we also never obtained cultures that formed typical cell aggregates [13] , [14] or mature metacestode vesicles ( data not shown ) . Finally , we also carried out short term treatment ( 24 , 48 , 72 h ) of metacestode vesicles with 50 nM and 100 nM BI 2536 concentrations , followed by 1 day recovery and 5 h EdU pulses . As shown in Figure 6 , already after 24 h in the presence of 50 nM BI 2536 , the number of proliferating cells within the germinal layer was reduced to ∼17% of the normal proliferating cell number , and it was further diminished after incubation for 72 h or at higher concentrations ( 100 nM; Figure 6 ) . Taken together , the experiments outlined above clearly indicated that BI 2536 specifically targeted the germinative stem cell population of metacestode vesicles and either led to stem cell killing or long-term mitotic arrest .
The pathology of AE is crucially linked to continuous asexual growth of the E . multilocularis metacestode stage within the intermediate host's liver , accompanied by metastasis formation in secondary organs [1] , [2] , [4] , [5] . Since somatic stem cells are typically employed for cellular proliferation in flatworms ( reviewed in [10] ) , it has already very early been suggested that a population of undifferentiated cells , which forms part of the Echinococcus GL , is responsible for both tumor-like , infiltrative growth of parasite larvae within host organs , and for metastases formation after distribution through the lymphatic system of the host [11] , [12] , [42] . We previously developed methods to isolate and cultivate these cells , and showed that they are capable of producing new metacestode tissue when kept in culture in the presence of host-derived feeder cells [13] . Very recently , we also showed that these cells exhibit a typical stem cell character and that subpopulations of these cells express several genes of the nanos and argonaute family that are typical components of the germline multipotency program of metazoan stem cells [14] . We also demonstrated that this cell type , called ‘germinative cells’ , is the only cell type capable of proliferation in E . multilocularis larvae and that it produces all differentiated cell types present in the metacestode [14] . Since germinative cells are capable of producing new metacestode tissue even when removed from their normal tissue context in the germinal layer [13] , this cell type thus constitutes a crucial target for the development of anti-Echinococcus drugs that aim to prevent parasite proliferation and metastasis formation . For decades , anti-AE chemotherapy has relied on benzimidazoles ( mostly albendazole ) [5] , and although other compounds with anti-parasitic activities are currently subject to intense research [8] , [9] , [43] , no reliable alternative to benzimidazole treatment is currently in the pipeline . Unfortunately , little information is available concerning direct effects of benzimidazoles on Echinococcus germinative cells . However , when we use albendazole in cell killing assays on freshly isolated primary cell cultures , which contain up to 80% germinative cells [14] , little or no effects are detected , even at high drug concentrations ( Hemer , Brehm , unpublished results ) . Indications for limited activity of benzimidazoles on parasite stem cells were also obtained in earlier studies by Ingold et al . [44] and Stettler et al . [45] who found that albendazole derivatives only affected the germinative cells ( called ‘undifferentiated cells’ in these publications ) at late time points of in vitro treatment of metacestode vesicles , and much less than other compounds such as nitazoxanide . These in vitro studies were recently verified in vivo by Küster et al . [46] . The fast recurrence of parasite growth in patients after discontinuation of albendazole chemotherapy [1] , [5] , [8] , which has to rely on surviving germinative cells [14] , could therefore be due to limited activities of benzimidazoles against this particular cell type . The molecular basis for this limited efficacy could be stem cell-specific expression of parasite β-tubulin isoforms that are resistant to inhibition by benzimidazoles [7] . We have previously characterized three E . multilocularis β-tubulin isoforms of which one , Tub-2 , displayed amino acid sequence motifs that indicated limited interaction with benzimidazoles [6] and in transcriptome analyses collected during the E . multilocularis genome project [28] , the Tub-2 encoding gene ( tub-2 ) displayed highest expression in parasite stages that are enriched in germinative cells ( Figure S3 ) . Furthermore , in preliminary transcriptome analyses on E . multilocularis larvae ( unpublished results ) , we already obtained evidence for stem cell-specific expression of tub-2 . Hence , apart from limited bioavailability of benzimidazoles at the site of infection [8] and adverse side effects due to high homologies between host- and parasite β-tubulin [6] , one of the drawbacks of benzimidazole chemotherapy could be limited efficacy against germinative cells since these express a potentially resistant β-tubulin isoform . In the present work , we present information on a druggable enzyme that fulfils a crucial role in Echinococcus germinative cell proliferation . Our structural analyses clearly identified EmPlk1 as a member of the Plk1-like subfamily of PLKs , with all protein domains and catalytic residues that are typical for this enzyme family at the corresponding positions . By heterologous expression in Xenopus oocytes , we also demonstrated that EmPlk1 is an active kinase that can induce meiosis resumption and GVBD . RT-PCR analyses , transcriptome data , and in situ hybridization experiments further indicated that EmPlk1 is specifically expressed in Echinococcus germinative cells that are present in the germinal layer of the metacestode and in developing protoscoleces . The stem cell-specific expression of EmPlk1 is further supported by preliminary deep sequencing transcriptome data of our group which show that HU treated , and thus stem cell depleted , metacestode vesicles [14] are dramatically reduced in emplk1 transcripts when compared to untreated metacestode vesicles ( unpublished data ) . Based on these data and on the conserved functions of Plk1-like kinases in other metazoans [15]–[18] , we propose that EmPlk1 fulfils an important function in Echinococcus germinative cells , particularly in dividing stem cells during G2/M phase transition . An important upstream interaction partner of human Plk1 is the kinase Aurora A , which phosphorylates , and thus activates , Plk1 at T210 [15] . Since we have shown that EmPlk1 also requires activation at a corresponding threonine residue ( T179 ) we propose that a similar activation mechanism also exists in Echinococcus cells , and according to the genome sequence [28] , the parasite indeed encodes a gene encoding an Aurora A-like kinase ( EmuJ_001059700 ) . Important downstream factors for human Plk1 are the tyrosine phosphatase Cdc25c ( also called M-phase inducer phosphatase ) which directs dephosphorylation of cyclin B-bound CDK1 ( cyclin-dependent kinase 1 ) , thus triggering entry into mitosis [15] , the transcription factor forkhead box M1 ( FoxM1 ) , which regulates the expression of a cluster of G2/M target genes , and the tumor suppressor p53 [15] , [16] . Orthologs to these factors are also present in the Echinococcus genome , such as an M-phase inducer phosphatase gene ( EmuJ_001174300 ) , a p53 ortholog ( annotated as p63; EmuJ_000098700 ) , and several genes encoding forkhead transcription factors ( e . g . EmuJ_000620400 ) . Notably , according to preliminary transcriptome data ( unpublished results ) , the genes encoding homologs to Aurora A , p53 , and several forkhead transcription factor genes are expressed in a germinative cell-specific manner and could thus , together with emplk1 , form a regulatory network that controls the Echinococcus stem cell cycle similar to the situation in humans [15] , [16] . In S . mansoni , Long et al . [22] recently characterized a Plk4-like PLK , named SmSak , which interacts with SmPlk1 in Xenopus oocytes , and which is co-expressed with SmPlk1 in the female ovary and vitellarium . These authors showed that SmSak can be activated following its interaction with SmPlk1 , indicating a potential role of SmSak in schistosome meiosis . SmSak was not , however , inhibited by PLK-inhibitors directed against Plk1-like family members . Together with their previous characterization of SmPlk1 [21] , these authors thus demonstrated that schistosomes employ an invertebrate typical set of PLKs , consisting of one member of the Plk1 and one member of the Plk4 sub-families . In our BLASTP analyses , we also noted the presence of gene encoding a second PLK in E . multilocularis ( EmuJ_000104700 ) . However , although the kinase domain of the encoded protein displayed similarity to PLK kinase domains , overall sequence similarity of this protein was highest with Plk4 subfamily members ( including SmSak ) and the protein obviously lacked conserved C-terminal PBDs ( data not shown ) . Of the 8 amino acid residues known to be involved in the binding of Plk1 inhibitors ( such as BI 2536 ) to human Plk1 [38] , which were all conserved in EmPlk1 , only one was conserved in the kinase domain of this putative EmSak . In the position corresponding to L101 of EmPlk1 , which in human Plk1 determines BI 2536 specificity [38] , the kinase domain of this protein contains a bulky residue ( phenylalanine ) . We therefore believe that the protein encoded by EmuJ_000104700 fulfils similar activities as SmSak , particularly in centriole duplication [22] , and might even interact with EmPlk1 , but that none of the activities of BI 2536 discussed below are due to inhibition of the kinase activity of this protein . Due to its overexpression in many human tumors , Plk1 has been extensively studied as a target for anti-cancer therapy and a number of compounds that either act as ATP-competitive inhibitors or interfere with Polo-box domain functions of Plk1 have already been identified [15] , [18] . One of the best studied compounds in this regard is the ATP-competitive inhibitor BI 2536 , which causes apoptosis and prometaphase arrest in a variety of tumor cell lines [15] , [18] , [47] , [48] . In several phase I and phase II clinical trials against lung or pancreatic cancer , BI 2536 was well tolerated , albeit with varying success rates , depending on the nature of the tumor [15] , [18] , [23] , [48]–[52] . In in vitro studies , BI 2536 usually displays activities against human tumor cell lines in an IC50 range between 5 and 175 nM [15] , [18] , [47] , [48] and , depending on the intravenous doses given , tolerable plasma concentrations of BI 2536 in clinical trials on cancer patients vary between 20 and 200 nM [23] , [49]–[52] . This is well within the range of BI 2536 activities that we observed herein against EmPlk1 and E . multilocularis larvae . As we have shown , concentrations as low as 20 nM BI 2536 almost completely inhibited the activity of EmPlk1 in Xenopus oocytes , and at higher concentrations ( 50 , 100 nM ) , EmPlk1 was no longer able to induce GVBD . Since all amino acid residues previously determined to mediate binding of BI 2536 to human Plk1 are also highly conserved in EmPlk1 , it is reasonable to assume that BI 2536-mediated inhibition of both enzymes follows a similar ATP-competitive mechanism . We further demonstrated that BI 2536 concentrations of 25 nM and higher were very effective in preventing metacestode vesicle formation from parasite germinative cells , and in depleting metacestode vesicles of germinative cells . The specific elimination of germinative cells in metacestode vesicles , which otherwise remained intact for several weeks , is yet another indicator for stem cell specific expression of EmPlk1 , and we propose that the BI 2536 effects we observed in the primary cell system are also due to EmPlk1 inhibition in the stem cell population . It is not yet clear whether BI 2536 treatment of parasite larvae induces germinative cell killing or just ( transient ) mitotic arrest . However , even weeks after treating metacestode vesicles with 50 or 100 nM BI 2536 , we never observed growth resumption or a re-population of the GL with germinative cells . Furthermore , it was not possible to set up proliferating parasite primary cell cultures from vesicles that had been treated with 50 or 100 nM BI 2536 . Together with observations that BI 2536 can induce apoptosis and severe phenotypes in human cancer cells [15] , [18] , we therefore hypothesize that BI 2536 treatment of E . multilocularis indeed led to germinative cell killing , or at least to permanent mitotic arrest . Taken together , we herein present a promising target for the development of anti-echinococcosis drugs that specifically affect the germinative ( stem ) cell system of the parasite and , thus , would ideally complement anti-parasitic activities of benzimidazoles . On the one hand , BI 2536 itself could already be administered to infected mice , combined with benzimidazoles , to study possible additive effects . Respective experiments are currently planned in our laboratory . Furthermore , BI 2536 could serve as a lead compound for the identification of drugs that are more specific to the parasite Plk1 when compared to human Plk1 . Although the kinase domains of both enzymes are homologous ( 62% identical residues ) , they are clearly more divergent than are host and parasite β-tubulins ( >90% [6] ) , and should contain structural differences that can be exploited for parasite-specific drug design . This possibility is supported by the fact that the second generation Plk1 inhibitor BI 6727 , which has similar affinities and a similar binding mode to human Plk1 as BI 2536 [27] , was less effective in preventing metacestode formation and in inhibiting EmPlk1 in the Xenopus system than BI 2536 and also shows 4- and 11-fold less selectivity against human Plk2 and Plk3 , respectively , than BI 2536 [27] . Despite these somewhat lower activities of BI 6727 in eliminating Echinococcus stem cells in vitro , it should not , however , be dismissed as a potential anti-echinococcosis drug due to its clearly improved pharmacokinetic profile and the fact that it can be given orally [27] . Finally , our characterization of EmPlk1 as a factor that governs the mitotic activity of E . multilocularis germinative cells will form a solid basis for further investigations into the regulation of the unique stem cell system of this parasite . | The lethal disease AE is characterized by continuous and infiltrative growth of the metacestode larva of the tapeworm E . multilocularis within host organs . This cancer-like progression is exclusively driven by a population of parasite stem cells ( germinative cells ) that have to be eliminated for an effective cure of the disease . Current treatment options , using benzimidazoles , are parasitostatic only , and thus obviously not effective in germinative cell killing . We herein describe a novel , druggable parasite enzyme , EmPlk1 , that specifically regulates germinative cell proliferation . We show that a compound , BI 2536 , originally designed to inhibit the human ortholog of EmPlk1 , can also inhibit the parasite protein at low doses . Furthermore , low doses of BI 2536 eliminated germinative cells from Echinococcus larvae in vitro and prevented parasite growth and development . We propose that BI 2536 and related compounds are promising drugs to complement current benzimidazole treatment for achieving parasite killing . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"cell",
"cycle",
"and",
"cell",
"division",
"cell",
"processes",
"tropical",
"diseases",
"parasitic",
"diseases",
"parasitology",
"developmental",
"biology",
"clinical",
"medicine",
"neglected",
"tropical",
"diseases",
"pharmacol... | 2014 | Targeting Echinococcus multilocularis Stem Cells by Inhibition of the Polo-Like Kinase EmPlk1 |
The gaseous hormone ethylene is one of the master regulators of development and physiology throughout the plant life cycle . Ethylene biosynthesis is stringently regulated to permit maintenance of low levels during most phases of vegetative growth but to allow for rapid peaks of high production at developmental transitions and under stress conditions . In most tissues ethylene is a negative regulator of cell expansion , thus low basal levels of ethylene biosynthesis in dark-grown seedlings are critical for optimal cell expansion during early seedling development . The committed steps in ethylene biosynthesis are performed by the enzymes 1-aminocyclopropane 1-carboxylate synthase ( ACS ) and 1-aminocyclopropane 1-carboxylate oxidase ( ACO ) . The abundance of different ACS enzymes is tightly regulated both by transcriptional control and by post-translational modifications and proteasome-mediated degradation . Here we show that specific ACS isozymes are targets for regulation by protein phosphatase 2A ( PP2A ) during Arabidopsis thaliana seedling growth and that reduced PP2A function causes increased ACS activity in the roots curl in 1-N-naphthylphthalamic acid 1 ( rcn1 ) mutant . Genetic analysis reveals that ethylene overproduction in PP2A-deficient plants requires ACS2 and ACS6 , genes that encode ACS proteins known to be stabilized by phosphorylation , and proteolytic turnover of the ACS6 protein is retarded when PP2A activity is reduced . We find that PP2A and ACS6 proteins associate in seedlings and that RCN1-containing PP2A complexes specifically dephosphorylate a C-terminal ACS6 phosphopeptide . These results suggest that PP2A-dependent destabilization requires RCN1-dependent dephosphorylation of the ACS6 C-terminus . Surprisingly , rcn1 plants exhibit decreased accumulation of the ACS5 protein , suggesting that a regulatory phosphorylation event leads to ACS5 destabilization . Our data provide new insight into the circuitry that ensures dynamic control of ethylene synthesis during plant development , showing that PP2A mediates a finely tuned regulation of overall ethylene production by differentially affecting the stability of specific classes of ACS enzymes .
Ethylene gas is a crucial regulator of numerous aspects of plant development and physiology , including germination , seedling growth and morphology , organ senescence and fruit ripening , as well as stress and defense responses [1] . The biosynthetic capacity for ethylene production is nearly ubiquitous throughout the plant body , but biosynthesis is generally maintained at low levels through regulatory circuitry that confers tight control while allowing rapid and dramatic increases under conditions such as wounding or fruit ripening . Ethylene biosynthesis levels change in response to endogenous developmental cues as well as light , temperature , pathogens and other exogenous signals ( reviewed in references [2]–[4] ) . Ethylene is derived from methionine via a well-characterized biosynthetic pathway ( reviewed in references [4]–[7] ) in which the first committed step , conversion of S-adenosyl methionine to 1-aminocyclopropane 1-carboxylate ( ACC ) , is performed by the enzyme ACC synthase ( ACS; see Figure 1 ) . The final step is the conversion of ACC to ethylene , CO2 and cyanide by ACC oxidase ( ACO ) . Under some conditions ( particularly in fruit and flowers ) , ACO activity may be rate limiting , but ACC synthesis is generally the rate-limiting step for ethylene production during vegetative growth . In seedlings , increasing ACS protein levels drive ethylene synthesis to high levels [8]–[10] . ACS isozymes are encoded by a gene family comprising three subclasses defined by the absence or presence of C-terminal phosphorylation motifs ( Figure 1 ) . Type 1 ACS isozymes carry target sites for mitogen-activated protein kinase ( MAPK ) phosphorylation; this MAPK motif lies immediately downstream from a calcium-dependent protein kinase ( CDPK ) phosphorylation site [11]–[13] . Type 2 isozymes carry only the CDPK target motif , and type 3 isozymes carry neither target site . Phosphorylation of the type 1 isozymes ACS2 and ACS6 by stress-responsive MAPKs ( MPK3 and MPK6 ) results in increased ethylene synthesis through protein stabilization [10] , [13] , [14] . Unphosphorylated type 1 isozymes are rapidly turned over via a 26S proteasome-dependent pathway , and the non-catalytic C-terminal region containing the CDPK and MAPK phosphorylation motifs is sufficient to confer instability on reporter protein fusions [10] . CDPK-mediated phosphorylation of LeACS2 , a tomato type 1 ACS , was recently shown to stabilize the enzyme , leading to increased ACS activity and ACC content in wounded tomato tissue [12] , [15] . Type 2 proteins are recruited by ETHYLENE-OVERPRODUCING1 ( ETO1 ) and the ETO1-like EOL1 and EOL2 proteins for ubiquitin-dependent proteolysis [8] , [9] , [16] , [17] . Ethylene overproduction in etiolated eto1 seedlings is caused by decreased proteolytic turnover of type 2 ACS isozymes . The dominant eto2 and eto3 mutations , which alter C-terminal amino acid sequences of ACS5 and ACS9 required for recognition by ETO1 and EOL proteins , also cause ethylene overproduction in etiolated seedlings . Each of these mutations stabilizes the ACSeto protein product by preventing its interaction with ETO1 [8] , [9] , [16] , [18] . The CDPK target motif of type 2 ACS proteins can be phosphorylated , but the role of phosphorylation in regulating ACS accumulation has not yet been established [9] , [11] . Although phosphorylation by CDPK and MAPK kinases has been linked to stabilization of ACS isozymes , ‘partner’ phosphatases acting on ACS isozymes have not been identified . The PPM-type protein phosphatase AP2C1 negatively regulates MPK6 , and overexpression of AP2C1 compromises wound-induced ethylene production and disease resistance [19] . Okadaic acid and calyculin , inhibitors of protein phosphatase 1 ( PP1 ) and PP2A , stabilize LeACS2 in tomato , and okadaic acid treatment causes the accumulation of LeACS2 phosphorylated at the CDPK site [12] . The heterotrimeric serine/threonine protein phosphatase PP2A was implicated in regulation of ethylene synthesis when the PP2A-deficient rcn1 mutant was shown to overproduce ethylene [20] . Because ethylene is a potent inhibitor of cell expansion in dark-grown seedlings , ethylene overproduction results in a characteristic short hypocotyl phenotype in rcn1 plants [20]–[22] . The RCN1 gene encodes one of three regulatory/scaffolding A subunits of Arabidopsis PP2A , and the pleiotropic rcn1 mutant phenotype results from a significantly reduced level of PP2A activity in rcn1 plants [23] , [24] . The experiments described here were designed to test the hypothesis that sustained C-terminal phosphorylation of ACS isozyme[s] due to decreased PP2A activity might account for ethylene overproduction in rcn1 plants . Our results show that PP2A negatively regulates the activity and accumulation of type 1 ACS isozymes . PP2A directly interacts with the ACS6 protein , and PP2A complexes dephosphorylate a carboxy-terminal ACS6 phosphopeptide in vitro . Dephosphorylation requires the PP2A complexes that contain the RCN1 scaffolding subunit . Elicitor-mediated activation of MPK6 , which up-regulates ethylene synthesis via type 1 ACS isozymes , has a reduced effect in the rcn1 mutant background , consistent with the model that the baseline accumulation of phosphorylated type 1 ACS proteins is increased in rcn1 plants . Genetic and molecular data also show that PP2A positively regulates the abundance of type 2 ACS proteins , revealing an unexpected role for PP2A in promoting accumulation of type 2 ACS isozymes . Our findings provide new insight into the finely tuned and phosphorylation-dependent regulation of ethylene synthesis .
Ethylene biosynthesis is enhanced in dark-grown rcn1 seedlings [20] , [22] . Because ACC synthesis is generally the rate-limiting step for ethylene production during vegetative growth , we asked whether ACS activity was increased in rcn1 mutant seedlings . We found that rcn1 mutants in both the Columbia ( Col ) and Wassilewskija ( Ws ) genetic backgrounds exhibited increased ACS enzymatic activities in dark-grown seedlings ( Figure 2 ) . In the Col genetic background , rcn1-6 and eto1 seedlings showed similar ACS activity levels . These data suggest that ethylene overproduction and reduced hypocotyl lengths in rcn1 mutant seedlings result at least in part from increased ACS enzymatic activity . The eto1 mutation causes ethylene overproduction by stabilizing type 2 ACS isozymes [8] , [9] , [16] , [17] . In both rcn1 and eto1 mutants , ethylene overproduction reduces hypocotyl elongation in dark-grown seedlings , causing a short hypocotyl phenotype . To determine whether PP2A-mediated regulation of ethylene synthesis requires the ETO1 protein , we assayed hypocotyl elongation in rcn1 eto1 double mutant seedlings . As expected , hypocotyl lengths in both single mutants were approximately one-half of those exhibited by the wild-type parents ( Figure 3A ) . Seedlings carrying both the rcn1 and eto1 mutations exhibited an extreme reduction in hypocotyl length , corresponding to 25% of the wild-type parent . Double mutant seedlings also exhibited a significant increase in ethylene production above the level observed in either single mutant parent ( p<0 . 001; Figure 3B ) . These data indicate that the rcn1 and eto1 defects in regulation of ethylene synthesis are additive , suggesting independent modes of action for PP2A and ETO1 . Although the ethylene overproduction defect of eto1 seedlings is more severe than that of rcn1 seedlings , the hypocotyl lengths of the two mutants are similar in our assays ( Figure 3B versus 3A ) . However , both the site ( or source tissue ) and timing of ethylene overproduction will affect the overall length of hypocotyls , and the rcn1 and eto1 mutants may differ significantly in these characteristics . To determine whether PP2A-mediated regulation of ethylene synthesis targets particular ACS isozymes , we asked whether acs loss-of-function mutations affect the short hypocotyl phenotype caused by PP2A inhibition . Wild-type seedlings treated with the phosphatase inhibitor cantharidin ( CT ) exhibit a characteristic inhibition of hypocotyl elongation similar to that observed in untreated rcn1 seedlings [21] , [23] , [25] . Hypocotyl elongation in the type 1 acs single mutants acs2 and acs6 and in the acs2 acs6 double mutant showed decreased cantharidin response ( Figure 3C ) . These data suggest that the effect of cantharidin on hypocotyl elongation is dependent on type 1 ACS isozymes . Similarly , ACS enzymatic activity in acs2 acs6 double mutant seedlings was insensitive to cantharidin treatment , while normal seedlings treated with cantharidin exhibited increased ACS activity relative to untreated controls ( data not shown ) . Interestingly , acs5 and acs9 loss-of-function mutants showed slightly increased cantharidin response ( Figure 3C ) , indicating that the type 2 isozymes ACS5 and ACS9 are not required for cantharidin-mediated inhibition of hypocotyl elongation . We also asked whether acs loss-of-function mutations affect ethylene overproduction in the presence of cantharidin . Wild-type seedlings treated with cantharidin exhibited a 76% increase in ethylene production ( Figure 3D ) . While cantharidin-treated acs6 seedlings exhibited a similar increase in ethylene synthesis , both acs2 and acs2 acs6 double mutant seedlings showed a reduced response to cantharidin treatment . Unlike the acs2 and acs6 mutants , acs5 and acs9 mutant seedlings exhibited a reduced baseline level of ethylene production . However , both type 2 acs mutants also showed a greater stimulation of ethylene synthesis in the presence of cantharidin ( Figure 3D ) . Both the decreased basal ethylene synthesis and the increased response to cantharidin were more pronounced in the acs5 acs9 double mutant , which exhibited a three-fold increase in ethylene production when grown in the presence of cantharidin ( Figure S1 ) . As in the hypocotyl elongation experiment described above , our data suggest that phosphatase inhibition acts through type 1 ACS isozymes to increase ethylene synthesis , while type 2 isozymes are not required for this effect . In addition , the data suggest that PP2A inhibition may have opposing effects on the activities of type 1 versus type 2 isozymes . This hypothesis is discussed in more detail below . Phosphorylation-mediated stabilization of type 1 ACS isozymes can be induced by treatment with the bacterial elicitor Flg22 [10] , [13] . Because both the rcn1 mutation and cantharidin treatment decrease PP2A activity levels , we expect that PP2A substrates will be hyperphosphorylated both in rcn1 plants and in cantharidin-treated wild-type plants . We reasoned that the stabilizing effect of Flg22 treatment might be lessened if steady-state ACS phosphorylation levels are increased in the rcn1 mutant . We measured Flg22-induced ethylene synthesis in wild-type ( Col ) and rcn1-6 mutant seedlings , and found that rcn1-6 seedlings indeed showed a reduced response to Flg22 treatment ( Figure 4 ) . While ethylene production rose by more than 2 . 5-fold ( 265% ) in wild-type Col seedlings , the elicitor-induced increase in rcn1-6 mutant plants ( 25% ) was not statistically significant . Analysis of variance ( ANOVA ) shows that the difference between Flg22-treated wild-type ( Col ) seedlings and rcn1-6 seedlings ( Flg22-treated or untreated ) also was not statistically significant . Flg22-induced ethylene synthesis in rcn1-6 plants was not rescued by a higher Flg22 dose; treatment with 200 nM Flg22 resulted in a 262% increase in ethylene production in the wild type , and less than a 30% increase in rcn1-6 plants ( data not shown ) . The attenuated effect of Flg22 treatment in rcn1-6 seedlings is consistent with the hypothesis that type 1 ACS isozymes are more phosphorylated and more stable when PP2A activity is reduced . To directly assess the effect of PP2A inhibition on ACS6 protein turnover , we assayed the stability of an epitope-tagged ACS6 protein [10] in wild-type and rcn1 mutant seedlings , and in wild-type plants grown in the presence of cantharidin . As expected , myc-ACS6 was rapidly turned over in dark-grown seedlings ( Figure 5A ) . The rcn1 mutation retarded the turnover of myc-ACS6 , as did cantharidin treatment ( Figure 5A ) . In contrast , the phosphomimic allele myc-ACS6DDD , which exhibits enhanced and phosphorylation-independent stability [13] , was equally stable in wild-type seedlings grown in the absence or presence cantharidin ( Figure 5B ) . Thus , the wild-type ACS6 protein is stabilized by phosphatase inhibition , while the phosphomimic ACS6DDD protein is immune to this effect . Immunoblots probed with anti-myc antibody revealed two poorly resolved bands near the molecular weight predicted for the myc-ACS6 protein . Extracts from rcn1 mutants and from cantharidin-treated wild-type plants showed enhanced accumulation of the upper band ( Figure 5A , 5C and 5D ) . To determine whether the upper and lower bands corresponded to phosphorylated and unphosphorylated myc-ACS6 , we isolated protein extracts from myc-ACS6-expressing wild-type and rcn1 plants and treated aliquots with alkaline phosphatase ( Figure 5C ) or with PP2A complexes immunoprecipitated from plants expressing an RCN1-YFP fusion protein ( Figure 5D ) . Treatment with either CIP or PP2A resolved the upper and lower bands into a single species running at the position of the lower band ( Figure 5C and 5D ) . These results are consistent with the hypothesis that both the rcn1 lesion and phosphatase inhibition result in the accumulation of a more stable phosphorylated ACS6 species in vivo . We asked whether the ACS6 protein interacts with PP2A complexes in planta . To address this question , we used a reciprocal co-immunoprecipitation approach . First we used anti-myc antibodies to immunoprecipitate ACS6 from protein extracts isolated from plants expressing myc-ACS6DDD ( Figure 6A ) . We probed these immunoprecipitates for regulatory A subunits of PP2A using anti-RCN1 antibodies that recognize all three A subunit isoforms [23] , [26] . Regulatory A subunits were detected in the pellet fraction even after a detergent wash , suggesting a stable interaction between ACS6DDD and PP2A . When anti-RCN1 antibodies were used for the reciprocal co-immunoprecipitation , immunoblotting revealed the presence of the myc-ACS6DDD protein in the immunoprecipitates , and the ACS6 signal was maintained through a stringent detergent wash ( Figure 6B ) . A weak but reproducible ACS6 signal was observed in anti-RCN1 immunoprecipitates isolated from plants expressing wild-type myc-ACS6; this signal was reduced after a gentle washing treatment , and was nearly undetectable after a stringent wash ( Figure 6C ) . These co-immunoprecipitation data suggest that PP2A directly interacts with ACS6 , and that interaction with the low-abundance wild-type isoform is unstable , while the interaction with the stabilized and more abundant ACS6DDD protein is more robust . While the ACS6DDD protein is an imperfect proxy for the phosphorylated wild-type ACS6 protein in this experiment , PP2A interaction may require motifs outside the MAPK phosphorylation site that are unaffected by the DDD substitution . For instance , in the case of PP2A-mediated dephosphorylation of the brassinosteroid-responsive transcription factor BRASSINAZOLE-RESISTANT1 ( BZR1 ) , a distinct binding motif is required for the PP2A interaction that leads to dephosphorylation of multiple sites; some or all of these sites lie outside the binding domain [27] . To determine whether PP2A acts directly on ACS6 , we asked whether immunoprecipitated PP2A complexes could dephosphorylate an ACS6 phosphosubstrate . Because both native and recombinant ACS6 proteins are very unstable , we used a synthetic C-terminal peptide in our dephosphorylation assays . A C-terminal ACS6 30mer containing the three MPK target motifs was phosphorylated with recombinant MKK4DD and MPK6 [13] and then used as a substrate for immunoprecipitated PP2A complexes isolated from transgenic plants expressing a YFP-tagged RCN1 ( Figure 7A ) . PP2A immunocomplexes showed dephosphorylation activity against the ACS6 C-terminal peptide ( Figure 7B ) . When the phosphatase inhibitor okadaic acid was added to the dephosphorylation reactions , activity dropped to background levels ( data not shown ) . In combination with our finding that ACS6 and PP2A interact physically ( Figure 6 ) , these data suggest that ACS6 is a bona fide PP2A substrate in etiolated seedlings . To determine whether a specific population of PP2A complexes may dephosphorylate ACS6 , we tested the activities of crude lysates extracted from plants carrying mutations in genes encoding the three scaffolding A subunits of PP2A . Plants carrying the rcn1 mutation maintain expression of the PP2AA2 and PP2AA3 scaffolds , while plants carrying the pp2aa2-1 pp2aa3-1 double mutant combination express only the RCN1 scaffold [26] . Okadaic acid-sensitive ACS6 dephosphorylating activity was observed in extracts from wild-type and pp2aa2-1 pp2aa3-1 seedlings , but not in extracts from rcn1 plants ( Figure 7C ) . In replicate experiments , extracts from pp2aa2-1 pp2aa3-1 mutants showed 78±7% of wild-type activity levels , while rcn1 mutants showed only background ( 0±1% ) levels of activity ( n = 4 ) . These data indicate that RCN1-containing PP2A complexes are required for dephosphorylation of the ACS6 C-terminus . In contrast to the results obtained with this ACS6 peptide substrate , rcn1 mutant plants exhibit approximately a two-fold decrease in ‘bulk’ PP2A activity ( measured against the model substrate myelin basic protein; [22] , [23] ) , while pp2aa2-1 pp2aa3-1 plants exhibit 89±5% of wild-type bulk activity ( J . Blakeslee , J . Heath and A . DeLong , unpublished ) . Thus our data indicate a specific requirement for the RCN1 scaffold in ACS6 dephosphorylation . These data are consistent with the specificity we observe in regulation of hypocotyl elongation; in contrast to the characteristic short hypocotyl phenotype of rcn1 mutant plants ( 58 . 6±4 . 1% of wild-type ) , pp2aa2-1 pp2aaa3-1 seedlings exhibit nearly normal hypocotyl lengths ( 96 . 7±3 . 6% of wild-type ) . We observed a strikingly different effect of PP2A inhibition on ACS5 protein levels . As expected , normal seedlings exhibited rapid turnover of a myc-tagged ACS5 protein ( Figure 8A ) . In rcn1 seedlings , myc-ACS5 accumulation was dramatically reduced ( Figure 8A ) . Direct comparisons of myc-ACS5 abundance in serial extract dilutions indicate that myc-ACS5 accumulation is reduced at least 25-fold in rcn1 seedlings ( see Figure S2B ) . We observed a similar decrease in myc-ACS5 accumulation when wild-type myc-ACS5 plants were treated with cantharidin ( see Figure S2D ) . Although the reduced baseline accumulation of ACS5 in this experiment precludes accurate measurements of turnover under conditions of phosphatase inhibition , ACS5 protein does not appear to be stabilized by the rcn1 mutation or by cantharidin treatment , but rather appears to be de-stabilized . These experiments employed an inducible myc-ACS5 construct [8] and reduced myc-ACS5 accumulation was apparent across a wide range of induction levels ( 25 to 200 nM Dexamethasone; see Figure S2A ) . At higher levels of induction , accumulation of the myc-ACS5 protein was more easily detected in rcn1 mutant seedlings . Although turnover in wild-type seedlings was impeded at this expression level , myc-ACS5 was clearly unstable in the rcn1 mutant plants ( Figure S2C ) . Strikingly , accumulation and turnover of the stabilized myc-ACS5eto2 protein product was not affected by the rcn1 mutation ( Figure 8B ) . This result argues that the decreased accumulation of wild-type myc-ACS5 results from ACS5 protein instability , rather than any effect on the Dex-inducibility of transgene expression in rcn1 plants . It also demonstrates that PP2A-dependent regulation of ACS5 accumulation requires the C-terminal sequences that are recognized by ETO1 . These observations are consistent with the increased cantharidin responsiveness of ethylene production and hypocotyl elongation we observed in acs5 and acs9 mutant plants ( Figure 3C and 3D ) ; in type 2 acs mutants , baseline ACS activity levels are reduced and the relative effect of stabilizing type 1 isozymes is exaggerated . Our data suggest that loss of PP2A activity simultaneously increases accumulation and activity of type 1 isozymes while reducing accumulation and activity of type 2 isozymes . Because normal baseline ACS activity levels are very low in dark-grown seedlings , stabilization of a single isozyme type has an obvious effect on ethylene production , even when other isozymes are destabilized . In effect , the increase in ethylene production due to stabilization of type 1 isozymes masks the destabilization of type 2 enzymes in hypocotyl elongation and ethylene biosynthesis experiments , and in the eto1 epistasis experiment .
Low levels of ethylene biosynthesis are characteristic of etiolated A . thaliana seedlings , and previous work has identified protein turnover mechanisms that limit the accumulation of ACS isozymes . In conjunction with the tight regulation of ACS mRNA levels [28] , [29] , these mechanisms constitute a stringent control system that regulates ethylene production [8] , [10] , [12] , [13] , [16] , [30] . MAPK-mediated phosphorylation antagonizes the turnover mechanism that controls the stability of type 1 ACS isozymes in seedlings [10] , [13] , [19] . Our data indicate that RCN1-containing PP2A complexes dephosphorylate and promote the turnover of type 1 ACS isozymes in etiolated seedlings , suggesting that PP2A-mediated protein dephosphorylation is an important counterbalance to MAPK action . Conversely , PP2A appears to positively regulate the accumulation of type 2 ACS isozymes . Thus the control systems for type 1 and type 2 isozymes are independently specialized , but both involve PP2A action ( Figure 9 ) . Under natural conditions , down-regulation of ethylene synthesis is necessary to allow the rapid hypocotyl cell expansion that ensures the emergence of seedling shoot tissues from the soil . Ethylene overproduction in plants with reduced PP2A activity results in a characteristic short hypocotyl phenotype [22] , [25] . Exploiting that phenotype in our genetic analysis , we found that the ACS2 and ACS6 genes are required , while the ETO1 , ACS5 and ACS9 genes are dispensable , for increased ethylene synthesis under conditions of PP2A inhibition . Direct analysis of ethylene production in acs loss-of-function mutants also demonstrated the requirement for type 1 but not type 2 isozymes . ACS enzyme activity levels are increased by rcn1 mutations and by cantharidin treatment . These results support a model in which PP2A inhibition allows accumulation of phosphorylated and stabilized type 1 ACS isozymes . Further support for this model comes from our analysis of elicitor-induced ethylene production in wild-type and rcn1 mutant plants . Wild-type plants exhibit a dramatic increase in ethylene production after Flg22 treatment , while rcn1 plants , in which baseline ethylene production is elevated above the wild-type level , show only a modest increase . Turnover of the wild-type ACS6 protein is retarded in rcn1 mutant plants and in cantharidin-treated wild-type plants , while the stabilized ACS6DDD protein shows little or no effect of cantharidin treatment . The RCN1 protein interacts with both wild-type ACS6 and with the stabilized ACS6DDD protein; as might be predicted for a substrate interaction , binding to the wild-type ACS6 protein appears quite unstable . Immunoprecipitated PP2A complexes dephosphorylate a MAPK- phosphorylated ACS6 C-terminal peptide . Finally , analysis of A subunit mutants shows that the RCN1 scaffolding subunit is required for dephosphorylation of the ACS C-terminal peptide , while loss of the PP2AA2 and PP2AA3 scaffolds has only a modest effect on dephosphorylation . This specific requirement for RCN1-directed dephosphorylation in vitro is mirrored by RCN1-specific regulation of hypocotyl elongation in vivo . Together these data suggest that PP2A complexes containing the RCN1 regulatory subunit dephosphorylate type 1 ACS isozymes , and that increased phosphorylation and stabilization of these enzymes allows increased ethylene synthesis in rcn1 seedlings . Recent work in tomato fruit indicates that LeACS2 , a type 1 ACS isozyme of tomato , is stabilized when phosphorylated on both the CDPK and MAPK phosphorylation target sites [12] . Treatment with a protein phosphatase inhibitor promotes the accumulation of LeACS2 that is phosphorylated at the CDPK target site , increasing ACS activity levels . ( The effect of protein phosphatase inhibition on phosphorylation at the MAPK sites was not directly assayed in that work . ) The effect of phosphorylation at the putative CDPK site of A . thaliana type 1 ACS isozymes has not yet been tested . However , substitution of phosphomimic residues in the MAPK site is sufficient to dramatically increase the stability and accumulation of ACS6 , suggesting that CDPK-dependent phosphorylation is not limiting for ACS6 stability in seedlings . The mechanism by which phosphorylation stabilizes ACS isozymes has not been clearly defined . The non-catalytic carboxy-terminal domain of ACS6 is sufficient to confer 26S-proteasome-dependent instability on GFP and luciferase reporters , and it has been suggested that this region acts as a flexible docking domain that extends from the catalytic core . The distribution of acidic and basic residues in this region influences the degree of stabilization observed in the phosphomimicking ACS6DDD mutant [10] , consistent with the idea that phosphorylation at both the CDPK and MAPK sites could contribute to type 1 isozyme stability . Phosphorylation at the CDPK target site in type 2 isozymes was postulated to affect interactions with the ETO1/EOL-containing E3 ubiquitin ligase complex , but non-phosphorylatable and phosphomimic alleles of ACS5 show normal interactions with ETO1 and its paralogs in yeast 2-hybrid assays [9] , indicating that modification at this site is not sufficient to regulate this critical interaction . Binding of 14-3-3 proteins to ACS isozymes also has been detected [31] and may play a role in phosphorylation-dependent stabilization . Unexpectedly , PP2A appears to play a positive role in regulating the accumulation of ACS5 , a type 2 isozyme . Thus the net result of phosphatase inhibition on ACS activity levels in wild-type plants represents the sum of two different effects: increased accumulation of type 1 isozymes and decreased accumulation of type 2 isozymes . The rcn1 defect dramatically reduced the accumulation of ACS5 in plants carrying an inducible transgene construct , indicating that PP2A affects some post-transcriptional mechanism required for ACS5 accumulation . Our data suggest that type 2 ACS proteins are less stable when PP2A activity is reduced , but it is unclear whether this mechanism involves direct action on type 2 isozymes or dephosphorylation of a component of the ETO1 complex ( Figure 9 ) . We have not yet determined whether ETO1 plays a role in PP2A-mediated ACS5 stabilization . Recent proteomic profiling has identified the ETO1-like EOL1 and EOL2 proteins as well as one representative of each ACS isozyme type ( ACS6 , 7 , 8 ) as 14-3-3 omega-binding clients , suggesting that these proteins are phosphorylated in vivo [31] . Since type 3 isozymes do not possess a C-terminal phosphorylation motif , these data suggest that phosphorylation in the conserved catalytic domain of some isozymes also may contribute to ACS regulation . The apparent enhancement of ethylene overproduction in cantharidin-treated acs5 and acs9 loss-of-function mutants indicates that PP2A function affects the activity of type 2 isozymes under native expression conditions as well . Although only ACS1 and ACS9 were found to make statistically significant contributions to control of hypocotyl elongation in 3-day old etiolated seedlings [32] , our analysis of ethylene production shows that both ACS5 and ACS9 play important roles in ethylene synthesis in 5-day old seedlings , with ACS2 and ACS6 contributing little , when PP2A activity levels are normal . For both isozyme classes , fine-tuning of the activity levels requires protein phosphorylation/dephosphorylation and involves RCN1-regulated PP2A function . Interestingly , when acs2 acs6 double mutants are treated with cantharidin , ethylene production is slightly increased ( Figure 3D ) . If overall ethylene production in acs2 acs6 double mutants were solely dependent on type 2 ACS isozymes , we would predict that phosphatase inhibition would decrease ethylene synthesis . Phosphorylation-dependent regulation of the poorly understood type 3 isozymes may contribute to the residual cantharidin-induced ethylene synthesis observed in acs2 acs6 double mutants . Additionally , recent analysis of single , double and multiple acs mutants has demonstrated that there is a complex interplay between ACS isozymes [32] , and it is possible that a compensatory mechanism is activated in acs2 acs6 double mutants . Earlier work suggests that ACS mRNA levels also are affected by reversible protein phosphorylation [33] . The data reported here for accumulation of ACS5 and ACS6 proteins are derived from transgenic lines that employ constitutive ( 35S::myc-ACS6 ) and glucocorticoid-inducible ( myc-ACS5 ) promoter fusions , and thus reflect effects of PP2A function that are independent of native ACS mRNA levels . Moreover , preliminary analysis of ACS mRNA levels suggests that ACS6 transcript levels are normal , while ACS5 and ACS9 transcript levels increase , in rcn1 mutant plants ( M . Soruco and A . DeLong , unpublished ) . Since our genetic analysis indicates that ethylene overproduction requires ACS2 and ACS6 , but not ACS5 and ACS9 , these results suggest that effects on mRNA accumulation do not account for the ethylene overproduction phenotype of rcn1 mutant seedlings . Similarly , analysis of enhanced LeACS2 accumulation after phosphatase inhibitor treatment indicates that mRNA levels remain unchanged while protein stability is increased [12] .
To generate dark-grown seedlings for physiological and biochemical analysis , A . thaliana seeds were surface-sterilized , suspended in 0 . 1% agar and stratified for 3 days at 4°C before plating . Seedlings were grown on 0 . 5× Murashige and Skoog ( MS ) salts with 1% sucrose and 1% agar . After sowing , seeds were given a 16-hour light treatment before transfer into the dark for germination and growth at 24°C on vertical plates . Hypocotyl elongation , ACS activity and protein accumulation phenotypes were scored 5 days post germination . For hypocotyl elongation assays , approximately 30 seeds were sown on each plate; at the conclusion of the growth period , plates were imaged on a flatbed scanner . Hypocotyl lengths were measured using Image J . Cantharidin responsiveness of hypocotyl elongation was assayed in four separate experiments; results shown in Figure 3 represent the aggregate values ± standard error . The rcn1-1 mutant [24] and the rcn1-1 RCN1-YFP transgenic line [21] are in the Wassilewskija ( Ws ) genetic background . The rcn1-6 mutant [21] and all other mutants and transgenic lines used in this work are in the Columbia ( Col ) genetic background . Transgenic 35S::myc-ACS6 , 35S::myc-ACS6DDD and the acs2 and acs6 mutant lines [10] as well as the acs2 acs6 double mutant [14] were the kind gift of S . Zhang ( University of Missouri , Columbia ) . In acs loss-of-function experiments , we also used the acs5-3 ( cin5 ) and acs9-1 alleles [28] , [30] . Crossing rcn1-1 and eto1-1 generated the rcn1 eto1 double mutant in a mixed Ws/Col background . Double mutant F3 families were compared to the parental single mutants and to single mutant sibling families segregating out of the cross . For ACS enzymatic activity assays , etiolated seedlings were harvested and ground in liquid nitrogen , resuspended in ACS protein extraction buffer [13] and assayed for ACS activity as described previously [8] . After chemical conversion of ACC to ethylene , 750 µl of headspace was transferred to a new vial for analysis on a Voyager portable gas chromatograph ( PhotoVac Inc . ) . All reactions were carried out in triplicate . For transgenic 35S::myc-ACS6 lines , MG-132 pre-treatment was used to promote protein accumulation [10] . For transgenic myc-ACS5 lines , seedlings were grown in the presence of low concentrations of Dexamethasone as previously described [8] . At the beginning of each turnover assay , seedlings were washed 3 times in liquid MS medium for 5 minutes and resuspended in liquid MS containing 1 mM cycloheximide . Samples were harvested and flash-frozen in liquid nitrogen in the dark at the specified time points and stored at −80°C until ACS stability was analyzed via immunoblotting with anti-myc antibody . For immunoblotting experiments , seedlings were ground to a fine powder in liquid nitrogen and boiled for 10 minutes with 4× SDS loading buffer ( 240 mM Tris pH 6 . 8 , 8% SDS , 40% glycerol , 0 . 04% bromophenol blue , 5% beta-mercaptoethanol ) . Extracts were centrifuged at 16 , 000× g for 15 minutes at 4°C and supernatants were harvested for immediate use or storage at −20°C . Extracts were separated by electrophoresis on a 10% SDS-polyacrylamide gel , then transferred to a PVDF membrane ( Millipore ) . Detection of proteins was performed using monoclonal anti-myc 9E10 ( Covance ) , antisera against phospho-enol pyruvate carboxylase ( anti-PEPC , Rockland ) or polyclonal anti-RCN1 antibodies [23] and standard chemiluminescence . For immunoprecipitation of PP2A complexes , RCN1-YFP seedlings were grown in the dark for 5 days on MS plates . Seedlings were then harvested and ground in liquid nitrogen , thawed in co-IP buffer ( 50 mM Tris , pH 7 . 5 , 100 mM NaCl , 0 . 3 M sucrose , 0 . 2% Triton X-100 , 2 µg/ml aprotinin and leupeptin ) . Extracts were centrifuged for 15 minutes at 16 , 000× g at 4°C to pellet debris , and the protein concentration of the supernatant was adjusted to 1 . 67 mg/ml with ice-cold co-IP buffer . For each precipitation , 250 µg of protein extract was incubated with 200 µl Protein A agarose ( Invitrogen ) plus 50 µl of 1 . 5∶100 dilution of anti-GFP antibody ( AbCam ) for 1 hour at 4°C . Immunoprecipitates were harvested by centrifugation at 1 , 000× g for 3 minutes at 4°C and washed twice in ice-cold co-IP buffer . For peptide dephosphorylation experiments , immunoprecipitates were resuspended in 200 µl of PPAB and aliquots were added to dephosphorylation reactions . For co-immunoprecipitation of ACS6 and PP2A , myc-ACS6 and myc-ACS6DDD seedlings were grown in the dark on MS plates for 4 . 5 days and then transferred to liquid MS media containing MG-132 for 16 hours to allow ACS protein accumulation [10] . Protein extracts were prepared and processed with anti-myc or anti-RCN1 antibodies as described above , and immunoprecipitates were eluted from the Protein A agarose with SDS loading buffer , boiled and analyzed by immunoblotting as described above . Protein extracts ( 62 . 5 µg total protein ) from MG-132 treated myc-ACS6 seedlings were treated for 10 minutes at 37° with a phosphatase or with extraction buffer ( 100 mM HEPES , pH 7 . 5 , 5 mM EDTA , 5 mM EGTA , 1 mM PMSF , 2 mM benzamidine , 2 µg/ml aprotinin and leupeptin ) alone . For CIP treatment , 5 units of alkaline phosphatase ( CIP , New England Biolabs ) were added . For PP2A treatment , anti-GFP antibodies were used to immunoprecipitate PP2A complexes from seedlings expressing the RCN1-YFP fusion as described above; 10 µl of PP2A immunocomplexes were added to the myc-ACS6 extract . After phosphatase treatment , samples were boiled in SDS loading buffer and the migration of myc-ACS6 was analyzed by immunoblotting . Recombinant His-tagged MKK4DD and MPK6 were purified using Ni-NTA affinity chromatography [13] and used to phosphorylate 500 µg of a biotinylated ACS6 peptide comprising the 30 C-terminal amino acids of ACS6 ( generous gift of S . Zhang , University of Missouri , Columbia ) in a buffer containing 200 µM ATP plus 12 µCi γ-33-P-ATP . Radiolabeled peptide ( 15 pmol/well ) was then bound to a streptavidin-coated 96-well plate ( Thermo Scientific ) and incubated with shaking at 4°C for 2 hours . Each well was washed 4 times for 5 minutes with PPAB ( 50 mM Tris-HCl pH 7 . 0 , 0 . 1 mM EDTA , 5 mM DTT , 0 . 01% Brij-35 ) . For immunocomplex assays , PP2A complexes were isolated from plants expressing an RCN1-YFP fusion protein [21] using either a polyclonal anti-GFP antibody ( AbCam ) or Protein A agarose alone . After a 60-minute immunoprecipitation and two stringent washes at 4°C , IP pellets were resuspended in PPAB and 20 µl aliquots were assayed for ACS6 dephosphorylation activity ( 4 replicate reactions per IP pellet ) . After 15 minutes at 30°C , reactions were stopped with 4× loading buffer and each supernatant was sampled for released counts . Immunoprecipitation fractions also were subjected to immunoblot analysis using anti-RCN1 and anti-C subunit antibodies to confirm isolation of PP2A immunocomplexes . For crude extract assays , dark-grown seedlings were ground in liquid nitrogen and resuspended in PPAB . Extracts were diluted to a protein concentration of 2 . 5 µg/ml in PPAB containing okadaic acid at final concentrations of 0 , 1 or 1000 nM , and 50 µl aliquots were added to 96-well plates pre-bound with 0 . 25 µg phospho-ACS6 peptide . Triplicate reactions were incubated at 30°C for 15 minutes before termination by addition of 4× loading buffer . Each supernatant was sampled for released counts , and the background activity observed at 1000 nM OKA was subtracted from the average values obtained in the presence of 0 and 1 nM OKA . For analysis of the effect of cantharidin on ethylene biosynthesis , seedlings were grown on 3 ml MS medium containing 3 µM cantharidin or a DMSO vehicle control in 22-mL gas chromatography vials for 5 days at 23°C in the dark . For Flg22 induction , seedlings were grown in long days at 23°C in 22-mL gas chromatography vials containing 3 ml MS medium for 12 days . Flg22 peptide ( final concentration 40 µM ) was added at day 12 and the vials were immediately capped and further incubated 4 hrs at 23°C in the light . In both experiments , the accumulated ethylene was measured by gas chromatography as described previously [30] . Flg22 peptide was the kind gift of S . Zhang , University of Missouri , Columbia . | Like animals , plants produce a number of substances that regulate growth and coordinate developmental transitions and responses to environmental signals . Ethylene gas is one such regulator of the plant life cycle , playing important roles in fruit ripening , pathogen defenses , and the regulation of cell expansion . Because overall plant form is determined largely by the degree and directionality of cell expansion , ethylene is a crucial regulator of morphology , and ethylene production must be maintained at low levels during phases of rapid cell expansion , such as early seedling growth . Recent work has identified molecular mechanisms that target ethylene biosynthetic enzymes for proteolytic degradation; this degradation plays a key role in controlling ethylene production . Here we exploit the molecular genetic resources available in the Arabidopsis thaliana system to identify a highly conserved protein complex that dephosphorylates target proteins as a new component of the mechanism that regulates degradation of ethylene-producing enzymes . Our findings show that protein phosphatase 2A plays a nuanced role in this regulatory circuit , with both positive and negative inputs into the stability of specific proteins that drive ethylene biosynthesis . This work enhances our understanding of the mechanisms that enforce adaptive levels of hormone production in plants . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"plant",
"biology/plant",
"growth",
"and",
"development",
"plant",
"biology/plant",
"biochemistry",
"and",
"physiology",
"genetics",
"and",
"genomics/plant",
"genetics",
"and",
"gene",
"expression"
] | 2011 | Protein Phosphatase 2A Controls Ethylene Biosynthesis by Differentially Regulating the Turnover of ACC Synthase Isoforms |
Human cytomegalovirus ( HCMV ) enters host by glycoprotein B ( gB ) -mediated membrane fusion upon receptor-binding to gH/gL-related complexes , causing devastating diseases such as birth defects . Although an X-ray crystal structure of the recombinant gB ectodomain at postfusion conformation is available , the structures of prefusion gB and its complex with gH/gL on the viral envelope remain elusive . Here , we demonstrate the utility of cryo electron tomography ( cryoET ) with energy filtering and the cutting-edge technologies of Volta phase plate ( VPP ) and direct electron-counting detection to capture metastable prefusion viral fusion proteins and report the structures of glycoproteins in the native environment of HCMV virions . We established the validity of our approach by obtaining cryoET in situ structures of the vesicular stomatitis virus ( VSV ) glycoprotein G trimer ( 171 kD ) in prefusion and postfusion conformations , which agree with the known crystal structures of purified G trimers in both conformations . The excellent contrast afforded by these technologies has enabled us to identify gB trimers ( 303kD ) in two distinct conformations in HCMV tomograms and obtain their in situ structures at up to 21 Å resolution through subtomographic averaging . The predominant conformation ( 79% ) , which we designate as gB prefusion conformation , fashions a globular endodomain and a Christmas tree-shaped ectodomain , while the minority conformation ( 21% ) has a columnar tree-shaped ectodomain that matches the crystal structure of the “postfusion” gB ectodomain . We also observed prefusion gB in complex with an “L”-shaped density attributed to the gH/gL complex . Integration of these structures of HCMV glycoproteins in multiple functional states and oligomeric forms with existing biochemical data and domain organization of other class III viral fusion proteins suggests that gH/gL receptor-binding triggers conformational changes of gB endodomain , which in turn triggers two essential steps to actuate virus-cell membrane fusion: exposure of gB fusion loops and unfurling of gB ectodomain .
Human cytomegalovirus ( HCMV ) , a member of the Betaherpesvirinae subfamily of the Herpesviridae family , is a leading viral cause of birth defects [1 , 2] and a major contributor to life-threatening complications in immunocompromised individuals . As one of the largest membrane-containing viruses , HCMV shares a common multilayered organization with all other herpesviruses , composed of an icosahedrally ordered nucleocapsid enclosing a double-stranded DNA genome , a poorly defined tegument protein layer , and a pleomorphic , glycoprotein-embedded envelope [3] . During infection , herpesviruses fuse their envelopes with cell membranes , resulting in the delivery of nucleocapsid into the cytoplasm of the host cells . This complex process requires a number of viral glycoproteins and host receptors functioning in a coordinated manner . Glycoproteins gB and gH/gL are conserved across all herpesviruses and are essential for virus entry into cells [4] . Receptor-binding to gH/gL-containing complexes—the composition of which differs among clinical and laboratory-adapted HCMV strains and across different herpesviruses [5]—triggers conformational changes of fusion protein gB , leading to fusion of the viral envelope with cell membrane [6] . This use of both a fusion protein and a receptor-binding complex for herpesvirus entry differs from many other enveloped viruses , which use a single protein for both receptor binding and membrane fusion . Averaging up to tens of thousands of particle images by single-particle cryoEM method has resolved in situ structures of capsid proteins [7–9] and the capsid-associated tegument protein pp150 [10] , up to atomic resolution [11] . However , such method is not applicable to the studies of herpesvirus gB and other glycoproteins due to their disorganized distribution on the pleomorphic viral envelope . Instead , the structures of gB ectodomains and various forms of gH/gL from herpes simplex virus ( HSV ) [12 , 13] , Epstein-Barr virus ( EBV ) [14 , 15] and HCMV [16 , 17] have been solved by x-ray crystallography . The gB ectodomain structures from these studies share structural similarities to other class III viral fusion proteins in their postfusion conformation [18–20] . Among these proteins , vesicular stomatitis virus ( VSV ) G is the only one whose ectodomain structure has been solved for both prefusion [21] and postfusion [20] conformations , thanks to its pH-reversibility between the two conformations and amenability to crystallization at both high and low pH conditions . At pH 6 . 3–6 . 9 conditions , VSV G has also been observed to exist in monomeric forms both in solution and on virion envelope , possibly representing fusion intermediates [22 , 23] . By contrast , the prefusion conformation of herpesvirus gB is metastable and its structure has been elusive ( even the recent crystal structure of the full-length HSV-1 gB is also in the postfusion conformation [24] ) . While cryo electron tomography ( cryoET ) of HSV-1 virions has revealed glycoproteins on the native viral envelope , poor contrast of cryoET reconstructions makes it difficult to distinguish different glycoprotein structures and conformations [25] . Recent efforts resorted to the use of purified HSV-1 gB-decorated vesicles to visualize the prefusion gB , but its domain assignments have been controversial due to difficulties in interpreting cryoET structures with poor contrast and signal/noise ratio ( SNR ) [26 , 27] . As a result , the mechanism underlying the complex process of receptor-triggered membrane fusion remains poorly understood for not only HCMV , but also for other herpesviruses . Recently , the technologies of electron-counting [28 , 29] , energy filter and Volta phase plate ( VPP ) [30] have significantly improved contrast and SNR of cryoEM images and their combined use in cryoET has led to resolution of two functional states of 26S proteasome in neurons [31] . In this study , we first demonstrated the ability to distinguish prefusion and postfusion conformations of the VSV G trimer ( 171 kD ) in situ by employing a combination of VPP , direct electron-counting , energy filtering and subtomographic averaging . Application of the same approach to HCMV virions has allowed us to identify different conformational states of HCMV gB ( 303 kD ) in their native virion environments and to determine the in situ structure of prefusion gB at a resolution of ~21 Å . Moreover , we also observed prefusion gB forming a complex with gH/gL in situ for the first time . Integration of these structures and knowledge of class III viral fusion proteins has led to a working model of how conformational changes drive membrane fusion during HCMV entry into host cells .
We first established the validity of our cryoET method of combining VPP , direct electron detection , energy filtering , and subtomographic averaging by obtaining in situ structures of class III viral fusion proteins with known structures . Towards this end , we took advantage of the relative simplicity of VSV in having a single 57kD glycoprotein , G , on the viral envelope , with its trimeric structures known for both prefusion and postfusion conformations; and used VSV as a gold standard to validate our method . For VSV at pH = 7 . 5 , tomograms reconstructed from tilt series obtained by 300kV Titan Krios equipped with VPP , energy filter and direct electron detection show excellent contrast , enabling the visualizations of G projecting from viral envelope , the helical nucleocapsid , as well as the internal densities corresponding to polymerases L ( Fig 1A and 1B ) . Two conformations of G are readily differentiable based on the height and shape of the ectodomain: the majority is long ( 12 . 5nm ) and slim , while the minority is short ( 8 . 7nm ) and fat ( Fig 1B ) . Subtomographic averages of 330 long-form particles and 65 short-form particles from five tomograms both contain a prominent ectodomain , with the long one ( ~28 Å resolution ) fit perfectly with the crystal structure of G ectodomain trimer in the postfusion conformation ( Fig 1C–1E ) and the short one with that in the prefusion conformation ( Fig 1G–1I ) . Similar structures were observed in a previous electron tomography study performed on negatively stained sample [32] . Both crystal structures of G contain five domains , DI through DV , despite drastic domain arrangements ( Fig 1E , 1F , 1I and 1J ) . The dramatically different appearances between the two conformations are primarily due to the refolding of the short loop ( residue 273 to 275 ) in DIII , resulting in the elongation of the central helix and a taller postfusion trimer . DIII form the trimeric core in both conformations , buried in the center of the cryoET density map ( Fig 1E and 1I ) . The other domains ( DI , DII and DIV ) undergo a rigid-body type rearrangement—only changing the relative orientations and locations while retaining their domain structures [21] ( Fig 1F and 1J ) . This analysis demonstrates that our cryoET approach incorporating the three cutting-edge technologies can distinguish the two forms of in situ structures of glycoprotein G and allows fitting existing domain structures of individual fusion protein into the density maps for functional interpretation . Next , we applied the same strategy established above to obtain in situ structures of gB and its interaction with gH/gL complex . We imaged virions of the highly passaged laboratory HCMV strain AD169 , taking advantage of its simplicity , as it has lost some glycoprotein genes and does not contain gH/gL/UL128/UL130/UL131A pentamers on its envelope [33] . We recorded cryoET tilt series of HCMV virions with and without VPP in a Titan Krios instrument equipped with an energy filter and a direct electron detector in electron-counting mode . Both the raw images in the tilt series and the reconstructed tomograms show significantly better contrast when VPP was used ( S1 Fig , S1–S4 Movies ) . Typical in virions obtained by high-speed centrifugation , the viral envelopes are pleomorphic and often exhibit membrane blebs likely due to mechanical stress during purification ( Fig 2A ) . [As discussed below , such mechanical stress might also be responsible for triggering some of the “spring-loaded”/higher-energy ( prefusion ) gB to its lower energy ( “postfusion” ) form , which were used as an internal control to validate our cryoET subtomographic averaging method . ] In the tomograms reconstructed from the tilt series obtained with VPP ( referred to as VPP tomograms ) ( Fig 2 ) , three types of enveloped viral particles are readily recognized: virions with C-capsid containing densely-packed dsDNA genome ( Fig 2B ) , non-infectious enveloped particles ( NIEPs ) with B-capsid containing a protein scaffold ( red arrows in Fig 2A ) or with empty A-capsid ( cyan arrow in Fig 2A ) . Inside C-capsids , the dsDNA molecule occupies evenly throughout the entire interior of the capsid with the 20 Å-diameter dsDNA duplex resolved ( Fig 2B ) —the first time such detailed features ever observed directly by cryoET . In B-capsids , the scaffolding protein ( pUL80 , up to 1000 copies/capsid [34 , 35] ) is organized into a density sphere with an outer and inner diameter of ~700 and ~400Å , respectively . In capsids devoid of genome DNA , a portal complex for DNA translocation is visible at one of the 12 vertices of the capsid ( Fig 2C ) . The viral envelope is pleomorphic ( Fig 2A ) and its membrane resolved into two leaflets 40Å apart ( Fig 2D ) , sporting sparsely and randomly located , and clearly identifiable glycoprotein spikes on the outer leaflet ( Fig 2A and Fig 3A ) . We used the following three pieces of evidence to establish the identifications of gB trimers on the viral envelope . First , among HCMV glycoproteins , gB is known to only exist as homotrimer with a combined mass of ~300 kD [36] and is the most abundant complex over 100 kD [37] . This mass is expected to occupy an estimated extracellular volume of ~300 nm3 . Among the density spikes decorating the outer leaflet of the viral membrane , only two differently shaped spikes with such volume were identified , suggesting that they might be gB trimer at different conformational states ( Fig 3B ) . Second , the two distinctive side-view shapes—one triangular , Christmas-tree like ( 71% ) and the other rectangular , columnar-tree like ( 29% ) ( Fig 3B ) —are similar to the side-views of the cryoET reconstructions of HSV-1 gB trimers on purified vesicles in their putative prefusion and postfusion conformations , respectively [27] . Third , we performed subtomographic averaging to these two types of spikes , respectively , in order to examine them with a higher SNR . Both of the averaged models exhibit apparent three-fold symmetry with the symmetric axis perpendicular to the plane of viral membrane , despite slight distortion arising from the inherent “missing wedge problem” of electron tomography ( S2 Fig ) . These three pieces of evidence all point to our tentative assignment of the Christmas tree-shaped and the columnar tree-shaped densities on the HCMV envelope as gB trimers in the prefusion and “postfusion” ( quotation marks are used here since the conformation is not really caused by fusion but likely triggered by mechanical stress during virion purification with high-speed centrifugation ) conformations , respectively . Indeed , as shown below , the available crystal structure of gB in the postfusion conformation matches perfectly with our final subtomographic average of the columnar tree-shaped density , further validating our assignments . As mentioned above , we performed subtomographic averaging to characterize the two putative gB conformations at a higher resolution . The significantly enhanced contrast afforded by imaging with VPP at a near-focus condition allowed the clear visualizations of different structures in the reconstructed tomograms . For direct comparison , we also obtained tilt series without using a VPP ( referred to as non-VPP tomograms ) . For the latter data , we had to use a significantly larger defocus value ( -4μm ) to improve image contrast and record much more tilt series ( 28 total ) in order to obtain a similar number of gB particles for subtomographic averaging due to greater difficulties in distinguishing different glycoprotein morphologies in the tomograms ( S1 Fig ) . In addition , the use of large defocus has necessitated correction for contrast transfer function ( CTF ) : the structure obtained without CTF correction contains phase-inverted , incorrect structure information beyond 25 Å ( S4E Fig ) , as reflected by the broken connections between the ectodomain and the viral membrane in the absence of CTF correction ( S3B , S4C and S4D Figs ) . In total , 350 particles of the columnar tree-shaped and 1509 particles of the Christmas tree-shaped densities were included for subtomographic averaging . For the columnar tree-shaped structure , all particles were extracted from the VPP tomograms due to ambiguities in distinguishing its slender shape from background noise in the non-VPP tomograms . For the Christmas tree-shaped structure , 874 particles , which came from VPP tomograms , were first used and 635 particles from non-VPP tomograms eventually were also included to further improve resolution . Three-fold symmetry was imposed subsequently to improve SNR and the resolution of the averaged structures . Fourier shell correlation ( FSC ) analyses indicate that the resolutions for the symmetrized 3D subtomographic average of the columnar tree-shaped and Christmas tree-shaped spikes are 26 Å and 21 Å , respectively , based on the gold-standard criterion ( S3A Fig ) . The subtomographic average of the columnar tree-shaped spike resolves the two leaflets of the bilayer viral envelope and a prominent ( 161Å in height ) ectodomain ( Fig 3C–3E , S5 Movie ) . The ectodomain density matches well with the crystal structure of the HCMV gB ectodomain trimer [16] ( Fig 3F–3H ) , validating our initial assignment of the columnar tree-shaped density as gB structure in its “postfusion” conformation and re-establishing the validity of our approach . The subtomographic average of our putative prefusion gB densities reveals the two leaflets of the bilayer viral envelope with prominent gB densities attached to both: a prominent ectodomain attached to the outer leaflet ( 130Å in height ) and a globular ( about 35Å in height and 26Å in width ) endodomain to the inner leaflet ( Fig 3I–3K ) . The ectodomain in the putative prefusion gB is shorter than that in the gB “postfusion” conformation and anchors to the membrane with three well-separated densities , forming a tripod ( Fig 3I–3K , S6 Movie ) . Although no crystal structure of prefusion gB is available to fit into our subtomographic average to directly confirm or refute this prefusion gB assignment , it is believed that herpesvirus gB bears structural and mechanistic similarities to other class III viral fusion proteins , which can be used to aid our assignment . Indeed , the postfusion conformation of HCMV gB ectodomain is similar to the postfusion conformations of all other class III viral fusion proteins [18] , including the postfusion VSV G ( Fig 1C–1F ) . The lower portion of the prefusion conformation of the VSV G trimer ( Fig 1G–1I ) has a tripod shape similar to the lower portion of the Christmas tree-shaped density ( Fig 3I ) . The prefusion VSV G trimer is shorter than—and undergoes drastic domain rearrangements towards—its postfusion conformation [20 , 21] ( Fig 1 ) ; likewise , the Christmas tree-shaped density is shorter than the columnar tree-shaped density . Taken together , these characteristic similarities to the prefusion structure of VSV G corroborate our initial assignment of the Christmas tree-shaped density as the in situ prefusion structure of HCMV gB trimer . Structure-guided sequence analysis ( Fig 4A ) indicates that each full-length gB protomer contains an N-terminal ectodomain ( residues 87–705 ) , a membrane proximal region ( MPR , residues 706–750 ) , a single transmembrane helix ( residues 751–771 ) and a C-terminal endodomain ( residues 772–906 ) . For the “postfusion” gB trimer , the ectodomain in the subtomographic average can be divided into a base in contact with the membrane , and two lobes—middle and crown—connected by a neck ( Fig 3F ) . The crystal structure of the ectodomain trimer shows that each protomer consists of five domains: DI , DII , DIII , DIV and DV ( Fig 3G and 3H ) [16] . Except for DV , these domains can be located in our subtomographic average of the “postfusion” gB ( Fig 3F ) . DI , each containing two fusion loops , is located at the base of the trimer; DII and DIV reside , respectively , in the middle and crown lobes , which are connected by DIII in the neck . DV contains a long loop connected by two short helices and is buried , thus is not resolved in our subtomographic average gB trimer due to the limited resolution . As detailed in the Method , we employed a combination of manual rigid-body fitting of known domain structures from the existing HCMV gB postfusion structure [16] , comparative modeling of DIII based on the homologous DIII from VSV G prefusion conformation [21] , followed by optimization by the molecular dynamics flexible fitting ( MDFF ) method [38] , to put forward a provisional domain arrangement model of the prefusion gB ( Fig 5 ) . DV was not considered in our domain modeling of HCMV gB prefusion conformation due to the lack of a template structure , since DV was truncated in the crystal structure of postfusion VSV G . MDFF not only optimized the chemical interactions among the fitted domains , but also improved overall model to map correlation coefficient from 0 . 83 to 0 . 94 ( Fig 5H and 5I , S7 Movie ) . The model from MDFF does not include the MPR ( residues 706–750 ) , which is proposed to lie between the ectodomain and the transmembrane helix ( Fig 4A ) and “mask” the fusion loops to prevent their premature ( non-productive ) association with lipid [39] . Helical wheel projection of the first 15 amino acids of the MPR shows an amphipathic helix ( Fig 4B ) whose hydrophobic side could interact with the fusion loops . This notion is consistent with our interpretation of DI in the subtomographic averages of both prefusion and “postfusion” conformations , with the fusion loops pointing to and in close proximity to the membrane . Among herpesviruses , gB and gH/gL are highly conserved and known to form a fusion machinery for virus entry [40] . Previous biochemical studies have indicated that gH/gL regulated fusion activity of gB [41] and might form a complex with gB in virions on the basis of co-immunoprecipitation experiments [42] . Besides gB trimer densities mentioned above , “L”-shaped spikes were also observed protruding outwards from the viral envelope , which we interpret as gH/gL complexes on the basis of size and shape similarities to the gH/gL crystal structure [13 , 17] . Moreover , among such “L”-shaped spikes , ~7% were observed to be in contact with the Christmas tree-shaped , prefusion gB trimer , forming a gB-gH/gL complex ( Fig 6B and 6C ) , while others were unbound . No “postfusion” gB trimer have been observed involving in gB-gH/gL complex . A subtomographic average was obtained by aligning and averaging 49 such gB-gH/gL complexes to investigate the contact sites between prefusion gB and gH/gL ( Fig 6C–6F ) , with a resolution around 30Å reported by calcFSC in PEET . The HCMV gH/gL crystal structure [17] fits well in the “L”-shaped density in the subtomographic average ( 0 . 75 of the cross-correlation coefficient between the cryoET map and the model filtered to 30Å , Fig 6D and 6E ) . This fitting , together with the predicted domain arrangement in the prefusion gB structure ( Fig 5H and 5I ) , reveals that DI of gB may contact the gH subunit of gH/gL ( Fig 6D ) . The contact sites on gB and gH are consistent with the gH-binding site on HSV-1 gB suggested by blocking gH binding to gB with SS55 and SS56 antibodies ( epitopes mapped to residues 153–363 of gB ) ( Fig 5H ) [43] and the gB-binding sites on gH/gL suggested by anti-gH/gL antibody LP11 for HSV [13] , respectively . Mutagenesis of gH cytotail has led to its proposed role of acting as a “wedge” to split the gB endodomain “clamp” to trigger gB ectodomain refolding [44] . Though the details of their interactions in the endodomain are yet to be resolved , this first observation of gH/gL complex making contact with prefusion gB in situ ( Fig 6 ) supports the notion that receptor binding to gH/gL triggers transformation of gB from prefusion to postfusion conformation .
Since the postfusion conformation of gB is energetically favorable and structurally more stable , it is not surprising that purified recombinant gB so far have all adopted the “postfusion” conformation [12 , 16 , 45] . Therefore , imaging gB in its native , virion environment by cryoET seems to be the necessary approach to obtain the in situ structure in its metastable , prefusion conformation . However , a major challenge in interpreting in situ cryoET structures is the intrinsic poor contrast of tomographic reconstructions due to the use of low electron dose in order to avoid radiation damage to specimen . Poor contrast makes it difficult to identify different molecules or structures for subtomographic averaging . Normally for cellular tomography without phase plate , one could image with a large defocus value to achieve better contrast , aiding in distinguishing densities with different characteristics for subtomographic averaging . However , such approach only offers limited improvements in contrast ( S1 Fig ) , and difficulties still exist in identifying the slender gB in postfusion conformation in our tomograms . This experience is consistent with two previous cryoET studies on HSV-1 gB structures , in which large defocus values were used to increase contrast to facilitate subsequent subtomographic averaging , yet the resulting structure either is at much lower resolution [26] than reported here or has led to controversial interpretations [27] . The greatly improved contrast afforded by VPP technology allowed the differentiation of various glycoprotein structures based on their characteristic appearances on the virion membrane ( Fig 7A and 7B; S1 Fig ) . Therefore , cryoET with VPP offers a clear advantage in resolving structures of proteins in the native environments , enabling their identifications and subtomographic averaging to obtain structures of multi-functional states , as also demonstrated by the existence of two states of 26S proteasome inside neurons [31] . A vital step of herpesvirus infections is the fusion of viral and cell membranes , a complicated process involving at least three conserved proteins—gB , gH and gL . The in situ structures of gB at both prefusion and “postfusion” conformations reported here can shed lights on conformational changes of gB during membrane fusion and inform how herpesvirus entry into cell ( Fig 7 ) . Prior to fusion , gB needs to be maintained at its inactive , metastable prefusion conformation ( Fig 7A ) . The maintenance of this metastable conformation possibly involves a properly-folded endodomain of gB , since removal of the endodomain caused gB ectodomain to adopt the postfusion conformation [46] . In addition , the direct observation in our cryoET structure of gB-gH/gL complex ( Fig 6 ) and its isolation by co-immunoprecipitation [42] both suggest that the metastable ectodomain of gB might also be stabilized through the interaction with the ectodomain of gH subunit ( Fig 7C ) . Host receptor-binding to gH/gL complex would trigger a conformational change in gH/gL cytotail and its dissociation from , and the destabilization of , the endodomain of gB , which in turn triggers the massive conformational changes of gB ectodomain to expose its fusion loops ( step 1 ) . Subsequently , DIII central helix extends , unfurling other domains and swinging the fusion loops to engage with the host membrane ( step 2 ) . Facilitated by the intrinsic fluidity in the plane of the membrane , the refolding of gB domains to the lower-energy , postfusion conformation , in which its ectodomain C-terminal end and the fusion loops must come together , leads to fusion of the two membranes and the release of viral DNA-containing capsid into cytoplasm ( step 3 ) . In the absence of receptor binding as in the situation of this study , mechanical stress to the membrane caused by such means as high-speed centrifugation could also destabilize the membrane-associated endodomain , triggering metastable prefusion gB to undergo the cascade of transformation events , possibly accompanied by the exposure of the fusion loops ( step 1 ) . Lacking host cell membrane , these events , with exposed fusion loops eventually encountering and inserting its hydrophobic moieties into the viral membrane , will be followed by refolding of other domains into the stable , “postfusion” conformation ( step 3 ) . Notably , the topology of the conformational change during step 2 to step 3 would preclude transiting from prefusion to postfusion conformation without breaking the three-fold symmetry . Indeed , monomeric intermediates of VSV G have been observed both in solution and on the surface of virions at intermediate pH conditions [22 , 23] . In our model , the fusion loops of prefusion gB point to and are in close proximity to the viral membrane , possibly buried within a hydrophobic “mask” of MPR , which is attached to the C-terminal end of the gB ectodomain crystal structure . This membrane-proximal location of the gB fusion loops is the same as that based on the cryoET structure of the HSV-1 gB/anti-fusion loop 2-antibody at 5nm resolution [26] and is consistent with the fusion loop locations in all known atomic structures of classes I and III viral fusion proteins , including influenza HA [47] , HIV env trimer [48] , VSV G [21] and others [6] . Notably , our model is in stark contrast to the exposed fusion loops assigned to the membrane-distal tips of the “short-form” HSV-1 gB structures [27] , which were obtained by cryoET of purified gB-containing vesicles . The ectodomain of the “short-form” vesicular HSV-1 gB structure is 15% shorter in height and 23% wider in diameter than that of our in situ HCMV gB structure , despite both sharing the Christmas tree shape ( S5 Fig ) . Superposition of the domain assignment obtained by the hierarchical fitting approach [27] into the “short-form” HSV-1 gB structure shows that the densities projecting from the lower whorl of the Christmas tree-shaped trimer were unaccounted for ( S5B Fig ) . Moreover , placing the same domain assignment into our in situ HCMV gB prefusion structure reveals that the fusion loops in this assignment are projecting out of the cryoET map , yet the leader density of the map is not accounted for ( S5C Fig ) . When filtering the pseudoatomic model to 25Å , the cross-correlation coefficient is 0 . 74 , as compared to 0 . 93 of our prefusion structure . We believe that an exposed fusion loop orientation of prefusion gB is unlikely for both chemical and biological reasons—exposed hydrophobic moieties are chemically unfavorable in solution and can lead to unproductive membrane insertion during infection . Indeed , the “short-form” HSV-1 gB structure was cautiously interpreted as an ambiguous “prefusion and/or intermediate” conformation [27] , probably to reconcile these contradictory considerations . Secondary structure prediction indicates that the endodomain is helix-rich ( ~50% ) ( Fig 4A ) . Our results suggest that gB endodomain undergoes significant conformational changes , from prominently visible/stable in the prefusion structure ( Fig 3I and 3K ) , to invisible/flexible in the “postfusion” structure ( Fig 3C and 3E ) . Proteolysis and circular dichroism analyses of the endodomain of the highly homologous HSV-1 gB posit that gB endodomain clamps the viral membrane and stabilizes gB in its prefusion conformation [44 , 49] . This proposed model is supported by studies on truncation and substitution mutations in endodomain [44 , 46] . The structured endodomain resolved in the recent crystal structure of full-length gB was thought to be similar to that in prefusion gB [24] . Detergent solubilization of the membrane may be responsible for the postfusion conformation of its ectodomain . Our observation of the endodomain structure of HCMV gB changing from a stable , prefusion conformation ( Fig 3I ) to a flexible , postfusion conformation ( Fig 3E ) is consistent with its proposed role in stabilization of gB prefusion conformation on native viral membrane [24] .
Human fibroblast MRC-5 cells ( ATCC ) were cultured in Eagle's Minimum Essential Medium ( EMEM , ATCC ) with 10% fetal bovine serum ( FBS , Omega scientific: FB-11 ) . Cells were grown in T-175 cm2 flasks to 90% confluence and infected with HCMV strain AD169 ( ATCC , Rockville , MD ) at a multiplicity of infection ( MOI ) of 0 . 1–0 . 5 , and incubated for about 7 days . Once the cells showed 100% cytopathic effect , the media were collected and centrifuged at 10 , 000 g for 15 min to remove cells and large cell debris . The clarified supernatant was collected and centrifuged at 60 , 000 g for 1 hour to pellet HCMV virions . Pellets were resuspended in 20mM phosphate buffered saline ( PBS , pH 7 . 4 ) , loaded on a 15%–50% ( w/w ) sucrose density gradient , and centrifuged at 60 , 000 g for 1 hr . After the density gradient centrifugation , three light-scattering bands were observed in the density gradient: top , middle and bottom . The middle band contained both HCMV virions and NIEPs ( particles with intact viral envelopes as judged by negative-staining EM ) and was collected , diluted in PBS and then centrifuged at 60 , 000 g for 1 hour . The final pellet was resuspended in PBS for further cryoET sample preparation . VSV virion ( Indiana serotype , San Juan strain ) samples were produced as previously described [50] . Particularly , the inoculum was passaged multiple times in Hela cells with a very low multiplicity of infection ( MOI ) , 0 . 001 , to suppress the truncated defective-interference particles . The full VSV particles were isolated in a sucrose gradient and the final inoculum was also plaque-purified in Hela cells . We then pelleted the VSV virions at 30 , 000g for 2 hours and resuspended them in PBS . The stock was subjected to another low speed centrifugation at 12 , 000g for 5min in a desktop centrifuge to remove large aggregates . After resuspension , the pellets were banded on a 10ml density gradient containing 0–50% potassium tartrate and 30–0% glycerol . The virions-containing band was collected , diluted in PBS , pelleted at 30 , 000g for 2 hours , resuspended in PBS and kept in 4°C refrigerator for further cryoET sample preparation . An aliquot of 2 . 5 μl of the sample mixed with 5-nm diameter gold beads were applied onto freshly glow-discharged Quantifoil Holey Carbon Grids . Grids were blotted and plunge-frozen in liquid ethane cooled by liquid nitrogen using an FEI Mark IV Vitrobot cryo-sample plunger and were stored in liquid nitrogen before subsequent usage . CryoEM imaging and cryoET tilt series acquisition were performed with SerialEM [51] on an FEI Titan Krios 300kV transmission electron microscope equipped with a Gatan imaging filter ( GIF ) , a Gatan K2 Summit direct electron detector , and with or without a Volta phase plate ( VPP ) . Tilt series were recorded by tilting the specimen covering the angular range of -66° to +60° ( starting tilt from -48° to +60° , then from -50° to -66° ) with 2° or 3° interval , with a nominal magnification of x53 , 000 ( corresponding to a calibrated pixel size of 2 . 6 Å ) and a cumulative electron dose of 100~110 e-/Å2 . Exposure time was multiplied by a factor of the square root of 1/cosα ( in which α = tilt angle ) , and the exposure time at 0° was set at 1 . 2s for the tilt step-size of 2° or 1 . 6s for the tilt step-size of 3° . Movies were recorded with the frame rate of 0 . 2 frame/s on a Gatan K2 Summit direct electron detector operated in counting mode with the dose rate of 8–10 e-/pixel/s . An energy filter slit of 20 eV was chosen for the GIF . For imaging with VPP , defocus value was targeted at -0 . 6μm . Note , one of the benefits of using a phase plate is that the CTF is insensitive to the sign of the defocus value being negative ( underfocus ) or positive ( overfocus ) [52] . VPP was advanced to a new position every tilt series , followed by a 2 min waiting for stabilization , and pre-conditioned by electron illumination with a total dose of 12 nC for 60s to achieve a phase shift of ~54° as previously described [53] . For tilt series obtained without VPP , the defocus value was maintained at around -4μm while other imaging parameters were kept the same as those for the tilt series with VPP . Frames in each movie of the raw tilt series were aligned , drift-corrected and averaged with Motioncorr [54] to produce a single image for each tilt angle . Both sets of tilt series , collected with and without VPP , were reconstructed with IMOD 4 . 8 software package [55] in the following six steps . All images in a tilt series were coarsely aligned by cross-correlation ( step 1 ) and then finely aligned by tracking selected gold fiducial beads ( step 2 ) . The positions of each bead in all images of the tilt series were fitted into a specimen-movements mathematical model , resulting in a series of predicted positions . The mean residual error ( mean distance between the actual and predicted positions ) was recorded to facilitate bead tracking and poorly-modeled-bead fixing ( step 3 ) . With the boundary box reset and the tilt axis readjusted ( step 4 ) , images were realigned ( step 5 ) . Finally , two tomograms were generated by weighted back projection and simultaneous iterative reconstruction technique ( SIRT ) method , respectively ( step 6 ) . For data collected without VPP , contrast transfer function ( CTF ) was corrected with the ctfphaseflip program [56] of IMOD in step5 . The defocus value for each image in one tilt series was determined by CTFTILT [57] , and the estimated defocus value of each image was used as input for ctfphaseflip . Subtomographic averaging was performed using PEET 1 . 11 [58 , 59] . High contrast SIRT tomograms were 4× binned by the binvol program of IMOD to facilitate particle picking . Particles were picked manually in IMOD as follows . For distinct conformations of VSV G and HCMV gB on viral envelope , two points ( head and tail ) in one contour were used to define one particle ( glycoprotein ) —head is the membrane-proximal end of the protrusion density while tail is the membrane-distal end . An initial motive list file , a RotAxes file and three model files containing the coordinates of head , centroid and tail for each particle were generated by stalkInit in PEET . In total , we manually picked 337 long-form particles from 5 VPP tomograms of VSV , and 350 columnar tree-shaped particles and 886 Christmas tree-shaped particles from 11 VPP tomograms of HCMV . Besides , 637 Christmas tree-shaped particles were picked from 28 non-VPP tomograms , averaged either alone or together with those from the VPP tomograms for prefusion gB . For the reconstruction of the long-form VSV G , subtomographic averaging was performed first with 4× binned SIRT tomograms using the sum of all particles as the initial reference . Through stalkInit , each particle’s tilt orientation ( i . e . , the axes normal to the membrane ) was already coarsely aligned to Y axis , but its twist orientation ( i . e . , the angle around the axis ) was randomized . Therefore , in the first refinement cycle , we set the angular search range 180° max ( -180° to 180° ) with 9° step in Phi ( Y axis ) , and 5° ( -5° to 5° ) max with 1° step in both Theta ( Z axis ) and Psi ( X axis ) , and search distance 3 pixels along all three axes . Due to the known symmetry of postfusion VSV G , the resulting averaged structure was then trimerized and used as the reference of the next refinement cycle . The trimerized structure was the sum of each refined particle and its two symmetrical copies—the two symmetrical copies have the same position and tilt orientation as the refined particle , but twist orientation differed by either 120° or 240° . For subsequent refinement cycles , the newly trimerized structure from the last refinement cycle was used as reference , with both angular and distance search ranges narrowing down gradually . After four refinement cycles , the averaged structure converged based on no further improvement in resolution . The following refinement cycles were performed with 2× binned tomograms reconstructed by weighted back projection , after up-sampling ( generations of 2× binned model files and updates of corresponding motive list files from the latest refinement cycle ) , with small search distance range ( 4 pixels ) and narrow angular search range ( -20° to 20° ) . The reference was updated from the averaged structure of the last refinement cycle ( trimerized ) . For particles with distance of <1 pixel and twist angle difference of <1° , the one representative with lower cross-correlation coefficient was treated as duplicate particle and removed during the refinement . The averaged structure , contributed by 330 particles , converged after eight refinement cycles and was filtered to the final resolution , calculated by calcFSC in PEET based on the 0 . 143 FSC criterion . Reconstructions of columnar tree-shaped and Christmas tree-shaped particles on HCMV envelope followed the same refinement procedure as the reconstruction of long-form VSV G , except that trimerization was only applied after three-fold symmetry became apparent in the averaged structures . With the removal of duplicate particles , the final averaged structures of the postfusion ( columnar tree-shaped ) and prefusion ( Christmas tree-shaped ) conformations were obtained from 350 particles and 1509 particles , respectively . Furthermore , gold-standard FSC calculations for the structures were performed afterwards by splitting the original dataset of each conformation into two independent groups . The same refinement procedure used above was applied to the two newly-generated groups independently . Upon the convergence of the averaged structures , FSC were calculated by calcUnbiasedFSC in PEET ( S3A Fig . ) . For the reconstruction of the short-form VSV G , 65 particles were manually picked from five tomograms with single point to define the centroid position . Each particle was manually rotated around X , Y , Z axes to a similar orientation ( both the tilt orientation and twist angle ) in IMOD slicer window . By slicer2MOTL in PEET , the initial motive list files for subtomographic averaging were generated from the corresponding X , Y , Z rotation degrees . For the Angular Search Range , small search range was set during all seven refinement cycles . The final subtomographic average was Gaussian filtered with width 7 using the “volume filter” tool in UCSF Chimera [60] . Due to the limited number of particles ( 49 particles ) , HCMV gB-gH/gL complex was reconstructed with the same strategy above . We used IMOD [61] to visualize reconstructed tomograms and UCSF Chimera to visualize the subtomographic averages in three dimensions . The crystal structures of prefusion VSV G ( PDB: 5I2S ) [21] , postfusion VSV G ( PDB: 5I2M ) [20] , HCMV postfusion gB ( PDB: 5CXF ) [16] and gH/gL part from HCMV pentamer ( PDB: 5VOB ) [17] were fitted into subtomographic averages of prefusion G , postfusion G , postfusion gB and gB-gH/gL complex , respectively , with the tool fit in map in Chimera . Segmentation and surface rendering for the membrane and tegument proteins were done by the tools volume tracer and color zone in Chimera . All membrane glycoproteins were placed back on the viral membrane according to their locations in the original tomogram . A published structure of HCMV capsid with inner tegument protein [11] was filtered to 10 Å and placed back at the same position of the capsid in tomogram . As outlined below , we employed a combination of initial manual fitting of known domain structures , followed by simulation with MDFF program [38] to generate a gB prefusion model based on our cryoET prefusion gB trimer density map and the existing gB ectodomain postfusion crystal structure ( PDB: 5CXF ) [16] . First , the ectodomain in the subtomographic averaged density map of prefusion gB trimer was segmented out and its symmetric axis obtained with Chimera’s “volume eraser” tool and “measure symmetry” command , respectively . Second , Chimera’s “fitmap” command with “global search” and 15Å-resolution options was used to refine 1000 initial random DI placements , resulting in 28 refined fitted positions , each with a correlation coefficient ( between the fitted model and the density map ) and a “clash volume fraction” value ( between symmetry-related copies ) . We chose the fitted position with the largest fitting score , defined as the correlation coefficient subtracted by the “clash volume fraction” penalty value ( Fig 5A ) . Third , we obtained our initial DIII by computationally mutating the DIII model from the existing hypothetic model of EBV prefusion gB [14] , as it is known to differ substantially from its postfusion conformation for both herpesvirus gB [14 , 62] and homologous VSV G [21] . Compared to that in the postfusion gB , the central helix α4 in DIII in the prefusion gB is bent in order to fit into the top of the Christmas tree-shaped density . This bent varies from only ~30° in our proposed HCMV gB prefusion structure to ~90° in VSV G ( Fig 5D ) [21] and ~180° in influenza HA [47] and HIV env [48] . This DIII model , and the models of DII and DIV from the gB postfusion crystal structure were manually fitted as rigid bodies into our prefusion gB trimer cryoET density to produce a composite model with the above obtained DI trimer model by referencing the prefusion VSV G crystal structure . Connecting loops were then added to this composite model through the Modloop server [63] . Fourth , the resulting trimer model was used as the initial model for MDFF simulations [38] with grid force scale of 0 . 3 . Secondary structure , cis peptide and chirality restraints were imposed during MDFF simulations . Simulations were performed with NAMD 2 . 12 [64] , using the CHARMM36 force field with CMAP corrections [65] . Secondary structures for residues 707–906 of gB were predicted with Phyre2 [66] . | Infection by herpesviruses leads to many human diseases , ranging from mild cold sores to devastating cancers . Human cytomegalovirus ( HCMV ) is among the most medically significant herpesviruses and causes birth defects and life-threatening complications in immuno-suppressed individuals . HCMV infection begins with cellular membrane fusion , a dynamic process involving receptor-binding to gH/gL complexes and drastic transformation of fusion protein gB trimer from the metastable prefusion conformation to the stable postfusion conformation . We have used cryo electron tomography incorporating cutting-edge technologies to observe the three-dimensional structures of gB and gH/gL in their native environments of HCMV particles . Visualizations of gB in both prefusion and postfusion conformations , together with structures of other class III viral fusion proteins and molecular dynamics flexible fitting ( MDFF ) , facilitate a prediction for the structure of HCMV gB in its prefusion conformation . With the further observation of the contact of prefusion gB with gH/gL complex , the conformational changes of gB from pre- to postfusion state lead to a better understanding of herpesvirus fusion mechanism . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"crystal",
"structure",
"viral",
"transmission",
"and",
"infection",
"condensed",
"matter",
"physics",
"microbiology",
"viral",
"structure",
"membrane",
"fusion",
"protein",
"structure",
"crystallography",
"cellular",
"structures",
"and",
"organelles",
"viral",
"entry",
... | 2018 | Different functional states of fusion protein gB revealed on human cytomegalovirus by cryo electron tomography with Volta phase plate |
Ubiquitination of the replication clamp proliferating cell nuclear antigen ( PCNA ) at the conserved residue lysine ( K ) 164 triggers postreplicative repair ( PRR ) to fill single-stranded gaps that result from stalled DNA polymerases . However , it has remained elusive as to whether cells engage PRR in response to replication defects that do not directly impair DNA synthesis . To experimentally address this question , we performed synthetic genetic array ( SGA ) analysis with a ubiquitination-deficient K164 to arginine ( K164R ) mutant of PCNA against a library of S . cerevisiae temperature-sensitive alleles . The SGA signature of the K164R allele showed a striking correlation with profiles of mutants deficient in various aspects of lagging strand replication , including rad27Δ and elg1Δ . Rad27 is the primary flap endonuclease that processes 5’ flaps generated during lagging strand replication , whereas Elg1 has been implicated in unloading PCNA from chromatin . We observed chronic ubiquitination of PCNA at K164 in both rad27Δ and elg1Δ mutants . Notably , only rad27Δ cells exhibited a decline in cell viability upon elimination of PRR pathways , whereas elg1Δ mutants were not affected . We further provide evidence that K164 ubiquitination suppresses replication stress resulting from defective flap processing during Okazaki fragment maturation . Accordingly , ablation of PCNA ubiquitination increased S phase checkpoint activation , indicated by hyperphosphorylation of the Rad53 kinase . Furthermore , we demonstrate that alternative flap processing by overexpression of catalytically active exonuclease 1 eliminates PCNA ubiquitination . This suggests a model in which unprocessed flaps may directly participate in PRR signaling . Our findings demonstrate that PCNA ubiquitination at K164 in response to replication stress is not limited to DNA synthesis defects but extends to DNA processing during lagging strand replication .
The accurate copying of a cellular genome and subsequent transmission of genetic material to two daughter cells occurs on a microscopic scale , but is nonetheless a prodigious task . Considering the difficulty of accomplishing this fundamental process for living cells , it is hardly surprising that evolution has selected for a complex and multi-layered system of checkpoints and redundancies that promote its completion under sub-optimal conditions [1 , 2] . Many of these processes are regulated by post-translational modification of proteins that act as molecular switches to regulate downstream responses . The replication clamp , proliferating cell nuclear antigen ( PCNA ) , is one such target for a variety of post-translational modifications that trigger an array of downstream effects . Known modifications include the covalent attachment of ubiquitin and the small ubiquitin-like modifier ( SUMO ) to specific lysine ( K ) residues [3] . SUMO modification of chromatin-bound PCNA , or sumoylation , occurs during an unperturbed S phase at K164 and–to a lesser extent–at K127 [3] . Sumoylation acts primarily to recruit the helicase/anti-recombinase suppressor of rad six 2 ( Srs2 ) , which prevents illegitimate recombination at replication forks [4–6] . Ubiquitination of PCNA occurs predominately at K164 , however , alternative attachment sites have been mapped in yeast and human cells [7–11] . In contrast to sumoylation , ubiquitination is induced by replication stress [3] . PCNA-K164 ubiquitination was initially identified as a response to template strand lesions , which stall the highly selective processive polymerases ( Pol- ) δ and Pol-ε [3] . Polymerase stalling leads to the accumulation of single-stranded ( ss ) DNA , which quickly becomes coated with replication protein A ( RPA ) [12] . This allows for the recruitment of the E2-E3 ubiquitin ligase complex radiation sensitive-6 and -18 ( Rad6-Rad18 ) to mono-ubiquitinate PCNA-K164 [13] . Mono-ubiquitin can be subsequently extended to K63-linked poly-ubiquitin chains by methyl methanesulfonate sensitive 2-ubiquitin conjugating 13-radiation sensitive 5 ( Mms2-Ubc13-Rad5 ) [3] . The length of the ubiquitin chain plays a crucial role in determining which of two PRR pathways is activated . Mono-ubiquitin facilitates an error-prone pathway for lesion bypass dependent on translesion polymerase activity [3 , 14–16] . Recent data indicates that replication past lesions by the mutagenic translesion polymerase Pol-ζ may continue for up to 1 kilobase beyond the lesion [17] . Alternatively , poly-ubiquitin chains enable a template switching pathway in which the nascent DNA of the sister chromatid acts as a template , allowing for lesion bypass and filling of ssDNA gaps [18] . This process is considered to be “error-free” , because it does not rely on the intrinsically mutagenic translesion polymerases of the error-prone pathway [19] . The precise mechanism of this branch is not yet well understood . In addition to the originally described function of PRR in DNA damage tolerance and lesion bypass , recent work has suggested that mutants with impaired replisome function also activate these pathways for replication of undamaged template strands [20–23] . We have previously demonstrated that PRR promotes the viability of mcm10 mutants in the absence of DNA damage [23] . To systematically explore the role of PCNA-K164 in response to intrinsic cellular dysfunction , we performed SGA analysis of a PCNA-K164 to arginine ( PCNA-K164R ) mutant . Interestingly , we found that the genetic interaction profile of the PCNA-K164R mutant closely resembled that of many alleles of lagging strand replication factors , including those involved in Okazaki fragment processing . This observation was particularly intriguing , as PRR has not been implicated in tolerating Okazaki fragment processing defects . As a result , we further investigated the activity of PCNA-K164-dependent pathways in mutants disrupting normal lagging strand replication . Specifically , we focused on the role of PCNA-K164 in cells deficient for the flap endonuclease radiation sensitive 27 ( Rad27 ) or enhanced level of genomic instability 1 ( Elg1 ) , a homolog of replication factor C ( RFC ) subunit Rfc1 [24] . Rad27 processes 5’ flaps generated during lagging strand replication when DNA synthesis by Pol-δ collides with the 5’ end of the RNA-DNA primer of the previous Okazaki fragment , displacing it into a small <10-nucleotide ( nt ) flap [25 , 26] . If the 5’ flap escapes processing by Rad27 and grows long enough to bind replication protein A ( RPA ) , it is then cleaved by Dna2 [26–28] . RPA binding of the flap serves both to inhibit Rad27 , and to recruit Dna2 [27] . Dna2 cleaves the long ~30-nt flap to a short flap ( 5–10 nt ) , which can then be further processed by Dna2 or Rad27 into a ligatable nick [25 , 27–29] . Although processing of long flaps must be relatively efficient under normal conditions , rad27Δ mutants exhibit a temperature dependent slow-growth phenotype [30] . This is best explained by an increased rate in DNA replication and concomitant increase in the formation of long flaps [30 , 31] . At the restrictive temperature of 37°C , rad27Δ mutants are unable to meet flap processing demands , resulting in lethality , whereas at the semi-permissive temperature of 35°C growth is merely impaired [30 , 32] . In the absence of complete flap removal–even at lower temperatures–Pol δ-exonuclease ( exo ) activity can resect the nascent 3’ end allowing the small 5’ flap to re-anneal and form a ligatable nick [26 , 33] . After ligation of the nick by DNA ligase I ( Cdc9 ) , PCNA is unloaded from chromatin by the Elg1:Rfc2-5 complex [34–36] . In the present study , we report that PCNA is ubiquitinated in rad27Δ and elg1Δ mutants . Whereas ablation of PRR is inconsequential in the elg1Δ strain , both translesion synthesis ( TLS ) and template switching promote rad27Δ viability , possibly by enabling alternative flap processing . Furthermore , the long RPA-coated flaps generated in the absence of Rad27 play an active role in promoting the ubiquitination of PCNA at K164 and initiating PRR .
To examine the global role of PCNA-K164 in the absence of exogenous DNA replication stressors , we performed SGA analysis of two independently isolated PCNA-K164R mutant clones ( identified as PCNA-K164R clone 1 and PCNA-K164R clone 2 , respectively ) against a library of temperature-sensitive ( TS ) alleles and a full genome ( FG ) array as previously described ( S1–S4 Tables ) [37 , 38] . Parallel analyses were performed against the TS array using a decreased abundance by mRNA perturbation ( DAmP ) allele of PCNA or a wild type ( WT ) allele as the query strain [39 , 40] . Since ubiquitination or sumoylation of K164 facilitates only a subset of PCNA functions , we anticipated that interactions identified in the PCNA-K164R SGA screens should represent a small part of those identified in the PCNA-DAmP analysis . Indeed the vast majority of hits identified in the K164R mutant screen with the TS array were also identified with the PCNA-DAmP allele ( for PCNA-K164R clone 1: 18/26 hits overlapped with PCNA-DAmP with p-value < 10−11 , and for PCNA-K164R clone 2: 11/14 hits overlapped with p-value < 10−8 as determined by Fisher’s test ) ( S1 and S2 Tables and Fig 1A ) . Furthermore , the negative genetic interactions were largely consistent with previous reports , including a requirement for PRR when thiol-specific antioxidant 1 ( TSA1 ) is deficient [41] . Mutants of TSA1 have reduced ability to neutralize reactive oxygen species , leading to increased DNA damage and synthetic sickness with PRR mutation [41] . We also observed a modest requirement for homologous recombination ( HR ) components in K164R mutants ( Fig 1A ) [42] . This gave us confidence that genes identified in the PCNA-K164R screens represented bona fide genetic interactions . The most informative results were revealed when examining the similarity of the PCNA-K164R SGA profiles with the interaction signatures of other mutants . Strong Pearson correlation coefficient ( PCC ) values between PCNA-K164R clones and rad5Δ and rad6Δ mutants were consistent with the known functions of K164 in facilitating PRR ( Fig 1B and Table 1 ) . Strikingly , PCNA-K164R also exhibited strong correlation with many mutant alleles of genes involved in lagging strand replication , suggesting that those mutants have replication defects similar to those in the PCNA-K164R mutants ( Fig 1B and Table 1 ) . To validate this observation in an unbiased manner , we performed a Gene Ontology ( GO ) enrichment analysis ( http://www . ebi . ac . uk/QuickGO/ ) . Alleles with SGA profiles similar to that of PCNA-K164R against the FG array ( PCC > 0 . 15 ) were significantly associated with leading and lagging strand replication GO terms ( Fig 1C ) . Interestingly , the list of enriched GO terms included “Okazaki fragment processing” , which has not been associated with PCNA-K164-dependent pathways ( Fig 1C ) . GO enrichment of SGA profiles similar to PCNA-K164R against the TS array ( PCC > 0 . 2 ) also showed nearly 25-fold enrichment with the “Okazaki fragment processing” term ( S6 Table ) . Because this initial analysis relied on existing GO annotations , we manually compiled an informed list of genes associated with leading and lagging strand replication for further analysis ( S7 Table ) . We found that PCNA-DAmP and PCNA-K164R profiles against the TS array were highly similar ( PCC > 0 . 2 ) to profiles of genes implicated in leading and lagging strand replication ( Fig 2A–2D ) . We confirmed these results when we compared the profile of a second PCNA-K164R query strain ( Table 1 and S1 Fig ) . The PCNA-WT profile did not show any significant similarities ( Table 1 and S1 Fig ) . Altogether , these findings suggested that PCNA-K164 may have an active role in lagging strand replication , even in the absence of exogenous DNA damage . Strong correlations with PCNA-K164R included pol1 mutants , which we previously described to activate PRR pathways , and pol3 mutants ( deficient in Pol-δ ) which have also been described to elicit PCNA ubiquitination and TLS ( Table 1 ) [20 , 23] . Additional strong correlations were observed for rfc5 , pol31 , rad27 , and elg1 mutants ( Fig 1B and Table 1 ) . These genes have been implicated in different steps of Okazaki fragment synthesis and processing , suggesting that PCNA-K164 is required at multiple junctions when lagging strand replication is impaired . This was surprising , as K164 ubiquitination of PCNA is dependent on the formation of ssDNA and is typically associated with DNA synthesis defects only [13] . The source of ssDNA–particularly in Okazaki fragment processing mutants , such as rad27Δ –was thus not immediately obvious . Nonetheless , these results led us to hypothesize that PCNA-K164-dependent pathways may be required to tolerate defects in lagging strand maturation . Because the function of K164 in PRR is dependent on its modification by ubiquitin , we hypothesized that lagging strand defects would result in chronic PCNA ubiquitination and activation of PRR pathways . To experimentally address this question , we assayed PCNA ubiquitination in rad27Δ and elg1Δ mutants , both of which had interaction profiles that correlated strongly with the PCNA-K164R alleles ( Fig 1B and Table 1 ) . rad27Δ is a temperature-sensitive allele , and for all subsequent experiments we shifted these mutants to 37°C for 3 h prior to analysis . To determine whether PCNA is ubiquitinated at K164 in rad27Δ and elg1Δ mutants , we purified histidine tagged PCNA ( His6-PCNA ) on Ni-NTA agarose and analyzed the eluates with PCNA- , ubiquitin- and SUMO-specific antibodies by western blot ( Fig 3A and 3B ) . PCNA was indeed ubiquitinated in both mutants and ubiquitin attachment was completely ablated when PCNA carried a K164R substitution ( Fig 3A and 3B ) , indicating that alternative attachment sites were not targeted . PCNA-K164R mutants also exhibited loss of K164-dependent sumoylation , consistent with previous reports [3 , 44] . Interestingly , when we introduced the PCNA-K164R mutation in elg1Δ cells , we reproducibly observed a minor PCNA-SUMO species of a slightly lower molecular weight than K164-SUMO ( marked by an asterisk ) , consistent with an earlier study that revealed an alternative sumoylation site ( Fig 3B ) [44] . As shown previously , K127-SUMO migrated markedly slower than K164-SUMO . Moreover , levels of K127-SUMO were elevated in PCNA-K164R mutants [3] . Both rad27Δ and elg1Δ exhibited increased PCNA sumoylation at K127 and K164 compared to wild type ( Fig 3A and 3B ) . Next we asked whether PCNA-K164 modifications were important for viability of these two strains . Spotting analysis revealed that rad27Δ mutants had a significant reduction in growth at the semi-permissive temperature of 35°C when combined with the PCNA-K164R ( pol30-K164R ) mutation , whereas elg1Δ cells exhibited no such sensitivity at any temperature tested ( Fig 3C ) , nor when they were exposed to UV light ( Fig 3D ) . In contrast , rad27Δ pol30-K164R double mutants were acutely sensitive even to low doses of UV , showing a severe reduction in growth after exposure to 1J/m2 ( Fig 3D ) . Together , these results suggested that K164-dependent pathways are important for the growth of rad27Δ , but not elg1Δ cells . Because ubiquitination of PCNA at K164 is necessary for both TLS and template switching , we sought to determine which of these pathways are active in rad27Δ cells . Spotting analysis revealed that pol30-K164R and rad18Δ mutations each significantly reduced the viability of rad27Δ cells at 35°C ( Fig 4A ) . The rad27Δ rad18Δ double mutant reproducibly appeared to have a slightly more severe growth defect than the rad27Δ pol30-K164R strain ( Fig 4A ) . We attribute this finding to the known fact that PCNA-K164 sumoylation suppresses HR , and therefore substitution of K164 upregulates HR [4 , 5] . To address whether sumoylation of PCNA-K164 affected the viability of rad27Δ mutants , we deleted SIZ1 , which encodes the SUMO E3 ligase that catalyzes this reaction . Consistent with published reports , rad27Δ siz1Δ double mutants did not exhibit any increased temperature sensitivity [42 , 45] . These results strongly suggest that the PCNA-K164 dependent phenotype is solely due to the loss of ubiquitination . To estimate the relative contribution of TLS and template switching to rad27Δ viability , we generated rad27Δ strains with rev3Δ or rad5Δ mutations rendering them deficient in TLS and template switching , respectively . In addition , we analyzed a rad27Δ rev3Δ rad5Δ triple mutant , defective in both branches . We found that rad27Δ rev3Δ double mutants did not display any significant growth defects , whereas the rad27Δ rad5Δ cells exhibited a mild but noticeable growth delay , suggesting that the template switching pathway is the more prominent of the two ( Fig 4B ) . However , loss of both pathways in the rad27Δ rev3Δ rad5Δ triple mutant resulted in a synergistic effect , resembling that of the rad27Δ rad18Δ double mutant ( Fig 4A and 4B ) . These results argue that the two branches of PRR are both active in rad27Δ cells and have partially redundant roles in promoting viability . The finding that REV3 affected the survival of rad27Δ mutants in the absence of RAD5 encouraged us to further examine the activity of the TLS branch of PRR . To accomplish this , we took advantage of the fact that TLS employs intrinsically mutagenic polymerases , which have a higher rate of nucleotide misincorporation combined with a lack of proofreading activity [46] . We predicted that TLS induced mutations would be dependent on K164 . Mutation of K164 to arginine disables DNA synthesis by pol-ζ and its binding partner Rev1 , which are responsible for the vast majority of TLS induced mutations [47] . Consistent with previous reports , fluctuation analysis revealed that rad27Δ mutants have a drastically increased rate of mutation ( Fig 4C ) [31 , 48] . Addition of the pol30-K164R allele led to a significant decrease in the mutation rate that accounted for 20–25% of total alterations , confirming that TLS was active in these cells ( Fig 4C ) . Because the pol30-K164R mutation also removes the suppressive effect of PCNA-SUMO on HR , an increase in gross chromosomal rearrangements may mask the decrease in mutation rate due to the loss of TLS . Therefore , the K164-dependent reduction is likely an underestimation of the contribution by TLS [4–6] . In agreement with this notion , deletion of the pol-ζ catalytic subunit REV3 results in a more severe reduction in the mutation rate than the pol30-K164R mutation ( Fig 4C ) . Nonetheless , our observations are consistent with a recent report that had estimated point mutations in rad27Δ mutants to account for ~40% of all genomic aberrations [48] . Our results support the idea that the majority of these single nucleotide variations are a result of translesion polymerase activity . To further explore how PCNA-K164 aids in cell survival , we analyzed activation of the Rad53 kinase , a downstream target of the mitotic entry checkpoint kinase 1 ( Mec1 ) , the homolog of human ATR ( ataxia telangiectasia mutated- and Rad3-related ) [49] . rad27Δ pol30-K164R double mutants showed increased phosphorylation of Rad53 relative to rad27Δ cells after they were shifted to the restrictive temperature of 37°C for 3 and 4 h . This was indicative of enhanced replication stress ( Fig 5A ) . Consistently , the double mutants also displayed a robust late S/G2 phase arrest at those time points ( Fig 5B ) . Since rad27Δ cells passed more proficiently through mitosis ( indicated by the higher G1 peak marked with a red arrow at 3 and 4 h in Fig 5B ) , we concluded that PRR pathways likely facilitated the completion of S phase and ultimately allowed for entry into mitosis . Therefore , without PRR cells have a reduced ability to tolerate Rad27 deficiency . Altogether , our findings support a role for PRR pathways in suppressing replication defects when Rad27 is absent . In contrast , elg1Δ pol30-K164R double mutants did not display any Rad53 activation or observable alterations in cell cycle distribution relative to elg1Δ ( S2 Fig ) . Previous work demonstrated that ubiquitination of PCNA at K164 by Rad6-Rad18 requires persistent regions of RPA-coated ssDNA [13] . These normally accumulate if the replicative polymerases are impeded [12] . However , in the context of Rad27 deficiency , the source of ssDNA was not readily apparent . Earlier studies established that in the absence of Rad27 , Okazaki fragment flaps are processed through a “long flap” pathway by the combined activities of Dna2 and Pol3-exo [25 , 26 , 33 , 50 , 51] . In this pathway short flaps become longer through enhanced strand displacement until they are sufficiently large to be bound by RPA [52 , 53] . Binding by RPA serves to recruit Dna2 and stimulate its nuclease activity , reducing the flap to approximately 5 nt [27 , 28] . Although Dna2 has been shown to be competent to subsequently cleave the remaining short flap in vitro [28] , Pol3-exo activity is clearly essential in rad27Δ mutants at all temperatures [33] . Pol3-exo is thought to resect the 3’ end of the preceding Okazaki fragment , allowing the remaining 5’ flap to re-anneal and form a ligatable nick [26 , 33] . It is conceivable that both Dna2 and Pol3-exo contribute to the resolution of short flaps in rad27Δ cells . The binding of RPA to long flap intermediates prior to processing by Dna2 led us to consider that long ssDNA flaps themselves could provide the stimulus for PCNA ubiquitination . We inferred that the close proximity of these RPA-bound structures to PCNA should allow for recruitment of the Rad6-Rad18 complex and subsequent PCNA ubiquitination . To test this hypothesis , we sought to modulate flap processing by overexpression of DNA2 [29] . Because the current model for Dna2 processing of 5’ flaps proposes that RPA binding occurs prior to cleavage , we expected that recruitment of Rad6 and Rad18 may not be significantly reduced upon DNA2 overexpression ( Fig 6A ) [27 , 54] , unless it would directly compete with the E2-E3 complex . Notably , overexpression of DNA2 did not reduce PCNA ubiquitination in rad27Δ ( Fig 6B ) . We also considered the possibility that Pol3-exo activity during long flap processing could generate ssDNA regions sufficiently large to bind RPA ( S3A Fig ) However , overexpression of an exonuclease-dead allele of POL3 ( pol3-01 ) failed to reduce PCNA ubiquitination in rad27Δ ( S3B Fig ) . Consistent with previous reports , pol3-01 expression was lethal in combination with rad27Δ ( S3C Fig ) [55] . We next sought to modulate flap processing in a manner that reduced the formation of long RPA-bound flaps . Multiple studies have demonstrated that overexpression of exonuclease 1 ( EXO1 ) rescues the DNA damage sensitivity of rad27Δ mutants [56–59] . Exo1 and Rad27 are both Rad2 family nucleases and crystal structures of their human homologs , FEN1 and EXO1 , reveal highly conserved mechanisms of substrate binding and cleavage [60–62] . Thus , we hypothesized that Exo1 , like Rad27 , may cleave flaps before RPA can bind to them . If this were true , Exo1 overexpression should reduce PCNA ubiquitination in rad27Δ cells ( Fig 6C ) . Indeed , overexpression of EXO1 from a galactose inducible promoter eliminated PCNA ubiquitination in rad27Δ mutants ( Fig 6D ) . Furthermore , this phenotype was dependent on the catalytic activity of EXO1 , as overexpression of a nuclease-dead exo1-D173A allele had no impact on PCNA ubiquitination ( Fig 6D ) [59 , 63] . Furthermore , EXO1 overexpression did not rescue the temperature sensitivity of rad27Δ ( S4 Fig ) . In summary , our results suggest that the majority of PCNA ubiquitination in rad27Δ is dependent on RPA-coated ssDNA intermediates , which recruit the Rad6-Rad18 complex and are degraded by Exo1 . To examine whether the effect of EXO1 overexpression on PCNA ubiquitination in rad27Δ mutants could be due to indirect suppression of ssDNA gap formation , we exposed EXO1 overexpressing cells to 50 J/m2 of UV light , which has been proven to cause ssDNA gaps ( Fig 7A ) [12] . Overexpression of EXO1 had no impact on the level of PCNA ubiquitination under these conditions ( Fig 7B ) . Moreover , we did not observe any differences in ubiquitination when cells were treated with 100 J/m2 of UV light , arguing that Exo1 did not act directly or indirectly to eliminate ssDNA regions ( Fig 7B ) . In support of this conclusion , overexpression of EXO1 had no impact on PCNA ubiquitination in cells harboring the temperature sensitive pol1-1 allele ( Fig 7C and 7D ) . This allele is thought to generate ssDNA regions as a result of reduced efficiency in the priming of Okazaki fragments ( Fig 7C ) [64 , 65] , and causes ubiquitination of PCNA at K164 at the non-permissive temperature of 35°C [23] . Taken together , these findings indicate that Exo1 does not suppress the formation of ssDNA gaps .
The primary replication defect in rad27Δ cells is caused by impaired processing of 5’ flaps generated during Okazaki fragment processing [31] . At elevated temperatures , DNA replication proceeds more rapidly , likely leading to an increase in the formation of long flap intermediates , which must bind RPA before they can be efficiently processed [27] . We speculated that these long RPA-coated flaps may serve as a platform to promote PCNA ubiquitination . To examine this hypothesis , we modulated flap length by overexpression of EXO1 , a close relative of RAD27 with a highly conserved mechanism of substrate binding and cleavage [60 , 61] . A number of prior studies have demonstrated that overexpression of EXO1 suppresses the intrinsic mutagenicity of the rad27Δ allele [56 , 57] . In particular , EXO1 overexpression suppresses the duplication of short direct repeats that have been hypothesized to result from longer flap structures generated in rad27Δ , which is consistent with Exo1 activity preventing long RPA-coated flap formation [56 , 57] . Thus , our finding that catalytically active Exo1 counteracts PCNA ubiquitination in rad27Δ , but has no effect on ubiquitination in pol1-1 cells or after UV treatment , argues that the DNA structures mediating ubiquitination in flap endonuclease deficient cells are different from ssDNA gaps ( Figs 6 and 7 ) . It has been speculated that long unprocessed flaps could participate in the replication stress response [10 , 52 , 69] . Campbell and colleagues found that constitutive Mec1 activation is responsible for dna2Δ lethality [69] . They hypothesized that Mec1 activation originated from long RPA-coated flaps that accumulate in these mutants [69] . Interestingly , EXO1 overexpression partially rescues the temperature sensitivity of a viable dna2-1 mutant , consistent with the notion that Exo1 acts upstream of long flap formation [70] . Another well-documented hallmark of Okazaki fragment processing mutants is profound instability of trinucleotide repeat ( TNR ) regions [71 , 72] . This raises the question as to whether a requirement for PCNA-K164 in rad27Δ mutants is specific to the replication of TNR regions . A previous study had linked error-free PRR to the suppression of TNR expansion in rad27Δ cells [73] . However , error-prone TLS did not appear to regulate TNR expansion at all [73 , 74] . Our finding that both TLS and error-free PRR are active in rad27Δ cells therefore leads us to conclude that the role of PCNA-K164 in these mutants is not restricted to genomic regions that encompass TNRs ( Fig 4B ) . Historically , PCNA ubiquitination and PRR were considered rescue pathways for template strand lesions that impair polymerase progression and require TLS or template switching for bypass [3] . Later work from the Shcherbakova group demonstrated that intrinsic replisome deficiencies in hypomorphic alleles of the replicative polymerases POL2 and POL3 also lead to PCNA ubiquitination and activation of TLS on undamaged DNA templates [20] . This important finding described ubiquitination of PCNA and activation of PRR in the absence of replication stressors that damage DNA . Nevertheless , the essential effect of DNA damaging agents and hypomorphic polymerases on replication is a disruption of DNA synthesis . Both therefore intuitively lead to ssDNA gaps , triggering PCNA ubiquitination and subsequent gap-filling by PRR . In contrast , our study describes PCNA ubiquitination under conditions in which DNA synthesis is not impaired . Rad27 catalyzes Okazaki fragment flap cleavage , which does not occur until after Okazaki fragment synthesis , yet we see a requirement for PRR to support the viability of Rad27 deficient cells . This distinction suggests that PCNA-K164 is active in mediating DNA processing defects that are unlinked to problems in DNA synthesis . It is tempting to speculate that PRR pathways are promoting processing of Okazaki fragments by Dna2 or potentially an alternative mechanism . Error-free template switching could allow for synthesis past multiple Okazaki fragments using the sister chromatid as a template and reducing the overall requirement for flap endonuclease . A role for PRR in promoting flap processing and thereby reducing the half-life of long flaps is consistent with our observation of elevated Rad53 phosphorylation in rad27Δ pol30-K164R double mutants ( Fig 5A ) . The mechanism by which PRR promotes flap processing or bypass is currently unclear . Similar to RAD27 , ELG1 has been described as an important protector of genome stability [24 , 75–78] . Recent reports have identified Elg1 as a crucial component of an alternative RFC complex that unloads PCNA from double-stranded DNA [34 , 35] . Our SGA screen revealed that the genetic interaction profile of elg1Δ correlates strongly with the PCNA-K164R and PCNA-DAmP profiles , leading us to investigate ubiquitination of PCNA at K164 in this mutant ( Figs 1B and 3B and Table 1 ) . Unlike rad27Δ , elg1Δ cells did not exhibit synthetic sickness with the K164R mutation and displayed no acute requirement for PRR pathways to tolerate intrinsic replication stress or mild UV treatment ( Fig 3C and 3D ) . We speculate that replication defects in elg1Δ are mimicking those present in PCNA-DAmP mutants in that both are limiting the amount of free PCNA in the nucleus that is available to load onto chromatin and engage in replication . In the case of the PCNA-DAmP allele this is simply the result of lower steady-state levels of PCNA protein , whereas in elg1Δ , PCNA is likely sequestered on DNA due to diminished unloading [34 , 35 , 79] . In summary , our results suggest that during the processing of Okazaki fragments via the long flap pathway , the flap itself is likely not an inert DNA processing intermediate , but may play an active role in signaling replication stress through PCNA . It is conceivable that under normal conditions a division of labor between long and short flap processing is required for efficient Okazaki fragment maturation [27 , 80–82] . In rad27Δ cells , the balance is pushed severely to the long flap pathway , leading to the accumulation of RPA bound ssDNA structures that can be eliminated by Exo1 . This becomes particularly problematic at elevated temperatures when the kinetics of DNA replication are increased and flaps are produced at a higher rate . The mechanism by which PRR is suppressing replication stress under these conditions is not clear at this time , but we speculate that it is helping to circumvent flap processing . These findings are salient in light of the relationship between the regulation of flap processing and cancer . Homozygous deletion of the RAD27 homolog FEN1 in mice is lethal , but heterozygous deletion in combination with mutation of the adenomatous polyposis coli ( Apc ) gene results in increased numbers of adenocarcinomas , enhanced tumor progression and decreased survival [83] . Mutations which reduce FEN1 activity have also been demonstrated to vastly increase cancer incidence in mouse models [84] . This study provides molecular evidence for the pathways contributing to mutagenesis when flap endonuclease function is compromised and gives insight into how cells sustain viability under these conditions .
All yeast strains used in this study are isogenic derivatives of wild type E133 cells , which were derived from CG379 [85] , with the exception of pol1-1 strains which are derived from SSL204 [23] . Strains with gene deletions were generated by PCR mediated gene disruption [86] . All clones were verified by PCR and sequencing of the modified locus . Strains carrying pol30-K164R mutations were generated by PCR mediated gene disruption as follows: continuous PCR fragments consisting of PCNA , its endogenous promoter and a LEU2 marker were amplified from pCH1654 ( a gift from L Prakash , UTMB ) and integrated at the endogenous PCNA locus . Integration and clonal homogeneity were verified by PCR and sequencing . All strains generated in this study are listed in S8 Table . His6-tagged PCNA strains were constructed using Yip128-P30-POL30wt ( gift from HD Ulrich , IMB Mainz ) . Plasmid variants with lysine mutations were constructed using the QuikChange Lightning ( Agilent Technologies ) site-directed mutagenesis kit . Briefly , the plasmid was linearized at an AflII restriction site in the LEU2 coding sequence and transformed . Clones were screened by PCNA western blot to ensure that His6-PCNA expression was equivalent to endogenous ( untagged ) expression levels . The endogenous copy of PCNA was then knocked out via PCR mediated gene disruption . In experiments using galactose inducible gene expression , liquid cultures were grown to OD600 = 0 . 600 at 25°C in raffinose containing medium . Galactose was then added to a final concentration of 2% and the cultures were shifted to 37°C for 3 h before collecting . Overexpression of POL3 and pol3-01 was induced by adding galactose to cells carrying the plasmids pBL336 and pBL336-01 , respectively ( gifts from D . Gordenin , NIEHS , originally constructed in P . M . J . Burgers laboratory , Washington University in St . Louis ) [26] . Expression of DNA2 was induced with galactose in cells carrying pgal-DNA2 ( a gift from R . Wellinger , Université de Sherbrooke , Québec ) [87] . EXO1 overexpression was induced by adding galactose to cells carrying pcDNA50 . 1 , a derivative of pRS316 that was referred to as gal-EXO1 ( a gift from K . Lewis , Texas State University ) [58] . The exo1-D173A variant was generated using the QuikChange Lightning ( Agilent Technologies ) site-directed mutagenesis kit . UV treatment ( 254nm ) of liquid cultures was applied using a UV crosslinker ( CL-1000 , UVP ) . Cultures were transferred to a sterile tray and treated in the crosslinker before being returned to flasks and cultured an additional 40 min before harvesting . A genome-wide screen for genetic interactions with four POL30 alleles as query strains ( PCNA-DAmP , PCNA-WT , PCNA-K164R Cl . 1 , and PCNA-K164R Cl . 2 ) was conducted at 30°C as previously described [38] . Because the screens are performed at 30°C they did uncover a synthetic interaction between PCNA-K164R and rad27Δ . Briefly , the query strains , marked with a nourseothricin ( NatMX4 ) resistance cassette and harboring the SGA haploid specific markers and reporter [38] , were mated to an array of 786 temperature-sensitive and 175 viable deletion mutants ( TS array: manuscript in preparation ) ( S1 and S2 Tables ) . Additionally , PCNA-K164R Cl . 1 and Cl . 2 were mated to an array of 3827 viable deletion mutants ( FG array ) . Nourseothricin- and geneticin-resistant heterozygous diploid mutants were selected and sporulated with MATa pol30 double mutants as described [38] . Results of this screen are also included in the supplementary material ( S3 and S4 Tables ) . Different PCC cutoffs were applied to the FG and TS array data ( 0 . 15 and 0 . 2 , respectively ) in order to enrich for the top 2% of all profile correlations . Cultures were grown to OD600 = 0 . 600 at 25°C and then shifted to 37°C for 3 h before harvesting ( with the exception of elg1Δ strains which remained at 25°C for 3 h after reaching OD600 = 0 . 600 ) . Cell pellets were frozen at -80°C . Briefly , cells were lysed under denaturing conditions and protein extracts were prepared as previously described [88] . Extracts were incubated rotating overnight at room temperature with Ni-NTA conjugated agarose ( Qiagen ) to bind His6PCNA . After incubation , His6PCNA-bound beads were washed with buffers of decreasing pH to increase stringency with successive washes . His6-PCNA was eluted from the beads with an EDTA-containing buffer and eluates were fractionated by SDS-PAGE . Purified PCNA and modified forms were then analyzed by western blot using specific antibodies against PCNA , ubiquitin , and SUMO . Whole cell protein extraction was accomplished by TCA precipitation as previously described and fractionated by SDS-PAGE [89] . Western blots were probed using the following antibodies; anti-PCNA at 1:4000 dilution ( S871 , a gift from B . Stillman , CSHL ) , anti-SUMO at 1:3000 dilution ( A gift from X . Zhao , MSKCC ) , anti-ubiquitin at 1:1000 dilution ( P4D1 , Covance ) , anti-Rad53 at 1:1000 dilution ( A gift from JFX Diffley , LRI , UK ) , anti-phospho-S129 H2A at 1:1000 dilution ( ab15083 , Abcam ) , and anti-tubulin at 1:5000 dilution ( MMS-407R , Covance ) . Mutation rates were estimated by measuring the forward rate of mutations at the CAN1 locus that confer resistance to canavanine [90] . Determinations were made from the median of at least 14 independent cultures for each strain . Cultures were inoculated from single colonies in 5 ml of YPD medium and grown to saturation for 5 days at 30°C . Cells were then washed and diluted to appropriate concentrations before plating on medium lacking arginine and containing canavanine ( 60 mg/L ) . Dilutions were also plated on non-selective YPD to obtain a viable cell count . Mutation rates were calculated using Drake’s formula as previously described [91 , 92] . Significance was determined by the Mann Whitney U test as previously described [43] . Relative cell viability was measured using an assay referred to as the “spotting assay” . In this assay , 10 ml cultures were grown to saturation for 4 days at 25°C . Cells were then harvested , washed with water , quantified , and diluted to equal volumes containing 2x107 cells . 10-fold serial dilutions were made from these 2x107 cells in a 96-well plate and then “spotted” on rich medium using a multi-pronged spotting manifold . Replicates were generated for incubation at various temperatures . UV treatment ( 254nm ) was applied where indicated after plating using a UV crosslinker ( CL-1000 , UVP ) . Plates were imaged after 1 . 5 days of growth except where indicated . Cell cycle progression was measured by flow cytometry as previously described [8] . Briefly , 1 ml of liquid culture was pelleted and fixed in ice-cold 70% ethanol overnight . DNA was stained with Sytox Green ( Invitrogen ) and cells were analyzed using a BD Accuri C6 flow cytometer ( BD Biosciences ) . Peaks were quantified using the quantification feature of the BD Accuri C6 software . | Genome duplication via the process of DNA replication is a prerequisite for cell division and underlies the propagation of all living organisms . This fundamentally important mechanism has been highly conserved throughout eukaryotic evolution , allowing us to use the relatively simple and genetically tractable Saccharomyces cerevisiae as a model to better understand DNA replication in human cells . Furthermore , there is strong evidence to suggest that defects in DNA replication are prominent contributors to mutation and genome instability , a hallmark of cancer . Not surprisingly , evolution has selected for mechanisms to mitigate the effects of defective replication and avoid the most harmful outcomes . Postreplicative repair ( PRR ) pathways are two such mechanisms with well described functions in promoting the completion of replication under adverse conditions . In this study , we utilized a non-biased genome wide genetic screen to systematically identify conditions under which PRR is required . Our findings indicate that in addition to previously described roles in rescuing DNA synthesis defects , PRR is also required in response to aberrant DNA processing . Specifically , we report a requirement for PRR in cells lacking RAD27 , the yeast homolog of the tumor suppressor FEN1 . These findings expand the known functions of PRR and reveal their importance in promoting the viability of cells lacking a known tumor suppressor . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Genetic Interactions Implicating Postreplicative Repair in Okazaki Fragment Processing |
Components of promyelocytic leukaemia ( PML ) nuclear bodies ( ND10 ) are recruited to sites associated with herpes simplex virus type 1 ( HSV-1 ) genomes soon after they enter the nucleus . This cellular response is linked to intrinsic antiviral resistance and is counteracted by viral regulatory protein ICP0 . We report that the SUMO interaction motifs of PML , Sp100 and hDaxx are required for recruitment of these repressive proteins to HSV-1 induced foci , which also contain SUMO conjugates and PIAS2β , a SUMO E3 ligase . SUMO modification of PML and elements of its tripartite motif ( TRIM ) are also required for recruitment in cells lacking endogenous PML . Mutants of PML isoform I and hDaxx that are not recruited to virus induced foci are unable to reproduce the repression of ICP0 null mutant HSV-1 infection mediated by their wild type counterparts . We conclude that recruitment of ND10 components to sites associated with HSV-1 genomes reflects a cellular defence against invading pathogen DNA that is regulated through the SUMO modification pathway .
Herpesvirus infections are controlled by acquired and innate defences involving cellular , humoral and cytokine mediated responses ( for reviews , see [1] ) . In recent years a concept has emerged of an additional antiviral defence mechanism that operates within individual cells . Unlike cytokine-mediated responses , intrinsic antiviral resistance involves the actions of pre-existing cellular proteins that , in the case of herpesviruses , act to repress viral transcription [2] , [3] , [4] . This defence is counteracted by viral regulatory proteins , for example the immediate-early ( IE ) proteins ICP0 of herpes simplex virus type 1 ( HSV-1 ) [5] , [6] , [7] , ie1 ( IE72 ) of human cytomegalovirus ( HCMV ) [8] , and HCMV virion component pp71 [9] , [10] , [11] , [12] , [13] , [14] . One aspect of intrinsic resistance concerns cellular nuclear sub-structures known as ND10 or promyelocytic leukaemia ( PML ) nuclear bodies , and a number of their major components , namely PML itself , Sp100 , hDaxx and ATRX . In HSV-1 infections , ICP0 overcomes the repressive properties of these proteins by inducing their degradation or dispersal [7] , [15] , [16] , [17] . ICP0 null mutant HSV-1 exhibits a greatly reduced plaque forming efficiency , but this defect is partially reversed in cells depleted of PML , Sp100 , hDaxx or ATRX [5] , [6] , [7] . A notable feature of PML and other ND10 components is their recruitment to novel ND10-like foci that are closely associated with parental HSV-1 genomes and early replication compartments during the initial stages of infection [18] , [19] . The recruitment of PML to the virus-induced foci is not dependent on de novo viral protein expression and occurs extremely rapidly , indicating that the cell responds to the entry of viral genomes into the nucleus [18] , [20] . Although the effect can be seen in wild type ( wt ) HSV-1 infections , it is short lived as the recruited proteins are rapidly degraded or dispersed through the effects of ICP0 . During infection with ICP0 null mutant HSV-1 , however , the ND10 proteins remain in these novel sites in a much longer-lived manner . The correlation between the biological activity of many ICP0 mutant proteins and their ability to counteract this recruitment process [21] suggests that this phenomenon reflects an aspect of intrinsic antiviral resistance . This model proposes that the recruited proteins generate a repressive environment that impedes viral transcription . This paper concerns the molecular mechanism by which ND10 components are recruited to the virus-induced foci and tests the hypothesis that the recruitment process contributes to intrinsic resistance to HSV-1 infection . The formation of the virus-induced ND10-like structures is distinct from that of normal ND10 in uninfected cells because it is not dependent on PML or indeed any of the major ND10 components that have so far been studied [5] , [6] , [7] , [20] . Here , we have used a depletion/reconstitution approach to analyze the molecular requirement for the recruitment of PML , Sp100 and hDaxx to HSV-1 genome-associated sites in newly infected cells . We found that in all cases the presence of a SUMO interaction motif ( SIM ) [22] is required for this property , and that the major SUMO modification sites of PML , but not Sp100 , are also required . The virus-induced ND10-like foci also contain SUMO-2/3 conjugates and PIAS2β , a SUMO E3 ligase , even in the absence of PML . Unlike their wt counterparts , SIM deletion mutants of PML isoform I and hDaxx are unable to repress ICP0 null mutant HSV-1 infection . These data demonstrate that SUMO modification pathways play a key role in the recruitment of ND10 proteins to HSV-1 genome associated sites . We propose that this SUMO-dependent cellular response is an important component of intrinsic cellular defence against foreign DNA that has entered the nucleus .
The recruitment of PML and other proteins to sites associated with parental HSV-1 genomes and early replication compartments can be detected by examination of cells at the periphery of developing viral plaques . The viral genomes enter the nucleus of newly infected cells in a directional and asymmetric manner , and remain close to the nuclear envelope . The viral IE transcriptional regulator ICP4 avidly binds to viral DNA , and therefore forms foci that , in cells at the early stages of infection , are commonly distributed along one interior edge of the nucleus . PML and other ND10 proteins then accumulate at novel sites that are closely associated with the viral genomes [18] , [19] . This phenotype allows the unambiguous identification of proteins that have been recruited to and accumulate at the virus-induced sites . A typical example of endogenous PML recruitment in an ICP0-null mutant HSV-1 ( ΔICP0 ) infected HepaRG cell expressing a control shRNA is shown in Figure 1A ( left ) in comparison with a similarly infected PML depleted cell ( Figure 1A , right ) . We investigated the molecular characteristics of PML that are required for this recruitment by adopting a depletion/reconstitution approach [23] . All the PML isoforms studied ( Figure 1B ) , expressed as EYFP fusions , were recruited to sites associated with ICP4 foci in cells containing endogenous PML ( Figure 1C ) . In cells depleted of endogenous PML , however , while PML isoforms I to V were recruited , PML . VI remained entirely in large foci that were not associated with ICP4 ( Figure 1D ) . The separated channels of the images in Figures 1C and 1D are shown in Figures S1 and S2 . This assay is not amenable to precise quantification as the extent of recruitment varies between cells and is most obvious when the ICP4 foci are small . Nonetheless , when recruitment occurs it is evident in all cells with ICP4 foci near the nuclear periphery . A large number of cells were examined in each experiment , and we have scored a protein as defective in recruitment only when its behaviour was clearly different from the appropriate control . As an example , compare PML . VI in the control and PML depleted backgrounds in Figures 1C and D . Although recruitment of PML proteins that we have scored as recruitment positive occurred to some extent in all relevant cells , we noted some differences in PML isoform behaviour . For example , PML . I was recruited more extensively than PML . V ( Figure 1D ) . We have not investigated the basis of these differences . The defect in PML . VI recruitment to virus-induced foci also occurred in analogous human fibroblast ( HF ) derived cells . PML . I co-localized with Sp100 in both control and PML depleted HFs ( Figure S3 ) , both PML . I and PML . VI were recruited to the virus-induced in control HFs ( Figure S4A ) , but PML . VI was not recruited in PML depleted HFs ( Figure S4B ) . PML isoform VI includes all the conserved exons present in the other isoforms , except exon 7a ( Figure 1B ) . Therefore exon 7a includes sequences that are required for recruitment of PML to the virally induced foci . The defect in PML . VI recruitment is not exhibited when endogenous PML is present because PML interactions mediated through the coiled-coil motif [24] , [25] enable the defective protein to be recruited in partnership with the endogenous isoforms . PML exon 7a encodes an 18 amino acid segment containing a SIM that has been implicated in proper ND10 assembly [26] . Mutants of PML . I and PML . IV that lack exon 7a ( Figure 2A ) . were expressed at levels similar to their wt versions , with similar patterns of SUMO modification , and co-localized with Sp100 in both control and PML depleted cells ( Figure 2B and C ) . While both mutants were recruited to virus-induced foci in the presence of endogenous PML , both were defective in recruitment in PML depleted cells ( Figures 2D and S5A ) . This defect of PML . I . Δ7a was confirmed in HF derived cells ( Figure S5B and C; the localization of this protein in uninfected HFs is shown in Figure S3 ) . EYFP-PML . I with substitution mutations in the SIM ( PML . I . mSIM ) exhibited properties very similar to those of PML . I . Δ7a in uninfected and ICP0-null mutant HSV-1 infected control and PML depleted cells ( Figure S6 ) . These data confirm that the lack of recruitment of PML . VI is due to the absence of the SIM in exon 7a . SUMO modification of PML is essential for ND10 assembly in uninfected cells [27] . Therefore we investigated whether this modification , in addition to the SIM , plays a role in PML recruitment to the virus-induced foci . Mutants of PML . I or PML . IV with lysine to arginine substitutions at residues 160 and 490 ( PML . I . KK and PML . IV . KK; Figure 3A ) exhibit substantial defects in SUMO modification , but some apparently modified species remain , especially in cells containing endogenous PML , and more so with PML . IV than PML . I . Modification of PML . I . KK in PML depleted cells was almost completely eliminated ( Figure 3B ) . PML . I . KK and PML . IV . KK localized to ND10 normally in cells expressing endogenous PML , probably through interactions with endogenous PML via their coiled-coil domains ( Figure 3C , upper row ) . In contrast , and consistent with previous work [27] , PML . I . KK exhibited a drastically altered localization in PML depleted cells , forming a reduced number of foci of increased size , some of which were in the cytoplasm . The nuclear foci contained greatly reduced amounts of Sp100 . The aberrant PML . IV . KK foci were predominantly nuclear , again with little co-localization with Sp100 ( Figure 3C , lower row ) . Residual co-localization with Sp100 of these PML mutants may be influenced by potential SUMO modification at other lysine residues . However , it is also affected by cell type since in the corresponding PML depleted HFs co-localization of PML . I . KK with Sp100 was more marked ( Figure S9A ) . To eliminate any residual SUMO modification of PML . I . KK we introduced additional mutations at either lysine 65 [28] or at lysine 616 , which lies in a good match ( LKID ) to the consensus SUMO modification site ( ΨKXE ) . Mutant PML . I . K123 ( K65R , K160R , K490R ) exhibited similar residual SUMO modification to that of PML . I . KK in cells expressing endogenous PML , whereas the K616R mutation virtually eliminated all modified bands in mutants PML . I . K234 ( K160R , K490R , K616R ) and PML . I . 4KR ( all indicated lysine residues substituted with arginine ) ( Figure S7 ) . We conclude that PML . I is not detectably modified at lysine 65 , but that lysine 616 is likely to be a SUMO modification site that is specific for PML . I and PML . IV ( because it lies in exon 8a which is present in these two isoforms only ) . We note that lysine 65 is very close to zinc coordinating cysteine residues in the RING finger , so SUMO modification here would likely affect both the overall architecture of the RING and its interaction interfaces . PML . I . 4KR co-localized with Sp100 in control but not PML depleted cells , in both HepaRG and HF backgrounds ( Figures S8A and S9A ) . These data support the hypothesis that the sporadic co-localization of PML . I . KK with Sp100 in PML depleted cells ( Figure 3C ) is due to the remaining potential for SUMO modification . As in the case of the SIM mutants , both PML . I . KK and PML . IV . KK could be recruited to virus-induced foci in cells expressing endogenous PML ( Figure 3D , upper row ) , but recruitment was either absent or at greatly reduced levels in PML depleted cells ( Figure 3D , lower row , and Figure S8B ) . Similar results were obtained in the equivalent HF derived cells ( Figure S9B ) , and with the 4KR mutant in both HepaRG and HF backgrounds ( Figures S8C and S9B ) . Therefore , in addition to the role of the SIM , mutation of the major SUMO modification sites of PML severely impedes the capacity of the protein to be recruited to sites associated with viral genomes , providing endogenous PML isoforms are absent . The defect of mutant PML proteins in recruitment to virus-induced foci could be explained by a decrease in intrinsic mobility . The dynamics of PML isoforms and SUMO modification deficient mutants in both the presence and absence of endogenous PML have been reported [18] , [29] , [30] , [31] , but the role of the SIM has not been investigated before . We analyzed the dynamics of all EYFP-linked PML isoforms , and the Δ7a and KK mutants of PML . I and PML . IV , in both control and PML depleted HepaRG cells . Typical FRAP recovery plots of PML . I and PML . I . KK in control cells are presented in Figure 4A , confirming previous conclusions that SUMO modification is required for the normal mobility of PML [30] , [32] . The data for all PML isoforms and mutants in both cell backgrounds are presented as their degree of fluorescence recovery at the 112-second time point ( Figure 4B ) . The results are broadly in agreement with previous work , with the exception that the relatively slow recovery of PML . V [30] was not reproduced here . The only major differences were between the SUMO modification deficient mutants and the wt forms . Importantly , the defect in recruitment to virus-induced foci of PML . VI and the PML . I and PML . IV mutants lacking exon 7a cannot be explained simply by reduced mobility , whereas in the case of the SUMO modification mutants this explanation cannot be excluded . Control and PML depleted cell lines that express EYFP-PML . I with a deletion of the coiled-coil or point mutations in the RING finger , B-Box 1 or B-Box 2 ( Figure 5A ) have been described previously [23] . Briefly , the RING finger mutant ( ΔRING ) co-localized with endogenous ND10 in control cells but was mostly nuclear diffuse in cells lacking endogenous PML , the coiled-coil deletion mutant ( ΔCC ) was nuclear diffuse in both cell backgrounds , the B-Box 2 mutant ( ΔBB2 ) formed foci in both types of cell but in neither case were these co-localized with Sp100 , and the B-Box 1 mutant ( ΔBB1 ) co-localized with Sp100 in aberrant ND10-like structures in control cells and formed foci in a proportion of PML depleted cells , none of which co-localized with Sp100 . SUMO modification of all tripartite motif ( TRIM ) mutants was highly compromised in both cell backgrounds [23] . The ΔRING and ΔBB2 mutants were recruited efficiently to the virus-induced foci in cells expressing endogenous PML , whereas the ΔBB1 and ΔCC mutants were not ( Figures 5B and S10A ) . The ΔBB2 mutant was also efficiently recruited in PML depleted cells ( Figure 5B ) , even though this mutant was not well SUMO-modified and did not efficiently co-localize with Sp100 in uninfected cells in either cell background [23] . Therefore the characteristics of PML required for nucleating an ND10-like structure in uninfected cells are distinguishable from those involved in recruitment to the novel virus-induced structures . As in cells expressing endogenous PML , the ΔBB1 and ΔCC mutants were not recruited into virus-induced foci in PML depleted cells ( Figure 5B ) . The most variable results were obtained with the ΔRING mutant in PML depleted cells . In most cells infected with ICP0 null mutant HSV-1 this mutant remained nuclear diffuse , but in some cells it was recruited with variable efficiencies ( Figure 5B ) . We conclude that whereas B-Box 1 and the coiled-coil are essential for PML recruitment to virus induced foci , B-Box 2 is dispensable , and inactivation of the RING greatly diminishes but does not always eliminate recruitment . Given that recruitment of PML to sites to HSV-1 induced foci is dependent on its SIM , we investigated whether recruitment of other ND10 proteins is also SIM dependent . Sp100A is the smallest and most abundantly expressed Sp100 isoform . It includes a SIM ( residues 323–326 ) , a major SUMO modification site at lysine 297 , and a region near the N-terminus that is required for localization to ND10 the HSR domain ( Figure 6A ) [33] , [34] . Control and Sp100-depleted HepaRG cells [6] were transduced with lentivirus vectors expressing EYFP-linked Sp100A and mutants lacking the HSR domain ( ΔHSR ) , the SUMO modification site ( K297R ) or the SIM ( mSIM ) , or a combination of both K297R and mSIM mutations ( ΔSSIM ) . Consistent with previous studies [33] , [34] , none of the mutant proteins exhibited the SUMO modification pattern characteristic of wt Sp100A ( Figure 6C ) . The mSIM mutant of Sp100A at least partially co-localized with PML in control cells , and also in Sp100 depleted cells expressing higher levels of the recombinant protein ( Figure 7B , leftmost panels ) , although it was nuclear diffuse when weakly expressed in Sp100 depleted cells ( not shown ) . Both wt and K297R mutant EYFP-Sp100A were recruited to the virus-induced foci in both control and Sp100 depleted cells ( Figure 7A and C ) , indicating that , unlike PML , the major SUMO modification site is not required for Sp100A recruitment . Although the HSR deletion mutant was not so recruited ( Figure 7C , right-most panels , and Figure S10B ) , it is possible that this deletion causes major structural defects because this mutant does not localize to ND10 in uninfected cells ( Figure 7B ) . The Sp100A . mSIM and ΔSSIM mutants were recruited to virus-induced foci in cells containing endogenous Sp100 , but not in its absence ( Figure 7C , leftmost panels , and Figure S10B; data not shown for ΔSSIM ) . These results indicate that the SIM of Sp100 , like that of PML , is required for recruitment , but in its absence the mSIM mutant can be recruited through an interaction with endogenous Sp100 , most likely through the HSR domain [34] . Like PML and Sp100 , hDaxx includes a SIM ( at its C-terminus; Figure 8A ) and is also recruited to HSV-1 induced ND10-like foci [18] , [35] . Both wt and mSIM mutant versions of hDaxx were expressed in control ( data not shown ) and hDaxx depleted cells ( Figure 8B ) . The wt protein co-localized with PML in ND10 ( Figure 8C , upper row , left ) , but the mSIM mutant was diffusely distributed in the nucleus ( Figure 8C , upper row , right ) . As with PML and Sp100 , wt but not SIM mutant EYFP-hDaxx was recruited to the virus-induced foci in hDaxx depleted cells ( Figures 8C and S10C ) . The requirement of the SIMs of PML , Sp100 and hDaxx for recruitment to the HSV-1 induced foci suggests that these proteins may be interacting with components of the SUMO conjugation pathway at these locations . Therefore we examined the behaviour of SUMO isoforms and the SUMO E3 ligase PIAS2β in this experimental system . We found that PIAS2β , as detected by an antibody that recognizes the endogenous protein , is an ND10 component ( Figure 9C , upper left ) . This suggests that PIAS proteins could be involved in ND10 assembly , consistent with the observation that ectopically expressed tagged PIAS1 localizes to ND10 in Vero cells [36] . In ICP0-null mutant HSV-1 infected cells , SUMO-1 , SUMO-2/3 and PIAS2β were clearly recruited to the virus-induced foci ( Figure 9A–C , right-hand panels ) . It could be argued that the presence of SUMO isoforms would be expected because both PML and Sp100 are recruited to the foci , and both are heavily SUMO modified . However , recruitment of SUMO-2/3 was readily evident in PML depleted cells ( in which SUMO modification of Sp100 is highly compromised ) ( Figure 9B ) , and although recruitment of SUMO-1 and PIAS2β was very weak at best in PML-depleted HepaRG cells ( Figures 9A , 9C and S11D ) , recruitment of PIAS2β remained strong in PML-depleted HFs ( Figure S11C ) . The presence of a SUMO E3 ligase in the virus-induced foci suggests that the recruited SUMO-2/3 signal could include newly generated SUMO conjugates , and it is possible that this activity drives the PML-independent but SIM-dependent recruitment of proteins such as hDaxx . Consistent with this hypothesis , we have found that the recruitment of PML and SUMO isoforms requires Ubc9 ( the sole SUMO E2 conjugating enzyme ) , and that the overall level of high molecular weight SUMO conjugates increases in ICP0 null mutant HSV-1 infections ( C . Boutell , D . Cuchet-Lourenço , E . Vanni , A . Orr , and R . D . Everett , unpublished data ) . We investigated the biological significance of the recruitment of PML and hDaxx to sites associated with HSV-1 genomes by comparing the plaque formation efficiencies of wt and ICP0 null mutant HSV-1 in various control and reconstituted cells . Depletion of either PML or hDaxx increases the plaque formation efficiency of ICP0 mutant HSV-1 while not affecting that of the wt [5] , [6] , [7] . As expected , the plaque forming efficiency of the wt virus was not significantly different in any of the cell lines tested here ( Figures 10A and 10B , upper histograms ) . Increased plaque formation of ICP0 null mutant HSV-1 was observed in cells depleted of either PML or hDaxx , and reintroduction of PML . I and hDaxx reversed these phenotypes partially and completely , respectively ( Figures 10A and 10B , lower histograms ) , confirming previous work [7] , [23] . However , reintroduction of the mSIM mutants of PML . I and hDaxx had no effect on ICP0 null mutant HSV-1 plaque formation ( Figures 10A and 10B , lower histograms ) . It may be relevant that mutation of its SIM reduces the transcriptional repression activity of hDaxx [35] . We have reported previously that PML . I . KK , PML . I . Δ7a are unable to reproduce the repressive effect of PML . I on ICP0 null mutant HSV-1 plaque formation [23] . The analogous experiments with Sp100 were not informative because reintroduction of Sp100A into Sp100 depleted cells did not reverse the effect of Sp100 depletion ( data not shown ) , possibly because it is the longer isoforms of Sp100 ( Sp100B , -C and -HMG ) that are thought to act as repressors , rather than Sp100A [37] , [38] , [39] . There is therefore a correlation between the recruitment of PML and hDaxx to foci associated with HSV-1 genomes and their involvement in intrinsic resistance to virus infection .
The recruitment of ND10 components to sites that are closely associated with parental HSV-1 genomes and early replication compartments is a dramatic cellular response to entry of the viral genome into the nucleus . It occurs very rapidly , being detectable as early as 30 minutes after addition of virus to a cell monolayer [20] , and it occurs independently of de novo viral protein synthesis , implying that it is the viral DNA itself ( perhaps in association with viral tegument proteins ) that signals the response [18] . These observations raise several important questions , including the nature of the mechanism underlying the recruitment , the biological significance of the response , and the wider implications of these events . We found that endogenous PML and Sp100 have dominant influences on the behaviour of introduced mutant forms of these proteins , obscuring the role of the mutated motifs ( Figures 1–3 , 5 and 7 ) . Expression of the reintroduced proteins at close to endogenous levels is also important , as when expressed in excess a protein may be unable to interact efficiently with limiting binding partners in the cell . With these issues overcome , we found that the recruitment of PML to the virus-induced foci depends on its SIM , despite the fact that the mSIM mutant is indistinguishable from the wt in terms of localization and SUMO modification in uninfected cells ( Figure S6 ) . Recruitment of PML to the virus-induced sites is also compromised by mutation of the major SUMO modification sites ( K160 and K490 ) and by alterations in the B-Box 1 , coiled-coil and RING motifs ( Figures 3 and 5 ) . It is unsurprising that elements of the TRIM are important for PML behaviour because these mutations greatly reduce SUMO modification [23] and are likely to have major consequences on PML structure and interactions . It has been reported that the SIM of PML is required for normal ND10 assembly [26] . Our finding that PML . I . mSIM and PML . I . Δ7a co-localize with Sp100 in PML depleted cells is not inconsistent with the data in this previous study , since the effect of removal of the SIM was observed only when the PML isoform used ( PML . III ) also lacked SUMO modification sites . Thus the SIM mutant of PML . III still co-localized with hDaxx and SUMO , even in PML ( −/− ) mouse fibroblasts [26] . That the SIM is not essential for ND10 localization is also consistent with previous studies utilizing PML . VI in PML depleted cells [8] , [23] , because this is a natural SIM deletion mutant . The SIMs of both Sp100 and hDaxx are also important for their recruitment to the virus-induced foci ( Figures 7 and 8 ) . Given that SUMO modification of Sp100A is not required ( Figures 6 and 7 ) , and hDaxx appears not to be detectably SUMO modified in our system ( Figure 8 ) , we conclude that it is the SIM rather than SUMO modification that is the common essential feature for recruitment of these proteins to the virus-induced foci . The conclusion that SUMO modification of Sp100A is not required for this behaviour is supported by the observation that Sp100 is not SUMO modified in PML depleted cells , yet is still recruited to the viral foci [5] , [6] . The importance of the SIM in recruitment to the virus-induced foci implies that PML , hDaxx and Sp100 are responding to an earlier SUMO-dependent event at these sites . This conclusion is supported by our recent data demonstrating that Ubc9 is required for efficient recruitment of PML to the virus induced foci ( C . Boutell , D . Cuchet-Lourenco , E . Vanni , A . Orr , and R . D . Everett , unpublished data ) . This raises the question of which factors these ND10 components are interacting with , through their SIMs , to enable their recruitment to the virus-induced foci . Recruitment of PML is not dependent on de novo viral gene expression [18] , implying that the initial event involves cellular proteins and their recognition of the viral genomes . Because the viral DNA is not chromatinized at the time of entry into the nucleus , its conformation is entirely different from cellular chromatin . It would be expected that many cellular proteins would become associated with the initially naked viral DNA , and one or more of these could initiate the SUMO-dependent recruitment process . The detection of PIAS2β in the novel foci implies that SUMO E3 ligases are involved in the assembly of the novel foci . This is the first report that PIAS2β is involved in ND10 biology , and it is intriguing that the protein has also been implicated in DNA damage and interferon pathways [40] , [41] . The mechanisms that underlie formation of the virus-induced structures are distinct in a number of respects from those required for normal ND10 assembly . For example , PML is not required for recruitment of Sp100 and hDaxx to the virus-induced foci [5] , [6] , [20] , and depletion of any one of these three proteins does not eliminate recruitment of the remaining two [6] , [7] . Some of the PML mutants also illustrate the differences between normal and virus-induced ND10 related structures: PML . I . ΔBB2 does not colocalize with Sp100 in uninfected cells [23] but is recruited very efficiently to the virus-induced foci ( Figure 5 ) . On the other hand , recruitment of ATRX to both normal and virus-induced ND10 structures is dependent on hDaxx [7] , illustrating that both specific protein-protein interactions and SIM dependent interactions are involved in building the virus-induced foci . PML-PML interactions through the coiled-coil and Sp100-Sp100 interactions through the HSR motif also influence assembly of the virus-induced structures ( Figures 2 , 3 and 7 ) . A picture emerges that the building of these foci involves multiple interactions between distinct proteins , protein dimerization events , and SIM-SUMO interactions . Our data imply a general importance for SUMO related pathways in the nucleation of the HSV-1 induced ND10-like foci . Because SUMO-2/3 and PIAS2β can also be recruited into these structures in a PML-independent manner ( Figures 9 and S11 ) , it is attractive to speculate that these events reflect ongoing SUMO conjugation events . We note that the genomes and replication compartments of many DNA viruses are closely associated with PML and ND10-like structures ( reviewed in [42] , [43] ) . It would be surprising if the principles revealed here concerning HSV-1 infection were not involved in ND10 association with other viral genomes , and there is evidence that this is the case in human cytomegalovirus infected cells [8] , [44] . Therefore the events reported here are likely to reflect a more general cellular response to foreign DNA entering the nucleus . It is possible that the cell is responding to viral genomes in a manner related to that of the DNA damage response . Recent evidence demonstrates that SUMO modification is intimately involved in the assembly of DNA damage response foci , and SUMO conjugates are present at these locations [45] , [46] . ICP0 inhibits the formation of these structures by inducing the degradation of RNF8 and RNF168 [47] , but it is possible that ICP0 might also impede any recruitment of SUMO conjugated proteins to DNA damage foci , as it does in regard to the viral genome associated ND10-like foci ( see below ) . These observations raise the intriguing possibility of commonalities between the DNA damage response , the assembly of HSV-1 induced ND10-like foci , and intrinsic resistance to HSV-1 infection . The observations that the SIM mutants of PML . I and hDaxx are unable to reproduce the repression of ICP0 null mutant HSV-1 infection conferred by the wt proteins ( Figure 10 ) imply that the recruitment of these proteins to the virus-induced foci is biologically significant and contributes to the repression of ICP0 null mutant HSV-1 infection . Recruitment of all ND10 component proteins so far studied is counteracted by ICP0 [7] , [18] , [20] . ICP0 inhibits PML recruitment by inducing its degradation [15] , [17] . This simple mechanism does not , however , explain why recruitment of all ND10 proteins is inhibited by ICP0 . Although ICP0 promotes the loss of SUMO modified Sp100 [16] this cannot explain why Sp100 recruitment is inhibited because the K297R mutant is still recruited ( Figure 7 ) . Furthermore , ICP0 does not promote the degradation of either hDaxx or ATRX [7] . Previous work has demonstrated that ICP0 induces the widespread degradation of SUMO conjugated proteins [15] . This activity provides an attractive explanation of how ICP0 inhibits recruitment of this group of proteins to virus-induced foci , since degradation of SUMO conjugates would eliminate SIM dependent interactions . These arguments suggest a direct link between SUMO-dependent pathways and the mechanism of intrinsic cellular resistance to HSV-1 infection that is counteracted by ICP0 . We propose that the cell responds to foreign DNA that enters the nucleus by stimulating SUMO conjugation events at sites associated with the introduced DNA , leading to recruitment of other proteins in a SIM dependent manner and resulting in a repressive environment . We note that there are several examples of factors involved in transcriptional repression that are regulated by SUMO modification [48] , and that SUMO modification pathways have been linked to a general cellular response to pathogens [49] . This concept is strengthened by our related studies ( C . Boutell , D . Cuchet-Lourenço , E . Vanni , A . Orr , and R . D . Everett , unpublished data ) that demonstrate the involvement of Ubc9 in intrinsic antiviral resistance , and that ICP0 has SUMO-targeted ubiquitin ligase activities that play an important role in its ability to counteract this resistance .
U2OS , HEK-293T and human fibroblast cells were grown in Dulbecco's Modified Eagles' Medium with 10% fetal calf serum ( FCS ) . Baby hamster kidney ( BHK ) cells were grown in Glasgow Modified Eagles' Medium with 10% new born calf serum and 10% tryptose phosphate broth . HepaRG cells [50] were grown in William's Medium E with 10% fetal bovine serum Gold ( PAA Laboratories Ltd ) , 2 mM glutamine , 5 µg/ml insulin and 0 . 5 µM hydrocortisone . All cell growth media contained 100 units/ml penicillin and 100 µg/ml streptomycin . Lentivirus vector plasmids expressing shRNAs , EYFP-hDaxx , and EYFP-PML isoforms I to VI , mutants of PML . I with lesions in the RING finger ( ΔRING ) , B-Box 1 ( ΔBB1 ) , B-Box 2 ( ΔBB2 ) , the coiled-coil motif ( ΔCC ) and at SUMO modification sites K160 and K490 were as described [6] , [7] , [23] . The following PML mutants were constructed using PCR splicing with mutagenic oligonucleotides: PML . I . Δ7a and PML . IV . Δ7a ( precise deletions of exon 7a in PML . I and PML . IV cDNAs ) ; PML . IV . KK ( K160R , K490R mutations in the PML . IV background ) ; PML . I . mSIM ( residues 566–569 VVVI mutated to VGGG ) ; PML . I . K123 ( mutations K65R , K160R , K490R ) ; PML . I . K234 ( K160R , K490R , K616R ) ; PML . I . K1-4 ( lysine residues K65 , K160 , K490 and K160 mutated to arginine ) . Lentivirus transduced HepaRG cells expressing EYFP-linked proteins were sorted by FACS , as described [23] . HFs expressing control and anti-PML shRNAs [5] and EYFP-PML . I , PML . VI , PML . I . Δ7a , PML . I . KK and PML . I . K1-4 were isolated using the same methodology . Lentivirus vectors expressing EYFP-Sp100 isoform A ( using a cDNA resistant to the anti-Sp100 shRNA ) and derivatives lacking the HSR region ( Sp100 . ΔHSR , deletion of residues 68–146 ) , the major SUMO modification site and the SIM combined ( Sp100 . ΔSSIM , deletion of residues 289–327 ) , the major SUMO modification site alone ( Sp100 . K297R ) and with point mutations in the SIM alone ( residues 323 to 326 IIVI changed to IGAG; Sp100 . mSIM ) were constructed using PCR splicing . A lentivirus vector expressing EYFP- hDaxx with lesions in the SIM ( residues 733 to 736 IIVI changed to IGAG ) was constructed by PCR directed mutagenesis . HepaRG cells expressing wt and mutant hDaxx were enriched by FACS . Lentivirus vector plasmids were co-transfected into HEK-293T cells with helper plasmids pVSV-G and pCMV . DR . 8 . 91 , then the supernatants were used to transduce HF or HepaRG cells . All shRNA vectors express puromycin for selection ( initially 1 µg/ml , then reduced to 0 . 5 µg/ml during subsequent passage ) , and all expressing a protein of interest confer G418 resistance ( selection initially 1 mg/ml , then reduced to 0 . 5 mg/ml during subsequent passage ) . Wild type HSV-1 strain 17syn+ and its ICP0 null mutant derivative dl1403 [51] were grown in BHK cells and titrated in U2OS cells . Derivatives of wt and ICP0-null mutant HSV-1 that include a β-galactosidase gene linked to the human cytomegalovirus immediate-early promoter/enhancer ( in1863 and dl1403/CMVlacZ ) were used for plaque assays , as described [5] . Cells in 24-well dishes at 1×105 cells per well were washed with phosphate buffered saline PBS ) before harvesting in SDS-PAGE loading buffer . Proteins were resolved on 7 . 5% SDS-gels , then transferred to nitrocellulose membranes by western blotting . The following antibodies were used: anti-actin mAb AC-40 ( Sigma-Aldrich ) ; anti-PML mAb 5E10 [52]; anti-Sp100 rabbit serum SpGH [53]; anti-hDaxx rabbit polyclonal D7801 ( Sigma-Aldrich ) ; anti-EGFP rabbit polyclonal ab290 ( Abcam ) . Cells on 13 mm glass coverslips were fixed using 1 . 5% ( v/v ) formaldehyde in PBS containing 2% sucrose then treated with 0 . 5% Nonidet P40 substitute ( EuroClone S . p . A . ) in PBS/10% sucrose . PML was detected with mAb 5E10 and ICP4 with mAb 58S . Rabbit polyclonal antibodies were used for Sp100 ( SpGH ) , hDaxx ( 07-471 , Upstate ) , SUMO-1 ( ab32058 , Abcam ) , SUMO-2/3 ( ab3742-100 , Abcam ) , PIAS2β ( gifted by Mary Dasso ) . The secondary antibodies were FITC conjugated goat anti-mouse IgG ( Sigma ) , Alexa 488 and Alexa 633 conjugated goat anti-rabbit and anti-mouse IgG , and Alexa 555 conjugated donkey anti-rabbit and anti-mouse IgG , ( Invitrogen ) . A glycerol-based mounting medium was used ( Citifluor AF1 ) . The samples were examined using a Zeiss LSM 510 confocal microscope with 488 nm , 543 nm and 633 nm laser lines and a ×63 Plan-Apochromat oil immersion lens , NA 1 . 40 . Exported images were processed using Adobe Photoshop with minimal adjustment , then assembled for presentation using Adobe Illustrator . Cells were seeded into modified 35 mm dishes with the central area replaced by coverslip glass ( MatTek Corporation ) . Fluorescence recovery after photobleaching ( FRAP ) was conducted using an LSM 510 META microscope with full environmental control . In each experiment , 3 PML foci in 10 different cells were subject to bleaching ( 100% power of the 514 nm laser , 10 reiterations ) , then 20 images were captured over a period of approximately two minutes during the recovery phase . Regions of interest were analyzed by subtracting the average pixel intensity in unbleached background areas and normalizing to any changes in overall intensity of a similar unbleached PML structure to control for any bleaching of the scanned areas during image acquisition . The data for each protein were assembled into an Excel file for graphical representation of the average plus standard deviation at each time point . | Viruses encounter several different defences that impede infection , including acquired immunity mediated by the immune system and innate immunity that includes the synthesis of antiviral proteins through the interferon pathway . In recent years , a third arm of antiviral defence has been described , named intrinsic immunity or intrinsic resistance , that is conferred by constitutively expressed cellular proteins . In the case of herpesviruses , intrinsic resistance involves the action of cellular repressors that restrict viral transcription once the viral genome enters the nucleus . Several studies have presented evidence that one aspect of intrinsic resistance involves cellular proteins that form distinct nuclear structures known as ND10 . Several ND10 components are known to accumulate rapidly at sites in close association with herpes simplex virus type 1 genomes . Here we report that this cellular response requires the ability of several of the proteins in question to interact with a small ubiquitin-like protein known as SUMO . In two such examples of these proteins , we show that their ability to interact with SUMO is required for their roles in repressing viral infection . We suggest that this SUMO-dependent pathway may underlie a more general mechanism by which cells protect themselves from invading foreign DNA . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"cellular",
"structures",
"mechanisms",
"of",
"resistance",
"and",
"susceptibility",
"molecular",
"cell",
"biology",
"cell",
"biology",
"virology",
"biology",
"microbiology",
"viral",
"replication",
"cell",
"nucleus"
] | 2011 | SUMO Pathway Dependent Recruitment of Cellular Repressors to Herpes Simplex Virus Type 1 Genomes |
Discovering genetic mechanisms driving complex diseases is a hard problem . Existing methods often lack power to identify the set of responsible genes . Protein-protein interaction networks have been shown to boost power when detecting gene-disease associations . We introduce a Bayesian framework , Conflux , to find disease associated genes from exome sequencing data using networks as a prior . There are two main advantages to using networks within a probabilistic graphical model . First , networks are noisy and incomplete , a substantial impediment to gene discovery . Incorporating networks into the structure of a probabilistic models for gene inference has less impact on the solution than relying on the noisy network structure directly . Second , using a Bayesian framework we can keep track of the uncertainty of each gene being associated with the phenotype rather than returning a fixed list of genes . We first show that using networks clearly improves gene detection compared to individual gene testing . We then show consistently improved performance of Conflux compared to the state-of-the-art diffusion network-based method Hotnet2 and a variety of other network and variant aggregation methods , using randomly generated and literature-reported gene sets . We test Hotnet2 and Conflux on several network configurations to reveal biases and patterns of false positives and false negatives in each case . Our experiments show that our novel Bayesian framework Conflux incorporates many of the advantages of the current state-of-the-art methods , while offering more flexibility and improved power in many gene-disease association scenarios .
Identifying genes associated with complex human diseases is the holy grail of human genetics . Years of research have proven this task to be very difficult , requiring substantial data and methodological resources , and far from being solved [1–4] . Recently , protein-protein interaction ( PPI ) networks have been successfully used to improve the power of detecting genes associated with genetic diseases [5] . Examples of success stories include Crohn’s disease and diabetes [6] , autism and related disorders [7 , 8] , multiple cancers [9] and many others [10 , 11] . Some of the first methods developed to take advantage of gene and protein interaction networks , such as dmGWAS , DAPPLE , NIMMI , iPINBPA , jActiveModules [12–16] , directly relied on the network annotations to iteratively search for neighbourhoods of associated genes while greedily exploring the network . More recently , several methods have used network diffusion to identify subnetworks associated with disease . Diffusion methods take into account the local topology of the network ( degrees of nodes , average random walk distance ) in order to propagate the heat ( association signal ) and then identify hot subnetworks . Hotnet and Hotnet2 [17 , 18] are examples of network diffusion methods successfully used to discover modules associated with several cancers . Although Hotnet2 was initially applied on gene-based statistics from cancer data , several recent publications used it successfully in Genome Wide Association Study ( GWAS ) context discovering subnetworks of genes statistically associated with diseases such as Ulcerative Colitis and ADHD [19] . The drawback of the previous methods is their reliance on the marginal statistics of each gene , such as p-values of marginal gene significance , to diffuse over the network . Recent methods [20 , 21] go beyond the diffusion of test statistics by ‘smoothing’ or propagating the data from each individual patient , i . e mutations , across the network . In this type of network propagation approach , genes neighbouring the gene with a mutation get some of its signal even if they are not mutated in the considered patient . For example , NBS [20] used this approach to identify relevant cancer subtypes and then selected subnetworks that had strong associations with these subtypes . In [21] , the approach was used to identify silent genes that are neighbours of mutated or differentially expressed genes . While this postulates an interesting hypothesis for how complex diseases arise , it is very hard to test such a hypothesis in a computational setting , as essentially all the evidence for the importance of these genes stems from the network with no direct support from the patients’ genetic data . More recently , [22] proposed an ILP algorithm for detecting subnetworks that are mutated in a large fraction of patients . This approach offers many advantages over methods using gene summary statistics: i ) it directly optimizes a function of interest ( explaining the most patients ) ; ii ) it takes advantage of underlying structures in the data such as mutual exclusivity between genes , such that solutions that include genes that are mutated in different groups of patients when taken together are considered preferable to sets of genes that are all mutated in the same patient . This type of structure was not directly exploited in methods such as [20 , 21] that used gene-based test statistics for diffusion . NBS [20] , ILP [22] and the silent-gene [21] methods mentioned above were successfully applied to TCGA cancer data . Unfortunately , this idea is not directly applicable to heritable diseases because there are many more germline genetic variants in cases and controls compared to thousands of somatic mutations specific to cancer . Additionally , none of the methods mentioned above account for the uncertainty in predicting which genes are associated with the phenotype , given the limited sample size . In this paper , we propose a biologically motivated hierarchical graphical model to identify sets of genes explaining complex human diseases from exome data . Probabilistic graphical models ( PGMs ) [23] are a framework to represent joint distributions and the conditional dependencies among variables . Hierarchical PGMs were previously used in computational biology , for example to identify regulatory network modules [24] and infer haplotype blocks [25] . Here we define a new PGM model to effectively address the problem of disease mechanism identification while incorporating PPI network knowledge into the framework . Similarly to Hotnet2 , our method is flexible enough to work on both cancer and hereditary diseases , and similarly to [22] , our method uses the full genetic data , not just summary statistics , taking into account complementarity and mutual exclusivity between genes to find the set of genes that explains the most patients . Most network-based approaches directly rely on the structure of the network when searching for gene subnetworks of interest . They either use the PPI network as a guide and directly select subnetworks from it , or they use it to diffuse information across genes . In addition to incorporating the characteristics of the current state-of-the-art methods , encoding the network as part of the structure of a probabilistic graphical model allows our method to be less susceptible to the noise intrinsic to protein-protein interaction networks . Our approach is able to find relevant sets of genes even when the sets are not or only partially supported by the network but are supported by the DNA aberrations in and across patients , limiting the influence of incompleteness and noise in the networks . Finally , our method returns marginal probabilities for each gene rather than a fixed associated subnetwork , keeping track of the uncertainty in predicting gene-disease associations . We tested our approach to identify a set of genes driving a complex human disease from exome data in a wide variety of settings . We first sampled various network structures comparing our probabilistic method to a very widely used diffusion method , Hotnet2 , as well as the most widely used rare-variant aggregation method SKAT-O [26] . We then compared the ability of both methods along with other network methods including dmGWAS , JActiveModule and PINBPA as well as enrichment method DAVID [27] , to recover true sets of genes previously identified in epilepsy , schizophrenia , autism and ovarian cancer from the simulated exome data . Our experiments indicate that our approach has higher sensitivity and precision than Hotnet2 in > 90% of the considered scenarios . We conclude that not relying on the network structure directly but encoding it as part of a structure in a Bayesian graphical model is a powerful new way to improve gene driver discovery in complex human diseases .
We assessed the performance of SKAT-O [26] , Hotnet2 applied on gene p-values from SKAT-O , and our graphical-model-based approach Conflux . Similarly to most other network-based methods , the output of Hotnet2 is a fixed set of genes ( subnetwork ) that is associated as a whole with the disease studied . A gene is either included in this list or not . In contrast , Conflux returns marginal probabilities for each gene . SKAT-O returns p-values for every gene . Thus , both Conflux and SKAT-O allow us to examine prioritization and ranking of genes . For this reason , we evaluate SKAT-O and Conflux in two ways: i ) selecting genes based on thresholds , SKAT-O p-values ( < 4 . 10−6 ) and Conflux marginal probabilities ( ≥ 0 . 2 ) , the values chosen to control for Type I error; ii ) selecting the top P ranked genes , where P is the number of causal genes . We assessed sensitivity , precision and the F-Measure ( a harmonic mean of precision and sensitivity ) for each considered method . Fig 1 presents the results across all simulations , each bar in the barplot summarizing 20 simulated disease scenarios . Fig 1A–1C show that Conflux has significantly higher sensitivity in detecting the causal genes compared to Hotnet2 in both thresholding and ranking regimes in 88% of scenarios . Fig 1D–1F show that Conflux has a consistently lower rate of Type 1 errors than Hotnet2 . We attribute these observations to the fact that Hotnet2 and many other diffusion methods focus on identifying sets of genes that form significant subnetworks together , not paying as much attention to the false positives that might also be part of that subnetwork . Conflux has two advantages over those methods: i ) it tries to maximize the number of patients explained while minimizing the number of selected genes; ii ) it keeps track of the uncertainty of each of the genes thus the gene might not be selected even if it is well connected within the subnetwork . Across all choices of sample sizes and number of causal genes , Conflux top ranked genes is the best performing approach in terms of power ( sensitivity ) and F-Measure , followed closely by Conflux with thresholded marginals ( ≥ 0 . 2 ) . Both significantly outperform Hotnet2 in all scenarios in terms of F-Measure . This is mostly due to Hotnet2 reporting more false positives . Analyzing the performance of SKAT-O’s top ranking genes , we can see that SKAT-O ranks causal genes high even if they do not reach significance , evident from the comparison of SKAT-O thresholded approach ( in red ) with SKAT-O top ranked genes ( in olive ) . Interestingly , as the complexity of the disease grows ( 20 and 50 genes ) , the sensitivity of Hotnet2 is the same or even lower than SKAT-O top ranked genes . At the same time , the precision of Hotnet2 is much higher than that of SKAT-O top-ranked genes . We conclude that Hotnet2 does a great job at selecting the true causal genes among SKAT-O top ranked and significant genes ( taken as input ) , which is helped by the knowledge encoded in the network structure . Top ranking versions of both SKAT-O and Conflux have higher F-Measure and sensitivity compared to the thresholding versions of the same methods across all scenarios . This means that i ) significance thresholding keeps both methods on the conservative side and ii ) it is beneficial to examine gene rankings and not just fixed gene sets . Both versions of Conflux perform better than the best ( gene ranked ) version of SKAT-O indicating that independently of the complexity of the disease and the number of available samples , Conflux has higher accuracy in prioritizing causal genes than the most widely used non-network gene prioritization method . As expected , we see that the problem of identifying the causal genes is more difficult for more complex diseases , i . e . the performance of all methods goes down for 20 and 50 causal diseases ( second and third columns of Fig 1 ) . The power/sensitivity of all methods and variations increased with larger sample sizes . Hotnet2’s precision increased from an average of 50% at n = 200 to an average of 90% at n = 800 . It is the only non-ranking method exhibiting this behaviour . This can be explained by the fact that it is constructing a subnetwork from SKAT-O top-ranked genes ( SKAT-O’s precision is affected by sample size ) and maximizes the significance of the whole subnetwork rather than its constituent genes independently . To summarize , our results on randomly simulated causal disease genes show that both network-based methods , Conflux and Hotnet2 , outperform SKAT-O , which does not use networks , in all of our simulations , indicating the importance of using networks in identifying interacting disease associated genes . Additionally , Conflux outperforms Hotnet2 according to the F-measure in all scenarios , which signifies the importance of using the full data in combination with the network rather than diffusing gene summary statistics . Local network topology plays an important role in how the information gets diffused across the network . When propagating messages/heat between genes , Conflux limits the incoming signal received by a gene from its neighbours , while Hotnet2 limits the heat coming out of a particular gene . This means that both methods penalize highly connected nodes but in different ways . We thus decided to investigate specific configurations of local network topologies and assess method performance in each of these scenarios . We examined three patterns: star , clique and chain and selected causal gene sets accordingly . We applied our method to the whole network , i . e . all genes , preserving the patterns of interest as significant , and examined genes identified by each of the methods . In all the simulations above , the disease causing genes were selected based on explicit assumptions such as taking them from local neighbourhoods in section on simulated complex diseases or local network configurations above . In this section , we avoid making such assumptions by taking real disease-associated gene sets reported in the literature . The exome data is still simulated the same way as in the previous sections . We assessed the performance of network-based methods on several gene sets associated with autism spectrum disorder ( ASD ) , epilepsy , schizophrenia [8] and ovarian cancer [30] in the literature . Each of these gene sets are subnetworks that can be found in iRefIndex network and were not originally found by Hotnet or Hotnet2 to be fair to all methods considered . We ran our experiments simulating a set of 400 patients and 400 controls . The results are summarized in Fig 5 .
It is becoming increasingly clear that networks are a powerful tool in performing gene-disease associations , improving the power of gene discovery . There are many ways in which networks could be incorporated into the analysis . The existing diffusion methods are effective , yet are subject to a few limitations . Our proposed Bayesian framework , Conflux , addresses limitations such as dealing with network noise and combining networks directly with the variant information rather than relying on precomputed summary statistics . It also proposes additional advantages , such as providing the ability to rank genes using marginal probability estimates . As researchers strive to integrate more information into a joint model of genomic data in an attempt to discover novel gene-disease associations , our powerful Bayesian framework is bringing flexibility and power to the modeling paradigm .
Conflux jointly models the outcome ( disease/no disease ) , the genotypes and the PPI network in a hierarchical graphical model . The graphical model is designed to fit the hierarchical structure of our problem by describing all variants in the data , all genes and all individuals ( cases/controls ) . The hierarchical structure aggregates variants into gene variables and gene variables into the phenotype . Fig 6 shows a factor graph corresponding to our model . We use variables specific to each individual and indicator variables that are shared across individuals . The variables X , D and Y correspond to the variants . The variables H , Q and G correspond to the genes . X and H are indicator variables for the variants and genes respectively , while D and Q are patient specific variables for the variants and genes respectively . Y and G are intermediate variables that are also patient specific variables for variants and genes , they represent the product of D and Q with the corresponding indicator variables ( X and H respectively ) . Table 1 describes all the listed variables . The PPI network is encoded as factors ( priors ) over the H variables ( Θ ) . For every interaction in the PPI network , we add a factor linking the pair of genes involved . Additionally , a regularization factor ρ is added over H , to encourage a set of disease-associated genes to be sparse . This factor is implemented in the form of cardinality potentials [31] . The likelihood of the model can be written as: P ( X , G , H , Q , p h e n o | D ) ∝ P ( p h e n o | G ) P ( G | H , Q ) P ( Q | X , D e x o m e ) P ( X ) P ( H ) ( 1 ) where the priors are defined as follows: P ( X ) are non-informative ( set to a constant value 0 . 5 ) , P ( H ) = θ ⋅ τ ⋅ ρ . There are two types of priors used for the H variables: The conditional probabilities are defined on the graph as the φ factors . The probability distribution of Q depends on the state of the intermediary variables Y . The relationship is described by the factor φ2 ( Q , Y ) = P ( Q | Y ) = P ( Q|X , D ) . For more details , see Section 2 . 1 of S1 Text . For a particular gene and individual , the G variable depends on Q and H as described by the factor: φ 3 ( G , H , Q ) = 1 [ G = H · Q ] . This factor encodes that a gene can only be relevant in a particular individual if the gene is shown to be relevant to the phenotype in general and if that particular individual has potentially harmful exome variants in that gene . Finally , the phenotype ( disease/no disease ) is given as an input . It relates to the number of active G variables in each patient: a patient is likely to have some affected genes while a healthy individual should have few or no affected genes . The expected number of affected genes for patients and the tolerated number of affected genes in healthy population depend on the disease and its complexity and are therefore taken as parameters in the function φ4 ( pheno , G ) = P ( pheno|G ) by our method . ( More on φ4 in Section 2 . 2 in S1 Text ) . For inference , we use loopy belief propagation , modified for efficiency ( see below ) , to jointly infer the marginal distributions of the unobserved variables in our graphical model . After convergence of the loopy belief propagation our method returns estimates of the posterior marginal probabilities for the H and G variables , along with the ranked list of most relevant aberrations ( coding variants ) in each affected individual according to the model . All computed messages are normalized and kept in log space for numeric stability . We used message damping with parameter α = 0 . 5 to improve the convergence behaviour of the algorithm . While there is no theoretical guarantee of convergence , we find that in practice dampening of the messages leads to convergence in every case we have considered . Variables related to the aberrations in each gene and patient are the most numerous , and their messages are the most expensive to compute and the slowest to vary . Conversely , the messages between the phenotype , the G and H variables are the fastest to change since these variables are tightly correlated . Therefore , every time we update the messages for X , Y and Q variables , we update the messages between G , H and phenotype variables up to 10 times or until their local convergence . Some of the factors connect a large number of nodes to one node: the φ4 factor connecting all G variables to phenotype , the regularization factor ρ connecting the H variables together , and the φ2 factor connecting Y nodes to a Q node . Since each of these factors mainly depends on the sum of its input variables , the computation becomes feasible by using the approach in Tarlow et al [31] , in which a binary tree structure is used , with the internal nodes representing the intermediary sums of variables , and then belief messages are computed across the tree . This divide & conquer approach reduces the complexity of messages computation to a O ( nlog2 ( n ) ) where n is the number of initial variables considered . The PPI network is used as prior over the H variables indicating whether a gene is relevant to the disease . It is encoded by the Θ factors , one for every pair of interacting genes . Θ encourage pairs of interacting genes to be active together for the same disease . There are difficulties associated with the implementation of this prior: To avoid having our network prior biasing Conflux results , we take the following measures . First , instead of the symmetric indirected factors usually used for modeling joint probabilities , we make the Θ factors into directed and asymmetric factors . The idea is that we want to encourage a gene to be ‘on’ if its neighbour has some evidence of being associated with the disease , but we do not want to encourage a gene to be ‘off’ because its neighbours are off . Consequently , if the neighbor is off it should be non-informative of the state of the considered gene . The directed factor between a gene i and its neighbour j is therefore: P ( Hi = 1|Hj = 0 ) = 0 . 5 , P ( Hi = 0|Hj = 0 ) = 0 . 5 , P ( Hi = 1|Hj = 1 ) = δ , P ( Hi = 0|Hj = 1 ) = 1 − δ . This is very similar to the heat diffusion in Hotnet2 which is also asymmetric , and only positive . Second , we parametrize the directed factor by the degree of the target gene . The goal is that the accumulation of signals from a large number of neighbours should never be enough to make a highly connected gene seem disease associated . So the parameter δ from the factor definition is a function of the degree of gene i in the network . δ is chosen to limit the amount of influence this gene can receive from any one of its neighbours . In Hotnet2 , the weights in the influence matrix also depend on the nodes degrees , but they depend on the degree of the source node ( the influencer ) while our δ depends on the degree of the receiving node . This approach allowed us to reduce the number of false positives across many of the presented examples . To compute δ for a particular degree d , we first compute the maximal contribution of the network to the posterior . In Conflux , every gene starts from a small prior τ and needs evidence from its own data ( coding variants ) to reach large enough marginals to be considered a contributor to the disease in question . The maximal possible contribution Mc is the contribution that would make the gene’s marginal reach 0 . 02 ( starting from the initial prior τ and supposing the gene have no signal ) . Limiting ourselves to this maximal contribution from the network guarantees that the network context on its own will never associate the gene with the disease ( marginal of at least 0 . 2 needed ) on its own . Messages between binary probabilistic variables are represented by the ratio of the probability of being active and the probability of being inactive . The maximal contribution Mc is given by: τ 1 - τ · M c 1 - M c = 0 . 02 1 - 0 . 02 ( 2 ) Once we define the maximal contribution , we compute the value of δ insuring that the sum of messages coming from a neighbourhood that is significantly enriched in active genes will be equal to the maximal contribution . Any neighbourhood that is less than significant would make less than the maximal contribution . A neighbourhood significance is assessed by a binomial with the probability of success in a single trial is equal to the prior τ and the number of trials is equal to the degree d . The significance threshold used is 0 . 05 divided by the total number of genes . For example , if the prior τ is 0 . 0025 and the degree d is 10 , significance is obtained if there are amin = 3 active neighbours and δ is chosen so that the sum of messages ( in log space ) coming from 3 active nodes ( parametrized by delta ) would be equal to the maximal contribution . In non-log space , the product by amin becomes a power , this gives: ( δ 1 - δ ) a m i n = M c 1 - M c ( 3 ) For the same prior τ = 0 . 0025 but with a higher degree d = 100 , significance is only obtained if there are amin = 6 active neighbours and δ is chosen so that the sum of messages coming from 6 active nodes would be equal to the maximal contribution . Therefore , higher degree nodes require more active neighbour genes to reach significance so they will have a lower δ . Additionally , we ensure that delta is never larger than 0 . 9 to limit the influence of one gene on another even if they are the sole neighbour . Note that the examples above are computed for a specific set of parameters and are given merely as illustration of the method . Up to this point our computations are theoretical , i . e . before the inference process and without looking at any real messages . It is possible that during the inference we observe more active neighbours than the minimal significant number amin we utilized to compute δ which would generate higher total contributions than the maximal contribution . Therefore , we also cap the sum of contributions to the maximal contribution during the inference process . To simulate a disease , we selected P genes [P is 10 , 20 or 50] that are close to each other on the protein-protein interaction network as being the causal disease mechanism . To do so , we randomly selected a seed gene in a iRefIndex Network [29] , then randomly picked other causal genes from first and second degree neighbourhoods . The first and second degree neighbourhoods around the seed gene must jointly have an acceptable size , i . e larger than the number of causal genes being simulated . Although these simulation assumptions ( causal genes being in relatively close proximity in the network ) may not always reflect the ( unknown ) reality of how real disease genes are distributed in PPI networks , they still provide a realistic enough framework to assess scenarios where PPI networks should be able to help to identify a disease mechanism . Note that in addition to these simulations we also used other configurations of causal genes ( star , clique and chain ) and literature reported disease subnetworks ( see the Results ) . For simulating coding variants , we used the European population model with the optimal parameters as in [28] . We considered the single nucleotide variants ( SNVs ) with selection coefficient ( indicating impact on the fitness ) greater than s = 0 . 001 as the deleterious SNVs and the remaining SNVs as neutral , as recommended in [32] . The generated sequence including the random missense SNVs was randomly split into blocks , each block corresponding to one of the genes in the iRefIndex network . Using this approach we generate coding variants ( the vast majority of which being rare variants ) in ≈ 12 , 000 genes for 900 , 000 individuals . For every individual in the simulated population , we count the number of harmful mutations within the disease mechanism and we select the patients as the top 3 , 000 individuals according to that sum . The rest of the population is considered healthy . We thus generated our case/control setting with sample size n for analyzing the disease by selecting n 2 cases from the patient subpopulation and n 2 controls from the healthy subpopulation . SKAT-O is the most widely used statistical test for the association of rare variants with a phenotype [26] . It combines a burden test with a variance component test ( SKAT [33] ) and has been proven to do well in terms of power and type 1 error . SKAT-O is resilient to the presence of neutral and protective variants . In our simulations , all the deleterious variants with high selection coefficient were rare variants . Therefore a gene-based test for aggregating rare variants such as SKAT-O is the appropriate baseline . Hotnet2 [18] takes gene-based statistics as an input . Though originally developed and used for cancer data , Hotnet2 was previously used in a GWAS setting identifying genes associated with complex human diseases through common variants [19] . In our example for complex human diseases due to rare rather than common variants , we used SKAT-O’s p-values ( after negative log transformation ) as input to Hotnet2 . As recommended in [19] we only selected the top scoring genes to have non-zero scores because Hotnet2 does not do well when there are many genes with low heat scores . We tested multiple ways of thresholding the p-values to improve the performance of Hotnet2 . First , we attempted to use FDR q-values and only selected those genes with FDR<0 . 05 . This created two problems: i ) Hotnet2 does not run with a limited number of genes ( less than 20 ) satisfying this FDR threshold; ii ) the performance of Hotnet2 dropped considerably in terms of sensitivity and became closer to that of thresholded SKAT-O . This is due to the fact that SKAT-O is underpowered and restricting the analysis to only those genes detected by SKAT-O limits the potential of Hotnet2 . We also attempted to use the local FDR curves [34] and look for inflection points as suggested in [19] . But in most examples this approach gave much worse performance for Hotnet2 and was very unstable from one simulation to the next: Hotnet2 was either too conservative ( low sensitivity ) or too imprecise ( precision under 0 . 1 ) . We chose to select the genes with p-values below a relaxed FDR criterion ( <0 . 5 ) . If the number of genes passing that criterion was less than 20 , which was the case for most of our experiments , we augmented it by taking genes with the smallest 20 p-values to ensure Hotnet2 can run . We also attempted taking the top 50 and top 100 genes , but that had a negative effect on Hotnet2 performance where the precision dropped to below 50% . Hotnet2 returns four sets of results corresponding to four different estimations of the method’s hyper-parameter δ . Given that there is no way to select the right set of results we always reported the best performing set returned by Hotnet2 . This might give an unfair advantage to Hotnet2 , but we did not find a more principled way of automatically selecting the right delta hyper-parameter for Hotnet2 . We extracted disease associated subnetwork from two publications . Hormozdiari [8] describes a combinatorial optimization method MAGI for building disease associated modules . From their supplementary data we selected 4 modules: Schizophrenia_M2 , Epilepsy_M1 , ASD_ID_M2 ( referred to as ASD1 in this paper ) and ASD_with_ID_M2 ( referred to as ASD2 ) . [30] construct informative gene subnetworks by integrating cancer gene expression profile with a PPI network and is applied on Ovarian cancer . We selected the module they report in their paper . | Networks and pathway-based methods are commonly used to improve the power of gene detection in associations with complex human diseases . Network diffusion approaches have shown their effectiveness and superior performance in cancer studies . Still , there are many problems such as noise and missingness with currently available human networks that bias the results of gene detection . We propose a novel graphical model-based method Conflux that overcomes several of the pitfalls of the existing state-of-the-art approaches while building on their successes . Conflux integrates genotype data with networks directly , using diffusion-like methods , but only as part of a structure in a probabilistic model to reduce the negative effect of the noise in the networks . This Bayesian framework allows Conflux to keep track of the uncertainty in the gene list that is being associated with the disease and consequently rank the genes with respect to our confidence in the association . It also allows for the discovery of gene sets that are not fully supported by the network if they have enough support in the data . These improvements result in a flexible approach that improves the power in many gene-disease association scenarios while reducing the number of false positives reported . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"genetic",
"networks",
"protein",
"interactions",
"protein",
"interaction",
"networks",
"cancers",
"and",
"neoplasms",
"oncology",
"mutation",
"network",
"analysis",
"genome",
"analysis",
"epilepsy",
"computer",
"and",
"informati... | 2017 | Incorporating networks in a probabilistic graphical model to find drivers for complex human diseases |
Alternative splicing is an evolutionary innovation to create functionally diverse proteins from a limited number of genes . SNAP-25 plays a central role in neuroexocytosis by bridging synaptic vesicles to the plasma membrane during regulated exocytosis . The SNAP-25 polypeptide is encoded by a single copy gene , but in higher vertebrates a duplication of exon 5 has resulted in two mutually exclusive splice variants , SNAP-25a and SNAP-25b . To address a potential physiological difference between the two SNAP-25 proteins , we generated gene targeted SNAP-25b deficient mouse mutants by replacing the SNAP-25b specific exon with a second SNAP-25a equivalent . Elimination of SNAP-25b expression resulted in developmental defects , spontaneous seizures , and impaired short-term synaptic plasticity . In adult mutants , morphological changes in hippocampus and drastically altered neuropeptide expression were accompanied by severe impairment of spatial learning . We conclude that the ancient exon duplication in the Snap25 gene provides additional SNAP-25-function required for complex neuronal processes in higher eukaryotes .
The evolution of more advanced organisms has required adaptation of genomes to be able to generate new gene functions . The original genome sequence analyses of different organisms surprisingly revealed that the number of genes in the human genome is only around 30 , 000 , versus 20 , 000 in much simpler organisms such as the nematode Caenorhabditis elegans [1] , [2] . However , a closer examination of these 30 , 000 identified human genes suggested that as many as one third of them might be false and the number of protein-coding genes in humans are close to the 19 , 000 found in the domestic dog [3] , [4] . Instead the increased protein complexity of higher eukaryotes appears to be the consequence of the same gene encoding several functional proteins . This has been accomplished by duplication of genes or gene segments and transcriptional and post-transcriptional regulation . The major contribution to protein diversity is alternative splicing and some 40–60% of all mammalian genes generate more than one protein [5] , [6] . Interestingly , as many as 10% of all genes in mammals contain tandemly duplicated exons , suggesting that exon duplication followed by functionally diverging mutations have been a fast and successful evolutionary mechanism to increase protein variety [7] , 8 . Duplicated exons are often subjected to mutually exclusive alternative splicing , incorporating only one of the two exons in the resulting polypeptide [8] . Regulated membrane fusion forms the basis for synaptic transmission but is also fundamental for appropriate release of hormones and modulatory neuropeptides [9]–[13] . One of the final steps prior to vesicle fusion with the plasma membrane is the formation of a trans-membrane soluble N-ethylmaleimide-sensitive factor ( NSF ) attachment protein receptor ( SNARE ) complex [14] , [15] . This specialized SNARE complex is bridging vesicles to the plasma membrane during regulated exocytosis and consists of three compartmentally defined proteins: The vesicle-associated VAMP2/synaptobrevin , a protein with a plasma membrane anchor syntaxin 1a , and the cytoplasmic synaptosomal-associated protein of 25 kD , SNAP-25 , that associates with membranes through palmitoylation . The three SNARE proteins are held together by strong protein-protein interactions , whereby the cytoplasmic domains form a four α-helix coiled-coiled bundle [16] . The detailed mechanisms mediating regulated exocytosis are still not fully elucidated but the current hypothesis is that SNARE proteins operate at the actual fusion event and have intrinsic capabilities to perform membrane fusion [17] , [18] . A possibility is that the SNARE proteins are candidates for adjusting thresholds for synaptic plasticity in more advanced neuronal systems . They form the central fusogenic core at the plasma membrane . It is notable that the number of SNARE proteins , and alternative isoform variants , have increased through evolution and their expression is strictly regulated both anatomically and temporally [19]–[23] . In higher vertebrates SNAP-25 is expressed as two developmentally regulated and complementary distributed splice variants termed SNAP-25a and SNAP-25b [23] . The alternative splicing is an obligate choice between two closely spaced tandemly arranged exon 5 sequences , and nine of 206 amino acids in the two polypeptides differ . The alternative splicing modifies a domain of the SNAP-25 protein spanning a quartet of cysteine residues that are substrates for post-translational palmitoylation and required for membrane targeting [24] , [25] . In mouse brain , SNAP-25a precedes SNAP-25b expression during development , but by the second postnatal ( PN ) week SNAP-25b becomes the major splice variant , concomitantly with a dramatic increase in SNAP-25 expression [26] . In fact , in adult mouse brain the SNAP-25b transcript represents more than 90% of total SNAP-25 mRNA [26] . Targeted disruption of the mouse Snap25 gene has demonstrated that complete removal of SNAP-25 results in total absence of evoked neuroexocytosis and embryonic lethality [27] . Separate overexpression of the two SNAP-25 isoforms in embryonic adrenal chromaffin cells from these SNAP-25 knock-out ( KO ) mice showed that the SNAP-25b isoform had a higher capability to stabilize the pool of primed vesicles than SNAP-25a , since the burst of Ca2+-evoked catecholamine release differed [28] . Recently , genome-wide scans and linkage analysis have indicated an association between polymorphisms in the human SNAP-25 gene and vulnerability to develop attention deficit hyperactivity disorder , ADHD [29] , [30] . In humans , different SNAP-25 alleles also demonstrate inheritance correlated to intelligence [31] , [32] . To specifically explore the physiological importance of the exon 5 duplication in the Snap25 gene we used a novel approach by developing SNAP-25b KO/SNAP-25a knock-in mouse mutants . The exon 5b was genetically eliminated and replaced with a second supplementary exon 5a , to maintain alternative splicing and normal expression levels , but allowing only SNAP-25a to be expressed . The SNAP-25b deficient mouse mutants exist in two versions , neo-containing with a Tkneo marker retained and neo-excised with the selection gene removed . Based on results from electrophysiological , behavioral and immunohistochemical experiments we conclude , that under physiological conditions , mice deficient in SNAP-25b have developmental defects , impaired short-term synaptic plasticity , a seizure-prone phenotype and malfunctioning cognitive performance . We suggest that the ancient duplication in the Snap25 gene followed by regulated alternative splicing between two similar but distinct exon 5 sequences is required for accurate synaptic function during development . Furthermore , a balanced expression of the two isoforms is a prerequisite for maintaining an operational neuronal network also during adulthood in advanced organisms .
In the Snap25 gene , exon 5a and 5b are closely spaced and differ in only nine of 39 amino acids [23] , [26] . To develop mouse mutants that only express SNAP-25a , a gene targeting vector with an additional exon 5a sequence , replacing the mouse exon 5b sequence , was generated ( Figure 1A ) . The vector construct spanning two exon 5a sequences arranged in tandem and exon 6 from the mouse Snap25 gene also contained a Tkneo selection cassette flanked by loxP recombination sites ( Figure 1A ) . Only the exon 5b sequence was changed to encode the exon 5a amino acids , and the original splicing signals for expression of the downstream exon 5 were kept intact . Three independent mouse lines were backcrossed on C57BL/6NCrl ( B6 ) mice for at least ten generations , thus establishing fully congenic strains . After intercrossing of heterozygous animals we found a mendelian distribution of genotypes ( 24% homozygous mutants of 125 mice ) , strongly indicating that our introduced genetic changes did not give rise to a prenatal lethal phenotype . However , after the second PN week all homozygous SNAP-25b deficient mice exhibited neurological defects and were sacrificed . Thus , we have developed a SNAP-25b deficient mouse mutant by replacing exon 5b with an additional exon 5a equivalent , thereby preventing expression of SNAP-25b but not the alternative splicing . We previously demonstrated that targeted insertion of a Tkneo selection cassette in the Snap25 gene impaired alternative splicing and repressed total gene expression [33] . Initially , we therefore investigated the level of SNAP-25 mRNA and protein expression in brain of neo-containing SNAP-25b deficient mutants and wild-type ( WT ) littermates at PN14 . Semi-quantitative RT-PCR analysis demonstrated that neo-containing SNAP-25b deficient mice expressed approximately 50% of the SNAP-25 mRNA levels present in WT littermates ( Figure 1B ) . In order to determine the relative ratio of SNAP-25a and SNAP-25b mRNA expression in brain at PN14 , isolated RNA from homozygous and heterozygous SNAP-25b deficient mutants and WT littermates was subjected to an RT-PCR assay based on the presence of exclusive restriction enzyme sites in exons 5a and 5b . At PN14 , SNAP-25b levels were five times higher than SNAP-25a in WT mice , heterozygous neo-containing mutants had equal amounts of both SNAP-25 mRNA isoforms , while homozygous neo-containing SNAP-25b deficient mutants only expressed SNAP-25a ( Figure 1C ) . Western blotting of protein homogenates from PN14-15 homozygous neo-containing SNAP-25b deficient mouse brains revealed that SNAP-25 protein levels were also lower in mutants when compared to WT littermates ( Figure 1D ) . SNAP-25b deficient mice expressed approximately 80% of normal SNAP-25 protein levels . Thus , there was no direct correlation between SNAP-25 mRNA and protein levels ( compare Figures 1B and 1D ) . The protein levels of the cellular SNAP-25 homolog , SNAP-23 , and the binding partner to SNAP-25 , syntaxin 1 , were not significantly different in mutants compared to control animals ( Figure 1D ) . Immunohistochemical analysis of SNAP-25 showed a similar appearance in homozygous neo-containing SNAP-25b deficient mutant and WT mouse brain at PN14 , and no obvious pathological changes were observed ( Figure 1E ) . Together , our results demonstrate that homozygous neo-containing SNAP-25b deficient mouse mutants only express SNAP-25a and that total SNAP-25 mRNA and protein levels are reduced . Unexpectedly , despite that neo-containing SNAP-25b deficient mutants express only 50% of normal SNAP-25 mRNA levels , protein levels of SNAP-25 are only reduced to 80% of WT littermates . Neo-containing SNAP-25b deficient mutants exhibited a severe phenotype that included a reduced gain in body weight after the first PN week ( Figure 1F ) . The growth-deficiency was not due to an inability of mutants to feed properly , as dissection revealed stomachs filled with milk ( data not shown ) . To further investigate developmental defects , bone growth was analyzed in PN14 mutants and WT littermates . Bone development was determined by measuring the hypertrophic , proliferative and reserve zones in hind limb sections ( Figure 1G ) . In both the femur and tibia , the hypertrophic zones were significantly reduced in homozygous neo-containing SNAP-25b deficient mutants when compared to WT littermates . The proliferative and reserve zones were not significantly different ( Figure 1G ) . Around PN10 homozygous SNAP-25b deficient mice were easily identified by their smaller size and extreme activity with hyperactive episodes . From PN11 , neo-containing SNAP-25b deficient mutants exhibited frequent episodes with tremors and seizure activity and were therefore sacrificed at PN14-17 . The early PN development also appeared postponed , indicated by that eye opening occurred approximately one day later compared to WT littermates and that the SNAP-25b deficient mutants also demonstrated inability to respond to sound concurrently in time when WT littermates acquired that ability . Heterozygous neo-containing SNAP-25b mutants were indistinguishable from WT littermates during early PN development . Our results show , that homozygous neo-containing SNAP-25b deficient mice , lacking SNAP-25b expression and having a moderate reduction in total levels of SNAP-25 protein , exhibit severe developmental and behavioral defects . The reduced SNAP-25 expression in neo-containing SNAP-25b deficient mutants could be due to the presence of the Tkneo selection gene . Therefore , neo-containing heterozygous SNAP-25b deficient mutants were crossbred with a global Cre transgene [34] . The Tkneo gene was excised and the Cre transgene was thereafter crossed out from the neo-excised SNAP-25b deficient mutants ( Figure 2A ) . Neo-excised SNAP-25b mutants were backcrossed onto a B6 background for more than ten generations . Intercrosses using heterozygous breeding pairs indicated the expected mendelian distribution ( 28% homozygous mutants born out of 222 mice ) . Unlike neo-containing homozygous SNAP-25b deficient mutants , the growth of neo-excised homozygous mutants between PN5 and 15 was not severely affected . No difference in body weight was observed between mutants and WT littermates when genders were mixed . However , when females were analyzed separately they demonstrated a small but significant reduction in body weight ( Text S1 , Figure S1A and S1B ) . After the first PN weeks the body weights were normalized and no significant differences were observed in adult females ( data not shown ) . Neo-excised SNAP-25b deficient mutants also demonstrated spontaneously occurring seizures although less frequent and not prior to young adulthood . In conclusion , in vivo excising of Tkneo from the targeted Snap25 gene rescues the most severe developmental defects observed in homozygous neo-containing SNAP-25b deficient mice . Contrary to homozygous neo-containing animals , neo-excised SNAP-25b deficient mutants showed SNAP-25 mRNA levels similar to WT littermates at PN14-15 ( Figure 2B ) . SNAP-25b mRNA in WT mice was , as expected , five times higher than SNAP-25a ( Figure 2C ) . In heterozygous neo-excised mutants the relative SNAP-25a/SNAP-25b mRNA ratio was 2∶1 , while SNAP-25b was absent in homozygous neo-excised mutants . In neo-excised mutants SNAP-25 protein expression was moderately but significantly increased to 111% compared to WT ( Figure 2D ) . SNAP-23 and syntaxin 1 protein levels were not altered from levels observed in control littermates . Our results demonstrate that removal of Tkneo restores SNAP-25 expression and moderately increases total SNAP-25 protein levels in brain of homozygous neo-excised SNAP-25b deficient mutants . In adult brain of WT mice , both splice variants of SNAP-25 are expressed but SNAP-25b is the predominant isoform comprising 90–95% of total SNAP-25 mRNA [26] . To investigate if the stability of SNARE complexes was altered in the brain of our neo-excised SNAP-25b deficient mutants compared to WT mice , we analyzed SDS-resistant SNARE complexes at different temperatures ( Figure 2E ) . Adult brain tissue homogenates only containing SNAP-25a ( neo-excised SNAP-25b deficient mutants , KO ) or predominantly SNAP-25b ( WT ) were incubated at different temperatures , either 4°C or heated , prior to gel electrophoresis and Western blotting . In non-heated samples , several immunoreactive bands were observed in SNAP-25b deficient and in WT brain homogenates ( Figure 2E ) . Except for a SDS resistant complex observed at ∼97 kD , identified as the ternary SNARE complex , additional intermediate complexes were detected . At 4°C and stepwise increasing temperatures ( 70°C–90°C ) the ratio of the quantified protein band migrating at ∼28 kDa ( free SNAP-25 ) to higher molecular weight SNAP-25 containing complexes indicated that the stability of SNARE complexes was reduced in SNAP-25b deficient mouse brain . Quantification of immunoreactive bands from WT and neo-excised SNAP-25b deficient mutants mice showed that the percentage of total SNAP-25 still present in the ternary complex differed significantly at 80°C , 85°C and 90°C [at 80°C , 71 . 2±7 . 2% and 44 . 0±8 . 0% ( *p<0 . 05 ) ; at 85°C 59 . 8±7 and 36 . 9±3 . 4 ( *p<0 . 05 ) , and at 90°C 48 . 3±12 . 3 and 4 . 5±2 . 2 ( **p<0 . 01 ) ] . The ternary complex almost disappeared in SNAP-25b deficient brain preparations treated at 90°C while the WT samples still showed a strong immunoreactive signal containing 48 . 3% of total SNAP-25 in non-disassociated complex ( Figure 2E ) . Boiling of the samples resulted in the loss of all immunoreactive bands except the monomeric SNAP-25 in both neo-excised SNAP-25b deficient and WT mice . Our results demonstrate that SNARE complexes isolated from brain of adult neo-excised SNAP-25b deficient mutants are less stable than SNARE complexes from adult WT mice . SNAP-25 is essential for evoked synaptic transmission [27] and its activation is dependent on cytoplasmic free Ca2+-concentrations [35] . The different stability of SNARE complexes found in neo-excised SNAP-25b deficient mutants compared to WT mice ( Figure 2E ) and that SNAP-25b demonstrates an increased association with plasma membrane fractions compared to SNAP-25a ( see Text S1 , Figure S2 ) suggest that SNAP-25a and SNAP-25b are differently associated with SNARE complexes close to , or immediately upstream of fusion . Therefore , we compared paired-pulse facilitation ( PPF , Figure 3E ) of AMPA receptor-mediated synaptic transmission in young WT , neo-containing and neo-excised SNAP-25b deficient mice ( KO ) at the Schaffer collateral-CA1 pyramidal neuron synapses of the hippocampus [36] , [37] . Two stimulus frequencies were used , 0 . 2 and 0 . 5 Hz and the interpulse interval ( IPI ) varied between 40 and 300 ms . Neo-containing SNAP-25b deficient mouse mutants ( n = 8 cells in 6 mice ) showed a reduction ( *p<0 . 05 ) in PPF at 0 . 2 Hz compared to WT mice ( n = 9 cells in 5 mice , Figure 3A ) . A decrease was also observed at 0 . 5 Hz for these mice ( *p <0 . 05 , n = 7 cells in 7 mice ) compared to WT littermates ( n = 9 cells in 7 mice , Figure 3C ) . Furthermore , neo-excised SNAP-25b deficient mouse mutants demonstrated a clear reduction in PPF at 0 . 2 Hz ( *p <0 . 05 , SNAP-25b KO: n = 9 in 4 mice; WT: n = 9 cells in 5 mice , Figure 3B ) but not at 0 . 5 Hz ( p = 0 . 085; n . s . , SNAP-25b KO: n = 9 cells in 6 mice; WT: n = 9 cells in 7 mice , Figure 3D ) compared to WT mice . Finally , we compared the PPF-ratios obtained from homozygous neo-containing and neo-excised SNAP-25b deficient animals . No significant differences were found at either 0 . 2 ( p = 0 . 92 , n . s . ) or 0 . 5 Hz ( p = 0 . 48 , n . s . ) . All recorded neurons were responsive to the presynaptic stimulation and there was no obvious rundown or difference in baseline responses , in contrast to what is seen in complete SNAP-25 KO mice [38] . Our results demonstrate that absence of SNAP-25b reduces PPF at Schaffer collateral-CA1 synapses in PN12-16 mice during low-frequency stimulation . In view of the above findings , we investigated the potential role of SNAP-25b in behavioral functions partly related to changes in long-term plasticity in the hippocampus and amygdala . The mice were examined in the elevated plus maze , a behavioral test primarily designed for analyzing anxiety-related behavior . Homozygous neo-excised SNAP-25b deficient mutants spent less time in the open arms compared to the corresponding control group ( ***p<0 . 001 , Figure 4A ) and more time in the closed arms during the 5 min observation period ( ***p<0 . 001 , Figure 4C ) . The total number of arm entries did not differ between mutants and control animals , which indicates that there were no differences in overall motor activity between the two groups ( Figures 4B and 4D ) . Acquisition and retention of spatial learning , which involve hippocampal mechanisms , were studied in the Morris water maze task . In the pre-training phase , there was no overall significant effect of genotype ( p = 0 . 09 , n . s . ) in the latency to navigate to the visible platform compared to the corresponding control group ( Figure 4E ) . However , a subsequent post-hoc analysis showed that homozygous neo-excised SNAP-25b female mice had a significant higher latency compared to the control mice on day 2 and on day 4 ( p<0 . 05 ) . Spatial acquisition was examined during five days of training and revealed highly significant differences between the groups with regard to escape latency ( p<0 . 001 , Figure 4F ) , indicating that neo-excised SNAP-25b deficient mice were impaired in their ability to acquire the spatial learning task compared to WT controls . In addition , neo-excised SNAP-25b deficient mutants had a longer swim distance compared to controls ( p<0 . 001 , Figure 4H ) . There were no overall differences with regard to swim speed between neo-excised SNAP-25b deficient mutants and the control groups ( Figure 4G ) . The analysis of percent swimming along the wall in the water tank , e . g . thigmotaxis revealed that both neo-excised SNAP-25b deficient males and females displayed a much higher thigmotactic behavior than the control animals ( p<0 . 001 ) . Young neo-containing SNAP-25b deficient mutants demonstrated periods with profound hyperactivity in their home cages , a behavior not observed in neo-excised SNAP-25b deficient mutants . Locomotor activity in adult neo-excised SNAP-25b deficient mutants was investigated in computerized locomotor cages . Analyses of spontaneous locomotor activity during the 60 min recording revealed that for all measures , e . g . motility , locomotion and rearing , the neo-excised SNAP-25 deficient mutants had lower ( p<0 . 01 ) locomotor activity than WT controls . Female mutants demonstrated a low activity level already during the first 10 min of recording , i . e . during the initial exploration of the locomotor cage ( data not shown ) . In summary , neo-excised SNAP-25b deficient mice , irrespective of gender demonstrate a higher anxiety index , a severe impairment in spatial learning with females being most severely affected , and a lower locomotor activity when compared to WT controls . In young neo-containing and neo-excised SNAP-25b deficient mouse mutants no obvious difference in structural and morphological appearance of the brain was observed by histological and immunohistochemical analysis ( Figure 1E , and data not shown ) . However , in the stratum lucidum ( SLu ) of the CA3 region of 4-month-old neo-excised SNAP-25b deficient mutants , SNAP-25-positive ( + ) mossy fibers had expanded and formed large bundles ( compare Figure 5E and 5F ) , separating the synaptophysin+ nerve endings into island-like structures ( Figure 5G and 5H ) . These morphological differences observed with immunohistochemistry were also evident in neo-excised SNAP-25b deficient mouse mutants at 2 months of age and appeared to progress with age . The CA1 and cerebellum were unaffected at all ages analyzed ( data not shown ) . Thus , the absence of SNAP-25b results in progressing pathological changes in brain areas important for cognitive functions . Epileptic activity causes dramatic changes in peptide expression in the hippocampal formation ( HF ) , in particular affecting the granule cell-mossy fiber system , termed ‘epilepsia peptidergic profile’ [39]–[42] . Since we observed seizure activity in our mouse mutants , we examined with immunohistochemistry the expression of cholecystokinin ( CCK ) , neuropeptide Y ( NPY ) and brain-derived neurotrophic factor ( BDNF ) in the HF of 8-week-old neo-excised SNAP-25b deficient mutants and WT littermates . Numbers of doublecortin ( DCx ) + migrating neuronal precursor cells were also inspected , as seizures may increase neurogenesis . WT mice displayed a wide expression of CCK immunoreactivity ( -ir ) in the HF , being strongest in SLu and supragranular layer of the dentate gyrus ( DG ) ( Figure 6A and 6E ) . Double-labeling experiments demonstrated that CCK+ terminals partially overlapped with SNAP-25-ir in SLu ( Figure 6I–6K ) ; however , SNAP-25-ir had a wider distribution in this layer . CCK-ir in SLu was virtually absent in homozygous neo-excised SNAP-25b deficient mutants ( Figure 6B and 6F , c . f . Figure 6A and 6E ) , but was increased in cortex ( Figure 6B ) . NPY+ fibers and interneurons were fairly evenly distributed in the WT mouse HF ( Figure 6C ) , but NPY-ir was strongly increased in SLu and the polymorph layer of neo-excised SNAP-25b deficient mutants ( Figure 6D and 6H , c . f . Figure 6C and 6G ) . NPY+ terminals/fibers in SLu of mutant mice overlapped with SNAP-25 , although SNAP-25-ir had a more extensive distribution ( Figure 6L ) . Diffuse NPY-ir was increased in the molecular layer of the DG , with lower levels in cortex of neo-excised SNAP-25b deficient mice ( Figure 6C , c . f . Figure 6D ) . BDNF-ir was weak in SLu and polymorph layer of the WT ( Figure 6M and 6O ) , but increased in mutant mice ( Figure 6N and 6P ) . In WT mice , DCx+ cell bodies were distributed as a single layer in the DG subgranular zone ( SGZ ) , while DCx+ processes had an even localization in the molecular and granular cell layers ( Figure 6Q and 6S ) . The number of DCx+ cell bodies was increased by 140% in the SGZ of neo-excised SNAP-25b deficient mice , including cells migrating through the granular cell layer , with a large increase in DCx+ fibers in the molecular layer ( Figure 6R and 6T , c . f . Figure 6Q and 6S and Text S1 , Figure S3 ) . No seizures were observed in juvenile neo-excised mutants prior to weaning , and in 3-week-old animals the expression of neuropeptides and BDNF did not differ compared to WT littermates ( data not shown ) . In conclusion , deletion of SNAP-25b results in dramatic alterations in neuropeptide levels in the HF , which together with increased levels of BDNF and DCx may provide an environment that promotes neuroprotection/neuroproliferation in response to disrupted synaptic transmission and seizure activity in neo-excised SNAP-25b deficient mutants .
In the SNARE complex with syntaxin and VAMP , SNAP-25 contributes with two amphipathic α-helices to the four-helix-coiled-coiled SNARE structure [16] . The nine amino acids that distinguish SNAP-25b from SNAP-25a encompass a region in the SNAP-25 protein that spans the last part of the N-terminal SNARE motif and the first part of the linker that separate the N- and C-terminal α-helices . Substitutions of two non-conservative amino acid differences in the SNAP-25a protein to resemble the SNAP-25b sequence followed by transient overexpression in mouse chromaffin SNAP-25 deficient cells changed the secretion capability of SNAP-25a to become similar to that of SNAP-25b [47] . In addition , it has been established that SNAP-25b containing SNARE complexes are more stable than those containing SNAP-25a [16] , and that SNAP-25b when overexpressed , is more efficient in driving fusion of catecholamine-containing vesicles from chromaffin cells [28] . We took advantage of the fact that homozygous neo-excised SNAP-25b deficient mutants only express SNAP-25a instead of predominantly SNAP-25b as in adult WT mouse brain . We have now been able to show for the first time that ex vivo SNARE complexes in brain homogenates from neo-excised SNAP-25b deficient mutants dissociated at lower temperatures than WT SNARE complexes ( mostly holding SNAP-25b ) . In this respect our attention was drawn to a SNAP-25b mouse mutant named the “blind-drunk mouse” ( Bdr ) with retinal degeneration ( caused by choice of the background strain ) and a mild phenotype with ataxia and impaired sensorimotor gating due to a dominant mutation in SNAP-25b [48] . The Bdr mouse expresses a SNAP-25 protein with even higher affinity for syntaxin than WT SNAP-25b . The SNAP-25b ( Bdr ) SNARE complexes , apparently more stable than complexes with WT SNAP-25b , result in impairment of both spontaneous and evoked release from cortical neurons and lack of facilitation during trains of high-frequency stimulation [48] . The outcome from studies of SNAP-25b ( Bdr ) , SNAP-25b and SNAP-25a containing complexes suggest that the difference in secretory phenotype might be dependent on the difference in complex-stability; and possibly on the interaction with accessory factors due to the presence of different amino acids exposed on the surface of the N-terminal amphipathic α-helix . SNAP-25a and SNAP-25b also exhibit differences in the linker region between the N- and C-terminal SNARE motifs , where a quartet of cysteines , implicated in membrane anchoring of the protein , exists in two different contexts [23]-[25] . We hypothesized that a difference in plasma membrane association between SNAP-25a and SNAP-25b also could add to the phenotype observed in our mutants . Therefore we performed sucrose density gradient fractionation of brain homogenates from adult neo-excised SNAP-25b deficient mutants ( only expressing SNAP-25a ) and WT littermates ( predominantly expressing SNAP-25b ) . In agreement with our assumption the mutants only had around 80% of the SNAP-25 protein levels found in WT plasma membrane fractions . Instead , in mutant brain more SNAP-25 was detected in low-density fractions , representing soluble protein and/or protein associated with small intracellular organelles . This difference in subcellular localization of the two SNAP-25 isoforms might contribute to the phenotype of the mutants . Whether that depends on the ability of the two isoforms to associate with membranes due to the altered organization of the cysteine residues that are substrates for palmitoylation , or , whether SNAP-25a and SNAP-25b reside in complex with additional secretory factors that guide the subcellular localizations needs to be further investigated . To explore a possible effect on presynaptic mechanisms we studied PPF at Schaffer collateral-CA1 pyramidal synapses . We observed a reduction of PPF in young neo-containing and neo-excised SNAP-25b deficient mutants during low-frequency stimulation ( 0 . 2 Hz ) , suggesting a specific effect of SNAP-25b-deficiency . During higher frequency stimulation ( 0 . 5 Hz ) , when Ca2+ may be expected to be elevated during longer time periods in the stimulated presynaptic terminals , a reduction of the PPF-ratio was observed in neo-containing experiments while there was only a non-significant tendency to a reduction in neo-excised SNAP-25b deficient mice . Whether this is a result of the presumed higher average Ca2+-concentration or due to other factors will need to be investigated in future studies . Normally , only few synaptic vesicles are “fusion-competent” and require no further modifications prior to exocytosis . Remaining vesicles close to the site of fusion need to be “primed” , that is undergoing ATP- and Ca2+-dependent maturation steps in order to be mobilized to the fusion-competent state [49] . It has been demonstrated that there is a 2–3 fold higher ability of SNAP-25b to keep vesicles in the primed state than SNAP-25a , a feature not due to facilitated priming but rather dependent on a lower de-priming rate [28] , [50] , [51] . Furthermore , SNAP-25b has been suggested to sustain the exocytotic bursts in a more efficient way than SNAP-25a in flash photolysis experiments with caged Ca2+ in chromaffin cells [28] . The results from the above mentioned investigations indicate that SNAP-25 is not only involved in the membrane fusion reaction but is also likely to play regulatory roles prior to exocytosis , such as in mobilization , docking and priming of vesicles [28] , [38] , [47] . Thus , the results from our PPF studies on SNAP-25b deficient mutants may represent an effect of reduced stability and availability of primed vesicles and are in line with the suggested weaker exocytotic bursts by synapses with SNAP-25a-containing SNARE complexes . Interestingly Bronk et al . [38] showed that there is no synaptic facilitation in the few cultured SNAP-25 KO hippocampal neurons that respond to extracellular stimulation . It may be that the two isoforms of SNAP-25 differ in their ability to sustain Ca2+-dependent types of facilitation including PPF . At physiological levels of SNAP-25 expression and elevated Ca2+-concentrations the replacement of SNAP-25b by SNAP-25a does not affect synaptic release enough to result in a statistically significant effect . However , there was a tendency to a reduction that needs to be investigated further . SNAP-25b is normally highly expressed in brain areas involved in cognitive function , such as the HF . Therefore we examined mutants in behavioral tasks dependent on hippocampal mechanisms . Neo-excised SNAP-25b deficient mutants demonstrated a higher anxiety index compared to WT mice , however , without alterations in overall locomotor activity in the elevated plus maze . This finding is consistent with the profound thigmotaxis observed in the computerized measurements obtained in the water maze task . In the Morris water maze task , homozygous neo-excised SNAP-25b deficient mutants were severely impaired in their ability to acquire the spatial learning task . This impairment cannot simply be explained by visual or motor disturbances although there was a trend for female neo-excised SNAP-25b deficient mutants to perform less well in the visible platform test . The failure of the mutants to consistently improve their performance in the visible platform test is probably related to the same mechanisms as those underlying the deficiency in spatial learning . It is important to stress that the mutants did not display impaired swim performance since they did not differ from WT in their swim speed . However , unlike the WT mice , the mutants displayed a high level of thigmotaxis in the water maze task . This suggests that the deficiency in spatial learning in the mutants could partly be related to their anxious phenotype evidenced by their profound thigmotaxis and/or related to attention deficits observed in the visible platform test . Both neo-containing and neo-excised SNAP-25b deficient mutants demonstrate spontaneously occurring convulsive seizures and freezing behavior . The neo-containing SNAP-25b mutants are most severely affected with frequent attacks debuting around PN12-13 , whereas in neo-excised SNAP-25b deficient mutants seizure activity is rare prior to young adulthood . These behavioral impairments were paralleled by increasingly pronounced anatomical and immunohistochemical changes , which are only observed after the debut of seizure activity . In neo-excised SNAP-25b deficient mice , mossy fibers appear notably enlarged and swollen with certain areas almost devoid of synaptophysin immunoreactivity , suggesting a locally decreased density of functional nerve endings . Increased NPY expression in mossy fibers after seizures is believed to be neuroprotective by dampening excitatory activity , while BDNF can regulate NPY expression [52] , [53] . Both NPY and BDNF have been suggested to promote dentate SGZ neurogenesis [54] , [55] , and their elevated expression may contribute to the increase in DCx+ migrating neuronal precursor cells observed in our mutants . It has been reported that cells generated in the adult dentate gyrus mature into functional neurons that integrate into hippocampal circuitry [55] , [56] , and it was recently shown that newly generated neurons after epilepsy exhibit dampening characteristics [57] . Taken together , it is conceivable that pathology develops in the SNAP-25-expressing mossy fiber area in response to aberrant presynaptic plasticity and synaptic contacts . However , a detailed ultrastructural study will be required to determine any alterations in synaptic morphology . We here demonstrate the physiological importance of the ancient exon 5 duplication in the Snap25 gene . The functions of SNAP-25a and SNAP-25b appear to complement each other in tuning the presynaptic exocytotic machinery towards different modes of release . For complex neuronal circuitries this modulation of release is necessary , and is also instrumental in protecting highly plastic brain areas from accumulating degenerative morphological changes with age . The development of higher brain functions during evolution has been complemented with expanding number of gene products , sometimes by gene duplications but primarily by increasing protein diversity and complexity via alternative splicing . The importance of alternatively spliced isoforms from other genes has previously been extensively analyzed in vivo , by hindering expression of selected exons using gene targeted mouse models [58] . This is the first time an exon knock-out/knock-in of tandem duplicated exons has been performed . We specifically investigated the physiological importance of removing one isoform while preserving total expression levels of the protein by a simultaneous knock-in of an extra copy of the remaining exon variant . Interestingly , a growing amount of reports are connecting SNAP-25 function with a wide variety of behavioral and neuropsychiatric disorders , as well as linking it with cognitive capability in humans [29]–[32] . Our present findings in the SNAP-25b deficient mouse mutants suggest that even small alterations in the strictly regulated temporal and anatomical expression of SNAP-25 isoforms could add to these variations .
To generate SNAP-25b deficient mouse mutants a targeting vector was generated where the exon 5b , located downstream of exon 5a in the Snap25 gene , was substituted with an additional exon 5a . A Tkneo gene surrounded by loxP repeats was introduced into the vector [34] ( Figure 1A ) . Homologous DNA recombination in embryonic stem cells ( ES cells ) was performed using standard procedures [59] . Chimeras that demonstrated germline transmission were chosen for establishing SNAP-25b KO mouse lines ( see Text S1 ) . Three mouse lines , bred independently from each other for at least ten generations on C57BL/6NCrl ( B6 ) mice , were termed “neo-containing” as the Tkneo gene inserted into the targeting vector was still present in the modified Snap25 gene . Two of the neo-containing lines were crossed with the Protamine-Cre recombinase transgene Prm1Cre [34] , resulting in in vivo excision of the Tkneo gene . The neo-excised lines were bred independently onto B6 background until congenic , and the PrmCre1 transgene crossed out . Genotyping was routinely performed by PCR and by Southern blotting when necessary . In vivo expression of the introduced chicken exon 5a was demonstrated by RT-PCR ( Text S1 ) . All animal breeding and studies were done in accordance with the guidelines from local ethical committees . Mouse brains , minus the cerebellum , were isolated after terminal CO2 anesthesia and frozen in liquid N2 . Total RNA was isolated from brain ( PN14-15 ) using the GenElute Mammalian Total RNA kit ( Sigma ) . For quantification of total SNAP-25 mRNA; 1 µg RNA , 10 µµ of each primer and a trace of [α32P] dCTP ( 3000 µCi/mmol , PerkinElmer Life Sciences ) were used in 25 µl reactions with the SuperScript III RT-PCR System ( Invitrogen ) . Semi-quantitative RT-PCR was performed with 20 cycles of amplification and SNAP-25 and glyceraldehyde-3-phosphate dehydrogenase ( GAPDH ) primers in the same reaction ( see Text S1 for primer sequences and PCR programs ) . The RT-PCR products were separated on an 8% polyacrylamide Tris-borate EDTA ( TBE ) gel that was dried , and detected using a phosphoimager ( BAS-1500 , Fujifilm ) . Signal intensities were quantified in Image Gauge V3 . 45 ( Fujifilm ) . Determination of SNAP-25a/b mRNA ratio expression was essentially performed as described [26] ( and Text S1 ) . Statistical analyses were made using Wilcoxon's signed-rank test . Mouse brains were homogenized in ( in mM ) : 20 HEPES , 1 MgCl2 , 250 D-sucrose , 2 EDTA and protease inhibitor cocktail ( Roche Diagnostics GmbH ) , pH 7 . 4 . For whole cell homogenates cells were lyzed with 1% NP-40 ( Sigma ) and 5 µg protein was run on 10% Tris-glycine/NU-PAGE gels ( Novex , Invitrogen ) followed by Western blotting . Primary antibodies used were a rabbit polyclonal antibody against SNAP-25 from Synaptic Systems ( 1∶20 , 000 dilution ) , a rabbit polyclonal anti-SNAP-23 ( 1∶1000 , Synaptic Systems ) , mouse monoclonals anti-syntaxin 1 , HPC-1 ( 1∶50 , 000 and 1∶100 , 000 ) and anti-α-tubulin , clone DM 1A ( 1∶45 , 000 ) , both from Sigma . Secondary antibodies were horseradish peroxidase-conjugated anti-rabbit and anti-mouse immunoglobulins ( IgGs ) from Dako Corporation and Rockland . Statistical analyses were made using Wilcoxon's signed-rank test for paired data . Pulverized adult WT and homozygous neo-excised SNAP-25 deficient ( KO ) mouse brain tissues were homogenized in buffer described for whole cell homogenates . SDS sample buffer [0 . 5M Tris-HCl ( pH 6 . 8 ) , 20% glycerol , 4% SDS , 10% 2-mercaptoethanol and 0 . 05% Bromophenol Blue] was added to equal amounts ( 20–40 µg ) of WT and KO homogenate before heating treatment at 70 , 75 , 80 , 85 , 90 and 100°C ( boiling ) , or kept at 4°C , for 20 min . Treated samples were immediately loaded and separated on 10% Bis-Tris NU-PAGE gels ( Invitrogen ) . SMI81 antibody ( 1∶500 , 000 ) was used to detect immunoreactivity of bands migrating as ternary SNARE complex or solitaire SNAP-25 protein . To measure the grade of disassembly of heat-resistant SNARE complex in WT and SNAP-25b KO tissue , percentage monomeric SNAP-25 protein of total SNAP-25 in ternary complex at 4°C was calculated and compared with the value calculated at a certain temperature . All experiments were performed at least three times . Results were analyzed with unpaired Student's t-test . For histological analysis , mice were anesthetized with isoflurane and transcardially perfused with PBS followed by freshly prepared 4% paraformaldehyde in PBS . Organs were post-fixed overnight , dehydrated in graded ethanol and embedded in paraffin according to standard procedures . 4 µm sections were stained with hematoxylin-eosin or cresyl violet . For immunohistochemistry of paraffin embedded tissue , sections were collected on Superfrost Plus slides , deparaffinized ( xylene ) , dehydrated ( ethanol ) and boiled by microwaving for antigen unmasking ( see also Text S1 ) . Antibodies used were mouse anti-SNAP-25 ( 1∶750 , SMI 81 , Sternberger Monoclonals ) , rabbit anti-SNAP-25 ( 1∶100 , Synaptic Systems ) , mouse anti-synaptophysin ( 1∶200 , SVP38 , Sigma ) , and anti-mouse IgG-Alexa 488 or anti rabbit IgG-Alexa 546 ( 1∶250 , Molecular Probes ) . Neuropeptide immunohistochemistry was performed as described [60] ( see also Text S1 ) . Antibodies used were mouse anti-SNAP-25 ( 1∶750 , 1∶2 , 000 SMI 81 , Sternberger Monoclonals ) , rabbit anti-SNAP-25 ( 1∶100 , Synaptic Systems ) , mouse anti-synaptophysin ( 1∶200 , SVP38 , Sigma ) , chicken anti-BDNF ( 1∶200 , Promega ) and goat anti-DCx ( 1∶100 , Santa Cruz Biotechnology ) . For tyramide signal amplification ( TSA+ , NEN Life Science Products ) antibodies used were rabbit anti-CCK ( 1∶8 , 000 ) , or rabbit anti-NPY ( 1∶3 , 000 ) ( for details see Text S1 ) . Corresponding secondary antibodies were HRP-swine anti-rabbit IgG ( 1∶200 , Dako ) , FITC-donkey anti-chicken , Cy3-donkey anti-goat and Cy3-donkey anti-mouse ( all 1∶100 , and from Jackson ImmunoResearch Laboratories ) . Hind limb specimens were fixed in 4% formaldehyde in phosphate buffer and embedded in paraffin . Sections , 4–5 µm thick , were stained with hematoxylin and eosin . Growth plates of tibia and femur were scanned using a Zeiss Axiovert 35M microscope fitted with a LSR Astro Cam type TE3/A/S digital camera . Images were analyzed using Concord software from Life Science Resources Ltd . The width of the different zones was determined at least in triplicate per section . Per sample , at least two cross sections were measured . Data was analyzed with unpaired Student´s t-test . PN12-16 mice were terminally anesthetized with 100% CO2 and sacrificed by decapitation . Brains were quickly removed and placed in ice-cold low Ca2+/ high Mg2+ artificial cerebrospinal fluid ( aCSF ) containing ( in mM ) : 124 NaCl , 5 KCl , 1 . 24 NaH2PO4 , 0 . 5 CaCl2 , 10 MgSO4 , 26 NaHCO3 , and 10 glucose , and oxygenated with 95% O2/5% CO2 ( pH 7 . 4 ) . Coronal hippocampal slices ( 350 µm thick ) were cut on a vibratome ( Leica VT1000S , Leica Microsystems ) and transferred to regular aCSF ( same composition as above but with 2 . 4 mM CaCl2 and 1 . 3 mM MgSO4 ) at RT . Patch-clamp electrodes ( 6–10 MΩ ) were filled with ( in mM ) : 135 Cs-methane sulphonate , 10 HEPES , 1 EGTA , 4 Mg-ATP , 0 . 3 Na-GTP , 2–5 QX-314 , and 8 NaCl ( pH 7 . 25; osmolarity 270–280 mOsm ) . Whole-cell voltage clamp ( −70 mV ) recordings of evoked excitatory postsynaptic currents ( EPSCs ) were obtained at RT from CA1 pyramidal neurons of the hippocampus , visualized with differential interference microscopy ( DIC ) using an Olympus BX50WI microscope . 50 µM picrotoxin ( Sigma-Aldrich ) was added to the aCSF to block GABAA receptor-mediated synaptic transmission . EPSCs were evoked by stimulation of Schaffer collaterals with fine concentric Pt/Ir bipolar stimulation electrodes ( Fredrik Haer and Co . ) placed in the stratum radiatum . Paired stimuli were delivered at IPIs from 40 to 300 ms at 0 . 2 and 0 . 5 Hz and data was collected after at least 5 minutes of baseline recording to make sure responses were stable . Statistical comparisons of PPF ratios between WT and SNAP-25b deficient mice were made with two-way repeated measurements analysis of variance ( ANOVA ) . For additional information , see Text S1 . The elevated plus maze analysis was performed as described previously [61] . For the water maze task , the animals were handled by the operator for a period of five days prior to the test and spatial learning and memory were examined as described before [62] . Locomotor activity was recorded by means of a multi-cage red- and infrared-sensitive motion detection system as described earlier for mice [61] . Data from the behavioral testing was analyzed by non-parametric statistics using Kruskal-Wallis ANOVA followed by Mann-Whitney U as the Post-hoc test . | In evolution , duplication of genes or gene segments appears to be an efficient way to add diverse functions in more complex organisms . The SNAP-25 protein plays an important role in mediating the release of neurotransmitters and hormones . SNAP-25 exists as two variants: SNAP-25a , which is present in early development , and SNAP-25b , which is most abundant from early adulthood and onwards . We have developed mouse mutants that only express SNAP-25a , but retain normal SNAP-25 levels by replacing the SNAP-25b segment in the Snap25 gene with an additional SNAP-25a copy . We show that SNAP-25b is required for early postnatal development and that a balanced expression of the two proteins is a prerequisite for maintaining an operational neuronal network during adulthood . Mice that only have SNAP-25a develop seizures , and show learning deficits and anxiety . Synaptic plasticity is impaired , and structural changes are observed in areas that are connected to such behavioral functions . In man , SNAP-25 function has been linked to behavioral and neuropsychiatric disorders , including attention deficit hyperactivity disorder , ADHD . Our present findings using genetic elimination of SNAP-25b suggest that even small alterations in the regulation of the Snap25 gene , resulting in a disturbed balance between SNAP-25a and SNAP-25b , lead to nervous system dysfunction . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"neuroscience/behavioral",
"neuroscience",
"cell",
"biology/cell",
"signaling",
"developmental",
"biology/developmental",
"evolution",
"cell",
"biology/neuronal",
"signaling",
"mechanisms",
"cell",
"biology/membranes",
"and",
"sorting",
"physiology/cell",
"signaling",
"neuroscien... | 2008 | An Ancient Duplication of Exon 5 in the Snap25 Gene Is Required for Complex Neuronal Development/Function |
Characterizing the fitness landscape , a representation of fitness for a large set of genotypes , is key to understanding how genetic information is interpreted to create functional organisms . Here we determined the evolutionarily-relevant segment of the fitness landscape of His3 , a gene coding for an enzyme in the histidine synthesis pathway , focusing on combinations of amino acid states found at orthologous sites of extant species . Just 15% of amino acids found in yeast His3 orthologues were always neutral while the impact on fitness of the remaining 85% depended on the genetic background . Furthermore , at 67% of sites , amino acid replacements were under sign epistasis , having both strongly positive and negative effect in different genetic backgrounds . 46% of sites were under reciprocal sign epistasis . The fitness impact of amino acid replacements was influenced by only a few genetic backgrounds but involved interaction of multiple sites , shaping a rugged fitness landscape in which many of the shortest paths between highly fit genotypes are inaccessible .
Predicting function and fitness of organisms from their genotypes is the ultimate goal of many fields in biology , from medical genetics to systems biology to the study of evolution [1–5] . Among the conceptual frameworks for understanding the genotype-to-phenotype connection is the fitness landscape , which assigns a fitness ( phenotype ) to every possible genotype ( sequence ) of a gene or genome under consideration [4 , 6] . The recognition of the importance of the fitness landscape stimulated the development of a variety of theoretical approaches to describe it , including its general shape and epistatic interactions between alleles , a key property which determines the complexity of the fitness landscape ( see [4] and references within ) . Before the advent of next-generation sequencing , experimental assays of the fitness landscape were few and could not address the issue at the sequence level . Recently , large-scale experimental assays described the shape of the fitness landscape a few mutations away from a local fitness peak ( see [7–10] and references within ) . Also , some assays involving a smaller number of genotypes considered combinations of mutations with established functional [11–17] or evolutionary [18–25] significance . Empirical evidence of the nature of large-scale fitness landscapes mostly comes from the study of genotypes incorporating random mutations [4 , 7–10] , the majority of which are deleterious [7–10 , 26] . Thus , our present knowledge of fitness landscapes is primarily driven by the study of deleterious mutations and their interactions , although local adaptive trajectories have also been considered [2 , 4 , 16 , 27–29] . Deleterious mutations were found to engage in synergistic epistasis , whereby the joint effect of multiple mutations was stronger than the sum of their individual effects [4 , 7–10 , 16] . Furthermore , sign epistasis among random mutations was mostly rare [5 , 7–10 , 16 , 30] , although some of these conclusions differ from study to study ( e . g . see [4 , 30] ) . Unfortunately , there are fundamental limitations to assaying the fitness landscape on a large or macroevolutionary level with random mutation libraries . The number of genotypes underlying the fitness landscape is the combinatorial set of all amino acids across the length of the protein [4 , 6] . For example , for the 220 amino acid protein coded by the His3 gene in Saccharomyces cerevisiae , the fitness landscape is a 220-dimensional genotype space with 20220 different possible sequences . Such immense spaces are both computationally and experimentally intractable . Fortunately , it may not be necessary to survey all genotypes to study the evolutionary-relevant section of the fitness landscape . Because the vast majority of mutations in protein sequences are deleterious [26] , a randomly sampled protein sequence is non-functional [31 , 32] . Here we propose an evolutionary approach for assaying fitness landscapes on a macroevolutionary scale in a high-throughput manner that avoids the random sampling of mostly non-functional sequences . The functionally and evolutionarily relevant section of the fitness landscape can be represented by the combination of extant amino acid states , those found in extant species . This approach applied previously on a limited scale [18–25] mitigates the problem of exploring a prohibitively large fitness landscape while highlighting the relationships between evolutionarily-relevant genotypes ( Fig 1A ) . Crucially , substitutions that have been fixed in evolution are fundamentally different from random mutations [26] , the former are either neutral or beneficial in at least some genetic contexts and represent the driving force of molecular evolution , while the latter are mostly deleterious and are primarily relevant on a microevolutionary scale . Therefore , current empirical data do not shed much light on the impact of interactions between amino acid states that were fixed in the course of evolution by natural selection . Combinations of extant amino acid states represent the area of the sequence space that considers all possible orders in which amino acid substitutions from evolution could have happened allowing one to assay a much wider functionally relevant area of the sequence space than approaches based on random mutagenesis of a single sequence ( Fig 1B and 1C ) .
We studied His3 , a gene coding for imidazoleglycerol-phosphate dehydratase ( IGPD , His3p ) , an enzyme essential for histidine synthesis . In a multiple alignment of His3 orthologues from 21 yeast species we identified 686 extant amino acid states ( S1 Supporting Information ) , which were evenly distributed across the His3p structure ( Fig 1D ) . These 686 states , which represent the end product of ~400 million years of evolution [33] ( Fig 1B and 1C ) correspond to ~1083 sequences , even a tiny fraction of which would be too many to survey . Thus , we sectioned His3 into 12 independent segments such that the full combinatorial set of amino acid states that occurred in His3 during yeast evolution comprised 10 , 000-100 , 000 genotypes per segment ( see Methods and S1A Fig ) . The 12 segments were of similar length , constrained by the molecular methods employed for library construction ( see Methods ) , and covered a diverse range of secondary structures and functional elements ( S1C Fig ) . For each of the 12 segments of His3 we performed an independent experiment surveying its fitness landscape . For each segment we used degenerate oligonucleotides to construct genotypes consisting of combinations of amino acids present in extant His3 sequences , and determined the fitness conferred by these genotypes by expressing them in a Δhis3 strain of S . cerevisiae and measuring the rate of growth ( S1B Fig ) . This way for each segment we assayed the fitness landscape of the genotype space that was traversed over the course of the last ~400 million years of evolution [33] . The segmentation of the His3 protein into 12 segments at first glance appears not to be relevant to understanding the fitness landscape of the entire protein . Indeed , it would have been more informative to survey combinations of extant amino acid states across the entire His3 sequence . Our choice to study independent segments was dictated by our preference for depth over width , in other words , we were more curious to consider combinations of extant amino acid states from distant orthologues rather than consider combinations of extant amino acids states from a few closely related species across the entire gene . Our choice of selecting random segments across the protein is justified in the same way as surveys of individual genes; segments may interact to form a functional protein while genes interact to contribute to the overall organismal fitness . Indeed , the lessons learned from fitness landscapes of random protein segments are likely to be scalable to the level of the entire protein to a greater extent than how fitness landscapes of an entire protein can be scaled to understand the fitness landscape of a genome . Across 11 experiments , we measured fitness for a total of 4 , 018 , 105 genotypes ( 875 , 151 unique amino acid sequences ) with high accuracy ( S2 Fig ) . Of these , 422 , 717 consist solely of combinations of extant amino acid states from His3 orthologues , while the remaining genotypes incorporate other amino acid changes ( S2 Supporting Information and Methods ) . Throughout the study , we considered the logarithm of fitness to allow for the sum of the impact of individual amino acid replacements on fitness to represent the fitness function . For one segment , 9 , the accuracy of our experiment was low , and it was not used in cumulative analyses . For each segment we measured fitness for 60% - 99 . 8% of all possible genotypes from the combinatorial set of selected extant amino acid states found in 21 yeast species and a smaller fraction of combinations found across all domains of life ( S2 Supporting Information ) , characterizing the evolutionary relevant fitness landscape ( Fig 1B ) . For segment 3 for instance , 11 out of 17 amino acid sites had more than one extant amino acid state: L145 = 2 , L147 = 2 , Q148 = 3 , K151 = 2 , V152 = 2 , D154 = 3 , L164 = 3 , E165 = 4 , A168 = 2 , E169 = 4 , A170 = 4 , with the full yeast combinatorial set consisting of 2*2*3*2*2*3*3*4*2*4*4 = 55 , 296 genotypes out of which we determined the fitness for 48 , 198 , or 87% of the possible yeast extant states combinations in our library . Our experimental design not only generated combinations of extant amino acids , but also allowed us to distinguish real variants from less abundant aberrations originating from sequencing errors ( S2 Fig ) . We further reduced genotyping errors by sequencing every variant twice through paired-end sequencing ( S2A Fig ) , and by using an error-correction algorithm ( see Sequencing error rate section ) . Most of the genotypes that can be created by the combination of extant amino acid states are not ancestral His3 protein segments . In a high-dimensional sequence space , the number of actual ancestral sequences is exponentially smaller than the number of combinations of extant states ( Fig 1A ) . Out of the 48 , 198 extant amino acid state combinations in segment 3 only seven sequences actually match the reconstructed ancestral states of the entire segment . The combinations of extant amino acid states represent the evolutionary-relevant segment of the fitness landscape because they represent all possible alternative evolutionary trajectories in sequence space that consist of the same substitutions that have occurred in evolution of the extant protein sequences under consideration . Furthermore , some of these seven sequences were likely found in a different context of the entire protein , so they do not represent the ancestral states of the entire His3 protein but rather the ancestral states found in segment 3 across different genetic context of the rest of the His3 protein . A substantial proportion of combinations of extant amino acid states led to genotypes with low fitness ( Fig 2 , Fig 3A and 3B , S3 Fig ) , an observation that takes into account the false discovery rate in our data ( S2 Supporting Information ) . This observation could be explained by i ) some extant amino acids having a universally deleterious effect , ii ) some amino acid states exerting a negative effect on fitness because of intergenic interactions with other S . cerevisiae genes , or iii ) by epistatic interactions between the extant amino acid states within His3 [34] . We exclude the possibility that some extant amino acid states had a universally lethal effect in S . cerevisiae background because no extant amino acid states were present only in unfit genetic backgrounds , genotypes conferring a fitness of zero . Indeed , no extant amino acid states had a universally strong deleterious effect , a decline of fitness by more than 0 . 4 ( Fig 3D ) , and the smallest number of fit backgrounds in which any extant state was found was approximately 300 ( Fig 3C ) . We use the 0 . 4 threshold because it corresponds to ~1% false discovery rate ( see Methods ) . We exclude the possibility that some extant amino acid states disrupt intergenic interactions because the complete His3 coding sequences from extant species fully complemented a His3 deletion in S . cerevisiae ( S4C Fig ) . Thus , the observed genotypes with low fitness can only be explained by epistatic interactions among extant amino acid states within the His3 gene in the same or different segments . Remarkably , 85% ( 330/389 ) of replacements between extant amino acid states had substantially different effects on fitness in different backgrounds ( Fig 3D ) . By contrast , only 15% of amino acid replacements are truly neutral , in the sense that they do not exert strong influence on fitness in any genetic background . Three quarters of the universally neutral amino acid replacements were observed in the disordered region of the protein ( 44/59 ) . Taken together , the His3 fitness landscape across the 11 segments with high accuracy was strongly influenced by epistasis on a macroevolutionary scale , i . e . the impact of an extant amino acid state on fitness often depends on the background in which it occurs [34–37] . An epistatic fitness landscape is rugged in the sense that evolving genotypes must avoid fitness valleys that emerge through deleterious combinations of amino acid states that may also be found in fit genotypes [19 , 20 , 37–38] . Characterizing the ruggedness and the mechanisms that determine the underlying epistasis becomes the primary challenge in understanding the fitness landscape of His3 . The ruggedness of the fitness landscape can be characterized by different measures of complexity of the underlying epistatic interactions . In the simplest case , epistasis may be unidimensional , in the sense that the fitness landscape can be described as a function of an intermediate variable , the fitness potential [40–42] . The fitness landscape is a function from the space of genotypes to fitness . In analogy with a scalar field , we can characterize the ruggedness of this function with standard measures of complexity if genotypes are arranged in a linear space . The simplest case is that of a linear predictor called the fitness potential: p = c1x1 + c2x2 + … + cnxn , where ci is a coefficient and xi is a binary variable that signifies the presence ( 1 ) or absence ( 0 ) of a given amino acid at a given position . By definition , ep describes a non-epistatic fitness landscape because the effect of every amino acid replacement is multiplicative , and it depends only on the associated c . Any other function of p leads to epistasis . If the f ( p ) function is “simple” , meaning that it has a small number of local extrema , such as a bimodal function , the epistasis is called unidimensional [42] . The limitation of simplicity of f ( p ) is necessary because any function f0 ( x1 , … , xn ) can be represented by a function f’ ( p ) and choosing appropriate coefficients c1 , … , cn in p . Thus , a simple f ( p ) leads to unidimensional epistasis because the entire genotype space can be reduced to a single dimension [42] . To quantitatively determine how well fitness differences between genotypes can be explained by unidimensional epistasis we used a deep learning approach to estimate the coefficients c for each allele x in the fitness potential and determine the unidimensional function f ( p ) that best approximated the fitness landscape . We used a dense neural network architecture composed of three layers . Each neuron in the architecture performed a linear transformation of its input and then applied a nonlinear ( sigmoid ) function . The single neuron of the first layer computes the fitness potential , which is then mapped to a fitness value obtained from f ( p ) , the function of the fitness potential which is found by the three layers of the neural network architecture ( see Methods; Fig 4 ) . For ten protein segments , f ( p ) converged to the same shape: a threshold function in which organismal fitness remains constant with decreasing fitness potential and then is rapidly reduced to lethal after a certain threshold ( Fig 5A ) . This analysis , coupled with previous observations of thresholds in fitness landscapes [7 , 10 , 43] suggests that macroevolutionary fitness landscapes may in fact represent instances of truncation selection . Remarkably , no fitness functions showed a defined optimum , implying a lack of stabilizing selection on His3 protein function and that it was maintained at an optimized state throughout its evolution . The ability of the cliff-like threshold fitness function [44] to predict fitness from genotype varied between the His3 segments from near perfect ( r2 = 0 . 97 ) in segment 7 , to relatively poor ( r2 = 0 . 44 ) in segment 5 ( S5 Fig ) . Thus , while the fitness landscape of His3 is approximately unidimensional for some segments , it has a higher degree of complexity for others . Ruggedness is a general property of fitness landscapes that quantifies the accessible paths of high fitness that connect fit genotypes [45–47] . A path between highly fit genotypes is inaccessible when one of the intermediate genotypes has low fitness [6 , 21–23 , 45 , 48] ( e . g . , for genotypes AB and ab , the intermediate are aB and Ab ) . Such instances also manifest in sign epistasis on the fitness landscape , that the same amino acid replacement may be beneficial or deleterious when occurring in a different genetic background [48 , 49] . To quantify the ruggedness of the His3 fitness landscape we identified instances of sign epistasis: replacements between extant amino acid states that were strongly beneficial in some backgrounds ( increasing fitness by at least 0 . 4 in absolute fitness ) or strongly deleterious ( decreasing fitness by at least 0 . 4 in absolute fitness ) in other backgrounds [48] . Some of these instances may be due to miscalled fitness of very few genotypes . Therefore , we considered a pair of extant amino acid states to be under sign epistasis only when sign epistasis was observed in a statistically significant number of different genetic backgrounds ( see Methods section Quantifying sign epistasis ) . An example of sign epistasis is the C141S replacement in the second segment that had an opposite effect on fitness depending on amino acid at site 143 ( I , V or T ) . The I143T replacement in turn exhibits sign epistasis depending on the amino acid at site 163 ( F , I , V or L ) ( Fig 6A ) . These epistatic interactions can be represented by a graph in which nodes represent a pair of extant amino acid states at a specific site and nodes are connected by edges if strong sign epistasis has been detected between them ( C141S - I143T - I163F ) ( Fig 5B ) . We found that 86 out of 128 ( 67% ) sites in our library exhibit sign epistasis and 46% ( 59/128 ) exhibit reciprocal sign epistasis with 8% ( 968/11597 ) of all pairs of sites exhibiting sign epistasis ( see S2 Supporting Information Table 5 ) . Most sites showed a sign epistatic interaction with multiple other sites ( Fig 6C , S6 Fig ) demonstrating that , although sign epistasis affects few genotypes , it leads to a fitness landscape that requires the interaction of multiple sites for proper characterization . Sign epistasis can appear when fitness is described by a unidimensional function of the fitness potential , for example , when the fitness landscape is a unimodal function with an optimum in an intermediate range of the fitness potential [45 , 49] . However , sign epistasis may also be a sign of multidimensional epistasis , when a unidimensional function of the fitness potential cannot fully describe genotype fitness [42] . Many genotypes were predicted poorly by a unidimensional function of the fitness potential ( S5A Fig ) . Two lines of evidence suggest that such genotypes reveal the presence of multidimensional epistasis . First , genotypes with a higher number of amino acid replacements influenced by sign epistasis were less well-predicted by a unidimensional fitness function ( Fig 5C and S7B Fig ) . Second , we explain a larger fraction of genotypes by using a more complex neural network architecture accommodating multiple fitness potentials instead of one . We found that increasing the number of neurons in the first layer of the neural network architecture , which is equivalent to increasing the number of independent fitness potentials , gradually improves the prediction power of the obtained models for most of the segments ( Fig 5D ) . Thus , adding dimensions to the function of fitness potential increases the prediction power of the model . For example , for a two-dimensional case fitness was described by f1 ( p1 , p2 ) with p1 = a1x1 + a2x2 + … + anxn and p2 = b1x1 + b2x2 + … + bnxn . For several His3 segments , a fitness function with multiple underlying fitness potentials described the fitness landscape more accurately than a simple unidimensional function of a single fitness potential ( S7A Fig ) . For instance , for these segments , fitness function of two fitness potentials described the shape with a higher degree of accuracy than a function of a single fitness potential ( Fig 5D and 5E ) . By contrast , epistasis in segment 7 is entirely unidimensional ( Fig 5D and 5E and S3 Supporting Information ) ; we do not see any improvement in the model’s predictive power when adding extra dimensions . On a smooth fitness landscape , evolution can proceed along any of the evolutionary paths connecting two fit genotypes , as none of the intermediate genotypes confer low fitness ( see Box 2 in [50] ) . Alternatively , the fitness landscape is rugged when it contains non-connected fitness peaks , such that there are no viable paths between some pairs of genotypes that confer high fitness [4 , 5] . In other words , the presence of deleterious intermediate genotypes between highly fit ones leads to inaccessibility of some evolutionary trajectories between extant or ancestral sequences [6 , 21–23 , 48] . The simplest explanation for the substantial ruggedness of the landscape observed in many of the His3 segments lies in the unidimensional threshold fitness function ( Fig 7A ) . On a threshold function , a combination of amino acid replacements that are all neutral in some genetic backgrounds can take a genotype beyond the fitness threshold through their additive effect on fitness potential , making some genotypes inaccessible for evolution ( Fig 7A ) . Between any two fit genotypes , the fraction of intermediate genotypes that are unfit depends on the fitness potential of the two parental genotypes ( Fig 7B ) . Evolution between two fit genotypes with high fitness potential can proceed unhindered because all intermediate genotypes also have high fitness potential and , consequently , high fitness . Conversely , when both fit genotypes are located close to the threshold , many of the intermediate genotypes between them have low fitness and many evolutionary paths between them are inaccessible ( Fig 7C ) . Thus , the cliff-like threshold fitness function is the major determinant of the observation that not all paths between two fit genotypes are accessible to evolution ( Fig 7B ) . We find that unfit intermediate genotypes are in genetic proximity with each other and are on a limited number of paths; the fraction of inaccessible paths is smaller than if the same number of unfit genotypes were distributed randomly in genotype space ( Fig 7D and 7E ) . The effect of synergistic epistasis dominates the His3 fitness landscape ( Fig 5A ) , affecting over 85% of amino acid replacements from our library that occurred in His3 evolution ( Fig 3D ) . This synergistic epistasis may reflect the folding free energy of the protein [10 , 51 , 52] , as evidenced by a weak to modest correlation between the fitness potential and the impact of amino acid replacements on the folding free energy of His3p ( S8 Fig ) . Similarly , instances of sign epistasis may also be explained by changes in protein stability; for example , in the 143T background C141S increased fitness and also had a positive effect on stability ( Fig 6B ) . Consistent with protein stability contributing to the observed sign epistasis we find that sites that exhibited reciprocal sign epistasis are close together in the His3p structure ( S8 Fig ) . Only a relatively distant His3 structure to S . cerevisiae was used in this analysis , with about 30% divergence , likely reducing the accuracy of the free energy estimates . An additive contribution to free energy can lead only to a unidimensional fitness function [51] , indicating that other non-additive mechanisms , such as catalytic activity or inter-subunit interactions , or non-additive contribution to protein stability must be responsible for the multidimensionality of the His3 fitness landscape . Epistasis may be caused by interaction among positions within a segment ( intra-segmental epistasis ) or by interaction of the segment with the rest of the S . cerevisiae His3 sequence ( inter-segmental epistasis ) . In other words , a specific genotype of one segment may be associated with low fitness because of interaction of amino acid states on that segment with amino acid states at other segments . To some extent , the contribution of inter- versus intra-segmental interactions can be decoupled . Given two fit genotypes ( e . g . , ABC & abc in one His3 segment ) , any unfit intermediate states ( e . g . , aBc in the same His3 segment ) must be due to intra-segmental epistasis because the rest of the protein remains constant . For each segment , we took as a measurement of intra-segmental epistasis all pairs of fit genotypes and calculated the proportion of unfit intermediate genotypes as a function of the Hamming distance between the two fit genotypes . We then compared this proportion with the total proportion of all unfit genotypes as a function of Hamming distance from S . cerevisiae , a measurement that includes both inter- and intra-segmental epistasis . We found three times more inter-segmental than intra-segmental epistasis ( S9 Fig ) , likely because a single segment provides a much smaller target space for interactions than the entire His3 protein . On the other hand , inter-segmental epistasis was just three times more frequent , rather than ten times more frequent , possibly due to higher probability of interaction of amino acid residues that are proximal in primary structure [53] . The proportion of sites under epistatic interactions increased exponentially with Hamming distance ( S9 Fig ) , analogous to Orr’s snowball , the accumulation of genetic incompatibilities in the course of speciation [34 , 54] .
The concept of the fitness landscape introduced by Sewall Wright ( Figs 1 and 2 in [6] ) is an indispensable tool for understanding multiple biological phenomena [1–5] . Experimental high-throughput assays of random mutations have begun to unravel some local properties of fitness landscapes [4] . Here , we described a fitness landscape on a macroevolutionary scale by focusing on amino acid states that have been put through the sieve of natural selection . We found that only 15% of amino acid states that were fixed in the evolution of His3 are universally neutral . For the remaining 85% , amino acid replacements had a profound influence on each other’s effect on fitness , providing an experimental confirmation that epistasis is one of the defining features of molecular evolution [36] . Substitutions that occur in evolution have properties vastly different from those of random mutations , which are mostly deleterious [26] . Therefore , the way in which combinations of extant amino acid states affect fitness may also be different from that of combinations of random mutations . Unexpectedly , we found that the interaction of extant amino acid states was dominated by synergistic epistasis in a manner similar to that previously found for random mutations [7–10 , 16] . However , the accumulation of random mutations leads to low fitness much faster than the accumulation of extant amino acid states ( compare Fig 2 from [16] and Fig 3B from [10] to Fig 2A ) . The experimental data showing that 85% of amino acid states found in extant species confer low fitness in a different genetic background lends strong support to the notion that epistasis is a key factor in protein evolution [34 , 36] . We showed that the fitness landscape of several segments of the His3 gene cannot be reduced to a single unidimensional forms of epistasis , with a function of multiple fitness potentials providing a more accurate description of the fitness landscape . By contrast , large-scale fitness landscapes incorporating multiple random mutations away from the wildtype sequence in a constant test environment have not displayed evidence of multidimensional epistasis [8–10 , 16]; however , it appears to be a more prevalent factor among amino acid replacements that have been subject to positive selection [12–16 , 19–23 , 37–39] . We also found that up to 67% of sites with an extant amino acid state were influenced by sign epistasis , resulting in a rugged fitness landscape and a limited number of fitness ridges connecting extant sequences for most His3 segments . However , sign epistasis substantially influenced fitness in only a small number of genotypes ( Fig 5D ) . Overall , the evolutionary-relevant section of the His3 fitness landscape is best described as a fitness ridge , with the crest of the ridge defined by a fitness potential . In some cases , the crest is multidimensional requiring several independent underlying fitness potentials . Evolution can proceed unhindered along the crest ( Fig 7A and 7C ) , however , pathway availability declines rapidly when evolution proceeds close to the edge of the fitness ridge .
The raw and processed data have been submitted to the NCBI Gene Expression Omnibus ( GEO;http://www . ncbi . nlm . nih . gov/geo/ ) under accession number GSE99990 . A virtual machine containing a running version of the data processing pipeline is available as a Docker image https://hub . docker . com/r/gui11aume/epi/ . The scripts to reproduce the figures are on Github at https://github . com/Lcarey/HIS3InterspeciesEpistasis . The His3 gene was selected for three principal reasons , it is short , conditionally essential and has not been known to be involved in protein-protein interactions . Studying 20220 variants of His3 is impossible , thus , we have chosen an approach to survey the fitness landscape in a manner that would elucidate the area most relevant to His3 evolution while managing the technical limitations of our experimental design . We considered amino acid states found in extant species , focusing on yeast species , which translated into a full combinatorial set of ~1083 unique genotypes . Technically , it is feasible to measure fitness on the order of 100 , 000 unique genotypes in a single growth experiment . Therefore , we split the His3 gene into 12 independent segments such that the full combinatorial set of extant amino acid states from 21 yeast species in each segment was 10 , 000 – 100 , 000 genotypes . We then considered the combinatorial library for each segment in an independent growth experiment , which allowed us to study a tractable section of the sequence space while considering trajectories across a vast part of the space connecting extant species ( Fig 1A ) . We constructed these combinations in 12 plasmid libraries and transformed them into a haploid His3 knockout S . cerevisiae strain . Growth rate ( fitness ) of yeast carrying different mutations in His3 was measured using serial batch culture in the absence of histidine . We split the His3 gene sequence into segments in a manner agnostic to the structure of the His3 protein ( S1A Fig ) . For technical reasons , a segment consisted of two variable regions with a constant region between them ( S1B and S2A Fig ) . All growth experiments were performed independently for each segment , with the exception of one experiment on a limited group of genotypes from each segment which was done for the normalization of fitness values across different segments ( S4 Fig ) . As a control , we measured the rate of growth of S . cerevisiae whose entire His3 gene sequence came from another distant species . We found that the replacement of an entire gene sequence of His3 leads to wild-type rates of growth of S . cerevisiae even when the His3 sequence comes from very distant yeasts , as far as S . pombe ( S4 Fig ) . Therefore , His3 appears to be an independent unit of the fitness landscape and is a good model for the study of fitness landscapes of an isolated gene . We isolated 197 strains from all segment libraries of extant amino acid combinations ( 9-26 strains per segment ) and used Sanger sequencing to determine the sequence . For each strain we performed 6 repeats of growth assay and calculated the average growth rate . Fitness values from competition and growth rates are highly correlated ( r = 0 . 82 , p = 10-48 ) . Correlation was significant and greater than 0 . 6 for all segments except segment 9 , where all selected genotypes appeared to be neutral ( S4 Fig ) . The individual sequences of the variants were recovered from pair-end reads with the following steps: the constant region between the two variable regions was identified by inexact matching allowing up to 20% errors using the Seeq library version 1 . 1 . 2 ( https://github . com/ezorita/seeq ) . The reads are not oriented because the Illumina sequencing adapters were added by ligation , so the constant regions were searched on both reads . Forward and reverse reads were swapped when a match was found on the reverse read . This ensured that all of the sequences are in the same orientation . For multiplexing purposes , the sample identity was encoded in the left and right primers used to PCR-amplify the variants . To demultiplex the reads , we used inexact matching with the candidate primers , allowing up to 20% errors . To merge the reads , the sequence of the reverse reads was reverse complemented and the constant region was searched by inexact matching allowing up to 20% errors . The position of the constant part in each read indicated how they must be stitched together . This approach was faster and less error-prone than using FLASH [56] . In the region of overlap , the consensus sequence was determined by picking the nucleotide with highest quality as indicated in the quality line of the fastq files . If 'N' persisted in the final sequence , the reads were discarded . The PCR primers were trimmed so that all the sequences of the same competition would start and end at the same location . Reads that did not have the constant region , that could not be oriented or that could not be demultiplexed were discarded . The remaining errors in the reads were corrected by sequence clustering . We used Starcode version 1 . 0 [57] with default parameters and allowing up to two errors . The corrected reads were translated using the genetic code . Variants encoding the same proteins were not merged; they were kept separate for downstream analyses . A running Docker virtual machine with commented scripts to replay the whole the process is available for download at https://hub . docker . com/r/gui11aume/epi/ . The total number of reads for 12 segments , 3 time points and 4 replicas are shown in S2 Supporting Information . Genotypes frequencies are defined as the number of reads for a given genotype divided by the total number of reads in that replicate . Mean frequency was calculated over 4 replicas to be used in further analysis . However , to eliminate influence of outliers the median was taken instead of mean if absolute difference between mean and median was greater than the median value . Only genotypes present in both technical replicas of both biological replicas with at least ten reads ( summed across all time points ) in each of them were kept . The length of the segment was designed so that each variable region was read twice by pair-end reads ( S2A Fig ) . This strategy led to a substantial reduction in the sequencing error rate because mismatches between the two reads were corrected to the nucleotide with the higher quality call . The raw Illumina sequencing error rate was estimated by measuring the frequency of mismatches between forward and reverse reads . The rationale of this estimate is that each mismatched nucleotide must be a sequencing error for at least one of the reads . The variant calling error rate was estimated by collecting groups of reads that differ by only one nucleotide in the constant region ( S2A Fig ) . Since the variable regions are identical , such reads come from the same variant and the different nucleotide in the constant region must originate from a calling error ( mutations in S . cerevisiae are negligible because they occur at a much lower rate ) . The frequency of such reads was used to approximate the per-nucleotide variant calling error rate . The raw Illumina error rate was computed by custom Python scripts and reads differing by one nucleotide were collected using the “sphere” clustering option of Starcode with a maximum distance of 1 ( see the Data access section for the code repositories ) . The results are summarized in S2 Supporting Information . The design strategy and the low error rate allow us to distinguish variants incorporated in the library from random sequencing errors . Each library contained on the order of 105 individual sequence variants , so that each library variant would be found at a frequency several orders of magnitude higher than variants created by sequencing errors . For example , for segment 7 , the number of nucleotide variants was 176 , 879 with 31 , 815 , 448 possible single mutants of these variants . The error rate in segment 7 was 0 . 04% per nucleotide , which translates into 2 . 4% of reads being miscalled . Thus , the expected frequency of a particular miscalled variant is 0 . 024 / 31 , 815 , 448 ≈ 8x10-10 , considerably smaller than the expected frequency of real variants 0 . 976 / 176 , 879 ≈ 6x10-6 . The estimated frequencies of library variants for all segments of His3 are reported in S2 Supporting Information . The final variant calling error rate was smaller than the numbers shown in S2 Supporting Information because errors were further corrected by sequence clustering using Starcode ( see Initial data filtering section ) . In line with the rationale above , low frequency erroneous reads are converted to the closest high frequency variant at a maximum Levenshtein distance of 2 . It is known that PCR can create new genotypes by template switching [58] . To test the magnitude of this effect , we took advantage of the two-block design of the variants and estimated the frequency of recombination between the left and the right variable regions . The structure of reads can be represented like this: AAAAAA-------BBBBBB . In this example , “A” represents the left part of the variable region of the segment , “-“ represents the invariable region and “B” represents the right part of the variable region . Insertions of two or more nucleotides are rare events caused by errors during the library synthesis; so , if the same insertion in a left half of the variant is associated to several variants on the right half , it is likely an occurrence of template switching ( the same holds for insertions in the right half of the variant ) . For example , if the region with A*AAA*AA-------BBBBBB , where “*” represents a deletion , is found with several different variants of the B segment then such a situation likely represents template switching . Focusing on the initial time point to avert the effect of selection , we counted a total of 11 , 454 variants with two or more insertions on either the left side or the right side . Among those , 76 had the same insertion as another variant . This means that > 98 . 6% of the variants were free of template switching . Extrapolating to the rest of the dataset , this means that the “leakage” of reads between variants is substantially lower than the magnitude of the observed epistasis and we can rule out this artefact as a potential explanation for our results . In either case , all errors in our experimental pipeline , including template switching , are taken into account when we calculate the false discovery rate . The major factors causing noise in genotype frequency measurements are sampling errors , PCR amplification errors and genetic drift during the competition . For all of these factors , the amount of error depends on the genotype frequency . Therefore , we estimated measurement errors as the function of genotype frequency . For a given segment , time point and a pair of biological or technical replicas for each genotype we calculated the mean frequency and the squared difference of frequencies from these two replicas . We sorted genotypes by mean frequency and grouped them such that each bin contains 5000 genotypes . We calculated the average frequency and the average squared difference in each bin . Additionally , squared error for frequency 0 was set equal to 12∙ ( ( 0 . 5Ni ) 2+ ( 0 . 5Nj ) 2 ) , where Ni and Nj are total read numbers in replicas i and j . Finally , by linear interpolation we obtained dependencies of squared differences as a function of frequency , sij2 ( f ) , where i and j are different replicas . Using squared differences from pairwise comparison of replicas we can estimate variance of mean frequency over four replicas . Let numerate replicas 1 , 2 , 3 , 4 where 1 , 2 are technical replicas of the first biological repeat and 3 , 4 are the technical replicas of the second biological repeat . Errors coming from the competition ( e . g . : genetic drift ) are shared for replicas 1 , 2 and for replicas 3 , 4 . Let’s call them Δfb1 and Δfb2 and their variances σb12 and σb22 , respectively . Technical errors of sampling from the population and from PCR are unique for each replica . Let’s call them Δfti , i=1 . . 4 and their variances σti2 , i=1 . . 4 , respectively . All variances are function of frequency and when writing σXi2 we assume σXi2 ( f ) . In the introduced notations the mean frequency over 4 replicas is: f=14∙ ( f1+f2+f3+f4 ) =14∙ ( ( f*+Δfb1+Δft1 ) + ( f*+Δfb1+Δft2 ) + ( f*+Δfb2+Δft3 ) + ( f*+Δfb2+Δft4 ) ) =f*+12∙ ( Δfb1+Δfb2 ) +14∙ ( Δft1+Δft2+Δft3+Δft4 ) , where f* is the true frequency . Applying basic properties of variance , the variance of mean frequency: σ2=14∙ ( σb12+σb22 ) +116∙ ( σt12+σt22+σt32+σt42 ) To estimate σb12 , σb22 , σt12 , σt22 , σt32 , σt42 we used squared differences from pairwise comparison of replicas calculated above s122 , s132 , s142 , s232 , s242 , s342: E[s122]=E[ ( Δft1−Δft2 ) 2]=σt12+σt22 E[s132]=E[ ( Δfb1+Δft1−Δfb2−Δft3 ) 2]=σb12+σt12+σb22+σt32 E[s142]=E[ ( Δfb1+Δft1−Δfb2−Δft4 ) 2]=σb12+σt12+σb22+σt42 E[s232]=E[ ( Δfb1+Δft2−Δfb2−Δft3 ) 2]=σb12+σt22+σb22+σt32 E[s242]=E[ ( Δfb1+Δft2−Δfb2−Δft4 ) 2]=σb12+σt22+σb22+σt42 E[s342]=E[ ( Δft3−Δft4 ) 2]=σt32+σt42 Therefore , the variance of mean frequency f can be found as: σ2=116∙ ( ( s132+s142+s232+s242 ) − ( s122+s342 ) ) Recalling that variance and squared differences are a function of frequency: σ2 ( f ) =116∙ ( ( s132 ( f ) +s142 ( f ) +s232 ( f ) +s242 ( f ) ) − ( s122 ( f ) +s342 ( f ) ) ) For each segment and time point we calculated the numerical function σ2 ( f ) . Then for each genotype having mean frequency fx we estimated its variance as σ2 ( fx ) We merged nucleotide genotypes that corresponded to the same amino acid sequence and summed their frequencies and variances . We filtered out all genotypes x that had any of following patterns: fxt0=0 , fxt1=0 , fxt2>0 or fxt0=0 , fxt1>0 , fxt2=0 or fxt0>0 , fxt1=0 , fxt2>0 . Fraction of such genotypes were <0 . 5% for all segments except S9 , for which it was 4 . 5% For further analysis , this amino acid dataset was used except when specified . Number of cells in a pool with particular genotype x after time interval t increases exponentially nxt=nx0∙Exp[sx∙t] , where sx is absolute fitness . Frequency of genotype x as well depends exponentially on absolute fitness with an additional multiplicative factor: fxt=nxtNt=nx0∙Exp[sx∙t]Nt=fx0∙Exp[sx∙t]Nt/N0 , where Nt and N0 are total cell numbers in a pool at time points 0 and t . Factor 1Nt/N0 reflects the total growth of population , it changes with time but is the same for all genotypes . Therefore , we can rewrite genotype frequency at time t as: fxt=fx0∙Exp[ ( sx−s0t¯ ) ∙t] , where s0t¯=1t∙Log ( NtN0 ) In the measured dataset for each genotype x we have 3 measurements of frequency fxt0 , fxt1 , fxt2 and their errors σ2 ( fxt0 ) , σ2 ( fxt1 ) , σ2 ( fxt2 ) . To estimate genotype fitness , we minimized relative squared errors of exponential fit as function of fitness sx and initial frequency fx0: ( sx , fx0 ) =argminsx , fx0 ( ( fxt0−fx0 ) 2σ2 ( fxt0 ) + ( fxt1−fx0∙Exp[ ( sx−s01¯ ) ∙t1] ) 2σ2 ( fx1 ) + ( fxt2−fx0∙Exp[ ( sx−s02¯ ) ∙t2] ) 2σ2 ( fx2 ) ) ( 1 ) This formula contains four parameters common for all genotypes from one segment: s01¯ , s02¯ , t1 , t2 . Further we will perform additional shifting and scaling of fitness values ( see next section ) , therefore , without loss of generality we could set s01¯=0 and t1 = 1 . Ideally , t2/t1 should equal 14; however , we noticed that this ratio does not hold for many segments and fitted k = t2/t1 from data instead of using value 14 . To find specific s02¯ and k for each segment we selected genotypes with high frequencies at t0 ( t0>25∙10−6 ) that corresponds to ~500-1000 reads per technical replicate . Each segment contains 103-104 genotypes that meet this criterion . We minimized eq . ( 1 ) for selected genotypes trying all possible combinations of ( s02¯ , k ) from a grid where s02¯ϵ[0 , 1 . 2] with step 0 . 01 and kϵ[1 , 14] with step 0 . 1 and choose ( s02¯ , k ) which gives minimal ( * ) . Finally , given ( s02¯ , k ) for each segment we found sx for each genotype . Errors for fitness values , sxstd , were estimated as standard error of best-fit parameter . For genotypes with frequencies pattern fxt0>0 , fxt1=0 , fxt2=0 fit of Eq ( 1 ) cannot be obtained . Therefore , we defined upper boundary for their fitness value as sxboundary=Log ( 1max ( N1t1 , N2t1 , N3t1 , N4t1 ) ) , where Nit1 , i=1 . . 4 are total read numbers at time point t1 in i-th replica . We scaled fitness such that lethal genotypes have fitness 0 and neutral genotypes have fitness 1 . We assumed that genotypes with a stop codon or frame shift are lethal . Thus , for each segment we linearly rescaled the fitness distribution so that 95% of genotypes with nonsense mutations have a fitness of 0 and so that the local maximum of the fitness distribution of genotypes with extant amino acids is 1 . The scaling around the local maximum led to the shift of fitness values of less than +/- 0 . 025 in each of the 12 segments compared to the measured wildtype strains and did not affect our results ( for scale , we called an amino acid change non-neutral if its effect on fitness was > 0 . 4 ) . All fitness values that became smaller than 0 were set to 0 . We used nucleotide synonymous sequences as an internal control . The error rate for a measurement of fitness of an amino acid sequence depends on the number of synonymous sequences , n , that were used to estimate it . Therefore , we estimated the false discovery rates separately for categories with n = 1 , . . 10 variants . For each amino acid genotype with more than n synonymous variants we merged random combination of n of its nucleotide genotypes and estimated fitness . We then calculated the difference between this fitness and the fitness of the corresponding amino acid sequence . We classified case as “false unfit” if difference was <-0 . 4 and as “false fit” if difference was >0 . 4 . The fraction of such cases gives us false discovery rates for genotypes having n synonymous variants . To get total false discovery rates for each segment we averaged “false unfit” and “false fit” rates for different n with weights equal to the fractions of genotypes in amino acid dataset which have n synonymous variant ( S2 Supporting Information ) . The high correlation between biological replicas ( S2 Supporting Information ) confirms high accuracy of our high-throughput experiments , with the exception of segment 9 . For each amino acid replacement , we calculated its fitness effect in different backgrounds . We estimated the fraction of backgrounds in which a replacement exhibits deleterious , beneficial and neutral effects . To get the fraction of backgrounds with neutral effects we utilized the approach of mixture distribution analysis from Sarkisyan et al . [10] . We assume that neutral replacements have the same distribution as the fitness effects of synonymous replacements and are caused by measurement noise . We then calculated the fraction of backgrounds in which mutations have a neutral effect as the overlap between the distribution of synonymous mutations and distribution of fitness effects across different backgrounds . In remaining backgrounds , the amino acid replacement was called to have a non-neutral effect . Among them we counted those with strong fitness effects , including deleterious mutations when the fitness effect was < -0 . 4 and beneficial when the fitness effect was > 0 . 4 . We concluded that a particular amino acid replacement exhibits a strong deleterious or beneficial effect in some background if the fraction of such backgrounds exceeded the false discovery rate . To predict the unidimensional fitness function based on additive contribution of extant amino acid states we used deep learning , a machine learning technique capable of constructing virtually any function , even with a simple neural network architecture [59] ( Fig 4B ) . To convert amino acid sequences into a binary feature matrix we used one-hot encoding strategy , in which each feature ( column in the matrix ) indicates the presence or absence of a particular amino acid state . For neural network implementation the TensorFlow library was used ( www . tensorflow . org/about/bib ) . To optimise the accuracy/overfitting ratio , we tested different combinations of neural network architectures and parameters . As a starting point , we selected a number of complex architectures , which describe our data but are prone to overfitting due to their large number of parameters . We then gradually reduced the number of layers and neurons to reduce the overfitting , while empirically controlling for accuracy . Our final architecture consists of three layers and 22 neurons in total ( Fig 4 ) . Each neuron performs a linear transformation of the input and subsequently applies a non-linear sigmoid activation function to the result . The output of the first layer is a single sigmoid of a linear transformation of the feature vector , i . e . σ ( c1Tx+b1 ) where x is the feature vector , c1 is the vector of coefficients , b1 is the bias and σ ( t ) = ( 1+e−t ) −1 . Looking forward , observe that c1Tx is a fitness potential of the genotype x ( see main text ) . The second layer decompresses the hidden nonlinear representation into 20 sigmoids , the combination of which is further linearly transformed with the only neuron of the third layer and wrapped into another sigmoid function: F ( x ) =σ ( ∑i=120c3 , i∙ ( c2 , i∙σ ( c1Tx+b1 ) +b2 , i ) +b3 ) In the formula above , c2 , i is the coefficient of the i-th neuron in the n-th layer , and b2 , i is the bias of the i-th neuron in the second layer ( the biases of the only neurons of the first and third layers are b1 and b3 , correspondingly ) . The key idea of our approach is that the number of neurons in the first layer of the neural network determines the number of linear combinations of mutations ( or fitness potentials ) used in order to predict the fitness of the variant . In other words , each neuron in the first layer assigns a single unique weight to every amino acid state in the dataset ( Fig 4 ) . Thus , the number of neurons in the first layer of the architecture is the dimensionality of epistasis in the model ( i . e . one in this case ) . The fitness potentials are then transformed by a nonlinear phase shift function constructed by the 22 neurons of the neural network . The simplicity of the architecture minimizes overfitting , which was further prevented by keeping 10% of the data as a test set ( in order to see how well the model performs on a fraction of the data it has never seen ) and early stopping ( training was stopped if the test accuracy did not improve for 10 epochs ) . The loss function that is being optimised is not convex , which leads to a high probability of getting stuck in different local minima . To ensure reproducibility , each of our models was constructed ten independent times using random train-test splits . The accuracies of the 10 constructed models varied by at most 2% . Each model was trained for under 100 epochs using mean squared error as the loss function . An adaptive learning rate method proposed by Geoffrey Hinton , RMSProp , was used as the optimiser [60] . This algorithm is a version of a mini-batch stochastic gradient descent , utilising the gradient magnitude of the recent gradients in order to normalise the current ones . All the weights were initialised using Xavier normal initialiser [61] . For analysis in Fig 7D , we first choose two fit “parental” genotypes , one randomly chosen genotype ( eg: ABE ) and the other parental genotype that is either S . cerevisiae wildtype genotype ( inter-segmental ) or another random fit genotype in the data ( intra-segmental ) ( eg: abe ) . The two genotypes in this example are Hamming Distance 3 apart ( HD = 3 ) . We next compute all ( 2HD-2 ) intermediate genotypes ( eg: AbC , aBc , et cetera ) and retain the subset that were experimentally measured . We represent the two parental genotypes and all measured intermediate genotypes as an undirected graph in which each genotype is a vertex . All genotypes one amino acid replacement apart are connected by an unweighted edge . The shortest possible path for a given pair of genotypes is of length HD . We find all shortest paths between the two parental genotypes using a breadth-first search . We next remove all vertices ( genotypes ) that are unfit , and recompute the number of shortest between the two parental genotypes . For example , in Fig 7A , there are six paths of length three if you take into account all genotypes , but only three paths of length three if you take into account only fit genotypes . For the analysis in Fig 7E , we first represent the two parental genotypes and all measured intermediate genotypes as an undirected graph in which each genotype is a vertex . All genotypes one amino acid replacement apart are connected by an unweighted edge . We can then compute the degree ( number of genotypes of distance one ) for each vertex ( genotype ) . We do so randomly drawing from all measured genotypes and using only unfit genotypes or using the same number but randomly chosen genotypes . For the randomly chosen genotypes , the value is the average over 1000 runs . For each amino acid replacement ( eg: C -> S at position 141 ) , we considered only those that exhibit a large fitness effect ( abs . difference > 0 . 4 ) comprising a set of amino acid replacements with large effects . For each amino acid replacement , we divided the genetic backgrounds into two categories: those in which the replacement caused a > 0 . 4 increase in fitness , and those backgrounds in which the replacement caused > 0 . 4 decrease in fitness . A single amino acid replacement can cause a large increase in fitness in some backgrounds and a large decrease in others due to two possible reasons: sign epistasis or experimental error . To differentiate the two cases , we identified secondary amino acid replacements that significantly alter the ratio of large increases to large decreases in fitness ( Fisher’s exact test , Bonferroni corrected p-value < 0 . 05 ) . We only consider a site to be under sign epistasis if there is a second site that alters the frequency of sign epistasis in a statistically significant manner , i . e . more frequently than expected by chance alone . We reconstructed ancestral amino acid states using maximum likelihood approach implemented in CODEML program of PAML 4 [62] . | An intuitive understanding of protein evolution dictates that , with the exception of adaptive substitutions , amino acid states should be freely exchangeable between the same gene from different species . However , the extent to which this assertion holds true has not been tested in a controlled experiment . Here , we show that whether an amino acid state can be exchanged between orthologues depends on other amino acid states in the same protein . Furthermore , we show that the mode of interaction of amino acid states is multidimensional . Assuming that amino acid replacements influence the protein in several independent ways substantially improves our ability to predict the effect of an amino acid state in a protein sequence that has not been observed in nature . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [] | 2019 | An experimental assay of the interactions of amino acids from orthologous sequences shaping a complex fitness landscape |
The extent and nature of genetic incompatibilities between incipient races and sibling species is of fundamental importance to our view of speciation . However , with the exception of hybrid inviability and sterility factors , little is known about the extent of other , more subtle genetic incompatibilities between incipient species . Here we experimentally demonstrate the prevalence of such genetic incompatibilities between two young allopatric sibling species , Drosophila simulans and D . sechellia . Our experiments took advantage of 12 introgression lines that carried random introgressed D . sechellia segments in different parts of the D . simulans genome . First , we found that these introgression lines did not show any measurable sterility or inviability effects . To study if these sechellia introgressions in a simulans background contained other fitness consequences , we competed and genetically tracked the marked alleles within each introgression against the wild-type alleles for 20 generations . Strikingly , all marked D . sechellia introgression alleles rapidly decreased in frequency in only 6 to 7 generations . We then developed computer simulations to model our competition results . These simulations indicated that selection against D . sechellia introgression alleles was high ( average s = 0 . 43 ) and that the marker alleles and the incompatible alleles did not separate in 78% of the introgressions . The latter result likely implies that most introgressions contain multiple genetic incompatibilities . Thus , this study reveals that , even at early stages of speciation , many parts of the genome diverge to a point where introducing foreign elements has detrimental fitness consequences , but which cannot be seen using standard sterility and inviability assays .
Explaining the present-day biological diversity requires an understanding of the speciation process . While we typically cannot observe speciation , we can ask to what extent are species incompatible if brought together to form hybrids . The founders of the Modern Synthesis typically argued that even the most recently diverged species accumulate enough fitness differences such that no large part of the genome can be shared between them ( e . g . [1]–[3] ) . This view of speciation argues that lots of loci with a wide range of effects on fitness should characterize the speciation process . E . Mayr championed a “genetic revolutions” version of this view , arguing that once separated from gene flow , most of the genome will undergo rapid coadaptive change , resulting in widespread fitness differences during speciation [3] , [4] . As a result , the Biological Species Concept ( BSC ) has historically emphasized the cohesiveness of the species , where most of the genome diverges as a single biological unit and the evolution of isolating barriers play a central role in protecting its “integrity” [2] , [3] , [5] . On the other hand , if adaptive functional divergence involves a limited number of loci , much of the genome could still penetrate across the species boundary during incipient stages of speciation . This is often described as the “genic view” of speciation and is argued to be especially applicable when speciation occurs with gene flow ( i . e . parapatric and sympatric modes of speciation , [6]–[8]; see Figure 1 in [6] ) . Recently , several studies have attempted to look for the so-called “genomic islands of speciation” ( e . g . [9]–[12] , see review in [13] ) . These assume that speciation with gene flow has occurred and that it will homogenize the genome except for a few genes involved in reproductive isolation and differential adaptation [6] , [8] , [14] . While earlier studies found support for the “islands” of speciation ( e . g . [9]–[11] ) , more recent comprehensive genome-wide screens are revealing a different picture [15]–[17] . Rather than having small genomic islands surrounded by mostly undifferentiated genomes , these incipient and sympatric races show widespread genomic differentiation , either being randomly distributed across the genome or clustered in the so-called “genomic continents” such as inversions or particular chromosomes ( see [15] for discussion ) . Other studies focus on identifying “speciation genes” that underlie reproductive isolation between closely related species ( see [5] for review ) . Historically , these studies have been interested in determining how many loci are involved in reproductive isolation [18]–[23] , and elucidating their identity and their evolution [24]–[31] . The great majority of these studies focus on the more easily measurable effects of sterility and inviability of hybrids . Many such sterility and inviability factors differentiating closely related species have been identified ( see reviews in [5] , pg . 302; [32] ) . While both approaches have made important contributions to understanding the genetics of speciation in nature , neither addresses the degree to which two genomes are genetically incompatible . Genome-wide scans show us the extent of sequence divergence across whole genomes , but they say nothing about whether these divergent sites carry fitness or functional consequences . Studies that search for speciation genes concentrate a priori on such effects as hybrid sterility and inviability , but ignore the rest of the genome for other fitness and functional differences between species . Perhaps genetic studies of natural hybrid zones and hybrid fitness come closest to estimating the true extent of genetic incompatibilities between incipient species ( e . g . [33] , [34] ) . Results from hybrid zones suggest that many fitness-related genes may differentiate genomes of even incipient races or recently diverged sibling species [33] , [35] , [36] . However , little has been done to determine whether these incompatibilities are associated with sterility or inviability effects or contain other fitness detriments . Further , the hybrid studies cannot identify specific genomic regions responsible for incompatibilities or determine the strength of selection associated with each of these genetic incompatibilities . Exploring these questions in a laboratory setting using genetic introgressions provides the best means to estimate the basic parameters of genetic incompatibilities on a genome-wide level . To approach this general question , the present paper focuses on recently diverged sibling species Drosophila simulans and D . sechellia . Molecular evidence indicates that they have diverged only about 250 , 000 years ago and thus represent fairly early stages of speciation [37] . For instance , these species have accumulated partial , but incomplete premating isolation and still produce fertile hybrid females in F1 and subsequent generations [18] , [38] . These sibling species have most likely speciated allopatrically; D . simulans likely evolving on the African continent , while D . sechellia has remained an island endemic to the Seychelles archipelago in the Indian Ocean [39] , [40] . Today , both species can be found in the Seychelles archipelago , but seem to occupy different islands [39] . Thus , we address our main question about genome-wide incompatibilities in a relatively young pair of taxa where whole-genomes were likely able to diverge without being impeded by substantial gene flow . To determine the extent and nature of genetic incompatibilities between D . simulans and D . sechellia , we have introgressed random genetic segments from D . sechellia into a D . simulans genome . We first ask if these random introgressions contain measurable sterility and/or inviability effects . If some of these introgressions do not show sterility or inviability , we can then ask whether these regions are selectively neutral upon introgression or whether they carry other deleterious fitness effects after long-term genetic competition experiments . If these random genomic introgressions turn out to be selectively neutral , this would indicate that genomic incompatibilities are typically restricted to previously described genes associated with such effects as sterility and inviability ( e . g . see [32] ) . However , if we find that most introgressions placed into a foreign genetic background experience strong fitness reduction and are selected out of the host population , it would imply that we are fundamentally underestimating the extent and possibly the type of fitness differences that accumulate between species . Thus our paper highlights the need to incorporate competition and other selection experiments to accurately test theories related to “genomic islands of speciation” .
The present study utilized 12 recombinant introgression lines ( RILs; henceforth referred to as “introgression lines” for short ) from Stuart J . Macdonald , Isabel Colson and David B . Goldstein ( Oxford University ) . Briefly , each line was made by genetically introgressing D . sechellia chromosomal fragments into a D . simulans genetic background ( for a detailed description of the construction of these lines see Materials and Methods ) . The introgressions were made homozygous by single-pair sib-mating for 18 generations , and 41 microsatellite genetic markers across X , 2nd , and 3rd chromosomes were used to map the regions of the sechellia introgressions . As those introgression lines have been maintained in Goldstein's laboratory for several years , we therefore tested whether each line was homozygous for the expected sechellia introgression ( henceforth referred to as “confirmed lines” ) or whether it did not contain the sechellia marker allele ( henceforth referred to as “unconfirmed lines”; see Figure S1 ) . The latter lines may have lost the introgression by stochastic or other processes during their years of maintenance . In total , 9 lines were confirmed to carry sechellia introgressions and 3 lines failed to show introgressions . To test whether the created introgression lines had any obvious inviability and/or sterility factors , we assayed overall fertility of each introgression line and compared it to the fertility of the experimental simulans strain that was used as the genetic background of introgressions ( Table 1 ) . Our results showed that while the introgression experiment clearly increased the variance in fertility among introgression lines ( one-way ANOVA: F = 9 . 05 , d . f . = 11 , p<0 . 0001 ) , the average fertility among lines was nearly identical to that of the experimental simulans strain ( Table 1 ) . Further , there was no evidence of significant fertility reduction in any of the lines studied using the posthoc Tukey-Kramer HSD test ( Table 1 ) . There was also no trend in fertility reduction among our introgression lines , with five out of the eight tested introgression lines actually having higher fertility than the experimental simulans strain ( Table 1 ) . Similarly , crosses between different introgression lines and between their F1 progeny either resulted in non-significant differences in fertility from the simulans strain or higher fertility relative to the simulans strain ( see Tables S2 and S3 ) . Therefore , we conclude that the present sechellia alleles placed in a simulans genetic background did not generate any detectable inviability and/or sterility effects . We can begin to address our main question as to whether these sechellia introgressions are equally fit to wild-type simulans alleles in a simulans genetic background . To test whether the introgressions had any other deleterious fitness affects , we set up 6 independent competition experiments to determine the evolutionary fate of sechellia alleles in a simulans genetic background . For each competition experiment , we crossed two different introgression lines ( see Methods for details ) . Combining the introgression lines together allowed us to control for any non-intentional effects of the introgression procedure ( e . g . to control for different levels of inbreeding ) . Competition crosses were of two types: The first set of experiments crossed a line containing a single confirmed introgression with another line that did not show evidence of the introgression . Thus in this experiment , sechellia alleles were competing with the wild-type simulans alleles at only a single genomic region ( henceforth referred to as “single-introgression experiment”; shown in Figure 1A–1C as black blocks ) . The second type of experiment crossed two lines , each containing a unique confirmed introgressed region on either the 2nd or 3rd chromosome ( henceforth referred to as “double-introgression experiment”; Figure 1D–1F ) . This allowed us to see if the introgressions on the 2nd and 3rd chromosomes interact when each competed against the wild-type simulans alleles ( see below ) . Our competition experiments revealed highly unexpected results based on the above lack of difference in fertility between introgression lines and the wild-type simulans line . We found that all of the sechellia marker alleles sharply decreased in their frequencies relative to the wild-type simulans alleles by generations six and seven ( Figure 2 ) . We found that from the starting 50% frequency , the sechellia marker alleles dropped to a range of 38% to 17% among different experiments . After this initial drop , the frequencies of sechellia alleles either: 1 ) kept further declining , 2 ) remained relatively unchanged or 3 ) actually increased over time in subsequent generations . This striking observation resulted in several conclusions . First , it showed that introgressed segments of sechellia into a simulans background does indeed carry strong deleterious fitness consequences . Second , it showed that these fitness effects cannot be detected by standard fertility measures above and were only revealed through long-term competition experiments . Third , it indicated that the marker alleles we were tracking either remained genetically linked to the deleterious alleles at surrounding fitness loci or became independent over time from these deleterious alleles due to recombination . We then tested whether the frequency declines of sechellia alleles are affected by having one or two confirmed introgressions during the competition experiment ( i . e . single introgressions versus double introgressions ) . Because the two sechellia introgressed segments are on different chromosomes , these are expected to assort independently during competition . We found that particular sechellia marker alleles that were either in the presence of a single or a double introgression had nearly identical frequency declines after six or seven generations ( t test: single avg . freq . gen . 6/7 = 0 . 284 , double avg . freq . gen . 6/7 = 0 . 276; F = 0 . 048 , p = 0 . 866; see also Figure 2 ) . Similarly , after 20 generations of competition , both types of introgression designs showed very similar sechellia marker frequencies ( t test: single avg . freq . gen . 20 = 0 . 158 , double avg . freq . gen . 20 = 0 . 214; F = 1 . 58 , p = 0 . 216; Figure 2 ) . Thus we did not detect any significant differences between single and double introgressions on fitness . Finally , we tested whether there is linkage disequilibrium between 2nd and 3rd chromosome marker alleles in double introgression experiments ( experiments D , E , and F in Figure 1 ) . Except for few cases in experiment D , experimental populations did not deviate significantly from linkage equilibrium ( p>0 . 05; Table S1 ) . Thus using this approach we failed to detect evidence for epistasis between 2nd and 3rd chromosome sechellia introgressions . In total , these results suggest that the observed fitness reduction during competition is not a consequence of combining two introgressions on 2nd and 3rd chromosome together . This implies that each sechellia segment is negatively epistatically interacting with the simulans genetic background on its own . To estimate the intensity of selection against sechellia alleles and the recombination rate between the marker and the surrounding fitness loci , we performed multiple-generation , computer simulations using maximum likelihood approaches . We assumed that each microsatellite marker is neutral and is linked at a recombination distance of c to a single deleterious allele with selection coefficient s and dominance h . All other aspects of the competition experiment , such as experimental population sizes , recombination only in females , etc . , were simulated accordingly ( see Materials and Methods for details ) . Because our main interest is to estimate c ( recombination rate; ranging from 0 to 0 . 5 ) and s ( selection against sechellia allele; ranging from 0 to 1 ) , we manipulated the dominance parameter , h . These estimates were not meant to precisely estimate s and c , but to give an idea of the scale of these values necessary to produce the observed marker allele frequency declines . The h parameter was assumed to equal either: 0 , 0 . 5 , 0 . 9 or 1 . Thus , we allowed sechellia allele to become increasingly dominant over the simulans allele from complete recessivity ( h = 0 ) to complete dominance ( h = 1 ) . In reality , it is not known which dominance best characterizes the sechellia-simulans allelic relationship , but as we will see below , our results are robust to changes in the dominance parameter . Maximum likelihood estimates of s and c were obtained by comparing the observed D . sechellia marker frequencies to computer-generated distributions based on simulations of introgression lines . Table 2 summarizes the simulation results based on contour plots in Figure S2 . Interestingly , we found that 7 of the 9 ( 78% ) maximum likelihood estimates of c have values that are very close to 0 , corresponding to very small physical distances ( Table 2 ) . This result has two possible interpretations: First , it may imply that the marker and fitness locus happen to be very close to each other in 78% of the experiments . Second , rather than a single fitness locus per segment ( as our simulation model assumed ) , the introgressed segments may be carrying multiple fitness loci with deleterious effects , thus preventing the single marker from recombining away from multiple deleterious interactions . Given that our markers were chosen randomly and that the sechellia segments are fairly large ( Figure 1 , Figure S1 ) , the chance that each randomly chosen marker locus happened to be so close to a single fitness locus with a deleterious effect seems very low . Instead these results most likely suggest that the introgressed sechellia segments probably carry multiple deleterious fitness alleles in a simulans background . Table 2 also shows that varying the dominance parameter , h , does not change the major results of the simulations , with essentially presence or absence of positive recombination across different lines . Finally , our computer simulations revealed that selection coefficients against sechellia alleles must be strong in order to explain the observed evolutionary changes ( Table 2 ) . On average , the selection intensity against sechellia alleles was s = 0 . 43 with a range of 0 . 28 to 0 . 65 . It can also be seen that the estimated selection coefficients were negatively correlated with the dominance of sechellia alleles ( R2 = 0 . 89 , F = 27 . 4 , p = 0 . 034 ) . This result is in general agreement with expectation of Haldane' sieve [41] , since if alleles are more recessive , in order to explain the observed frequency declines , they must have stronger selection coefficients ( note however that we are dealing with negative selection rather than positive as in [41] ) . However , even under completely dominant assumption , the selection strength against sechellia alleles is on average still high ( s = 0 . 37; Table 2 ) . In total , our simulations indicated that multiple incompatibilities likely exist within the great majority of our introgressed segments and that these factors have substantial negative fitness consequences that cannot be detected by standard fertility tests above . Determining exactly why sechellia alleles declined in frequency in our competition experiments is beyond the scope of this paper . However , we did perform one additional experiment focusing on whether introgression lines have reduced mating success relative to the original simulans strain ( see Materials and Methods for details ) . These results showed that individuals ( combined males and females ) from 6 out of 8 ( 75% ) introgression lines did indeed have lower relative mating success compared to individuals from the simulans strain ( Table 3 ) . While suggestive , this result is not statistically significant ( sign test: one-tailed p = 0 . 14 ) . On average , simulans individuals comprised 53% of the total matings relative to 47% of the introgression individuals , which did turn out to be slightly significant ( Wilcoxon test: χ2 = 6 . 4 , p<0 . 011 ) . It is particularly the introgression males that are strongly outcompeted by simulans males ( a 12% differential in fitness; Wilcoxon test: χ2 = 10 . 6 , p<0 . 0011 ) . Introgression females have the same mating success as simulans females ( Wilcoxon test: χ2 = 0 . 03 , p = 0 . 87; Table 3 ) . Unfortunately , performing such an experiment does not allow us to adequately control for different overall levels of inbreeding between our introgression lines and our simulans line , a factor known to influence mating behavior in Drosophila ( e . g . [42] ) . Thus , presently , we cannot conclude that mating behavior differences were responsible for the observed inferiority of sechellia alleles in a simulans background ( see Discussion for additional possibilities ) .
Adaptive evolution within species largely rests on the basic parameters of genetic architecture of fitness-related traits [3] , [46] , [47] , [48] , [49] , [50] . Such parameters as the level of genetic interactions ( epistasis ) , the number of genes and their effects and the pleiotropic byproduct of genes will determine how much fitness and functional divergence is expected between species . If most phenotypes and developmental systems are governed by complex genetic architectures , whose genes are organized into epistatic networks that also have pleiotropic effects , we would expect that even incipient species would exhibit a multitude of fitness and functional differences between their genomes that cannot be easily broken down by subsequent gene flow [3] , [6] , [18] , [51] , [52] . This highly co-adaptive view of speciation was strongly favored by E . Mayr who even suggested that speciation will sometimes lead to veritable “genetic revolutions” due to the large-scale reorganization of allelic selective pressures as a result of new independent mutations and a change in epistatic interactions between new and existing alleles in each isolated population [3] . However , if most fitness-related traits and developmental systems are governed by few loci of additive and non-pleiotropic major effect , then it is conceivable that incipient speciation would only involve a handful of divergent loci with the rest of the genome being highly penetrable to gene flow [6] . The fact that we observed genetic incompatibilities with every random genetic introgression from D . sechellia into D . simulans suggests that the genetic basis of speciation is likely to be highly polygenic and epistatic between these young species . Our competition results are particularly striking because we showed that while these introgressions are viable and fertile on their own , they nevertheless rapidly decline in frequency when they compete against wild-type alleles for multiple generations . We studied two obvious components of fitness that could have been potentially involved in the inferiority of D . sechellia introgressions . These included both premating ( mating success ) and postmating ( fertility ) assays in our introgression lines ( D . simulans background+D . sechellia introgressed segment ) relative to the experimental D . simulans strain . Our results did not detect significant fertility effects of introgression since we initially showed that fertility is not lower in the introgression lines compared to the D . simulans strain . This finding indicates that the observed competitive exclusion of D . sechellia introgressions is unlikely a result of “weak” sterility and/or inviability factors since these would have generated lower fertility in introgression lines . Therefore , the cause of D . sechellia introgression inferiority is likely to be in other components of fitness . We also used multiple-choice mating trials to assess relative mating success of introgression lines against D . simulans strain . While individuals from introgression lines had a tendency to have lower mating success compared to the D . simulans line , this trend was not significant . Moreover , we could not control for inbreeding effects on mating success with this approach . Taken together , these assays could not identify a clear mechanism by which D . sechellia alleles were outcompeted from the D . simulans genetic background in our experiments . At this point we can only speculate that other as of yet unknown aspects of fitness particularly involved in soft-selection or competitive ability must be responsible for these fitness incompatibilities between these genomes . What is presently unclear is which biogeographical conditions of speciation will facilitate the rapid accumulation of genetic incompatibilities . In our work we have shown that fitness incompatibilities are fairly extensive between 250 , 000 year old allopatric sibling species . Because these species most likely diverged in allopatry , their genomes are expected to have accumulated incompatibilities at more or less homogeneous rates over time without much gene flow [6] . Will younger sibling species also show similar patterns ? Will parapatric or sympatric modes of speciation favor a more limited accumulation of genetic incompatibilities than what we have observed ? While earlier studies of sequence divergence using small number of markers generally found “genomic islands of speciation” ( e . g . [9] , [11] ) , more recent analyses of incipient parapatric and sympatric forms show more extensive sequence differentiation [15]–[17] . However , it is still largely unknown whether any of these sequence differences will translate to fitness divergence and genetic incompatibilities ( but see [15] ) . Future work will gain further insights into the evolution of genetic incompatibilities by extending our genetic competition experiments to even more incipient cases of speciation and those that have likely speciated with gene flow . This appears to be a more accurate way to assess which view of speciation is likely to be correct . It will also determine under which circumstances extensive genetic incompatibilities accumulate between two genomes . Follow-up studies may also reveal the causes of non-sterility and non-inviability genetic incompatibilities that are likely to be observed in such long-term competition experiments .
The recombinant introgression lines ( RILs ) were kindly provided by Stuart J . Macdonald , Isabel Colson and David B . Goldstein ( Oxford University ) . The construction and genotype checking of these introgression lines are briefly described here . D . simulans females from the “sim132” ( European Drosophila Stock Centre , Umeå ) line were crossed to D . sechellia males from the “sec S9” ( Mid-America Drosophila Stock Center ) , and the resultant F1 females were backcrossed to D . simulans males . The subsequent F2 males were individually crossed to either three simulans females ( P cross , P ) or three F1 females ( H cross , H ) and further made homozygous by single-pair sib-mating for 18 generations ( SJ Macdonald , pers . comm . ) . Figure S1 illustrates the genotype for each introgression lines based on the information provided by SJ Macdonald . In total , 41 microsatellite markers , i . e . , 8 , 16 , and 17 markers on the X , 2nd and 3rd chromosomes respectively , with an average interval of about 8 cM [53] are used in the initial genotyping . There are much fewer introgression fragments with smaller sizes on the X chromosome compared to the two autosomes ( SJ Macdonald , pers . Comm . ) . Only 3 of the 12 lines ( 6H , 16H , and 94P ) carry a small X chromosomal introgression ( Figure S1 ) . We therefore focused on the two autosomes for the competition experiments . Before all experiments , we genotyped these 12 introgression lines by using one microsatellite marker per introgressed segment and found 9 lines ( 6H , 12H , 25H , 28H , 60H , 29P , 62P , 78P , and 94P ) showed the expected sechellia alleles ( these lines are referred to as “confirmed lines”; see Figure S1 for specific location of each marker in each confirmed line ) . The other three lines ( 16H , 37P , and 129P ) showed no evidence of sechellia alleles at the genotyped locus ( Figure S1 ) . Nevertheless , these lines may still carry some parts of the sechellia introgression that could not be assessed by our genotyping . Therefore we will refer to the latter three lines as “unconfirmed lines” . To see if these introgressions had any obvious viability and/or sterility effects , we assayed the overall fertility of each introgression line relative to the original wild-type D . simulans line without introgressions . This was done by measuring the number of offspring produced by each introgression line and comparing it to the fertility of the D . simulans strain . All fertility assays were performed by setting up 10 replicates of three pairs of males and females in small vials for each tested line . We allowed the mating pairs to lay eggs for 15 days , at which point all adults were cleared . We then counted the number of F1 progeny to determine fertility . To test for significance , we first confirmed that the fertility data did not significantly deviate from normal distribution using a Goodness of Fit test ( Shapiro-Wilk test: W = 0 . 987 , p = 0 . 8666 ) . We then analyzed the whole dataset using a one-way ANOVA . To determine which specific introgression lines were significantly different from each other and from the wild-type simulans strain , we used a Tukey-Kramer HSD test that takes into account multiple testing . All tests were performed in JMP software ( SAS ) . To determine if there were any other fitness effects of sechellia alleles in a simulans genetic background , we performed a multi-generational competition experiment lasting twenty generations . We set up six independent competition crosses between different introgression lines: ( A ) 37P×25H , ( B ) 78P×16H , ( C ) 6H×129P , ( D ) 62P×29P , ( E ) 94P×28H and ( F ) 12H×60H . Combining the introgression lines together allowed us to control for any non-intentional effects of the introgression procedure ( i . e . all lines entering the competition experiment went through the same introgression procedure ) . The detailed procedures of the cross are as follows: ( 62P×29P as an example ) : 50 virgin females of 62P were crossed to 50 males of 29P and 50 virgin females of 29P were crossed to 50 males of 62P . The resultant F1 progeny of the two bottles were mixed and allowed to lay eggs to produce a large number of F2 progeny , which were transferred to five bottle replicates . There were approximately 300–500 flies in each bottle . The flies were allowed to lay eggs for 4 days and then collected in 100% ethanol . For the next generation , when enough flies ( 300–500 ) emerged , they were transferred to a new bottle . The same procedure applied to other crosses . In total , we set up five replicates for each one of the six distinct populations . The population sizes were kept at 300–500 for each replicate . All experiments were done at 22±1°C with a 12 hr–12 hr light–dark cycle . Samples of around 40 flies were taken from each bottle at generations 6 or 7 , 14 , and 20 . For each introgression segment , we examined one microsatellite marker in that region . The microsatellite markers are: A: DMU25686 ( cytological position: 93F14 ) ; B: DRODORSAL ( 36C8 ) ; C: DROGPAD ( 47A9 ) ; D1: AC005732 ( cytological position 24C9 ) ; D2: DMRHO ( 62A2 ) ; E1: DMMP20 ( 49F13 ) ; E2: DMCATHPO ( 75E1 ) ; F1: DS00361 ( 54B5 ) ; F2: DMU43090 ( 99D5 ) [53] , [54] . From the genotyping results , allele frequencies were calculated for each bottle replicate and for whole experiment sets . We developed an individual-based computer simulation model of our competition experiments performed above . The purpose of the simulation was to manipulate the presumed selection pressures against sechellia alleles relative to simulans alleles , the dominance of sechellia alleles' fitness effects and the recombination rate between selected loci and marker loci . The goal was to determine which combination of selection pressures and recombination rates was best at explaining our observed results . The simulations tracked either D . simulans or D . sechellia alleles at both the marker loci and the selected loci for each individual in the population . Each simulation involved 10 , 000 iterations with fixed values of selection coefficient ( s ) , recombination distance ( c ) , and dominance ( h ) . Each iteration ran for 20 generations with population sizes fixed at 150 males and 150 females each generation . Each generation was divided into the following stages: selection , recombination ( females only ) , and reproduction . In the selection stage: individuals homozygous for the D . simulans allele all survived , heterozygous individuals survived with probability = hs , and individuals homozygous for the D . sechellia alleles survived with probability = s . Reproduction began with recombination in females which occurred with probability = c . During random mating , male and female haplotypes were randomly selected from the population to make 150 males and 150 females for the next generation . Sampling took place in generations 7 , 14 , and 20 where 20 males and females were removed from the population after selection and before random mating . The allele frequencies of the D . simulans alleles at the marker loci were recorded from these sampled flies . The simulation ran for 10 , 000 iterations . Distributions were created for each of the three sampled generations ( 7 , 14 , and 20 ) . The allele frequencies were assigned to one of 40 bins ( 0– . 025 , . 025– . 05 , … , . 975–1 ) , and bin counts were incremented for each iteration . Observed allele frequencies for each marker were then compared to the three distributions . For a given generation , all of the bins containing observed frequencies were added together and divided by the total number of iterations . The log of this ratio was treated as a likelihood estimate for the parameters s , c , and h for that marker . Simulations were run for all combinations of c = 0 . 0001 , 0 . 001 , 0 . 01 , 0 . 1 , 0 . 2 , 0 . 3 , 0 . 4 , and 0 . 5; s = 0 . 1 , 0 . 2 , 0 . 3 , 0 . 4 , 0 . 5 , 0 . 6 , 0 . 7 , 0 . 8 , and 0 . 9; and h = 0 , 0 . 5 , 0 . 9 , and 1 . The Maximum likelihood estimate for each marker was the set of s , c , and h that yielded the highest likelihood value . Linkage disequilibrium ( LD ) analyses were carried out for double introgression experiments ( experiments D , E , and F in Figure 1 ) . The null hypothesis is that genotypes at one locus assort independently from genotypes at the other locus . The exact test for the LD was performed by using M . Raymond and F . Rousset's GENEPOP software package ( http://genepop . curtin . edu . au/genepop_op2 . html ) . We performed this probability test using a Markov chain with parameters of dememorization number = 1000 , number of batches = 100 and number of iterations per batch = 1000 . To assay whether introgressed segments caused a reduction in the mating success of their individuals relative to wild-type D . simulans genotype , we applied a multiple-choice mating experiment design similar to [55] . All flies were fed red or green colored food 14–18 hours prior to the experiment . The food coloration alternated between replications and had no effect on D . simulans mating choice ( data not shown ) . Experiments were started within one hour after the beginning of the light cycle and conducted at 22±1°C . Sixty 4-day-old virgin flies from each sex of D . simulans S132 line and the introgression line were simultaneously released into a Plexiglas cage ( 14 . 5″ L×10″ H×9 . 5″ W ) with fly food in a 14-cm diameter Petri dish . The copulating pairs were aspirated out of the cage for identification by the food coloring in the guts . We let the mating trials run for 1 hour or until 60 matings ( 50% of possible copulations ) had occurred , whichever came first ( as recommended by [56] to avoid bias ) . Several comparisons were replicated at least three times to determine overall reliability in the mating behavior . We then calculated the percentage of matings by D . simulans individuals relative to introgression line individuals and also the relative percentage of matings by each sex of each line . | Determining the extent of genomic incompatibilities is a pivotal issue in understanding the process of speciation . A controversial topic that has recently sparked debate is whether there are few isolated genetic regions ( so-called “genomic islands of speciation” ) or extensive genetic regions ( “genomic continents of speciation” ) responsible for species divergence . To answer this question , most work has focused on species divergence at the DNA sequence level . Here , we present a new perspective by shifting the focus to the fitness and functional aspects of foreign genomic introgression . To illustrate our point , we performed an introgression experiment on two sibling species , D . sechellia and D . simulans . After introgressing random genomic segments of D . sechellia into D . simulans genetic background , a 20-generation competition experiment revealed that , even at the early stages of speciation , there are virtually always detrimental fitness consequences to introducing random foreign elements from one genome to another . This implies that incipient speciation may be characterized by widespread accumulation of genomic incompatibilities rather than a few isolated genes . This study shows that we should move beyond the sterility and inviability assays in order to understand the full extent of genetic incompatibilities during speciation . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"evolutionary",
"processes",
"population",
"genetics",
"biology",
"evolutionary",
"biology",
"genomic",
"evolution",
"evolutionary",
"genetics",
"evolutionary",
"theory"
] | 2012 | Incompatibility and Competitive Exclusion of Genomic Segments between Sibling Drosophila Species |
A major goal of evolutionary developmental biology ( evo-devo ) is to understand how multicellular body plans of increasing complexity have evolved , and how the corresponding developmental programs are genetically encoded . It has been repeatedly argued that key to the evolution of increased body plan complexity is the modularity of the underlying developmental gene regulatory networks ( GRNs ) . This modularity is considered essential for network robustness and evolvability . In our opinion , these ideas , appealing as they may sound , have not been sufficiently tested . Here we use computer simulations to study the evolution of GRNs' underlying body plan patterning . We select for body plan segmentation and differentiation , as these are considered to be major innovations in metazoan evolution . To allow modular networks to evolve , we independently select for segmentation and differentiation . We study both the occurrence and relation of robustness , evolvability and modularity of evolved networks . Interestingly , we observed two distinct evolutionary strategies to evolve a segmented , differentiated body plan . In the first strategy , first segments and then differentiation domains evolve ( SF strategy ) . In the second scenario segments and domains evolve simultaneously ( SS strategy ) . We demonstrate that under indirect selection for robustness the SF strategy becomes dominant . In addition , as a byproduct of this larger robustness , the SF strategy is also more evolvable . Finally , using a combined functional and architectural approach , we determine network modularity . We find that while SS networks generate segments and domains in an integrated manner , SF networks use largely independent modules to produce segments and domains . Surprisingly , we find that widely used , purely architectural methods for determining network modularity completely fail to establish this higher modularity of SF networks . Finally , we observe that , as a free side effect of evolving segmentation and differentiation in combination , we obtained in-silico developmental mechanisms resembling mechanisms used in vertebrate development .
A major goal of evolutionary developmental biology ( evo-devo ) is to understand how multicellular body plans of increasing complexity have evolved , and how the underlying developmental programs are encoded in the genome and gene regulatory network ( GRN ) . Modern evo-devo research shows more and more that a shared developmental toolkit of signaling , adhesion and transcription factor genes are essential for the development of organisms ranging in body plan complexity from cniderians to arthropods and vertebrates [1]–[3] . Therefore the current paradigm is that body plans of increasing complexity are the result of increases in the complexity of regulation of this similar set of genes [1] , [2] , [4]–[9] combined with increases in the number of variants of certain developmental toolkit genes [10]–[12] . As a consequence , a strong focus in current evo-devo research is on changes in spatio-temporal gene expression patterns and the differences in architecture of the developmental networks generating them . Network characteristics that are considered key for the evolution of increasingly complex body plans are modularity , robustness and evolvability . It is frequently argued that developmental GRNs are typically modular , i . e . that different functions are performed by largely independent network parts [2] , [13]–[16] , and that this is the key property responsible for both network robustness and evolvability . The idea is that modularity reduces pleiotropy , allowing for the malfunctioning of or tinkering with network parts involved in one function without producing failure in other functions [2] , [13]–[16] . Although this reasoning sounds appealing and intuitively correct , little research has been done to explicitly test the roles and relationships of developmental network modularity , robustness and evolvability in the evolution of complex body plans . Indeed , we argue that it is currently unclear how modular developmental networks are , how such modularity evolves , and how this modularity looks . Today , only a limited number of developmental GRNs have been studied in considerable detail . These studied networks are mostly involved in the patterning of a single organ or developmental phase , without detailed knowledge on their relationships with the rest of the developmental network [17]–[22] . As a consequence , although these networks have often been claimed to be modular , it is currently hard to fully assess the modularity of developmental networks . Based on theoretical studies it has been argued that evolution should neither be expected to produce nor to preserve architectural modular networks . This follows from the fact that modular networks form only a small subset of the possible network architectures capable of performing a particular function [23] . Indeed , theoretical studies aimed at investigating the evolution of architecturally modular networks have had to use quite specific fitness targets to obtain modular networks [24]–[27] . On the other hand , it has previously been shown for other genome [28] , [29] and network [30] properties that these may arise as a neutral side effect of the mutational dynamics rather than requiring an adaptive explanation . Similar suggestions have been made for network modularity [31] , [32] . With regards to the appearance of modularity , note that in its most general sense network modularity is defined fairly functional -different functions are performed by largely independent network parts- but is currently most frequently measured entirely architectural -different modules of genes that are more densely connected with genes within the module than genes in different modules [33]–[35] . However , it is recently being suggested that functional or dynamic rather than architectural network modularity may be most relevant for network functioning and evolution [36]–[40] . Note that architectural and functional modularity do not necessarily overlap . This might mean that different , more functionally oriented methods to measure modularity are needed [36] , [37] , [40] . Recently , several such methods have been proposed , among which clustering of genes with similar expression in network attractors [36] , or with similar knockout effects [40] , or with a function in the same specific process [37] . Thus , currently both the extent and shape of developmental network modularity remain unclear . In addition , it is not well known whether evolution of this modularity requires selection for robustness or evolvability or arises neutrally . The goal of the current study is to use computer simulations to investigate what type of network architecture and properties evolve during the evolution of complex body plan patterning . This will allow us to check to what extent evolved developmental networks are modular , whether network modularity is related to increased robustness and evolvability , and what exactly network modularity looks like . In our simulations we select for segmented and differentiated body plans . Segmentation and extensive anterior posterior domain differentiation are considered key innovations of the bilaterian clade , and have been extensively studied both experimentally and theoretically . This will allow us to compare our in-silico evolved developmental networks with actual biological patterning networks and results of previous simulation studies . Furthermore , by independently selecting for segmentation and domain formation we enhance the chances for modular networks to evolve . We study the different types of evolutionary trajectories that arise , and compare them with respect to network robustness , evolvability and modularity and the type of developmental mechanism they produce . Quite interestingly , we find that there are only two distinct evolutionary strategies to evolve a segmented and differentiated body plan , each resulting in a distinct developmental mechanism . In one strategy , first most segments and only then domains evolve ( SF strategy ) , while in the other segments and domains evolve more or less simultaneously ( SS strategy ) . In addition , we show that in the SF strategy , a complex time transient is responsible for domain differentiation , while a genetic oscillator produces regular body segments . In contrast , in the SS strategy , a complex time transient generates both the body segments and domains . We find that imposed indirect selection for robustness causes the SF strategy to evolve much more frequently than the SS strategy . Furthermore , the SF strategy was also found to be more evolvable . The different types of expression dynamics involved in segmentation and domain formation , together with the larger robustness and evolvability of SF networks suggests that they may also be more modular . However , frequently used , purely architectural modularity scores suggest that the two network types are equally non-modular . Pruning of non necessary network parts that potentially obscure architectural modularity did not change these results . Furthermore , changing model parameters such that less densely connected networks evolve also did not produce architecturally modular networks . Therefore , we also used a more functionally oriented method . Specifically , we take into account the fact that the networks generate both segments and domains and investigate whether or not there are relatively independent network parts responsible for these two processes . Using this approach we could demonstrate that while SS networks generate segments and domains in an highly integrated manner , SF networks generate segments and domains in a more modular manner . Our results show that evolved developmental networks are not necessarily highly modular , robust or evolvable . However , upon significant selection for robustness , networks that are more modular , robust and evolvable will dominate . Our results thus confirm the relationship between modularity , robustness and evolvability . Our results also show that the type of modularity that evolved could not be detected by frequently used , automated , purely architectural algorithms , but required a more functionally oriented method . Beslon recently reported similar results [40] . Importantly , these results suggest that for the detection of biologically meaningful modularity purely architectural methods are less suitable and approaches ( also ) taking into account network dynamics and function should be preferred . Intriguingly , we find that the patterning mechanism employed by our SF networks shares key characteristics with vertebrate somitogenesis and axial patterning , even though this was not a specific aim of our study or explicit part of our model design .
Briefly , we use an individual based , spatially embedded model of a population of evolving embryo-organisms ( Figure 1 ) . The organisms consist of a one dimensional row of 100 cells , similar to the approach followed in [41]–[43] . The organisms have a genome that contains genes coding for transcription factors ( TFs ) and upstream regulatory regions with transcription factor binding sites ( TFBS ) [44] , [45] . Genes have a certain type , indicated with a number ranging from 0 till 15 . There can be multiple genes of the same type . The gene types can be subdivided into a few functional categories . Gene type 0 is a maternal gene . Its expression is not controlled by the organism , but instead is imposed to give rise to a morphogen wavefront . This wavefront moves from the anterior to the posterior of the embryo , switching the expression from gene type 0 from a level of 100 to 0 . Gene type 5 is a gene that the organisms can use to indicate the boundaries of body segments . Differential expression of gene types 8 till 15 can be used to subdivide the body into functionally different regions ( domains ) . Finally , gene types 1 till 4 , 6 and 7 are general transcription factors . By assigning gene type 5 to segmentation and gene types 8 till 15 to differentation the evolving segmentation and differentiation processes are not forced to be coordinated but can in principle use completely disjoint sets of genes . The genome codes for a gene regulatory network , with genes corresponding to nodes , and TFBS defining the activating and repressing influence of genes on each other . These regulatory links have a non-evolving impact strength of +1 or −1 , respectively . At the beginning of development , gene expression in each cell of an organism is initialized with gene types 1 to 4 having an expression level of 100 and all other genes having an expression level of 0 . Subsequent gene expression dynamics and protein levels are governed by the GRN and are modeled with ordinary differential equations using a similar approach as in [41] . The gene expression pattern present at the end of development is used to determine the number of segments and domains the body is patterned in . A segment boundary is defined as a position in space where the expression of the segmentation gene switches from a high to a low level or vice versa . A domain is defined as a region in space where cells express the same combination of differentiation genes at a high level . The minimum length for a segment and domain is 7 cells , allowing for a maximum number of 14 segments and domains . To ensure stable differentiation , we compare gene expression at the end of development with that 20 time steps before . For each cell that has different gene expression levels at these two time points a fitness penalty is applied . In addition , to prevent excessive genome growth small fitness penalties are applied for each gene and TFBS present in the genome ( See Table S1 in Text S1 ) . At the start of evolution the population is initialized with a group of 50 identical organisms in a field of size 30×30 . These organisms have a genome containing a single copy of each gene type in a randomized order and with an average of 2 TFBS , randomly drawn from the possible types of TFBS , upstream of each gene . Evolution occurs through mutations on the genome and fitness dependent reproduction . We apply gene duplications and deletions , TFBS duplications and deletions , and changes in the type and weight ( activating or repressing ) of TFBS . Note that in contrast to some previous studies [41]–[43] we do not evolve gene expression rates , protein decay rates , or TF activation and inactivation threshold levels here . Tournament selection is used to determine which organisms may reproduce . Death occurs with a constant probability of 0 . 5 . After an initial transient population sizes plateau at around 600 individuals . As explained , we are interested in the robustness , evolvability and modularity of the evolved developmental GRNs . To give evolution the freedom to evolve networks producing segments and differentiation domains either in a modular or integrated manner , we choose our fitness function such the number of segments and domains contribute independently to fitness ( i . e . we use rather than e . g . ) . As a side effect of this choice , evolution is also free to evolve only segments or domains , rather than both . For our analysis we select those simulations that were successful in evolving both a significant number ( ≥7 ) of segments and domains . To determine the evolutionary history of a developmental mechanism and its underlying GRN we traced the ancestry of the final fit evolved individuals . We performed a total of 50 simulations using the default parameter settings of our model ( see Table S1 in Text S1 ) . We analyzed the networks that successfully evolved segmentation and differentiation in terms of evolutionary strategy followed ( whether segments and domains evolve sequentially , simultaneously , or something in between ) , network size and architecture ( number of genes and connections , positive feedback loops , attractors ) and generated developmental dynamics ( type of spatiotemporal gene expression patterns and how this generates segments and domains ) . Furthermore we evaluated the robustness , evolvability and modularity of different evolved network types . First , to determine robustness of different evolved network types we performed three additional series of 50 simulations . We increased mutation rate , added gene expression noise , or added variability in morphogen wavefront speed ( see Text S1 ) . From the frequency with which the different evolutionary strategies ( SS or SF ) occur we determine their relative robustness . Second , we performed a total of 140 simulations to find how network types differ with respect to evolvability . Here , we first performed 20 simulations with a fitness target of 6 segments and 6 domains . From these we selected 6 successful networks that differed in type ( SS or SF ) . These were each used as a starting point for 20 independent simulations with a fitness target of 9 segments and 9 domains ( see Results ) . From differences in rates of success of evolving to this second target we determine the relative evolvability of the different network types . Finally , we determined the modularity of the different network types . Here we used a range of approaches . First , we determined the architectural modularity of the evolved networks using algorithms that try to find the optimal modularity score or Q value for a network . To ensure that our results were not biased by the particular details of the algorithm used , we used two different methods applying different heuristics . The first uses Newman's leading eigenvector method to determine optimal modularity [33] , [34] , the second method uses a random walk approach to determine Q values [35] . Furthermore , to allow interpretation of the thus found Q values , we determined Q values for not only random and architecturally modular networks , but also for neutrally evolved networks . These neutrally evolved networks serve as a benchmark against which to test whether there is selection for architectural modularity in our simulations . However , architectural network modularity can easily be obscured by the presence of non-functional or redundant genes and regulatory interactions . Therefore , we pruned the original evolved networks to a minimal essential core network ( see Text S1 ) and also determined Q values for these core networks . Furthermore , architectural modularity may be obscured by the particular model parameter setting used , when these tend to cause the evolution of densely connected networks . To determine whether this was the case , we performed 3 additional series of simulations in which the impact of TFBS deletion rates on modularity was tested . In the first two series , TFBS deletion rates were increased either twofold or fivefold , while all else was kept the same as in the default simulations . In the last series of experiments , a core network with a relatively high Q value was selected from the set of default simulations . This core network was subsequently taken as a starting point for continued evolution simulations with a fivefold higher than normal TFBS deletion rate . Finally , as an alternative to these automatic , purely architectural methods of determining network modularity , we also assessed modularity in an alternative way . Here we used the core networks as a starting point to determine the minimal networks needed for either segmentation or differentiation alone . To determine how modular a network is we subsequently asses three points . First , we check how well the minimum networks are capable of autonomously reproducing the original segment or domain pattern . Second , we determine how well they can produce one thing ( segments ) without as a side effect also accidentally producing the other thing ( domains ) . Finally , we assess the amount of overlap between the two minimum networks . Thus , we assess how functionally autonomous and how functionally and architecturally independent these network parts are . The method thus takes into account prior knowledge about network function ( they generate both segments and domains ) and considers both functional and architectural aspects of modularity . If the minimum segment and domain networks function are good at reproducing either only the original segment or the original domain pattern and contain only a few overlapping genes and connections , we will classify the network as modular . In contrast , Q value based algorithms may fail to detect modularity if modules share not only connections but also a few genes .
Figure 2A schematically shows the phase space of possible evolutionary trajectories of evolving both segments and domains . In it we show 3 theoretically possible extreme trajectories: 1 ) all segments evolve before domains evolve; 2 ) the opposite , all domains evolve before segments evolve , 3 ) the intermediate , segments and domains evolve simultaneously . In our analysis we focus on those 30 simulations ( out of the total of 50 ) in which ≥7 segments and ≥7 domains evolved . We find that in 10 of these simulations ( 33% ) first most segments and then domains evolved . In Figure 2B the evolutionary trajectory of 5 of these simulations is shown . Each trajectory shows the maximum number of segments and domains in the population as a function of evolutionary time . In the 20 other simulations ( 67% ) segments and domain numbers increased more or less simultaneous over evolutionary time . Figure 2C shows the trajectories of 5 of these simulations . None of the simulations first evolved most domains and then segments . A detailed overview of the results of the 10 SF simulations and 20 SS simulations can be found in Tables S4–S9 of Text S1 . These results are summarized in Table 1 . When comparing network architecture , we find that SF networks are simpler , with similar numbers of genes but significantly lower connectivity . With regards to the network's developmental output , we find that the two alternative strategies attain very similar overall fitness levels . However , SF type networks produce body plans with more segments then domains , whereas the SS type networks do exactly the opposite . In addition , the segments produced by SF networks are much more regularly sized than those produced by SS networks . Indeed , the developmental gene expression dynamics generated by the two network types differ significantly . Figure 3 shows final evolved networks together with the generated intracellular gene expression dynamics , developmental space-time plot , and the final gene expression pattern for both an example SS ( Figure 3A ) and SF ( Figure 3B ) network . We see that the evolved SS GRN is quite complex , containing 24 nodes and 72 connections ( Figure 3A , top row ) . The network produces a complex time transient of gene expression ( Figure 3A , bottom row ) that upon passage of the maternal morphogen wavefront ( gene type 0 , arrow ) is converted into a stable gene expression pattern . We furthermore observe that the gene types that become stably expressed at a location depend on the time when the wavefront passes . As a consequence , the complex time transient is translated into a temporally stable , but spatially diversified gene expression pattern ( Video S1 ) . The space-time plot ( Figure 3A , top row ) shows another representation of this process . We recognize the anterior to posterior progression of the morphogen wavefront as a distinct diagonal pattern , and see how it transforms the time varying gene expression into a stable spatial pattern ( Video S2 ) . If we look at the gene expression pattern at the end of development ( Figure 3A , top row ) we see that a spatially alternating expression of the segmentation gene ( gene type 5 ) produces 7 body segments of different sizes . The combination of spatially varied expression of the identity genes ( gene types 8 till 15 ) produces a total of 10 domains , also of varying sizes . The SF network is indeed simpler , containing 23 genes and 57 connections ( Figure 3B , top row ) . The networks produces a complex time transient of gene expression ( Figure 3B , bottom row ) in which a subset of genes ( gene types 2 , 5 , 7 , 10 , 12 , 13 and 15 ) display an oscillatory dynamics that we did not observe for the SS network . As for the SS network , the passing by of the morphogen wavefront converts the time-varying gene expression into a stable , spatially varied expression pattern ( Video S3 , Video S4 ) . However , in this case the oscillatory dynamics of genes 2 , 15 , 5 and 7 are translated into a regular , alternating expression pattern , allowing gene type 5 ( segmentation gene ) to produce 12 regularly sized segments ( Figure 3B , top row ) . This mode of producing segments resembles the process of somitogenesis in vertebrates . In addition , the non-oscillatory dynamics of genes 3 , 6 , 8 9 and 11 are converted to 4 continuous , staggered expression regions ( Figure 3B , top row ) . This expression pattern resembles the typical expression pattern of Hox genes along the anterior posterior axis of bilaterian animals . As genes 8 till 15 all are identity genes , the combination of the alternating expression of gene 13 and 15 and the continuous staggered expression of genes 8 , 9 and 11 produce a total of 7 different domains ( if multiple regions express the same set of identity genes only the first counts as a domain ) . Similar results were found for other SS and SF networks . Thus , while SS networks use a complex time transient to produce both segments and domains , SF networks use a similar complex time transient to produce domains , while using oscillatory dynamics to produce regularly sized segments . In later sections we discuss further details of these developmental dynamics in the context of network modularity . We found that under the default parameter settings ( Table S1 in Text S1 ) the SS strategy evolved more frequently than the SF strategy . Next , we investigated how the propensity of the two evolutionary strategies is affected by adding noise to our simulations . Previous research has shown that robustness evolves as a result of increased mutational or gene expression noise [46] . Here we thus assume that increased noise , independent of the type of noise , produces indirect selection for robustness . By assessing the frequency with which the different strategies evolve under increased noise we investigate which of the two strategies is more robust . We performed 3 series of 50 simulations . In the first series mutation rate was increased by a factor 10 . In the second the propagation speed of the maternal morphogen gradient was varied between individuals within a 30% range . In the third series 5% gene expression noise was incorporated . Table 2 shows the percentage of successful simulations and how often the different evolutionary trajectories were followed . Note that we did not observe any additional types of evolutionary trajectories , i . e . first evolving domains and then segments . We see that for all 3 additional series of simulations a shift occurred from SS as a dominant evolutionary strategy to SF as a dominant evolutionary strategy . Thus indirect selection for robustness favors the SF type networks , suggesting that these are more robust . Next we determined whether the two network types also differed in evolvability . It is frequently thought that a special selection regime is required for the evolution of evolvability [2] , [13]–[16] . An often used approach is to impose indirect selection for evolvability by alternating between different selection regimes [44] , [47]–[49] . Clearly , such a back and forth alternation between selection criteria is hardly realistic in a developmental context . However , it has been shown that robustness and evolvability of GRNs is strongly correlated [50] , [51] . It is thus interesting to investigate whether the differences in robustness we observed between the two evolutionary strategies are correlated with differences in evolvability . Specifically , we tested for differences in the evolutionary potential of the two network types for evolving new segments and domains . To do this , we first performed 20 simulations in which we selected for 6 segments and 6 domains ( Figure 4 ) . From these simulations we selected the successful ones . Next , we selected 3 SF and 3 SS simulations . From these 6 simulations we extracted the genome of a finally evolved , fit individual . Each of these 6 genomes were used as input for a series of 20 independent simulations in which now selection for 9 segments and 9 domains was imposed . Finally , we compared the success rates of these 6 series of simulations ( Table 3 ) and whether these differed significantly ( pairwise t-test ) ( Table 4 ) . We see that simulations started with SF type genomes have a considerably higher success rate than simulations started with SS type genomes ( Table 3 ) and that these differences are significant ( Table 4 ) . In contrast , simulations started with different genomes but of the same strategy type have much more similar success rates ( Table 3 ) , differences being not or hardly significant ( Table 4 ) . Differences in success rate are thus not due to random differences between genomes from different simulations , but rather are due to the more fundamental differences between genomes evolved following SF versus SS type evolutionary trajectories . Clearly , genomes evolved in a SF trajectory have a higher evolvability for inventing new segments and domains . These results imply that increased network evolvability can occur as a byproduct of selection for robustness , rather than requiring selection for evolvability itself . Note that it remains an interesting question for further research whether other types of evolvability have also increased . Particularly relevant would be whether the ease with which segmentation and differentiation patterns are maintained if embryo size changes , the ease with which celltypes within domains can be changed , or the ease with which segment and domain numbers can decrease are also increased . As a final part , we investigated whether the differences between the SS and SF evolutionary and developmental strategies are reflected in further differences between their evolutionary dynamics .
In this paper we investigated the in-silico evolution of complex body plans that are both segmented and show anterior-posterior differentiation . An implicit assumption of our study thus is that extensive body plan differentation and segmentation tend to evolutionary co-occur . We base this on the fact that most unsegmented , relatively simple animals such as cniderians possess only a small number of different Hox genes and body domains . In contrast , more complex animals with a larger set of Hox genes and more extensive anterior posterior patterning are either segmented , or show signs of past segmentation [55]–[58] . Note that we made no further assumptions on the order in which segmentation and differentiation evolved , or on whether they evolved once or multiple times [58]–[64] . However , the main aim of the current study was not to settle any of the above issues , but rather to use this setup to study whether or not modular developmental networks evolved . We furthermore investigated how evolution of developmental network modularity depends on indirect selection for robustness . In addition , we studied whether evolved modularity and robustness influence future evolvability . Indeed , we could have used a much more general fitness criterion for body plan patterning , for example maximizing the number of celltypes [65]–[67] or the amount of positional information [68] , to study these issues . Instead , we decided to use a more specific fitness criterion that ‘invites’ modularity to evolve , by independently selecting for two functions , segmentation and differentiation . Furthermore , we wished to study segmentation and differentiation as these are considered two major innovations in bilaterian body plan patterning and thus have been extensively studied both experimentally and theoretically . Evolution was successful in generating body plans that were both significantly segmented and differentiated in 60% of our simulations . This demonstrates two things . First , complex body plan evolution is possible but not trivial . Second , this evolution can be achieved without any coding sequence evolution , by allowing evolution to rewire the regulatory interactions between a simple set of developmental toolkit genes and to duplicate and reuse these genes . Our results thus agree with the argued importance of regulatory evolution [1] , [2] , [4]–[8] and duplication and divergent usage of existing gene categories [10]–[12] in body plan evolution . Interestingly , we found that our successful simulations could be divided into only 2 distinct evolutionary scenarios . In 66% of successful simulations segment and domain numbers increased more or less simultaneously during evolution . The evolved developmental networks produced a complex gene expression transient that upon passage of the wavefront was translated into a stable , spatially differentiated expression pattern producing both segments and domains . In the other 33% of successful simulations , first the number of segments increased substantially before the number of domains increased . The evolved SF networks generate gene expression dynamics consisting of a combination of regular oscillations and a complex time transient . The oscillatory dynamics are responsible for producing segments , whereas the complex transient generates domains . Under default parameter settings the segments simultaneous evolutionary strategy is dominant . However , we find that adding noise , thus producing indirect selection for robustness , causes the segments first evolutionary strategy to become the dominant strategy . We furthermore demonstrate that the SF networks also have a higher evolutionary potential for evolving new segments and domains . Based on the observed differences in expression dynamics , robustness and evolvability we hypothesized that SF networks may also be more modular than SS networks . However , when applying commonly used , purely architectural modularity algorithms similar modularity scores were found for SS and SF networks . Furthermore , these scores were below those of neutrally evolved networks and very close to those of random networks , indicating that no selection for the type of modularity measured by these algorithms occurred . Only by using our functional knowledge of the networks ( they should generate both segments and domains ) , and taking both functional ( different network parts should independently generate either segments or domains ) and architectural ( these network parts should be largely non-overlapping ) aspects of modularity into account could we establish differences in modularity between SS and SF networks . We found that SS networks generated segments and domains in a rather integrated manner , while SF networks operate in a more modular fashion . However , the found modularity was not 100% . Indeed , the SF subnetworks needed to generate either segments or domains share a small subset of their genes and regulatory interactions . Furthermore , a subset of the domains can only be generated in a segment dependent manner . Still , SF networks are considerably more modular than SS networks . Our results agree with the often heard suggestion that selection for robustness favors modular GRNs and that these modular GRNs tend to be more evolvable [2] , [13]–[16] . Furthermore , our findings demonstrate the importance of considering functional aspects of biologically relevant network modularity [36]–[39] . We observed two additional interesting differences between the SS and SF evolutionary strategies . First , while genome size is uncorrelated with body plan complexity for the SS networks , for SF networks not total but core genome size is correlated with organismal complexity . Second , we observed that the complexity and functionality of SF networks changed during evolution in a much more incremental fashion than did the SS networks . Both these differences are likely to contribute to the larger robustness and evolvability of SF networks . We never observed a domains first segments later evolutionary strategy . In hindsight this is easy to understand . Segments can be generated through two alternative mechanisms . The first , applied in SF networks , uses a segmentation gene oscillator to produce regular segments independent of any domains . The second , used in the SS networks , creates segments by linking segmentation gene expression to the expression of domain forming genes . In this latter case , once a differentiation gene has a spatially varied expression pattern , evolution of a single regulatory link to the segmentation gene suffices to produce segments . Because of this easiness of using domains to make segments , we never observe early evolution of domains with a later evolution of segments . Previous simulation studies on the evolution of body plan patterning have modeled the evolution of either segmentation [41] , [42] , [69] or differentiation [43] , [65]–[67] , [70] alone . The major aim of these studies was to gain an understanding of how natural developmental mechanisms might have evolved . As a consequence these studies focused on the resemblance between in-silico evolved network architectures and those found in nature [41]–[43] . Below we compare our results both to the findings of these earlier studies and to developmental networks found in nature . It should however be kept in mind that in our study this resemblance was neither an explicit aim nor part of our model design . As discussed above , SS networks generate a single complex gene expression transient that produces both segments and domains . In contrast , SF networks generate both oscillatory dynamics and a complex time transient , the first responsible for producing segments and the second responsible for generating domains . The translation of oscillatory dynamics by a wavefront into a regular segmentation pattern is called the clock-and-wavefront mechanism for segmentation . It was first suggested by Cooke and Zeeman [71] and has been extensively modeled [72]–[76] . This mechanism is responsible for vertebrate somitogenesis [77]–[81] , arthropod short germband segmentation and annelid segmentation [64] , [82]–[84] . It is suggested to be the ancestral mode of segment formation [60] , [62] , [85] . Recently , Francois and co-workers [41] found that selection for body plan segmentation in the presence of a propagating morphogen wavefront always leads to the evolution of a clock-and-wavefront type mechanism . In contrast , we find that under selection for both segmentation and differentiation either a clock-and-wavefront type segmentation mechanism or a mechanism in which segmentation depends on the expression of domain forming genes may evolve . In the latter case , segments arise downstream of the differentiation process , with different segments arising from different combinations of domain forming genes . This mechanism very crudely resembles the long germband , Drosophila type of segmentation [86]–[88] . However , in our model segments are formed sequentially rather than simultaneously . The fact that we do not observe a hierarchy of mutual repressors as has been observed in simulations of long germband type patterning [42] , [43] is most likely due to this sequential rather than simultaneous patterning . Our results suggest that key to understanding Drosophila segmentation is not just considering that the process occurs simultaneously rather than sequentially , but to also take into account that the segmentation and differentiation processes are tightly integrated . We found that both SS and SF networks use a complex gene expression transient to produce different domains , and in case of the SS network also different segments . In addition , we found for the SF network that the domains produced were of a continuous staggered nature , somewhat similar to the Hox gene anterior posterior expression domains . In a previous study , Francois and Siggia [43] explicitly selected for such a Hox like differentiation pattern . They found that in case of a propagating morphogen wavefront , a special timer gene was needed to control the order and location in which genes were switched on . The expression level of this timer gene slowly accumulated in the time preceding the passage of the wavefront , thus allowing a translation of wavefront passage time into timer gene expression level and finally expression of a different set of downstream genes . In contrast , in our study we obtained anterior-posterior differentiation without the need for such a timer gene , by combining the presence of alternative attractors with a long and complex time transient . Together this ensures convergence to different attractors at different times of wavefront passage , thus also producing sequential spatial differentation . Experimental data suggest that the initial Hox gene activation occurring during the primitive streak phase is temporally colinear and may involve timing mechanisms such as chromosomal looping , ordered opening of chromatin domains and cluster level activator and repressor regions [89]–[94] . In contrast , the Hox gene activity in the presomitic mesoderm and during somite formation appears to be under more individual gene level regulatory control [93] , [95] , [96] and coordinated with the somitogenesis clock and morphogen wavefront [94] , [96]–[103] . Indeed , in our segment first simulations we find that the segmentation and patterning processes both depend on the morphogen wavefront ( Figure 6B , middle and bottom row ) , and that they require some coordination ( see Figure S10 in Text S1 ) . This resemblance to vertebrate axial patterning evolved for free , as it was neither part of our fitness criterion nor of the model design and is a side effect of considering the combined evolution of segmentation and differentiation . Furthermore , it demonstrates that the evolution of natural developmental mechanisms such as vertebrate axial patterning is neither a very unlikely event nor a completely random outcome , but a type of solution that can be expected . | An important question in evolutionary developmental biology is how the complex organisms we see around us have evolved , and how this complexity is encoded in their DNA . An often heard statement is that the gene regulatory networks underlying developmental processes are modular; that is , different functions are carried out by largely independent network parts . It is argued that this network modularity allows both for robust functioning and evolutionary tinkering , and that selection thus produces modular networks . Here we use a simulation model for the evolution of animal body plan patterning to investigate these ideas . To allow for the evolution of modular networks we independently select for both body plan segmentation and differentiation . We find two distinct evolutionary trajectories , one in which segments evolve before domains , and one in which segments and domains evolve simultaneously . In addition , the two evolved network types also differ in terms of developmental dynamics . We show that indirect selection for robustness favors the segments first type networks . Furthermore , as a free side effect , these more robust networks are also more evolvable . Finally , we take into account both functional and architectural aspects to determine the modularity of the network types . We show that segments simultaneous networks generate segments and domains in a integrated manner , whereas segments first networks use largely independent modules to generate segments and domains . Finally , although mimicking natural developmental mechanisms was not part of our model design , the segments first developmental mechanisms resembles vertebrate axial patterning mechanisms . This resemblance arises for free , simply from considering segmentation and differentiation in combination . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"evolutionary",
"biology",
"computerized",
"simulations",
"developmental",
"biology",
"mathematical",
"computing",
"mathematics",
"theoretical",
"biology",
"biology",
"nonlinear",
"dynamics",
"biophysics",
"physics",
"systems",
"biology",
"computer",
"science",
"computer",
... | 2011 | Evolution of Networks for Body Plan Patterning; Interplay of Modularity, Robustness and Evolvability |
Negative regulation of receptor signaling is essential for controlling cell activation and differentiation . In B-lymphocytes , the down-regulation of B-cell antigen receptor ( BCR ) signaling is critical for suppressing the activation of self-reactive B cells; however , the mechanism underlying the negative regulation of signaling remains elusive . Using genetically manipulated mouse models and total internal reflection fluorescence microscopy , we demonstrate that neuronal Wiskott–Aldrich syndrome protein ( N-WASP ) , which is coexpressed with WASP in all immune cells , is a critical negative regulator of B-cell signaling . B-cell–specific N-WASP gene deletion causes enhanced and prolonged BCR signaling and elevated levels of autoantibodies in the mouse serum . The increased signaling in N-WASP knockout B cells is concurrent with increased accumulation of F-actin at the B-cell surface , enhanced B-cell spreading on the antigen-presenting membrane , delayed B-cell contraction , inhibition in the merger of signaling active BCR microclusters into signaling inactive central clusters , and a blockage of BCR internalization . Upon BCR activation , WASP is activated first , followed by N-WASP in mouse and human primary B cells . The activation of N-WASP is suppressed by Bruton's tyrosine kinase-induced WASP activation , and is restored by the activation of SH2 domain-containing inositol 5-phosphatase that inhibits WASP activation . Our results reveal a new mechanism for the negative regulation of BCR signaling and broadly suggest an actin-mediated mechanism for signaling down-regulation .
B lymphocytes are a key component of the immune system and responsible for generating antibody responses against foreign invaders . B-cell–mediated antibody responses are activated by signals generated from B-cell antigen receptor ( BCR ) and from T helper cells through antigen presentation . Antigen binding induces self-aggregation of BCRs into microclusters and BCR association with lipid rafts , which lead to the recruitment of signaling molecules to BCRs . First the tyrosine kinases Lyn and Syk are recruited followed by phospholipase Cγ2 ( PLCγ2 ) , phosphatidyinositol-3-kinase , Bruton's tyrosine kinase ( Btk ) , and the guanine nucleotide exchange factor Vav for the GTPases Rac and Cdc42 , which activate signaling cascades [1] , [2] . BCR microclusters grow over time and subsequently merge into each other , resulting in the formation of a BCR central cluster at one pole of the cell [3]–[5] . After initial signaling activation , inhibitory signaling molecules , including the tyrosine and phosphatidylinositol phosphatases SHP , SHIP , and PTEN , are activated , down-regulating signaling [6]–[9] . Defects in the negative regulation of BCR signaling are associated with losses of B-cell self-tolerance and increases in the susceptibility to autoimmune diseases [10] , [11] . However , the molecular details of the negative regulation of BCR signaling have not been well defined . The self-clustering of surface BCRs into microclusters is an essential event for triggering signaling activation and a target for regulation . While surface BCRs have been shown to exist as tight but inhibitory oligomers at the nanoscale before activation [12] , [13] , antigen-induced coalescence and transformation of the nano-clusters into microclusters is required for BCR activation . Spontaneous formation of BCR microclusters induced by actin depolymerization leads to signaling activation in the absence of antigen [14]–[16] . BCRs with high affinity to an antigen cluster induce signaling with faster kinetics and to higher levels than those with low affinity to the antigen [17] , [18] . Conversely , the co-engagement of the BCR with the inhibitory coreceptor FcγRIIB by antigen–antibody complexes inhibits both BCR clustering and signaling [19] , [20] . We have recently shown that while the formation of BCR microclusters induces signaling , the coalescence of BCR microclusters into the BCR central cluster is associated with signaling attenuation at the B-cell surface . Both the attenuation of BCR signaling and the coalescence of BCR microclusters into the central cluster are inhibited in B cells where the gene of SH2 domain-containing inositol 5-phosphatase ( SHIP-1 ) is specifically deleted [21] . These results suggest that the formation of the BCR central cluster is a down-regulatory mechanism for BCR signaling . BCR self-clustering at the B-cell surface depends on actin reorganization . Actin can regulate BCR clustering by controlling the lateral mobility of surface receptors and B-cell morphology [5] , [15] , [17] , [22] . Perturbing the cortical actin increases the lateral mobility of surface BCRs and facilitates BCR self-clustering and BCR signaling [15] , while stabilizing the actin network does the opposite [16] . In response to membrane-associated antigen , B cells undergo actin-dependent spreading and contraction . B-cell spreading expands the area of contact between the B-cell surface and the antigen-tethered membrane , thereby increasing the number of antigen-engaged BCRs [17] . The merger of BCR microclusters into the central cluster appears to depend on B-cell contraction following spreading since the BCR central cluster fails to form when cell contraction is inhibited [21] . Conversely , BCR clustering and B-cell spreading are regulated by BCR signaling . B cells with genetic deletion of signaling molecules , CD19 , PLCγ2 , Btk , Vav , or Rac , exhibit impaired BCR clustering and B-cell spreading [23]–[25] . We have demonstrated that the stimulatory kinase Btk is essential for the activation of the actin regulator Wiskott–Aldrich syndrome protein ( WASP ) , B-cell spreading , and BCR clustering . In contrast , the inhibitory phosphatase SHIP-1 inhibits WASP activation by suppressing Btk activation , which promotes B-cell contraction and the coalescence of BCR microclusters into the central cluster [21] . Furthermore , actin reorganization is essential for BCR internalization [26] , which is not only the initiation step of antigen processing but also a mechanism for down-regulation of receptor signaling . Therefore , actin can provide both positive and negative feedback to BCR signaling . WASP is an actin-nucleation–promoting factor that is specifically expressed in hematopoietic cells [27] , [28] . WASP mutations that ablate its expression cause a severe and complicated X-linked immune disorder , WAS [29]–[31] , which demonstrates the critical role of actin in immune regulation . WAS patients suffer from recurrent bacterial infections , which are associated with defective T-independent antibody responses against polysaccharide and impaired maturation of T-dependent antibody responses [30] , [32]–[34] . In addition , a large portion of WAS patients develop autoimmune diseases , which are associated with a higher risk of leukemia and lymphoma [35] , [36] . In the WASP knockout ( KO ) mouse model , there is no significant defect in T- and B-cell development , except for a reduction in marginal zone B cells [37]–[39] . The in vitro activation of T cells but not B cells from WASP KO mice is decreased [40] . WASP-deficient B cells from both mice and WAS patients exhibit defective migration [34] , [37] , [38] . We have found that there were significant reductions in cell spreading , BCR clustering , surface tyrosine phosphorylation [21] , and BCR internalization [41] in WASP KO B cells as compared to those of wild type ( wt ) B cells . Recent studies have clearly demonstrated a critical and B-cell–intrinsic role for WASP in controlling B-cell self-tolerance [39] , [41] . However , the molecular mechanism underlying the loss of self-tolerance in WASP-deficient B cells has not yet been elucidated . In addition to hematopoietic-specific WASP , B cells , like all immune cells , also express the ubiquitous N-WASP . These two proteins share ∼50% sequence homology [42] and are reported to have the same cellular function , activating actin polymerization and branching by binding to Arp2/3 complexes [43] , [44] . They are capable of linking cell signaling to actin dynamics via their GTPase binding ( GBD ) , proline rich ( PRD ) , and pleckstrin homology ( PH ) domains . The binding of GTP–Cdc42 and phosphatidylinositol-4 , 5-biphosphate ( PI ( 4 , 5 ) P2 ) releases the proteins from an autoinhibitory conformation [43] , [45] , [46] . Signaling-induced phosphorylation of WASP and N-WASP at tyrosines in the GBD domain and serines in the VCA ( verprolin homology , cofilin homology , and acidic ) domain is required for their optimal activity [47]–[49] . In B cells , Btk is responsible for activating WASP by activating Vav , a guanine nucleotide exchange factor for Cdc42 , stimulating the production of PI ( 4 , 5 ) P2 , and inducing the phosphorylation of WASP [50] . While WASP and N-WASP have been well studied individually , how these two proteins function when coexpressed is largely unknown . Recent studies have demonstrated that both WASP and N-WASP are critical for the development and function of B cells [38] . However , why both WASP and N-WASP are required and how N-WASP and WASP functionally coordinate with each other during BCR activation is not well understood . In this study , we examined the cellular function of N-WASP and its functional relationship with WASP during BCR activation using WASP KO , B-cell–specific N-WASP KO , and double KO mice , as well as primary human B cells from WAS patients . We found that while both WASP and N-WASP are required for optimal signaling activation and internalization of the BCR , the two proteins have distinct functions . N-WASP is critical for the down-regulation of BCR signaling . N-WASP promotes signaling attenuation by facilitating B-cell contraction and coalescence of BCR microclusters into a central cluster and by mediating BCR internalization . Surprisingly , WASP and N-WASP functionally suppress each other and are inversely regulated by stimulatory and inhibitory signals . These results reveal a new function for N-WASP in the negative regulation of receptor signaling and demonstrate a unique functional relationship between N-WASP and WASP during BCR activation .
To understand the function of N-WASP in BCR activation , we examined the activation status of N-WASP in relation to WASP activation in human peripheral blood ( PBMC ) and mouse splenic B cells in response to BCR cross-linking . We labeled and cross-linked surface BCRs using Alexa Fluor 546–labeled , monobiotinylated Fab′ fragment of anti-mouse or human IgG+M antibody ( AF546–mB-Fab′–anti-Ig ) that was tethered to planar lipid bilayers by streptavidin to mimic membrane-associated antigen or mB-Fab′–anti-Ig plus soluble streptavidin to mimic soluble antigen . We chose this activation system because it activates the BCR on mouse and human B cells in a similar way and because it has similar capability of inducing BCR activation as bona fide Ag , such as hen egg lysozyme ( HEL ) tethered to lipid bilayers that activates B cells from MD4 transgenic mice ( Figure S1 ) . Active WASP and N-WASP were detected using antibodies specific for their phosphorylated forms that did not show cross-reactivity between phosphorylated WASP and N-WASP ( Figure S2 ) . Human and mouse B cells were activated with membrane-tethered or soluble mB-Fab′–anti-Ig plus streptavidin before staining for phosphorylated WASP ( pWASP ) and N-WASP ( pN-WASP ) . Using total internal reflection fluorescence microscopy ( TIRFM ) , we analyzed the distribution and the relative levels of pWASP and pN-WASP at the surface of B cells in contact with the mB-Fab′–anti-Ig–tethered lipid bilayer ( B-cell contact zone ) . We found that both pWASP and pN-WASP were detected in the contact zone of mouse ( Figure 1A , B ) and human B cells ( Figure 1C , D ) , and they have a punctate appearance . As Ag–BCR complexes coalesced and merged into a central cluster accompanied by B-cell contraction , most of the pWASP staining ( Figure 1A , C ) , but not pN-WASP ( Figure 1B , D ) , moved away from the center to the edge of antigen–BCR central clusters . The mean fluorescence intensity ( MFI ) of pN-WASP in the B-cell contact zone increased over time similar to that of pWASP . However , the pN-WASP MFI reached its maximal level 2 min later than pWASP , concurrent with the pWASP levels returning to the basal levels ( Figure 1E , G ) . Flow cytometry analysis was used to determine the overall levels of pWASP and pN-WASP in response to mB-Fab′-anti-Ig plus soluble streptavidin . We found similar patterns of WASP and N-WASP activation , where the MFI of pN-WASP peaked 3–5 min later than that of pWASP until the pWASP levels decreased ( Figure 1F , H ) . These results indicate that N-WASP is transiently activated at locations where BCRs interact with antigen similar to WASP , but its activation does not peak until the level of WASP activation decreases . Previous studies have shown that WASP is dispensable for antigen-induced BCR clustering and B-cell morphological changes [21] , [41] , implying a compensatory role for N-WASP in WASP KO B cells . To investigate this hypothesis , we utilized WASP KO mice ( WKO ) , B-cell–specific N-WASP KO mice ( cNKO ) , and double KO mice where N-WASP is selectively deleted in B cells ( cDKO ) , established previously by Westerberg et al . [38] . Using reverse transcription PCR and Western blot , we were unable to detect any mRNA and protein of N-WASP in B cells sorted from the splenocytes of cNKO and cDKO mice ( Figure S3 ) , confirming the effective deletion of loxP flanked n-wasp gene by CD19Cre . We examined the effect of WASP and/or N-WASP KO on BCR clustering and B-cell morphology in response to membrane-tethered Fab′–anti-Ig . Surface BCR clustering was analyzed by TIRFM . As we have shown previously [16] , [21] , upon being bound by the BCR , AF546–mB-Fab′–anti-Ig clustered , coalesced , and formed a polarized central cluster at 7 min ( Figure 2A ) , and the total fluorescence intensity ( TFI ) of AF546–mB-Fab′–anti-Ig in the contact zone of littermate control B cells ( CD19Cre/+ or N-WASPFlox/Flox ) increased over time ( Figure 2D ) . There was no significant BCR clustering and accumulation in the contact zone of B cells interacting with transferrin ( Tf ) -tethered lipid bilayer ( Figure 2A , D ) , indicating that Fab′–anti-Ig aggregates reflects BCR clustering . In WKO and cNKO B cells , the TFI of labeled BCRs in the contact zone was significantly decreased compared to that of littermate control B cells ( Figure 2D ) . While BCR accumulation in the contact zone of WKO and cNKO B cells was decreased to similar levels , the BCRs showed distinct distribution patterns . BCRs in the contact zone of WKO B cells formed a central cluster smaller than that of control B cells ( Figure 2A ) , while in cNKO B cells they appeared punctate , failing to merge into a central cluster ( Figure 2A ) . Treating MD4 B cells stimulated with membrane-associated HEL with the N-WASP inhibitor wiskostatin resulted in similar phenotypes as seen in cNKO B cells ( Figure S1 ) . The deletion of both wasp and n-wasp genes caused a further decrease in the BCR TFI in the B-cell contact zone , similar to the levels in unstimulated B cells ( Figure 2A , D ) . Similarly , treating WKO B cells with the N-WASP inhibitor wiskostatin ( Figure 2D ) [51] or A20 lymphoma B cells with siRNAs targeted to WASP and N-WASP ( Figure 2B , F ) reduced the BCR TFI in the contact zone to levels similar to that in cDKO B cells . Furthermore , the BCR TFI in the contact zone of human B cells was decreased by wiskostatin treatment to levels similar to that of cNKO mouse B cells ( Figure 2C , H ) . We examined the effect of WASP and/or N-WASP KO on B-cell morphology by quantifying the changes in the contact area between B cells and Fab′–anti-Ig–tethered lipid bilayer , using interference reflection microscopy ( IRM ) . In littermate control B cells , the contact area increased rapidly during the first 3 min , indicating cell spreading , and then decreased after reaching a maximal spread area , indicating cell contraction ( Figure 2A , E ) . The change of WKO B-cell contact area over time was qualitatively similar to that of control B cells , but with a smaller magnitude of increase ( Figure 2A , E ) . Both the kinetics and magnitude of the increase in the contact area of cDKO B cells were dramatically slower and smaller than those of control and WKO B cells ( Figure 2A , E ) . Surprisingly , the contact area of cNKO B cells was significantly larger and decreased much later ( 7 min ) than that of control B cells ( Figure 2A , E ) . Again , treating WKO B cells with the N-WASP inhibitor wiskostatin ( Figure 2E ) or A20 B cells with WASP/N-WASP siRNAs ( Figure 2B , G ) reduced the B-cell contact area to sizes similar to that of cDKO B cells . Treating human primary B cells ( Figure 2C , I ) or HEL-stimulated mouse MD4 B cells ( Figure S1 ) with wiskostatin caused an increase in B-cell spreading and a delay of B-cell contraction , similar to what we observed in cNKO B cells ( Figure 2A , E ) . Taken together , these results indicate that both WASP and N-WASP are indispensable for optimal BCR clustering and B-cell spreading , but N-WASP exhibits two opposing functions , supporting BCR clustering and B-cell spreading in the absence of WASP and promoting the coalescence of BCR microclusters into a central cluster and B-cell contraction in the presence of WASP . N-WASP exhibits similar functions in mouse and human primary B cells . The effects of WASP and/or N-WASP KO on BCR clustering and B-cell morphology suggest their involvement in BCR signaling . To test this hypothesis , we analyzed the impact of WASP and/or N-WASP KO on tyrosine phosphorylation ( pY ) at the cell surface in response to membrane-tethered Fab′–anti-Ig using TIRFM . Similar to what we have shown previously [21] , pY was first detected at BCR microclusters at early times during the interaction of littermate control B cells with membrane-tethered Fab′–anti-Ig ( ∼3 min ) and then at the outer edge of the BCR central cluster at later times ( ∼7 min ) ( Figure 3A ) . The MFI of pY staining rapidly increased upon BCR binding , peaked at 3 min , and then decreased ( Figure 3E ) . The distribution and levels of pY in the contact zone of WKO B cells followed a qualitatively similar pattern as in control B cells , but the increasing magnitude of pY MFI in the contact zone of WKO B cells was significantly smaller than that of control B cells ( Figure 3B , E ) . Double KO of WASP and N-WASP caused a further reduction in the levels of pY in the B-cell contact zone ( Figure 3D , E ) . However , the pY staining in the contact zone of cNKO B cells remained punctate and colocalized with BCR clusters at 7 min ( Figure 3C ) . The peak level of pY in the contact zone of cNKO B cells was similar to that of control B cells , but its attenuation was significantly delayed ( Figure 3E ) . These results suggest that N-WASP is involved in both stimulation and attenuation of BCR signaling . Our previous studies show a two-phase relationship between BCR clustering and signaling , where BCR aggregation into small clusters stimulates signaling activation , but the merger of small clusters into a central cluster leads to signaling attenuation at the cell surface [21] . The effects of cNKO on BCR clustering and B-cell contraction led us to hypothesize that N-WASP may regulate signaling via modulating the clustering of surface BCRs . To investigate this hypothesis , we determined the relative size of BCR clusters based on the TFI of BCR labeling in individual clusters and the relative signaling levels of BCR clusters based on the fluorescence intensity ratio ( FIR ) of the pY to the BCR in individual clusters . A nonparametric regression method , LOWESS , was used to examine the trend between the size and signaling level of BCR clusters . In littermate control B cells , the FIR of pY to BCR increased as the size of BCR clusters increased when their sizes were relatively small . After the clusters reached a certain size , the FIR decreased as the sizes of BCR clusters further increased ( Figure 3F ) . Compared to control B cells , BCR clusters in the contact zone of cNKO B cells were limited to smaller sizes , and the FIR of pY to BCR in individual microclusters of cNKO B cells was much higher than that of control B cells ( Figure 3G ) . These results show that the delayed attenuation of tyrosine phosphorylation in cNKO B cells is associated with the inhibition of the growth of BCR clusters , suggesting that N-WASP can down-regulate BCR signaling by promoting the growth and coalescence of BCR microclusters into the central cluster . In order to confirm the roles of WASP and N-WASP in BCR signaling , we determined the effect of WASP and/or N-WASP KO on the phosphorylation of stimulatory kinase Btk ( pBtk ) and inhibitory phosphatase SHIP-1 ( pSHIP-1 ) in response to membrane-tethered Fab′–anti-Ig by TIRFM and calcium flux in response to soluble Fab′–anti-Ig plus streptavidin by flow cytometry . Similar to the effect of WASP and/or N-WASP KO on pY , the MFI of pBtk was significantly reduced in the contact zone of WKO B cells and further reduced in that of cDKO B cells , compared to that of littermate control B cells ( Figure 3H ) . However , the MFI of pBtk in the contact zone of cNKO B cells was not only significantly higher at the 3-min peak time , but the attenuation of pBtk was also significantly delayed , compared to that of control B cells ( Figure 3H ) . In contrast , the MFI of pSHIP-1 in the contact zone was significantly increased in WKO B cells but reduced in cNKO B cells ( Figure 3I ) . Double KO also caused a decrease in pSHIP-1 levels in the B-cell contact zone , but the magnitude of the decrease was similar to that in cNKO B cells ( Figure 3I ) . Consistent with the changes in levels of pBtk and pSHIP-1 in the contact zone , the level of calcium influx was reduced in both WKO and cDKO B cells , with a much more dramatic reduction in cDKO than WKO B cells . In contrast , calcium influx was increased in cNKO B cells ( Figure 3J ) . These results collectively indicate critical roles for WASP and N-WASP in regulating BCR signaling in response to both membrane-associated and soluble antigen , and suggest dual functions for N-WASP in positive and negative signaling regulation . The negative regulatory function of N-WASP in BCR activation implies a role for N-WASP in controlling B-cell self-tolerance . To investigate this possibility , we analyzed the serum levels of anti-nuclear and anti–double strand ( ds ) DNA antibody . Using an immunofluorescence test , we found that the sera of 50% cNKO mice were positive with anti-nuclear antibody ( n = 4 ) , compared to none of littermate control mice at 6 mo of age ( Figure 4A ) . Consistent with this result , quantitative ELISA analysis detected an elevated level of anti-dsDNA antibody in the sera of cNKO mice , compared to littermate control mice at 6 and 9 mo old ( Figure 4B ) . However , we did not detect any significant increase in the levels of anti-nuclear and anti-dsDNA antibody in the sera of WKO and cDKO mice compared to those of littermate control mice ( Figure S4 ) . Since the n-wasp gene deletion in cNKO mice is B-cell specific , our data indicate a critical and B-cell–intrinsic role for N-WASP in maintaining B-cell tolerance . Since the major function of WASP and N-WASP is to activate actin polymerization , we examined the effects of WASP and/or N-WASP KO on the distribution patterns and levels of F-actin in the B-cell contact zone using phalloidin staining and TIRFM analysis . When littermate control B cells spread on Fab′–anti-Ig–tethered lipid bilayer , F-actin formed smaller clusters throughout the contact zone and partially colocalized with BCR clusters . When B cells contracted and BCR microclusters coalesced into a central cluster , F-actin accumulated at the outer edge of the contact zone ( Figure 5A ) . The level of F-actin in the contact zone of control B cells rapidly increased over time and peaked at 3–5 min when B-cell spreading reached maximal magnitude , followed by a significant reduction at 7 min when B cells contracted ( Figure 2E and 5E ) . Gene KO of WASP or both WASP and N-WASP did not significantly change the distribution pattern of F-actin ( Figure 5B , D ) but reduced the level of F-actin accumulation in the B-cell contact zone ( Figure 5E ) . The reduction was much more drastic in cDKO B cells and WKO B cells treated with the N-WASP inhibitor wiskostatin than in WKO B cells ( Figure 5E ) . In contrast , the level of F-actin accumulation in the contact zone of cNKO B cells was significantly increased at 5 min and remained high at 7 min ( Figure 5E ) . F-actin clusters formed in the contact zone of cNKO B cells appeared much more prominent than those in control B cells , and they displayed sustained colocalization with BCR clusters ( Figure 5C ) up to 7 min . These results suggest that N-WASP not only synergizes with WASP in generating and mobilizing F-actin to BCR clusters during signal activation , but is also critical for removing F-actin from the contact zone during B-cell contraction and surface signaling attenuation . Since WASP and N-WASP share the same cellular function , activation of actin polymerization by binding to Arp2/3 , we next asked how these two molecules exhibit opposing roles in actin remodeling during BCR activation . We analyzed the behavior of Arp2/3 in response to membrane-tethered Fab′–anti-Ig and its spatial relationship with pWASP and pN-WASP at the B-cell surface using TIRFM and Arp2-specific antibody . We found that Arp2 was readily recruited to the contact zone of littermate control B cells , and the timing of its recruitment , levels , and distribution patterns were similar to those of F-actin ( Figure 5F , I ) . Consistent with the effect of WASP and/or N-WASP KO on F-actin accumulation , the MFI of Arp2 staining decreased slightly in the contact zone of WKO B cells and was significantly reduced in that of cDKO B cells , but increased in that of cNKO ( Figure 5F , I ) . This result supports the notion that antigen-induced F-actin accumulation in the B-cell contact zone is mediated through the activation of Arp2/3 by WASP and N-WASP . The spatial relationship of Arp2 with active WASP and N-WASP were analyzed using Pearson correlation coefficients between the staining of Arp2 and pWASP or pN-WASP in the B-cell contact zone . The results showed that pWASP exhibited a significantly higher level of colocalization with Arp2 than pN-WASP in control B cells ( Figure 5G , H , J ) . WASP KO caused a significant increase in the colocalization between pN-WASP and Arp2 , close to the colocalization level of pWASP with Arp2 in control B cells ( Figure 5H , J ) , but N-WASP KO did not change the level of colocalization between pWASP and Arp2 ( Figure 5G , J ) . These data suggest that activated WASP predominately colocalizes with Arp2/3 while inhibiting the colocalization of active N-WASP with Arp2/3 at the B-cell surface . BCR activation induces receptor internalization , which attenuates receptor signaling by removing the receptor from surface signaling microdomains . Since BCR internalization requires actin reorganization [26] , we investigated whether WASP and N-WASP are involved in this process . BCR internalization was evaluated qualitatively by the colocalization of surface-labeled BCRs with the late endosomal marker LAMP-1 using immunofluorescence microscopy and quantitatively by the amount of surface-labeled BCRs remaining at the cell surface after internalization using flow cytometry . In littermate control B cells , 70% of surface-labeled BCRs disappeared from the cell surface after internalization ( Figure 6D ) , and they colocalized with LAMP-1–labeled late endosomes ( Figure 6A , C ) that coalesced in response to signaling [52] . In WKO B cells , the colocalization of surface-labeled BCRs with LAMP-1 was slightly decreased and the amount of the BCR remaining on the cell surface was increased ( Figure 6A , C , D ) , indicating a reduction in BCR internalization . In cDKO B cells , the colocalization of the BCR with LAMP-1 was decreased from 0 . 4 to 0 . 1 , and the amount of the BCRs remaining at the cell surface increased ( ∼80% ) , compared to those in littermate control B cells ( ∼30% ) , showing a dramatic reduction of BCR internalization ( Figure 6A , C , D ) . Similarly , double knockdown of WASP and N-WASP by siRNA reduced the colocalization of the BCR with LAMP-1 in A20 cells ( Figure 6B–C ) . Noticeably , cNKO B cells showed a similar level of reduction in the colocalization of BCR with LAMP-1 and a similar level of increase in surface BCRs as those observed in cDKO B cells , indicating that cNKO causes a decrease in BCR internalization similar in magnitude as cDKO . These results demonstrate that N-WASP plays a major role in this process . The involvement of both WASP and N-WASP in BCR activation suggests a functional coordination between these two proteins . To understand their functional relationship , we compared the activation levels and kinetics of one of the two proteins in the presence and absence of the other using TIRFM and flow cytometry . Both the MFI of pWASP in the contact zone of cNKO B cells ( Figure 7A , C ) and the cellular MFI of wiskostatin-treated mouse B cells ( Figure 7E ) were increased compared to those in untreated control B cells . Conversely , the pN-WASP levels in the contact zone of WKO B cells ( Figure 7B , D ) and in WKO B cells ( Figure 7F ) were significantly higher than those in littermate control B cells . In both cases , the time taken for pN-WASP to peak was not changed . While the levels of pN-WASP and pWASP were changed , the overall protein levels of N-WASP and WASP did not change in WKO and cNKO B cells ( Figure S5 ) , indicating that the increased phosphorylation level is not due to an increase in protein expression . Consistent with the data from mouse models , treating human B cells from healthy subjects with the N-WASP inhibitor wiskostatin resulted in an increase in the cellular level of pWASP ( Figure 7G ) . Furthermore , the level of pN-WASP was increased in PBMC B cells from WAS patients that did not express or expressed low levels of WASP , compared to that of healthy human controls ( Figure 7H ) . These results indicate that WASP and N-WASP negatively regulate each other during BCR activation in both mouse and human B cells . The activation of WASP and N-WASP in response to BCR cross-linking suggests that BCR signaling triggers their activation . We have previously shown that Btk is responsible for activating WASP , while SHIP-1 suppresses WASP activation by inhibiting Btk [21] . To investigate whether N-WASP activation is controlled in the same manner as WASP , we determined the effect of Btk or SHIP-1 deficiency on antigen-triggered phosphorylation of N-WASP using xid mice where the PH domain of Btk contains a point mutation that blocks Btk activation [53] and B-cell–specific SHIP-1 KO mice [21] , [54] . In sharp contrast with the effect of Btk and SHIP-1 deficiency on WASP activation , the MFI of pN-WASP was increased in the B-cell contact zone and B cells from xid mice ( Figure 8A–C ) , while it was decreased in the B-cell contact zone and B cells from SHIP-1 KO mice ( Figure 8D–F ) . These results suggest that the activation of WASP and N-WASP is regulated inversely by BCR signaling: Btk , which activates WASP , suppresses N-WASP activation , while SHIP-1 , which inhibits the activation of Btk and WASP , promotes N-WASP activation .
In this study , we demonstrate that in addition to the overlapping function with WASP in signaling activation , N-WASP plays a unique role in the down-regulation of BCR signaling at the cell surface . This is shown by enhanced and/or prolonged tyrosine and Btk phosphorylation , increased calcium influx , and reduced SHIP-1 phosphorylation in cNKO B cells . Importantly , the enhanced signaling in response to in vitro antigenic stimulation is associated with elevated levels of anti-nuclear and anti-dsDNA autoantibodies in the serum of cNKO mice . Since N-WASP is exclusively deleted from B cells , the increased autoantibody and BCR signaling is the result of B-cell–intrinsic defects . These results indicate that N-WASP–mediated signal attenuation of surface BCRs is critical for the maintenance of B-cell self-tolerance . Our studies find that there are more severe defects in BCR clustering , B-cell spreading , and signal activation in cDKO B cells than in WKO B cells , indicating overlapping functions between N-WASP and WASP in these processes . This is consistent with their shared cellular function in promoting actin polymerization . However , WASP KO alone results in significant decreases in these processes , even though they are much less severe than those in cDKO B cells . These results suggest that N-WASP is unable to completely compensate for WASP in WKO B cells . Our data further show that the inhibition of the attenuation of BCR surface signaling is concurrent with increases in B-cell spreading , delays in B-cell contraction and the coalescence of BCR microclusters into the central cluster , and a blockage of BCR internalization in cNKO B cells . These results suggest that N-WASP promotes signaling attenuation via modulating BCR clustering , B-cell morphology , and BCR internalization . Similar to what we have previously shown in SHIP-1−/− B cells [21] , BCR clusters in cNKO B cells remain smaller in size and exhibit much higher levels of tyrosine phosphorylation than those in littermate control B cells , while cNKO B cell contraction is delayed in comparison with littermate control B cells . These data collectively demonstrate that B-cell contraction , which can provide a force for the merger of small BCR microclusters into polarized central clusters , is an important mechanism for down-regulating BCR signaling at the cell surface . The exact mechanism by which the size of BCR clusters regulates receptor signaling is unknown . Our results show that the levels of pBtk and pSHIP-1 are increased and decreased , respectively , in the contact zone of cNKO B cells , suggesting an association of the activation of positive and negative signaling molecules with the size of BCR clusters and N-WASP expression . N-WASP has been suggested to have a role in regulating the cellular location and activation of SHIP [55] . The dominant role of N-WASP in BCR internalization enables the removal of antigen–BCR complexes from the cell surface , which leads to both surface signaling attenuation and antigen presentation that provides another layer of control over B-cell activation by T cells . While N-WASP has both an overlapping role with WASP in signaling activation and a unique role in signaling down-regulation in B cells , we demonstrate here that N-WASP exhibits the two opposing functions under different circumstances . N-WASP predominantly displays its function for signaling down-regulation in B cells that have normal expression of WASP , since cNKO B cells have no defects in B-cell spreading and signaling activation . However , it switches to signaling activation function in B cells lacking WASP expression , since cDKO B cells have more severe signaling defects than WKO B cells . The different functionalities of N-WASP in the presence or absence of WASP expression suggests that WASP is involved in the functional switch of N-WASP . Indeed , we found that in littermate control B cells , the pWASP level increases first , and the pN-WASP level does not peak until pWASP returns near to its basal level . In the absence of WASP , the pN-WASP level rises earlier and significantly higher in WKO B cells than that in littermate control B cells . Conversely , in the absence of N-WASP , the pWASP level increases earlier and is sustained longer in cNKO B cells than that in littermate control B cells . Moreover , we found similar results in human B cells from healthy subjects and WAS patients . These results indicate that WASP and N-WASP mutually suppress each other for activation in both human and mouse B cells . Previous studies have shown that WASP and N-WASP share the same activation mechanisms , including the interaction with the GTPase Cdc42 or Rac via the GBD domain and PI ( 4 , 5 ) P2 via the PH domain , and tyrosine and serine phosphorylation [43] , [46] , [56] . We have previously shown that Btk is responsible for the activation of WASP in B cells by activating guanine nucleotide exchange factor Vav , increasing PI ( 4 , 5 ) P2 , and inducing WASP phosphorylation [50] . A surprising finding of this study is that N-WASP is not activated by the same pathway as WASP , since the level of pN-WASP is increased rather than decreased in Btk-deficient B cells and is decreased rather than increased in SHIP-1−/− B cell as seen for pWASP . This suggests that SHIP-1 instead of Btk is involved in N-WASP activation . During BCR activation , Btk is activated before SHIP-1 [2] , [10] , [57] , which provides an explanation for the activation of WASP before N-WASP . SHIP-1 inhibits Btk activation via dephosphorylating PI ( 3 , 4 , 5 ) P3 , the docking site of Btk at the plasma membrane [58] , consequently suppressing WASP activation . The molecular mechanisms by which SHIP-1 activates N-WASP are unknown . Based on our data , we postulate that releasing N-WASP from WASP suppression by inhibiting Btk-dependent WASP activation is one possible mechanism for SHIP-1 to facilitate N-WASP activation . In addition , SHIP-1 potentially regulates the location and activation of N-WASP via the signaling adaptor protein Grb2 . Grb2 has been reported to bind both N-WASP and the SHIP adaptor protein Dok [42] , [59] , [60] . Further , the product of SHIP-1–mediated hydrolysis PtdIn ( 3 , 4 ) P2 has been shown to regulate N-WASP by recruiting Tks5-Grb2 scaffold [61] . SHIP may also indirectly regulate the subcellular location and activity of kinases that phosphorylate N-WASP . These possible mechanisms remain to be further examined . Studies accumulated from the last decade have clearly demonstrated that WASP and N-WASP share the same cellular function: stimulating actin polymerization by activating Arp2/3 [44] , [46] , [62] . This raises the question of how these two proteins could possibly have opposing roles in B-cell morphology and BCR clustering and activation . It should be noted that almost all of the studies so far have examined the cellular function of WASP and N-WASP individually in the absence of their homolog . Consistent with these studies , we found that WASP and N-WASP appear to play a similar role in actin accumulation at the B-cell contact zone , shown by a smaller decrease in the level of F-actin in the contact zone of WKO B cells than that of cNKO B cells . Surprisingly , N-WASP KO alone causes a significant and sustained increase in the level of F-actin in the B-cell contact zone , which is inconsistent with the current dogma that N-WASP functions as an actin-nucleation–promoting but not inhibitory factor . Our examination of the spatial relationship of activated WASP and N-WASP with Arp2/3 shows that pWASP has a significantly higher level of colocalization with Arp2 than pN-WASP , and that WASP KO increases the colocalization between pN-WASP and Arp2 , but N-WASP KO does not lead to further increases in the colocalization of pWASP with Arp2 . These results suggest a competition between WASP and N-WASP for binding to Arp2/3 complexes . Since WASP is activated and recruited to the B-cell contact zone by Btk first , this allows WASP to win the competition for binding to Arp2/3 , consequently competitively inhibiting the binding of N-WASP to Arp2/3 . How N-WASP reduces the amount of F-actin in the B-cell contact zone and inhibits B-cell spreading is an interesting question . When Takenawa's group discovered N-WASP , N-WASP was identified as an actin depolymerizing protein since the VCA domain of N-WASP was capable of depolymerizing F-actin in vitro in the absence of Arp2/3 [42] . This raises the possibility that the VCA domain of active N-WASP , when it is not bound by Arp2/3 as it loses the competition to activated WASP , can depolymerize F-actin at the B-cell contact zone . In addition , N-WASP has a unique function in tethering F-actin to budding and moving vesicles [63] , [64] . We have previously demonstrated that the membrane fission step of BCR endocytosis requires actin [26] . The dominant role of N-WASP in BCR internalization implicates that N-WASP is responsible for this actin-dependent step , probably by recruiting F-actin from the B-cell contact zone to and/or activating actin polymerization at budding vesicles containing BCRs [65] , [66] . Furthermore , the involvement of N-WASP in SHIP-1 activation reported here and the capability of N-WASP to interact with BCR adaptor proteins , such as Grb2 [42] , suggest that N-WASP may inhibit actin polymerization and B-cell spreading by facilitating or enhancing the activation of inhibitory signaling molecules . WASP deficiency due to gene mutations causes an X-linked immune disorder , exhibiting immune deficiency , autoimmunity , and lymphoma [30] , [32] , [36] . While defects in other immune cells contribute to the disease , recent studies have demonstrated that B-cell–intrinsic defects are critical for the development of autoimmunity in mouse models [39] , [41] . Here we show that WASP and N-WASP behave similarly in mouse and human B cells , including the sequential activation of N-WASP and WASP and the mutual regulation between the two . The B-cell–intrinsic roles of N-WASP in the down-regulation of BCR signaling and B-cell tolerance demonstrated here suggest a critical contribution of N-WASP to disease development . It is possible that without the competitive inhibition of WASP , the signaling promoting function of N-WASP is enhanced and its signaling attenuation function is reduced , leading to deregulation of BCR and B-cell activation . This hypothesis will be pursued in our future studies . It is well known that the genetic background influences the characteristics of the mouse immune system and the susceptibility of mice to autoimmune diseases [67] , [68] . Due to triple genetic manipulations , the mouse models used in this study were comprised of a mixed background of C57BL/6 and 129Sv . Given our breeding strategy ( described in the Materials and Methods section ) , the CD19Cre/Cre mice on a C57BL/6 background bring B6 genes into WASP−/− and N-WASPFlox/Flox mice on a 129Sv background . The CD19 Cre allele from CD19Cre/+ mice generates a systematic bias for B6 mouse chromosome 7 , where CD19 is located , while n-wasp and wasp genes are located in chromosome 6 and X chromosome of mice , respectively . We utilized CD19Cre/+ mice for the final crossing step , which enables us to generate CD19+/+ littermates with similar numbers of B6 alleles as CD19Cre/+ littermates , thereby providing littermate controls . By using more than 15 sets of littermate controls to compare with WKO , cNKO , and cDKO mice , we found a consistent and significant increase in the level of serum autoantibody in cNKO mice as well as increased spreading of cNKO B cells . While CD19Cre/+ C57BL/6 mice would provide an additional control for ruling out any contribution of genetic background to the results , our data with 15–18 littermate controls improves confidence that the dosage of B6 genes is not biasing the results regarding the negative regulation mediated by N-WASP . Taking the results of this study and previous studies together enables us to propose a working model for the functional coordination of WASP and N-WASP during BCR activation ( Figure 9 ) . Antigen binding to the BCR induces an early activation of Btk ( Figure 9A ) that in turn activates and translocates WASP to the cell surface . Activated WASP stimulates actin polymerization by binding to Arp2/3 , which modulates BCR lateral mobility and drives B-cell spreading . Together , these facilitate BCR clustering and signaling ( Figure 9B ) . The activation of SHIP-1 after the initial signaling inhibits Btk activation , which decreases the level of active WASP and releases N-WASP from the suppression of WASP . The activated N-WASP reduces the surface level of F-actin probably by depolymerizing and/or transferring F-actin to BCR containing budding vesicles . Reductions in F-actin at the B-cell contact zone allow B-cells to contract , which promotes the coalescence of BCR microclusters into a central cluster . Tethering F-actin to endocytosing vesicles by N-WASP is essential for the endocytosis of antigen–BCR complexes . The formation of BCR central clusters and BCR endocytosis lead to the down-regulation of BCR signaling at the cell surface ( Figure 9C ) . While molecular details of this working model require further investigation , this study reveals a novel function of N-WASP in the down-regulation of BCR signaling and a unique functional coordination between WASP and N-WASP during receptor signaling .
Wild-type ( wt ) ( CBA/CaJ ) , xid ( CBA/CaHNBtkxid/J ) , MD4 Ig transgenic , and CD19Cre/+ ( B6 . 129P2 ( C ) -Cd19tm1 ( cre ) Cgn/J ) mice were purchased from Jackson Laboratories ( Bar Harbor , ME ) . B-cell–specific SHIP-1 knockout mice ( CD19Cre/+ SHIP-1Flox/Flox ) were kindly provided by Dr . Silvia Bolland at NIH [21] , [54] . WASP knockout ( WASP−/− CD19+/+ N-WASPFlox/Flox , WKO ) , B-cell–specific N-WASP knockout ( CD19Cre/+ N-WASPFlox/Flox , cNKO ) , WASP and N-WASP double conditional knockout ( WASP−/− CD19Cre/+ N-WASPFlox/Flox , cDKO ) mice were bred as previously described [38] . These mice are generated by breeding WKO mice on a 129Sv background , N-WASPFlox/Flox mice on a 129Sv background , and CD19Cre/Cre mice on a C57BL/6 background . The data were generated from breeding littermates of CD19Cre/+ N-WASPFlox/Flox with CD19Cre/+ N-WASPFlox/Flox for littermate control mice ( CD19+/+ N-WASPFlox/Flox ) and cNKO mice ( CD19Cre/+ N-WASPFlox/Flox ) and of WASP−/− CD19Cre/+ N-WASPFlox/Flox with WASP−/− CD19Cre/+ N-WASPFlox/Flox for WKO ( WASP−/− CD19+/+ N-WASPFlox/Flox ) and cDKO ( WASP−/− CD19Cre/+ N-WASPFlox/Flox ) . The littermates of cNKO breeding that have neither WASP nor N-WASP deficiency were used as controls ( CD19+/+N-WASPFlox/Flox ) . B-cell lymphoma A20 IIA1 . 6 cells ( H-2d , IgG2a+ , FcγIIBR− ) were cultured and splenic B cells were isolated as previously described [50] . Human peripheral blood mononuclear cells ( PBMCs ) were collected from WAS patients , age-matched healthy controls , and healthy adults . B cells were isolated from human PBMCs using a Miltenyi Human B-cell negative selection kit by AutoMACS ( Miltenyi Biotec Inc . , Auburn , CA ) . Mono-biotinylated Fab′ fragment of anti-mouse or human IgM+G antibody ( mB-Fab′–anti-Ig ) was generated from the F ( ab′ ) 2 fragment ( Jackson ImmunoResearch , West Grove , PA ) using a published protocol [69] . The planar lipid bilayer was prepared as described previously [70] , [71] . Liposomes were made by sonicating 1 , 2-dioleoyl-sn-Glycero-3-phosphocholine and 1 , 2-dioleoyl-sn-Glycero-3-phosphoethanolamine-cap-biotin ( Avanti Polar Lipids , Alabaster , AL ) in a 100∶1 molar ratio in PBS . Coverslip chambers ( Nalge Nunc International , Rochester , NY ) were incubated with the liposomes before coating with 1 µg/ml streptavidin ( Jackson ImmunoResearch ) and 2 µg/ml AF546–mB-Fab′–anti-Ig mixed with 8 µg/ml mB-Fab′–anti-Ig antibody . For a nonantigen control , surface BCRs were labeled with AF546-Fab–anti-Ig ( 2 µg/ml ) . The labeled B cells were then incubated with biotinylated holo-transferrin ( Tf; 16 µg/ml , Sigma , St . Louis , MO ) tethered to lipid bilayers by streptavidin . Images were acquired using a Nikon TE2000-PFS microscope equipped with a 60× , NA 1 . 49 Apochromat TIRF objective , a Coolsnap HQ2 CCD camera ( Roper Scientific ) , and two solid-state lasers of wavelength 491 and 561 nm . Interference refection images ( IRM ) , AF488 , and AF546 were acquired sequentially . B cells were incubated with AF546–mB-Fab′–anti-Ig–tethered lipid bilayers at 37°C , and then were then fixed with 4% paraformaldehyde , permeabilized with 0 . 05% saponin , and stained for phosphotyrosine ( Millipore ) , phosphorylated Btk ( BD Bioscience , San Jose , CA ) , SHIP-1 ( Cell Signaling Technology , Inc . , Danvers , MA ) , WASP ( S483/S484 or Y290 ) ( Bethyl Laboratory , Inc . , Montgomery , TX or Abcam , Cambridge , MA ) , N-WASP ( Y256 , Millipore ) , and AF488-phalloidin . The B-cell contact area was determined based on IRM images using MATLAB software ( The MathWorks , Inc . Natick , MA ) . The total and mean fluorescence intensity in the B-cell contact zone was determined using Andor iQ software ( Andor Technology , Belfast , UK ) . Background fluorescence generated by antigen or secondary antibody was subtracted . For each set of data , >50 individual cells from three independent experiments were analyzed . The intracellular calcium flux was measured by flow cytometry using the calcium-sensitive dyes Fluo4 AM and Fura Red ( Invitrogen ) using manufacturer-recommended protocols . The relative levels of intracellular calcium were determined by a ratio of Fluo4 to Fura Red emission using FlowJo software ( Tree Star , Inc . , Ashland , OR ) . B cells were incubated with FcR blocking antibodies ( BD ) and then with PE-Cy7–anti-CD19 ( BD ) at 4°C . B cells were activated with F ( ab′ ) 2–anti-Ig ( M+G ) ( 10 µg/ml , Jackson ImmunoResearch ) at 37°C . The cells were fixed , permeabilized , and stained with anti-phosphorylated WASP ( S483/S484 ) and N-WASP ( Y256 ) antibodies for mouse B cells , anti-phosphorylated WASP ( Y290 ) ( Abcam , Cambridge , MA ) and N-WASP ( Y256 , Millipore ) antibodies for human B cells , and anti-WASP or N-WASP ( Santa Cruz , CA ) antibody for total protein . Stained cells were analyzed by a BD FACS Canto . Anti-phosphorylated WASP antibody showed no significant staining in B cells from WKO mice and anti-phosphorylated N-WASP antibody showed no staining in B cells from cNKO mice ( Figure S2 ) , indicating that there is no cross-reactivity of these two antibodies between phosphorylated WASP and N-WASP . B cells were pretreated with wiskostatin B ( 10 µM , EMD Bioscience , Gibbstown , NJ ) for 1 h at 37°C . The inhibitor was also included in the incubation media . Anti-nuclear antibody in sera was tested using the ANA slide test kit from MBL-Bion ( Des Plaines , IL ) . The serum levels of anti-dsDNA antibody were quantified by ELISA using a published protocol [41] . B cells were incubated with FcγR blocking antibodies ( BD ) and then with PE-Cy7-anti-CD19 ( BD ) at 4°C . CD19 positive cells were sorted with BD FACS Aria II , and mRNAs were extracted using Trizol ( Invitrogen ) . RT-PCR was carried out using the SuperScript III One-Step RT-PCR System ( Invitrogen ) . WASP and N-WASP were amplified using specific primers from Santa Cruz Biotechnology , and β-tubulin was amplified as a control . Lysates were generated from sorted CD19 positive cells and analyzed by SDS-PAGE and Western blotting . Blots were probed with WASP- ( Cell signaling ) and N-WASP–specific antibodies ( Santa Cruz ) , and β-tubulin–specific antibody ( Sigma ) as loading controls . Statistical significance was assessed by the two-tailed student's t test using Prism software ( GraphPad Software , San Diego , CA ) . | Mechanisms to shut down B-cell activation are necessary to ensure termination of an immune response when an infection has been cleared . When this negative regulation goes wrong , it can also lead to autoimmunity . To understand how this inhibitory process is regulated , here we utilized knockout mice containing B cells that are deficient for proteins potentially involved in their negative regulation . We focus on Wiskott–Aldrich syndrome protein ( WASP ) , a key cytoskeletal regulator of hematopoietic cells , and neural WASP ( N-WASP ) , which shares 50% homology with WASP and is ubiquitously expressed . Our study shows that mouse B cells that lack N-WASP protein are activated to a greater level and for longer periods than B cells that express this protein . Furthermore , in mice where B cells do not make N-WASP , the numbers of self-reactive B cells are elevated . We went on to identify molecules that promote or inhibit N-WASP activation and to examine the cellular mechanisms by which N-WASP inhibits B-cell activation . Based on these findings we propose that N-WASP is a critical inhibitor of B-cell activation and serves to suppress self-reactive B cells . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2013 | N-WASP Is Essential for the Negative Regulation of B Cell Receptor Signaling |
The mammalian suprachiasmatic nuclei ( SCN ) contain thousands of neurons capable of generating near 24-h rhythms . When isolated from their network , SCN neurons exhibit a range of oscillatory phenotypes: sustained or damping oscillations , or arrhythmic patterns . The implications of this variability are unknown . Experimentally , we found that cells within SCN explants recover from pharmacologically-induced desynchrony by re-establishing rhythmicity and synchrony in waves , independent of their intrinsic circadian period We therefore hypothesized that a cell's location within the network may also critically determine its resynchronization . To test this , we employed a deterministic , mechanistic model of circadian oscillators where we could independently control cell-intrinsic and network-connectivity parameters . We found that small changes in key parameters produced the full range of oscillatory phenotypes seen in biological cells , including similar distributions of period , amplitude and ability to cycle . The model also predicted that weaker oscillators could adjust their phase more readily than stronger oscillators . Using these model cells we explored potential biological consequences of their number and placement within the network . We found that the population synchronized to a higher degree when weak oscillators were at highly connected nodes within the network . A mathematically independent phase-amplitude model reproduced these findings . Thus , small differences in cell-intrinsic parameters contribute to large changes in the oscillatory ability of a cell , but the location of weak oscillators within the network also critically shapes the degree of synchronization for the population .
Circadian clocks generate the near 24-h oscillations that orchestrate daily behaviors in organisms throughout the kingdoms of life [1] . In mammals , the suprachiasmatic nucleus ( SCN ) , a bilateral structure of 20 , 000 neurons in the ventral hypothalamus , functions as the master pacemaker with circadian cells driving rhythms in behavior and physiological processes , such as sleep-wake , locomotor activity , temperature , and hormone release [2] . It was hypothesized that every SCN neuron acts an autonomous clock , using molecular feedback loops to generate daily rhythms in gene expression and cellular output in the absence of external signals [3] , [4] , [5] , [6] , [7] . For example , the Period2 ( Per2 ) gene , a clock gene found in humans and other animals , shows daily rhythms in transcription that appear to depend on daily repression by complexes including its protein ( PER2 ) [8] , [9] . Recent data , however , highlight that , when isolated , SCN neurons exhibit a range of behaviors including damped or unstable circadian oscillations [10] , [11] . Therefore , although all cells may be capable of autonomous rhythmicity , they require stabilization from the SCN network to function as robust circadian oscillators . The potential source or sources of this cell-intrinsic variability , as well as its potential impact , are unknown . Whether the intrinsic properties of SCN oscillators independent of , or interactions amongst groups of oscillators within , the SCN network , or both , are responsible for the overall behavior is a current area of research [12] , [13] . To first test the hypothesis that intrinsic differences between cells may affect how they resynchronize to each other , we followed daily oscillations of PERIOD2 protein levels in single SCN cells before , during , and after pharmacological blockade of intercellular signaling . The results revealed individual cells that differed in their intrinsic amplitude , level of gene expression , circadian period and ability to sustain rhythmicity , none of which predicted the cells' behaviors as they resynchronized to the population . Instead , we found that oscillations resumed and cells joined the rhythmic population at specific circadian phases , ultimately revealing the previously described daily waves of gene expression across the SCN [6] , [14] , [15] . Recent work has further suggested that the phase relationships of SCN cells across the network could be important for robust rhythmic behavior of the tissue [13] , [16] , [17] . To understand the complex behaviors of SCN cells , many studies have employed computational models . Both deterministic models detailing the molecular processes driving oscillations in single cells and stochastic models investigating the effects of noise on the system have aided in the understanding of mechanisms generating circadian rhythmicity in mammals [18] , [19] , [20] . Multi-cellular network models have been constructed from these single oscillators to describe synchrony across the SCN tissue , entrainment to light-dark cycles , and phase shifting behavior [21] , [22] , [23] , [24] . Network models have also probed regional differences in the SCN [25] , [26] and the phenomenon of splitting , in which synchronized regions in the SCN can oscillate with the same period but opposite phases [27] . We were interested in the relationship between cell-intrinsic rhythmicity and tissue synchronization , and found two major implications in the literature . First is that “smaller is better”: damped oscillators [23] , [24] , [28] and oscillators with short relaxation times [29] synchronize efficiently . Additionally , a recent study of fibroblast cells shows that cellular oscillators have small , but sustained amplitudes , and that their proximity to a bifurcation allows them greater control over their period [30] . The authors note that this could be advantageous for peripheral oscillators that need to be entrained by the pacemaker and suggest that similar properties in pacemaker cells could aid synchrony . Second , network topology also affects the quality of synchrony , and specifically , small-world type network topologies are beneficial for synchrony [22] , [26] . It has not been shown , however , why small oscillations are good for synchrony or how cell-intrinsic behaviors and network topology together affect synchrony . Using a mathematical model provides us the flexibility to explain biological phenomena without constraints found in the physiology , e . g . the type , number , and location of oscillators within a network . We sought to address this by first assessing the roles of intracellular processes on intrinsic properties , such as rhythmic ability and phase-responsiveness . Next we assessed the effects of individual cell properties on network synchronization , and finally , how the location of key cells within the network affects synchrony . We hypothesized that intracellular properties and intercellular interactions contribute to the resynchronization behaviors we observed in the tissue data . To test this prediction , we used a computational model to simulate clock gene transcription-translation feedback loops in single cells and found that small changes in parameter combinations produce the range of intrinsic oscillations observed in SCN cells . When placed in a network , these cells were able to synchronize , meaning that they were capable of adjusting their phases to align with the population . To understand this phenomenon , we computed velocity response curves ( VRCs ) for these cells [31] , [32] . VRCs predict the phase velocity , i . e . how fast phase changes in response to intercellular signals . For our model , the VRCs suggested that cells with weaker oscillations could adjust their phase velocity more readily than cells with strong oscillations . These results were consistent with previous results that “smaller” is better to initiate synchrony , but with an alternative definition of smaller – we studied the effects of rhythmic , but low-amplitude ( weak ) cells , rather than initially rhythmic cells that lose amplitude , and eventually , all rhythmic ability , over the long-term ( damped ) . We therefore tested the prediction that inclusion of weak circadian cells , which are highly responsive when isolated , would improve a network's ability to synchronize . We hypothesized that as weak cells establish rhythmicity and synchrony in the network , they lose responsiveness , becoming strong oscillators when coupled . By using a model of 400 coupled , heterogeneously oscillating cells , we found that increasing the proportion of weak oscillators or placing weak oscillators at more connected nodes in the network allowed for improved resynchronization .
Recent reports have shown that when SCN explants are treated with tetrodotoxin ( TTX ) , a blocker of voltage-gated Na+ channels , the circadian rhythms of single cells gradually drift out of phase from one another [6] , [10] , [33] , [34] . To understand the relative contributions of cell-intrinsic and network properties to these synchronization dynamics , we examined the bioluminescence recorded from single cells ( n = 123 across two nuclei , slice 1; n = 90 within one nucleus , slice 2; for details see Text S1 ) in SCN explants from homozygous PERIOD2::LUCIFERASE ( PER2::LUC ) knock-in mice [35] during and after TTX treatment ( Fig . 1 ) . Although all cells appeared to gradually drift out of phase , only some expressed sustained circadian rhythms while others slowly or rapidly lost rhythmicity until the TTX was removed , at which point they began to regain rhythmicity and , eventually , synchrony . Looking at the timing of recovery of oscillations in slice 1 , we found that approximately one-third of the cells that regained rhythms showed significant circadian oscillations within the first 35 h after removal of TTX . During the next 10 h another group of cells , similar in number , became circadian and began to synchronize to the first group . The remaining cells showed significant circadian rhythms starting around 45 h after removal of TTX , with the final cells entering by 96 h . Interestingly , the initial cohort of cells regained rhythmicity closely in phase while later cells regained rhythmicity with more broadly dispersed phases ( Figs . 1B; Text S1; Rayleigh tests performed at the entrance time of the last cell in each cohort; Cohort 1 , n = 39 cells , r = 0 . 68; Cohort 2 , n = 43 cells , r = 0 . 55; Cohort 3 , n = 32 cells , r = 0 . 43 ) . In the second explant , we found a similar gradual restoration of rhythmicity to individual cells after TTX was removed ( slice 2; Text S1 ) . There was also a spatial pattern in each nucleus of the slices: lateral cells regain rhythmicity earlier than or phase lead medial cells ( see Table S2 and Figure S10 in Text S1 ) . In addition , lateral cells are , on average , smaller in amplitude than medial cells . This suggested a spatial organization of amplitude in the network during synchrony recovery . This led us to ask if there was something intrinsically different about the oscillations in cells that became rhythmic earlier or later after TTX was removed . We reported previously that SCN cells uncoupled by TTX display diverse circadian behaviors both in terms of amplitude and period [10] . We acknowledge the possibility that TTX treatment itself can alter a cell's amplitude; however , we will assume that amplitude during TTX is reflective of amplitude that is independent of other feedback from other cells , and as such , is intrinsic to a cell . To determine whether or not intrinsic behaviors explain early or late restoration of rhythms , we compared amplitudes and periods of cells in TTX-treated SCN explants to the time when their rhythms reemerged and to the quality of synchrony within the group of circadian cells . In both slices , we found no significant correlations ( R2 values of <0 . 2 , Text S1 ) between intrinsic circadian properties , such as mean bioluminescence , total bioluminescence , bioluminescence amplitude and period , and when a cell joined in oscillations within a resynchronizing SCN network . We conclude that intrinsic properties alone fail to explain the dynamic emergence of rhythms and resynchrony of individual cells . Therefore network properties likely participate along with these intrinsic behaviors in synchrony . To explore the relationship , if any , between the intrinsic properties of the cells within the context of the network , we implemented a mathematical model . First , we sought to reproduce the diversity of characteristics of isolated cells ( i . e . PER-driven bioluminescence with patterns that could be described as strongly rhythmic , weakly rhythmic or arrhythmic over multiple days ) by identifying potential molecular determinants of these circadian phenotypes . We utilized an existing model of the mammalian molecular clock to simulate SCN neurons [18] and focused on four parameters that regulate the output we recorded in the biological data ( PER2::LUC ) : the rate of transcription of the Period ( Per ) gene , or translation , phosphorylation or degradation of the PERIOD ( PER ) protein . We categorized each cell as arrhythmic , weak ( rhythmic but low in amplitude ) , or strong ( rhythmic and high in amplitude; see Materials and Methods ) . We found that changing any of the four parameters by at least 10% moved simulated cells from arrhythmic to weakly rhythmic to sustained circadian gene expression ( Fig . S2A ) . Regardless of whether they were varied alone or in combination , these parameters recapitulated the phenotypes found in SCN explants ( Figs . S1 , S2 , S3 ) . We used a multi-dimensional visualization technique to evaluate the relative contributions of the four parameters to rhythm generation [36] , providing a novel analysis of sensitivity of strength and sustainability of circadian oscillations to specific parameter combinations . By nesting parameter combinations into stacks , we arranged our data set with a large number of parameters in two dimensions that could be displayed easily ( Materials and Methods ) . Based on the position of cells across the parameter space visualization , we found that small changes in rates of transcription of Per mRNA and degradation of PER protein produced larger effects than changes in translation and phosphorylation of PER on the circadian phenotype of simulated cells ( Figs . S1 and S2B ) . Per rhythmicity was similarly more sensitive to Bmal1 transcription and BMAL1 degradation than BMAL1 translation and phosphorylation ( Fig . S2B ) . We ensured that individual model cells accurately represented individual cells from the slice . Amplitude was of particular importance because here we tested the effect of weak oscillators for the first time . When we compared the circadian periods and amplitudes of simulated and recorded cells we found no correlation between period and amplitude for either the model or the slices ( Fig . S4; R2<0 . 02 for all ) . Further , the periods were similarly distributed ( slice 1 std . dev . = 2 . 1 h , slice 2 std . dev . = 2 . 1 h , model std . dev . = 2 . 1 h ) and the amplitude distributions were dominated by small values in both the model and the slices ( Fig . S4 , Text S1 ) . This suggests that the period and amplitude values in model cells faithfully mimic behaviors we observe in the slice during TTX treatment . Neither this independence of period and amplitude , nor the dominance of small amplitudes has been described in other computational models . Here we are explicit in our modeling that the intrinsic amplitude is much smaller than the in-network amplitude . We concluded that by specifying small differences in key circadian parameters between cells , our simulated cells accurately represented the diverse rhythmic abilities , as well as realistic intrinsic properties such as period and amplitude , of SCN cells . Another relevant property of a circadian oscillator is how it will adjust its phase velocity ( speed ) following a perturbation . We calculated the velocity response properties of the simulated cell set , including both strong and weak cells . Fig . 2C shows representative velocity response curves ( VRCs ) to a signal , where curves are plotted as a function of phase of oscillation . From the curve , we see that if the signal arrives early in the day ( around circadian time , CT , 0 ) the cell will speed up , and if it arrives late in the day ( between CT6 and CT12 ) , the cell will slow down . To measure the cell's ability to shift , we computed the area under the absolute value of the VRC . We compared this VRC area to intrinsic oscillator amplitude ( the sum of the peak to trough amplitude of all states ) and found it inversely correlated with velocity response ( Fig . 2D; R2 = 0 . 85 ) . This indicates that oscillators with small intrinsic amplitude are more likely to have larger velocity response and therefore greater phase shifting ability compared to high-amplitude cells . Interestingly , we found that small oscillators in both the simulation and slice 1 have a broader distribution of periods compared to strong cells . The VRC results suggest a functional strategy to overcome this period variability: weaker cells are better at shifting their phase . To test empirically if and how the proportion of weak oscillators contribute to the synchronization properties of a network like the SCN , we modeled a network of 400 SCN cells with diverse oscillatory abilities , including different periods and amplitudes , as well as network connections . Specifically , each cell had a unique set of parameters selected randomly to establish a population with defined proportions of arrhythmic , weak and sustained oscillators . We chose to include both local and global coupling between cells based on recent theoretical work [22] . Each cell was connected to its four nearest neighbors and 20% of cells connected to cells beyond their immediate neighbors ( Fig . S5 ) . Coupling was achieved in the model by simulating release of vasoactive intestinal polypeptide ( VIP ) , a known synchronizer in the SCN [37] , from all cells . Each network ( n = 56 independent runs for each condition ) was populated with 400 characterized cells and its overall response to uncoupling and recoupling was measured by calculating the synchronization index ( SI ) of all cells ( see Materials and Methods ) . On average , we found that networks with more weak oscillators ( total oscillator amplitude < = 8 . 4 a . u . ) reliably reached higher levels of synchrony ( SI> = 0 . 7 at days 15–18 ) with approximately 5-fold higher synchrony in networks comprised of mostly weak , compared to mostly strong , oscillators ( Fig . 3A–B; ANOVA between populations , p<0 . 001 ) . We found that networks with only strong oscillators failed to resynchronize ( Fig . 3A–B; SI = 0 . 2 at days 15–18 ) . We conclude that weak , highly shiftable cells can promote synchrony . To test the importance of location within the network , we evaluated synchrony in networks of 50% weak and 50% sustained oscillators in which weak oscillators were assigned to hubs , i . e . nodes with more than the average of 10 outputs ( range = 4–39 outputs ) . We found that when weak cells were placed in the more connected nodes of the network ( n = 56 independent network runs for each condition ) , the population reached approximately 5-fold greater synchrony compared to networks with strong cells at these nodes or networks with oscillators distributed to nodes randomly ( Fig . 3C–D ANOVA between populations , p<0 . 001 ) . The quality of resynchrony , therefore , depended on both the number and placement of weak , shiftable oscillators in the network . To test whether our findings regarding weak oscillators extend to damped oscillators , we repeated the simulations using cells that lose amplitude , and eventually , all rhythmic ability ( see Materials and Methods ) . We found that the effects of weak circadian cells and of damped cells were nearly identical . For example , increasing the percentage of damped cells ( Fig . S6A ) or placing damped cells at network hubs ( Fig . S6B ) enhanced synchrony . Further , we verified that our results were not sensitive to our definition of weak . For the simulations used to generate Fig . 3 and Fig . S6 , the weak cells were the smallest 30% in intrinsic amplitude . We repeated all simulations varying the percentage of rhythmic cells classified as weak . For each of these cut-offs , we measured the largest difference in SI between weak and damped cells ( range = +0 . 07–0 . 69 ) . The closer the weakly circadian cells were to the bifurcation , the more they acted like damped cells . We observed that as long as the cut-off is less than 50% , weak cells are similar to damped cells . To demonstrate that these behaviors could be generalized to other oscillatory systems , we constructed a phase-amplitude model , which functions as a reduced version of our mechanistic model ( Text S2 ) . We wanted to know if the benefit of weak cells for synchrony was evident in simpler systems and if a reduced model could further our understanding . The reduced model also showed that inclusion of more low amplitude , or small , oscillators or strategically placing them at more highly connected nodes increased synchrony . Thus , these results were robust across model compositions and types , and indicate that larger phase adjustments by small oscillators will , in general , produce higher synchrony .
Although physiologists and anatomists have described differences between SCN cells including their circadian amplitude , phase and waveform [38] , [39] , [40] , the functional role of oscillator heterogeneity has been little studied . For example , the intrinsic daily oscillations of SCN neurons can be sustained , damped , or absent [10] , [11] , [41] , but the consequences of these diverse circadian phenotypes remain unknown . Here , we found that the resynchronization of SCN cells following pharmacological blockade of cell-cell signaling involves waves of cells becoming rhythmic and adjusting their phases to join the daily cycling of the population . Previous theoretical studies have suggested that damped cells can aid network synchrony by entraining to a wider range of periods [23] , [24] and relaxation oscillators can entrain faster if they have shorter relaxation rates or more spike-like waveforms [29] , but have also highlighted that it is not yet possible to distinguish whether SCN cells should be modeled as damped oscillators or low amplitude , sustained oscillators [28] . What are the potential sources of these differences ? By tuning a computational model , we found that small changes in a small set of parameters could produce a realistic distribution of cells that varied not only in period length , as has been generated previously [18] , [21] , but in qualities of the oscillations themselves . Using non-biased minimization techniques to represent multi-dimensional parameter space , we found that parameters associated with transcription rate and protein degradation of the Period gene were more likely to contribute to circadian changes than other parameters . We speculate that genetic differences in and the environmental modulation of these key rate constants between SCN cells could underlie the heterogeneity in their circadian properties . For example , it has been shown that the amplitude of Per transcription is altered in the absence of VIP [42] and that the stability of PER protein against degradation affects circadian period [43] . Because we found that the amplitude of our model cells is reliably and inversely related to their ability to adjust their phase velocities in response to natural signals , we tested the impact of both low ( weak ) and high ( strong ) amplitude oscillators on network synchrony . Previously , Bernard and colleagues suggested that a network comprised of damped circadian oscillators is capable of synchronizing and maintaining rhythmicity , and hypothesized that damped oscillators , when synchronized , induced rhythmicity in the population [24] . Locke and colleagues then performed a parameter optimization , searching to maximize the ability of a network of oscillators to synchronize . The best-synchronized networks were composed of damped cells [44] . Together , these results suggested that the driving force coupling cells together could arise from some inherent property found in damped cells . In a similar fashion , we sought to identify inherent characteristics in both biological and modeled weak oscillators , including relationships between intrinsic amplitude and intrinsic shiftability . Published models using damped oscillators have been unable to mathematically quantify shiftability . By studying weak circadian oscillators , we measured larger changes in oscillator speed for smaller amplitude cells . Measurements of shiftability now provide a tool to study for the first time the kinetics of resynchronization . We posit that weakly oscillatory cells can send signals to other weakly oscillatory cells to readily adjust their phases . As the system synchronizes , the cells gain amplitude and thus lose the ability to make dramatic shifts . This suggests a strategy for neurons to resynchronize . The system can move from being sensitive to perturbations to being robust against them through a process of cell-cell amplification of rhythm amplitudes [33] . In our model networks we demonstrated that the total number of a specific oscillator type is critical and that there is an effect of the degree of connectivity of certain oscillator types on synchrony , such that , networks with more and more highly connected weak oscillators have improved synchrony during the recovery period following a perturbation . We concluded that heterogeneity arises from both cell intrinsic and network contributions , including the network topology and number of weakly circadian cells . The model does not account for all dynamics of resynchrony that we observed in the data , which will be addressed in the future . For example , though we observed populations of cells consistently ahead of or behind the mean phase of the network simulations , we observed no spatial pattern in these phase differences; the more homogeneous connections in the model networks led to most cells becoming rhythmic at the same time and together tighten in phase . Future work will use modeling to understand if and how spatial heterogeneity in network connections causes spatial patterns in the phase of oscillators across the slice . Future work will also take into account stochasticity in cell behavior . Preliminary results ( data not shown ) indicate that incorporating white noise into tissue simulations has no effect on the role of weak oscillators . How the evolving differences within oscillators and amongst oscillator populations carry over to the behavior of networks is an open question for investigation . We return to the issue of whether rhythmicity and synchrony are due to intrinsic cell properties or are dependent on cell location and network structure . Recent studies have examined phase heterogeneity within the SCN [12] , [13] and have concluded it is not a function of cellular properties . Foley and colleagues summarized their findings as “the tissue is the issue” – that placement within the SCN network ( based on assigned phase ) dictates whether and how an SCN neuron will oscillate [13] . We extend this to hypothesize specifically that cells , which are intrinsically different in their ability to maintain strong or weak rhythms , will impact the population rhythm differentially ( e . g . the quality of synchronization increases with more weak oscillators ) , but also depending on their location within the network ( e . g . cells at hubs have greater influence ) . Other theoretical studies have emphasized that the number of connections between cells could modulate the degree of synchrony in the network and argued for region-specific placement of particular oscillator types ( e . g . sustained cells in the dorsal SCN and arrhythmic or gated cells in the ventral SCN ) [25] , [45] , [46] . We find no evidence for specialized , localized populations of oscillators in the resynchronizing SCN slice following the removal of TTX . In contrast , our model shows the importance of weakly rhythmic , highly responsive oscillators at hubs where they can send coordinated phase information broadly throughout the network , becoming less responsive as they increase in amplitude , and that this is critical for improved synchrony . It is thought that SCN neurons establish rhythmicity and synchrony amongst each other and with the external light-dark environment late in gestation [47] , [48] . Because these features are likely critical for the survival [49] , we posit that the composition of the SCN , including a continuum of oscillator behaviors and connections , allows the tissue to adjust to shifts in environmental timing cues . These properties may be universal to all networks that include weak oscillators .
Single cells measured in SCN slices reported in this study were recorded as previously published [10] . Briefly , SCN explants from neonatal PER2::LUC mice were cultured for 3 days on MilliCell-CM ( Millipore ) membrane pieces in CO2-buffered medium supplemented with 10% newborn calf serum ( Invitrogen ) before being inverted onto polylysine/laminin coated coverslip dishes . All procedures were approved by the Washington University Animal Studies Committee and complied with NIH guidelines . We conducted recordings in air-buffered medium containing 0 . 1 mM beetle luciferin ( BioThema ) at 37°C beginning at day 2 after slice transfer to coverslip dishes We temporally ( 1 h integration time ) and spatially ( 2×2 pixel resolution ) bioluminescence counts using a Versarray 1024 CCD camera ( Princeton Instruments ) . Following 6 days of baseline recording , we treated organotypic SCN explants with 0 . 5 µM tetrodotoxin ( TTX , Sigma ) as previously described [10] . TTX remained in the medium for 6 days before the medium was removed and we washed explants with 1 full volume exchange of fresh medium . Recording then continued for at least 6 days to examine rhythms as cells resynchronized after the restoration of cell-cell communication . We used NIH ImageJ software to process all images by first subtracting background levels and then measuring pixel intensity over time in a region of interest above each cell . Cells were tracked manually from frame-to-frame and across treatments to account for any tissue movement . Cells were initially scored as rhythmic or arrhythmic if their gene expression rhythm oscillated with a period between 15 and 35 hours that was statistically significant by both Chi-squared periodogram [50] and FFT-NLLS [51] . We also used Wavos to determine period and phase information from the single cell traces [52] . A version of a previously published 16-ordinary differential equation model of the mammalian circadian clock was used to simulate rhythms in single model cells [18] . We altered parameters for rates of transcription , translation , phosphorylation , and degradation of either Period or Bmal1 genes , leaving 50 other parameters set to published basal values [18] , and measured rhythms in gene output . We simulated 720 hours of gene expression from each cell , using initial conditions from a representative , high-amplitude sustained cell . To measure the sensitivity of circadian cycling to clock gene parameters , we organized results from single cell simulations using clutter based dimensional reordering ( CBDR ) , which applies minimization and dimensional stacking algorithms described below . These methods allow visualization of the underlying structure of clock parameter space and gauge the influence of tested parameters relative to output behavior . We utilized published Matlab code [36] to minimize differences between output scores ( strong , weak , arrhythmic ) and cluster behaviors together . The code arranged parameter combinations iteratively until the minimization requirement , i . e . cells with like behavior , were clustered together , was fulfilled . First , the code scans one pair of parameters over a range of values while the remaining parameters are set to basal values . Then we label this grid based on the output for each combination and add it to a larger montage of other parameter pairs . For a useful visualization , the code orders these dimensional stacks to group similar outputs together . Given a unique behavior and parameter combination for each cell , we minimize the stack so that differences between regions of varying outputs are small ( in this case , strong , weak , or arrhythmic patterns in gene expression ) , and this provides an order ranking of “higher” versus “lower” parameters in the stack . Changes in parameter value that produce larger effects in output phenotype are higher in the stack order . A velocity response curve ( VRC ) predicts the effect of parametric perturbation on the phase velocity of the oscillator . For a cell in the SCN , there is a single parameter ( vsP ) that is manipulated by VIP signaling . Hence , we consider the VRC associated with vsP , mapping the circadian time of VIP signaling to its effect on the phase velocity . Cells with higher-magnitude VRCs can be sped up or slowed down more by VIP signals than cells with lower magnitude VRCs . To quantify the “shiftability” of a cell , we compute the area under the absolute value of the VRC . A VRC may be computed for any cell with a parameter set allowing for limit cycle oscillations . For details regarding computation , see [31] . Mathematically , the individual cells we have modeled can be categorized as rhythmic ( those that converge to a periodic orbit ) or arrhythmic ( those that converge to a steady-state solution ) . However , simulations of single cells display a spectrum of behaviors , with some showing lower or higher amplitude than others . Using the peak to trough amplitudes of all model components ( i . e . by summing the amplitudes of all states ) , we separated the rhythmic cells into two categories: 1 . Weak cells are rhythmic with small amplitudes and 2 . Strong cells are rhythmic with larger amplitudes . For some simulations , we needed damped oscillators , which form a subset of the arrhythmic oscillators . We used the total amplitude at the end of a 15-day simulation ( starting from high-amplitude initial conditions ) to create the additional categories: 1 . Flat cells are arrhythmic and have the smallest amplitudes in their final pseudo-cycle and 2 . Damped cells are arrhythmic and have the largest amplitudes in their final pseudo-cycle . For most simulations in the paper , we defined the smallest 30% ( n = 228 ) of the oscillatory cells as weak and the remaining 70% ( n = 595 ) as strong . For simulations that needed to distinguish between flat and damped , we chose the cut-off so there would be the same number of damped cells and weak cells . Model cells were coupled together by VIP signaling , simulated as a drive on the rate of Per transcription , as previously published [21] . In our model , 20 percent of the 400 neurons were capable of sending a VIP signal and all neurons could respond to VIP . Connections between cells were organized with a small world network topology as in [22] where each VIP cell was coupled to its four nearest neighbors and then had a probability of sending unidirectional long-range connections to other cells in the network . We set the connection probability to p = 0 . 05 , resulting in a synchronized system with a range of 4 to ∼40 outgoing connections in most networks . To mimic the TTX experiments , we simulated 6 days with VIP-mediated coupling followed by 6 days with coupling eliminated and then reinstated for 6 days . We assessed the intrinsic circadian expression of each cell as well as the rate of resynchronization of each cell and the ensemble . The network connections and parameter values for each cell did not change throughout the simulation . The synchronization index ( SI ) provides a real-time measure of the phase dispersion across a population of oscillators , which ranges from 1 ( all cells peak in phase ) to 0 ( all cells peak at uniformly-distributed times of the day ) . We defined SI at each time t by the radius r of the complex order parameter [53] according towhere N is the number of cells , φj ( t ) is the phase of the jth cell at time t , and ψ ( t ) is the average phase of all cells . We compute the instantaneous phase of each cell ( simulated or real ) by applying the continuous wavelet transform using a Morlet wavelet [54] to its trace of Period mRNA . The phase of the cell over time may be recovered from the ridges of the transform , which are extracted using a straight-forward algorithm ( [52] , [54]; Wavos Package ) . Briefly , the continuous wavelet transform ( CWT ) produces a complex-valued field over scales ( which may be mapped to instantaneous frequency ) and translations ( which may be mapped to time ) . The magnitude of the complex number at a given translation and scale may be interpreted as the strength of oscillation of the signal at the frequency given by the scale and the time given by the translation . The phase of the complex number at a given translation and scale gives the phase of that oscillatory component . By selecting points with contiguous scales across a range of translations that maximize the magnitude of the CWT ( the “wavelet ridge” ) , we may extract the dominant frequency of the oscillator over time , and from those points extract the phase evolution of the oscillator from the angles of the CWT coefficients . Because the wavelet analysis requires a window ( in time ) around the model state in question , it is unable to calculate the phase during the first and last 34 hours of each simulation . We treat each experimental condition separately , which mean there are gaps in the SI plotted in Fig . 3 . | Circadian rhythms are daily , near 24-h oscillations in biological processes that nearly all organisms on Earth experience . Single cells contain a molecular clock that drives circadian rhythms in physiology and , when many cells synchronize in a population , daily behaviors . We hypothesized that small differences in intrinsic cellular properties allow for a diversity of circadian periods and amplitudes across cells . We observed circadian cells and their synchrony before , during , and after limiting communication between cells and then compared their intrinsic properties to their resynchronization behavior . We found that arrhythmic , weakly oscillating , and self-sustained circadian cells rejoined the rhythmic population independent of their cell-intrinsic oscillations . Using a mechanistic computational model of circadian cells , we found that resynchronization could be enhanced by including more weak oscillators or by placing weak oscillators at more connected nodes in the network . We conclude that intrinsic properties ( e . g . oscillator weakness and responsiveness ) and network structure ( e . g . positions of weak oscillators ) can independently buffer tissue rhythms from perturbations . This reveals how cellular and network properties impose rules on systems of circadian cells that must achieve synchrony from a desynchronized state , for example during perinatal development or when forced to overcome societal constraints on sleep-wake behavior , such as working early or late shifts . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"systems",
"biology",
"biophysic",
"al",
"simulations",
"biology",
"computational",
"biology",
"neuroscience",
"neurophysiology"
] | 2012 | Weakly Circadian Cells Improve Resynchrony |
Mosquito-borne viral diseases cause significant burden in much of the developing world . Although host-virus interactions have been studied extensively in the vertebrate host , little is known about mosquito responses to viral infection . In contrast to mosquitoes of the Aedes and Culex genera , Anopheles gambiae , the principal vector of human malaria , naturally transmits very few arboviruses , the most important of which is O'nyong-nyong virus ( ONNV ) . Here we have investigated the A . gambiae immune response to systemic ONNV infection using forward and reverse genetic approaches . We have used DNA microarrays to profile the transcriptional response of A . gambiae inoculated with ONNV and investigate the antiviral function of candidate genes through RNAi gene silencing assays . Our results demonstrate that A . gambiae responses to systemic viral infection involve genes covering all aspects of innate immunity including pathogen recognition , modulation of immune signalling , complement-mediated lysis/opsonisation and other immune effector mechanisms . Patterns of transcriptional regulation and co-infections of A . gambiae with ONNV and the rodent malaria parasite Plasmodium berghei suggest that hemolymph immune responses to viral infection are diverted away from melanisation . We show that four viral responsive genes encoding two putative recognition receptors , a galectin and an MD2-like receptor , and two effector lysozymes , function in limiting viral load . This study is the first step in elucidating the antiviral mechanisms of A . gambiae mosquitoes , and has revealed interesting differences between A . gambiae and other invertebrates . Our data suggest that mechanisms employed by A . gambiae are distinct from described invertebrate antiviral immunity to date , and involve the complement-like branch of the humoral immune response , supressing the melanisation response that is prominent in anti-parasitic immunity . The antiviral immune response in A . gambiae is thus composed of some key conserved mechanisms to target viral infection such as RNAi but includes other diverse and possibly species-specific mechanisms .
Arthropod-borne viruses ( arboviruses ) are a significant health burden across the world . They represent an emerging and resurgent group of pathogens [1] , many of which are transmitted by mosquitoes including Dengue Fever ( DEN ) , Yellow Fever ( YF ) , West Nile Virus ( WNV ) and Chikungunya ( CHIKV ) . The development of control strategies to combat the spread of these viruses requires a detailed knowledge of host-pathogen interactions in both the vertebrate host and invertebrate vector . Targeting human pathogens , for example malaria parasites , within their insect vectors has been the focus of intense research towards identification of novel targets for transmission blocking interventions . Understanding the molecular mechanisms of immunity to pathogens within insect vectors could reveal potential candidates for such interventions . Extensive research has been carried out into insect immune responses to bacterial , fungal and parasitic infections; however , it is only recently that invertebrate antiviral immunity has received analogous attention . Initial studies have used Drosophila melanogaster as a model system , as the power of genetics and the extensive knowledgebase in Drosophila have been invaluable in establishing the foundations for insect antiviral immunity research . However , the biology of arboviruses is tightly linked to the physiology of haematophagous arthropods , and as such research in model organisms may not be fully relevant to the transmission of viruses and associated vector defence . A forward research approach is required to effectively study the vector responses to arboviruses , utilizing findings in Drosophila as guidance . Mosquitoes launch robust immune responses against a variety of pathogens: recognition of pathogen associated molecular patterns ( PAMPS ) leads to activation of immune signalling pathways associated with production of potent anti-microbial peptides ( AMPs ) or cascades that lead to pathogen lysis , phagocytosis , melanisation or cellular encapsulation by hemocytes , the white blood cell equivalents [2] . To date three signalling pathways have been implicated in mosquito antiviral immunity . The JAK/STAT pathway , a known antiviral signalling pathway in mammals [3] , appears to have a conserved function in Aedes aegypti . JAK/STAT related genes are differentially regulated in response to DENV infection [4] , and Ae . aegypti can be made more or less susceptible to DENV through silencing of DOME ( receptor of the JAK/STAT pathway ) and PIAS ( negative regulator of the JAK/STAT pathway ) respectively [5] . In addition , 18 genes downstream of the Ae . aegypti JAK/STAT pathway are regulated by DENV infection , two of which have been shown to be DENV antagonists [5] . The RNAi pathway has been demonstrated to limit viral infection in several mosquito vector-virus combinations . AgAGO2 ( a member of the RISC complex ) is an antagonist of ONNV in A . gambiae [6]; AeAGO2 , AeDCR2 and AeTSN ( all members of the RNAi pathway ) are Sindbis virus ( SINV ) antagonists in Ae . aegypti [7]; AeDCR2 had also been shown to be a DENV antagonist [8] . The presence of viRNA ( siRNA that is specific to viral genomes ) has been demonstrated in Ae . aegypti infected with SINV and DENV [9] , [10] , and recombinant viruses encoding suppressors of RNAi have been shown to increase mortality , increase viral titres and lower the build-up of viRNAs in infected mosquitoes [9] , [10] . Finally , Toll pathway related genes are differentially regulated in response to both SINV and DENV infection in Ae . aegypti [4] , [11] . Activation and inhibition of the Toll pathway has been demonstrated to respectively decrease and increase susceptibility to different DENV strains in different Ae . aegypti strains showing the importance of the Toll pathway in mosquito antiviral immunity [4] , [12] . Whereas the Aedes and Culex mosquitoes transmit numerous viruses , Anopheles mosquitoes ( the principal vectors of malaria ) are known to be the primary vectors of only O'nyong-nyong virus ( ONNV ) . ONNV is a positive ( sense ) strand single stranded RNA ( +ssRNA ) virus of the Alphavirus family , with reported epidemics in West Africa in the 1960s and 1990s [13]–[18] . Viral replication of ONNV in A . gambiae is shown to be slow and restricted in tissue tropism compared to most vector-virus combinations [19] . Permissiveness to infection has been shown , in part , to be regulated by RNAi , and inhibition of RNAi results in high susceptibility to viral infection [6] . Here we have profiled the global transcriptional responses of A . gambiae to ONNV infection of the hemolymph to identify viral responsive genes , and then used RNAi silencing to test a selection of identified genes for antiviral function . Our results confirm that in A . gambiae the RNAi pathway is a key antiviral mechanism , however , the JAK/STAT and the Toll pathway do not have a significant role in regulating systemic ONNV infection . We further identify four viral responsive genes with novel functions in mosquito antiviral immunity . Patterns of immune gene expression coupled with co-infections of A . gambiae with the rodent malaria parasite P . berghei suggest that viral infection inhibits parasite melanisation . Overall , we demonstrate that A . gambiae uses a combination of conserved antiviral pathways , including RNAi , and novel uncharacterised mechanisms to target ONNV infections .
5′ONNVic-eGFP plasmid was kindly provided by Dr Brian Foy , Colorado State University . 5′ONNCiv-eGFP infectious clones were generated as described in [19] with some modifications . RNA generated in vitro from the infectious clone template was purified using the RNeasy mini kit ( Qiagen ) . RNA concentration and purity was ascertained using a Nanodrop ( Labtech International ) . 2 µg of RNA were transfected into a confluent culture of VERO cells in a T75 flask using the Transmessenger transfection reagent ( Qiagen ) . Cells were observed for cytopathic effects and GFP expression at 24 hours post transfection . At 72 hours post infection cells were scraped and filtered through a 0 . 22 µm filter , aliquoted and stored at −80°C . 250 µl of first passage 5′ONNVic-eGFP was used to infect a large culture of confluent VERO cells . At 72 hours post infection cells were scraped , filtered through a 0 . 22 µm filter , aliquoted and stored at −80°C . Second passage virus was used in all experiments . Adult mosquitoes were maintained as described in detail by Sinden and co-workers [20] . In brief , mosquitoes were reared and maintained at 28°C , 65–70% relative humidity with a 12 hour light/dark cycle . Adult mosquitoes were fed on sterile filtered and autoclaved 10% fructose solution and used for experimental purposes when 1 or 2 days old . Newly emerged female mosquitoes were inoculated with the required dilution of second passage 5′ONNVic-eGFP in MEM ( Invitrogen ) , using a pulled capillary glass needle and a Nanoject ( Drummond Scientific ) . Inoculated mosquitoes were kept in cohorts of 30–50 and maintained as described by [20] . Inoculated mosquitoes were double-contained to prevent escape . Pools of ∼30 whole mosquitoes were homogenised in 200 µl of Drosophila Schneiders medium ( Gibco ) . Homogenates were centrifuged at 3000 g for 30 minutes at 4°C to pellet debris . Supernatant was transferred to a new 1 . 5 ml eppendorf tube and centrifuged at 5000 rpm for a further 30 minutes at 4°C . The supernatant was filtered through a 0 . 2 um filter , and 140 µl of the filtrate was used for viral RNA extraction using the Qiagen viral RNA extraction kit according to the manufacturer's instructions . 10 µl of vRNA was used to generate cDNA using the Superscript II kit ( Invitrogen ) . To ascertain the abundance of viral RNA , or viral genome copy number , an absolute quantification method was used . cDNA was generated from vRNA extracted from a sample with a known viral titre , calculated using standard plaque assay . A standard curve the sample was generated using neat , 1∶5 , 1∶10 . 1∶50 , 1∶100 and 1∶500 dilutions of cDNA . Qrt-PCR was carried out using SybrGreen reagents ( Applied Biosystems ) and primers against the nsP3 ONNV gene ( Table S3 ) . The standard curve was used to calculate the viral genome copy number of an unknown sample by mapping the CT value to that of the standard curve , giving the viral genome copy number . Total RNA was extracted from pools of ∼10 mosquitoes in TRIzol ( Invitrogen ) 4 days after dsRNA treatment . cDNA was generated from total RNA using the Superscript II kit ( Invitrogen ) . Primers were designed against GOIs such that there was no overlap with dsRNA probes ( Table S3 ) including the S7 gene that is constitutively expressed in the mosquito . Qrt-PCR was carried out using SybrGreen reagents ( Applied Biosystems ) . cDNA input was normalised using the abundance of S7 in each sample . Once normalised , gene KD efficiency was calculated as a relative % decrease in transcript abundance compared to a control KD sample . Standard plaque assays were carried out as described in [21] . In brief , individual mosquitoes were homogenised in 270 µl of Drosophila Schneiders medium ( Gibco ) and filtered through 0 . 22 µm filters . 10-fold serial dilutions of each sample were added in duplicate to confluent monolayers of VERO cells in 24 well plates and immobilised using an agar nutrient solution . Cells were stained after 4 days incubation at 37°C using 200 µl of 5 mg/ml Thiozolyl Blue Tetrazolium Bromide ( MTT ) ( Sigma ) in PBS . Plaques of dead cells were counted and used to calculate the plaque forming units ( PFU ) /mosquito . Total RNA extracted from whole homogenates of A . gambiae mosquitoes was amplified and labelled using the Low RNA Input Amplification kit ( Agilent , UK ) . In brief: 2 µg of total RNA was used in a random primed reverse transcription reaction to generate cDNA . After amplification by conversion to cDNA , cDNA was transcribed to copy messenger RNA ( cmRNA ) incorporating either Cy-3UTP ( for the reference sample ) or Cy-5UTP ( for the test sample ) fluorescent nucleotide analogs . cmRNA quality and labelling efficiency was assessed by spectrophotometry using a Nanodrop ( Labtech International ) . If cmRNA yield was sufficient and Cy-3UTP or Cy-5UTP labelling was successful , 825 ng of RNA was hybridised to the Agilent 4X44K array in 2× GEx-hybridisation buffer HI-RPM at 60°C for 17 hours . Hybridised slides were washed with GE wash buffer 1 at RT for one minute and GE wash buffer 2 at 37°C for one minute , to remove excess labelled cmRNA prior to scanning . Microarrays were scanned using a GenePix semiconfocal microarray scanner ( AXON Instruments , Foster City , CA ) Gene Pix Pro 6 . 1 was used to record feature signal intensity , to eliminate local backgrounds , for grid alignment and manual inspection of feature quality . Average feature diameter was calculated and features lying outside three standard deviations of the mean were excluded from analysis . The ratio of feature intensity verses local and global backgrounds were calculated and features not exceeding background intensities were excluded from analysis . Features were normalised using Genespring 6 . 1 ( Axon instruments ) by locally weighted linear regression methods ( Lowess ) . Feature intensities of the three biological replicates were averaged . T-test p-values were calculated , and normalised data was filtered to exclude data with p-values greater than 0 . 05 . Data was further filtered to include only genes showing 2-fold and greater regulation . Candidate genes were selected based on several criteria , including gene ontology , and known roles of orthologous genes . Microarray data has been submitted to the open access Vectorbase database ( www . vectorbase . org ) . Primers were designed ( Table S3 ) for 200–600 bp sections of genes of interest , with a T7 promotor sequence ( GAATTAATACGACTCACTATAGGGAGA ) added to their 5′ ends . Polymerase chain reaction ( PCR ) was carried out using cDNA derived from A . gambiae mosquitoes and PCR products were sequenced to confirm correct amplification for each probe . PCR amplicons were used to synthesise dsRNA using the T7 MEGAscript kit ( Ambion ) . Concentration of dsRNA was adjusted to 3 µg/µl and stored at −80°C until use . P . berghei ANKA clone 259c12 was maintained in Theiler's original mice ( Harlan , UK ) as described in [20] . All animal work was carried out by Dr Tibebu Habetewold and Kasia Sala . Mice were infected by intraperitoneal ( IP ) injection of 100–200 µl of P . berghei infected blood . For mosquito infections , three days after passage with infected blood mice were terminally anaesthetised with an intramuscular ( IM ) injection of 0 . 05 ml/10 g body weight of Rompun ( 2% stock solution , Bayer ) , Ketastet ( 100 mg/ml ketamine , Fort Dodge Animal Health Ltd ) and PBS in a 1∶2∶3 ratio . Newly emerged adult G3 mosquitoes were intrathoracically inoculated with ∼1640 PFU 5′ONNVic-eGFP . Inoculated mosquitoes were maintained at 27°C for 48 h . Mosquitoes were starved of sugar for 4–5 hours prior to blood feeding . Mosquitoes were fed on a terminally anaesthetised P . berghei infected mouse , maintained at 19°C for 72 h post blood feeding to allow successful parasite development and were subsequent maintained at 27°C to allow for optimal viral replication . Unfed mosquitoes were removed between 24 and 48 h post blood feeding , when the blood bolus is clearly visible through the abdomen of the mosquito . Seven days post blood feeding , mosquito midguts were dissected and fixed in 4% PFA . Fixed midguts were mounted in Vectorshield ( Vectorlabs ) on glass slides with sealed coverslips . Live oocysts expressing GFP were counted using fluorescence , and melanised ookinetes were counted using light microscopy . For plaque assay experiments and P . berghei oocysts/ookinete quantification , results were subject to the Man Whitney U-test for statistical significance . Significance was accepted where P<0 . 001*** , P<0 . 01** , P<0 . 05* . For analysis of changes in P . berghei melanisation prevalence , results were subject to the Chi Squared test for statistical significance , where P<0 . 001*** . Statistical significance in microarray experiments was calculated using the T-test comparing normalised ( Lowess ) expression data . Differential regulation was considered were fold change in expression was greater than 2 and P<0 . 05 over three biological replicates . This study was carried out in strict accordance with the United Kingdom Animals ( Scientific Procedures ) Act 1986 . The protocols for maintenance of mosquitoes by blood feeding and for infection of mosquitoes with P . berghei by blood feeding on parasite-infected mice were approved and carried out under the UK Home Office License PLL70/6347 awarded in January 2008 and PPL70/7185 awarded in November 2010 . The procedures are of mild severity and the numbers of animals used are minimized by incorporation of the most economical protocols . Opportunities for reduction , refinement and replacement of animal experiments are constantly monitored and new protocols are implemented following approval by the Imperial College Ethical Review Committee . The experimental procedures for the ONNV work were approved by the HSE and the Imperial College GM Safety Committee .
Using an infectious clone of ONNV encoding enhanced GFP ( 5′ONNVic-eGFP ) under the control of a duplicated viral subgenomic promoter ( provided by B . D . Foy , AIDL , Colorado state University [19] ) , we characterized infection of ONNV in adult G3 A . gambiae mosquitoes . Adult mosquitoes were intrathoracically inoculated with ∼1640 PFU/mosquito . Viral RNA ( vRNA ) was extracted from 10 pooled mosquitoes every day over 9 days and quantitative real-time PCR ( qrt-PCR ) was used to calculate the viral genome copy number/mosquito ( Figure 1A ) . Viral titre increased slowly until 5 days post infection ( DPI ) , when infection rapidly increased , peaking at 7–8DPI , and then subsequently decreased to low levels at 9DPI . Plaque assays using individual mosquitoes at 7DPI showed that the prevalence of infection was ∼90% ( data not shown ) . GFP expression was also monitored at 1 , 4 and 9 DPI by fluorescence microscopy of live , cold anaesthetized mosquitoes ( Figure 1B ) . GFP expression , most commonly visible through the eyes ( Figure 1C ) and occasionally through the thorax , was visible in only ∼20% of mosquitoes at 4DPI and ∼25% of mosquitoes at 9DPI . The discrepancy in infection prevalence between plaque assays and GFP observations is attributed to the mosquito cuticle that provides a barrier to GFP detection and together with strong autofluorescence leads to underestimation of the infection prevalence in whole mosquitoes . In dissected mosquitoes at 7DPI , patterns of infection and tissue tropism were in agreement with those previously published using the same strain of A . gambiae mosquitoes and the same infectious clone of ONNV [19] . Additionally GFP expression was commonly seen in the midgut musculature of infected mosquitoes ( Figure 1D ) similar to what has been previously observed in other vector-alphavirus combinations [22] , [23] . In order to investigate the responses of A . gambiae to systemic viral infection , we utilised a genome-wide microarray platform to profile gene expression during a time-course of ONNV infection of the hemocoel . Three time points were selected for analysis: 1DPI ( representing initial introduction of virus into the hemocoel ) , 4DPI ( where virus has replicated , is being released from infected cells and is infecting new tissues ) and 9DPI ( where infection levels have significantly dropped ) . Transcriptional profiling of whole mosquito homogenates from infected versus mock-infected mosquitoes revealed a large number of viral responsive genes . Initial exposure to virus ( 1DPI ) triggered the differential regulation of 66 genes ( 53 upregulated and 13 downregulated ) , increasing to 211 genes ( 119 upregulated and 92 downregulated ) at 4DPI and dropping to 23 genes ( 20 upregulated and 3 downregulated ) at 9DPI ( Figure 2 ) . A full list of regulated genes is presented in Table S1 . Genes were grouped into functional categories based on gene ontology ( GO ) terms , orthologous gene function and literature reviews . These categories span a wide range of cellular and physiological processes including metabolism , RNA degradation , signalling , and cell division; however , the most striking category pertains to genes with putative immune functions , particularly at 1DPI ( 30% of regulated genes at 1DPI , 18% at 4DPI and 26% at 9DPI ) . Overall , 45 genes with putative immune functions were differentially regulated following ONNV infection . Grouping these genes based on gene ontology and putative function ( Table S2 ) revealed genes with roles in all aspects of immunity , including pathogen recognition , complement-like proteins , immune signalling pathway components , humoral cascade regulators and effector genes . The majority of genes ( 39/45 ) were upregulated , consistent with the hypothesis that viral infection triggers immune signalling in A . gambiae . A surprisingly small number of genes from immune signalling pathways known to respond to viral infection in other invertebrates were differentially regulated; comparison of viral responsive genes with those downstream of the Toll and IMD pathway in A . gambiae [24] and the JAK/STAT pathway in A . gambiae ( unpublished data ) demonstrated very little overlap in gene expression indicating that these pathways are not activated by ONNV infection . In fact at 4DPI genes involved in the RNAi , JAK/STAT and IMD pathways were downregulated suggesting inhibition of these signalling pathways . Downregulated genes encode the Tudor-SN ( TSN ) , a component of the RNAi RISC complex , the janus kinase HOP of the JAK/STAT pathway , and IKKb , a positive regulator of the IMD pathway . In contrast to decreased IKKb transcripts , DPT and CEC3 ( AMPs thought to be downstream of the IMD pathway ) were upregulated at 1DPI and 4DPI . Genes encoding putative recognition receptors were upregulated at 1DPI and 4DPI; two MD2-like receptors ( ML1/9 ) , three galectins ( GALE6-8 ) , one fibrinogen-like protein ( FREP50 ) and GNBPB1 all increased in transcript abundance . Additionally a large number of genes encoding proteins implicated in humoral immunity were upregulated , consisting of LRIMs and complement-like Thioester-containing proteins ( TEPs ) . Different LRIMs and TEPs were regulated during the different phases of infection; LRIM1/4 and TEP5 at 1DPI; LRIM7 and TEP4/9/10/12 at 1DPI and 4DPI; LRIM10 and TEP14 at 4DPI and LRRD7 at 4DPI and 9DPI . A number of clip-domain serine proteases and their inactive homologs ( CLIPs ) and C-type lectins ( CTLs ) were upregulated including two known inhibitors of melanisation ( CTLMA2 and CLIPA2 ) . The roles of the other regulated CLIPs and CTLs are not known , although they probably function in the modulation of signalling that regulates humoral responses . Genes encoding additional putative immune effectors were upregulated , the majority of which at 4DPI , including two hydrogen peroxidases , a glutathione peroxidase and three lysozymes . Additionally apoptosis related genes also responded to viral infection . Downregulation of the inhibitor of apoptosis-1 ( IAP1 ) and upregulation of Caspase-6 ( CASPS6 ) at 4DPI suggests that apoptosis may be triggered . Transcriptional profiling highlighted immune genes that respond to infection , however , whether these genes have genuine antiviral functions could not be inferred from expression profiling alone . To identify genes that have roles in A . gambiae antiviral immunity we developed an RNAi and qrt-PCR based assay to measure the effects of gene knockdown ( KD ) on viral titres . Mosquitoes were co-inoculated with dsRNA corresponding to a gene of interest and ∼3000 PFU of 5′ONNVic-eGFP . The viral RNA genome copy number per mosquito was calculated 7DPI using qrt-PCR . 19 genes were selected from the viral responsive immune genes identified in our transcriptional analysis and from the classical immune signalling pathways . DsRNA corresponding to AgAGO2 and the ONNV nsP3 gene were included as positive and negative controls respectively , while dsRNA corresponding to the Escherichia coli LacZ gene was used as a reference to calculate percentage changes in viral infection loads . As expected KD of AgAGO2 resulted in increased 5′ONNVic-eGFP titres and nsP3 silencing resulted in decreased 5′ONNVic-eGFP titres , indicating that our screening method was accurate ( Figure 3 ) . Interestingly , silencing of genes from the JAK/STAT and the Toll pathways ( HOP , STAT1-2 , PIAS , REL1 , CACT ) , which are known to be involved in antiviral immunity in Ae . aegypti and D . melanogaster , as well as of the IMD pathway ( REL2 ) did not have an effect on 5′ONNVic-eGFP titres or yielded data that were highly variable and therefore inconclusive . However , 5 of the 10 tested viral responsive immune genes identified in our gene profiling experiments appeared to be viral antagonists and were selected for further investigation: the putative recognition receptors ML1 and GALE8 , and the antimicrobial peptides LYSC4 , LYSC6 and CEC3 . In order to confirm that the 5 genes identified in our qrt-PCR screen have a statistically significant impact on ONNV infection we carried out plaque assays to measure the viral titres in individual gene silenced mosquitoes . Silencing 4 of the 5 tested genes consistently affected viral titres ( Figure 4 ) . ML1 KD resulted in a 6 . 2 fold increase ( P<0 . 0001 ) , LYSC4 KD resulted in a 6 fold increase ( P<0 . 0001 ) , LYSC6 KD resulted in a 5 . 4 fold increase ( P<0 . 001 ) and GALE8 KD resulted in a 2 fold increase ( P = 0 . 0163 ) in median viral titres at 7DPI . The prevalence of infection was similar for all genes tested ( LacZ 89% , ML1 KD 95% , GALE8 KD 95% , LYSC4 KD 89% and LYSC6 KD 93% ) . Qrt-PCR was used to confirm the reduction of ML1 ( 91% ) , GALE8 ( 80% ) and LYSC4 ( 82% ) transcripts in mosquitoes after dsRNA treatment and to confirm the transcriptional profile of these genes after systemic infection ( Figure S1 ) ; LYSC6 was not assayed . Several recent studies have shown that the A . gambiae humoral immune system exists in a delicate state of balance that can be diverted to lysis or melanisation against malaria parasites [25] , [26] . To investigate whether the differential regulation of several humoral immune factors observed during infection with ONNV , including CLIPs , CTLs , LRIMs and TEPs , can influence this balance we utilised A . gambiae co-infections with ONNV and the rodent malaria , P . berghei . Newly emerged female mosquitoes were injected with ∼1640 PFU 5′ONNVic-eGFP or mock infected and blood-fed 48 h later on a mouse infected with P . berghei . This experimental design resulted in parasite ookinetes traversing the midgut wall and entering the hemolymph approximately 4 days post ONNV infection . Seven days post blood-feeding the mosquito midguts were dissected and parasite oocysts and melanised ookinetes were counted . Additionally , midguts were scored for ONNV infection of the midgut musculature . Figure 5A shows the oocyst and melanised ookinete distribution in virally infected and mock infected mosquitoes . The results revealed that priming with ONNV results in an approximately 40% reduction in the number of live oocysts in the virally infected mosquitoes showing viral midgut infections , although this decrease was not statistically significant using the Man Whitney test . However , a statistically significant decrease in the numbers of melanised ookinetes was observed ( P = 0 . 001 ) . The prevalence of melanised ookinetes also significantly decreased from 35% in mock infected to 18 . 5% in virally infected mosquitoes ( Figure 5B ) ( P<0 . 001 using the Chi squared test ) .
Viruses as obligate intracellular pathogens represent a unique challenge to the immune system and require sophisticated mechanisms of recognition and targeting . Completing their lifecycle within host cells and use of host cell membranes limits the number of signatures that the immune system can recognise as non-self . In this study we have attempted to elucidate the components of the immune system employed by A . gambiae mosquitoes to target ONNV through transcriptional profiling of infected mosquitoes and gene silencing experiments . Our observations of slow viral replication and restricted tissue tropism within the A . gambiae mosquito host are consistent with those observed for ONNV infections in the past [19]; it has been suggested that the robust RNAi response observed in A . gambiae mosquitoes [27] may contribute to the poor vectorial capacity of these mosquitoes in comparison with other typical vector-virus combinations [6] . Although not carried out in this study , dissemination rates for ONNV infection in A . gambiae have also been shown to be low compared to other vector-virus combinations [19] . This indicates that A . gambiae is a poor vector of ONNV , and may not be the natural vector of the disease outside of epidemics [28] . Our transcriptional profiling of ONNV infected mosquitoes has identified a large number of viral responsive genes . A significant proportion of these genes have no known function , indicating that A . gambiae may utilise non-classical immune mechanisms to target viral infection . Of the putative immune genes that were viral responsive , a perhaps surprisingly small number were associated with the Toll or JAK/STAT pathways . It has been shown in D . melanogaster and Ae . aegypti that both the Toll and JAK/STAT pathways have important roles in antiviral immunity [4] , [5] , [29] . In our study ONNV infection fails to induce expression of components of these two pathways; in fact HOP is downregulated at 4DPI . Additionally , there is very little overlap between viral responsive genes and those known to be downstream of the Toll pathway in A . gambiae [24] and genes identified through microarray analysis of HOP KD A . gambiae mosquitoes ( unpublished data ) . Indeed through our gene silencing experiments we have observed that activating or inhibiting both pathways has little effect on ONNV titres . These data indicate that not only does systemic ONNV infection fail to trigger Toll and JAK/STAT signalling , but that genes downstream of these two pathways do not target ONNV infection . A possible contributing factor to the surprising difference observed between Ae . aegypti and A . gambiae responses to viral infection may be the route of infection used in the experimental designs . The JAK/STAT and the Toll pathways have been shown to be important in regulating flavivirus DENV midgut infection after an infectious blood meal , which is the natural route of infection . In our study , intrathoracic inoculation was used to infect mosquitoes with ONNV in order to overcome the restricted tissue tropism and very limited dissemination rates observed following oral infection with ONNV , thus achieving higher levels and prevalence of systemic infection . Although a role of the JAK/STAT and the Toll pathway in midgut defense against ONNV is possible , our results demonstrate that once the virus enters the mosquito homocoel , immune responses other than the JAK/STAT and the Toll pathways are involved in regulating systemic antiviral immunity . Other studies investigating immune responses of insects to alphavirus infection have also suggested or demonstrated that the JAK/STAT and Toll signalling pathways do not target alphaviral infection . Sanders et al [11] conducted microarray analysis of SINV infected Ae . aegypti; despite suggesting a role for the Toll pathway during early infection , they found no differential regulation of JAK/STAT or Toll pathway components . Fragkoudis et al ( 2008 ) [30] found that SFV in all likelihood does not trigger classical immune signalling pathways in Ae . albopictus cells , and in fact infection inhibits activation of the JAK/STAT , Toll and IMD pathways , probably through reducing host cell gene expression . The difference observed between immune gene regulation and function using Ae . aegypti/DENV and A . gambiae/ONNV may also be a feature of flavivirus verses alphavirus infection respectively . In addition to Toll signalling , a second NF-κB signalling pathway , the IMD pathway , regulates numerous genes that target Plasmodium parasites and bacteria [24] , [31] . Our expression profiling shows that IKKb , a positive regulator of the IMD pathway , is downregulated at 4DPI , suggesting that the pathway may be inhibited . In contrast , upregulation of LRIM1 and CEC3 , which are downstream targets of the A . gambiae IMD pathway [31] , as well as of a homolog of the AMP Diptericin that is downstream of the IMD pathway in Drosophila [32] , may suggest activation of the pathway . Indeed a previous study has demonstrated that the IMD , but not the Toll pathway , has an antiviral function during SIN infection of Drosophila [33] . It is possible that the IMD , and not the Toll pathway , responds to alphavirus infection , with the reverse being true for flavivirus infection . Nevertheless , silencing REL2 , the NF-κB factor of the IMD pathway , has no significant effect on ONNV titres , suggesting that the contribution of this pathway to ONNV infection , whether inhibited or activated , is minor . RNAi has been demonstrated in a number of invertebrates to target and limit viral infection [6]–[10] . Our transcriptional profiling does not show any induction of expression of RNAi components , similar to observations of D . melanogaster infected with DCV [29] and Ae . aegypti infected with DENV [4] . Presumably the components of the RNAi pathway are constitutively expressed to levels sufficient to target replicating viruses . We confirm that RNAi is a key antiviral mechanism in A . gambiae mosquitoes through silencing of the nsP3 viral gene and inhibition of RNAi by silencing AgAGO2 . Interestingly transcriptional profiling revealed that TSN , a component of the RISC complex in the classical RNAi pathway , is downregulated at 4DPI . It is possible that this downregulation is mediated by ONNV , as inhibition of RNAi would be advantageous for infection . ONNV and other alphaviruses are thought not to encode a direct suppressor of RNAi [34] as seen in other insect viruses such as the B2 protein of the FHV ( Flock House Virus ) [35]; however , it is possible that viral gene products may interfere with host gene expression . For example , 90% of Semliki Forest Virus ( SFV; a closely related alphavirus ) nsP2 protein localises to the nucleus of infected cells where its function is unknown but could modulate expression of host genes [36] . In addition to the downregulation of TSN , two further genes involved in the miRNA and piRNA pathways , DCR1 and AgAGO5 , and a group of DEAD-box helicases are downregulated at 4DPI . There is evidence that the piRNA pathway has antiviral functions in Drosophila , as Piwi mutant flies are more susceptible to WNV than wild-type flies [37] . The DEAD-box helicases , although having diverse functions in RNA metabolism , are closely related to RIG-I and the RIG-I-like receptors ( RLRs ) that have well defined roles in viral RNA recognition in mammalian systems [38] . In addition , the human DEAD-box helicase DDX3X also has antiviral roles , and multiple viruses have been shown to interact with this protein and modulate its function [39] . The downregulation of these genes suggests an intriguing function in antiviral immunity . However , silencing of three helicases , including DCR1 , has no effect on ONNV titre . In addition to the immune signalling pathways , A . gambiae mosquitoes have a humoral branch of the immune system that recognises and eliminates invading pathogens . A ternary complex of two proteins of the LRIM family , LRIM1 and APL1C , and the complement-like TEP1 has been shown to target invading Plasmodium parasites for lysis or melanisation [25] , [40] , [41] . As our transcriptional analysis showed that LRIM1 as well as several other members of the LRIM and TEP families are upregulated following infection , we investigated whether these humoral responses triggered by viral infection can interfere with Plasmodium infections . ONNV and P . berghei co-infections of A . gambiae were timed such that parasites traverse the midgut and enter the hemolymph 4 days after ONNV infection , thus encountering virus-induced humoral immune responses in the hemolymph . Our results reveal that parasite melanisation is significantly inhibited in the presence of ONNV . These results are consistent with our transcriptional analysis that shows upregulation of two important negative regulators of melanisation at 4DPI: CTLMA2 and CLIPA2 [42] , [43] . An observed simultaneous decrease in the number of surviving parasites is not dramatic suggesting that upregulation of LRIM1 alone is not sufficient to cause significant parasite lysis , and that parallel upregulation of APL1C and the complement effector protein TEP1 would be needed . However , among additional TEPs upregulated by ONNV infection are TEP4 and TEP9 , which have been recently shown to also form complexes with LRIM1 [44] . It remains to be investigated whether the alternative LRIM1/TEP complexes promote antiviral responses such as virus lysis or clearance of infected cells . Four novel ONNV antagonists have been identified through our RNAi screen; ML1 , GALE8 , LYSC4 and LYSC6 . ML1 is one of two MD-2 like receptors upregulated by ONNV infection and a known P . falciparum antagonist [45] . The MD2 protein forms part of an LPS sensing mechanism in mammals [46] , [47] . In addition to responses to LPS , MD2-TLR4 signalling triggers the expression of pro-inflammatory cytokines in response to ebola envelope protein [48] . TLR4 signalling has also been linked to several other viral infections including the vesicular stomatitus virus , respiratory syncytial virus , mouse mammary tumor virus [48] and Kaposi sarcoma herpesvirus [49]; however the role of MD2 in these interactions is not clear . A TLR binding partner for the MLs has yet to be identified in flies and mosquitoes . One hypothesis is that the ML proteins may act as an extracellular surveillance system that recognise viral PAMPs and lead to signalling via a Toll receptor . At least 10 Toll receptors have been identified to date in A . gambiae , but their role in immune responses is yet unclear [50] . Beta-galactoside binding galectins are found in many organisms and display a complex repertoire: the multiple isoforms and their observed plasticity in sugar binding suggests substantial diversity in their glycan recognition properties [51] . There are 10 putative galectins in A . gambiae [50] , 3 of which are upregulated by ONNV infection , GALE6-8 . All three galectins are part of a mosquito specific expansion of the Galectin family , including also GALE4 and GALE5 . This expansion maybe due to the haematophagous lifestyle of mosquitoes , and subsequent exposure to a disparate group of blood-borne pathogens compared to D . melanogaster , including viruses . Therefore , the upregulation of several galectins in response to ONNV infection suggests that this group of mosquito specific galectins have antiviral roles . Galectins are known to function at several levels of antiviral defence , from initial recognition and blocking of envelope and fusion glycoproteins to the activation and amplification of the innate and adaptive immune responses [51] . In mammals , Galectin 1 cross-links the N-glycans displayed in the envelope proteins of Nipah and Hendra viruses ( paramyxoviruses that induce syncytia in infected cells ) , directly blocking cell infection and cell-cell fusion [51] . In addition , Galectin expression is regulated by Herpesvirus 1 , Newcastle disease [52] , Epstein Barr Virus [53] , Hepatitus C virus [54] and Human papiloma virus ( HPV ) [55] while Galectin 3 secretion and carbohydrate binding increase upon Herpesvirus 1 infection [56] . Three lysozymes are also induced by ONNV infection and two of those , LYSC4 and LYSC6 , appear to have antagonistic effects against ONNV infections . Although lysozymes are classical antibacterial proteins that function through perturbation of cell membranes [57] , many of them are also shown to have antiviral immune functions including human urinary Lysozyme C , a lysozyme from chicken egg whites , human milk lysozyme and human neutrophil lysozyme , all of which have anti-Human Immunodeficiency Virus ( HIV ) activity [58] . In addition , a lysozyme from a marine organism is shown to inhibit Pseudo Rabies Virus ( PRV ) growth in cell culture [59] . Although , the mechanism of lysozyme antiviral activity is not clear , it is likely that they also function through membrane perturbation . In summary , this study is the first step in elucidating the antiviral mechanisms of A . gambiae mosquitoes , and has revealed interesting differences between A . gambiae and other invertebrates . The finding that two pathways with known antiviral roles in other invertebrate-virus systems do not significantly modulate systemic ONNV infection indicates that A . gambiae may use other immune mechanisms to recognise and fight viral infections . Our data suggest that these mechanisms involve the complement-like branch of the humoral immune response , and that the melanisation response that is prominent in anti-parasitic immunity is suppressed . The antiviral immune response in A . gambiae is thus composed of some key conserved mechanisms to target viral infection such as RNAi but includes other diverse and possibly species-specific mechanisms . | Mosquito-borne viral diseases are found across the globe and are responsible for numerous severe human infections . In order to develop novel methods for prevention and treatment of these diseases , detailed understanding of the biology of viral infection and transmission is required . Little is known about invertebrate responses to infection in mosquito hosts . In this study we used a model system of Anopheles gambiae mosquitoes and O'nyong-nyong virus to study mosquito immune responses to infection . We examined the global transcriptional responses of A . gambiae to viral infection of the mosquito blood equivalent ( the hemolymph ) identifying a number of genes with immune functions that are switched on or off in response to infection , including complement-like proteins that circulate in the mosquito hemolymph . The switching on of these genes combined with co-infection experiments with malaria parasites suggests that viral infection inhibits the melanisation pathway . Through silencing the function of a selection of viral responsive genes , we identified four genes that have roles in A . gambiae anti-viral immunity; two putative recognition receptors ( a galectin and an MD2-like receptor ) ; two effector lysozymes . These molecules have previously non-described roles in antiviral immunity , and suggest uncharacterised mechanisms for targeting viral infection in A . gambiae mosquitoes . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"virology",
"immunology",
"biology",
"microbiology"
] | 2012 | Anopheles gambiae Antiviral Immune Response to Systemic O'nyong-nyong Infection |
Taenia solium is a zoonotic cestode parasite which causes human neurocysticercosis . Pigs transmit the parasite by acting as the intermediate host . An intervention was implemented to control transmission of T . solium by pigs in Dalit communities of Banke District , Nepal . Every 3 months , pigs were vaccinated with the TSOL18 recombinant vaccine ( Cysvax , IIL , India ) ) and , at the same time , given an oral treatment with 30mg/kg oxfendazole ( Paranthic 10% MCI , Morocco ) . The prevalence of porcine cysticercosis was determined in both an intervention area as well as a similar no intervention control area , among randomly selected , slaughter-age pigs . Post mortem assessments were undertaken both at the start and at the end of the intervention . Participants conducting the post mortem assessments were blinded as to the source of the animals being assessed . At the start of the intervention the prevalence of porcine cysticercosis was 23 . 6% and 34 . 5% in the control and intervention areas , respectively . Following the intervention , the prevalence of cysticercosis in pigs from the control area was 16 . 7% ( no significant change ) , whereas no infection was detected after complete slicing of all muscle tissue and brain in animals from the intervention area ( P = 0 . 004 ) . These findings are discussed in relation to the feasibility and sustainability of T . solium control . The 3-monthly vaccination and drug treatment intervention in pigs used here is suggested as an effective and practical method for reducing T . solium transmission by pigs . The results suggest that applying the intervention over a period of years may ultimately reduce the number of tapeworm carriers and thereby the incidence of NCC .
Neurocysticercosis is a serious medical condition caused by infection in the brain or other nervous tissue with the larval stage of the parasite Taenia solium . The life cycle of the parasite involves pigs and humans in a prey-predator cycle . Humans are the obligatory definitive host for T . solium and harbour the adult tapeworm in the small intestine . Tapeworm eggs are released with the faeces and , if they are ingested by pigs , the larval cysticercus stage develops , principally in the muscle tissues . The life cycle is completed when humans eat insufficiently cooked , infected pig meat , leading to the development of a tapeworm . The serious medical consequences of T . solium infection arise because the eggs released by a tapeworm carrier are not only infective for pigs but can also cause cysticercosis if accidentally ingested by humans . In humans the cysticercus larvae commonly encyst in the brain , causing neurocysticercosis , a frequent symptom of which is epilepsy . The full life cycle of T . solium is perpetuated where sanitation conditions are poor , pigs have access to human faeces or food contaminated with human faeces , and where pork is ingested raw or poorly cooked . Hence , the full life cycle of T . solium is restricted to populations living in many of the poorest countries of the world . Encystment of T . solium in the brain of humans is responsible for the parasite causing 29% of seizure cases in areas where T . solium transmission occurs [1] . Human cysticercosis is one of a small number of diseases that have been formally recognised as being capable of being eradicated [2] . Improvements in sanitation and pig rearing practices in developed countries have led to a cessation in T . solium transmission , however attempts to institute cysticercosis control measures in poor communities have had limited success [3] . Control measures for T . solium that have been evaluated include treatment of human taeniasis cases with niclosamide or praziquantel , improvement in sanitation and other practices through public education , vaccination and medication of pigs , and improvement in pig rearing and meat inspection practices [3] . A major limitation to achieving a long-lasting reduction in neurocysticercosis has been the sustainability of T . solium control activities . Lightowlers and Donadeu [4] presented a logical model for control of T . solium transmission by pigs using a combination of vaccination and medication . Combined use of the TSOL18 vaccine [5] and oxfendazole treatment [6] in all pigs at 3-monthly intervals was predicted to lead to a cessation of T . solium transmission by slaughter-age pigs within a year of initiation of the program [4] . T . solium neurocysticercosis is a major medical concern in Nepal where it has been determined to cause the highest burden of disease due to a parasitic infection [7] . Sah et al . [8] identified the Banke District in mid-western Nepal as having a high prevalence of porcine cysticercosis . Recently two new commercially produced and registered products manufactured according to Good Manufacturing Practice guidelines have become available for control of cysticercosis in pigs—the TSOL18 vaccine ( Cysvax ) produced by Indian Immunologicals Limited , India , and Paranthic 10% an oxfendazole formulation manufactured by MCI Santé Animale , Morocco , which is specifically registered for the treatment of porcine cysticercosis . These new products provide the opportunity for an assessment of a 3-monthly pig vaccination and treatment program on T . solium transmission . Here we describe the impact of these T . solium control measures implemented in pigs in the Banke district of Nepal .
The study was conducted in Udaypur Village Development Committee ( VDC ) and Hirminiya & Betahani VDC of the Banke district in Nepal ( 81°37’E to 81°42’E , 27°90N to 28°20’N ) . The objective was to determine changes in the prevalence of porcine cysticercosis in slaughter-weight pigs following implementation of a 3-monthly vaccination and medication program . The investigation was a pilot study , controlled , community-level , prospective intervention undertaken in the pig population . Two distinct areas were involved each containing a total of at least 200 pigs and assigned at random as a non-intervention area or an area where a combined intervention involving TSOL18 vaccination and oxfendazole treatment of pigs was undertaken . The intervention area ( Udaypur VDC ) comprised a total of 115 households among the Dalit community which kept pigs , while the control area ( Betahani/Himiniya VDC ) comprised 118 Dalit households . The numbers of pigs in the two areas in September 2017 was 279 ( Udaypur ) and 218 ( Betahani/Himiniya ) . The study was designed to be able to identify an 80% reduction in the prevalence of porcine cysticercosis in slaughter-weight pigs . Sample size calculations were undertaken using a one-sided likelihood ratio test at the 5% significance level using SAS 9 . 3 ( SAS Institute , Cary North Carolina , USA ) with the TWOSAMPLEFREQ command in the PROC POWER procedure . Assuming an initial prevalence of infection of 20% , sample sizes of 55 animals were required in each area at the start and end of the trial in order to meet the desired statistical power . The eligibility criteria for animals that were enrolled in the study were indigenous breed pigs >8 weeks of age , not heavily pregnant and not clinically ill . In the treatment area , animals that were destined for slaughter within 3 weeks were excluded in compliance with the withholding period for the oxfendazole formulation that was used [9] . At both the start and the end of the trial , a random selection of enrolled slaughter weight animals was assessed for cysticercosis by necropsy examination ( detailed below ) , as this is the only reliable , sensitive and accurate method that is available to diagnose porcine cysticercosis [10] . In the intervention area , the day of first treatment administration to pigs was defined as Day 0 . Pigs were enrolled and received their first treatment administration within a target period of 15 days . Subsequent administrations of treatments to pigs , including enrolment and treatment of new pigs , occurred at intervals of three months and on each occasion were completed within a target period of 15 days . The study had a duration of 12 months and hence involved 4 interventions . The study was approved by the Nepal Veterinary Council and was conducted adhering to the Council’s requirements for animal husbandry . For the animals that met the eligibility criteria , written informed consent was obtained from the farmers . In the intervention area 114 pig rearing households agreed to participate; three households declined to participate . In both the intervention and control areas , detailed information was recorded about farming and animal management practices . Farmers provided , or estimated , the age of every pig . In the intervention area , farmers were questioned about the origin and age of all new pigs and also about what had happened to any animal that had been present for one intervention but absent at a subsequent intervention visit . Approximately 30% of piglets died before the age of 3–4 months . Classical Swine Fever ( CSF ) also caused significant mortality . A minority of farmers vaccinate with commercial CSF vaccine–this generally being the only veterinary attention that the animals received . The intervention team consisted of a registered veterinary doctor , two veterinary technicians and 2–4 pig catchers . After obtaining the consent of the owner , all pigs meeting the enrolment criteria were caught and numbered tags applied to both ears . Animal weight in kilograms was estimated using a measuring tape and the formula [Girth2 x Length/400]/2 . 2 . The dose for oxfendazole ( 3ml/10kg Paranthic 10% , MCI Sante Animale , Morocco ) was calculated according to the animal weight ( 30mg/kg body weight ) and was applied per os . Concurrently , 1ml TSOL18 vaccine ( 150μg TSOL18 recombinant protein in mineral oil adjuvant; Cysvax , Indian Immunologicals Limited , India ) was administered intramuscularly in the left side of the neck behind the base of the ear , prior to release of the animal . A different needle was used for every vaccination . The procedures were undertaken swiftly and efficiently in order to minimize stress on the animals . Animals having a weight consistent with that at which pigs are commonly sold or slaughtered in the communities were selected at random and purchased . Slaughter weight was determined through advice from the farmers; the mean weight of animals that the farmers indicated were available for slaughter ( and which were purchased for post mortem analyses ) was 70kg , although individual animal weights ranged from 35kg to more than 175kg . Necropsy procedures undertaken on 110 slaughter-weight pigs at the start of the intervention are described by Sah et al . [8] . Similar procedures were undertaken for the post mortem analyses at the end of the trial but with some variations , as follows . All staff involved in the post mortems were blinded as to whether the animals were from the control or intervention areas . Animals from both areas were necropsied in random order . The animals were transported to a licensed commercial abattoir in Nepalgunj Municipality , Banke where they were euthanized by slaughter house staff according to normal commercial practice processes . The viscera were removed and the heart , liver , lungs , both kidneys and the full diaphragm retained in numbered containers . The organs and the two halves of the carcase , including the head , were refrigerated overnight at 4°C , after which the carcase was skinned . The tongue , masticatory muscles ( both right and left sides ) and brain were removed and retained . The muscles from each side of the carcase were dissected from the bones and kept separately . Except in cases of very heavy infection , all the retained organs and muscles of the right hand side of the carcase were sliced by hand at intervals of approximately 3mm and examined meticulously for the presence of T . solium cysticerci or other lesions . During the necropsies undertaken at the end of the trial , when no cysticerci were detected in the tongue , masticatory muscles , diaphragm , brain or muscles from the right hand side of the carcase , the muscles of the left hand side of the carcase were also sliced . Cysticerci were recorded as viable when they appeared as translucent vesicles filled with transparent fluid and having a visible white scolex . Non-viable lesions were recorded separately in cases where vesicles were non-translucent , containing a dense white or yellowish fluid and having no scolex and in cases of fibrosed or calcified lesions . Suspect , non-viable lesions that were not calcified were placed into RNA-later ( Sigma ) and investigated by PCR analysis of a fragment of the mitochondrial 12S rDNA gene using the restriction enzymes DdeI and HinfI or HpaI , as described by Rodriguez-Hidalgo et al . [11] , Devleesschauwer et al . [12] and Dermauw et al . [13] . In carcases that contained thousands of cysts , all of the heart , liver , kidneys , lungs , diaphragm , tongue , masticatory muscles and brain were sliced and counted as above . The remaining carcase musculature was weighed and representative samples from different muscle sites were selected representing approximately 1kg . This sample was weighed accurately and then sliced and counted as above and the number of cysts in the carcase muscles estimated from the total muscle weight . The definition of a confirmed case of cysticercosis which was adopted by Sah et al . [8] was also used here . An animal was determined to be a confirmed case of porcine cysticercosis if one or more viable T . solium cysticerci was found in the muscle and or the brain , or if more than one non-viable lesion was detected in the muscles and/or brain . Animals having only non-viable lesions in organs that are not typical locations for T . solium ( eg the liver , lungs or kidneys ) , and which could not be confirmed as being T . solium by DNA analyses , were excluded . Direct comparisons of infection prevalence at the start and end of the trial were made on the basis of infections detected in the various organs examined as well as cysts found in the right hand side of the carcase , as this was the procedure used for the post mortems undertaken at the start of the trial [8] . Raw data was transcribed into pre-formatted Excel spreadsheets suitable for importation into Genstat 18th edition . Statistical analysis was undertaken to evaluate the effect of treatments on the prevalence of T . solium cysts at post mortem . Prevalence results were compared within and between study groups , at baseline and end of study , using a two-sample binomial test . A generalised linear model with logit link function ( logistic model ) for binary data was also used to confirm results and to provide standard error estimates and confidence intervals around prevalence figures ( Genstat 18th edition ) .
At the start of the trial the total pig population in the control and intervention areas was 805 pigs . In the intervention area there were 313 pigs in total , of which 227 met the inclusion criteria . In the intervention area , a total of 4 rounds of pig vaccination and oxfendazole treatments were carried out between August 2016 and May 2017 . A total of 585 pigs were treated during the trial , with 209 , 209 , 207 and 203 receiving treatment in first , second , third and fourth interventions , respectively ( Tables 1 and 2 ) . Interventions delivered to individual pigs are summarized in Table 3 . Of the 209 animals receiving the vaccination and drug treatment during the first intervention period , 95 animals were absent at the time of the second intervention . The reasons for absence were: unable to be caught ( 12 animals ) , 2 had died , 32 had been sold or consumed locally , and 33 were otherwise ineligible for inclusion in the second round of interventions ( 4 were ill; 29 in late pregnancy ) . The remaining 16 animals were absent for unknown reasons although several were present and treated at a subsequent intervention time . Thirteen animals were not available for the second round of interventions but were present and received treatment either during the third intervention period ( 12 animals ) or fourth intervention period ( 2 animals ) . Three animals which were vaccinated and drug treated at the second intervention were absent at the third intervention period but were present and treated during the fourth intervention . The number of animals which were unable to be caught decreased as the trial progressed such that approximately 95% of eligible animals received their vaccinations and oxfendazole treatments during the third and fourth rounds of intervention . No adverse reactions to either vaccination or oxfendazole treatment were noted by the field staff or reported by the farmers ( who were asked specifically about the issue ) . No injection site lesions were noted . Many farmers however were reluctant to have ear tags placed on their animals , especially since some animals developed ear infections following the first intervention . There was no reluctance on the part of the farming community to their animals being either vaccinated or given anthelmintic drench . Farmers were pleased to see worms voided in the feces after the animals were treated . The pigs that underwent post mortem examination at the end of the trial were from 8 to 48 months of age and weighed from 35 to 175 kilograms , mean 89kg . The number of interventions and individual treatments received by the animals in the intervention area that underwent post mortem at the end of the study , are summarized in Table 4 . Thirty three animals from the intervention area were examined at most mortem , among which one animal had received a single intervention treatment , 16 animals had received 2 intervention treatments , 13 had received 3 treatments and 1 animal had received all 4 treatments . The numbers of animals recorded as having T . solium infection from the control and intervention areas , assessed at necropsy at the start and the end of the trial , are summarized in Table 5 . Among the animals necropsied at the start of the trial , nineteen out of 55 ( 34 . 6% ) pigs were positive from the intervention area , and thirteen out of 55 ( 23 . 6% ) pigs were positive from the control area ( not significant , p = 0 . 207 ) . The total prevalence of porcine cysticercosis ( PC ) was 29 . 1% . Approximately 9–10 months after the first intervention , 33 pigs from the intervention area and 35 pigs from the control area were subjected to post mortem to determine the presence of T . solium cysts . Zero out of 33 ( 0%; 95%CI 0–10 . 6% ) pigs were positive from the intervention group and six out of 35 ( 17 . 1%; 95% CI 6 . 6–33 . 7% ) pigs were positive from the control group ( significant , p = 0 . 025 Fisher’s Exact test 2-tailed ) . The pre-intervention prevalence of infection was significantly greater than the post intervention prevalence in the intervention area ( p<0 . 001 ) . In the control area there was no significant difference between the baseline and end of study prevalence of infection with T . solium ( p = 0 . 424 ) . The number of both viable and non-viable cysts was counted according to the criteria specified for a confirmed case of T . solium infection , including the number in the full carcass estimated by doubling the cyst number found in the skeletal muscles of the right hand side of the carcase . The baseline post mortems revealed 8 , 347 ( viable 7 , 379: non-viable 968 ) cysts in animals from both the control and intervention areas , with 7 , 039 cysts ( viable 6 , 694: non-viable 345 ) in 19 pigs ( average per infected animal 370 . 5±537 . 5 ) from the intervention area and 1308 cysts ( viable 685 , non-viable 623 ) from 13 pigs ( average 100 . 6±145 . 0 ) from the control area . There were fewer cysts found at the end of study post mortems , with 120 cysts identified in six pigs ( average 20 . 0±24 . 0 ) from the control area and none found to be infected in the intervention area . All animals that were recorded as having no T . solium infection detected in the heart , masseters , tongue , right hand carcase musculature , brain , liver or lungs had the remaining carcase musculature ( the left hand side ) sliced to determine whether there was any infection in the carcase at all . Two further animals were identified as being infected from the control area; one with a single viable cyst and one animal with two viable cysts only in the skeletal muscles from the left hand side of the carcase . No infection was detected , either viable or non-viable cysts , in any of the 33 animals from the intervention area after complete dissection of the carcases ( Table 5 ) . A total of 145 pigs that had no vaccination or drug treatment ( 110 animals at the start of the trial plus 35 from the control region at the end of the trial ) were subjected to detailed necropsy . This included careful slicing of the entire brain . Nine of these animals had one or more viable T . solium cysticerci in the brain . All those with cysts in the brain also had viable cysts in one or more muscle tissues . None of the 33 pigs from the intervention area had any T . solium cysts in the brain . Analysis of the characteristics of animals that were found to be infected with T . solium and for which reliable age data were available in comparison to the total number of animals that were necropsied at baseline and from the control area at the end of the trial , found no significant relationship between the age of the animals and the proportion of infected animals in the age range between 7 and >19 months of age ( Pearson’s correlation coefficient 0 . 189 , P = 0 . 76; Fig 1 ) . The proportions of infected animals in the different age classes were: 7–9 months , 2/10; 10–12 months , 14/50; 13–15 months , 7/38; 16–18 months , 6/24; ≥19 months , 5/21 ) . There was also no significant relationship between the presence of T . solium infection and the weight of the animals ( P = 0 . 288 , F-test; P = 0 . 09 , t-test ) . Similarly , no significant relationship was evident between the percentage viability of cysts found in individual animals and the age of the pigs ( Pearson’s correlation coefficient 0 . 188 , P = 0 . 28 ) , nor between the intensity of infection ( total number of cysts ) and the age of the animal ( Pearson’s correlation coefficient 0 . 133 , P = 0 . 45 ) .
Following implementation of 3-monthly treatments of the pig population over a 10-month period in a T . solium endemic region of Banke District , Nepal , transmission of the parasite was eliminated among the animals that were assessed at the end of the study . Comparison of the number of infected animals found in the intervention and control areas indicates that the intervention led to a significant reduction in porcine cysticercosis ( P = 0 . 004; Table 5 ) . This change in the risk of T . solium transmission by pigs is also evident when comparing the starting prevalence of infection in the intervention area with the prevalence of infection in the same area at the end of the intervention ( P<0 . 001 ) . There was a reduction in the prevalence of T . solium prevalence between the start and the end of the trial in the control area , however this was not significant ( P = 0 . 424 ) . Fewer animals were able to be purchased for necropsy at the end of the trial than had been intended . Nevertheless , significant differences were seen between the prevalence of infection at the start and end of the trial , as well as between the intervention and control areas at the end of the trial , due to the higher prevalence of infection than expected at the start and the magnitude of the intervention’s impact on cysticercosis transmission . The 3-monthly vaccination and oxfendazole treatment regime was implemented over an approximately 10-month period prior to the post mortem assessments being undertaken at the end of the trial . As the animals selected for assessment were based on them being of a size and age at which pigs from the area are generally sold or slaughtered , most of the animals assessed had been present for at least two of the interventions ( Table 4 ) . Animals indicated in Table 3 to have received less than two treatments were mostly those sold for slaughter shortly after the program was implemented , animals treated as piglets only at the time of the fourth intervention and hence not of slaughter weight at the completion of the trial , or animals that had died subsequent to receiving their initial treatment . At each treatment time , more than 90% of eligible pigs received the treatment ( Table 1 ) . The treatment schedule which was assessed was determined to be the most effective in an area where animals are consumed from 7 months of age [14] . While this frequency of treatment presents a limitation due to the costs involved , reducing the frequency of treatments , even by a single additional month to 4-monthly , may lead to over 50% of the slaughter-age animals being capable of transmitting T . solium [4] . Although immunity following a single immunization with TSOL18 has not been investigated specifically , available evidence suggests that stimulation of immunity may require two immunizations with the currently available vaccine [15] . Excellent secondary responses to the vaccine are seen when the interval between vaccinations is between one and four months [16] , hence the vaccination scheme adopted in this trial would be predicted to provide a high level of immunity . The level of protection achieved in this trial is similar to what was achieved in a previous field trial undertaken in Cameroon which involved a cohort of animals , rather than an on-going intervention program that was implemented in this trial in Nepal [17] . Application of a strategy combining vaccination and medication in pigs was based on characteristics of the developmental biology of the parasite and other relevant factors [14 , 17]; however , the contributions of the vaccination or medication components alone were not evaluated here . An estimate of the age at which animals were sent for slaughter was determined from the information obtained about animals that were present for the second intervention but had been sold prior to the third intervention , together with those that were present at the time of the third intervention but had been sold prior to the fourth intervention . Excluding animals that were absent for reasons such as them having died , being pregnant or ill , this information provided the age of the animals at which they were sold . On this basis , the typical age at slaughter of pigs in these communities was found to be 14 months when the animals were at least 60kg . Data obtained from the animals that were not part of the intervention provide valuable information about the age at which pigs acquire T . solium infection in a natural endemic situation . Very little information is available concerning this topic . Among the animals that were confirmed to have T . solium infection , no significant relationship was evident between the age of the animals and the proportion that was infected ( Fig 1 ) . Assuming that the infections acquired in young pigs persist , these data suggest that pigs acquire infection relatively early in life and that additional infections do not accumulate as the animals age . A hypothesis that would be consistent with these data would be that pigs older than approximately 1 year are relatively resistant to infection . Age-related resistance to infection is recognised in the intermediate hosts of other Taenia species [18] . None of the animals from the intervention area in Nepal were found to have T . solium in the brain tissue , whereas 9 of 145 untreated pigs were found to have cysts in the brain . This difference is not statistically significant , however the absence of cysts in vaccinated and treated pigs is consistent with the results of previous trials with the TSOL18 vaccine in which vaccinated animals also had no cysts in the brain [5 , 17 , 19] . Cysts in the brain of pigs are not killed by treatment of the animals with oxfendazole [20] , however the available evidence suggests that the TSOL18 is effective in reducing the number of cysticerci in the brain as well as in muscle tissues . The post mortem investigations undertaken at the start of the trial in Nepal included slicing of half the carcase musculature in addition to the heart , masseters , diaphragm , tongue , brain , liver , kidneys and lungs . Post mortems undertaken at the end of the trial involved the same procedures , such that direct comparisons could be made between results from the two sets of data . However , for all animals in which no cysticerci were found during the post mortems carried out at the end of the trial , the remaining musculature ( left hand side carcase ) was also sliced . In the case of the animals from the intervention area , none was recorded as having any cysts in the entire carcase musculature or other tissues that were examined . In the 36 control animals , two additional infected animals were identified , one having a single viable cyst and the other having two viable cysts in the left hand side musculature , but no cysts elsewhere . In this group of 36 infected animals from the control area most had light infections . Identification of infected animals by slicing muscles only from one side of the carcase , rather than the entire carcase musculature , would have missed 25% of the infected animals ( 2 of 8 infected ) . Chembensofu et al . [21] found that slicing predilection sites plus only one side of the carcase musculature would have missed 16% of the infected animals in their study undertaken with naturally infected pigs in Zambia . During the post mortems all carcase lesions and lesions in the brain , liver , kidneys and lungs were examined for the presence of a cysticercus . No cysticerci were found other than in striated muscle tissue and the brain . Necrotic lesions and other suspect lesions were investigated for the presence of taeniid or T . solium DNA . No T . solium lesions were identified by these methods other than in the brain and striated muscle . These data are consistent with the tissue distribution of T . solium cysticerci in many previous studies , including the comprehensive investigations undertaken by Boa et al . [22] on naturally infected pigs in Tanzania , the majority of which were heavily infected . These data , however , contrast with those reported by Chembensofu et al . [21] , who found large numbers of viable , DNA-confirmed cysticerci in the liver , lungs and kidneys of many pigs from Zambia . There is no clear explanation for this discrepancy . Potentially effective control measures for T . solium have been available now for decades and yet there have been few programs implemented as specific strategies that have led to a sustained reduction in neurocysticercosis [3] . Feasibility and sustainability of control measures have been the stumbling block to controlling T . solium . The requirement for a 21-day withholding period after oxfendazole treatment of pigs , creation of necrotic lesions in the meat of drug-treated , infected pigs , and difficulties with reliably predicting the time of sale or slaughter , prevent a treat-immediately-before-slaughter approach being used in pigs to control T . solium . To be effective , the frequency with which intervention would need to be undertaken in the pig population is governed by the rapid turnover of pigs in the communities and constant introduction of new , susceptible animals into the population due to pigs breeding throughout the year . Combining both vaccination and oxfendazole as a preventative treatment for porcine cysticercosis has several advantages . Firstly , the drug treatment eliminates any viable cysts that may be in an animal’s musculature prior to the animal being protected after vaccination . Secondly , the drug treats many nematode and trematode infections , as well as cysticercosis , likely providing a health and productivity boost to the treated animals [23 , 24] . Oxfendazole treatment does not provide any protection for uninfected pigs against subsequent exposure to the T . solium , hence combined use with the vaccine provides both treatment as well as prevention from subsequent infection . After treatment of an infected pigs with oxfendazole necrotic lesions are evident in the musculature for a period of at least several weeks; the great majority disappearing within a period of 3 months [6 , 24 , 25] . It seems likely that some of the animals from the intervention area in Nepal that underwent post mortem investigation would have been infected with T . solium prior to them being fully vaccinated . However no non-viable lesions were detected in the muscles of the animals that had participated in the interventions . The three-monthly treatment regime that was implemented in the trial appears to have allowed sufficient time for any lesions that were the result of the death of parasites in muscles after medication to be resorbed before the animals reached slaughter age . A limitation to the use of oxfendazole as a treatment for porcine cysticercosis is the requirement for a 3 week withholding period after treatment before slaughter due to the presence of drug residues in the tissues [9] . In the intervention described here , all animals ≥2 months of age were treated ( other than sows near parturition ) . Farmers were requested to not sell or kill the treated animals for 3 weeks after each treatment . This imposes a significant burden on the farmers , especially when the procedure is repeated every 3 months , and it is difficult to monitor compliance . Also , the farmer’s requirements about selling animals can change rapidly; a family illness or other unforeseen event can impose an urgent need to sell animals . The T . solium transmission modelling presented by Lightowlers and Donadeu [4] predicted that a 3-monthly program of vaccination plus oxfendazole treatment of pigs between one and 7 months of age would eliminate T . solium infection entirely from pigs >7 months of age such that they would not require further oxfendazole treatment . Cessation of drug treatment of animals that are approaching slaughter age would reduce or prevent the risk that animals with high levels of drug residues could be consumed as well as reducing the cost . The intervention program that was applied in the trial in Nepal involved animals of all ages ( >2 months ) . Introduction of a new 3-monthly treatment program in an area would necessarily involve all pigs to start , so as to treat and protect all the existing animals . However , immunity induced by 2 immunizations with the Cysvax vaccine , together with a natural resistance to T . solium infection in animals >1 year of age ( mentioned above ) , may allow a continuing program involving vaccination plus oxfendazole medication to be effective if it were only implemented in animals up to 7 months of age [4] . The intervention that was undertaken in this trial in Nepal was relatively simple . Groups of 5–6 persons travelled by motorbike . The most time-consuming aspect of the intervention was catching the animals; having caught a pig , vaccination and drug treatment took just a few moments . The older animals were generally the more difficult and time consuming to catch . Based on the experience gained in conducting this trial in Nepal , teams of 5–6 persons who were undertaking a similar , but on-going cysticercosis control program implemented in pigs up to 7 months of age , would be able to vaccinate and drug treat approximately 100 pigs per day in Dalit communities such as those in the Banke district . The three-monthly intervention scheme adopted here was predicted to , and did , lead to the cessation of the risk of T . solium transmission by the vaccinated and drug-treated animals . Any intervention limited to the pig population would not immediately affect the incidence of cysticercosis in humans because it would take time for the prevalence of human T . solium taeniasis to decline as new cases of taeniasis were prevented due to the absence of cysticercosis in pigs . Calculations based on the rate of re-establishment of taeniasis following mass treatment of communities [26] suggest that T . solium tapeworms have a lifespan of 2–3 years; a lifespan of less than 5 years is also suggested by epidemiological evidence [27] . If this were the case , implementation of an on-going intervention only in pigs would lead to a substantial reduction in , or elimination of , the incidence of human cysticercosis within about 2–3 years . Alternatively , a single treatment of the human population for taeniasis after porcine cysticercosis was controlled , would lead to a more immediate reduction in the incidence of human cysticercosis [3 , 4] . Having achieved a substantial level of reduction in T . solium transmission by pigs , it may not be necessary to continue with a high frequency of interventions in the pig population in order to maintain a low level of parasite transmission by pigs . Hence , the relatively intense 3-monthly intervention trialled here may not be required to be a practice that is required to be sustained for a long term . Although it was not tested here , evidence about the duration of protection afforded by the TSOL18 vaccine [17] and T . solium transmission modelling [4] would suggest that a 3-monthly vaccination and drug treatment regime would be effective if applied only to animals up to 7 months of age , with re-vaccination only of animals kept for long periods , for example , for breeding purposes . We propose that this could be an effective , relatively simple and feasible control strategy for T . solium which could be applied to reduce the transmission of T . solium by pigs , and the results suggest that applying the intervention over a period of years may ultimately reduce the number or tapeworm carriers and thereby the incidence of NCC in Nepal and elsewhere . The feasibility of this approach has been enhanced by the availability of an effective vaccine and medication , with both becoming available , for the first time , as commercial products licensed for use in pigs for T . solium cysticercosis . Vaccination and medication of pigs in this trial led to cessation of the potential for T . solium transmission by the animals that had received the 3-monthly treatments , however the trial had a number of limitations . It was a small study in a single region . The trial was also undertaken for a relatively short period and would be expected to have little if any effect on the incidence of neurocysticercosis in the study areas because the intervention was not undertaken for a sufficiently long period to lead to a substantial decrease in the prevalence of T . solium taeniasis [26] . Also , the logistics , costs and requirement of a cold-chain for the vaccine may be limitations to this as an intervention strategy . Further consideration of this approach , for control of T . solium transmission , would benefit from a cost-benefit analysis being undertaken on a program implemented for a longer period . The analysis would require an assessment of the impact on the incidence of neurocysticercosis in the population participating in the program . | Neurocysticercosis is a disease caused by a parasitic infection of the brain . The parasite responsible , Taenia solium , is transmitted by pigs where human sanitation is poor and pigs roam freely . Neurocysticercosis is responsible for many cases of epilepsy in people living in poor , developing countries . The feasibility and sustainability of implementing control measures have been major impediments to reducing the incidence of neurocysticercosis . Recently , two new commercial products have become available which together offer the possibility of interrupting the parasite’s transmission by pigs–the TSOL18 vaccine ( Cysvax , IIL , India ) and an oxfendazole formulation ( Paranthic 10% , MCI , Morocco ) licensed for use in pigs for the treatment of cysticercosis . Here we describe the impact of implementing vaccination plus drug treatment of pigs in the Banke district of Nepal . The intervention eliminated the risk of transmission of T . solium by the animals vaccinated and treated during the trial . Application of the vaccination and drug treatment program used here , possibly with strategic use of anthelmintics also in the human population , is an effective option for reducing the incidence of neurocysticercosis in Nepal and elsewhere . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"neurocysticercosis",
"immunology",
"tropical",
"diseases",
"geographical",
"locations",
"vertebrates",
"parasitic",
"diseases",
"animals",
"mammals",
"animal",
"slaughter",
"preventive",
"medicine",
"pharmaceutics",
"neglected",
"tr... | 2019 | Implementation of a practical and effective pilot intervention against transmission of Taenia solium by pigs in the Banke district of Nepal |
The envelope glycoproteins of primate lentiviruses , including human and simian immunodeficiency viruses ( HIV and SIV ) , are heterodimers of a transmembrane glycoprotein ( usually gp41 ) , and a surface glycoprotein ( gp120 ) , which binds CD4 on target cells to initiate viral entry . We have used electron tomography to determine the three-dimensional architectures of purified SIV virions in isolation and in contact with CD4+ target cells . The trimeric viral envelope glycoprotein surface spikes are heterogeneous in appearance and typically ∼120 Å long and ∼120 Å wide at the distal end . Docking of SIV or HIV-1 on the T cell surface occurs via a neck-shaped contact region that is ∼400 Å wide and consistently consists of a closely spaced cluster of five to seven rod-shaped features , each ∼100 Å long and ∼100 Å wide . This distinctive structure is not observed when viruses are incubated with T lymphocytes in the presence of anti-CD4 antibodies , the CCR5 antagonist TAK779 , or the peptide entry inhibitor SIVmac251 C34 . For virions bound to cells , few trimers were observed away from this cluster at the virion–cell interface , even in cases where virus preparations showing as many as 70 envelope glycoprotein trimers per virus particle were used . This contact zone , which we term the “entry claw” , provides a spatial context to understand the molecular mechanisms of viral entry . Determination of the molecular composition and structure of the entry claw may facilitate the identification of improved drugs for the inhibition of HIV-1 entry .
Entry of HIV-1 into target cells involves the interaction of the surface glycoprotein gp120 ( designated SU ) with the cell surface receptor CD4 [1] , a binding-induced structural change [2] in gp120 that creates the binding site for a cellular seven-transmembrane-helix co-receptor protein [3] , followed by conformational changes [4] in the transmembrane glycoprotein gp41 ( designated TM ) that allow formation of the “pre-hairpin” conformation [5 , 6] . Insertion of the fusogenic portion of the TM polypeptide into the target cell membrane ultimately leads to fusion of the viral and target cell membranes [7] . Substantial insights into the entry mechanism have come from cell and structural biological studies , which suggest that the molecular species that initiates the steps leading to fusion is a trimer of envelope glycoprotein heterodimers [8] . Electron tomography is a powerful approach for determining the three-dimensional ( 3-D ) structures of large and heterogeneous sub-cellular assemblies at resolutions that are typically one to two orders of magnitude higher than those that can be currently achieved using light microscopy [9–11] . Because these assemblies are not generally amenable to analysis by crystallographic approaches , electron tomography provides tools to bridge the gap between cellular and molecular structure . Although the highest resolutions have been obtained from tomographic analyses of thin , unstained specimens in a near-native state at cryogenic temperatures , significant information has also been derived from tomographic imaging of chemically fixed and stained specimens that allows structural investigation of the interior of thick cellular specimens . Here , we have used electron tomographic approaches to analyze the 3-D architectures of simian immunodeficiency virus ( SIV ) and HIV-1 virions incubated with CD4+ T lymphocyte target cells to identify structural features of cell-bound viruses trapped at a stage prior to entry into target cells .
To validate the approach used for structural analysis of infected cells using electron tomography of stained , plastic-embedded specimens , we first carried out analysis of unstained purified viruses using cryo-electron tomography , and compared the resulting 3-D structures with those obtained from analysis of free virions in stained , plastic-embedded specimens . For ease of comparative analysis , we used an SIV strain that expresses a high level of envelope glycoprotein as a consequence of the effects of the 9-kDa C-terminal truncation in TM [12 , 13] . Purified virion samples were rapidly vitrified by plunge-freezing in liquid ethane cooled to ∼ −180 °C . The frozen hydrated specimens were imaged at liquid nitrogen temperatures using a 300 kV electron microscope equipped with a field emission gun and an energy filter . Four 1-nm-thick slices from a reconstructed tomogram are shown in Figure 1 . The images reveal the distribution of the envelope glycoprotein spikes on the surface and a glimpse of the internal core of the virus . Since these are from unstained virus preparations , the contrast in the images arises from the intrinsic distribution of mass in the virus . A segmented 3-D rendering of one of the four virions shown in Figure 1 is presented in Figure 2A , with the surface spikes and the internal core highlighted . Substantial variation was evident between virions in the numbers of glycoprotein spikes on the surface , with some displaying as many as 100 spikes , while others were largely devoid of spikes . Inspection of the 3-D structures of several individual viral spikes shows that they have shapes that are often narrow ( ∼60 Å ) in the region closest to the membrane , and wider ( ∼120 Å ) at the distal end . This size is consistent with that expected from a trimer of envelope glycoprotein molecules protruding from the surface of the viral membrane , with the TM trimer packed in the narrower end of the viral spike . There is noticeable structural heterogeneity in individual spikes , which could arise either from the low signal-to-noise ratios inherent to cryo-electron tomography , the effect of the missing wedge in data collection , and/or from genuine conformational variability in the viral spikes [14] . To evaluate the distribution of the viral spikes on the surface of the virus , we developed automated feature extraction tools to identify the locations of individual spikes , as illustrated in Figure 2B . In this SIV preparation , spikes are distributed throughout the surface without obvious clustering , with an average spacing of ∼200 Å between neighboring spikes ( Figure 2B ) . Images of individual spikes can be averaged together with compensation for the missing wedge to improve the signal-to-noise ratio; however , an averaged structure does not capture the structural heterogeneity in spike structure , and the resulting averaged image can vary depending on the approach used to align the spikes to each other . Different averaged structures and differing atomic models for the SIV spike have been recently reported by Zhu et al . [15] and by Zanetti et al . [16] ( see commentary in [14] ) . Here , we show the structures of SIV envelope glycoprotein spikes without averaging solely to provide a reference for the subsequent structural analysis of viruses captured in contact with target cells . The viruses we ( Figure 1 ) and Zhu et al . [15] have imaged are variants of SIVmac239 with a truncated TM glycoprotein . The SIVmneE11S virus analyzed by Zanetti et al . [16] is closely related; however both sets of viruses have high levels of envelope glycoproteins as a consequence of truncation of the cytoplasmic tail of the TM glycoprotein . To compare the overall architecture of viral spikes observed by cryo-electron tomography of purified SIV virions to that seen in fixed , plastic-embedded SIV , we obtained tomograms from 150-nm-thick sections prepared from plastic-embedded specimens of SIV-infected T cells . The outlines of viruses can be readily identified in single slices from these cellular tomograms ( Figure 3A–3C ) and in the segmented representation of the tomogram ( Figure 3D ) . The spikes on the surface of these virions are ∼120 Å high and similar in appearance to those obtained from unstained viruses obtained using low-dose cryo-electron tomography . While the tomograms from stained specimens report on the 3-D distribution of the stain rather than the intrinsic density of the virus , the demonstration that viral spikes analyzed by both methods have similar dimensions provides confidence for interpretation of tomograms of contact regions between virions and the surface membrane of target cells obtained using fixed , plastic-embedded specimens , as described below . In addition to seeing mature virus particles with characteristic morphologies [17] in the extracellular medium ( Figure 3 ) , we also detected , at a lower frequency , regions of the cell surface where viruses appeared to be captured in close contact with the cell membrane ( Figure 4D ) . This contact region , as visualized in projection , displayed a characteristic pattern of striated densities connecting the viral membrane to the cell membrane . In some instances , the ( concave ) curvature of the cell membrane was found to follow the ( convex ) curvature of the viral membrane at the region of contact ( Figure 4E ) . The viruses involved in these contacts generally showed evidence of mature cores and did not display the characteristic thick layer of uncleaved Gag protein seen in immature viruses ( Figure 4B ) , indicating that they were not immature virions in the process of budding ( Figure 4A ) or free mature viruses ( Figure 4C ) . To evaluate virus–cell contact regions potentially involved in infection events more unambiguously , we pre-incubated uninfected CD4+ T cells susceptible to SIV and HIV-1 infection with high concentrations of infectious virus at 4 °C , then warmed the cells to 37 °C and carried out rapid fixation after waiting for periods ranging from 15 min to 3 h . The rationale for pre-incubation at 4 °C was to allow binding , but not fusion , of virions to target cells , while warming to 37 °C was intended to allow progression of fusion and virus entry over time scales not long enough for viral replication [18] . Images of the virus–cell interface from these acutely infected cells ( Figure 4F ) display exactly the same type of cell–surface contact observed in the chronically infected cells ( Figure 4D and 4E ) ; the profiles shown are representative of the images obtained from over 200 different virus–cell contact regions imaged in cells fixed at different times after warming . We conclude from these observations that the contact zones found in chronically infected cells ( Figure 4D and 4E ) are therefore likely to represent infection events involving mature viruses binding back to target cells . Next , we investigated the 3-D spatial architecture of a typical virus–cell contact region such as those shown in Figure 4D and 4E using electron tomography . Two transverse views of a 3-D reconstruction of an entry claw in a chronically infected cell ( at different depths in the tomogram ) are shown in Figure 5A and 5B , while a top view along the plane of contact is presented in Figure 5C . Distinct rods of density in the contact region spaced at ∼150 Å can be observed , as indicated by the arrows . The same type of contact structure is also observed in cells subjected to acute infection , as shown in the tomographic slice ( Figure 5D ) and segmented version of the entire 3-D volume ( Figure 5E ) of a virus–cell contact in acute infection . Based on our findings , and its probable connection to viral entry ( discussed below ) , we refer to this contact region as the viral “entry claw” . Between five and seven rods of density , arranged in a closely packed pattern , were typically observed in the entry claws that we imaged . Of the fourteen tomograms of entry claws that we analyzed in detail in 3-D , one was observed with four rods of density , seven were comprised of five rods , five had six rods , and two had seven rods . These findings are supported by visual inspection of these contact regions recorded in 2-D projection views . The variation in claw profile could arise , at least in part , due to the fact that not all contact regions were fully captured in the 100-nm-thick sections . Overall , each rod is ∼100 Å long and ∼100 Å wide with a center-to-center distance ranging from 140 Å to 170 Å between individual rods . The overall width of the claw ranges from ∼350 Å to 450 Å and forms the outlines of a neck-shaped region at the junction of the viral and target cell membranes . The spacing of these rods is slightly closer than the average spacing of envelope glycoprotein spikes observed in free viruses ( ∼200 Å ) , suggesting that rearrangement of spikes within the viral membrane may be involved in formation of the structure . A unique feature observed in association with most viral entry claws we imaged was the virtual absence of visible spikes on the rest of the viral surface . This is remarkable given that the SIV viruses used in some of our studies display as many as ∼100 spikes distributed over their surface upon release from infected cells ( Figures 1 and 2 ) . We believe that a likely explanation of our observations is that the spikes remote from the region of contact may have been shed from the particles . It is formally possible , although improbable , that there exists a small percentage of viruses that only have a few viral spikes , and these are the ones that are preferentially engaged in entry claw formation . We cannot resolve at present whether both gp120 and the TM glycoprotein have been lost from virions involved in these docking interactions , or whether only gp120 has been lost , leaving the TM glycoprotein behind . We carried out a number of additional control experiments to verify that the virus–cell contact shown in Figures 4 and 5 represent specific contact events relevant to viral entry . The typical frequency of entry claw observation in our experiments was ∼ten events/300 imaged cell sections ( corresponding to ∼15 entry claws/cell , assuming an ∼100-nm section and an average cell radius of ∼10 μm ) . When incubation of cells with virus was carried out in the presence of an anti-CD4 antibody known to block viral entry ( anti-Leu3a; [19] ) , no instance of viral contact was found despite extensive screening of several hundred cells , indicating that the presence of the antibody inhibited close interactions of the kind seen in Figure 4 . We conclude that cell surface CD4 is required for formation of the unique architecture observed in the contact region . Entry claw structures similar to those observed in Figures 4 and 5 were observed when we used a wild-type SIV mac239 virus expressing full length TM glycoprotein ( Figure 6A ) . The average number of spikes in this viral isolate is known to be about ten times less than those in the tail-truncated version [13] . When cells were fixed only after the 4 °C incubation , and without warming to 37 °C , no evidence of entry claw formation was found ( 0/300 imaged cells ) , even in instances where viruses were observed in proximity of the cell membrane ( Figure 6B ) . While viral binding to the cell surface is not expected to be restricted to the period of incubation at 4 °C , the lack of entry claw formation in the absence of warming suggests that the latter step is required for its stabilization . Similarly , no entry claw structures were found when incubation of viruses was carried out in the presence of the CCR5 antagonist TAK779 ( Figure 6C ) or the peptide inhibitor C34 ( Figure 6D ) . Together , these experiments provide strong evidence that formation of the entry claw requires interaction of the envelope glycoprotein and the cell surface receptors known to be involved in viral entry . To determine whether the structural features observed in the contact region are unique to the SIV used in our studies , we analyzed the virus–target cell contact region after incubating CD4+ target cells with HIV-1MN virions . Cells were processed using the same approaches employed for the SIV studies described above . 2-D and 3-D images from the HIV-1-infected cells ( Figure 7 ) demonstrate that the same type of architecture that is observed for contact between SIV and T cells is also observed for contact of HIV-1 . Figure 8A provides a schematic interpretation of the tomographic analysis presented here , showing viruses in different stages of maturation and in contact with T cells to form the entry claw . Density corresponding to the core was observed in some but not all viruses engaged in entry claw formation after both short and long periods of incubation , although we note that since many contacting viruses would only have been captured partially in the 100-nm sections , the absence of core-like density cannot be used as definitive diagnosis for the absence of a core in the contacting viruses . Our working hypothesis is that each of the rods is derived from a single viral spike , although their precise molecular composition remains to be determined . The spacing of these rods is slightly closer than the average distribution of envelope glycoproteins observed in free viruses , suggesting that rearrangement of spikes within the viral membrane could be involved . Although extensive labeling experiments will be necessary to determine the precise composition of the entry claw , it is interesting to note that that the observed average length ( ∼100 Å ) of the rods of density connecting the viral and cell membranes is consistent with the expected dimensions of a potentially fully extended state of TM , representing the pre-hairpin intermediate [6] .
Viral binding and entry are key processes in the biology of AIDS virus infection and represent critical steps for potential prophylactic intervention by vaccine-induced neutralizing antibodies , or therapeutic intervention by inhibitors of binding or viral fusion . The observations we present here of the size and composition of the SIV and HIV-1 entry claw are in agreement with previous studies suggesting that viral entry is likely to be mediated by an oligomeric assembly of viral spikes and cell surface receptors . Experimental support for the cooperative interaction of multiple envelope glycoproteins in the entry of influenza [20] , Semliki Forest virus [21] , rabies virus [22] , and baculovirus [23] has already been documented . In the case of HIV , there is also evidence that between four and six co-receptors ( CCR5/CXCR4 ) and multiple CD4 molecules are likely to be present at the fusion pore [24 , 25] . The combination of these studies and the results of our structural analysis suggest that the entry claw is a specific macromolecular structure associated with initiating fusion and viral entry . What stage of the fusion does the entry claw represent ? Some models for viral fusion generally invoke the formation of a wide-necked pore via a hemi-fusion intermediate [26] . Despite thorough screening , we have not yet detected structures that appear to represent this type of fusion event where one might expect a partially fused virion in the act of transferring the viral genome-containing core into the cell . One possible explanation for this could be that the fusion event itself is very rapid , and therefore not detected in our experiments . In support of this idea , there is considerable evidence that once viruses bind to target cells , formation of sufficient numbers of ternary complexes of Env , CD4 , and co-receptor is likely to be the rate-limiting step in viral entry [27] . This initial stage of viral–cell interaction corresponds at the biochemical level to generation of the pre-hairpin intermediate [6] , and at the cellular level to the “temperature-arrested stage” [28] , where viruses can remain attached to cells at temperatures that are suboptimal for fusion . In this model ( Figure 7B ) , steps subsequent to initial docking must include , at a minimum ( although not necessarily in this sequence ) , helix bundle formation , creation of the correct geometry for fusion , formation of a hemi-fusion intermediate leading to merger of the outer leaflets of the viral and target cell membranes , and the fusion event itself , which results in delivery of the viral genome into the cell . There is an alternative model ( Figure 7C ) for how fusion might occur that is potentially consistent with recent studies that have concluded that a single fusion competent envelope glycoprotein trimer may be sufficient to support HIV-1 entry [29] . At first glance , these studies would appear to contradict the findings reported here , which consistently show multiple viral spikes acting in concert to form the entry claw . However , the key point is that in the analysis carried out by Sodroski and co-workers [29] , the number of functional , fusion-competent trimers was modulated by varying the relative proportion of functional and defective ( fusion-incompetent ) glycoproteins present on surface of the virus , without alteration in the average number of spikes per virion . These non-functional envelope glycoproteins were mutants fully capable of CD4 binding , but defective for cleavage to generate fusogenic TM polypeptides . Thus , multiple trimers might participate in formation of an entry claw structure , creating a stabilized scaffold to dock the virus to the cell membrane , regardless of their fusion competence . If fusion were to occur via local puncture of the membrane at the location of the single functional spike , presumably via the formation of a fusogenic , six–helix bundle intermediate [6] , it could account both for the structural role of multiple viral spikes in forming the claw to create the scaffold to dock the virus to the cell membrane , and the finding that one functional trimer is adequate for viral entry . The electron tomographic analyses show that dimensions of the pore that could be formed between the anchors of the claw could be as wide as ∼300 Å . An opening of this size is comparable to the dimensions of an intact viral core [30] and may thus be large enough to transport the core of the virus into the cell even without invoking any additional pore expansion . In this mechanism , the entry claw scaffold and the viral envelope could continue to remain attached on the cell surface after loss of viral core . Although evidence of staining from viral cores was observed in entry claw contacts after both short and long incubation periods , at present , we cannot definitively distinguish between the two models in Figure 8 . The discovery of the entry claw raises many fundamental questions about viral entry . What is the molecular composition and stoichiometry of the components forming the entry claw ? Do other complex factors such as membrane reorganization and actin rearrangement play a role in its formation or stabilization ? Is persistence of the entry claw dictated by the speed of the topological rearrangements needed for formation of the helix bundle that triggers membrane fusion ? Does the composition of the claw change following initial contact ? Are there other structural intermediates that can be trapped by using selected gp120 or gp41 mutants with altered functional properties , or by the addition of antibodies or ligands to the cell surface receptors involved ? Can the fusion event itself be captured and visualized ? What is the role , if any , of interactions between the C-terminal tail of the envelope glycoprotein ( TM ) and the Gag matrix in rearrangements of viral spikes at the point of contact ? Continuing improvements in technologies for cryo-electron tomography , combined with X-ray crystallographic analysis of the envelope glycoprotein in its various conformations , offer the promise of making significant advances towards answering these questions .
Viruses were isolated from chronically infected cells by density gradient centrifugation and further purified by using an anti-CD45 affinity column to remove contaminating microvesicles [31] . Purified viral suspensions were deposited on Quantifoil grids ( Quantifoil , http://www . quantifoil . com ) , mixed with a solution of colloidal gold , and plunge-frozen using a Vitrobot device ( FEI Company , http://www . vitrobot . com ) . Grids were imaged at liquid nitrogen temperatures using a Polara field emission gun electron microscope ( FEI Company ) equipped with a 2k × 2k CCD placed at the end of Gatan energy filter , and operated at 300 kV . Tilt series for tomographic reconstruction were acquired using the FEI tomography software package and reconstructed using either weighted back-projection or SIRT procedures as implemented in IMOD [32] and Inspect 3D ( FEI Company ) , respectively . Typically , for the cryo-electron tomographic experiments , electron doses of ∼120 electrons/Å2 were used , and images were collected at 2 . 5-degree intervals . Visualization and semi-automated segmentation was carried out using software tools implemented in the program Amira ( TGS , http://www . tgs . com ) . For acute infection studies using SUPT1/CCR5 CL . 30 cells as targets , replicate aliquots of cells were resuspended at 5 × 105 cells in 200 μl in microcentrifuge tubes in RPMI 1640 with 10% ( v/v ) heat-inactivated fetal calf serum . The cells of this human T cell line naturally express CD4 and CXCR4 and have been engineered by retroviral transduction to express CCR5 [33] . Cells were incubated at 4 °C for 1 h , with or without monoclonal anti-Leu3a ( Becton Dickinson , http://www . bd . com ) antibody at a final concentration of 10 μg/mL . Pre-cooled , sucrose gradient purified concentrated SIV or HIV-1 was then added to the cells , using approximately 2 μg of p28CA or 1 μg of p24CA equivalent per 5 × 105 cells ( corresponding to approximately 1 × 1010 to 2 × 1010 virions and a nominal multiplicity of infection of between 0 . 1 and 1 , based on titration in SUPT1/CCR5 CL . 30 cells ) . The viruses used were a SIV variant with a truncated TM glycoprotein and increased virion envelope glycoprotein content , SIVmac239/251 tail ( lot P3973 , produced from SUPT1/CCR5 CL . 30 cells ) , a wild-type SIV with a full-length TM glycoprotein and lower envelope glycoprotein content , SIVmac2339 ( lot P4118 , produced from CEM X174 ( T1 ) cells ) , or wild-type HIV-1MN ( lot P4091 , produced from SUPT1 cells ) . The virus SIVmac239/251 tail and the SUPT1/CCR5 CL . 30 cells were provided by J . Hoxie ( University of Pennsylvania ) . Cells were incubated with virus for 60 min at 4 °C , then warmed to 37 °C and maintained at this temperature until the conclusion of the experiment . After specific incubation periods , samples were fixed by addition of glutaraldehyde , to a final concentration of 2 . 5% by volume , in 0 . 1 M cacodylate buffer ( pH 7 . 4 ) ( Electron Microscopy Sciences , http://www . emsdiasum . com/microscopy/default . aspx ) . Cells incubated with virus in the presence of non-antibody inhibitors were processed in exactly the same way , except for the presence of either the TM peptide–derived fusion inhibitor SIVmac251 C34 ( 5 uM ) , or the CCR5 antagonist TAK779 ( at 10 μM ) . For chronic infection samples , similar procedures were used for initial virus incubation , with an innoculum of SIVmac239/251 tail ( lot P3973 ) corresponding to ∼600 ng p28CA equivalent and 6 × 109 virions , and a nominal multiplicity of infection of ∼0 . 05 . Inoculated cells were washed twice , resuspended at 1 × 106 cells/mL , and cultured for 3 d , at which time fresh medium was added and the cells incubated for an additional 4 d . After aspiration of the bulk of the supernatant , the cell suspension was fixed with glutaraldehyde as in the case of acutely infected cells . Fixed cells were treated with reduced osmium ( 1:1 mixture of 2% aqueous potassium ferrocyanide ) as described previously [34] , embedded in 2% agar , dehydrated in ethanol , and embedded in Epon resin . Then , 100-nm- to 150-nm-thick sections were collected on copper grids and stained with lead citrate . Electron tomography was carried out at room temperature using a Tecnai 12 electron microscope ( FEI Company ) operated at 120 kV . Methods for data collection and reconstruction were the same as those used for cryogenic specimens described above . To study the patterns of viral spike distribution , viral membranes were first segmented semi-automatically using energy-based segmentation algorithms that allow incorporation of shape and smoothness constraints to facilitate segmentation . Individual spikes were located automatically on the surface of the viral membrane , and their locations stored for subsequent computational analysis . Localization of viral membranes is most reliable in the central portion of viruses where the membranes are aligned to the best spatially resolved plane in the tomograms; we therefore used only a band comprising about one third of the total viral surface area for quantitative analysis of spike distribution . Individual spikes were located automatically by template matching using a spherical template with a diameter of ∼10 nm . The search was restricted to identify features originating on the membrane surface and radiating outward . This procedure assigns a template-fitness value to all points on the membrane; this parameter is locally maximized at the location of viral spikes ( represented with arrows pointing in the normal direction , see Figure 2 ) . Once all spikes are located , distances between pairs of positions were measured along the surface of the viral membrane . To analyze the distribution of distances , we computed for all spikes the distance at which the three closest spikes were located , and plotted this distribution as a histogram . | Retroviruses such as simian immunodeficiency virus and HIV-1 enter target cells by exploiting the interaction between their surface glycoproteins and cell surface receptors . Knowledge of the structures of these glycoproteins and of the molecular details of their interaction with cell surface receptors is of fundamental interest in understanding viral entry mechanisms . Electron tomo-graphy is a powerful approach to determining the three-dimensional structures of large and heterogeneous sub-cellular assemblies such as virus–cell contact regions that cannot easily be analyzed by high-resolution structural methods such as X-ray crystallography . Here , we have used electron tomographic approaches to show that SIV and HIV-1 viruses make contact with T cells via a unique structure that we term the viral “entry claw” , which is typically composed of about six clustered rods of density that span the contact region . Investigation of the structure of the entry claw and the factors that promote its formation could lead to new insights into the design of more effective drugs to inhibit HIV entry . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"viruses",
"biophysics",
"biochemistry",
"virology"
] | 2007 | Electron Tomography of the Contact between T Cells and SIV/HIV-1: Implications for Viral Entry |
CD8+ cytotoxic T lymphocytes ( CTLs ) are critical for clearing many viral infections , and protective CTL memory can be induced by vaccination with attenuated viruses and vectors . Non-replicating vaccines are typically potentiated by the addition of adjuvants that enhance humoral responses , however few are capable of generating CTL responses . Adjuplex is a carbomer-lecithin-based adjuvant demonstrated to elicit robust humoral immunity to non-replicating antigens . We report that mice immunized with non-replicating Adjuplex-adjuvanted vaccines generated robust antigen-specific CTL responses . Vaccination by the subcutaneous or the intranasal route stimulated systemic and mucosal CTL memory respectively . However , only CTL memory induced by intranasal vaccination was protective against influenza viral challenge , and correlated with an enhancement of memory CTLs in the airways and CD103+ CD69+ CXCR3+ resident memory-like CTLs in the lungs . Mechanistically , Myd88-deficient mice mounted primary CTL responses to Adjuplex vaccines that were similar in magnitude to wild-type mice , but exhibited altered differentiation of effector cell subsets . Immune potentiating effects of Adjuplex entailed alterations in the frequency of antigen-presenting-cell subsets in vaccine draining lymph nodes , and in the lungs and airways following intranasal vaccination . Further , Adjuplex enhanced the ability of dendritic cells to promote antigen-induced proliferation of naïve CD8 T cells by modulating antigen uptake , its intracellular localization , and rate of processing . Taken together , we have identified an adjuvant that elicits both systemic and mucosal CTL memory to non-replicating antigens , and engenders protective CTL-based heterosubtypic immunity to influenza A virus in the respiratory tract . Further , findings presented in this manuscript have provided key insights into the mechanisms and factors that govern the induction and programming of systemic and protective memory CTLs in the respiratory tract .
Vaccination is the most effective tool for protecting humans and animals from infectious diseases . [1–4] However , despite decades of research , there are no broadly protective vaccines against seasonal influenza A viruses ( IAV ) , and effective vaccines against most other respiratory viruses do not exist . The most effective IAV vaccines currently licensed in the U . S . depend upon the generation of neutralizing antibodies targeting IAV hemagglutinin ( HA ) antigens . [5] These neutralizing antibodies are capable of eliciting varying levels of protective immunity to specific viruses in certain demographics . However , HA is also the most frequently mutated of the IAV proteins , and the immunity resulting from this year’s vaccine strain may not confer immunity against strains emerging during the current and subsequent influenza seasons . Therefore , vaccine strains must be adjusted annually to match HA predicted for the next influenza season . Even with annual administration , humoral immune responses tend to be short-lived , cross-protection against strains with minor HA mutations is highly variable , and there is progressively less protection against heterosubtypic or heterotypic viruses . [5–8] As a result , current public health policy is largely dependent on annual re-vaccination for seasonal IAV , and pandemic disease surveillance , outbreak containment , and the activation of an emergency vaccine development pipeline aimed at producing a vaccine bespoke for the virus of interest . [9 , 10] IAV vaccines that elicit cell-mediated immunity ( CMI ) or balanced CMI and antibody responses are promising alternatives to antibody-only strategies . [5 , 11–18] Because of their capacity to selectively target and kill IAV-infected cells , CD8+ cytotoxic T lymphocytes ( CTLs ) play a crucial role in the initial clearance of influenza virus infections and are the primary target for most CMI vaccination strategies . [14 , 15 , 19] [20] Unlike most neutralizing antibodies , CTLs intrinsically target a variety of IAV structural epitopes such as nucleoprotein peptides that are substantially less mutable and more broadly conserved than HA , and they can generate long-lived memory cells capable of mounting cross-protective recall responses . [18 , 21–23] Important experimental studies of cell-mediated immunity in mice demonstrate that , separately , influenza-specific memory CTLs and TH cells are sufficient to protect against heterosubtypic influenza challenge . [24] Additionally , naturally occurring cross-protective memory CTL responses in humans are potent enough to be a confounding factor in the evaluation of human IAV challenge studies , and there is significant evidence that pre-existing cross-protective cell-mediated immunity mitigated the effects of the most recent pandemic H1N1 influenza outbreak . [25] These findings strongly suggest that CTL-mediated immunity may provide the means for universal vaccinations for IAV and other respiratory viruses . Currently licensed CTL-generating vaccines require the presence of a replicating antigen such as an attenuated virus or vector . However , replicating antigens are contraindicated in several key target groups and have the potential for causing clinically significant disease by several mechanisms . [26 , 27] By contrast , inactivated viruses and their subunits are comparatively safe , and can be used in at-risk populations . However , non-replicating antigens are intrinsically poor immunogens that primarily generate humoral responses , even when adjuvants are added to enhance immunogenicity . [26 , 28] Few modern adjuvants have been reported to safely elicit cell-mediated immunity , and none of those are currently licensed for routine use in the U . S . [8 , 15 , 26 , 28–38] Most of these adjuvants function as simple ligands for pattern recognition receptors , or are immune stimulating complexes that act on a variety of receptors on many cell types . Adjuplex ( ADJ ) is a carbomer-lecithin-based adjuvant demonstrated to elicit robust humoral immunity and T-cell responses to subcutaneous IAV subunit vaccines in mice . [39] Here we report that intramuscular , subcutaneous , and intranasal administration of ADJ-adjuvanted non-replicating antigens generated robust antigen-specific CTL responses in mice . Systemic CTL memory was induced regardless of whether mice were vaccinated via subcutaneous , intramuscular or intranasal routes , however , only intranasal vaccination provided protection against influenza viral challenge , and correlated with an enhancement of CD103+ CD69+ CXCR3+ TRM and airway memory T cells in the lung . Thus , ADJ can be used to induce systemic and/or mucosal CTL memory to non-replicating antigens , and by tailoring the route of vaccine delivery we can engender CTL-based protective immunity against systemic and mucosal pathogens .
The carbomer-lecithin based adjuvant ADJ was previously demonstrated to induce humoral responses in a wide range of species . [39–44] . Here , using the tractable chicken ovalbumin ( OVA ) antigen model , we investigated whether ADJ ( confirmed to be endotoxin-free ) can elicit antigen-specific CD4 and CD8 T cell responses . OVA-specific naïve CTLs and helper T ( TH ) cells are infrequent in naïve mice , and to enhance assay sensitivity we initially employed adoptive transfer techniques . To assess the activation and expansion of antigen-specific CD8 T cells , we adoptively transferred OVA SIINFEKL-specific naïve TCR transgenic OT-1 CD8 T cells ( Thy1 . 1 ) into congenic Thy1 . 2 B6 mice . One day after cell transfer , mice were immunized by intramuscular ( I/M ) injections with 100 μg OVA in PBS with and without ADJ 20% v/v or Alum 50% v/v . On day 7 after immunization , activated OT-I cells in spleen were enumerated by flow cytometry . Data in Fig 1A and 1B illustrate the potent activation and clonal expansion of donor OT-1 CD8 T cells in spleen of ADJ mice ( 35–100 fold higher ) , as compared to OT-I CD8 T cells in mice vaccinated with PBS or Alum . To assess the activation of OVA-specific CD4 T cells , Ly5 . 1+ve naïve monoclonal I-Ab-restricted OVA 323-339-specific TCR Tg OT-II CD4 T cells were adoptively transferred into congenic Ly5 . 2/B6 mice . 59 One day after cell transfer , mice were immunized by I/M injections with OVA in PBS with and without ADJ or Alum by I/M injection , as above . On day 7 after immunization , activated OT-II cells in spleen and DLNs were enumerated by flow cytometry . The total number of donor CD44HI OT-II CD4 T cells in spleen of ADJ mice were 5-12-fold greater than in PBS or Alum groups ( Fig 1C ) . To examine whether ADJ stimulates expansion of polyclonal OVA-specific CD8 T cells , B6 mice were vaccinated by IM injection with OVA with and without adjuvants , as described above . On day 7 after immunization , the numbers of IFNγ-producing OVA epitope SIINFEKL-specific CD8 T cells in spleen were quantified by intracellular cytokine staining ( ICCS ) . As shown in Fig 1D , IFNγ-producing SIINFEKL-specific CD8 T cells were barely detected in spleens of mice from PBS or Alum groups . By contrast , SIINFEKL-specific cytokine-producing CD8 T cells constituted approximately 12% of the CD8 T cells in spleens of mice from the ADJ group . Additionally , the total number of SIINFEKL-specific CD8 T cells in spleen of ADJ-OVA-immunized mice was markedly greater ( 50 fold ) than in the spleen of PBS-OVA- or Alum-OVA-immunized mice ( Fig 1E ) . We next investigated the degree to which CTL responses to ADJ-OVA vaccination compared to CTL responses elicited by replicating pathogens . Recombinant Listeria monocytogenes ( LM-OVA ) and vaccinia virus ( VV-OVA ) expressing the OVA have been demonstrated to stimulate strong polyclonal CD8 T cell responses to OVA epitopes . [45–49] We compared the immunogenicity of ADJ-OVA with LM-OVA and VV-OVA . On days 7 and 23 after immunization , the number of SIINFEKL-specific IFNγ-producing CD8 T cells in spleen of ADJ-immunized group was similar to those in LM/OVA and VV/OVA-immunized mice ( Fig 1F ) . Collectively , these data clearly demonstrate that akin to live vaccines , vaccination with ADJ elicited potent monoclonal and polyclonal cell-mediated immune responses to the model antigen OVA . [45–49] Inflammation plays a key role in regulating the differentiation of effector and memory CD8 T cells[50] . Therefore , we next sought to determine the effect of ADJ concentration on inflammation and CTL responses to IM vaccination with OVA , and then compare these to CTL responses elicited by subcutaneous ( SQ ) injection . Mice were immunized IM or SQ with 10 μg OVA in PBS or with ADJ at concentrations of 1 , 5 , 10 , and 20% v/v , and spleen and tissues from injections sites were collected 8 days later . Data in Fig 2A illustrates a near-linear positive correlation between increasing concentrations of IM ADJ and increasing frequency of INF-γ-producing SIINFEKL-specific CTLs in the spleen . Following IM injection with 20% ADJ , frequencies of IFN-γ+ CTLs peaked at 3 . 2% , however differences between groups that received greater than 1% ADJ were not statistically significant ( p>0 . 05 ) . As demonstrated in the histological images of the injection sites , there was a significant correlation between increasing dose of ADJ and the extent and cellularity of the inflammatory infiltrate at the injection site . The infiltrate was predominantly composed of vacuolated to foamy histiocytes with fewer neutrophils and small lymphocytes , and small foci of coagulative myofiber necrosis were present at the highest doses of ADJ ( Fig 2B–2F ) . As shown in Fig 2G , the correlation between increasing ADJ concentration and the magnitude of CTL responses to SQ vaccination was similar to IM vaccination . Although the difference between responses to 5% and 10% ADJ concentration was not significant , the differences between responses to 1% vs . 5% , 5% vs . 20% , and 10% vs . 20% were significant ( p<0 . 05 ) . Following SQ vaccination , the ADJ-dose-related increase in the frequency of OVA-specific CTLs again correlated with an increase in the extent and cellularity of the inflammatory infiltrate at the injections , although to a lesser degree than that resulting from IM vaccination . At the highest SQ doses of ADJ , the inflammation extended from the subcutaneous adipose tissue into adjacent muscle ( Fig 2H–2L ) . Based on our results , no more than 5% ADJ is recommended for parenteral use in mice . Taken together , regardless of the route of immunization , 5 or 10% ADJ elicited strong CD8 T cell activation with moderate to low injection site inflammation . A pair-wise comparison of IM versus SQ vaccine responses to 10ug OVA at concentrations of 5 , 10 , and 20% ADJ revealed that the route of vaccination had little effect on the magnitude of the splenic primary CTL responses ( Fig 3A ) . Using the SQ route , we then performed a matrix titration with increasing doses of ADJ with 1 , 3 , or 10 μg OVA ( Fig 3B ) . Statistically significant increases ( p<0 . 05 ) in the frequency of IFN-γ-producing SIINFEKL-specific CTLs in the spleen were observed between 1 μg OVA + 0% ADJ , 10 μg OVA + 5% ADJ , and 10 μg OVA + 10% ADJ . There were no significant differences ( p>0 . 05 ) between any doses lower than 5% ADJ+10 μg OVA when compared to baseline ( PBS group , 1 μg OVA ) . Therefore , we considered responses to only 2 vaccine combinations , 10 ug OVA + 5% ADJ and 10 ug OVA + 10% ADJ , to be above the threshold for reliable detection of an adjuvant effect on systemic CTL responses to SQ vaccination in this system . As there were no significant differences between the magnitude of primary CTL responses to 10 μg OVA at 5 and 10% ADJ , we elected to use 5% ADJ +10 μg OVA for subsequent SQ vaccine studies . Having established that SQ vaccinations with ADJ-OVA can generate strong primary CTL responses , we sought to determine the degree to which the primary CTLs would differentiate into CTL memory . Mice were vaccinated once SQ with 10 μg OVA and either ADJ ( 5% v/v ) , Alum ( 50% v/v ) , or 10 μg ODN-1826 ( CpG ) . Unlike Alum , CpG has been previously demonstrated to generate CTL responses to OVA and influenza virus proteins . [51–53] At 8 and 90 days after vaccination , spleen and vDLN were collected , and the frequency , number and phenotype of SIINFEKL-specific CTLs were characterized by flow cytometry ( Fig 4 ) . The frequency and number of IFN-γ-producing SIINFEKL-specific splenic CTLs generated by ADJ at D8 were significantly ( p<0 . 05 ) greater than 3-fold compared to Alum , and elevated nearly 3–fold compared to CpG ( Fig 4A ) . Additionally , significant differences ( p<0 . 05 ) at D8 were observed in the absolute numbers of polyfunctional CTLs co-producing IFN-γ , IL-2 , and TNF-α resulting from ADJ as compared to Alum or CpG ( Fig 4A ) . At the peak of the CD8 T-cell response , based on the cell surface expression of KLRG-1 and CD127 , effector cells can be classified into two subsets , the short lived effector cells ( SLECs; KLRG1HI/CD127LO ) and memory precursor effector cells ( MPECs; KLRG1LO/CD127HI ) . Data in Fig 4D show that the relative proportions and total numbers of SLECs were higher in the ADJ group , as compared to the Alum and CpG groups . Thus , ADJ appears to promote the differentiation of SLECs , but the total numbers of MPECs were comparable between CpG and ADJ groups ( Fig 4B ) . Next we quantified the frequencies and numbers of SIINFEKL-specific memory CD8 T cells at days 90 after immunization ( Fig 4A ) . Although ADJ stimulated greater expansion of antigen-specific CD8 T cells at day 8 ( Fig 4A ) , there were no differences ( P<0 . 05 ) in the number of memory CD8 T cells between the groups by D90 ( Fig 4A ) . In order to assess memory CD8 T cell-dependent protective immunity 90 days post-vaccination , mice from all 3 groups were challenged by intranasal administration of SIINFEKL-expressing recombinant influenza A/PR/8/34-OT-I ( PR8-OT-I , Fig 4C and 4D ) . We chose PR8-OT-I to focus the protection studies on the role of CTLs , as the MHC-I-restricted SIINFEKL peptide was the only shared epitope between the vaccine antigen and challenge virus . [54] At D6 post-infection , the frequency of IFN-γ-producing CTLs in the ADJ group were higher compared to PBS-OVA and ALU-OVA vaccinated mice ( Fig 4C ) . Surprisingly , despite the strong secondary CTL responses in the lungs ( Fig 4C ) , there was no difference in lung viral titers between the groups ( Fig 4D ) . Collectively these data indicated that , a single SQ vaccination with ADJ/CpG/Alum stimulated systemic CTL memory that failed to augment viral control following a mucosal challenge with influenza virus . Data in Fig 4A–4D suggested that a single parenteral immunization might not elicit sufficient number of memory CTLs to confer protective immunity in the respiratory tract . Therefore , we next sought to determine the effect of prime-boost vaccination on the number of memory CTLs , and the degree to which these cells conferred protective immunity in the lungs . As above , mice were prime-vaccinated SQ with OVA and either ADJ , Alum , or CpG , and the vaccines were repeated 3 weeks later . At 90 days post-boost , the spleen and vDLN were collected , and the frequency and number of SIINFEKL-specific CTLs were characterized by flow cytometry ( Fig 4E ) . At day 90 , the numbers of IFN-γ-producing SIINFEKL-specific CD8 T cells in spleen were significantly ( p<0 . 05 ) greater in the ADJ group , as compared to Alum . Notably , at D90 , the number of polyfunctional memory CTLs were significantly ( p<0 . 05 ) elevated in the ADJ group to approximately 40 times greater than Alum and CpG ( Fig 4E ) . We next assessed the protective capacity of the memory CTLs generated by prime-boost vaccinations . At 90 days post-boost , mice from all 3 groups were challenged by IN administration of PR8-OT-I ( Fig 4F–4H ) , and 6 days later we quantified CD8 T-cell responses and viral titers in the lung . The number of IFN-γ-producing SIINFEKL-specific CTLs in the lung was greater in the CpG and ADJ groups than Alum and PBS-OVA , however only ADJ was significantly ( p<0 . 05 ) increased over PBS-OVA ( Fig 4G ) . Surprisingly , despite the presence of high numbers of antigen-specific CD8 T cells in the lungs of ADJ mice , viral loads in the lungs were not significantly different between any treatment groups ( Fig 4H ) . Collectively this indicates that prime-boost SQ ADJ vaccination can enhance the magnitude of systemic antigen-specific polyfunctional memory T cells and the secondary CD8 T-cell recall responses in the lungs , but this does not enhance viral control following influenza virus challenge . The preceding studies demonstrated that prime-boost ADJ expanded systemic CTL memory following prime-boost vaccination , however the expanded CTLs did not enhance viral control following respiratory viral challenge . We next investigated the nature of CTL memory in the lungs following prime-boost SQ or intranasal ( IN ) vaccination . First we tested whether IN administration of ADJ was tolerated in mice . For vaccination , 3 groups of four 8-10-week old mice were administered intranasally with PBS or PBS plus 10% ADJ . Mice were observed for the first 30 minutes after vaccination , then every 8 hours for 24 hours , then daily until day 7 . A subset of mice from each treatment group was boosted with identical vaccines 21 days after priming , and observed daily for 21 more days . No change in behavior or appetite was observed , and the mice did not lose weight during the first 7 days . Lungs were collected from PBS- and ADJ-vaccinated mice at 1 and 7 days after prime , and 21 days after boost . Histological evaluation of the lungs found no significant abnormalities at any time point ( S1 Fig ) . Thus , ADJ alone did not lead to pulmonary lesions or discernible distress or disease following IN administration to mice . To determine the effect of SQ and IN vaccinations on CTL memory in the lungs , we vaccinated separate groups of mice via SQ or IN routes . The SQ-vaccinated mice received 10 μg OVA in PBS + 5% ADJ while the IN-vaccinated group received 10 μg OVA in + 10% ADJ , and all vaccines were boosted 21 days later . At 21 days post-boost , the frequency , number and phenotype of SIINFEKL-specific CTLs in lung and spleen were characterized by flow cytometry ( Fig 5 ) . IN vaccination resulted in a significant ( p<0 . 05 ) 2 . 5-fold increase in the absolute number of CD8+ lymphocytes in the lung compared to SQ , while no differences in number of CD8+ lymphocytes were observed in the spleen ( Fig 5A ) . The number of SIINFEKL-tetramer+ CTLs in the lungs in the IN group was 8-fold larger than SQ ( p<0 . 05 ) , while the number of tetramer+ cells in the spleen was more than 7-fold larger following SQ vaccination ( p<0 . 01 , Fig 5B ) . In the lung , the absolute number of CD103+ CTLs was 4 times larger in the IN group than in the SQ group ( p<0 . 001 ) , however there were no differences between the two groups in the numbers of CD103+ CTLs in the spleen ( Fig 5C ) . Notably , IN vaccination resulted in a 17-fold greater number of CD103+ CD69+ CXCR3+ resident memory-like CTLs ( TRM ) compared to SQ ( p<0 . 05 , Fig 5D ) . In contrast , intermediate numbers of TRM-like cells that were not significantly different were found in the spleen in both groups . In summary , compared to prime-boost SQ-ADJ vaccination , IN vaccination with ADJ generated substantial populations of CD103+ CD69+ CXCR3+ TRM-like CTLs in the lungs . By comparison , SQ-ADJ vaccine elicited larger frequencies and numbers of memory CTLs in the spleen . Next , we compared functional cytokine-producing CD8 T cells in the lung and spleen of vaccinated mice . In the lung , following IN vaccination , the frequency of IFN-γ-producing SIINFEKL-specific CTLs was 7 fold greater than SQ ( p<0 . 05 ) , while the frequency in the spleen was nearly four-fold higher in the SQ group compared to the IN group ( p<0 . 05 , Fig 5E ) . Similarly , the absolute number of IFN-γ-producing SIINFEKL-specific CTLs in the lungs of IN group was four-fold greater than SQ ( p<0 . 01 ) , while the numbers in the spleen was more than four-fold greater in the SQ group compared to the IN group ( Fig 5F ) . Differences were more pronounced in the polyfunctional CTLs co-producing IFN-γ , IL-2 , and TNF-α . The frequency of these triple cytokine-producing CTLs in the lung in the IN group was four-fold greater than the SQ group ( p<0 . 05 , Fig 5G ) , and translated into 2 times as many triple cytokine+ cells ( p<0 . 001 , Fig 5H ) . Conversely , the frequencies of triple cytokine+ cells in the spleen following SQ vaccination were 5 fold greater than IN ( p<0 . 05 ) , however the absolute number of cells was not significantly different . Remarkably , when taken together , the data in Fig 5 indicate that IN and SQ vaccinations elicit mucosal and systemic CTL memory respectively . We next compared the protective recall capacity of memory CD8 T cells between IN and SQ vaccinees . Mice were prime-boost vaccinated via SQ or IN routes as above . At 21 days post-boost , mice were infected by IN administration of PR8-OT-I , and 6 days later we quantified secondary CD8 T-cell responses and viral titers in the lungs . The number of CD8+ T lymphocytes in the spleen and lungs was the same whether mice were vaccinated with ADJ via the SQ or IN route ( S2 Fig ) . Surprisingly , we found that IFN-γ+ CTL recall responses in the lungs were of similar magnitude in both the ADJ SQ- and IN-vaccinated groups , and these responses were 3-fold or more greater than SQ without ADJ ( p<0 . 05 ) and IN without ADJ ( p<0 . 001 ) ( Fig 5I ) . This demonstrated a clear adjuvant effect in the ADJ groups , and strong recall responses regardless of administration route . Importantly however , despite the similarities in the magnitude of the CTL recall responses , only immunity generated in the ADJ-IN group protected against virus , decreasing viral titers by nearly 100 fold ( Fig 5J ) . To address whether improved viral control in intranasally vaccinated mice was linked to altered distribution of secondary CTLs in the lung airways and parenchyma , we vaccinated groups of mice with ADJ-OVA via SQ or IN routes . At 21 days after vaccination , mice were challenged with PR8-OT-I . Six days after challenge , we quantified SIINFEKL-specific secondary CTLs in airways ( bronco-alveolar lavage; BAL ) and lungs . Data in S3 Fig show that there were ~2-fold more SIINFEKL-specific CTLs in the BAL of IN vaccinated mice than in SQ vaccinated mice . Thus , enhanced influenza viral control following IN vaccination was associated with increased numbers of CTLs in the airways . Data in Fig 5 show that IN vaccination with ADJ-OVA elicited markedly greater number of TRM-like CD8 T cells in the lungs . In addition to the induction of TRM , strategic positioning of memory T cells in the tissues is of critical importance in engendering protective immunity . Using intravascular staining technique in combination with adoptive transfer of OVA-specific OT-I CD8 T cells , we assessed whether SQ and IN vaccination differed in terms of the localization of memory T cells in the lung vasculature , parenchyma or airways . As illustrated in Fig 6 , IN vaccination with ADJ-OVA elicited a significantly ( p<0 . 01 ) greater number of vascular and non-vascular memory OT-I cells in the lungs , as compared to those in SQ vaccinated mice . Memory OT-I CD8 T cells were barely detectable in the airways of SQ ADJ-OVA mice ( Fig 6 ) . In striking contrast , IN vaccination with ADJ-OVA potently elicited a significantly ( p<0 . 01 ) greater number of memory T cells in the airways ( Fig 6 ) . Thus , IN ADJ-OVA vaccination elicits strong CD8 T cell memory in lung vasculature , parenchyma and airways . We next evaluated whether protection afforded by IN ADJ vaccination with the model antigen OVA , illustrated above , would extend to pathogen-associated antigens . We repeated the IN prime-boost vaccine protocol described above , but replaced OVA with varying concentrations of beta-propiolactone ( BPL ) -inactivated influenza A virus . The inactivated influenza virus , PR8-Tex H3N2 , is a reverse-genetics-derived virus containing HA and NA genes from A/Texas/50/2012 H3N2 with the structural genes from A/PR/8/34 H1N1 influenza virus . Typically , commercial influenza virus vaccines are standardized based on the amount of HA . Therefore , viral protein concentrations in our vaccine preparations were normalized based on the amount of HA . It should be noted that the vaccine preparation contains other influenza viral proteins including nucleoprotein ( NP ) , polymerase acidic protein ( PA ) , neuraminidase and matrix proteins , in addition to HA . The inactivated preparation of PR8-Tex virus containing different concentrations of HA1 protein with and without 10% ADJ was used to immunize mice by the IN route; mice vaccinated with ADJ alone served as negative control . Twenty-one days after the initial vaccination , the vaccines were repeated to complete the prime-boost protocol . At 21 days post-boost , mice were challenged by IN inoculation with heterosubtypic influenza A/PR/8/34 H1N1 . Six days after infection , mice were euthanized and lungs were collected to analyze virus-specific CTL responses and viral titers . The number and phenotype of CTLs specific for the influenza NP epitope ASNENMETM ( NP366 ) were assessed by using MHC I tetramers and ICCS for IFN-γ ( Fig 7 ) . The absolute number of NP366-tetramer+ cells in the 3 μg and 10 μg HA+ADJ groups was 10–25 times larger than the other groups ( all p<0 . 05 , Fig 7A ) . In contrast , the absolute number of tetramer+ cells generated by 1 μg HA , 1 μg HA+ADJ , and 3 μg HA were no different than ADJ alone ( Fig 7A ) . However , 3 μg HA+ADJ yielded a 20-fold increase in the number of tetramer+ cells over 3 μg HA alone , and was similar to 10 μg HA+ADJ . Likewise , the absolute number of CD103+ CD69+ CXCR3+ CTLs in the lungs was 2–3 times larger in the 3 μg and 10 μg HA+ADJ groups ( Fig 7B ) . Thus , ADJ potently enhanced the immunogenicity of an inactivated influenza A virus vaccine and elicited potent CTL recall responses to experimental heterosubtypic influenza A virus infection . We also found dose-dependent increases in the magnitude of the CTL responses as measured by frequency and number of IFN-γ+ cells . Fig 7C compares the frequencies of NP366- and PA224-specific IFN-γ-producing CTLs for each treatment group . Consistent with the increase in the numbers of tetramer-binding virus-specific CD8 T cells , the recall responses of IFN-γ-producing CD8 T cells were markedly enhanced in the 3 μg HA+ADJ and 10 μg HA+ADJ groups ( Fig 7C and 7D ) . Additionally , the majority of the NP366- and PA224-specific IFN-γ+ cells were also positive for CD107a ( LAMP-1 ) , a marker of degranulation and indicator of cytolytic capacity ( Fig 7E ) . The increased magnitude of the CTL response at higher doses of HA+ADJ was reflected in markedly decreased lung viral titers ( Fig 7F ) . Viral titers at 1 μg HA and 3 μg HA were similar to ADJ alone , while the group given 1 μg HA+ADJ had a 1 log decrease than ADJ alone ( p<0 . 0001 ) , and the groups given 3μg HA+ADJ and 10 μg HA+ADJ were similar with nearly 3-log decrease in titer compared to ADJ alone ( p<0 . 0001 ) and 2-log decrease compared to 1 μg HA+ADJ ( p<0 . 05 ) . Virus was not detected in 1 mouse each from the 1μg HA , 1 μg HA+ADJ and 3 μg HA groups , while virus was only detected in 2 of 10 mice in the 3 μg HA+ADJ and 10 μg HA+ADJ groups . Thus , the adjuvanted vaccines provided significant protection against heterosubtypic influenza viral challenge . In summary , the addition of ADJ to the vaccines resulted in enhanced viral control following influenza virus challenge , and doses of 3 and 10 μg HA+ADJ resulted in viral titers below the level of detection in 8 of 10 and 7 of 10 mice respectively . This protection correlated with increases in the number of NP366- and PA224-specific CTLs with increased IFN-γ-production , and higher expression levels of CD107a , CD69 , and CD103 . We next investigated possible mechanisms underlying the CTL-activating effects of ADJ . MyD88 is a common adapter protein in the signaling pathway for all toll-like receptors except TLR3 . MyD88 is important for the activation of DCs and other innate cells , and the production of inflammatory cytokines such as IL-18 and IL-1β . Furthermore , MyD88 deficiency has adverse effects on T-cell proliferation , and TH1 differentiation and effector function . [55] Cohorts of WT and MyD88-/- mice were vaccinated SQ with 10 μg OVA and 5% ADJ in 50 μl PBS . On day 8 after vaccination , spleens were collected and the frequency , number and phenotype of SIINFEKL-specific CTLs generated by the prime vaccination were characterized by flow cytometry . The frequency of SIINFEKL-specific CTLs ( Fig 8A ) in spleen of MyD88-/- mice was similar to those in WT mice . However , consequent to increased splenic cellularity in vaccinated MyD88-/- mice , the total number of SIINFEKL-specific CD8 T cells in these mice was higher than in WT mice ( Fig 8B ) . The percentages of SIINFEKL-specific CD8 T cells producing IFN-γ ( Fig 8C ) in WT and MyD88-/- mice were nearly identical . We also analyzed whether MyD88 deficiency affected the differentiation of effector subsets in the spleen . Strikingly , the relative proportions of transition effectors ( TE; KLRG-1HICD127HI ) and SLECs among SIINFEKL-specific CD8 T cells were decreased in the MyD88-/- mice ( Fig 8D ) . Collectively these data demonstrated that MyD88 deficiency has little effect on CTL priming or the magnitude of initial responses to SQ ADJ vaccination , however it skewed the response towards the less terminally differentiated KLRG-1LO phenotype . These data indicate that ADJ activation of CTLs is MyD88-independent , however MyD88 does play a role in CTL differentiation , as previously reported . [55] For our initial investigation into the effect of ADJ on DC populations in vivo , mice were vaccinated by SQ injection with 10 μg OVA in 50 μl of PBS with and without either 5% ADJ . The vDLN were collected at 48 hours after vaccination and analyzed by flow cytometry . Activated conventional DCs were identified as CD11c+ GR-1- MHC-IIHI cells with FSc/SSc parameters greater than the lymphocyte gate ( Fig 9A ) . At 48h , there was an increase in the number of activated conventional DCs in vaccine draining lymph nodes ( vDLNs ) of ADJ mice , as compared to those in PBS mice ( Fig 9B ) . These data suggested that ADJ increases conventional DCs in the draining lymph nodes . We next investigated the effect of ADJ on recruitment and activation of inflammatory cells to the airways and lungs following IN vaccination . Mice were vaccinated by IN inoculation of 50μl PBS with and without 10% ADJ , and bronco-alveolar lavage ( BAL ) fluid and lungs were collected 24 hours later . At 24 h after vaccination , ADJ induced significant alterations in multiple innate cell populations in the lung airways ( BAL fluid ) , and exerted modest effects on select populations in the lung ( Fig 10 ) . Absolute cell counts in the BAL fluid and lungs were not different ( p<0 . 05 ) between groups ( S4 Fig ) . In the BAL , neutrophils accounted for 4% of cells in the ADJ group , but were barely detectable in the PBS group ( p<0 . 001 , Fig 10A ) . Conversely , alveolar macrophages comprised >50% of BAL cells in the PBS group , but only 15% in the ADJ group ( p<0 . 01 , Fig 10B ) . Notably , the percentages of exudative macrophages ( Fig 10C ) , inflammatory monocytes ( Fig 10E ) and inflammatory DCs ( Fig 10F ) in the BAL and lungs of ADJ mice were significantly ( p<0 . 05 ) greater than in PBS mice . ADJ did not alter the percentages of CD103+ ( Fig 10H–10J ) or the CD103- DCs . However , the expressions of MHC-II ( Fig 10K ) , CD40 ( Fig 10L ) and CD86 ( Fig 10M ) on CD103+ DCs in the BAL of ADJ mice were significantly ( p<0 . 05 ) higher than in the PBS mice . Expression levels of cell-surface surface molecules MHC-II ( I-Ab ) , CD40 , and CD86 were used to evaluate the activation of antigen-presenting cells in the BAL fluid and lung . Strikingly , the CD103+ DCs in the ADJ BAL had significantly higher expression of MHC-II , CD40 , and CD86 than PBS alone ( all p<0 . 01 , Fig 10K–10M ) . Expression of MHC-II , CD40 , and CD86 on alveolar macrophages was unaffected , and levels were similar in the in BAL fluid and lung ( Fig 10N ) . Taken together , ADJ significantly augmented in the lung airways ( BAL ) , the activation of CD103+ DCs , which are known to play a prominent role in cross presentation of exogenous antigens to CD8 T cells . [56] In order to investigate the effects of ADJ on antigen uptake , processing and presentation by DCs , we next pursued in vitro studies using DC2 . 4 cells , an immortalized murine DC-like cell line . DC2 . 4 cells are considered to be similar to immature dendritic cells in vivo , and have been used extensively in studies of antigen uptake and processing , and are capable of presenting antigen on class I and class II MHC molecules , and priming naïve T cells . [57–59] For our initial studies , we cultured DC2 . 4 cells in growth media containing 200 μg/ml FITC-OVA with or without 1% ADJ for 1 or 4 hours . We evaluated the expression of cell-surface markers CD11b , CD11c , CD80 , CD86 , CD40 , and IFN-γR1 at both time points . ADJ treatment did not alter the expression levels of CD80/CD86 but increased the expression of CD40 ( Fig 11C ) . CD11c and IFN-γR1 were not detectable at any time point . In a parallel experiment , we pulsed DC2 . 4 cells with OVA or OVA with ADJ/LPS for 30 minutes . At different time points after the pulse , we quantified MHC II expression on DC2 . 4 cells by flow cytometry . As shown in Fig 11A and 11B , MHC-II expression was detected in all treatment groups at 30 minutes , with 40% of cells in the LPS group , and 20–25% of cells in the LPS-OVA and ADJ-OVA groups , while it was expressed by less than 2% of cells in the media-only group ( Fig 11B ) . Over the first 2 hours , frequency of MHC-II+ cells rapidly decreased to less than 15% in the LPS-OVA and LPS-ONLY groups , however the frequency in the ADJ-OVA group rapidly increased to over 50% of cells at 2 hours , remained greater than 40% at 24h , and was slightly elevated over the other groups at 48h . Thus , data in Fig 11A–11C suggested that ADJ is a potent activator of DC2 . 4 cells . Because of the large disparity in MHC-II expression at 24h after pulse , we repeated the pulse-chase experiment with only media , FITC-OVA , and FITC-OVA-ADJ , and looked at the overall expression of class I and II MHC molecules , and specifically looked for differential expression of MHC-I molecules bearing SIIFNKEL peptide at 24h after pulse . Expression of all MHC-I alleles , as evaluated with pan-reactive antibodies against H2-Db and H2-Kb , was not significantly different between groups ( p>0 . 05 ) ( Fig 11D ) . Expression of the SIINFKEKL-bearing MHC-I molecules was evaluated with antibodies specifically targeting the H2-Kb/SIINFEKL complex . Interestingly , there were no differences between media and the OVA-only groups , however the ADJ-OVA group exhibited a modest but reproducible increase in Kb/SIINFEKL expression ( Fig 11E ) . These data suggested that ADJ might enhance antigen presentation by DCs to CD8 T cells . DQ-OVA is OVA labeled with a self-quenched BODIPY-FL dye . The primary green fluorescence of DQ-OVA only occurs after proteolytic cleavage , indicating active antigen processing . A secondary red florescence can be detected if the degradation products form adequately large aggregates , or excimers , within the cells . To determine whether ADJ affected antigen processing , we pulsed DC2 . 4 cells with DQ-OVA with and without ADJ for 30 minutes , and evaluated the primary and excimer fluorescence by flow cytometry 24 and 48h later . As illustrated in Fig 12A and 12B , the magnitude of primary and excimer fluorescence was significantly increased from 24 to 48h in both groups . Both primary fluorescence and excimer fluorescence in the ADJ group at 24h were twice the magnitude of the PBS group , though differences between groups were smaller at 48h . These data along with data in Fig 11 suggested that ADJ potently increases or accelerates antigen processing and presentation by DC2 . 4 cells . To further investigate antigen processing and intracellular localization in DC2 . 4 cells , we repeated the DQ-OVA pulse-chase experiment and analyzed the cells by confocal microscopy at 3 , 6 , and 24h post-pulse . As illustrated in Fig 12C , the amount of primary and excimer fluorescence and the intracellular localization of the antigen were dynamically altered by the presence of ADJ . In the absence of ADJ , DQ-OVA was primarily localized to punctate cytoplasmic aggregates randomly distributed throughout the cells with localization of some larger aggregates in perinuclear regions in some cells ( Fig 12C ) . The majority of the punctate aggregates exhibited green primary florescence with the larger perinuclear aggregates exhibiting more mixed ( yellow ) and excimer fluorescence ( orange-red ) . In contrast , the ADJ-treated cells exhibit a homogenous to amorphous distribution of green primary fluorescence throughout the cytoplasm , with fewer to larger globoid aggregates tending toward mixed yellow fluorescence , and striking bright orange , irregularly segmental bands of fluorescence subtending the cell membrane ( Fig 12C ) . By 6 hours , the green punctate aggregates in PBS group were larger and slightly less well defined , and were accompanied by slightly increased in amorphous green background fluorescence ( Fig 12C ) , while the ADJ group generally exhibited a mix of membrane-associated orange linear to irregular orange aggregates ( Fig 12C ) . The PBS group exhibited minor changes by 24 hours , primarily characterized by an increase in homogeneous green cytoplasmic fluorescence , and larger and less distinct cytoplasmic aggregates with increased orange fluorescence ( Fig 12C ) . In contrast , the homogenous green cytoplasmic fluorescence in the ADJ group was completely replaced by large irregular orange aggregates irregularly distributed under the cell membrane , and randomly within the cytoplasm ( Fig 12C ) . Further , the cells in the ADJ group more frequently assumed a unipolar morphology with the sub-membranous aggregates often concentrated away from an eccentric nucleus . Confocal imaging of these pulse-chase experiments demonstrate that treatment with ADJ alters antigen uptake and processing within DC2 . 4 cells , the intracellular localization of the antigen , and the cytomorphology of the DC2 . 4 cells . To directly assess whether ADJ enhances the ability of DCs to activate and drive clonal expansion of naïve antigen-specific CD8 T cells , we pulsed DC2 . 4 cells with OVA mixed with PBS or ADJ for 30 minutes . Subsequently , antigen-pulsed DC2 . 4 cells were cultured with CFSE-labeled OT-I TCR transgenic CD8 T cells for 72 hours . Data in Fig 12D illustrate that DCs exposed to ADJ or OVA induced minimal proliferation of OT-I CD8 T cells . In striking contrast , DC2 . 4 cells exposed to OVA plus ADJ stimulated robust proliferation of OT-I CD8 T cells . Collectively data presented in Figs 11 and 12 indicate that ADJ treatment enhanced the ability of DC2 . 4 cells to uptake , process and present antigen to naïve CD8 T cells .
Here we report that ADJ , a Carbomer-lecithin-based adjuvant elicits potent mucosal and systemic cell-mediated immune responses to non-replicating antigens . We find that systemic CTL memory induced by parenteral immunization fails to confer protective immunity against influenza virus in the respiratory tract . Strikingly however , IN immunization with an experimental antigen or inactivated influenza A virus adjuvanted with ADJ induces CTL memory in the respiratory tract and confers robust protective immunity to influenza virus challenge . These findings provide key insights into the induction of CTL memory-dependent protective immunity by non-replicating antigens in the respiratory mucosa . Our investigations into the kinetics of the CTL responses to prime-only and prime-boost protocols using the model antigen OVA revealed that booster vaccinations can dramatically enhance the magnitude of secondary effector responses . We also found that systemic and respiratory CTL memory resulting from SQ prime-boost vaccination is capable of mounting substantial recall responses in the spleen and lung , but these responses did not protect against challenge with influenza virus . Unlike antibodies that can rapidly and widely diffuse into areas of inflammation , memory CTLs are a heterogeneous population with divergent capacities for tissue trafficking and effector function and the initial programming of different CTL memory subsets affects the nature of their recall responses . [60] Why CTL recall responses generated by SQ vaccination with ADJ did not mediate enhanced viral control following respiratory viral challenge is unknown . It is possible that the nature of the systemic CTL memory is such that it could provide robust protection against systemic challenge . Alternatively , enhanced viral clearance occurs in ADJ/OVA vaccinated group at later time points after challenge ( beyond day 6 ) . We are designing additional experiments to explore these possibilities . Because the phenotype of respiratory CTL memory is likely influenced by the nature of the initial challenge , we next investigated the viability of the intranasal route of vaccination with ADJ in mice . We find that SQ vaccination yielded substantial systemic CTL memory pools and limited respiratory CTL memory , while IN vaccination yields the reverse . Given the disparity in the anatomic distribution of the CTL memory resulting from SQ and IN vaccines , the magnitude of recall CTL responses in the lungs after influenza challenge was nearly identical with IN and SQ vaccination . However , further analysis show that improved viral control following IN vaccination and not SQ vaccination is associated with greater number of virus-specific CTLs in the airways . It remains to be determined whether airway CTLs are descendants of airway memory cells and/or memory cells found in the pulmonary vasculature , lung parenchyma or circulation . However , the association of greater number of memory T cells and secondary CTLs in airways with viral control in IN vaccinated mice is consistent with a report that airway memory CD8 T cells are necessary and sufficient for protection against influenza virus challenge . [61] It has been previously reported that CTL-based mucosal immunity is greatly enhanced by the establishment of TRM , a unique subset of CTL memory that persists in the lungs . [62] TRM is characterized by expression of the cell surface molecules CD103 and CD69 , which facilitate the retention of memory CTLs within the lung parenchyma and airway epithelium . [62] Indeed , protection by IN vaccination not only correlated with a significantly larger pool of CTL memory cells within the lungs , but also with enhancement of TRM . Further , a greater proportion of the CTLs resulting from IN vaccination were capable of co-producing IFN-γ , TNF-α , and IL-2 suggesting a greater functional capacity as well . Significantly , only IN and not SQ vaccination elicited substantial number of memory CD8 T cells in the airways . Taken together , data presented in this manuscript strongly suggests that enhanced influenza virus control following IN vaccination with ADJ-OVA is linked to induction of TRMs and airway memory CD8 T cells . To address the concerns that responses to ADJ-OVA vaccines may not be representative of vaccines formulated with pathogen-associated antigens , we expanded our vaccine-challenge system to pathogen-associated antigens while minimizing confounding humoral immunity . For the antigen , we chose the BPL-inactivated A/PR8xTexas H3N2 strain of influenza virus , and A/PR/8/34 H1N1 strain for challenge . The PR8xTexas strain was derived from the PR8 strain , and is genetically identical to PR8 with the exception of HA and NA genes . We found that IN ADJ-PR8xTexas vaccines did indeed elicit robust CTL memory and recall responses in the lungs that were similar in magnitude to those previously generated by OVA-based vaccines . We observed that a minimum antigen dose is required to elicit these responses , and that higher doses did not necessarily elicit correspondingly larger CTL memory pools . The higher doses also provide substantial protection against heterosubtypic challenge , and the protection afforded by vaccine containing 3 μg HA ( and other viral proteins ) is equivalent to that of the 10 μg HA-containing vaccine preparation . In our analysis , we find that the 3 μg HA-containing vaccines generate slightly larger CTL responses than vaccines containing 10 μg HA , and that this correlates with increases in the number of virus-specific CD8 T cells producing IFN-γ and amount of IFN-γ produced by each cell , number and degranulation . Further , increased IFN-γ production correlated with increases in CD107a , which indirectly measures the capacity for degranulation and cytolytic function . Lastly , the 3 μg dose of HA yields the largest absolute number of CTLs expressing CD103 and CD69 , suggesting that this dose is optimum for protection in this model , and that antigen dose may positively or negatively affect the generation of TRM . Why vaccination with greater amount of viral antigens reduces the number of memory CD8 T cells ( especially TRMs ) remains unknown . It is possible that higher levels of antigens might drive terminal differentiation of effectors and diminish the development of memory T cells in the respiratory tract . Because the antigen dosage of currently licensed influenza vaccines is based on HA content , we also formulated our vaccines based on HA concentration rather than the concentration of NP or other structural proteins . As NP protein may be twice as abundant as HA , it is likely that the actual concentration of antigen recognized by CTLs is substantially greater than suggested by HA concentration . [63] Therefore subsequent investigations would benefit from quantitation of specific antigens so that optimal doses can be determined . Further , our intent was to focus on the cell-mediated immune responses to the immunodominant NP epitopes in this model , and altered the HA and NA of the challenge virus to limit the contribution of potentially protective humoral responses targeting these proteins . We do recognize , however , that protection in this model is less constrained to CTL responses than the OVA model . ADJ can elicit potent antibody responses , and epitopes such as the conserved M1 and M2 proteins may be targets for protective humoral immunity . [64 , 65] Nonetheless , we have clearly demonstrated that ADJ can elicit robust cell-mediated immune responses to non-replicating antigens . The mechanisms by which adjuvants influence the generation of the CTL responses to non-replicating antigens are unclear . [14 , 31] Some adjuvants appear to function primarily via signaling through innate pattern-recognition receptors ( PRRs ) . This includes CpG DNA , which depends upon signaling via TLR9 receptors to activate antigen-presenting cells such as plasmacytoid dendritic cells to upregulate of antigen cross-presentation . [35 , 52 , 53] Others , such as the immune-stimulating complex adjuvant Iscomatrix™ , may function independent of TLRs , yet but activate inflammasome signaling and alter intracellular localization of antigen in ways that facilitate antigen cross-presentation . [66 , 67] How ADJ elicits CTL responses is unknown . Recent studies on the immune-stimulating properties of ADJ found evidence that ADJ did not induce TLR or NLR signaling in vitro . [68] This does not preclude a role for these signaling pathways in the various interactions underlying the generation of CTL responses in vivo , therefore we repeated our SQ vaccine experiments in MyD88-deficient mice . Virtually all TLR signaling except TLR-3 is abolished in these mice , enabling us to evaluate the role of TLRs in CTL activation and primary responses . We find that primary CTL responses to ADJ were not affected by MyD88 deficiency . We did however note alterations in the SLEC:MPEC differentiation states of the primary CTLs in MyD88-deficient mice . Thus , ADJ might engage MyD88 signaling to promote differentiation of effector CD8 T cells . APC activation affects both antigen-processing and migration to the secondary lymphoid tissues where naïve CTLs are activated . In the lymphoid tissues , the CTL activation depends on the amount , and strength , and duration of antigen signaling . [69] Therefore , we next looked at the effect of ADJ on APC activation , antigen uptake and processing , and migration in vivo . We find that SQ ADJ vaccination quickly increases the numbers of conventional DCs . A more extensive characterization of the immune cells in the airways ( BAL ) and lung following IN ADJ vaccination revealed much more profound changes in the composition of the inflammatory cell populations . Key alterations in the ADJ BAL were the recruitment of neutrophils , depletion of alveolar macrophages ( aMΦ ) , and increased frequencies of exudative macrophages , inflammatory monocytes and inflammatory DCs . These findings suggest a strong pro-inflammatory environment in the BAL of ADJ-vaccinated mice . The aMΦ were not activated in either group , but the small population of CD103+ DCs in the ADJ group were strongly activated , expressing high levels of CD40 , CD86 , and MHC-II . In contrast to the BAL , the ADJ-treated lung displayed only modest alterations in the composition of innate immune cells . In preliminary experiments , transcription levels of inflammatory cytokines within the lung tissue revealed upregulation of IL-1 , type I interferon , and TNF-α were strongly upregulated compared to PBS . Thus ADJ treatment induces a pro-inflammatory environment and significant alterations in the innate cell populations in the lung , with a dramatic shift from aMΦ to inflammatory DC’s , yet it is the small populations of CD103+ DCs that are highly activated . Since , migratory CD103+ DCs in lungs and intestines ) are crucial for cross presentation in vivo[70 , 71] , it is possible that ADJ enhances the priming of CTLs in the lungs by activating this cell type . To gain insights into ADJ’s mechanism of action , we investigated the effect of ADJ on the DC2 . 4 cell line , which is capable of direct and cross-presentation of antigen , and can fully activate CD4+ and CD8+ T cells . [57 , 59] In an experiment where DC2 . 4 cells were continuously exposed to ADJ , DC2 . 4 cells displayed increased CD40 expression , a strong indicator of activation . In a pulse-chase model , the increase in CD40 expression persisted longer and to a greater degree in these cells , significantly longer than with LPS exposure . Interestingly , in the same experiment , ADJ markedly upregulated MHC-II expression , in the absence of antigen . While we did not find similarly dramatic alterations in global MHC-I expression on these cells , but ADJ did induce a small but detectable increase in the subset of MHC-I bearing SIINFEKL peptide . Further , in comparison to untreated cells , ADJ-treated DC2 . 4 cells potently stimulated the proliferation of naïve OT-I CD8 T cells in vitro . In sum , ADJ appears to augment antigen processing and presentation by DCs , at least in vitro . Perhaps more impressively , in vitro studies with DQ-OVA found that antigen processing was upregulated by ADJ , with ADJ-treated cells exhibiting far greater primary and excimer fluorescence as evaluated by flow cytometry . Confocal imaging confirmed these findings , and demonstrated that the increased fluorescence was also associated with aberrant intracellular localization of the DQ-OVA . One of the putative key features of the Iscomatrix adjuvant is the capacity to rapidly translocate antigen to the cytosol . [67] It is possible that the diffuse green fluorescence spread throughout cells in the ADJ-treated group at 3 and 6 h post-treatment reflects a similar effect . The etiology of the irregular linear aggregates of DQ-OVA under the cell membrane in the ADJ group is unclear . Further investigations into the nature and behavior of ADJ-induced alterations in antigen uptake and processing are ongoing , with particular focus on co-localization studies and identification of the structures contributing to the excimer fluorescence observed with flow cytometry and confocal microscopy . Still , the molecular mechanisms underlying the capacity for ADJ to elicit CMI are not clear . As the effects of ADJ appear to be MyD88-independent , and independent of other commonly recognized PRRs such as NLRs in vitro , additional mechanisms must be investigated . The components of ADJ themselves , polyacrylic acid polymers ( carbomer ) and soy lecithin may guide future studies . Polyacrylic acid ( PAA ) , widely used in pharmaceuticals , has been previously shown to have potent antiviral effects in mice . [72 , 73] This effect is tied to its chemical structure and its ability to elicit type I interferons when administered at a range of doses by multiple routes . [74] The lecithin component of ADJ is largely composed of membrane phospholipids , primarily phosphatidylcholine and phosphatidylinositol . [75] In ADJ , lecithin is formulated as a nano-emulsion , and speculated to fuse with cell membranes similar to liposomes . [39] An additional possibility is that the inflammatory environment at the vaccination site results in the oxidation of these phospholipids . This could occur in the extracellular environment or within phagocytes , prior to being released during cell death . Oxidized membrane phospholipids are potent immunostimulatory molecules and signal through scavenger receptors CD36 and SR-B1 on macrophages and other innate cells . [76] Notably , CD36 signaling is TLR and integrin-independent , and involves signaling through fyn , p38 map kinase , JNK1 and JNK2 . [77] In addition , CD36 signaling via SRC kinases leads to NFkB signaling and expression of IL-1 and TNF-α via NLRP3 inflammasome activation . [78] Thus the inflammatory responses to PAA and oxidized phospholipid-induced activation of CD36 are highly consistent with the findings in our study , particularly MyD88 independence , and rapid upregulation of type I interferons and TNF-α at the vaccination site . Additional mechanisms to consider are activation of the classical complement by interactions between phosphatidylcholine and C-reactive protein , and the abundant availability of phosphatidylcholines as a substrate for arachidonic acid metabolites . [79] Given the complex machinations of ADJ’s effect on innate immunity it is likely that one or more of these pathways are involved at different points in the generation of CTL memory . In many of our studies , we compared the adjuvanticity of ADJ with Alum and CpG . While it is abundantly clear that ADJ is superior to Alum in eliciting primary , memory and recall CD8 T cell responses , ADJ and CpG induced comparable CD8 T cell responses to SQ vaccines . Regardless of the adjuvant used , CD8 T cell memory induced by SQ vaccination failed to provide protection against influenza virus in the respiratory tract . This reinforces the idea that vaccination by parenteral routes might not effectively program antibody independent CTL memory-dependent protective immunity in the respiratory tract . In this study , we did not compare the relative efficacies of ADJ and CpG in inducing protective immunity in the lungs following IN vaccination . Using a mouse model similar to ours , it has been shown that IN vaccination with CpG-OVA induced memory CTLs that reduced influenza virus replication in the lungs by ~2 logs , as compared to no adjuvant controls . [80] Thus , ADJ and CpG might program comparable levels of CTL-dependent protective immunity to influenza virus in the lungs . Future experiments will compare CTL-dependent protective immunity induced by ADJ and CpG . Collectively , our studies present Adjuplex as a potent provocateur that manipulates key facets of the innate response to effectively generate cell-mediated immunity to non-replicating antigens . It does so by promoting the recruitment and activation of antigen-presenting cells to sites of vaccination , and induces local production of inflammatory cytokines leading to APC activation and likely further APC recruitment . Activated APCs in turn exhibit alterations in antigen uptake and processing , and enhanced trafficking of DCs to the DLN . The magnitude and character of the resulting CTL responses are strongly influenced by antigen and adjuvant dose , and route of vaccination . Indeed , vaccine route played a key role in the capacity for ADJ vaccines to generate protective mucosal immunity . The molecular mechanisms by which ADJ works remain to be discovered , however future investigations will likely yield vital insights into the conditions required for cell-mediated immune responses to non-replicating antigens .
Six- to eight-week-old C57BL/6 ( B6 ) mice were purchased from the National Cancer Institute ( Bethesda , MD ) or from restricted-access SPF mouse breeding colonies at the University of Wisconsin-Madison Biotron Laboratory . OT-I TCR Tg mice carrying Thy1 . 1 allele , OT-II TCR Tg mice carrying the Ly5 . 1 allele , and MyD88-deficient B6 . 129P2 ( SJL ) -Myd88tm1 . 1Defr/J mice on the C57BL/6 background were provided by Dr . Bruce Klein ( Department of Pediatrics , School of Medicine , University of Wisconsin-Madison , Madison , WI ) . [81 , 82] All mice were housed in specific-pathogen-free conditions in the animal facilities at the University of Wisconsin-Madison ( Madison , WI ) . All experiments were performed in accordance with the protocol ( Protocol number V1461 ) approved by the University of Wisconsin School of Veterinary Medicine Institutional Animal Care and Use Committee ( IACUC ) . The animal committee mandates that institutions and individuals using animals for research , teaching , and/or testing much acknowledge and accept both legal and ethical responsibility for the animals under their care , as specified in the Animal Welfare Act ( AWA ) and associated Animal Welfare Regulations ( AWRs ) and Public Health Service ( PHS ) Policy . Listeria monocytogenes expressing chicken ovalbumin as a full-length protein ( LM-OVA ) was provided by Dr . Hao Shen ( University of Pennsylvania School of Medicine , Philadelphia , PA ) . Mice were infected with 5x104 CFU LM-OVA per mouse by tail-vein injection . Recombinant vaccinia virus expressing chicken ovalbumin as a full-length protein ( VV-OVA ) was provided by Dr . Jack Bennink ( National Institutes of Health , Bethesda MD , ) . [83] Mice were infected with 5 × 105 PFU VV-OVA per mouse by intraperitoneal injection . Influenza virus strain A/PR/8/34 H1N1 ( PR8 ) and strain A/H1N1/PR/8/34 H1N1–OT-I ( PR8-OVA ) , which expresses the SIINFEKL peptide of chicken ovalbumin , were a kind gift from Dr . Paul Thomas ( St . Jude Children’ Research Hospital , Memphis , TN ) . [54] Influenza A strain PR8xTexas H3N2 ( PR8-Tex ) , a reassortant virus composed of ( A/PR8/H3N2 influenza virus containing HA [H3] and NA [N2] proteins from A/Texas/50/2012 ) was generated by reverse genetics in the Kawaoka Laboratory , as previously described[84] . The PR8-Tex virus was amplified by passage in eggs , and inactivated with 0 . 1% beta-propiolactone as previously described . [85] Inactivated virus was purified by sucrose gradient ultracentrifugation , and loss of infectivity was evaluated by inoculation into eggs . The viral genome was confirmed by sequencing , and the virus concentration was determined by Western blot for the hemagglutinin HA1 domain . For infection challenge studies , mice received a single intranasal inoculation of 500 PFU of PR8-OVA , and were humanely euthanized 6 days after infection . Lung tissues were frozen at −80°C in plain RPMI 1640 immediately after euthanasia for virus quantification . Tissues were rapidly thawed and homogenized in the RPMI media , and cleared supernatants were titrated on MDCK cells using standard methods . For adoptive transfer experiments , spleens were harvested from OT-I TCR transgenic Thy1 . 1+ or OT-II TCR Tg Ly5 . 1+ mice . Spleens were mechanically processed into a single-cell suspension and erythrocytes were lysed by incubation with 0 . 9% NH4Cl for 1 minute . Then splenocytes containing 105−106 OT-I or OT-II TCR transgenic T cells were transferred to Thy1 . 2+ or Ly5 . 2+ mice , respectively , and mice were vaccinated 24 hours later . Hen egg white ovalbumin grade V ( OVA ) was purchased from Sigma-Aldrich ( St . Louis , MO ) . ODN 1826 CpG oligonucleotide adjuvant was purchased from InivivoGen ( San Diego , CA ) , and was reconstituted in sterile phosphate-buffered saline ( PBS ) . Adjuplex ( endotoxin-free ) was provided by Advanced BioAdjuvants , LLC ( Omaha , NE ) . Imject™ Alum ( ALM ) was purchased from Thermo Fisher Scientific ( Pierce , Rockford , IL ) . OVA antigen was prepared by dissolving crystallized OVA in sterile phosphate-buffered saline and passage through a 0 . 2 μm syringe filter . Ovalbumin or inactivated viruses were mixed with adjuvants by forceful pipetting and vortexing until homogenous , then aliquoted into 0 . 5 cc tuberculin syringes with 28g needles for intramuscular and subcutaneous injection , or individual 50 μl aliquots for intranasal inoculation via 200 μl pipette . For intramuscular vaccines , 25 μl of vaccine was injected bilaterally into the tibialis muscles as previously described . [86 , 87] For subcutaneous vaccines , the tail base was cleaned with 70% ethanol and 50 μl of the vaccine was administered subcutaneously . For intranasal vaccinations , mice were briefly anesthetized with 3% isoflurane in oxygen and the vaccine was slowly inoculated into the nares . The immortalized DC2 . 4 dendritic cell-like line was a gift of Dr . Kenneth Rock ( Department of Pathology , University of Massachusetts Medical School , Worcester , MA ) DC2 . 4 cells were maintained in DMEM high glucose ( 4500 mg/L , Life Technologies ) supplemented with 10% fetal bovine serum , 100 U/ml penicillin G , 100 g/ml streptomycin sulfate , and 50 μM 2-ME . For in vitro cell assays , FITC-conjugated chicken ovalbumin ( FITC-OVA ) and DQ-OVA ( Molecular Probes ) were purchased from Life Technologies , Inc . and reconstituted in DC2 . 4 growth media . 1 x 106 DC 2 . 4 cells were plated per well in a 96-well plate and grown overnight at 37C 5% CO2 . The media was aspirated and replaced growth media only , or growth media containing FITC-OVA ( 200μg/mL ) , DQ-OVA ( 200 ug/mL ) , FITC-OVA + LPS ( 5 μg/mL ) with and without Adjuplex 1% V/V . For pulse-chase experiments , the treatment media was washed off after 30 minutes and completely replaced with growth media every 24 hours . For other experiments , DC2 . 4 cells were incubated in the treatment media for the indicated time . After treatment , cells were harvested , washed 3 times , stained for viability with an amine-reactive dye as indicated below , incubated with Fc-Block ( BD Biosciences , San Diego , CA ) in PBS at a 1:200 dilution for 15 minutes , and stained with fluorochrome-conjugated antibodies as below . Single-cell suspensions of mononuclear cells from lymph nodes , spleen , lung , and bronchoalveolar lavage were prepared using standard techniques . Briefly , prior to antibody staining , some cells were stained for viability with either Fixable Viability Dye eFluor® 506 or eFluor® 780 ( eBiosciences , San Diego , CA ) , or Ghost Dye™ Red 780 ( Tonbo Biosciences , San Diego , CA ) according to manufacturer’s instructions . Fluorochrome-labeled antibodies against the cell-surface antigens Thy1 . 1 , Thy1 . 2 , Ly5 . 1 ( CD45 . 1 ) , Ly5 . 2 ( CD45 . 2 ) , CD4 , CD8a , CD8b , CD44 , CD62L , KLRG-1 , CD127 , CD103 , CD69 , CXCR3 , CD11b CD11c , CD40 , CD80 , CD86 , Siglec-F , F4/80 , Gr-1 , Ly6C , Ly6G , and intracellular antigens IFN-γ , TNF-α , IL-2 , CD107a , T-bet , and Eomes were purchased from BD Biosciences ( San Jose , CA ) , Biolegend ( San Diego , CA ) , eBioscience ( San Diego , CA ) , or Tonbo Biosciences . The antibody recognizing the H2-Kb:SIINFEKL complex was purchased from eBiosciences . [88] The anti-granzyme B antibody was purchased from Invitrogen ( Grand Island , NY ) . Fluorochrome-conjugated H2-Kb tetramers bearing the ovalbumin peptide SIINFEKL ( OT-I ) , and H2-Db tetramers bearing influenza nucleoprotein peptide ASNENMETM ( NP366 ) and acidic polymerase peptide SSLENFRAYV ( PA224 ) were obtained from the NIH Tetramer Core Facility ( Emory University , Atlanta , GA ) . Cells were incubated with tetramer for 60 minutes on ice in the dark , and with antibodies for 30 minutes on ice in the dark . Intravascular staining for vascular CD8 T cells in the lungs was performed as previously described . [89] Briefly , 5 minutes prior to euthanasia , mice were infused with fluorochrome-labeled anti-CD8β antibodies . Cells from lungs were stained with anti-CD8α and other surface markers . Cells positive for both CD8α and CD8β were considered as vascular CD8 T cells . For intracellular cytokine staining , 1x106 cells per well were plated on flat-bottom tissue-culture-treated 96-well plates . Cells were stimulated for 5 hours at 37C in the presence of human recombinant IL-2 ( 10 U/well ) , and brefeldin A ( 1 μl/ml , GolgiPlug , BD Biosciences ) , with one of the following peptides: SIINFEKL , NP366 , PA224 ( thinkpeptides® , ProImmune Ltd . Oxford , UK ) at 0 . 1ug/ml , or without peptide . After stimulation , cells were stained for surface markers , and then processed with Cytofix/Cytoperm kit ( BD Biosciences , Franklin Lakes , NJ ) . Permeabilized cells were transferred to FACS buffer for acquisition , while surface-stained cells were fixed with 2% paraformaldehyde in PBS for 20 minutes , then transferred to FACS buffer . All samples were acquired on FACSCalibur , LSR II , or LSRFortessa ( BD Biosciences ) analytical flow cytometers . Data were analyzed with FlowJo software ( TreeStar , Ashland , OR ) . For confocal microscopy , 1x104 DC2 . 4 cells were plated in growth medium on 12mm diameter #1 . 5 coverslip glasses ( Warner Instruments , Hamden , CT ) in 24-well tissue culture plates ( Corning Costar , Sigma Aldrich ) . Cells were grown overnight , and treated as described for activation and antigen-processing assays . After treatment , the coverslip glass was washed three times with Dulbecco’s PBS containing Ca++/Mg++ ( DPBS , Life Technologies ) , fixed in 1% PFA in DPBS for 10 minutes , washed 3 times with DPBS , permeabilized with 0 . 1% saponin at room temperature , and incubated with DAPI ( Life Technologies ) at 1ug/ml for 5 minutes at room temperature . Cover glass was mounted in Vectashield ( Vector Laboratories ) and mounted on Permafrost microscope slides ( Thermo Scientific ) . Cells were imaged on a Leica SP8 confocal laser-scanning microscope within 48 hours of mounting . Following euthanasia , tissues surrounding IM and SQ injection sites were collected en-bloc and fixed in 10% neutral phosphate-buffered formalin ( NBF , Sigma-Aldrich ) . Lungs were perfused in situ by intratracheal administration of 750 μl NBF , then the trachea was ligated , and the lungs and heart were removed en bloc and immersed in NBF . Preserved tissues were paraffin embedded , replicates of 5-μm-thick sections were prepared for each tissue , and sections were stained with standard hematoxylin and eosin . Tissue sections were evaluated histologically by a board-certified veterinary anatomic pathologist ( DJG ) , and photomicrographs were created with an Olympus BX41 microscope , DP71 camera system , and cellSens software ( Olympus , Tokyo , Japan ) . Data statistics were calculated with Prism version 6 . 0g for Mac OS X ( GraphPad Software , La Jolla California USA , www . graphpad . com ) . Student’s two-tailed t-test , and one-way ANOVA analyses were used to calculate the statistical significance of differences between groups , and significance was defined at p < 0 . 05 . | Current respiratory-virus vaccines typically employ non-replicating antigens and rely solely on the generation of humoral responses for protection . Viruses such as influenza can mutate and escape these responses , thereby limiting immunity and necessitating revaccination . Cell-mediated immunity ( CMI ) could provide broader protection by targeting viral components that infrequently mutate , however non-replicating vaccines capable of inducing CMI are not available . Impediments to vaccine development include an incomplete understanding of the nature of protective respiratory CMI and a lack of vaccine adjuvants capable of eliciting CMI to non-replicating antigens . Using a mouse model , we characterized the protective immunity afforded by CMI responses to non-replicating vaccines formulated with the adjuvant Adjuplex . We found that vaccination via either the subcutaneous or intranasal route was capable of inducing potent CMI responses . However , only intranasal vaccination protected against challenge with heterosubtypic influenza viruses . This protection correlated with enhancement of T cells with a resident-memory phenotype in the lungs . Additionally , mechanistic studies showed that Adjuplex affects antigen-presenting cells via activation and alteration of antigen uptake , processing , and presentation . The current studies: ( 1 ) identified an adjuvant that elicits protective CMI to respiratory viral pathogens; ( 2 ) suggested that stimulation of protective CMI in the respiratory tract requires intranasal vaccine delivery . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"blood",
"cells",
"flow",
"cytometry",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"immune",
"physiology",
"pathology",
"and",
"laboratory",
"medicine",
"spleen",
"immunology",
"neuroscience",
"learning",
"and",
"memory",
"vaccines",
"preventive",
"med... | 2016 | Effective Respiratory CD8 T-Cell Immunity to Influenza Virus Induced by Intranasal Carbomer-Lecithin-Adjuvanted Non-replicating Vaccines |
The contribution of regulatory versus protein change to adaptive evolution has long been controversial . In principle , the rate and strength of adaptation within functional genetic elements can be quantified on the basis of an excess of nucleotide substitutions between species compared to the neutral expectation or from effects of recent substitutions on nucleotide diversity at linked sites . Here , we infer the nature of selective forces acting in proteins , their UTRs and conserved noncoding elements ( CNEs ) using genome-wide patterns of diversity in wild house mice and divergence to related species . By applying an extension of the McDonald-Kreitman test , we infer that adaptive substitutions are widespread in protein-coding genes , UTRs and CNEs , and we estimate that there are at least four times as many adaptive substitutions in CNEs and UTRs as in proteins . We observe pronounced reductions in mean diversity around nonsynonymous sites ( whether or not they have experienced a recent substitution ) . This can be explained by selection on multiple , linked CNEs and exons . We also observe substantial dips in mean diversity ( after controlling for divergence ) around protein-coding exons and CNEs , which can also be explained by the combined effects of many linked exons and CNEs . A model of background selection ( BGS ) can adequately explain the reduction in mean diversity observed around CNEs . However , BGS fails to explain the wide reductions in mean diversity surrounding exons ( encompassing ∼100 Kb , on average ) , implying that there is a substantial role for adaptation within exons or closely linked sites . The wide dips in diversity around exons , which are hard to explain by BGS , suggest that the fitness effects of adaptive amino acid substitutions could be substantially larger than substitutions in CNEs . We conclude that although there appear to be many more adaptive noncoding changes , substitutions in proteins may dominate phenotypic evolution .
Only about 1% of the mammalian genome encodes proteins [1] . Purifying selection is apparent in protein-coding sequences , where both diversity within species and divergence between species is markedly reduced . Additionally , selection prevents deleterious mutations in protein-coding genes from rising in frequency , which creates an excess of rare variants . In some mammalian species , there is also evidence of widespread adaptive evolution in proteins , manifest as an excess of substitutions compared to that expected under a neutral model [2]–[4] . However , little is known about selection operating on regulatory sequences present in the noncoding fraction of the genome or about the strength of selection operating on adaptive substitutions in general . Studies of interspecific divergence of noncoding DNA sequences among mammalian species imply that only ∼5% of the noncoding genome is conserved , and therefore is likely to be subject to purifying selection [5] . Conversely , a high proportion ( as much as 80% ) of the genome has been classified as “functional” by virtue of displaying reproducible biochemical signatures [6] , but the evolutionary significance of this has been challenged [7]–[8] . The overall evolutionary rate and fitness change associated with advantageous and deleterious mutations in noncoding DNA are essentially unknown . The question of the relative contributions of protein-coding versus noncoding DNA to genetic variation for fitness and adaptive evolution is not new . King and Wilson [9] argued that changes in proteins ( also known as structural changes ) are unlikely to explain the myriad phenotypic adaptations that distinguish humans and chimpanzees , and proposed that regulatory change in noncoding sequences affecting the timing and tissue-specificity of gene expression must therefore dominate adaptive evolutionary change . There is empirical evidence from genetic mapping experiments both supporting [10]–[12] and contradicting this view [13] . On theoretical grounds , it has been argued that regulatory change should cause fewer harmful pleiotropic effects than mutations affecting proteins [10] , and there is evidence in Drosophila and mammals from patterns of diversity within and between species suggesting the presence of positive selection in noncoding DNA [14]–[17] . Here , we study genome-wide nucleotide diversity within the house mouse subspecies M . m . castaneus and nucleotide divergence from two outgroup species , M . famulus , and Rattus norvegicus ( the brown or Norway rat ) . Genome sequences of wild mice were obtained from individuals sampled from their ancestral range in NW India [18]; see Materials and Methods . The effective population size of house mice from this region has been estimated to be nearly 106 [2] . Nucleotide diversity levels approach those seen in some invertebrates and some signals of natural selection are stronger than seen in humans , making wild house mice a powerful system for studying natural selection in the mammalian genome [16]–[18] . We use these data firstly to estimate and compare the distributions of fitness effects ( DFEs ) for deleterious mutations occurring in protein-coding genes and conserved noncoding elements ( CNEs ) , many of which have functions regulating gene expression . Secondly , we attempt to quantify the relative numbers of adaptive coding and noncoding substitutions based on nucleotide divergence between M . m . castaneus and the two outgroup species . However , it is important to consider not only the rate , but also the fitness effects of adaptive substitutions , since a category of sites showing a low rate of adaptive substitution could make a larger contribution to adaptation if the substitutions tend to have stronger effects on fitness . The relative contributions of structural and regulatory adaptive fitness change has not previously been explored in any depth . To address this issue , we investigate patterns of nucleotide diversity surrounding protein-coding exons and CNEs . As a consequence of genetic linkage , diversity is expected to be reduced close to sites where advantageous mutations regularly spread to fixation ( causing selective sweeps ) , but may also be reduced as a consequence of purging of deleterious mutations ( background selection ) . We attempt to quantify the relative impact of these two forces on diversity at sites closely linked to exons and CNEs .
We calculated mean nucleotide diversity ( π ) and mean divergence ( d ) from the brown rat for nonsynonymous sites , synonymous sites and UTRs of protein-coding genes and CNEs and the flanks of CNEs ( Figure 2A ) . Diversity of synonymous sites and CNE flanks approaches 1% , consistent with previous surveys of protein-coding genes [2] and CNEs [17] . Diversity is therefore ∼10-fold higher than in humans and higher than in other wild house mouse populations [3] , [18] , [27] . As expected , mean nucleotide divergence between mouse and rat and diversity within M . m . castaneus at CNEs and nonsynonymous sites are much lower than the corresponding values for synonymous sites and CNE flanks ( Figure 2A ) , consistent with net purifying selection . This is corroborated by an excess of rare variants and a more negative Tajima's D than at synonymous sites ( Figure 2A ) . We obtained the folded site frequency spectrum ( SFS = the distribution of minor allele frequencies across sites ) for M . m . castaneus from each of our focal classes of sites ( nonsynonymous sites , UTRs and CNEs ) and for tightly linked , putatively neutral sites . We then used these SFSs to estimate the distribution of fitness effects ( DFE ) of deleterious mutations in the focal classes ( Table 1 , S1 , S2; Figure S1 ) using the program DFE-alpha [28] ( see Materials and Methods ) . To do this , we first fit a demographic model to the neutral SFS by maximum likelihood with a step change in population size . Then , assuming the estimated demographic parameters from this model ( Table S1 ) , we fit a gamma DFE of new deleterious mutations to the selected SFS . This model makes the assumption that advantageous mutations are sufficiently rare as to make a negligible contribution to polymorphism . The inferred gamma DFEs for each selected site class are highly leptokurtic ( Table 1 ) , i . e , estimates of the shape parameter of the distribution , β , are all less than 0 . 19 . This implies that most nonsynonymous mutations are strongly deleterious ( we infer that 77% have fitness effects greater than Nes = 10 ) , and that there are relatively few nearly neutral nonsynonymous mutations ( ∼20% have fitness effects less than Nes = 1 ) ( Table 1 ) . In particular , this estimated proportion of nearly neutral amino acid-changing mutations is markedly lower than estimates for human populations [28]–[30] which is likely to be a consequence of the larger recent effective population size of wild house mice . However , nearly neutral mutations in UTRs and CNEs are relatively abundant and strongly deleterious mutations are less frequent ( we estimate that ∼65% and ∼44% of mutations have fitness effects less than Nes = 1 , whereas ∼25% and ∼37% have fitness effects greater than Nes = 10 , for UTRs and CNEs respectively ) . If the DFE is highly leptokurtic ( as in these cases ) , the mean effect of a deleterious mutation is highly sensitive to the frequency of strongly deleterious mutations that are essentially absent from the data , and cannot be accurately estimated . However , the estimated proportions of mutations with selective effects in different ranges in more robust to the model assumptions [28] , [31] . We used DFE-alpha , incorporating an extension of the McDonald-Kreitman test [32] , [33] , to estimate the proportion of nucleotide differences ( α ) in protein-coding genes , UTRs and CNEs that were driven to fixation by positive selection ( Materials and Methods ) . To do this , we calculate the mean fixation probability of a deleterious mutation relative to a neutral mutation ( un ) from the inferred DFE ( Table 1 ) , and use this to estimate the expected number of fixed differences between mouse and rat attributable to neutral and slightly deleterious mutations [32] . An excess of observed substitutions compared to that expected can be quantified as α , the fraction of adaptive substitutions , or ωa , the rate of adaptive substitution relative to the rate of neutral substitution ( ds , Materials and Methods ) . We obtained high estimates of α and ωa for all site classes , indicating that there has been widespread adaptive evolution in proteins , UTRs and CNEs ( Table 1 , Fig . 1B , Table S2 ) . In order to check the robustness of our estimates of the DFE and α and ωa , we investigated alternative neutral reference classes of sites and fitted a more complex demographic model to the neutral site data . Similar results are obtained if we assume AR sites within introns as a neutral reference ( Table S3 ) . We then investigate an alternative demographic model incorporating two step changes in the population size . This gave similar fits to the neutral SFSs as the single step change model ( Table S4 ) . Our results were not substantially affected by the choice of the neutral reference for CNEs , and we obtained similar results for CNEs located near to and far from exons ( Table S5 ) . Previous work suggest that estimates of the DFE and the rate of adaptive evolution from DFE-alpha are substantially robust to mis-specification of the demographic model [28] , [32] , as long as the estimated demographic parameters provide a good model fit to the observed folded neutral SFS . For similar reasons , estimates of the rate of adaptive evolution are robust to the presence of genetic linkage , although this can lead to misinference of the demographic parameters [34] . We investigated the extent to which our estimates of α and ωa depend on our assumption that the DFE is gamma distributed by fitting a second model of the DFE in which selective effects are divided into three discrete bins [31] . In this model we allow three bins of mutation effects at s = 0 ( i . e . neutral ) , s = s2 ( estimated ) and s = 1 ( i . e . , lethal ) and estimate the relative proportion of mutations with these effects ( p1 , p2 and p3; Table S6 ) . Our estimates of α and ωa depend on un , which is obtained from the DFE ( Materials and Methods ) . For both the gamma model and the discrete effects model , our estimates of un are similar , although the resulting estimates of α and ωa are somewhat lower when using the discrete model ( Table 1 , S6 ) . The results are also robust to the choice of outgroup ( Table S2; see Text S1 ) . The total rate of adaptive substitution per generation ( na ) attributable to a particular class of sites is proportional to its rate of adaptive substitution ( ωa ) : ( 1 ) where nt is the total number of sites in that class ( Table 1 ) and ds is the divergence for the neutral reference class of sites . A comparison of na estimates among nonsynonymous sites , UTRs and CNEs therefore suggests that the majority ( >70% ) of adaptive nucleotide substitutions in the murid genome occur in CNEs and that only about 20% occur at nonsynonymous sites in protein-coding genes ( Table 1; Figure 2C ) . It has been suggested that the pattern of nucleotide diversity around selected sites can provide information about the rate and selective effects of adaptive mutations [35] , [36] . Diversity is expected to be reduced at sites partially linked to an adaptive substitution , so we would expect to observe reductions in diversity around substituted sites if some fraction of substitutions were driven by positive selection . In our data we see substantial reductions in mean π in the regions flanking nonsynonymous sites that have experienced a nucleotide substitution between M . m . castaneus and the closely related M . famulus ( Figure 3 ) . We also see slightly increased d ( mouse-rat divergence ) near nonsynonymous substituted sites , indicative of locally correlated rates of nonsynonymous substitution , which may reflect local variation in constraint or mutation rate . Notably , we still see reductions in mean π when controlling for d . However , the reduction in mean π/d around substituted nonsynonymous sites is very similar to that around substituted synonymous sites , as observed by Hernandez et al . [36] ( Figure 3; Text S1 ) . Furthermore , a similar drop in mean π/d is also observed around nonsynonymous sites that have not experienced a substitution . As discussed below , these patterns can be explained by the presence of many linked selected sites jointly contributing to reductions of π/d , obscuring the contributions of individual substitutions , and/or effects of background selection . Thus , there appears to be limited information present in measures of mean π/d in flanking regions to infer the strength of positive selection at substituted sites , at least in wild house mice and hominids , unless the effects of selection on linked sites are taken into account . Information about the frequency and strength of selected mutations may also be contained in the pattern of nucleotide diversity and divergence surrounding functional genomic elements , such as exons or CNEs [37] . Both diversity and divergence are expected to be influenced to similar extents by variation in the mutation rate or negative selection acting in the flanking sequences . However , diversity is also expected to be influenced by selection at linked sites ( either in the form of hitchhiking or background selection ) . Thus , it is possible in principle to tease apart the contribution of linked selection to diversity by controlling for divergence . We see reductions in diversity and divergence in the flanks of CNEs , indicative of direct negative selection , consistent with a previous study [17] , and indicating that our defined CNEs may not include the entire underlying functional elements ( Figure 1 ) . However , we observe substantial reductions in π/d , around both exons and CNEs , suggesting that there is an effect of positive or negative selection in exons and CNEs on diversity in the linked sequences flanking them [36] , [38] , [39] . Notably , mean π/d in the flanks of both exons and CNEs is well approximated by a negative exponential function , and the fit of this model implies that π/d is reduced over a ∼10-fold wider area around exons than around CNEs ( Figure 1 ) . We obtained similar results if sequences that are most likely to experience direct negative selection are excluded from exon and CNE flanks ( Text S1 and Table S7 ) , supporting the conclusion that the observed reductions reflect the action of selection on the linked functional elements . When interpreting these patterns , it is important to account for the spatial and length distribution of selected elements in the genome . In particular reductions in π/d ( Figure 1 ) will be influenced by the following factors: 1 . Selection on exons can contribute to the diversity reductions observed in the flanks of CNEs and vice versa ( i . e . , the panels in Figure 1 are not independent from each other ) . 2: CNEs and exons vary dramatically in length . 3: Both CNEs and exons are clustered ( Figure S2 ) , such that more than 80% of exons lie within 10 Kb of another exon and more than 75% of CNEs are within 1 Kb of another CNE . 4: Exons tend to cluster with CNEs ( Figures S3 ) . This implies that reductions in diversity around exons and CNEs ( which extend over ∼100 Kb and ∼10 Kb , respectively ) are likely to be affected by selection , not only on the nearest exon or CNE , but many other closely linked CNEs and exons as well . To address these issues , we attempted to model π/d calculated within 200 bp or 1 Kb non-overlapping windows throughout the genome ( at non-exonic and non-CNE sites ) . Although we attempted to fit a variety of simpler models ( Text S1 ) , the best fitting model ( model C , as measured by r2 ) was one where we considered the effects of all linked selected sites on π/d , rather than just the effect of the nearest exon and CNE ( Table S8 ) . In this model , we reasoned that π/d for any given non-overlapping window depends on the neutral ( unreduced ) level of π/d and the combined reductions attributable to all linked selected bases ( in exons or CNEs ) . We assumed that these diversity reductions decay exponentially around each selected base within an exon or CNE and that the individual reductions combine multiplicatively to give the total predicted reduction ( see Materials and Methods ) . We estimated two exponential rate parameters , for exonic and CNE sites , simultaneously . The results from this model imply that a single exonic or CNE site reduces diversity at a linked neutral site by 0 . 0013% and 0 . 019% , respectively . Therefore , in the absence of any influence from other linked selected exons or CNEs , we predict a reduction in π/d of only 0 . 23% and 1% , for an average exon ( 176 bp ) or CNE ( 54 bp ) , respectively . Note that these predicted reductions , are much smaller than the observed average reductions of 16% and 14% around exons and CNEs respectively ( Figure 1 ) . The discrepancy can be explained by the contribution of multiple selected elements on the observed reductions in π/d . To further investigate this finding , we obtained π/d predictions from model C for each genomic window , binned these values according to distance from the nearest exon/CNE , and plotted the average for each bin against distance ( Figure 4 ) . These predictions give a good fit to the observed data . This model also implies that reductions in diversity attributable to a single exonic base are ∼5× wider on average than those attributable to a single conserved noncoding base ( parameters p3 and p5 for model C in Table S8 ) . Our results suggest that the substantial reductions in diversity around CNEs and exons could be explained by the cumulative effects of many linked selected sites . Can these reductions in neutral diversity be explained solely as a result of deleterious mutation ? To investigate this , we attempted to model background selection ( BGS ) [40] , a process whereby reductions in diversity result from selection against deleterious mutations at linked sites [41] . We used the approach and software described by McVicker et al . [39] to predict reductions in diversity throughout the genome as a result of deleterious mutations occurring within exons and CNEs . In this model we assume a mutation rate of 3 . 79×10−9 , and a recombination rate of 0 . 528 cM/Mb [42] . We also assume that reductions from each element type ( exons and CNEs ) combine multiplicatively and that the effects of deleterious mutations in CNEs and exons follow an exponential distribution , the parameters of which we estimate by least-squares ( Materials and Methods ) . We subsequently tested the fit of a gamma distribution of deleterious mutation effects with a range of shape parameters ( 0 . 25 , 0 . 75 and 2 . 0 ) and found that the fit to the data ( as measured by the r2 ) was very similar across the range of shape parameters tested ( data not shown ) . The best-fitting BGS model suggests that deleterious mutations at nonsynonymous and CNE sites have absolute mean selection coefficients of 4×10−5 and 2×10−5 , respectively ( Table S8 ) , implying mean scaled selection coefficients ( Nes ) of ∼44 and ∼22 , respectively ( assuming an effective population size in mice of 5 . 6×105 and dominance coefficients of 0 . 5 ) . BGS explains nearly as much variation ( in terms of r2 ) as model C described above . The results are also consistent in that they indicate that the substantial drops in mean diversity around exons and CNEs are caused by the cumulative effects of many selected elements , and that only small reductions are attributable to a single average length exon or CNE ( i . e , the BGS model predicts a drop of ∼1% for both ) . We binned predictions of relative diversity according to distance from the nearest exon/CNE and plotted average for each bin against distance ( Figure 4 ) . Notably , although there is a good match between the average predicted diversity and the observed mean diversity in the flanks of CNEs , the width of the predicted reductions in diversity around exons fits the observations poorly . Note , however , that we also infer that there is substantial adaptive evolution in both CNEs and exons , so the observed diversity patterns are unlikely to be solely a consequence of BGS .
We used genome-wide polymorphism data to estimate the DFE of new deleterious mutations for non-synonymous sites , UTRs and non-coding elements conserved across mammals . The inferred DFEs are highly leptokurtic in all cases . Under a model of mutation-selection balance , the equilibrium genetic variance is proportional to the product of the mean effect of a deleterious mutation and the genomic mutation rate . We obtain much higher estimates of the mean effect of a nonsynonymous mutation than a CNE mutation ( Table 1 ) . This suggests that standing genetic variation for fitness could be dominated by variation in protein-coding genes . Precise partitioning of the variation is not possible , however , because estimates of the mean effect of a deleterious mutation depend on the frequencies of very rare , strongly selected alleles that are virtually absent from the data [43] . We calculated two measures of adaptation , the rate adaptive substitution relative to neutrality ( ωa ) and the fraction of substitutions that are adaptive ( α ) . The method [32] attempts to account for the contribution of slightly deleterious mutations to polymorphism and divergence and the impact of recent demographic change . Consistent with previous work [2] , [16] , [17] , estimates of α and ωa are quite high , suggesting that ∼30% and ∼20% of nonsynonymous and CNE/UTR substitutions , respectively , are driven to fixation by positive selection ( Figure 2 ) . These estimates disregard slightly advantageous mutations , but their contribution to the folded SFS is essentially indistinguishable from that of slightly deleterious mutations [43] . These values are considerably higher than corresponding estimates for humans [15] , [30] , [32] , which may reflect the substantially higher Ne in wild house mice . By multiplying estimates of ωa by the number of sites in each category , we can calculate the numbers of adaptive substitutions in protein-coding genes , UTRs and CNEs . Only ∼20% of adaptive substitutions are inferred to occur in coding sequences , and the vast majority of the remaining 80% occur in noncoding DNA , mostly CNEs . This is largely driven by the far higher number of sites in CNEs than in protein-coding genes , since estimates of ωa are quite similar among the site categories . Estimates of the number of adaptive noncoding substitutions are likely to be underestimates if there are elements that experience high rates of adaptive evolution , or elements specific to the murid lineage that we failed to identify on the basis of conservation . Additionally , there are many mammalian elements identified as part of the ENCODE project [6] lacking conservation that show reduced diversity in humans [44] . This has been interpreted as a signature of weak purifying selection [44] , but a more plausible explanation is that the effect is attributable to selection on linked sites ( see below ) . Sattath et al . [35] have demonstrated the existence of reductions in nucleotide diversity in D . simulans close to sites that show an amino acid substitution between D . melanogaster and D . simulans . In contrast , synonymous substitutions do not show diversity reductions , implying that recent selective sweeps at nonsynonymous sites have purged diversity at linked sites . A parallel study in humans by Hernandez et al . [36] observed reductions in nucleotide diversity of similar magnitudes close to synonymous and nonsynonymous substitutions between human and chimpanzee . In our study in wild mice , we see a similar pattern to that observed in humans . Furthermore , we have also shown that reductions in diversity of a similar magnitude can be observed close to non-substituted sites . These observations suggest that diversity reductions around recently substituted non-synonymous sites in mammals are principally caused by selective sweeps or BGS at closely linked functional sites rather than at the focal sites themselves . The contrast between Drosophila and mammals is likely to be explained by the 10-fold narrower scales over which diversity is purged close to amino acid substitutions in D . simulans , compared to humans and mice [35] , [36] , resulting in little power to apply the Sattath et al . /Hernandez et al . approach in hominids or murids . This is corroborated by our observation that diversity reductions surrounding exons and CNEs are caused by the influence of selection on multiple , linked elements and that little of the reductions can be explained by selection within the focal elements themselves . Hernandez et al . [36] argue that their observations suggest a lack of complete selective sweeps in humans . Although this may be true , the conclusion does not necessarily follow from their analysis . To investigate whether the BGS model on its own can explain the drops in diversity around exons and CNEs , we obtained predictions of relative diversity in the genome using the approach and software described by McVicker et . al . [39] to find the parameters of a distribution of selection coefficients for nonsynonymous and CNE sites that best fit the genome-wide distribution of nucleotide diversity . We found that the best-fitting model accurately predicts the depth of the reductions in diversity in the immediate flanks of exons and CNEs , but under-predicts the width of reductions observed in the flanks of exons ( Figure 4 ) . An explanation for our failure to accurately predict the width of reductions in π/d in the flanks of exons is that recurrent selective sweeps may play a role in explaining the observed patterns of diversity around exons . We investigated whether the inferred DFE from the BGS model is consistent with the amount of diversity observed within exons and CNEs themselves . We calculated the diversity predicted at nonsynonymous sites and CNEs , from the DFEs obtained from the best fitting BGS model of McVicker et al . [39] . Assuming that alleles act additively , the estimated DFEs predict a nucleotide diversity of 0 . 32 and 0 . 45 within exons and CNEs , respectively . The BGS model therefore overpredicts diversity at nonsynonymous sites ( the observed values are 0 . 14 for 0-fold sites and 0 . 18 for 2-fold sites ) , but is close to the observed value of 0 . 42 for CNEs . We then used ML estimates of the DFE obtained from DFE-alpha to predict drops in diversity around exons and CNEs under a BGS model ( Material and Methods ) . The resulting r2 of the model is lower than the best fitting model , but not substantially so ( 1 . 57 compared to 1 . 88 for 1 Kb windows and 0 . 35 compared to 0 . 44 for 200 bp windows ) . Furthermore , when binning data by distance from the nearest exon or CNE and averaging π/d within bins , the predicted mean π/d is almost indistinguishable from that obtained from the best fitting model ( Figure S4 ) . Thus , we conclude that diversity data from within selected elements is largely consistent with deleterious mutations causing the observed reductions in π/d observed in the flanks of CNEs , but is inconsistent with reductions observed in the flanks of exons . This analysis can only qualitatively compare the fit of different models , and it is not possible to exhaustively explore all models . However , we have found that two models to predict diversity in the flanks of exons and CNEs that use very different information ( the best fitting BGS model and a BGS model parameterised using DFEs inferred from DFE-alpha ) give similar predictions , and both provide a poor fit to the pattern of diversity reduction around exons . We have obtained evidence that there are many more adaptive noncoding than coding nucleotide substitutions . This does not necessarily imply , however , that noncoding substitutions dominate adaptive change , because the fitness effects of coding substitutions could be larger than those of noncoding mutations . Patterns of diversity around sites under positive selection shed some light on this issue , but for two reasons , it is not possible to reach a firm conclusion . First , the rate of adaptive fitness change ( ΔW ) depends on the distribution of fitness effects of advantageous mutations ( f ( sa ) ) . Consider a model in which adaptive mutations occur at a rate μa at na sites and are fixed with probability u ( sa ) : ( 2 ) Assuming that selection is strong relative to genetic drift , equation ( 2 ) becomes ( 3 ) which is proportional to the additive genetic variance for fitness from new advantageous mutations . Prediction of ΔW therefore requires knowledge of the average squared effect of an advantageous mutation , but we have not attempted to estimate this . Second , adaptive evolution may be driven by complete selective sweeps originating from recent , novel mutations , from partial sweeps involving older , standing genetic variation or from some combination of these . The pattern of diversity around sites that have experienced positive selection is influenced by whether sweeps have been full or partial [45] . Recent evidence suggests that partial sweeps comprise a substantial proportion of adaptive events in humans [46] . If we assume that adaptation is dominated by complete sweeps , our results would imply that positive selection is substantially stronger on coding than on CNE mutations . This is firstly because BGS fails to account for the wide mean diversity reductions around exons ( unlike CNEs ) , and secondly because the width of a region purged of variation from new mutations by positive selection is predicted to be proportional to sa [37] . Diversity is reduced by 50% around single exonic and CNE sites at distances of ∼38 Kb and 3 Kb , respectively , implying that the strength of selection on advantageous mutations in exons could be substantially higher than in CNEs . Our results therefore lend some support to the idea that , although there are many more adaptive noncoding changes , the net effect of coding change may exceed that of noncoding fitness change . However , it is not possible to exclude the existence of infrequent substitutions dominating adaptive fitness change involving alleles with large fitness effects that do not show up in the average patterns of diversity drops around selected sites .
Of 38 M . m . castaneus individuals from Himachal Pradesh , India [18] , 21 were identified as belonging to a single cluster , labelled “North-West” , based on a STRUCTURE analysis [47] of 60 microsatellite loci . Protein-coding gene sequences of 15 of these individuals [2] suggested that one individual ( H10 ) showed signs of inbreeding , having a substantially higher proportion of homozygous SNPs than other individuals , and was excluded from further analysis . We selected ten individuals for whole genome sequencing ( individuals H12 , H14 , H15 , H24 , H26 , H27 , H28 , H30 , H34 and H36 ) . We also sequenced a single individual of Mus famulus , originating from Tamil Nadu , India ( locality Kotagiri ) , obtained from the Montpellier Wild Mice Genetic Repository , as an alternative outgroup to the rat . For each individual , five standard Illumina paired-end sequencing libraries were made with fragment sizes from 300–550 bp . We generated between 21–42× mapped sequence coverage ( average 29× ) across the samples ( Table S9 ) . The libraries were run at a mixture of 76 , 100 and 108 bp read lengths on the Illumina GAIIx and HiSeq platforms . In order to check the identity of all of the Illumina sequencing lanes , we used a set of SNPs previously identified by Sanger sequencing of the same individuals [2] . This was done by using SAMtools mpileup ( v0 . 1 . 16 ) in conjunction with BCFtools and GLFtools [48] to generate genotype likelihoods for each sample . All lanes were confirmed to be the correct genotype . The M . m . castaneus Illumina sequencing reads were aligned to NCBIM37/mm9 unmasked reference genome with SMALT ( http://www . sanger . ac . uk/resources/software/smalt/ ) using the following parameters: -k 13 -s 6 . The individual lane BAM files were merged to the library level , PCR duplicates were marked using Picard ( http://picard . sourceforge . net/ ) , and a single merged BAM file was produced per sample . All of the sequencing data are available from the European Nucleotide Archive via accession ERP000231 . We created bcf ( genotype likelihood ) files for each chromosome from the individual BAM files using ‘samtools mpileup’ with options -D -S -g -m 2 -F 0 . 0005 -P ILLUMINA [48] . We then used ‘bcftools view’ with options -A -g to obtain SNP calls for every site in the genome . bcftools allows the specification of a prior site frequency spectrum ( SFS ) , which can improve genotype calls at each site . We obtained an approximate prior SFS for the genome using an iterative approach ( see http://samtools . sourceforge . net/mpileup . shtml ) . We used bcftools to estimate a posterior SFS for all sites on chromosome 1 , then used the SFS as a prior ( using option -P ) for a second call to bcftools , and iterated until the prior and posterior converged . The final posterior SFS was then used as a prior to obtain genotype calls for the whole genome , which were used to obtain site frequency spectra for specific genomic regions . We called all genotypes using an approximate M . m . castaneus reference sequence , which is identical to the NCBIM37/mm9 reference sequence , but with all SNPs at a frequency of >0 . 5 replaced with the major allele observed amongst the M . m . castaneus individuals . This reduced the number of SNP calls representing fixed differences between the mouse reference and the M . m . castaneus sequences and reduced the number of triallelic SNP calls ( which can arise if a variant in M . m . castaneus also has a fixed difference to the reference ) . We filtered genotype calls having no mapped reads or where the Hardy-Weinberg equilibrium P-value reported by SAMtools ( from a χ2 based test ) was less than 0 . 0002 . In principle , our approach for inferring the SFS ( by iterating the prior and posterior SFSs ) should be close to optimal , since it makes most efficient use of the information on each individual's genotype . However , we obtain very similar inferred SFSs for the genome as a whole if we infer the allele frequency at each site by maximum likelihood using SAMtools [48] or by simply using the genotype calls for each individual ( Figure S5A ) . We also investigated the effect on the inferred SFSs of removing sites with low coverage ( measured as total coverage across all individuals ) ( Figure S5B ) . As expected , given that mean coverage was ∼30× and more than 80% of sites had coverage >10× in all individuals ( Table S9 ) , there was very little effect of removing sites with low coverage ( up to a total coverage of 100 reads across all individuals ) . We observed a skew towards low frequency variants as the severity of filtering of the data increased . However it is possible that the sites that pass these stringent filters represent a biased set of sites in terms of diversity and allele frequency , as argued previously [49] . M . famulus is divergent from M . m . castaneus and the M . m . musculus reference sequence and assembly of its genome sequence is therefore worth some special consideration . Specifically , divergent regions in the genome will reduce the efficacy or accuracy of the final sequence because reads with too many differences to the reference cannot be mapped properly to the reference genome . To mitigate this effect , we used an ‘iterative mapping’ approach where successive rounds of read mapping are conducted and after each iteration a new genome sequence is generated for use in the next iteration . In effect , we are converting the original reference genome to a M . famulus reference genome by changing the divergent sites to match those from M . famulus . Therefore , regions of high divergence where reads cannot be aligned initially may eventually be assembled as divergent sites are eliminated from the reference . We aligned each of the lanes of data to the reference genome independently using BWA v0 . 5 . 9 [50] . We used SAMtools v0 . 1 . 16 mpileup to call variant SNPs and converted variant positions in the reference to match the all high quality homozygous variant calls ( genotype quality , GQ>40 ) . With this approach we ignored all shared polymorphisms that are heterozygous in our M . famulus sample and more importantly we also ignore potential indel divergence . We discarded indels and SNPs neighbouring indels to avoid converting regions of the genome where read mapping has erroneously generated indels due to repeats and to retain the same position indices of genomic features as the reference genome . The new reference was then used to repeat this process . The most improvement in terms of positions covered and reads mapped occurred in the first and second iteration , after which the gains made with successive iterations plateaued . We carried out a total of five iterations over which the number of reads mapped increased from 72 . 6% to 84 . 0% and the median coverage improved from improved from 23× to 25× . After five iterations we called the final genotype of the M . famulus genome using the same methods described for the M . m . castaneus . We obtained gene coordinates from the Ensembl database version 62 for a total of 18 , 761 autosomal protein-coding genes , which are orthologous between mouse and rat . We used these to obtain the exonic sequence of the canonical spliceform of each gene for mouse and rat , and constructed sequences for M . famulus and M . m . castaneus based on the genotype calls . In order to preserve the reading frame , we constructed alignments using the translated amino-acid sequences and back-translated to the DNA sequence . We considered two classes of nonsynonymous sites: zero-fold and two-fold degenerate ( for two-fold sites , only transversions were considered as nonsynonymous ) . However , in order to be able to construct an SFS for two-fold nonsynonymous we require an appropriate estimate of the number of invariant sites . To do this , we calculate the approximate fraction of two-fold sites at which a mutation would be nonsynonymous . We assume all transversions are nonsynonymous and calculate the ratio of transitional and transversional mutations from that observed at four-fold degenerate sites across all genes in a comparison of M . m . castaneus and rat . UTRs were identified as sequences annotated as being transcribed , but not forming part of a coding sequence . We identified conserved noncoding elements ( CNEs ) in the mouse genome using phastCons [21] . We ran phastCons on multiz alignments of 28 vertebrate genomes based on Build 36 . 1/hg18 version of the human genome ( see http://hgdownload . cse . ucsc . edu/goldenPath/hg18/multiz28way ) . We restricted the phastCons analysis to the data from placental mammals only and also excluded information from the mouse/rat lineage , thereby avoiding an ascertainment bias which could affect divergence within the mouse/rat lineage . We used parameters for phastCons that have been tuned to produce ∼5% conserved elements in the genome ( expected-length = 45 , target-coverage = 0 . 3 , rho = 0 . 31 ) and that were used to produce the conservation scores and “most conserved” tract for the UCSC genome browser . Having identified CNEs in the human genome , we mapped the human coordinates onto the NCBIM37/mm9 mouse reference genome using the UCSC liftOver tool . We subdivided CNEs into those proximal to exons ( within 20 Kb of an exon , pCNEs ) and those distal to exons ( dCNEs , more than 20 Kb away from any exon ) . A 20 Kb cutoff was chosen such that dCNEs would be located outside of the flanks of exons that show reduced levels of π/d ( see Figure 1 ) . As such , dCNEs can be considered to be outside the shadow of protein-coding exons , allowing us to examine patterns of nucleotide diversity within CNEs and their flanks independently from protein-coding exons and their flanks . We attempted to obtain an appropriate neutral reference sequence to use when inferring rates of adaptation from our analysis of the SFS for CNEs . Ideally , this neutral reference sequence should be free from direct selection and tightly linked to the focal sequence such that the neutral reference and focal sequence can be assumed to share the same genealogy . We used sections of the genome near to CNEs , but far enough away such that mean divergence to rat approximates that for ancestral repeats ( 0 . 168 at non-CpG-prone sites ) which occurs at ∼500 bp away from CNEs on average ( Figure 1 ) , Assuming that ancestral repeats are a good a priori candidate for neutrally evolving sites , as has been previously suggested [26] , then this would imply that these regions are , on average , evolving neutrally . For each CNE we selected two sections split equally between a section upstream of the CNE and a section downstream of the CNE , each offset from the CNE start/end by 500 bp , such that the total length of the two sections was equal to the length of the CNE . From these sections of sequence , we masked any other CNEs and any annotated exons . Remnants of transposable element repeat sequences that were inserted prior to the split of mouse and rat ( ancestral repeats ) provide a good choice for putatively neutral sequence , as mutations within these sequences are unlikely to affect host fitness . In order to identify potential ancestral repeat sequences , we obtained a set of coordinates of all identified repeat sequences in the mouse genome from Ensembl and removed from this set any repeat sequences that failed to align well with rat ( those whose alignments contained >50% gaps or whose mean divergence , after applying a Kimura 1980 [51] correction for multiple hits , exceeded 1 ) . Folded SFSs for a class of selected sites ( nonsynonymous sites , UTRs or CNEs ) and a putatively neutral reference class were analysed by DFE-alpha [28] , [32] . Our linked neutral reference class of sites , which were four-fold degenerate sites or intronic ARs for protein-coding genes and UTRs and sequences flanking CNEs ( masked for any annotated exons or other CNEs ) for CNEs . DFE-alpha uses maximum likelihood to fit a demographic model to the neutral site data , involving a step change in population size . Assuming the estimated parameters for the demographic model ( Table S1 ) , a gamma distribution of fitness effects ( DFE ) of new deleterious mutations is fitted to the selected site class . We assume that strongly advantageous mutations are sufficiently rare as to make a negligible contribution to polymorphism . In this analysis , it is not possible to infer the frequency of slightly advantageous mutations , since their contribution to the estimated DFE is effectively indistinguishable from that of slightly deleterious mutations [43] . Their contribution to estimates of α are therefore disregarded . From the estimated DFE parameters , the mean fixation probability , un , of a deleterious mutation relative to a neutral mutation is calculated ( Table 1 ) . The proportion of adaptive substitutions is α = ( dn−dsun ) /dn , and the relative rate of adaptive substitution is ωa = ( dn−dsun ) /ds , where dn and ds are nucleotide divergences for the selected and neutral site classes , respectively , corrected for the contribution of polymorphism under the assumption of equal nucleotide diversity in the focal species and the outgroup [52] . Estimates for nonsynonymous sites are generally slightly lower than a previous study based on a small set of genes [2] , but confidence limits overlap when using the same outgroup species ( M . famulus ) ( Table S2 ) . Attempts have previously been made to infer the strength of positive selection by comparing reductions in π/d around selected and putatively neutral sites that have experienced a substitution between related species [35] , [36] . The principle of this approach is that if a large fraction of substitutions at selected sites have been driven to fixation by recent positive selection from new variation , then there will be a signature of reduced diversity around the locations of these substitutions , but this reduction is not expected to be observed ( or to be as strong ) in the regions surrounding neutral substitutions . Quantifying reductions in π/d around nonsynonyomous and synonymous substitutions allows this hypothesis to be tested if synonymous substitutions are assumed to be neutral . Synonymous substitutions also provide a suitable control , since they are interdigitated with nonsynonymous sites . To estimate reductions in π/d in the flanks of nonsynonymous and synonymous substitutions , we firstly identified the locations of all substitutions within coding sequences between M . m . castaneus and M . famulus ( excluding any that were segregating in either species ) . Coding substitutions were identified using the Ensembl genome annotation while only considering canonical spliceforms . From the annotation , we identified substitutions at four-fold degenerate sites ( which were considered synonymous ) and those at zero-fold degenerate sites ( considered nonsynonymous ) . Then , for 1 Kb non-overlapping windows up to 100 Kb away from each substituted site , we obtained estimates of neutral π , divergence from rat ( d ) and π/d ( by excluding any sequence located within an exon or a CNE ) . Estimates of divergence between mouse and rat were obtained using the chained and netted whole genome alignments of mouse and rat ( using rat genome version rn4 ) . We averaged estimates of each statistic over windows at equivalent distances from each zero-fold or four-fold substituted site , and plotted these averages as a function of distance . For comparison , we also calculated π/d around non-substituted sites . In order to investigate the influence of selection acting within exons and CNEs on patterns of nucleotide diversity ( π ) in other parts of the genome , we first divided the genome into non-overlapping windows of 200 bp or 1 Kb . For each window , we excluded any sites belonging to exons ( protein-coding or UTR ) or CNEs and then calculated estimates of π , divergence from the rat ( d ) , and π/d . As above , divergence was obtained using the chained and netted whole genome alignments of mouse and rat . Windows for which π or d could not be calculated due to missing data ( e . g . those lacking divergence data due to deletions in rat , insufficient coverage in M . m . castaneus or because they completely overlapped a CNE or exon ) were excluded from further analysis . In total 72% of windows of length 200 bp and 80% windows of length 1 Kb were retained for further analysis . For each non-overlapping window , we calculated a genetic position on each chromosome ( in cM ) based on a genetic map [53] . Note that for ease of fitting of the models described below , we also scaled these genetic distances by multiplying by a constant factor of 1 , 708 , 728 ( the average number of bp/cM in the genome over chromosomes 1 to 19 , based on the genetic map ) such that their magnitude was comparable to a physical scale in bp . We also obtained positions on the same scale for the start of end of each exon and CNE . To investigate patterns of π/d around exons and CNEs we attempted to model π/d calculated for non-exonic and non-CNE sites within 200 bp or 1 Kb non-overlapping windows throughout the genome ( described above ) as functions of distance to exons and CNEs . To do this we fitted a variety of models described below . In all cases we only considered data for positions within the neighbourhood of exons and CNEs ( positions within 100 Kb of an exon and 10 Kb of a CNE , a selection that includes 64% and 63% of the useable non-overlapping windows described above ) . Firstly , we fitted a number of simple linear models ( using lm in the R statistical package ) where π/d is modelled as a linear function of distance or log-distance to the nearest exon and nearest CNE ( either on a physical or genetic scale ) . We tested whether adding length of the nearest exon and CNE as predictors substantially improved the fit of this model , since it might be expected that the observed reductions in diversity depend on the number of nearby selected sites . Secondly , we investigated a more biologically informative range of models where reductions in diversity decrease exponentially with distance from the nearest exon and CNE , following [36] . These non-linear models were fitted using nls in the R statistical package . We fitted functions of the form π/d∼p1 ( 1−p2 . exp ( -x/p3 ) ) , where x is the ( physical or genetic ) distance from an exon or CNE and p1 , p2 and p3 are estimated parameters . Under this model , p1 can be interpreted as an estimate of the neutral or unreduced value of π/d as x tends to infinity , p2 as the reduction in π/d when x = 0 and p3 as quantifying the distance over which neutral π/d is recovered . We also investigated two more models that allow us to take into account the individual effects of all partially-linked selected exonic and CNE sites on π/d for a given genomic window . In our first model , rather than assuming that diversity around selected elements decreases exponentially , we assume that diversity around all selected sites within selected elements decreases exponentially , and that effects from multiple selected sites on a given neutrally evolving genomic position combine multiplicatively . Therefore , π/d for a single neutral site can be written as a product over all linked selected sites:where p1 is the “neutral” or unreduced value of π/d , p2 is the reduction in diversity observed at a single selected site and p3 measures the rate of decay . Assuming that individual reductions in π/d are small , then this can be approximated as:or for two classes of mutational effects ( i . e . , reductions due to exonic and CNE sites ) :where there n and m are numbers of exonic and CNE sites respectively . We fitted this model to the data by least squares , by minimising the sum of squared deviations between the observed and predicted estimates of π/d for each non-overlapping window . For computational efficiency , we used predictions calculated at the midpoint of each non-overlapping window as a proxy for the average over the window . In our second model we attempted to predict π/d for each window by considering the reduction expected as a result of BGS , a process whereby variation is purged as a result of negative selection on linked loci [41] . We obtained estimates of the reductions in neutral diversity expected for exons and CNEs at the midpoint of each of our genomic windows using the approach and software described in ref . 39 . For this model we assume a constant mutation rate , which we estimated by dividing a Jukes-Cantor-corrected estimate of divergence between mouse and rat at four-fold degenerate sites of 0 . 182 by an estimate of the number of generations separating mouse and rat . We assume that mouse and rat diverged either 12 MYA and have on average two generations per year , giving a separation of 48 generations and a mutation rate estimate of 3 . 79×10−9 . We account for the fact that only a fraction of changes at exonic sites are nonsynonymous by multiplying the mutation rate for exonic regions by the average fraction of nonsynonymous sites per exon ( 0 . 718 ) . We assume a recombination rate of 0 . 528 cM/Mb [42] . Predictions for the reduction in neutral diversity were obtained separately for exons and CNEs and we assume that reductions from each element type combine multiplicatively to give the total reduction for a given genomic position . To compare the predicted reductions to our estimates of π/d we scaled the predicted reductions by multiplying by the unreduced value for π/d which we estimate by minimising the sum of squared differences between the scaled predicted values and the observed ( equivalent to a linear regression of the the observed on the predictions with a zero intercept ) . The sum of squared deviations between the observed and scaled predicted reductions can be used to obtain an r2 for the model . To obtain our best fitting model of BGS , we explored a range of exponential DFEs for both exons and CNEs ( over several orders of magnitude ) and found the parameters of these distributions that maximized the r2 of the model . We also obtained predictions for the reduction in neutral diversity expected from the DFEs inferred for nonsynonymous sites and CNEs from DFE-alpha ( for non-synonymous sites , we use parameters inferred for zero-fold degenerate sites ) . In this case , due to numerical instability it was not possible to obtain predictions from the BGS model using the ML β parameters obtained from DFE-alpha ( i . e . 0 . 11 for zero-fold degenerate sites and 0 . 16 for CNEs ) . To circumvent this issue we obtained estimates for the mean effect of a deleterious mutation using DFE-alpha assuming a fixed β = 0 . 25 ( Nes = 1 . 1×103 for zero-fold sites and Nes = 16 for CNEs ) . Similar results were also obtained when using DFE-alpha estimates where we set β = 1 . 0 , suggesting the predictions from the BGS model are not sensitive to the shape parameter assumed ( data not shown ) . All animal work followed the legal requirements , was registered under number K-13/14-05 ( Evolutionary Genomics ) , and was approved by the animal ethics commission of the University of Cologne ( Germany ) . | We present an analysis of the genome sequences of multiple wild house mice . Wild house mice are about ten times more genetically diverse than humans , reflecting the large effective population size of the species . This manifests itself as more effective natural selection acting against deleterious mutations and favouring advantageous mutations in mice than in humans . We show that there are strong signals of adaptive evolution at many sites in the genome . We estimate that 80% of adaptive changes in the genome are in gene regulatory elements and only 20% are in protein-coding genes . We find that nucleotide diversity is markedly reduced close to gene regulatory elements and protein-coding gene sequences . The reductions around regulatory elements can be explained by selection purging deleterious mutations that occur in the elements themselves , but this process only partially explains the diversity reductions around protein-coding genes . Recurrent adaptive evolution , which can also cause local reductions in diversity via selective sweeps , may be necessary to fully explain the patterns in diversity that we observe surrounding genes . Although most adaptive molecular evolution appears to be regulatory , adaptive phenotypic change may principally be driven by structural change in proteins . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2013 | Contributions of Protein-Coding and Regulatory Change to Adaptive Molecular Evolution in Murid Rodents |
Norepinephrine , a neuromodulator that activates β-adrenergic receptors ( βARs ) , facilitates learning and memory as well as the induction of synaptic plasticity in the hippocampus . Several forms of long-term potentiation ( LTP ) at the Schaffer collateral CA1 synapse require stimulation of both βARs and N-methyl-D-aspartate receptors ( NMDARs ) . To understand the mechanisms mediating the interactions between βAR and NMDAR signaling pathways , we combined FRET imaging of cAMP in hippocampal neuron cultures with spatial mechanistic modeling of signaling pathways in the CA1 pyramidal neuron . Previous work implied that cAMP is synergistically produced in the presence of the βAR agonist isoproterenol and intracellular calcium . In contrast , we show that when application of isoproterenol precedes application of NMDA by several minutes , as is typical of βAR-facilitated LTP experiments , the average amplitude of the cAMP response to NMDA is attenuated compared with the response to NMDA alone . Models simulations suggest that , although the negative feedback loop formed by cAMP , cAMP-dependent protein kinase ( PKA ) , and type 4 phosphodiesterase may be involved in attenuating the cAMP response to NMDA , it is insufficient to explain the range of experimental observations . Instead , attenuation of the cAMP response requires mechanisms upstream of adenylyl cyclase . Our model demonstrates that Gs-to-Gi switching due to PKA phosphorylation of βARs as well as Gi inhibition of type 1 adenylyl cyclase may underlie the experimental observations . This suggests that signaling by β-adrenergic receptors depends on temporal pattern of stimulation , and that switching may represent a novel mechanism for recruiting kinases involved in synaptic plasticity and memory .
Long-term potentiation ( LTP ) in the hippocampus has long been studied as a mechanism underlying mammalian learning and memory . At least two mechanistically distinct phases of LTP have been characterized , including an early-phase LTP , which decreases over the course of two hours , and a late-phase ( L-LTP ) , which endures for more than two hours . Though 1 s of 100 Hz electrical stimulation of Schaffer collaterals produces only early-phase LTP in area CA1 , pretreatment of β-adrenergic receptors ( βARs ) with the selective agonist isoproterenol followed by 1s of 100 Hz electrical stimulation produces L-LTP [1–5] . Similarly , 3 min of 5 Hz stimulation fails to induce LTP by itself , but in the presence of isoproterenol , the same stimulation induces robust L-LTP [2–6] , often called β-LTP . Because L-LTP resembles memory storage in its requirement for protein synthesis , understanding the role of βARs in producing L-LTP may shed light on molecular mechanisms of memory storage . Several studies have identified molecular components involved in β-LTP [2–5] . Both βARs and the requisite calcium influx through N-methyl-D-aspartate receptors ( NMDARs ) couple to the cAMP signaling pathway , though via different intermediaries ( Fig 1A ) . The calcium ( bound to calmodulin ) stimulates adenylyl cyclase types 1 and 8 ( AC1 and AC8 ) which catalyze cAMP production [7 , 8] . Stimulation of βARs activates the Gs subtype of GTP binding protein , which stimulates adenylyl cyclase isoforms [9] . AC1 and AC8 , abundantly expressed in CA1 pyramidal neurons [10–13] , are required for L-LTP induction and long-term memory [14] . The synergistic activation of AC1 by simultaneous Ca2+ and Gs signals in AC1-expressing HEK293 cells [15] as well as synergistic cAMP-mediated transcription in cultured hippocampal neurons [16] suggests that NMDA and isoproterenol would enhance cAMP production during β-LTP , but this has not been demonstrated . The regulation of cAMP downstream of adenylyl cyclases is largely carried out by phosphodiesterases ( PDEs ) , which are regulated by cAMP-dependent kinase ( PKA ) . Type 4 PDEs ( PDE4 ) comprise the major cAMP-degrading PDE family in the hippocampus [17] . PKA phosphorylation of PDE4s increases their activity [18 , 19] forming a cAMP-PKA-PDE4 negative feedback loop , which is a significant contributor to cAMP signaling dynamics downstream of βARs [20 , 21] . To investigate how NMDARs and βARs contribute to the cAMP and PKA underlying β-LTP , we combined FRET-based live-cell imaging of cAMP in cultured rat hippocampal neurons with a spatial mechanistic model of the cAMP signaling network in a hippocampal CA1 pyramidal neuron . Unexpectedly , when NMDA was applied after the onset of isoproterenol in experiments , rather than generate synergistic elevations of cAMP , the cAMP was attenuated compared to that of NMDA alone . This attenuation of NMDA-induced cAMP following isoproterenol was not sufficiently explained by either PKA or PDE4 in the model . Instead , our results suggest that PKA-mediated Gs-Gi switching following βAR activation may underlie the attenuation of NMDA-induced cAMP following isoproterenol pretreatment .
Primary hippocampal cell cultures were prepared from brains of E18 Sprague Dawley rats as previously described [22] . Briefly , surgically dissected hippocampi were enzymatically and mechanically dissociated and the resultant cell suspensions were plated on coverslips coated with poly-L-lysine ( Sigma ) and maintained in Neurobasal medium ( Invitrogen ) supplemented with B27 ( Invitrogen ) . The medium was partially changed once a week . At 5–9 days in vitro ( the day before the experiments ) neurons were transiently transfected with the Epac1 based FRET sensor for cAMP [23] using Transfectin ( Biorad ) transfection reagent . The experiments were performed on an inverted Olympus IX 70 microscope using a 60xNA , 1 . 4 oil-immersion objective . The microscope was equipped with a CCD camera ( Sensicam QI , PCO , U . S . A . ) , a software-controlled monochromator ( Polychrome IV , TILL Photonics , Germany ) , and an optical beam-splitter device ( Multispec Microimager; Optical Insights , U . S . A . ) . All filters and dichroics were from Chroma Technology . Live images were acquired for 200–300 ms at 3 s intervals . The day of the experiment , coverslips were mounted in an imaging chamber at room temperature and maintained in a modified Hank’s balanced salt solution ( HBSS ) as follows: 137 mM sodium gluconate , 5 mM potassium gluconate , 0 . 6 mM Na2HPO4 , 0 . 6 mM KH2PO4 , 5 . 5 mM glucose , 20 mM HEPES , 1 . 4 mM calcium gluconate pH 7 . 4 ( gluconate was used to replace chloride to avoid the unequal quenching of CFP and YFP due to chloride ion entry during NMDA stimulation ) . Images were acquired using TILLvisION v3 . 3 software and then processed off-line using ImageJ . Cells received either the NMDA alone stimulation , or the NMDA after ISO stimulation , both for control experiments , and in the presence of either H89 or rolipram . When isoproterenol was pre-applied , the NMDA was then applied between 2 and 5 minutes later , after the response to isoproterenol reached a plateau . FRET changes were measured as changes in the background-subtracted 480/545 nm fluorescence emission intensities on excitation at 430 nm and expressed as R/R0 , where R is the ratio at time t and R0 is the ratio at time = 0 s . The amplitude of response was calculated as ΔR/R0 , where ΔR = R–R0 and expressed in bar graphs as % FRET ratio change ( %ΔR/R0 ) . All data are presented as means and SEM . Student’s t tests ( two-tailed ) were performed using SAS ( SAS Institute ) to evaluate statistical significance ( P ≤ 0 . 05 ) . When variances were unequal , the Satterthwaite method for variances of the samples was used . Pharmacological stimuli , N-methyl-D-aspartic acid ( NMDA , 300 μM ) , Glycine ( 3 μM ) , 3-Isobutyl-1-methylxanthine ( IBMX , 100 μM ) , isoproterenol ( ISO , 1 μM ) , rolipram ( 1 μM ) , dopamine ( 20 μM ) , H-89 dihydrochloride hydrate ( H89 , 10 μM ) , all from Sigma , were prepared in stocks and diluted to the final concentration ( indicated in brackets ) in the bath . We created a spatial , mechanistic model of the NMDAR and βAR activated signaling pathways in hippocampal CA1 pyramidal neurons by modifying an existing model of the signaling pathways underlying L-LTP ( Tables 1–3 ) . The morphology of the model represents the region of interest of the cultured hippocampal neurons used for imaging ( Fig 2A ) . Thus , we modeled one neurite and half of the soma ( for computational efficiency , Fig 1B ) with diameters based on the morphology of the cultured neurons . These values are similar to that reported for the apical dendrite and soma of reconstructed neurons in NeuroMorpho . org . The morphology is discretized with 0 . 9 μm voxels for the cytosol , with one layer of 0 . 3 μm submembrane voxels and one layer of 0 . 6 μm voxels adjacent to the submembrane voxels . A subset of the molecules in the system is diffusible ( Table 3 ) . Membrane associated proteins , such as βAR , G proteins , AC1 , AC8 and PKA holoenzyme are organized as multi-protein complexes by A-Kinase-Anchoring-Proteins ( AKAPs ) [24 , 25] . The AKAPs are made implicit in this model by colocalizing at the membrane the molecules comprising these multi-protein complexes . In addition , for most simulations the PDE4s were considered anchored [26] , and thus do not diffuse . One set of simulations ( dynamic recruitment of PDE4s ) required additional biochemical reactions ( Table 4 ) . Dynamic recruitment of PDE4 to cell membranes occurs in an activity-dependent manner [61] , but the specific mechanisms by which PDE4 is recruited to the membrane are unknown , though may involve β-arrestin . In the model , we assume that a fraction of the PDE4 ( called PDE4D ) is anchored to a non-diffusible cytosolic anchoring protein ( APcyt ) but is released by a cooperative , cAMP-dependent mechanism . The released PDE4D diffuses and binds to an anchoring protein localized to the submembrane ( APsm ) , thus completing the recruitment . In the dynamic recruitment simulations , one third of the total PDE4 was the PDE4D form . To demonstrate that the results are not dependent on the cAMP-dependent release mechanism , we ran additional simulations using elevation of Gsβγ instead of cAMP as the trigger . Though the Gsβγ trigger prevented PDE4D recruitment in response to NMDA , neither of these dynamic recruitment mechanisms could account for the reduction in the NMDA response after ISO pretreatment . For the final set of simulations , the biochemical reactions of the signaling pathways were modified by adding receptor desensitization and Gs-Gi switching ( Fig 1C ) , [62 , 63] . Table 5 provides the rate constants governing PKA phosphorylation of βARs , followed by activation of the Gi subtype of GTP binding protein . A single phosphorylation event decouples the βAR from Gs , but only the fully phosphorylated βAR can bind Gi . We assume βARs are phosphorylated with cooperativity and in a distributive manner , i . e . with PKA dissociating from the receptor after each phosphorylation event , which together enable an ultrasensitive response [64–66] . For most simulations , βARs require PKA phosphorylation at 4 sites [67] for Gi binding; however , for a subset of simulations , only 2 sites were phosphorylated to produce switching . The rates of Gi activation and hydrolysis were adjusted to produce a low basal quantity of GiαGTP . The reactions and kinetics for binding of GiαGTP to AC1 were derived from [68] . Bath application of NMDA alone was simulated by injecting calcium , at t = 50s at a rate of 5 molecules per ms for 500 s , to create an intracellular Ca2+ concentration of ~1 . 4 μM [60] . Bath application of isoproterenol was simulated by injecting isoproterenol at t = 50s , at a rate of 2 . 15 molecules per ms for 1 s , to create an isoproterenol concentration of 1 . 0 μM . For a subset of simulations , to create higher or lower concentrations of isoproterenol , a higher or lower injection rate was used . Bath application of NMDA after ISO applied the isoproterenol stimulus at 50s , followed by the NMDA calcium stimulus at 170s . The signaling pathways are simulated using a well-validated , efficient , mesoscopic stochastic reaction-diffusion algorithm , NeuroRD [71] , version 2 . 1 . 9 . Thus , the noise and fluctuations in the simulations are caused by the stochastic simulation technique . Because of this stochastic variability , some simulations are repeated to generate means and SEM . All simulations use a time step of 2 . 5 μs . Simulation output is processed using NRDPost ( to calculate average concentration for defined regions in the morphology ) . The simulation and output processing software and the files used for the model simulations ( S1 File ) are freely available from modelDB ( http://senselab . med . yale . edu/modeldb/showmodel . cshtml ? model=184731 ) . The responses in bar graphs for both model and experiments were calculated as follows . The response to NMDA alone was measured as the peak response to NMDA application alone . The response to ISO was measured as the peak response to ISO detected prior to NMDA application . The NMDA after ISO response was the peak response to NMDA , detected after both NMDA and ISO application , minus the ISO response during the 10 sec immediately prior to NMDA application . For all cases , the peak response was the mean value measured during a 10 sec window surrounding the peak . A synergistic effect implies that the peak response to NMDA after ISO is larger than the sum of the ISO response plus NMDA alone response . Equivalently , synergy is suggested if the mean NMDA after ISO response plotted in the graphs is larger than the response to NMDA alone . Peak responses for experiments are tabulated in S1 Data .
In CA1 pyramidal neurons , pretreatment of βARs facilitates several NMDAR-dependent forms of LTP [1–5 , 72] . The mechanism underlying the facilitation is suggested by previous research demonstrating that the catalytic activity of AC1 increases synergistically when GsαGTP and Ca2+/calmodulin signals coincide [15 , 16] , but a synergistic increase in cAMP has not been demonstrated in hippocampal neurons . To investigate the effects of βAR and NMDAR interactions underlying βAR-dependent L-LTP , we performed live-cell imaging of cAMP in cultured hippocampal neurons expressing the FRET sensor Epac1-camps ( Fig 2A ) . To approximate βAR activation followed by electrical stimulation , we bath applied isoproterenol and added NMDA when the isoproterenol-induced FRET change reached a plateau ( henceforth called the NMDA after ISO stimulus ) . FRET imaging of cAMP did not reveal a synergistic increase in cAMP in response to NMDAR and βAR stimulation . Isoproterenol by itself induced relatively weak cAMP responses that were similar in amplitude in the neurites and soma ( n = 10 , P = 0 . 648; Fig 2D ) . NMDA alone induced relatively robust cAMP responses , with average responses in neurites significantly higher than those in the soma ( n = 46 , P<0 . 0001; Fig 2C and 2D ) . However , when the NMDA was applied after the ISO stimulus , a synergistic response was not observed . In some neurons , isoproterenol pretreatment led to an NMDA-induced cAMP response similar to that of NMDA alone ( Fig 2B1 ) ; in other neurons , isoproterenol pretreatment attenuated the NMDA-induced cAMP to below that of NMDA alone ( Fig 2B2 ) . Note that in all cases the NMDA after ISO response is measured as the difference between the response to isoproterenol pre-treatment and the response to the combined ISO+NMDA application . Thus , if the two responses were additive , the NMDA after ISO response would be the same as in response to NMDA alone . A synergistic effect would produce an NMDA after ISO response larger than the response to NMDA alone . Statistical analysis revealed that the average NMDA-induced cAMP response of the NMDA after ISO stimulus was significantly attenuated relative to that of NMDA alone in the neurites but not in the soma ( NMDA alone , n = 46; NMDA after ISO , n = 10; neurites: P = 0 . 03; soma: P = 0 . 337; Fig 2D ) . In addition , we observed an inverse relationship between isoproterenol- and NMDA-induced cAMP responses in the soma , such that as the cAMP response to isoproterenol increased , that of subsequently applied NMDA decreased ( Fig 2E ) . Since we did not observe synergistic cAMP production in these neurons , we hypothesized that additional negative feedback mechanisms were operating downstream of βARs in these neurons to limit the subsequent NMDA-induced cAMP . Because PDE4s are the predominant negative feedback regulators of cAMP signaling in hippocampal neurons [17] , we investigated their role in the isoproterenol-mediated attenuation of the NMDA response . We bath applied a subsaturating concentration ( 1 μM ) of the specific PDE4 inhibitor rolipram prior to NMDA alone , and the NMDA after ISO stimulus . Rolipram did not increase the average cAMP response to isoproterenol or to NMDA alone in either neurites or soma ( Fig 3A ) . Nonetheless , rolipram prevented the decrease in cAMP response caused by isoproterenol pretreatment in the neurites ( rolipram + NMDA , n = 19; rolipram + NMDA after ISO stimulus , n = 13; neurites: P = 0 . 341; Fig 3A1 ) . In other words , in the presence of rolipram , the peak neurite response to NMDA after ISO ( %ΔR/R0 = 26 . 7 ) approximately equaled the sum of the isoproterenol response ( %ΔR/R0 = 7 . 2 ) plus the NMDA response ( %ΔR/R0 = 23 . 8 ) . These data suggest that PDE4s may be involved in reducing the NMDA-induced cAMP response following isoproterenol pretreatment . PKA phosphorylation of PDE4 can enhance its hydrolytic activity ~2-fold [18 , 19] , acting as a negative feedback regulator of cAMP [20 , 21] . If this mechanism is operating in hippocampal neurons , then inhibiting PKA should prevent the attenuation of the NMDA response by prior isoproterenol application . Application of the specific PKA inhibitor H-89 ( 10 μM ) prevented the reduction in the NMDA response caused by prior application of isoproterenol in the neurites ( NMDA in H89 , n = 11; NMDA after ISO stimulus in H-89 , n = 8; p = 0 . 138; Fig 3B1 ) , and allowed the ISO pretreatment to enhance the soma response to NMDA ( P = 0 . 0095; Fig 3B2 ) . Note that inhibition of PKA with H-89 did not alter cAMP responses to isoproterenol alone in either neurites or soma ( ISO , n = 10; H-89 + ISO , n = 8; neurites: P = 0 . 603; soma: P = 0 . 315; Fig 3B ) but robustly decreased NMDA-induced cAMP responses in both neurites and soma ( NMDA , n = 46; H-89 + NMDA , n = 11; neurites: P < 0 . 001; soma: P < 0 . 001; Fig 3B ) . The latter is consistent with the known function of PKA phosphorylation in increasing the fractional Ca2+ influx through NMDARs in CA1 pyramidal neurons [73 , 74] . Nonetheless , in the neurite in the presence of H-89 , the peak response to NMDA after ISO equaled the sum of the isoproterenol response plus the NMDA response , as observed with rolipram . Therefore , the experiments suggest that both PDE4s and PKA may be involved in the attenuation of NMDA-induced cAMP following isoproterenol pretreatment . To better understand how isoproterenol-induced enhancement in PDE4 activity might lead to an attenuation of the cAMP response to NMDA , we adapted a previously validated , spatial mechanistic model of signaling pathways in CA1 pyramidal neurons [75] and evaluated whether downstream mechanisms alone , i . e . , the cAMP-PKA-pPDE4 negative feedback loop ( Fig 1A ) , can indeed account for the experimental observations . We ran the same stimulation combinations of NMDA with and without isoproterenol pretreatment . Using this initial model we verified in control simulations that the cAMP responses to isoproterenol alone ( Fig 4A—initial part of NMDA after ISO trace ) and NMDA alone ( Fig 4A ) agreed with the experiments . Indeed , both the dynamics and the existence of a soma to neurite gradient were similar to experiments . The fluctuations in the model traces are due to the stochastic nature of the molecule interactions in the small submembrane region . The standard deviation of these signals ranges from 0 . 4 to 0 . 8%ΔR/R0 . The darker , less noisy , superimposed trace shows the concentration in the cytosolic compartments . Because the mean values are similar for cytosolic and submembrane traces , only the less noisy cytosolic traces are shown in the remainder of the graphs . In addition to similarity in cAMP responses , the time course of pPDE4 activity ( Fig 4B ) was consistent with that shown by others [19] . Note that the lag in phosphorylation of PDE4 in the soma is due to the lower surface to volume ratio in the soma . The adenylyl cyclase is in the membrane , whereas the PDE4s are throughout the morphology; this higher adenylyl cyclase to PDE4 ratio in the neurite causes a higher neurite cAMP and PKA activity , and faster phosphorylation of the PDE4s in the neurite . We tested the hypothesis that the cAMP-PKA-PDE4 negative feedback loop was involved in the attenuation by evaluating the NMDA after ISO stimulus in the model . However , the NMDA-induced cAMP response following isoproterenol was synergistic , not attenuated , relative to the cAMP response to NMDA alone ( Fig 4A ) . In other words , the difference between the peak NMDA after ISO response and the isoproterenol response ( %ΔR/R0 = 34 . 9 ) was greater than the NMDA alone response ( %ΔR/R0 = 9 . 8 ) ; also , the peak NMDA after ISO response ( %ΔR/R0 = 40 . 9 ) was greater than the sum of the NMDA alone and ISO alone responses ( %ΔR/R0 = 16 . 0 ) . The cause of the synergistic response to NMDA after ISO was the greatly increased cAMP production by the GsαGTP bound adenylyl cyclase ( Fig 4C ) , which is not sufficiently compensated by the increase in pPDE4 , because PKA and pPDE4 also increase in response to NMDA alone ( Fig 4B and 4D ) . We further evaluated whether the cAMP-PKA-PDE4 negative feedback loop could explain the experimentally observed attenuation of the NMDA response after ISO by assessing two other mechanisms that could enhance PDE4 activity in an activity-dependent manner . First , to see if an enhanced activity of pPDE4 could underlie the attenuation , we simulated the effect of a several-fold increase in pPDE4 activity , as may occur due to SUMOylation [76] . Increasing the activity of pPDE4 did not eliminate the enhanced response to NMDA after ISO application , because enhanced pPDE4 activity also decreased the NMDA alone response ( Fig 4E1 ) , even when combined with an increased rate of GsαGTP hydrolysis ( Fig 4E1 ) . Combining enhanced pPDE4 activity with an increase in the rate at which PKA phosphorylates PDE4 reduced , but did not eliminate , the enhanced response to NMDA after ISO application ( Fig 4E2 ) . Since previous work has shown that PDE4s are anchored [26] , and that different PDE4 isoforms distribute differentially in cells [77] , we repeated these simulations with four times the concentration of submembrane PDE4 compared to cytosolic PDE4; however , the cAMP response to the NMDA after ISO stimulus remained larger than that to NMDA alone ( Fig 4E2 ) , indicating that the mechanisms integrated in this initial model were not sufficient to reflect the responsible pathways . A second mechanism of enhancing apparent PDE4 activity is via dynamic recruitment of the PDE4D5 subtype to the plasma membrane [26 , 78 , 79] . Bringing PDE4D5 in close proximity to adenylyl cyclases following βAR activation can increase the specificity and efficiency of PDE4D5 activity and could thus strongly oppose subsequent cAMP generation at the plasma membrane . To test if this could explain the attenuation , we implemented dynamic recruitment of PDE4s in the model ( see Methods; Table 4 ) such that a fraction of the PDE4 ( called PDE4D in our model ) was recruited to the submembrane region following isoproterenol pretreatment . The activity of the membrane-bound PDE4D was either the same as the cytosolic PDE4D , or increased , as may occur when PDE4D binds anchoring proteins at the submembrane [80] . Fig 5B shows an increase in PDE4D in the submembrane with stimulation , demonstrating that the simulated dynamic recruitment is indeed successful . However , the NMDA-induced cAMP response following isoproterenol was still much greater than that of NMDA alone , even with enhanced activity of the membrane-bound PDE4D ( Fig 5A ) . As observed with the basic model , enhanced plasma membrane PDE4D activity does not reproduce the experimental results because the enhanced plasma membrane PDE4D reduces the NMDA alone response , though much less than observed with the basic model ( Fig 5A and 5D ) . Because dynamic recruitment of PDE4D with enhanced membrane-bound PDE4D activity seemed promising , we performed additional simulations of this model variant with faster or slower rates at which PKA phosphorylates PDE4 and different rates at which PDE4D diffuses from the cytosol to the membrane . Increasing the rate at which PKA phosphorylated PDE4 indeed reduced the NMDA response after ISO , though not enough to account for the experimental observations , whether with or without enhanced activity of plasma membrane PDE4 ( Fig 5C ) . Slowing the diffusion constant of PDE4D ( Fig 5D ) had only a small effect , because of the small diameter of the dendrite . Similar to the cAMP-dependent recruitment of PDE4D , Gβγ-dependent recruitment of PDE4D ( Fig 5D ) was unable to lower the cAMP response to NMDA to a value lower than the cAMP response to NMDA after ISO . Thus , we conclude that the negative feedback loop of PKA phosphorylation of PDE4 , while effective , is insufficient to fully explain the suppression of synergistic cAMP generation induced by NMDA after ISO in hippocampal neurons . As an alternative to mechanisms acting downstream of adenylyl cyclases in the cAMP signaling network , we considered mechanisms operating upstream of adenylyl cyclases to regulate cAMP . One possibility is regulation of βARs , which are known to undergo two modes of desensitization , one mediated by PKA and another by G protein-coupled receptor kinases ( GRK ) [81] . In particular , PKA phosphorylation of βARs can lead to a “switch” in βAR coupling from the Gs to Gi subtype of GTP binding proteins [62 , 82] . Activated Gi proteins can then release GiαGTP and Giβγ subunits that can directly inhibit the catalytic activity of AC1 [70 , 83] , thus reducing cAMP production . However , the effects of PKA-mediated desensitization of βARs on NMDA-induced cAMP have yet to be investigated . To evaluate if PKA-mediated desensitization of βARs could explain the attenuation of NMDA-induced cAMP following isoproterenol pretreatment , we added PKA-dependent Gs-Gi switching and GiαGTP inhibition of AC1 to the model ( see Materials and Methods; Fig 1C; Table 5 ) . We implemented PKA phosphorylation of βARs on four serine residues , as these have been identified as PKA phosphorylation sites in vitro [67] . Fully phosphorylated βARs bind the Gi subtype of G protein instead of the Gs subtype , producing GiαGTP , which can then bind to and inhibit AC1 . Simulations demonstrate that switching can explain the reduction in NMDA induced cAMP following isoproterenol application . In control simulations , we observed that there was little difference in isoproterenol- or NMDA-induced cAMP ( Fig 6A ) with the addition of Gs-Gi switching and GiαGTP inhibition of AC1; however , when we simulated the NMDA after ISO stimulus , Gi robustly inhibited the NMDA-induced cAMP increase ( Fig 6A and 6B ) . After isoproterenol pretreatment , the elevation in cAMP produced by NMDA stimulation was similar to that observed experimentally and smaller than to NMDA alone . This response was robust to parameter variations , as similar results were obtained with a model where PKA phosphorylation of βARs could occur only on two sites ( Fig 6B ) . In addition , the attenuation of the NMDA response after ISO was observed for a range of affinities of Gi for the phosphorylated receptor ( Fig 6C ) . Gs-Gi switching differs qualitatively from enhanced PDE4 activity in that switching only occurs consequent to the ISO application and does not affect the NMDA alone response ( Fig 6C ) . Simulations of a 4 min delay between NMDA and ISO application ( Fig 6D ) produces too strong a decay of the ISO response using the default parameters; however a lower Gi binding rate , e . g . 0 . 2x , yields a much smaller decay of the ISO response while still attenuating the subsequent NMDA response . The attenuation of the NMDA response after ISO also was observed for a range of isoproterenol concentrations ( Fig 6E1 ) , though the NMDA response after ISO increased with lower concentrations of isoproterenol . The amount of GiαGTP bound to AC1 in response to different isoproterenol concentrations reveals the mechanism underlying this observation . The time course and strength of inhibition of AC1 by is proportional to the concentration of isoproterenol ( Fig 6E2 ) , and a fast increase in GiαGTP is needed to inhibits the subsequent peak cAMP response to NMDA . To further explore the mechanisms involved in producing the observed results we performed several additional simulations . To evaluate the specific roles of PKA in the model , we either blocked PKA phosphorylation of βAR or blocked all PKA activity . Either blocking PKA phosphorylation of βARs or inhibiting total PKA activity ( e . g . with H89 ) prevented the attenuation of the NMDA-induced cAMP following isoproterenol ( Fig 7A ) . This latter result , that inhibiting total PKA activity blocks the attenuation of the cAMP response caused by isoproterenol pretreatment , is similar to experiments using the PKA inhibitor H89 . The difference between inhibiting total PKA and preventing PKA phosphorylation of βARs shows the contribution of pPDE4 ( with the 2x increase in activity ) to reducing the cAMP response to NMDA after isoproterenol . Two components of the PKA-mediated desensitization of βARs could be producing the observed attenuation of the NMDA response after ISO . First , decoupling of βAR from Gs may remove the synergistic activation of AC1 by reducing the production of GsαGTP , independent of the inhibition of AC1 by GiαGTP . Second , the direct inhibition of AC1 by GiαGTP may reduce AC1 activity to a level below that of NMDA alone . Simulations in a model in which GiαGTP did not bind to the phosphorylated βAR ( Fig 6C , Gi bind rate = 0 . 0x ) allow the NMDA response after ISO to exceed the NMDA alone response , demonstrating that decoupling the receptor from Gs to Gi by itself is not sufficient . An alternative method for evaluating the role of Gs decoupling is to replace PKA phosphorylation of βAR with GRK mediated desensitization , which is responsible for the transience of the isoproterenol-induced cAMP response in HEK293 cells [84 , 85] . GRK-mediated desensitization of β2ARs leads to the recruitment of β-arrestin to the receptor , which is required for receptor desensitization via internalization , recycling , and degradation [86] . To see if GRK-mediated desensitization of βARs could explain the attenuation of the cAMP response to NMDA after ISO , we implemented GRK-mediated desensitization of βARs in the model in the absence of Gs-Gi switching ( Table 6 ) , together with dynamic recruitment of PDE4D to the membrane ( and 10x enhanced activity of plasma membrane PDE4D ) . The GRK-mediated desensitization produced a response to NMDA after ISO ( Fig 7A ) that was larger than the response to NMDA alone , and was unable to reproduce the experimentally observed ( Fig 2D ) attenuation of the NMDA response after isoproterenol pretreatment . In summary , desensitization of the βAR decreased the synergistic activation of AC1 , but did not reduce the cAMP response to NMDA after ISO to a level 25% less than that of NMDA alone . Thus , in addition to receptor decoupling from Gs , inhibition of AC1 by GiαGTP is required . Though the switching model agrees with experiments regarding a role of protein kinase A , the model is unable to reproduce the experimental observation that rolipram prevents the reduction in the NMDA response after isoproterenol pretreatment . Rolipram in the model was implemented as inhibition of a fraction of the PDE4 , both because the affinity of rolipram for PDE4 depends on the isoform and to reproduce the experimental observation that rolipram causes only a small increase in cAMP basal level . Simulations show that rolipram slightly enhances the cAMP response to either isoproterenol , or to NMDA alone ( Fig 7B ) , similar to experiments . The consequence of enhanced cAMP in response to ISO is a slightly enhanced PKA phosphorylation of the remaining PDE4 , leading to a similar or slightly reduced NMDA response , which is opposite of that shown by experiments . The same result occurs whether rolipram inhibits 5% or 10% of the PDE4 . Nonetheless , we cannot rule out that inhibition of a particular nanodomain of PDE4 in the model would be able to produce the experimental observations . We propose that PKA-mediated Gs-to-Gi switching of βARs and GiαGTP inhibition of AC1 might underlie the reduction in the NMDA-induced cAMP response following isoproterenol pretreatment . This result implies that attenuation of the NMDA response will be blocked by the Gi inhibitor pertussis toxin [62] and that the attenuation of the NMDA response will not be observed subsequent to stimulation of Gs coupled receptors that do not exhibit switching . This latter model prediction is consistent with the results of additional experiments in which we used dopamine instead of isoproterenol to stimulate Gs coupled dopamine D1/D5 receptors in hippocampal cultures . In these experiments , there was no evidence that dopamine attenuated the subsequent cAMP response to NMDA ( Fig 7C ) . Additional simulations were performed to explore the implications of the model and make additional , experimentally-testable predictions . Because switching occurs only in response to isoproterenol application , and does not occur in response to NMDA application , the order and timing of agonist application will influence the cAMP response . Thus , we performed simulations with NMDA applied either prior to or simultaneously with isoproterenol application . In addition , we applied pairs of transient stimulation pulses of the same agonist . Fig 8A1 shows that application of isoproterenol simultaneous with ( or after ) the NMDA application produces a synergistic cAMP response . The peak response of 47 . 7% ΔR/R0 was considerably greater than the sum of the NMDA alone ( %ΔR/R0 = 15 . 4 ) and isoproterenol alone ( %ΔR/R0 = 6 . 6 ) responses . Even the model with enhanced PDE4 ( the model with dynamic recruitment and 10x activity of plasma membrane PDE4D from Fig 5A1 ) exhibits a larger response when NMDA is applied simultaneous with or 15–30s prior to isoproterenol ( Fig 8A2 ) . This synergy is caused by enhanced activity of AC1 when bound to both Gs and calcium-calmodulin . Because both models produce similar peak responses under these conditions , an experiment with simultaneous application of the two agonists will neither support nor refute the switching model , and instead will test a critical underlying assumption of the model: that a single pool of AC1 responds to both NMDA and isoproterenol . Simulations with a delay between NMDA and ISO application do reveal a difference: a 60s delay yields a reduced response to ISO in the enhanced PDE4 model , but not the switching model . This reduction is due to prior NMDA producing enhanced PDE4 through cAMP and PKA activity . The response to paired pulses is another experiment that can validate the switching model . Fig 8B and 8C shows the response to the paired pulse protocol , for both the switching model ( Fig 8B ) and the enhanced PDE4 model ( Fig 8C ) . Both models give a similar response to paired isoproterenol pulses ( Fig 8B2 and 8C2 ) : responses to subsequent pulses of isoproterenol are reduced with larger time intervals . The two models differ in their response to NMDA ( Fig 8B1 and 8C1 ) : the enhanced PDE4 model exhibits a significant reduction in the response to the second NMDA pulse as the time between pulses is increased , because both ISO and NMDA produce the cAMP that leads PDE4D recruitment to the membrane . The reduction does not occur in the switching model because NMDA does not produce receptor decoupling . In summary , the response to paired isoproterenol pulses will be different than the response to paired NMDA pulses in the switching model , but not in the enhanced PDE4 model , because both NMDA and isoproterenol activate PKA and enhance PDE4D equally .
In this work , we used a combination of FRET imaging of cAMP dynamics and spatial mechanistic modeling of cAMP signaling pathways to investigate the contribution of βAR signaling pathways to cAMP dynamics . We demonstrated that isoproterenol pretreatment of cultured hippocampal neurons leads to a reduced cAMP response to NMDA application . This result was unanticipated because AC1 is synergistically activated by Ca2+ and isoproterenol when applied simultaneously , as measured by cAMP in HEK293 cells [16] , or cAMP-mediated transcription in cultured hippocampal neurons [15] . We showed that mechanisms both upstream and downstream of adenylyl cyclase oppose the synergistic activation of AC1 and contribute to the observed reduction in NMDA-induced cAMP caused by prior isoproterenol application . Downstream of adenylyl cyclase , the cAMP-PKA-PDE4 negative feedback loop contributes modestly to the attenuation of NMDA-induced cAMP after isoproterenol . Upstream of adenylyl cyclase , PKA phosphorylation of βARs followed by Gs-Gi switching and GiαGTP inhibition of AC1 is required to overcome the enhanced cAMP production by adenylyl cyclase stimulated by GsαGTP and Ca2+/calmodulin . These mechanisms are qualitatively different in that the downstream , PDE4 feedback loop is activated by both NMDA and isoproterenol , whereas the upstream βAR feedback loop is activated only by isoproterenol . Therefore , the upstream feedback loop suppresses the NMDA response after isoproterenol , but not the NMDA alone response . While in the model we have assumed GiαGTP is the G-protein subunit responsible for blocking AC1 , the results do not preclude Giβγ inhibition of AC1 . Indeed , the latter has been suggested to be a more potent inhibitor than the former [70 , 83] . Regardless of which Gi subunit is involved , both upstream and downstream mechanisms are critically dependent on PKA activity . Both experiments and simulations revealed a gradient of cAMP from the neurites to the soma . The gradient was observed after NMDA alone , but was reduced or absent in response to NMDA after ISO application . These observed gradients are consistent with those previously reported [88] , despite being measured on a smaller spatial scale . In that study , gradients were measured over a spatial scale of 100 μm for both simulations and experiments , whereas our gradients appear across a 20 μm long structure . The source of the gradients in both studies is the larger surface to volume ratio of the neurites ( dendrites ) as compared to the soma: the membrane location of adenylyl cyclase versus PDE4s located in the entire volume produces a greater ratio of production to degradation for neurites ( dendrites ) as compared to the soma . The spatial aspect of the model also contributed to the delay in dynamic recruitment of PDE4D from the cytosol to the submembrane region . Though dynamic recruitment could not completely reproduce the experimental results , it did indeed produce a small reduction in the NMDA response after isoproterenol application . Amongst the preponderance of PKA substrates in hippocampal neurons , our results suggest that PKA phosphorylation of βARs is crucial for mediating the attenuation of NMDA-induced cAMP by isoproterenol pretreatment . PKA phosphorylates a number of different targets in CA1 pyramidal neurons which are implicated in plasticity [89]; however , the phosphorylation of the majority of these PKA targets leads to activity that would presumably promote rather than attenuate cAMP generation . For example , PKA phosphorylation of NMDARs [73] or L-type voltage-gated Ca2+ channels [90] increases Ca2+ influx through these channels , which then enhances the activation of Ca2+/calmodulin-stimulated adenylyl cyclases . Therefore , a robust mechanism for the reduction of cAMP is needed to overcome this array of PKA effects . There are comparatively few known PKA targets that lead to reduced cAMP signaling in CA1 pyramidal neurons . One such mechanism is PKA inhibition of AC8 [91]; however , the modest inhibition of AC8 by PKA observed in HEK293 cells ( ~30% reduction of FRET after 3 min forskolin stimulation ) is likely too weak to reduce NMDA-induced cAMP under our conditions . This is compounded by the relatively small contribution of AC8 to the cAMP response in the first place , due to its relatively low affinity for Ca2+/calmodulin ( ~800 nM vs . ~150 nM for AC1 ) . Nonetheless , our results cannot preclude the possibility that PKA phosphorylation of both AC1 and AC8 contributes to the experimental observations . The negative feedback loop of PKA phosphorylation of PDE4s cannot completely account for the experimental results in part because this mechanism leads to reduction in the NMDA alone response , and in part because this limits PKA activity itself , and thus limits the amount of pPDE4 . Indeed , for a negative feedback loop to allow a large but transient response ( required to uphold the isoproterenol response and repress the subsequent NMDA response ) , a time delay followed by rapid activation of the negative feedback loop is required [92–94] . The addition of dynamic recruitment to the model produced a moderate time delay , but PDE4 enhancement still began during the isoproterenol and NMDA alone pulses . Several mechanisms , including further enhancement of pPDE4 activity by SUMOylation [76] , may produce the requisite delay . Though we included the effect of SUMOylation , by allowing up to 40 fold increase in activity of plasma membrane PDE4 , this effect was instant . In contrast , a delay in activation of SUMO may have been able to produce the experimental results . An alternative to SUMOylation is proffered by a recent study showing that CaMKII phosphorylation of PDE4 increases its activity in cardiac myocytes [95] . This suggests that a 10–20 fold increase in PDE4 activity of dual PKA/CaMKII phosphorylated PDE4 might provide a delay in PDE4 enhancement tied to the NMDA delay . Another possible mechanism involves the ultrasensitive switch dynamics of ERK activation [65 , 66] . Delayed phosphorylation of PDE4 by ERK accompanied by a large enhancement in PDE4 activity would provide the needed delay in PDE4 enhancement . ERK indeed phosphorylates some PDE4 isoforms [96] , though the most common result is inhibition [97] . ERK itself could be activated via PKA phosphorylation of the βAR followed by either switching or arrestin recruitment [61] , via PKA phosphorylation of B-Raf [98] , or through other pathways not involving PKA . If ERK is involved in producing the experimental results , then MEK inhibitors should block the smaller NMDA response after ISO , and biochemical assays could be employed to demonstrate both an increase in ERK phosphorylation and the reduction in phosphorylated ERK when PKA inhibitors are applied . Though the model implements the spatial detail of submembrane location for membrane bound molecules , nanodomain mechanisms may be operating in the experiments that were omitted from the model . One nanodomain mechanism involves more specific localization of PDE4 subtypes , and extrapolates from the known differential affinity of rolipram for different PDE4 subtypes [99] . This mechanism assumes that PKA phosphorylation of PDE4s is limited to those anchored in a nanodomain around the NMDA receptor , and that rolipram specifically inhibits that NMDA-associated-subset of PDE4s . Such a nanodomain of PKA phosphorylated PDE4s might yield a model response to rolipram similar to that of experiments . Another nanodomain involves localization of different pools of adenylyl cyclases . If the pool of adenylyl cyclase activated by isoproterenol were distinct from the pool of adenylyl cyclase activated by NMDA , there would be no synergistic activation of AC1 by isoproterenol and NMDA . The existence of adenylyl cyclase nanodomains could be tested experimentally: In the absence of such nanodomains , the model predicts that NMDA application simultaneous with ISO application would produce a synergistic increase in cAMP . If experiments reveal an absence in synergy , then either the dominant subtype of adenylyl cyclase or the spatial location of adenylyl cyclases in the model needs modification . Furthermore , in the absence of synergy , the enhanced PDE4 activity provided by PKA phosphorylation ( Fig 4E ) might be sufficient to reduce NMDA induced cAMP production . Since the discovery of Gs-Gi switching after PKA phosphorylation of β-adrenergic receptors , the signaling pathways downstream of switching have been characterized in several cell types [62 , 63 , 100 , 101] , but have only recently been considered in neurons . Prior research on HEK293 cells showed that ERK activation was dependent both on PKA activity and on pertussis toxin sensitive G proteins [62] . In CHO cells , norepinephrine-stimulated ERK activation specifically requires PKA phosphorylation of β-adrenergic receptors [63] . In the hippocampus , recent experiments [102] employing novel βAR antagonists suggest that switching is involved in the LTP underlying memory storage . Specifically , theta-burst LTP is not blocked by propranolol , but LTP is blocked by the complete antagonist ICI 118551 . Propanolol is an antagonist that blocks cAMP production but not ERK activation in response to isoproterenol [103] , suggesting that the role of βAR activation in LTP is to promote ERK activation . In addition , genetic disruption of PKA anchoring to the β-adrenergic receptors produces deficits in both PKA phosphorylation of β-adrenergic receptors and ERK activation [102] . Our study uses a different approach to arrive at a similar conclusion that switching occurs in hippocampal neurons , a concept that may spur a novel line of research into alternative mechanisms of ERK activation underlying memory . Our model suggests that ultrasensitive or switch-like behavior may be important for Gs-Gi switching in these neurons . Since the intracellular C-terminal tails of β2ARs have four consensus PKA phosphorylation sites [67] , serines 261 , 262 , 345 , and 346 , it is plausible that the mechanism of switching is dependent on up to 4-site PKA phosphorylation of βARs , and thus we tested models of 1- , 2- , and 4-step phosphorylation of βARs . The sites were phosphorylated in a step-wise , or distributive , manner by PKA in the presence of phosphatase activity ( see Table 5 ) , conditions that are considered essential for producing switch-like behavior [64–66] . In addition , cooperativity between the phosphorylation sites was required , as the 1-site model did not exhibit switch-like behavior . These requirements are similar to those suggested for switch-like behavior at synapses [104] . The timing of receptor activation is important in determining the resultant cAMP signaling dynamics . When NMDA and isoproterenol are applied simultaneously , the induced cAMP response is synergistic ( Fig 8A ) , similar to that reported in HEK293 cells [16] . However , as the experiments show , when isoproterenol precedes NMDA by several minutes , the cAMP response consistently is sublinear . Therefore , the signaling pathways activated by βAR depend on the temporal pattern of stimulation . βAR may enhance cAMP under some temporal conditions , and undergo switching and Gi production under other conditions . Thus , during β-LTP experiments , isoproterenol application may not be contributing cAMP and PKA activation . Instead , our results , together with those of [102] , suggest that isoproterenol is enhancing ERK activation through Gi proteins . More importantly , βAR may be signaling through two different pathways in vivo . Thus , firing of noradrenergic locus coeruleus neurons just prior to CA3 neurons during behavior may enhance cAMP synergistically through Gs production , whereas the enhanced hippocampal memory formation by exogenous norepinephrine [72] , or the chronic release of norepinephrine during stress [105] may be acting through the switching pathway . | Noradrenaline is a stress related molecule that facilitates learning and memory when released in the hippocampus . The facilitation of memory is related to modulation of synaptic plasticity , but the mechanisms underlying this modulation are not well understood . We utilize a combination of live cell imaging and computational modeling to discover how noradrenergic receptor stimulation interacts with other molecules , such as calcium , required for synaptic plasticity and memory storage . Though prior work has shown that noradrenergic receptors and calcium interact synergistically to elevate intracellular second messengers when combined simultaneously , our results demonstrate that prior stimulation of noradrenergic receptors inhibits the elevation of intracellular second messengers . Our results further demonstrate that the inhibition may be caused by the noradrenergic receptor switching signaling pathways , thereby recruiting a different set of memory kinases . This switching represents a novel mechanism for recruiting molecules involved in synaptic plasticity and memory . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"phosphorylation",
"medicine",
"and",
"health",
"sciences",
"enzymes",
"drugs",
"enzymology",
"neurites",
"simulation",
"and",
"modeling",
"fluorophotometry",
"lyases",
"pharmacology",
"cellular",
"structures",
"and",
"organelles",
"neuronal",
"dendrites",
"research",
"an... | 2016 | Control of βAR- and N-methyl-D-aspartate (NMDA) Receptor-Dependent cAMP Dynamics in Hippocampal Neurons |
Influenza Virus ( IV ) pneumonia is associated with severe damage of the lung epithelium and respiratory failure . Apart from efficient host defense , structural repair of the injured epithelium is crucial for survival of severe pneumonia . The molecular mechanisms underlying stem/progenitor cell mediated regenerative responses are not well characterized . In particular , the impact of IV infection on lung stem cells and their regenerative responses remains elusive . Our study demonstrates that a highly pathogenic IV infects various cell populations in the murine lung , but displays a strong tropism to an epithelial cell subset with high proliferative capacity , defined by the signature EpCamhighCD24lowintegrin ( α6 ) high . This cell fraction expressed the stem cell antigen-1 , highly enriched lung stem/progenitor cells previously characterized by the signature integrin ( β4 ) +CD200+ , and upregulated the p63/krt5 regeneration program after IV-induced injury . Using 3-dimensional organoid cultures derived from these epithelial stem/progenitor cells ( EpiSPC ) , and in vivo infection models including transgenic mice , we reveal that their expansion , barrier renewal and outcome after IV-induced injury critically depended on Fgfr2b signaling . Importantly , IV infected EpiSPC exhibited severely impaired renewal capacity due to IV-induced blockade of β-catenin-dependent Fgfr2b signaling , evidenced by loss of alveolar tissue repair capacity after intrapulmonary EpiSPC transplantation in vivo . Intratracheal application of exogenous Fgf10 , however , resulted in increased engagement of non-infected EpiSPC for tissue regeneration , demonstrated by improved proliferative potential , restoration of alveolar barrier function and increased survival following IV pneumonia . Together , these data suggest that tropism of IV to distal lung stem cell niches represents an important factor of pathogenicity and highlight impaired Fgfr2b signaling as underlying mechanism . Furthermore , increase of alveolar Fgf10 levels may represent a putative therapy to overcome regeneration failure after IV-induced lung injury .
Influenza viruses ( IV ) may cause primary viral pneumonia in humans with rapid progression to lung failure and fatal outcome , and treatment options for this sometimes devastating disease are limited [1 , 2] . Histopathology and clinical features of IV-induced lung injury in humans resemble those of other forms of ARDS ( acute respiratory distress syndrome ) and are characterized by apoptotic and necrotic airway and alveolar epithelial cell death , loss of pulmonary barrier function and severe hypoxemia [1 , 3 , 4] . IV primarily infect cell subsets of the upper and lower respiratory tract . In the latter , these are particularly ciliated and goblet cells , club cells and alveolar epithelial cells type II ( AECII ) [5–7] . Injury of lung epithelial cells is induced by both direct viral cytopathogenicity and unbalanced immune responses [8–11] . The initiation of well-coordinated programs of inflammation termination and of regeneration of the injured distal lung epithelium are a prerequisite for the re-establishment of proper gas exchange . Absence or imbalance of these responses may at best result in chronically organizing infiltrates and aberrant or excess remodeling with tissue fibrosis , associated with long-term pulmonary organ dysfunction in ARDS survivors [12 , 13] , or in fatal outcome at worst . However , the cellular communication patterns and molecular networks underlying regeneration of the distal lung compartment after severe pathogen-associated injury are incompletely understood to date . In particular , the distinct mechanisms of interaction between injury-causing pathogens with components of regenerative signaling pathways within the lung stem cell niche , determining outcome of the repair response , have not been studied in detail . Alveolar re-epithelialization after injury was shown to involve different populations of endogenous , organ-resident stem/progenitor cells , which express lineage markers of distal lung epithelium such as club cell-specific protein ( CC10/scgb1a1 ) or surfactant protein C ( SPC/sftpc ) , are quiescent under normal conditions and proliferate during repair [14] . More recent reports revealed that intrinsically committed distal airway stem cells ( DASC ) expressing keratin 5 ( krt5 ) and the transcription factor p63 were found to contribute to de novo generation of both bronchiolar and alveolar tissue after formation of cell “pods” in a murine model of IV infection [15 , 16] . Vaughan et al . defined lineage-negative , integrin ( β4 ) +CD200+ epithelial progenitors as the source of p63/krt5+ amplifying cells regenerating airways and alveoli , highlighting integrin ( β4 ) +CD200+ epithelial cells as important progenitors regenerating the distal lung following IV-induced injury [17] . During regeneration processes , the lung stroma likely plays a key role by maintaining the distinct microenvironment of the stem cell niche , involving extracellular matrix , direct cell-cell contacts and autocrine or paracrine mediators . These signals initiate and co-ordinate self-renewal , fate determination and terminal differentiation of stem/progenitor cells . Different subsets of resident lung stromal/mesenchymal cells have been attributed a role in these processes , including parabronchial smooth muscle cells [18] , Sca-1high lung mesenchymal cells [19 , 20] or a human vimentin+ lung fibroblast population [21] . Signals involved in these cross-talk events include , among others , the paracrine fibroblast growth factors ( Fgfs ) , which regulate cell survival , proliferation , differentiation , and motility . In particular , Fgf7 and Fgf10 and their common tyrosine kinase receptor Fgfr2b ( fibroblast growth factor receptor 2b ) , are indispensable for distal lung development including branching morphogenesis [19 , 22–24] . Fgfr2b signaling is also re-activated in stem cell niches of the adult lung after different forms of injury to regenerate the epithelium [23 , 25 , 26] . The regulation of ligand and receptor expression of the Fgf7/10-Fgfr2b network in the context of lung repair after infectious injury , however , is not well understood . In the current study , we demonstrate that a highly proliferating EpCamhighCD24lowintegrin ( α6β4 ) highCD200+ distal lung epithelial cell population represents a primary target of pathogenic IV . This population highly enriched cells expressing key characteristics of distal lung epithelial stem/progenitor cells mediating bronchiolar and alveolar repair . Of note , IV tropism to these cells significantly reduced their regeneration capacity by impairment of β-catenin-dependent Fgfr2b signaling . These data for the first time demonstrate that the extent of lung stem/progenitor cell infection by IV is a hallmark of pathogenicity as it critically impacts on lung regeneration capacity after severe IV injury . Moreover , IV-induced regeneration failure could be counteracted by intratracheal application of excess recombinant Fgf10 , suggesting recruitment of the non-infected Fgfr2bhigh stem cell fraction for repair as putative novel treatment strategy to drive organ regeneration in patients with IV-induced ARDS .
It is well established that IV infect different subsets of the airways and alveoli , particularly ciliated and goblet cells , club cells and AECII [5–7] . However , recent advances in the field resulted in the definition of more specialized subsets of lung epithelial cells , some of which display stem/progenitor cell characteristics and contribute to repair of the injured organ [17 , 19 , 27] . To address which of these epithelial cell compartments were infected by IV , we fractionated distal lung epithelial cells into different subsets , after dissection of large airways and vessels and depletion of leukocytes and endothelial cells , according to surface expression levels of EpCam and integrin α6 [19] , and the lineage markers CD24 ( differentiated airway epithelial cells ) [19] , CC10 ( club cells ) , pro-SPC ( AEC II ) and T1α ( AEC I ) , by flow cytometry . We identified a high-frequent EpCamlowα6low fraction ( 91 . 3 ± 1 . 8% ) and a low frequent EpCamhighα6high fraction , the latter of which consisted of a CD24low and CD24high population ( 1 . 7 ± 0 . 3% and 6 . 3 ± 1 . 8% , respectively , Fig 1A ) . The majority of the most abundant EpCamlowα6low cells showed a granular cytoplasm typically observed in AEC II , with approximately 95% of the cells expressing the AEC II signature pro-SPChighT1αneg and around 5% expressing an AEC I signature ( SPCnegT1α+ ) ( Fig 1B ) . EpCamhighα6highCD24high cells contained pro-SPCnegCC10neg differentiated small airway epithelial cells ( SAEC , 70% ) , composed of both β-tubulin+ ciliated and mucin5AC+ goblet cells , and pro-SPCnegCC10+ club cells ( 30% ) ( Fig 1C ) . EpCamhighα6highCD24low cells were cells of homogeneous morphology and stained positive for the stem cell antigen Sca-1+ ( Fig 1D ) . To analyse which of these epithelial cell subset were targeted by IV , we infected C57BL/6 mice using 500pfu of IV strains of increasing pathogenicity , i . e . low-pathogenic H3N2 ( x-31 ) , pandemic pH1N1 strain ( A/Hamburg/04/09 ) , causing mild to moderate lung injury at this dose in mice , and the highly pathogenic mouse-adapted PR/8 strain [28 , 29] . Quantification of the infection rates by staining for IV nucleoprotein ( NP ) revealed that EpCamlowα6highCD24high and EpCamlowα6low cells were infected with a frequency of ∼11% and ∼6% , respectively , by d4 pi after PR/8 infection , a time point where PR/8 replication in the lung reaches a peak [28] . Subfractionation into differentiated alveolar and airway epithelial cells revealed that rates of PR/8 infection in AEC I and AEC II ranged at ∼8% and ∼4% , whereas club cells and ciliated/goblet cells displayed similar PR/8 infection rates of ∼10% . Of note , EpCamhighα6highCD24low cells were infected by IV to high amounts ( around 15% of all EpCamhighα6highCD24low after PR/8 infection ) , and the proportion of infected epithelial cell subsets at d4 pi correlated with the level of pathogenicity of the IV strain used ( Fig 1E and 1F ) . To further address whether PR/8 revealed increased tropism to EpCamhighα6highCD24low cells , AEC , SAEC and EpCamhighα6highCD24low cells were flow-sorted , seeded into culture plates at equal numbers and infected ex vivo with x-31 , PR/8 and pH1N1 at an MOI of 2 , respectively . After one replication cycle ( 6h ) , excluding de novo infection by progeny virions , the infection rate was determined , reflecting the capacity of each virus strain to infect the respective cell population . Similar to our in vivo results , we observed that PR/8 reveals an enhanced tropism to EpCamhighα6highCD24low cells ( S1 Fig ) . Given that the EpCamhighα6highCD24lowSca-1pos cells which revealed the highest rates of infection were previously described as epithelial stem/progenitor population giving rise to airway and alveolar epithelium [19] , we aimed to further characterize their phenotype and stemness properties . Further analyses using established stem cell markers [17] revealed that they were integrin β4+CD200+ , a signature which has been confined to a distal lung stem cell phenotype known to engage the krt5/p63 regeneration program [17] ( Fig 1D ) . In accordance , these cells were negative for krt5 and p63 in healthy lungs , but highly upregulated krt5 and p63 gene expression after IV-induced injury ( Fig 1G ) , suggesting that they contain epithelial stem/progenitor cells ( EpiSPC ) . To verify these stem/progenitor cell characteristics ex vivo , EpCam+ cell fractions were flow-sorted and seeded in organotypic 3D cultures [30] . As opposed to AEC I/II , and SAEC/club cells , EpiSPC developed typical large organoid spheres with cystic or saccular outgrowth in the presence of the growth factors Fgf10 and Hgf ( hepatocyte growth factor ) , a characteristic feature of lung stem/progenitor cells [19] ( S2A Fig ) . A robust clonogenic potential of EpiSPC was demonstrated by repetitive cycles of serial passaging of digested organospheres by single-cell sorting and detection of de novo sphere formation after one week , respectively ( S2B Fig ) , and by use of clonality assays where tdtomato+ EpiSPC were mixed with wildtype ( wt ) EpiSPC at a defined ratio and cultured in matrix , resulting in pure tdtomato+ and tdtomatoneg colonies indicative of clonal expansion ( S2C Fig ) . Given that Fgf10-Fgfr2b signaling is indispensable for epithelial stem cell outgrowth ex vivo [18 , 23 , 30] , we aimed to further define whether EpiSPC expansion and differentiation were Fgf10-dependent . To understand cellular cross-talk mechanisms involved in activation of the regenerative Fgfr2b axis after IV-induced lung injury we sought to define the predominant cellular source of Fgfr2b ligands , Fgf7 and 10 under homeostatic conditions and in the acute phase of IV-induced lung injury , at the peak of EpiSPC proliferation ( d7 pi ) . Lung digests of mock- or PR/8-infected mice ( d7 pi ) were therefore fractionated by FACS sorting into four main lineages , including endothelial cells ( CD31+CD45negEpCamneg , R1 ) , leukocytes ( CD31negCD45+EpCamneg , R2 ) , epithelial cells ( CD31negCD45negEpCam+ , R3 ) and CD31negCD45negEpCamnegSca-1high cells ( S3A Fig , left , R4 ) . mRNA expression of Fgfr2b ligands in these four populations revealed that both Fgf7 and Fgf10 expression was significantly increased in the R4 fraction ( CD31negCD45negEpCamnegSca-1high ) compared to the other three major lineages of the lung ( endothelial cells , epithelial cells , leukocytes ) , independent of IV infection ( S3A Fig , right ) . Previous data suggested that Sca-1high expression in EpCamneg cells was associated with the fibroblast lineage [20] . To address whether Fgf10-expressing resident mesenchymal cells ( rMCs ) would support organosphere generation , flow-sorted EpiSPC and rMC were co-cultured for several days in absence of growth factor supplementation . As shown in S3B Fig , presence of rMC was sufficient to drive early organosphere formation at d5 of culture . Furthermore , rMCs mediated saccular outgrowth of EpiSPC spheres at d10 and formation of lung-like structures at d16 of culture , as compared to EpiSPC mono-cultures in supplemented medium , and this response was abrogated early in the cystic phase when Fgf10 was blocked by neutralizing antibodies ( S3D Fig ) . Of note , co-culture of EpiSPC with either flow-sorted CD45+ ( R2 ) or CD31+ ( R1 ) cells did not result in organosphere formation ( S3C Fig ) . Finally , lung-like structures derived from EpiSPC-rMA co-cultures significantly upregulated markers of terminal airway and alveolar cell differentiation , such as T1α and β-tubulin ( S3E Fig ) . Together , these data indicate that EpiSPC both self-renew and differentiate to distal lung epithelial cell subsets in an Fgf10-dependent manner during organotypic culture . Analyses of the proliferative response of various distal lung epithelial cell populations after PR/8 infection revealed that the EpiSPC population showed the highest proliferation capacity compared to the AEC and SAEC subsets between d7 to d14 after injury ( Figs 2A and S4 ) . Of note , comparison of EpiSPC and AEC proportions under homeostatic conditions and at d7 pi revealed an increase of the EpiSPC pool from 1 . 7 ± 0 . 3% to 5 . 7 ± 1 . 8% after IV- infection , whereas the high frequent AEC pool is reduced from 91 . 3 ± 1 . 8% to 84 . 7 ± 4 . 4% . Quantification of phosphatidylserine externalization by Annexin V expression of non-infected and PR/8 infected wt mice at d7 pi , a time point where apoptotic epithelial injury is most prominent in the lungs [8 , 28] , revealed that EpiSPC were resistant to IV-induced apoptosis , whereas the other EpCam+ populations showed high levels of apoptosis in response to infection ( Fig 2B ) . These findings suggest that a damage-resistant cell population with high proliferative capacity is contained in the EpiSPC fraction , which might contribute to renewal of both bronchiolar and alveolar epithelial tissue [19] . Given that Fgf10 is an indispensable growth factor for EpiSPC outgrowth ex vivo and that genetic deletion of Fgf10 or of the Fgf10 receptor Fgfr2b results in failure of embryonic lung development [25] , we speculated that this pathway might be reactivated to drive EpiSPC proliferation for epithelial regeneration after infectious injury in adult mice . In fact , Fgfr2b surface expression was significantly upregulated on EpiSPC in the course of severe PR/8 infection compared to mock-infected mice , most prominent at d7 post infection ( pi ) when the proliferative response was highly increased ( Fig 2C ) . To decipher the functional role of Fgfr2b and its ligands Fgf10 and Fgf7 in this EpiSPC renewal response , transgenic mice with inducible overexpression of either soluble dominant negative Fgfr2b or overexpression of the ligand Fgf10 , were PR/8-infected and proliferation of all EpCam+ subsets was quantified at d7 pi . Attenuation of Fgfr2b signaling by doxycycline induction of a dominant-negative Fgfr2b ( scavenging soluble Fgf10 ) resulted in significant impairment of EpiSPC proliferation capacity compared to non-induced littermates , whereas AEC and SAEC revealed no or little , Fgfr2b-independent proliferation ( Fig 2D ) . Similarly , the proliferating proportion of EpiSPC was significantly increased in mice with induced overexpression of Fgf10 at d7 pi , compared to non-induced litters ( Fig 2E ) . Of note , Fgf7-/- mice exhibited only slightly but not significantly reduced EpiSPC proliferation in comparison to Fgf7+/+ mice ( Fig 2F ) . To verify that the Fgf10-Fgfr2b axis is indeed a key pathway in the epithelial regenerative response of the distal lung following IV-induced injury , we determined re-establishment of barrier function and outcome in doxycycline ( dox ) -induced versus non-induced Rosa26rtTA/+;tet ( O ) sFgfr2b/+ mice . Blockade of Fgfr2b signaling resulted in significantly increased lung permeability ( as determined by alveolar albumin leakage ) during the repair phase at d14 pi ( S5A Fig ) , indicating that this pathway is crucial for re-establishment of gas exchange function . Concomitantly , the surviving sFgfr2b overexpressing mice showed decreased body weight compared to controls during the regeneration phase at d11 to d20 pi , and did not fully regain weight until d21 ( S5B Fig ) . Together , these findings demonstrate that IV infection induces activation of an Fgfr2b-dependent signaling pathway , which largely mediates the EpiSPC proliferative response and barrier repair after IV-induced injury . Given that the EpiSPC cell fraction harbored stem/progenitor cells crucial for Fgfr2b-drived lung repair and at the same time represented primary targets of high pathogenic PR/8 in the distal lung , we sought to address whether infection of these cells would result in an impaired renewal response . We therefore infected C57BL/6 mice using 500pfu of IV strains of increasing pathogenicity , i . e . x-31 , pH1N1 , and the highly pathogenic PR/8 [28 , 29] . After 21 days , when mice had apparently recovered from IV infection , x-31 and pH1N1 infected mice showed a restored distal lung epithelial architecture , whereas mice infected with the highly pathogenic PR/8 still presented with thickened alveolar walls and incomplete re-epithelialization ( Fig 3A ) , suggesting that IV of high pathogenicity impaired the regenerative response of the lung . Quantification of Ki67+ fractions in infected ( NP+ ) versus non-infected ( NPneg ) EpiSPC within the same PR/8 infected lungs revealed that the proliferative response was lost in infected EpiSPC ( Fig 3B ) , associated with loss of Fgfr2b upregulation in NP+ EpiSPC ( Fig 3C ) . Finally , infection of flow-sorted EpiSPC ex vivo with increasing doses of PR/8 followed by organotypic culture resulted in significantly reduced formation of organospheres depending on the multiplicity of infection ( MOI ) applied ( Figs 3D and S6 ) . This was not due to infection-induced death of EpiSPC ( as analysed by live/dead staining ) . Importantly and in line with these data , intratracheal transplantation of non-infected ( viral hemagglutinin-neg; HAneg ) EpiSPC , flow-sorted from the lungs of IV-infected tdtomato+ mice , into IV-infected wildtype mice , resulted in integration into and in de novo generation of distal lung tissue ( including AEC I ) between d7-d14 post transplantation , whereas infected ( HA+ ) EpiSPC or SAEC showed incorporation into distal lung tissue , but only limited expansion and did not give rise to distal lung tissue ( Figs 3E , S7A and S7B ) . These data indicate that EpiSPC indeed contain precursors of alveolar tissue in vivo , and IV infection of the EpiSPC niche results in defective tissue repair after IV-induced injury ( Figs 3E and S7B ) . Of note , transplanted flow-sorted EpiSPC of non-infected tdtomato+ mice can be visualized at d14 post transplantation in the lung tissue of IV-infected wt mice , but do not expand to generate tissue de novo ( S7B and S7C Fig ) , suggesting that factors expressed within the stem cell niche during IV-induced injury play a crucial role in early activation of quiescent EpiSPC for tissue repair ( e . g . via IV-induced upregulation of the p63/krt5 regeneration program , Fig 1G ) . We next addressed the putative mechanism of inhibition of Fgfr2b upregulation in infected EpiSPC . Previous data revealed that Fgfr2b expression is dependent on Wnt/β-catenin signaling in the developing lung [31] . Indeed , conditional knockout of β-catenin by tamoxifen treatment in adult distal lung epithelial cells from Rosa26ERTCre/ERTCre;Ctnnb1flox/flox mice grown ex vivo ( Fig 4A , left ) resulted in impaired upregulation of fgfr2b mRNA expression after PR/8 infection ( Fig 4A , right ) . Recent data suggest that β-catenin is involved in expression of interferon-dependent genes and that IV block β-catenin transcriptional activity in vitro as part of an antiviral escape strategy [32] . In fact , activation of the Wnt/β-catenin pathway by LiCl resulted in widely reduced IV replication in ex vivo cultured distal lung epithelial cells , whereas inhibition increased replication , as demonstrated by immunofluorescence and quantification of the viral m segment expression ( Fig 4B ) . Concomitantly , expression of the β-catenin target genes Axin and Ccnd1 , and of Fgfr2b , was reduced by ∼50 to 100-fold in infected ( HA+ ) compared to non-infected ( HAneg ) EpiSPC flow-sorted from PR/8-challenged mice at d7 pi ( Fig 4C ) . These data indicate that EpiSPC tropism of IV represents a key factor of pathogenicity . IV interfere with β-catenin-dependent gene transcription in infected EpiSPC , likely to escape β-catenin anti-viral properties , which results in impaired Fgfr2b expression and reduced renewal capacity in infected EpiSPC . To evaluate whether alveolar deposition of excess Fgf10 would counteract the impaired Fgfr2b-mediated renewal response in IV infected mice by increased recruitment of the non-infected Fgfr2bhigh EpiSPC , we applied recombinant Fgf10 or PBS to IV-infected C57BL/6 mice at d6 pi . Indeed , Fgf10 treatment resulted in significantly increased proliferation of EpiSPC compared to PBS-treated mice at d7 pi ( Fig 5A ) . IV-infected mice showed a severely disturbed lung architecture with distinct cellular infiltrates , areas of extensive atelectasis and loss of epithelial cells at d10 pi , which was partially reverted by Fgf10 treatment ( Fig 5B , left ) . By d21 , Fgf10-treated mice revealed an almost normal lung structure with re-epithelialized bronchioli and alveoli ( Fig 5B , right , arrowheads ) , whereas PBS-treated controls still presented with areas of atelectasis , cellular infiltrates and epithelial injury , indicating failure of epithelial renewal and persisting injury-associated inflammatory responses ( Fig 5B , right , arrows ) . To verify that Fgf10 indeed impacted re-establishment of bronchiolar and alveolar epithelial structures , E-cadherin stainings of lung sections were performed and revealed that Fgf10 treatment resulted in complete re-establishment of the epithelium at d21 pi , associated with increased numbers of Ki67+ proliferating cells , whereas PBS-treated mice showed only partial epithelial renewal ( Fig 5C ) . These findings were confirmed by quantification of the total numbers of EpCam+ cells in these treatment groups ( Fig 5D ) . Of note , Fgf10-treated murine lungs showed increased expression of krt5 , a marker of stem cell-induced repair , at d21 pi ( Fig 5E ) [16 , 17] . Finally , Fgf10-mediated epithelial repair resulted in improved barrier function at d14 pi , and improved survival until d21 compared to controls ( Fig 5F and 5G ) , highlighting the therapeutic potential of Fgf10 to improve EpiSPC-dependent epithelial regeneration .
Repair of the injured lung epithelium including structural and functional re-establishment of alveolar barrier function after severe IV pneumonia is crucial for recovery and survival . In this study , we demonstrate that a fraction of distal lung epithelial cells phenotyped as EpCamhighα6β4highCD24lowSca-1highCD200+ EpiSPC drive epithelial renewal processes involving Fgf10/Fgfr2b-mediated signaling . Of note , the impaired alveolar regeneration after infection with highly pathogenic IV observed in mice and reported in humans [3] was associated with increased viral infection rates of EpiSPC in the distal lung compared to AEC and SAEC , and IV-induced inhibition of β-catenin-dependent gene transcription impaired regenerative Fgfr2b-signaling in these progenitor cells . Whereas transplantation of non-infected EpiSPC isolated from murine lungs , resulted in integration into lung tissue and de novo generation of distal lung epithelium including AEC I , previously infected EpiSPC did not give rise to lung tissue in the distal lung . These data highlight that tropism of IV to subsets of the lung stem cell niche may be a crucial determinant of IV pathogenicity resulting in severe impairment of Fgfr2b-mediated lung regeneration , and likely persistent failure of barrier function and worsened outcome . Regeneration of the epithelial compartment of the distal lung was shown to involve different stem/progenitor cell populations , including p63+krt5+ lineage-negative , β4+ epithelial progenitors or distal airway stem cells ( DASCp63/krt5 ) [16 , 17] , α6β4high alveolar cells , and more lineage-committed CC10+ or SPC+ populations [14 , 17 , 33 , 34] . Our data show that the EpiSPC population phenotyped as EpCamhighα6β4highCD24lowSca-1+CD200+ [19 , 35] contained cells with stem characteristics as verified by organoid outgrowth in 3D cultures , and clonogenic potential in presence of growth factors including Fgf10 . Furthermore , EpiSPC gave rise to cells expressing markers of terminally differentiated airway and alveolar epithelium in matrigel , suggesting that they are precursors of bronchiolar and alveolar epithelium . EpiSPC displayed high proliferation capacity after bronchio-alveolar injury caused by IV infection , as opposed to other distal lung epithelial cell populations . This renewal response was associated with strong induction of the p63/krt5 regeneration program , found to be crucial for distal lung repair [17 , 36] . Furthermore , Fgf10-treated mice increased krt5 expression in their lungs , suggesting that the cells we identify as EpiSPC contribute to the p63/krt5 pool , and that expansion or generation of krt5+ cells is dependent on Fgf10 . Conflicting data exist on the capacity of lineage-committed cells to be progenitors of differentiated distal lung cells [17 , 37 , 38] . Zheng et al suggested that most of the newly induced p63+ cells in the IV-damaged distal lung compartment might be derived from CC10+ cells , whereas a recent report defined them as lineage-negative [17 , 39] , highlighting that the contribution of different stem/progenitor populations to alveolar repair may be injury-specific , dependent on the region and extent of injury , and on microenvironmental factors regulated in the context of defined types of damage . With respect to recent data highlighting the AEC II pool as stem cell niche of the alveolar epithelium , we found that the EpCamlowα6low AEC II fraction proliferated only to a limited extent after IV infection in vivo . However , our data do not fully exclude contribution of an AEC II progenitor to the alveolar regeneration process , as demonstrated for bleomycin-induced damage [33] , particularly as AECII constitute a highly abundant cell population of the lung , and even a small proliferating AECII fraction could still contribute to epithelial repair . Clearly , EpiSPC proliferation after IV-induced injury largely depended on Fgfr2b and its ligand Fgf10 , as demonstrated by use of Rosa26rtTA/+;tet ( O ) sFgfr2b/+ , and Rosa26rtTA/+;tet ( O ) Fgf10/+ mice , whereas Fgf7 did not substantially contribute . Given that fgf7-/- mice are viable and do not display gross lung abnormalities as compared to Fgf10-/- and Fgfr2b-/- mice [25] , and given that organoid formation from EpiSPC ex vivo is strictly dependent on Fgf10 but does not require Fgf7 [30] , we conclude that Fgf10 rather than Fgf7 mediates EpiSPC renewal , although both ligands share the same receptor . A recent report highlights different functions of these ligands with respect to Fgfr2b processing and recycling [40] , suggesting that the prolonged proliferative response observed after exogenous Fgf10 application at d21 pi might be associated with Fgf10-induced maintenance of Fgfr2b expression on EpiSPC . With respect to cellular origin of Fgfr2b ligands within the stem/progenitor cell niche , an EpCamnegSca-1high cell population of non-leukocyte and non-endothelial lineage [24] was found to be the primary source and supported for lung organoid formation in 3D culture . The lung mesenchyme is known to be the cellular origin of Fgf10 during lung organogenesis and postnatally [23 , 24] , and therefore represents a key orchestrator of the EpiSPC niche . Cellular responses of Fgf10 in the developing lung include epithelial progenitor cell maintenance and prevention of epithelial differentiation [41] . Our data confirm a central role of Fgf10 in survival and proliferation in adult lung EpiSPC ex vivo and in vivo , and demonstrate that neutralization of Fgf10 in EpiSPC-mesenchymal cell co-cultures results in inhibition of sphere outgrowth at a very early stage . Canonical β-catenin signaling was found to induce expression of the Fgfr2b gene in the developing lung epithelium [31] . A key finding reported here is that IV infect EpiSPC and interfere with β-catenin-dependent gene transcription of Fgfr2b , resulting loss of Fgfr2b upregulation in the infected fraction of EpiSPC , which are thus unable to proliferate to promote repair . It has been recently demonstrated that β-catenin is indispensable for expression of type I interferon in response to IV infection in different cell lines . Many viruses have evolved gene products during co-evolution with their hosts by which they can induce and control various responses of their host cell , in particular early innate immune pathways . Mechanistically , a direct interaction of the viral NS1 protein , a potent antagonist of host innate antiviral responses [42] , with the Wnt receptor frizzled upstream of canonical β-catenin signaling was discussed [43] . More recent publications suggest an interaction of Influenza or Sendai Virus-induced host cell components of the NF-κB pathway or of IRF3 , respectively , with nuclear β-catenin to repress β-catenin-dependent gene transcription [32 , 44] . Additionally , another report suggests that expression of the pandemic 1918 IV polymerase subunit PB2 increases virulence by inhibition of the Wnt signaling cascade which impacts on regeneration of the inflamed lung tissue [45] . Our own data clearly support the concept that the canonical β-catenin pathway is anti-viral , as demonstrated in studies using primary distal lung epithelial cells ex vivo infected with IV in presence of a β-catenin activator or inhibitor . This suggests that inhibition of β-catenin-dependent gene transcription is a conserved strategy of viral immune escape , which additionally results in blockade of renewal programs in EpiSPC . Our data furthermore provide a comprehensive , FACS-based quantification of infection rates of various lung epithelial cell compartments of the distal murine lung , and particularly indicate that the extent of EpiSPC infection by different IV strains represents an important , previously undefined factor of viral pathogenicity . In addition , the finding that EpiSPC as opposed to differentiated epithelium [10] are not subjected to infection-associated apoptosis but survive ( likely due to constitutive expression of maintenance/renewal-associated survival pathways ) , raises the question whether viral ´imprinting´ will cause changes of transcriptional or epigenetic programs of stem cell plasticity resulting in long-term epithelial dysfunction . Altogether , we provide evidence that pathogens such as IV severely affect the progenitor cell-mediated , Fgfr2b-dependent repair of the distal lung epithelium , and that intratracheal treatment of pathogen-injured lungs with excess Fgf10 , to recruit the non-infected , Fgfr2bhigh EpiSPC fraction , promoted epithelial renewal without inducing aberrant repair [46] at the dose used . Fgf10 might therefore represent a putative treatment option to foster organ repair and re-establish gas exchange function after IV-induced and possibly other forms of ARDS .
Wildtype C57BL/6 mice were purchased from Charles River Laboratories . CMV-Cre mice [47] were crossed with rtTAflox mice [48] to generate mice expressing rtTA under the ubiquitous Rosa26 promoter . This constitutive Rosa26rtTA/rtTA mouse line was then crossed with tet ( O ) sFgfr2b/+ or tet ( O ) Fgf10/+ responder lines to generate Rosa26rtTA/+;tet ( O ) sFgfr2b/+ and Rosa26rtTA/+;tet ( O ) Fgf10/+ double heterozygous animals on a mixed genetic background , allowing ubiquitous expression of dominant-negative soluble Fgfr2b [49 , 50] or of Fgf10 [18 , 46] by administration of doxycycline-containing normal rodent diet with 0 . 0625% doxycycline ( Harlan Teklad ) . Mice were genotyped as described previously [46 , 50–52] . Fgf7-/- mice were obtained from Jackson Laboratory and backcrossed for several generations on a C57BL/6 background ( strain #4161 ) . B6 . 129 ( Cg ) -Gt ( ROSA ) 26Sortm4 ( ACTB-tdTomato , -EGFP ) Luo/J ( mTmG ) mice in C57BL/6 genetic background , a tamoxifen-responsive driver mouse line , and Ctnnb1flox/flox mice were obtained from Jackson Laboratory ( strains #7676 , #3309 and #4152 ) and the latter bred to generate homozygous Rosa26ERTCre/ERTCre;Ctnnb1flox/flox mice on a C57BL/6 background allowing induction of a β-catenin knockout by application of tamoxifen . Mice were housed under pathogen-free conditions and experiments were performed according to the regional institutions´ guidelines . The following antibodies were used for flow cytometric analyses , cell sorting or immunofluorescence: CD49f PE or Pacific Blue ( clone: GoH3 ) , CD326 ( EpCam ) APC-Cy7 or FITC ( clone G8 . 8 ) , CD24 PE-Cy7 ( clone: M1/69 ) , Ly-6A/E ( Sca-1 ) PerCP/Cy5 . 5 or Pacific Blue ( clone: D7 ) , CD31 Alexa fluor 488 or PE ( clone: MEC13 . 3 ) , CD45 FITC or APC-Cy7 ( clone: 30-F11 ) , CD200 PE ( clone: OX-90 ) , T1α/podoplanin APC ( clone: 8 . 1 . 1 . ) and corresponding isotype controls syrian hamster IgG ( clone SHG-1 ) ; all Biolegend . Influenza A virus nucleoprotein ( NP ) FITC ( clone: 431 , Abcam ) , Fgfr2b ( clone: 133730 ) and corresponding isotype control IgG2a ( clone: 54447 , both R&D Systems ) , p63 ( Life Span Biotechnology ) , CC10 ( clones T-18 and S-20 ) and isotype-matched normal goat IgG ( Santa Cruz Biotechnology ) , CD104 Alexa fluor 647 ( clone: 346-11A , AbD SeroTec ) , Ki67 FITC or PE ( clone: B56 ) and corresponding isotype control IgG1κ FITC or PE ( clone: MOPC-21 , both BD Bioscience ) , Annexin V Alexa fluor 647 ( Invitrogen ) , E-cadherin ( clone: DECMA-1 , Abcam ) , p63 Alexa fluor 555 ( clone: P51A , Bioss ) , Podoplanin ( clone: RTD4E10 , Abcam ) or corresponding isotype control syrian hamster IgG ( clone: SHG-1 , Abcam ) , beta IV tubulin ( clone: ONS . 1A6 ) and corresponding isotype control IgG1 ( clone: CT6 , both Abcam ) , cytokeratin 5 FITC ( Bioss ) , Uteroglobin ( clone: EPR12008 , abcam ) and isotype-matched monoclonal rabbit IgG ( abcam ) , purified Ki67 ( Thermo Scientific ) , purified pro-surfactant protein C ( Millipore ) , biotinylated mucin 5AC ( clone: 45M1 , Abcam ) , anti-influenza NP ( clone: 1331 , Meridian Life Science ) . Secondary antibodies used were anti-Streptavidin APC ( Becton Dickinson ) , anti-rabbit Alexa fluor 488/555 , anti-goat Alexa fluor 647 , anti-mouse Alexa fluor 555/647 , anti-rat Alexa fluor 488/647 , anti-hamster Alexa fluor 488 ( all Molecular Probes ) . Magnetic separation was performed using biotinylated rat anti-mouse CD45 , CD16/32 and CD31 mAb ( BD Bioscience , Pharmingen ) . For ex vivo neutralization assays anti-fgf10 ( clone: C-17 ) or normal goat IgG ( Santa Cruz Biotechnology ) were used at a concentration of 5μg/ml . For flow cytometric analyses , cells were routinely stained with 7-AAD ( Biolegend ) or fixable live/dead stain reagents ( Molecular Probes ) for dead cell exclusion . Mice were anaesthesized and intratracheally inoculated with 500pfu ( plaque forming units ) of A/PR/8/34 ( H1N1 , mouse-adapted ) , A/x-31 ( H3N2 ) , or A/Hamburg/5/09 ( pandemic H1N1 ) , grown and quantified in Madin Darby Canine Kidney ( MDCK ) cells ( obtained from American Type Culture Collection ) and diluted in a total volume of 70 μl in sterile PBS-/- . In the treatment approach , 5 μg recombinant Fgf10 ( R&D Systems ) dissolved in sterile PBS-/- or PBS-/- alone were intratracheally applied to IV infected mice . Venous blood and BALF were collected as described previously [53] . Lung permeability was determined by i . v . injection of 100 μl FITC-labeled albumin ( Sigma-Aldrich ) and quantification of FITC fluorescence ratios in BALF and serum with a fluorescence reader ( FLX800 , Bio-Tek instruments ) as described elsewhere [28] . In selected experiments , 20 , 000 flow-sorted HA+ or HAneg EpiSPC from IV infected mTmG mice were intratracheally applied into IV infected wt mice at d7 pi . Virus titers from BALF were determined by immunohistochemistry-based plaque assay on confluent MDCK cells in 6-well plates in duplicates as previously described [54] . Infected mice were monitored 1–3 times per day and a morbidity score was calculated from weight loss , general appearance , breathing frequency/dyspnea , and body temperature . Mice with a score ≥ 20 were moribund , sacrificed and classified as dead in mortality studies . Lung homogenates of distal lung cell suspensions were obtained by instillation of dispase ( BD Biosciences ) and 0 . 5 ml low-melting agarose ( Sigma ) through the trachea into the HBSS ( Gibco ) perfused lung , followed by incubation in dispase for 40 min as previously described [55] . After gelling of the agarose and removal of the agarose-filled trachea and proximal bronchial tree , the lung was homogenized ( GentleMACS , MACS Miltenyi Biotech ) in DMEM/2 . 5% HEPES with 0 . 01% DNase ( Serva ) and filtered through 100μm and 40μm nylon filters . Cell suspensions were incubated with biotinylated rat anti-mouse CD45 , CD16/32 and CD31 mAb for 30 min at 37°C followed by incubation with biotin-binding magnetic beads and magnetic separation to deplete leukocytes and endothelial cells prior to flow cytometric analysis and cell sorting or to further culture . Multicolor flow cytometry or high speed cell sorting was performed with an LSR Fortessa or an Aria III cell sorter using DIVA software ( BD Bioscience ) . For analytical measurements 1–5 x 105 cells were freshly stained with fluorochrome-labeled antibodies for 20 min at 4°C in BD FACS buffer . For intracellular stainings ( NP , proSPC ) , permeabilization of cells was achieved by previous incubation with 0 . 2% saponin in PBS-/- for 15 min at 4°C , followed by incubation with anti-NP FITC , anti-proSPC , or respective isotype control mAbs for 20 min at 4°C . When non-labeled primary mAb were used , a fluorescent labeled secondary Ab was added and incubated for 20 min at 4°C in FACS buffer . Doublets and dead cells were routinely excluded from the analyses ( the latter by using 7AAD ) . Annexin V staining was performed on fresh , non-permeabilized cells . Prior to antibody incubation , cells were washed and resuspended in Annexin V buffer ( BD Bioscience ) and incubated with Annexin V 647 and further mAbs in Annexin V buffer for 20 min at 4°C . The stained cells were washed and resuspended in Annexin V buffer . Cell sorting was performed with an 85 or 100μm nozzle . Single cell sorting was performed using the automated cell deposition unit ( ACDU ) with a 24-well plate and 12mm cell culture inserts ( 0 . 4μm pore size , Millipore ) . Purity of flow-sorted cells was always > 95% . Mono- or co-culture of EpiSPC and rMC was performed as described previously [30] . In brief , flow sorted cells were counted , resuspended and mixed with growth factor reduced matrix ( BD Biosciences ) diluted with EpiSPC medium ( α-MEM , 10% FCS , 1x pen/strep , 1x insulin/transferrin/selenium , 2mM L-glutamine , 0 . 0002% heparin ) at a 1:1 ratio . Cell suspensions were seeded in 12mm cell culture inserts ( 0 . 4μm pore size , Millipore ) in a 24-well plate and incubated for 5 min at 37°C , 5% CO2 to allow gelling . EpiSPC medium was then added to the bottom wells of the plate . In selected experiments , 50ng/ml recombinant Fgf10 and 30ng/ml recombinant Hgf ( both R&D systems ) or anti-Fgf10 or control Ab were added to the medium at day 2 of culture . For matrix digestion a preheated dispase/collagenase I ( Boehringer , Gibco ) mixture ( 3 mg/ml ) was added and a single cell suspension was obtained for re-seeding or single-cell sorting . Images were taken with a DM IL LED microscope and a corresponding camera MC170 HD ( Leica ) . To obtain lung cryosections , lungs were perfused with HBSS and filled with 1 . 5ml TissueTek ( Sakura ) mixed with PBS-/- at a 1:1 ratio as described [56] . Lungs were removed , snap-frozen and 4–10 μm sections were prepared using a Leica cryotome . In selected experiments , lungs were filled with a TissueTek/PBS-/- mixture containing 1% paraformaldehyde and were incubated in 1% paraformaldehyde after removal . Lung cryosections were stained with Hematoxylin-Eosin or fixed with 4% paraformaldehyde for 20 min , blocked with 0 . 05% Tween 20 , 5% BSA , 5% horse serum in PBS-/- for 30 min and stained with fluorochrome-labeled mAb diluted in PBS-/- , 0 . 1% BSA , 0 . 2% Triton X-100 for 2 h . After washing , secondary mAbs were added for 2 h , followed by mounting with Dapi containing mounting medium ( Vectashield , Vector Labs ) . Epithelial or mesenchymal cells cultured in chamber slides ( Nunc ) were fixed in a 1:1 ratio of cold acetone/methanol for 5 min and blocked with 3% BSA in PBS for 30 min prior to staining . Cytospins were additionally stained with Pappenheim stain . Analysis was performed with a Leica DM 2000 or with the Evos Fl Auto ( Invitrogen ) microscope . Primary murine distal lung cells contained >90% epithelial cells as determined by FACS . Cells were grown in 24-well plates ( Greiner ) or in chamber slides ( Nunc ) in DMEM enriched with HEPES , L-Glutamine , FCS , and pen/strep , until confluency and infected with PR/8 at the indicated MOI , as described previously [55] . PR/8 was diluted in PBS-/- containing BSA and pen/strep and added to the cells for 1 h , until the inoculum was removed and changed to infection medium ( DMEM supplemented with BSA , pen/strep , L-Glutamine and trypsin ) for further incubation . For inhibition of β-catenin , isolated distal lung epithelial cells of Rosa26ERTCre/ERTCre;Ctnnb1flox/flox mice were treated with 1 μM tamoxifen ( Sigma-Aldrich ) for 24 hours prior to PR/8 infection . Wildtype cells were treated with either 50 mM LiCl ( Abcam ) or 10 μM XAV939 ( Abcam ) directly after PR/8 infection , followed by RNA extraction or immunofluorescence staining . For infection of FACS-sorted lung cells ( AEC , EpiSPC , SAEC ) , the cells were counted , seeded in wells or infected in tubes for 1 hour with the given virus strain and MOI , and were either seeded in matrix , or further incubated at 37°C , 5% CO2 with EpiSPC medium and processed for FACS analysis at the given time point pi . RNA from sorted or cultivated cells was isolated using RNeasy Kit ( Qiagen ) according to manufacturer's manual . cDNA synthesis was performed , as described previously [56] or with RiboSPIA kit ( NuGen ) according to manual . Actin or ribosomal protein subunit S-18 ( RPS-18 ) expression served as normalization controls for the qRT-PCR , and the reactions were performed with SYBR green I ( Invitrogen ) in the AB Step one plus Detection System ( Applied Bioscience ) . The following intron spanning primers were used: Actin ( FP , 5′-ACCCTAAGGCCAACCGTGA-3′; RP , 5′-CAGAGGCATACAGGGACAGCA-3′ ) , Rps-18 ( FP , 5’- CCGCCATGTCTCTAGTGATCC-3′; RP , 5’- TTGGTGAGGTCGATGTCTGC-3′ ) , p63 ( FP , 5’-CAAAGAACGGCGATGGTACG-3′; RP 5’-CCTCTCACTGGTAGGTACAGC-3′ ) , Krt5 ( FP , 5’-CCTTCGAAACACCAAGCACG-3′; RP 5’-AGGTTGGCACACTGCTTCTT-3′ ) , β-tubb ( FP , 5’-CCACCACCATGCGGGAAA-3′; RP , 5’-CTGATGACCTCCCAGAACTTG-3′ ) , Fgf10 ( FP , 5’-CCATGAACAAGAAGGGGAAA-3′; RP 5’-CCATTGTGCTGCCAGTTAAA-3′ ) , Fgf7 ( FP , 5’- TCGCACCCAGTGGTACCTG-3′; RP , 5’- ACTGCCACGGTCCTGATTTC-3’ ) , Axin2 ( FP , 5’-AAGCCCCATAGTGCCCAAAG-3′; RP , 5’-GGGTCCTGGGTAAATGGGTG-3′ ) , Fgfr2b ( FP , 5’- AAGAGGACCAGGGATTGGCA-3′; RP , 5’- GTACGGTGCTCCTTCTGGTTC-3′ ) , Ctnnb1 ( FP , 5’- ACTTGCCACACGTGCAATTC-3′; RP , 5’-ATGGTGCGTACAATGGCAGA-3′ ) , Ccnd1 ( FP , 5’- GCGTACCCTGACACCAAT-3′; RP , 5’- GGTCTCCTCCGTCTTGAG-3′ ) , Pdpn ( FP , 5’-CCCCAATAGAGATAATGCAGGGG-3′; RP , 5’-GCCAATGGCTAACAAGACGC-3′ ) , Influenza Virus M segment ( 5’-GGACTGCAGCGTAGACGC-3′; 5’ CATCCTGTTGTATATGAG-3′; 5’-CATTCTGTTGTATATGAG-3′ ) . The relative gene abundance compared to the reference gene Actin or Rps-18 ) was calculated as dCt value ( Ctreference−Cttarget ) . Data are presented as dCT , ddCt ( dCtreference—dCttarget ) or fold change of gene expression ( 2ddCt ) . Animal experiments performed at the UGMLC were approved by the regional authorities of the State of Hesse ( Regierungspräsidium Giessen; reference numbers 100–2012 , 48–2013 , 26–2013 , 09–2009 ) and conducted according to the legal regulations of the German federal animal protection law ( Tierschutzgesetz ) . All data are given as mean ± SD . Statistical significance between 2 groups was estimated using the unpaired Student’s t test or ANOVA and post-hoc Tukey for comparison of 3 groups and calculated with GraphPadPrism . A p value less than 0 . 05 was considered significant . | Influenza Virus ( IV ) pneumonia causes disruption of the alveolar epithelial barrier , leading to edema formation which severely affects gas exchange . Recovery after IV-induced acute lung injury includes not only innate immune responses to clear virus infection , but also tissue renewal processes to restore the alveolar barrier . We demonstrate that highly pathogenic IV particularly infect cells of the epithelial stem/progenitor cell niche , which significantly impairs repair processes within the damaged lung , as demonstrated by transplantation experiments using intrapulmonary delivery of infected versus non-infected stem cells into IV-injured mice . Analyses of the underlying virus-host interactions reveal that IV infection of epithelial stem/progenitor cells results in blockade of a pathway involving Wnt/β-catenin and fibroblast growth factor receptor 2b ( Fgfr2b ) activation within a regenerative epithelial-mesenchymal cell signaling network . Our data therefore suggest that tropism of IV to the distal lung stem cell niche represents an important factor of pathogenicity . Therapeutic intrapulmonary application of excess Fgf10 to induce Fgfr2b signaling in the non-infected stem cell niche was found to counteract IV-induced repair failure and to restore barrier function , representing a promising treatment to foster epithelial cell regeneration after IV-induced injury . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"flow",
"cytometry",
"organoids",
"medicine",
"and",
"health",
"sciences",
"organ",
"transplantation",
"biological",
"cultures",
"cloning",
"surgical",
"and",
"invasive",
"medical",
"procedures",
"epithelial",
"cells",
"endocrine",
"physiology",
"developmental",
"biology"... | 2016 | Influenza Virus Infects Epithelial Stem/Progenitor Cells of the Distal Lung: Impact on Fgfr2b-Driven Epithelial Repair |
Optimal management of eumycetoma , a severely debilitating chronic progressive fungal infection of skin , disseminating to bone and viscera , remains challenging . Especially , optimal antifungal treatment and duration are ill defined . We conducted a monocentric retrospective study of 11 imported cases of eumycetoma treated by voriconazole or posaconazole for at least 6 months . Response to treatment was assessed through evolution of clinical and magnetic resonance imaging ( MRI ) . ( 1→3 ) ß-D-glucan ( BG ) and positron emission tomography using [18F] fluorodeoxyglucose ( PET/CT ) results were also assessed . Identified species were Fusarium solani complex ( n = 3 ) ; Madurella mycetomatis , ( n = 3 ) , and Exophiala jeanselmei , ( n = 1 ) . Moreover , two coelomycetes and one phaeohyphomycetes strains without species identification were retrieved . Serum BG and PET/CT were abnormal in 7/8 and 6/6 patients tested , respectively . Patients received last generation azoles for a mean duration of 25 . 9±18 months . Complete response ( major clinical and MRI improvement ) was observed in 5/11 patients , partial response ( minor MRI improvement or stable MRI findings ) in 5 and failure ( MRI evidence of disease progression ) in one , with a 73±39 [6–132] months mean follow-up . Relapse occurred in 2 patients after treatment discontinuation . Optimal outcome was associated with fungal species , initiation of last generation triazole therapy ( <65 months since first symptoms ) , negative serum BG and PET/CT normalization . MRI , PET/CT and serum BG appear as promising tools to assess optimal time of antifungal treatment for eumycetoma .
First described in 1642 by Kaempfer in his dissertation in the University of Leiden and then by John Gill as “Madura Foot” in 1842 , mycetoma is a chronic progressive and pseudotumoral granulomatous infection of skin , subcutaneous tissues and ultimately bone or viscera caused by fungi ( eumycetoma ) or bacteria ( actinomycetoma ) [1] . Young male adults of low socioeconomic status particularly manual workers in poor resource areas are the worst affected . Eumycetoma prevails in the belt that stretches between the 15th South and 30th North parallels , especially in Sudan and India , where drought could favors fungal growth in plant material including acacia and cow dung . Rural barefoot activities favor fungus transmission to human through subcutaneous contaminated thorn pick , which then spread locally or through the lymphatic system , and rarely through the bloodstream [2] , [3] . The eumycetoma clinical triad consists in painless subcutaneous mass , sinus formation and sero-purulent discharge that contains grains , aggregates of the fungal hyphae [4] . Among black grained-mycetoma , always of fungal origin , Madurella mycetomatis is the most prevalent causative agent [5] . Among white grained-mycetoma , Scedosporium boydii , Acremonium falciforme and Fusarium spp have been most frequently reported [5] . Eumycetoma is not self-healing and spontaneously leads to severely debilitating limb lesions with severe socioeconomic consequences [6] , making it currently one of the neglected tropical disease according to WHO [7] . Optimal management of eumycetoma remains ill defined and mostly relies on surgery and prolonged oral first generation antifungal treatment , when available . Indeed , surgery alone is associated with a 10–25% relapse rate [8] . In contrast , since 1984 , itraconazole , ketoconazole and terbinafine have been evaluated in limited open prospective trials ( i . e . maximal total number of patients recruited being 20 ) , in association or not with surgery , and disclosed a 70–80 % response rate , with an acceptable tolerance after up to 24 months of therapy and less than 10% relapse [9]–[13] . More recently , preliminary reports mainly from non-endemic countries using posaconazole and voriconazole have demonstrated in vitro activity and suggested encouraging in vivo efficacy [14]–[22] . In order to better assess last generation triazoles efficacy and their minimal treatment duration , we conducted a multicentric retrospective study from 2002 to 2013 of 11 adult patients with eumycetoma treated by voriconazole or posaconazole for at least 6 months with a follow-up of at least 6 months at the Centre d’Infectiologie Necker-Pasteur , Paris , France .
This study was conduced in compliance with the Institutional Review Board Paris Necker . In accordance with French law regarding retrospective studies , oral consent was obtained from each patient . Through the French National Reference Center for Invasive Mycoses and Antifungals ( NRCMA , Institut Pasteur , Paris ) , we retrospectively collected all adult cases of proven eumycetoma treated with voriconazole or posaconazole for a minimum period of 6 months and evaluated each of them at least once at the Centre d’Infectiologie Necker-Pasteur from January 2002 to December 2013 . All patients underwent magnetic resonance imaging ( MRI ) and/or positron emission tomography using [18F] fluorodeoxyglucose ( PET/CT ) before recent triazoles start and then every 6 to 12 months according to physician opinion and patient evolution . Moreover , ten CT scanner were performed among 4 patients . Six chest and abdominal CT scanner were performed in the patient 5 with lung involvement . A standardized form was used to collect information regarding age , sex , origin , place of contamination , medical background , immunodepression , date and localization of first symptoms , date and modalities of microbiological and/or pathological diagnosis , prior antifungal treatment and surgery , current triazole therapy , dosage regimen , route of administration , trough serum levels and side effects potentially attributable to one of the two tested triazoles . Excised grains were incubated at 30°C on liquid blood agar medium and Sabouraud glucose adding with chloramphenicol during at least 7 days . All isolates were identified by phenotypic methods ( macroscopic and microscopic aspect on Sabouraud , PDA and Malt extract media , growth at 37°C , determination of conidiogenesis by using slide culture on Malt extract agar for Exophiala sp . and Fusarium sp . isolates ) , and by sequencing of the ITS and D1/D2 regions of the gene coding ribosomal RNA by using universal primers ( V9D [23]/LS266 [24] and NL1/NL4 [25] primers respectively ) . Serum ( 1→3 ) ß -D-Glucan ( BG ) levels were determined with the Fungitell test kit ( Associates of Cape Cod , Inc . , Cape Cod , MA ) , according to the manufacturer's instructions . The results of a kinetic colorimetric assay performed at 37°C were read at 405 nm for 40 min . The BG concentrations in samples were calculated automatically by using a calibration curve established with standard solutions ranging from 6 . 25 to 100 pg/ml . This assay is reported continuously for results between 31–500 pg/mL , and as>500 pg/mL for values above this range . BG levels higher than 80 pg/ml were considered to be positive , as defined by the manufacturer . Serum assays were performed in duplicate A drug assay was performed for 26 samples among 8 patients using a previously published high-performance chromatography-UV detection method [26] . Immunodepression included history of diabetes , cancer , chronic renal disease , HIV infection , autoimmune disease or immunosuppressive therapy . Search for immune deficiency included anti-HIV antibodies , T , B and NK lymphocytes phenotyping and protein electrophoresis . Organ involvements were defined by abnormal MRI or surgical appearance compatible with eumycetoma . Osteitis was defined by T1 weighted hypointense and T2 weighted hyperintense signal of bone . Clinical response to treatment was assessed by a clinical score based on the presence of pain , inflammation signs and spontaneous drainage . Clinical responses were classified as “major” defined by a score equal to zero or “partial” defined by a decrease in clinical score . Stable clinical response and clinical failures were also considered . Biological response was assessed through BG serum levels . Major , minor , stable and worsened BG responses were defined by normalization , decreasing , unchanged or increasing values , respectively . MRI response , based on the Mycetoma Skin , Muscle , Bone Grading already reported [27] , was assessed by a single expert radiologist ( SP ) through comparison of site , size and contrast enhancement of main lesions . MRI response was notified as major , minor , stable or failure in case of pathologic hyper T2 signal complete disappearance , improvement by at least 50% , stability or in case of new lesions occurrence , respectively . The presence of a “Dot in the circle” pattern , i . e . conglomerate areas of small round discrete T2 weighted hyperintense lesions surrounded by a low-signal-intensity rim with central dot , highly suggestive of mycetoma [27] , [28] was also analyzed . PET/CT response relied on maximum Standard Uptake Value ( SUV ) comparison in a single nuclear medicine department ( CM ) . Major , minor , stable responses and failure were defined by negativation , improvement ( more than 30% decrease ) , no change ( less than 30% change ) or worsening ( more than 30% increase ) of max SUV values , respectively . End of treatment ( EOT ) time was defined as time of treatment discontinuation or last available evaluation . EOT response was defined as “complete response ( CR ) ” in case of clinical score negativation , and major MRI response , as “failure” if one of these parameters remained unchanged or had deteriorated or as “partial response ( PR ) ” otherwise . Risk for underdosing was defined by low posaconazole or voriconazole trough concentration ( <1 µg/mL ) or obvious non-adherence reported by the physician in charge of patient . Continuous data were described with descriptive statistics , including mean±SD and/or median [range] as appropriate and categorical data with frequencies ( % ) . Categorical data were analyzed by univariate analysis with Fisher's exact test as appropriate and continuous data by nonparametric Mann-Whitney test . Univariate analysis was used to identify factors associated with overall complete response at end of treatment ( EOT ) . P ≤ . 05 was considered statistically significant . Antifungal treatments which duration was shorter than six months were excluded from statistical analysis . Statistical analyses involved use of SPSS software .
Eleven cases of proven eumycetoma were identified during the study period ( Table 1 ) . Median age at the time of first symptoms was 28 . 8 [10 . 0–56 . 3] years . All patients were from African descent ( Senegal , n = 3 , Mali , n = 2 , Brazil , Martinique , Tchad , Mauritania , Togo , and Mayotte , n = 1 each ) , native from Western or Central Africa ( 8/11 ) . All but one patient had been likely contaminated in Western or Central Africa , and 4/11 reported a preceding trauma with thorn or stone . Initial sites of lesion were mainly foot or ankle ( 9/11 ) . Secondary skin and soft tissue infection was reported in 3/11 patients . Because of debilitating progression of the disease , 5/11 patients had to leave their job . At diagnosis , most of them ( 6/11 ) had cardiovascular risk factor , mainly hypertension . Two patients had chronic HBV infection and 4 patients type II diabetes or chronic kidney disease . HIV serology was negative , gammaglobulin serum levels and T , B , NK lymphocytes counts were normal in all patients . Consanguinity was present in 3/11 patients . Diagnosis was established in France in all but one patient with a median time of 58 [7–318] months since first symptoms . Grains , mostly of black color ( 8/11 ) were seen in every histopathological examination thereby confirming mycetoma . Microscopic examination and mycological culture were positive in 7/11 and 10/11 cases , respectively . Identification was possible to the species level in 7/10 cases , through exclusive phenotypic methods in 4 cases ( Fusarium solani complex , n = 2; Madurella mycetomatis , n = 2 ) and ITS 1/2 sequencing in 3 cases ( Madurella mycetomatis , n = 1; Fusarium solani complex , n = 1; Exophiala jeanselmei , n = 1 ) . In 4 cases , identification was only possible to the class level of Coelomycetes ( patient 7 and 11 ) , and Phaeohyphomycetes ( patient 1 and 3 ) . Coelomycetes class identification relied on ITS 1/2 sequencing . Phaeohyphomycete class identification relied on the presence of pigmented molds on microscopic histopathological examination , with negative cultures ( patient 3 ) and phenotype analysis of colonies with inconclusive ITS 1/2 sequencing results ( patient 1 ) . Before last generation azole therapy , 5 patients had already received antifungal treatments such as ketoconazole ( n = 4 ) , itraconazole ( n = 2 ) , fluconazole or terbinafine ( n = 1 , each ) ( see Table 2 ) . Primary clinical failure had occurred with fluconazole , or with shorter than 3 months treatment regimens . Eight patients had already undergone a surgical lesion resection that induced median clinical remission duration of 162 [19–288] months before relapse . At the time of initiation of last generation triazoles , all patients were symptomatic and had pain due to osteitis , local inflammation or purulent discharge ( see Table 2 and Figure 1 ) . MRI was abnormal in all 11 cases , showing soft tissue , osteolytic bone ( mainly talus , calcaneus and metatarsal ) , muscle , articular and visceral lesions ( lung , diaphragm and kidney ) in 11 , 8 , 5 , 4 and 1/11 cases respectively . A “dot in the circle” pattern was noticed in soft tissue in 7/11 cases ( picture 3A ) . Serum BG was tested in 8/11 cases and was positive in 7/8 cases ( median value 305 [80–500] pg/ml; normal<80 pg/ml ) ) . There was no correlation between BG value , lesion size evaluated by MRI or CT scanner nor SUV max . All 6 patients studied had abnormal PET/CT with a median SUV max of 6 . 6 [4 . 9–15 . 2] . Last generation triazoles were initiated as primary therapy ( n = 8 ) or as secondary therapy following failure from prior treatment ( n = 3 ) ( see Table 2 ) . 6/11 , 3/11 and 2/11 patients were treated with voriconazole ( 200 to 350 mg BID outside meals ) , posaconazole ( 400 mg BID with food ) or switch from posaconazole to voriconazole for failure , respectively . Three patients were treated with combination antifungal therapy including terbinafine ( n = 2 ) or flucytosine ( n = 1 ) . Additional small eumycetoma surgical excision was performed in 5/11 patients . At last evaluation , after a mean and median uninterrupted duration of 22 . 2±18 . 3 and 18 months respectively , treatment had been discontinued for completion or toxicity in 5/11 and 1/11 patients , respectively , or was ongoing in 5 cases . EOT response was complete , partial and null in 5/11 ( 45 . 4% ) , 5/11 ( 45 . 4% ) and 1/11 ( 9 . 1% ) patients , respectively ( see Table 2 and Figure 2 and 3 ) . Among partial responders , 3/5 had a negative clinical score and 2/5 still complained of pain . All 3 patients with negative clinical score had minor MRI improvement or stable MRI findings . The patient with failure had stable clinical score but MRI evidence of progression of mycetoma to lung and paraspinal region . Complete responders ( CR ) received at least 9 months of uninterrupted last generation azole treatment . Six patients whose triazoles treatment was discontinued had a mean follow up of 73±39 [6–132] months . Among them , relapse occurred in 2 patients , 8 and 11 months after treatment discontinuation , respectively . EOT major clinical , MRI , PET/CT and BG responses were achieved in 8/11 , 4/9 , 4/8 and 2/6 cases , respectively after a mean duration of triazole therapy of 25 . 9±18 . 0 months . In all patients , slight MRI contrast enhancement persisted in soft tissue at last evaluation . Side effects were reported in 3/11 patients treated with voriconazole and included chronic cholestatic hepatitis , transitory visual disturbance ( related to high voriconazole trough of 5 . 1 µg/mL ) and muscle pain with CPK elevation . Of note , none of the patients presented skin lesion , pain suggestive of fluorosis or significant drug-to-drug interaction . CR slightly differed from non-CR regarding patient's history , eumycetoma disease or treatment related characteristics . CR patients tended to be older at first symptoms ( median age of 40 . 6 years [10–56 . 3] and 22 . 5 years [18]–[31] , respectively , p = 0 . 13 ) and more often naive of antifungal treatment ( 5/5 ( 100% ) and 3/6 ( 50% ) patients , respectively , p = 0 . 38 ) . CR presented a trend towards earlier diagnosis of eumycetoma ( median time to diagnosis 36 [7–268] and 90 . 0 [36–318] months respectively , p = 0 . 18 ) , earlier last generation azole treatment ( median time to new azole treatment of 65 [9–290] and 267 [114–515] months respectively , p = 0 . 13 ) and more combination therapies ( 2/5 ( 40% ) and 1/6 ( 16 . 7% ) patients respectively , p = 0 . 38 ) . None of CR patients was infected with Fusarium solani or unidentified black fungi , whereas 5/6 ( 83 . 3% ) of non-CR were infected with one of those species or unidentified black fungi ( p<0 . 05 ) . Frequent lack of fructification of collected strains explained difficulties to obtain readable MIC values for all patients . MIC values were available in 5 patients , 3 with CR and 2 with PR . Patients with CR tend to have been infected with a fungus exhibiting lower triazole MIC values than those found in patients with PR or failure ( median and range MIC 0 . 125 µg/L [0 . 014–8] and 6 µg/L [4]–[8] , respectively , p = 0 . 4 ) Remembering the small population size described here , patients only treated with posaconazole had nevertheless a higher rate of complete response than that found in other patients ( 3/3 ( 100% ) vs 2/8 ( 25% ) respectively , p = 0 . 06 ) . CR received longer duration of uninterrupted last generation azole treatment ( 30±23 . 6 and 15 . 7±10 . 6 months , respectively , p = 0 . 18 ) . CR also had less azole under dosage risk factors ( 1/5 ( 20% ) and 4/6 ( 66 . 7% ) respectively , p = 0 . 24 ) . EOT BG and PET/CT major responses were associated with complete response at EOT ( p = 0 . 06 and p<0 . 05 respectively ) . All of the two relapses occurred in the 2 non-CR patients and consisted in local inflammatory signs reappearance occurring after a mean duration of triazoles therapy of 21±4 . 2 months .
Eumycetoma remains one of the most neglected infectious diseases around the world . In high-income countries , imported eumycetoma frequently presents as a severe and unknown disease , which optimal management by trained specialists should now relies on one of the last generation triazoles , posaconazole or voriconazole , that unfortunately aren’t available yet in endemic countries . Beyond essential basic pharmacokinetic patients explanations , optimal use of these drugs should include personalized dosage adapted to therapeutic drug monitoring results and at least a 9-month duration based on clinical , BG , MRI and PET/CT follow up assessment before discontinuation . | Eumycetoma is a severe chronic progressive fungal infection of skin and ultimately bone or viscera that affect mainly people with low economic status . Optimal treatment of this condition relies on medical and often surgical therapy and remains challenging because of lack of gold standard therapy and high rate of relapse after treatment discontinuation . In this retrospective study we assessed whether modern triazoles ( voriconazole and posaconazole ) suited to eumycetoma treatment and if tailored treatment duration , based on serial evaluation of a serum biomarker of fungal infection ( Serum ( 1→3 ) ß-D-Glucan , ( BD ) ) , magnetic resonance imaging ( MRI ) or positron emission tomography using [18F] fluorodeoxyglucose ( PET/CT ) would be useful . We found that modern triazoles were efficient treatments of eumycetoma , allowing complete or partial response in 10/11 of patients , without significant side effects . Moreover , patients with treatment discontinuation based on normalization of BD , MRI or PET/CT seemed to have better long-term outcome than those with clinical cure but still abnormal BD or imaging results . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"infectious",
"diseases",
"subcutaneous",
"mycoses",
"medicine",
"and",
"health",
"sciences",
"mycetoma",
"fungal",
"diseases",
"soft",
"tissue",
"infections"
] | 2014 | Last Generation Triazoles for Imported Eumycetoma in Eleven Consecutive Adults |
A common challenge in systems biology is quantifying the effects of unknown parameters and estimating parameter values from data . For many systems , this task is computationally intractable due to expensive model evaluations and large numbers of parameters . In this work , we investigate a new method for performing sensitivity analysis and parameter estimation of complex biological models using techniques from uncertainty quantification . The primary advance is a significant improvement in computational efficiency from the replacement of model simulation by evaluation of a polynomial surrogate model . We demonstrate the method on two models of mating in budding yeast: a smaller ODE model of the heterotrimeric G-protein cycle , and a larger spatial model of pheromone-induced cell polarization . A small number of model simulations are used to fit the polynomial surrogates , which are then used to calculate global parameter sensitivities . The surrogate models also allow rapid Bayesian inference of the parameters via Markov chain Monte Carlo ( MCMC ) by eliminating model simulations at each step . Application to the ODE model shows results consistent with published single-point estimates for the model and data , with the added benefit of calculating the correlations between pairs of parameters . On the larger PDE model , the surrogate models allowed convergence for the distribution of 15 parameters , which otherwise would have been computationally prohibitive using simulations at each MCMC step . We inferred parameter distributions that in certain cases peaked at values different from published values , and showed that a wide range of parameters would permit polarization in the model . Strikingly our results suggested different diffusion constants for active versus inactive Cdc42 to achieve good polarization , which is consistent with experimental observations in another yeast species S . pombe .
Mathematical models provide a more quantitative description of biological systems compared to qualitative arrow diagrams . A major tool of mathematical modeling is differential equations representing the dynamics of various components of the system which may be a cell , organism , or ecosystem [1–3] . In models describing cellular dynamics , the components are typically different protein species , and their changing levels depend on the biochemical reactions between the species [4 , 5] . If spatial dynamics , such as diffusion and advection , are present and considered , partial differential equation ( PDE ) models must be used; otherwise , the system can be modeled by ordinary differential equations ( ODE ) . One of the challenges in modeling is identifying the parameters from data [6 , 7] . For cellular models these parameters include the kinetic rate constants in the various reaction terms , as well as initial conditions of the modeled species . Two important tools of parameter analysis are sensitivity analysis and parameter estimation . Parameter sensitivity analysis is used to quantify the degree to which each parameter affects an output of interest . Two general types of sensitivity analysis , local and global sensitivity analysis , have been widely used . In local sensitivity analysis , sensitivities are evaluated at a single parameter set , whereas in global sensitivity analysis , sensitivities are evaluated across the entire parameter space . These analyses have been widely applied , such as to models in epidemiology [8–12] , signalling pathways [13–15] , physiology [16] , and wound healing [17] . For parameter estimation , two major approaches are Bayesian and maximum likelihood [7 , 18] . The primary difference between these two is that Bayesian methods infer a probability distribution for the parameters based on the available data , whereas maximum likelihood methods will provide a single-point estimate . In systems biology , parameter estimation has been widely applied , via both Bayesian inference [11 , 19–21] and optimization methods [22–25] . An important advantage of the Bayesian approach is a more explicit representation of the uncertainty in the parameter estimates; however , that usually comes at a price of high computational cost for estimating the distribution by extensive sampling . In general , global sensitivity analysis and parameter estimation both require sampling of the parameter space . For systems with large parameter counts , this can become very challenging due to the curse of dimensionality . Too many parameters can make sampling of the parameter space computationally intractable , especially for partial differential equation models that are expensive to solve . Many advances have been made in reducing computational cost in the field of uncertainty quantification ( UQ ) , which is concerned with the characterization and reduction of uncertainty in mathematical models [26–28] . Polynomial approximation has proven to be a key tool in uncertainty quantification . System outputs can be approximated by an orthogonal polynomial expansion ( polynomial chaos ) , allowing for straight-forward computation of statistical quantities [29] . In this paper , we apply a method for parameter sensitivity analysis and parameter estimation that uses polynomial approximation to significantly reduce the computational cost for large problems . A key step in the proposed method is the construction of a polynomial surrogate model . This surrogate model allows for sampling methods to be applied without the need to solve the full system for each sample . The use of surrogate models ( e . g . support vector machines ) for biological systems has been explored previously in [16] , and the use of polynomial approximations for uncertainty analysis has been investigated in [30 , 31] . To demonstrate the capability of the proposed method , we apply it to models of yeast cell polarization . Cell polarization is the process by which intracellular species ( e . g . proteins ) become asymmetrically localized , which is fundamental to cellular processes such as cell division , differentiation , and movement [32 , 33] . Failure in polarization can lead to cell death or dysfunction , and abnormal cell polarity is characteristic of cancer and may contribute to tumor initiation [34] . Cell polarization has been extensively studied in the budding yeast S . cerevisiae due to its pronounced polarity and genetic tractability [35 , 36] . The models analyzed in this paper describe polarization in response to pheromone during mating in budding yeast . We consider two models: an ODE model for only one module of the system ( the heterotrimeric G-protein cycle ) , and a spatial model that incorporates a larger signaling pathway as well as membrane diffusion of the proteins . We will refer to these models as Model 1 and Model 2 , respectively . Model 1 was proposed in [37] and has eight kinetic rate parameters , six of which have been experimentally measured or approximated from the literature . The remaining two parameters were estimated in [37] via an optimization method . This model is used to demonstrate the method and for comparison with the previous results . Model 2 is a mechanistic reaction-diffusion model , which is an extension of the model considered in [38] . This model has 35 unknown parameters . Parameter sensitivity analysis and parameter estimation have not previously been performed for this model , in part due to the large number of parameters . We seek to utilize polynomial surrogate models to quantify the effects of the parameters on polarization and to infer the biologically reasonable parameter values . It should be noted that the results of parameter sensitivity and parameter estimation are dependent on the assumed model structure . In systems biology there is often significant uncertainty in the model structure itself . Some work has been done on quantifying the structural uncertainty in models of biological networks and reconstructing networks from data [39–41] . However , this is beyond the scope of the present work and this source of uncertainty is not addressed in this paper . The structure of this paper is as follows . We first present the mathematical methods for surrogate model construction and how to perform parameter sensitivity analysis and parameter estimation using a polynomial surrogate . We then demonstrate the methods on Model 1 , performing sensitivity analysis and estimation in two cases: first , varying only the two free parameters , and second , varying all eight parameters . We then present Model 2 and use sensitivity analysis to significantly reduce the parameter count . Bayesian parameter estimation is then performed in the reduced parameter space . We discuss the computational savings afforded by the use of a polynomial surrogate for parameter estimation in Model 2 . Finally , we discuss biological implications of the results and future applications of the polynomial surrogates in Bayesian model analysis .
Biological systems often possess many parameters whose true values are unknown . In order to gain an understanding of the effects of each parameter , we need to sample the parameter space . However , sampling a high-dimensional space is a difficult task . For example , in the next section we consider a large PDE model with 35 parameters . In this case , even with only two sample points in each dimension we would need 235 ∼ O ( 1010 ) samples , and each sample requires solving a PDE system . This makes direct sampling of the PDE impractical . Instead , we may choose a scalar response function that quantifies an output of interest and by assuming that this response function depends smoothly on the parameters , a polynomial can be fit using far fewer sample points . Since we are performing parameter estimation , the response function depends not only on what quantity is of interest but also on what experimental data are available . If multiple response functions are of interest ( for example , different time points or different values of some input ) , there are two options—one can either increase the number of variables in the polynomial or use multiple polynomials . For example , if measurements are taken at several time points t1 , … , tk , then either t may be introduced as a variable of the polynomial or a polynomial Pi can be fit for each time point ( i = 1 , … , k ) . The choice can be made based on computational cost . If data are sparse , it is usually best to fit multiple polynomials , which is the approach taken in this work . Once the polynomial is established , we can use it as a surrogate for the full model so that sampling of the parameter space is far less expensive . To perform the polynomial fitting , we use an orthogonal polynomial basis from the generalized polynomial chaos ( gPC ) approach [26 , 29] . Thus the choice of basis for the polynomial space depends on the assumed probability distribution of the parameters . For the examples considered in this work , we assume that the parameters are independent and identically distributed , and uniformly distributed in a fixed range . This leads to the use of a Legendre polynomial basis . All parameters are mapped to a standard reference interval of [−1 , 1] . We do not consider any other distributions , but the same principles can be applied if the parameters have a Gaussian distribution ( Hermite polynomials ) , Gamma distribution ( Laguerre polynomials ) , or Beta distribution ( Jacobi polynomials ) . Recall that the number of basis functions for the set of polynomials of degree up to d in n variables is ( n + d n ) . The polynomial coefficients can be solved for in a number of ways , depending on the number of samples available . If the number of samples is exactly ( n + d n ) , then the coefficients can be solved for by direct interpolation . This case should generally be avoided as interpolation is notoriously prone to instability . If the number of samples is greater than ( n + d n ) , least squares approximation can be used . If the number of samples is less than ( n + d n ) , which is the case of interest for large problems , one may use compressed sensing methods to solve for the coefficients [42] . This approach has been well established for UQ problems [43 , 44] . The samples can be chosen in a variety of ways ( e . g . uniform random sampling , sparse grids , Latin hypercube sampling , etc . ) . A quasi-optimal sampling scheme for least squares polynomial fitting has been explored in [45] . In the applications presented here , we use uniform random sampling . Details of the polynomial fitting are presented in Algorithm 1 . Algorithm 1 Polynomial fitting algorithm . 1 . Determine the desired polynomial degree and how many samples can reasonably be obtained . 2 . Sample the parameter space using the sampling method of your choice . The sampling method may depend on whether you are undersampling or oversampling ( e . g . for oversampling , you may want to use quasi-optimal points for least squares [45] ) . 3 . Using the samples from step 2 , set up a linear system Ax = b where x is the vector of polynomial coefficients , A is a matrix whose entries are the basis polynomials evaluated at the sample points ( each row corresponds to one sample , each column corresponds to one basis polynomial ) , and b is a column vector of the model output at the sample points . 4 . Solve for the coefficients . If undersampling , perform compressed sensing with ℓ1-minimization . If oversampling , perform least-squares fitting . The accuracy of the polynomial can be estimated by cross-validation . In cross-validation , the model is evaluated at additional sample points that were not used in the polynomial fitting . The model output can then be compared with the polynomial value at those points to determine the error . One may also perform k-fold cross validation in which the total set of sample points is partitioned into k equally sized subsets; call them Ωi , i = 1 , … , k . Cross-validation is then performed k times . For each i , the samples in Ωi are used to evaluate the error and the remaining samples are used to fit the polynomial . The acceptable level of error will depend on the particular application . Once the polynomial surrogate model is constructed , it can be used to perform parameter sensitivity analysis and parameter estimation ( Fig 1 ) . Any sensitivity or estimation method can be applied using the polynomial surrogate model to decrease computational cost . In the work presented here , the methods are as follows . We define the sensitivity of a response function z ( p1 , … , pn ) to a parameter pj as S j = E ( ∂ z ∂ p j ) . We refer to Sj as the sensitivity coefficient for pj . Note that , while the partial derivative is typically used for local sensitivity analysis , the expectation makes this a global measure of sensitivity since ∂ z ∂ p j is integrated over the entire parameter space . Using the surrogate model , the parameter sensitivities can be analytically computed by taking partial derivates and evaluating S j = ∫ ∂ z ∂ p j d ρ , where ρ is the probability measure associated with the n-dimensional parameter space . We can then assess the importance of each parameter based on its sensitivity . If the response is not sensitive to a parameter pj , then the dynamics of the model will likely remain unchanged if pj is fixed . Further , pj may be non-identifiable so that multiple values can produce an equally good fit to data . Thus , we may use the sensitivity analysis to decrease the parameter count by fixing those parameters that have small sensitivity coefficients . For parameter estimation , we use Markov chain Monte Carlo ( MCMC ) method with Metropolis-Hastings algorithm [46] . MCMC is a method for sampling the posterior distribution of the parameters—that is , the parameter distribution that corresponds to the distribution of the provided data , given an assumed prior distribution . For the prior distribution , we use the parameter distribution that was assumed in the construction of the surrogate polynomial ( in this case , uniformly distributed within a range ) . This Bayesian approach to parameter estimation provides both the most probable parameter set ( or sets ) as well as a characterization of the parameter uncertainty . MCMC methods have become a popular choice for parameter estimation in biological systems [21 , 47 , 48] . However , these methods are often prohibitively expensive for computationally intensive models , since each sample in the Markov chain requires a model evaluation . By using the polynomial surrogate , the cost is greatly reduced . Further , it has been shown that in the generalized polynomial chaos framework , the polynomial fit and the resulting posterior distribution have similar convergence properties [49] . Thus , if the error in the polynomial fit is small , we expect the error in the posterior distribution to also be small . A key question is knowing when the MCMC has converged , meaning that the distribution of the Markov chain samples has converged to the posterior distribution . Several convergence diagnostics for MCMC have been proposed [50 , 51] . We employ a simple test which is to run multiple Markov chains from different initial parameter sets and compare the resulting distributions . Roughly speaking , if the independent chains stabilize at the same distribution , then the MCMC has converged . Since the chains are independent , they can be run in parallel to save computing time . We choose MCMC over alternative sampling methods [20 , 52 , 53] because of its efficiency . Since MCMC is based on a Markov chain , the samples tend toward higher probability areas of the parameter space in contrast to schemes that may sample the entire space . All codes have been made publicly available on GitHub in the repository https://github . com/chingshanchou/UQ-Yeast-Mating-Model . The yeast strain CGY-021 is a derivative of W303-1A and contains the bar1Δ mutation that prevents α-factor degradation by deletion of the Bar1 protease . GFP has been integrated genomically at the C-terminus of Ste20 to create a Ste20-GFP fusion protein that is a fluorescent reporter for active Cdc42 [54] . The genotype of the strain CGY-021 is MATa , can1-100 , ade2-1 , leu2-3 , -112 , his3-11 , -15 , trp1-1 , ura3-1 , bar1::hisG , ste20Δ::STE20-GFP-HIS5 . Cells were cultured in YPD ( yeast extract-peptone-dextrose ) media supplemented with adenine . Cells were treated for 60 minutes with 10 nM α-factor and then fixed with formaldehyde . Visualization was performed using a 60x objective ( NA = 1 . 4 ) on an Olympus Fluoview 1000 Spectral confocal microscope . The resulting images were analyzed in Matlab and the membrane fluorescent intensity was quantified over the periphery of the cell to generate the polarization profile that was averaged over 20 cells and converted into a polarization factor value .
To demonstrate our methods , we first consider a simple model: an ODE model of the heterotrimeric G-protein cycle taken from [37] . These equations represent the first stage of the system that senses the input ligand ( L ) α-factor: d [ R ] d t = - k R L [ L ] [ R ] + k R L m [ R L ] - k R d 0 [ R ] + k R s , ( 1 ) d [ R L ] d t = k R L [ L ] [ R ] - k R L m [ R L ] - k R d 1 [ R L ] , ( 2 ) d [ G ] d t = - k G a [ R L ] [ G ] + k G 1 [ G d ] [ G b g ] , ( 3 ) d [ G a ] d t = k G a [ R L ] [ G ] - k G d [ G a ] , ( 4 ) where the k’s are reaction rates . Here , [Gd] = Gt − [G] − [Ga] and [Gbg] = Gt − [G] , with Gt being the total number of G-protein molecules per cell . The model output is the fraction of free Gβγ ( Gbg/Gt ) , and the time unit is seconds . The model contains 9 parameters ( 8 rate constants and Gt ) , 7 of which were determined in [37] from experimental measurements and information from the literature . The remaining two parameters ( kGa and kGd ) were fit to data in [37] via least squares minimization . These parameter values are given in S1 Table . We focus first on this two-parameter problem , and use the proposed methods to corroborate the published parameter estimates . Later , we will allow all eight kinetic parameters to vary to determine if the same parameter estimates are obtained in the larger parameter space . In the 2-dimensional sensitivity analysis and parameter estimation , we will assume that the parameters kGa and kGd are log-uniformly distributed in the intervals [10−7 , 10−3] and [10−3 , 10] , respectively , which span the relevant ranges for the parameters . To capture the spatiotemporal dynamics of yeast cell polarization during mating , one needs a mechanistic spatial model . In this model , protein spatial dynamics are driven by two processes: surface diffusion on the cell membrane and reactions with other proteins in the system . This leads to a system of reaction-diffusion equations , similar to the model presented in [38] . The first six equations represent the dynamics of the heterotrimeric G-protein cycle , and the remaining equations represent the dynamics of the Cdc42 G-protein cycle . The distance unit is μm , the time unit is seconds , and concentration is measured as the number of molecules per unit surface area or volume ( except for the ligand L , which is measured in nM ) . ∂ [ R ] ∂ t = D R ∇ m 2 [ R ] - k R L [ L ] [ R ] + k R L m [ R L ] - k R d 0 [ R ] + p s k R s ( 5 ) ∂ [ R L ] ∂ t = D R L ∇ m 2 [ R L ] + k R L [ L ] [ R ] - k R L m [ R L ] - k R d 1 [ R L ] ( 6 ) ∂ [ G ] ∂ t = D G ∇ m 2 [ G ] - k G a [ R L ] [ G ] + k G 1 [ G d ] [ G b g ] ( 7 ) ∂ [ G a ] ∂ t = D G a ∇ m 2 [ G a ] + k G a [ R L ] [ G ] - k G d [ G a ] ( 8 ) ∂ [ G b g ] ∂ t = D G b g ∇ m 2 [ G b g ] + k G a [ R L ] [ G ] - k G 1 [ G d ] [ G b g ] ( 9 ) ∂ [ G d ] ∂ t = D G d ∇ m 2 [ G d ] + k G d [ G a ] - k G 1 [ G d ] [ G b g ] ( 10 ) ∂[ C24m ]∂t=DC24m∇m2[ C24m ]+k24cm0 ( Gbgn* ) [ C24c ]+k24cm1 ( B1* ) [ C24c ]−k24mc[ C24m ]−k24d[ Cla4a ][ C24m ] ( 11 ) ∂ [ C 42 ] ∂ t = D C 42 ∇ m 2 [ C 42 ] - k 42 a [ C 24 m ] [ C 42 ] + k 42 d [ C 42 a ] ( 12 ) ∂ [ C 42 a ] ∂ t = D C 42 a ∇ m 2 [ C 42 a ] + k 42 a [ C 24 m ] [ C 42 ] - k 42 d [ C 42 a ] ( 13 ) ∂ [ B 1 m ] ∂ t = D B 1 m ∇ m 2 [ B 1 m ] + k B 1 c m [ C 42 a ] [ B 1 c ] - k B 1 m c [ B 1 m ] ( 14 ) ∂ [ C l a 4 a ] ∂ t = k C l a 4 a ( C 42 a t * ) - k C l a 4 d [ C l a 4 a ] . ( 15 ) The coefficients are given by B 1 * = B 1 t * 1 + ( γ G b g n * [ B 1 m ] ) - h , B 1 t * = ∫ S [ B 1 m ] d s S A , γ = S A 2 ∫ S [ B 1 m ] d s , G b g n * = 1 1 + ( δ ( G b g n ) ) - q , δ = S A ∫ S ( G b g n ) d s , ( G b g n ) = [ G b g ] [ G ] 0 , C 42 a t * = ∫ S [ C 42 a ] d s S A , p s = [ C 42 a ] C 42 a t * if C 42 a t * > 0 , else p s = 1 , where SA is the surface area of the cell . The initial conditions are given by [ R ] 0 = R t / S A , where R t is the total amount of R , [ G ] 0 = G t / S A , where G t is the total amount of G , [ C 42 ] 0 = C 42 t / S A , where C 42 t is the total amount of C 42 , [ R L ] 0 = 0 , [ G a ] 0 = 0 , [ C 24 m ] 0 = 0 , [ C 42 a ] 0 = 0 , [ B 1 m ] 0 = 0 . [ G d ] = [ G ] 0 - [ G ] - [ G a ] , [ G b g ] = [ G ] 0 - [ G ] . The conservation equations are V · [ C 24 c ] = C 24 t - ∫ S [ C 24 m ] d s , V · [ B 1 c ] = B 1 t - ∫ S [ B 1 m ] d s , where C24t and B1t are the total amounts of C24 and B1 respectively , V is the volume of the cell , and [C24c] and [B1c] are the concentrations of C24 and B1 , respectively , in the cytoplasm . Thus the total amounts of Bem1 and Cdc24 are conserved . Estimates from previous work and ranges for the parameters are given in Table 3 . In our numerical simulations , the cell membrane is simulated as a circle centered at the origin with radius 2 μm . The pheromone input is administered as a gradient from the positive x-direction with midpoint of 10 nM and slope of 0 . 1 nM/μm . The surface diffusion of a quantity W on a circle is given by ∇ m 2 W = W s s where s is an arc length parameter , ds2 = dx2 + dy2 . The computational domain is parametrized by α ∈ [0 , 2π] , where α denotes the angle from the negative x-axis . The numerical method utilizes a second order finite difference discretization for the spatial derivatives and an implicit Crank-Nicolson method for the time derivative . The spatial mesh consists of 400 equally spaced points . Each simulation is run to steady state ( t = 1 , 000s ) . More detail about the numerical method can be found in the Supplementary Material ( S2 Text ) .
Typically , MCMC requires a model evaluation at every iteration . Since our Markov chain length for the 15-parameter model was 2 × 106 , we would require 2 × 106 evaluations of the PDE model to steady state . It would likely take even more iterations for the MCMC to converge for the full 35-parameter model . The PDE is solved with an implicit method implemented in Fortran , and each evaluation takes 40-60 minutes of CPU time . Thus , the full MCMC would require at least ∼200 years of CPU time . Further , MCMC is not inherently parallelizable , although advancements have been made in parallel MCMC methods [63–66] . Using the polynomial surrogate , we are able to practically eliminate the cost of MCMC by evaluating only a polynomial at each MCMC iteration . Computing a chain of length of 2 × 106 takes only a few hours in MATLAB . In place of this cost , we must evaluate the full PDE model at the sample points used to fit the polynomial . For our full 35-parameter model , we use 5000 sample points to fit a polynomial to perform the sensitivity analysis . We then are able to reduce the parameter count to 15 , and use 6000 additional samples to fit a polynomial in the reduced parameter space . Thus we require 11 , 000 model evaluations in total . There is also some cost to fit the polynomial via ℓ1-minimization , which is of the order of several hours . The time required to evaluate the polynomial is considered to be negligible compared to the time required to solve the PDE . Thus we have a roughly 180-fold reduction in computational cost compared to the MCMC without a polynomial surrogate . In addition , the samples are independent so that the model evaluations to produce these samples can easily be computed in parallel . The computational savings in the ODE test model are not as dramatic , since the ODE model is inexpensive to solve . In numerical tests for the 2-parameter ODE model with a 10th degree polynomial surrogate , we found a 20% reduction in CPU time in evaluating the polynomial vs . evaluating the model directly . In the 8-parameter model with a 5th degree polynomial surrogate , we found a more than 10-fold reduction in CPU time; we believe the greater reduction in cost is afforded by the lower polynomial degree . The computational savings afforded by using polynomial surrogates will vary depending on the ODE solver , the degree of the polynomials , and the time step required to solve the ODE . Whether a problem warrants the use of surrogate models will generally depend on the cost of evaluating the original model , the number of sample ( data ) points required for accurate parameter estimation , and the polynomial degree required to fit the model output . The primary challenge with the method is constructing accurate polynomials . As we demonstrate in the ODE example , more sample points and a higher degree polynomial produce greater accuracy . One concern is the ability of the surrogate polynomials to describe highly nonlinear relationships between parameters and outputs arising from bifurcations . If the model output is discontinuous with respect to the parameters , for example , then the model output will not be well-approximated by polynomials . This issue may exist in the PDE model presented here , since it has previously been shown that the model for some parameter values possesses multistability contributing to the polarization [60]; thus the steady state behavior is discontinuous with respect to the initial conditions . The 5th degree polynomial surrogate produces an error of 0 . 1 to 0 . 2 . It is likely we can reduce the error by employing more sample points or by using a higher degree polynomial . Alternatively , one can take advantage of Design of Experiments methods [67] to pick more informative sample points to decrease the error . However , we will still not be able to capture the discontinuous nature of the model output . Another issue is that one may make false assumptions in determining a response function . In the PDE model we choose a response function that quantifies the cell polarization at steady state , and thus we are assuming that the system settles to a steady state . While this seems to be a reasonable assumption for the system presented here , this may not always be the case . If a system has periodic solutions rather than a stable steady state in some region of the parameter space , then one would need to carefully consider how to build an appropriate response function . Unfortunately , it is not always clear a priori whether such solutions exist for a given system . Finally , a third issue is the combinatorial increase in the number of polynomial coefficients as the number of parameters increases . The 5th degree polynomial for the 35 parameter model possesses 658 , 008 coefficients and a 100 parameter 5th degree polynomial would possess over 75 million coefficients . For the PDE model we employ compressed sensing methods ( ℓ1-minimization ) that allow undersampling to fit higher dimensional polynomials from larger models with fewer sample points . It is possible to adopt advanced sparse regression methods such as ℓ1-ℓ2 minimization [68] to further reduce the number of required model evaluations . A second approach is explore optimal sample set design such as the optimal sample selection strategy [45] that , for any given number of samples ( model evaluations ) , finds the parameter sample points to provide a polynomial surrogate nearly as accurate as the one obtained by a much larger number of model evaluations . In the yeast G-protein ODE model , the parameter distributions inferred from the time-course and dose-response data are consistent with the parameter estimates and experimental measurements from [37] . For example , the peaks for kGa and kGd are very close to the previous maximum likelihood estimates . Interestingly , the parameter estimates for kRL and kRLm are close to the measured values found in [37] , but they are at least one to two orders of magnitude larger than the estimated values from three other groups [69–71] . One possible explanation is the use of fluorescent analogs of α-factor in some of the earlier work , but this discrepancy needs to be addressed in future work . The PDE model shows broad distributions for nearly all 15 parameters examined indicating that a wide range of parameter values are compatible with good polarization of active Cdc42 . The fact that the feasible region of the parameter q , representing the cooperativity of the interaction between Gβγ and Cdc24 , spans the full range from 1 to 100 demonstrates that the high value ( q = 100 ) previously used in the model [38] is not necessary for polarization , and that lower values ( e . g . q = 1 to 10 ) are almost equally probable . These lower cooperativity values corresponding to smaller Hill exponents are more plausible from a mechanistic standpoint . In addition , several parameters ( k24cm1 , k42a , k42d , Dc42 , and Dc42a ) show peaks at one or the other side of the distribution indicating that the previous estimates may miss the most likely parameter range . The diffusion constants Dc42 and Dc42a were assigned the same value in our previous model [38 , 61] , but in this work Dc42 shows a preference for higher values , whereas Dc42a shows a preference for lower values . Recent measurements by Bendezú et al . [72] in the fission yeast S . pombe found that inactive Cdc42 had a 10-fold faster diffusion rate than active Cdc42 consistent with the trends in our parameter distributions . This work also highlights the inability of the current PDE model to produce the sharp polarization peak of active Cdc42 observed in cells . One explanation is that the model is missing important dynamics or positive feedback mechanisms that enhance cell polarization . In the future , we plan to include additional spatial dynamics such as the polarized transport of Cdc42 to the front of the projection , which is absent from the model . The broadness of the obtained parameter distributions also implies that the current data are insufficient to obtain tight parameter estimates . In this study we focused on identifying parameter values that would produce polarization in the model versus an unpolarized state . Further data can be collected tracking the spatial dynamics of the other species in the model such as Gβγ , Cdc24 , and Bem1 in both wild-type and mutant yeast strains . The additional data along with model modifications should result in narrower parameter distributions and a better fit to the total system dynamics . In our analysis , we presented only the sensitivity measure S j = E ( ∂ z ∂ p j ) . The advantage of this sensitivity measure is its simplicity; it is easy to compute analytically when z is a known function and it usually provides a good measure of sensitivity when the relationship between z and pj is monotonic . However , if the relationship between z and pj is non-monotonic or highly nonlinear , Sj may not be a desirable measure . In these cases , other measures of sensitivity may be a better choice such as variance-based sensitivity measures or the partial rank correlation coefficient [8 , 73] . The use of polynomial chaos expansions to approximate variance-based sensitivities has been explored previously in [31] . Other derivative-based sensitivity measures have also been proposed [74] , which can be computed analytically using the polynomial chaos expansion . Polynomial surrogates may also be used in methods for parameter estimation not addressed in this paper . In principle , polynomial surrogates can be applied to any type of model for which parameter ranges are known , and for any sampling-based method that requires model evaluations . By fitting polynomials to the quantities for which data is available , every model evaluation in a computational method can be replaced by a polynomial evaluation . While we have demonstrated this here only in the context of a Markov chain Monte Carlo method , the same principles may be used to accelerate the computations involved in other Bayesian methods for parameter estimation , such as rejection sampling and sequential Monte Carlo . Yet another potential application of polynomial surrogates is to accelerate methods for Bayesian model selection . The idea behind Bayesian model selection is that we can recover a probability distribution for a model index parameter m enumerating different models , providing information on the likelihoods of the candidate models given the available data . In essence this is still a parameter estimation problem , and established methods for parameter estimation can be adapted for model selection . Polynomial surrogates can be used to accelerate these methods which include Bayesian rejection sampling , sequential Monte Carlo , population annealing , and MCMC [20 , 52 , 53 , 75 , 76] . Model selection is of great importance in systems biology since uncertainty in the model structure may significantly impact the conclusions of parameter inference [39] . | Mathematical models in systems biology often have many parameters , such as biochemical reaction rates , whose true values are unknown . When the number of parameters is large , it becomes computationally difficult to analyze their effects and to estimate parameter values from experimental data . This is especially challenging when the model is expensive to evaluate , which is the case for large spatial models . In this paper , we introduce a methodology for using surrogate models to drastically reduce the cost of parameter analysis in such models . By using a polynomial approximation to the full mathematical model , parameter sensitivity analysis and parameter estimation can be performed without the need for a large number of model evaluations . We explore the application of this methodology to two models for yeast mating polarization . A simpler non-spatial model is used to demonstrate the techniques and compare with published results , and a larger spatial model is used to demonstrate the computational savings offered by this method . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"markov",
"models",
"simulation",
"and",
"modeling",
"fungi",
"model",
"organisms",
"probability",
"distribution",
"mathematics",
"statistics",
"(mathematics)",
"algebra",
"experimental",
"organism",
"systems",
"cellular",
"structures",
"and",
"organelles",
"polynomials",
... | 2018 | Parameter uncertainty quantification using surrogate models applied to a spatial model of yeast mating polarization |
Given current neglect for Chagas disease in public health programs in Mexico , future healthcare and economic development policies will need a more robust model to analyze costs and impacts of timely clinical attention of infected populations . A Markov decision model was constructed to simulate the natural history of a Chagas disease cohort in Mexico and to project the associated short and long-term clinical outcomes and corresponding costs . The lifetime cost for a timely diagnosed and treated Chagas disease patient is US$ 10 , 160 , while the cost for an undiagnosed individual is US$ 11 , 877 . The cost of a diagnosed and treated case increases 24-fold from early acute to indeterminate stage . The major cost component for lifetime cost was working days lost , between 44% and 75% , depending on the program scenario for timely diagnosis and treatment . In the long term , it is cheaper to diagnose and treat chagasic patients early , instead of doing nothing . This finding by itself argues for the need to shift current policy , in order to prioritize and attend this neglected disease for the benefit of social and economic development , which implies including treatment drugs in the national formularies . Present results are even more relevant , if one considers that timely diagnosis and treatment can arrest clinical progression and enhance a chronic patient's quality of life .
Chagas Disease ( CD ) is caused by the flagellated protozoan parasite Trypanosoma cruzi ( T . cruzi ) [1] , vectored by triatomine insects known as kissing bugs . The parasite is transmitted most often via the bug's feces , and to a much lesser extent via blood transfusion , congenital or alimentary transmission , and organ transplant or laboratory accident [2] , [3] . The disease is endemic in 21 Latin-American countries and the United States , although human migration has expanded at-risk populations for most transmission modes in previously considered non-endemic countries [4] . In Mexico , more than 71 , 000 , 000 inhabitants are at direct risk in both rural and urban areas for vector transmission from one of 18 vector species [5] , [6] . The current prevalence is not well documented , although most estimates suggest between 0 . 013%–3 . 12% of the Mexican population are seropositive [7] , [8] and 650 , 000 chronic cases are currently in some form of clinical care in one of the many health care systems [9] . The first National Seroepidemiology Survey in México , found a 1 . 6% seroprevalence of antibodies to T . cruzi ( 66 , 678 samples tested ) at the national level . The highest prevalence was found in Chiapas ( 5 . 0% ) , Oaxaca ( 4 . 5% ) and the south-east region , followed by the central plains of the temperate Huasteca region , which includes the states of Hidalgo ( 3 . 2% ) , San Luis Potosí ( 2 . 5% ) , Veracruz ( 3 . 0% ) and Tamaulipas ( 1 . 6% ) . However , a limitation of that study was its poor coverage of rural areas , which may have led to a significant underestimate of the current prevalence of the infection and disease [10] . Blood transfusion risk also exists , the review of 64 , 969 blood donors in 18 states of Mexico , demonstrated a 1 . 5% seropositivity , with prevalence ranging from 0 . 2% in Chihuahua up to 2 . 8% in Hidalgo . A more recent study of blood donations in the Social Security system ( IMSS ) , highlights a similar profile and suggests that in urban populations , 0 . 4% are seropositive [7] . About 2000 inhabitants each year could be at risk of infection with T . cruzi via blood transfusion [11] . Analysis of the economics surrounding a disease can generate information essential for decision-making and evidence-based adoption of specific prevention and control policies . This is particularly useful for health sector authorities in order to generate greater social benefit with a lower cost to the health system [12] . It is also fundamental for creative programming and financing of prevention and control strategies in the face of economic crises and in relation to social and economic development . Direct medical costs to the health system for support therapy for chronic CD cases are remarkable and clinical interventions in the chronic phase raises the costs because it consists of specialized medical care such as palliative and corrective cardiac and digestive surgery [13] . If we consider the indirect costs due to loss of productivity , the burden of CD increases due its impact on individuals in their most productive years [14] . According to the first WHO Report on Neglected Tropical Disease ( CD ) , in Latin America 752 , 000 working days per year were estimated to be lost due to premature deaths due to CD . The economic cost of CD in terms of lost productivity was estimated at US$ 1 . 2 billion each year for the seven countries of the Southern Cone . In Brazil , worker absentee affected by CD represents an estimated minimum loss of US$ 5 . 6 million per year [15] . In Mexico , there is only one published study that estimates the cost of CD treatment; the calculations were based on 13 clinical records at a tertiary level hospital , and hence cost estimates cannot be extrapolated for the entire country or for all healthcare systems [16] . The present study aims to estimate the current costs of treating a chronic CD case detected and treated early vs an undetected case among the salaried population ( 47% of the Mexican population [17] ) , and the direct and indirect costs and effects simulated since birth to death using a cohort Markov model .
We constructed a Markov decision model based on previous publications [18] , [19] , to simulate the natural history of a CD cohort and to project the associated short and long-term clinical outcomes . Professional software was used to construct the model ( TreeAge Software , Williamstown , Massachusetts ) . Most recently , several published and ongoing studies have demonstrated that having negative serology after treatment is not a guarantee for remaining seronegative over time [20] , [21] . However , given the current lack of evidence validating seroconversion with parasitological clearance and therapeutic cure , we refer in this study to the endpoint for treatment as “no progression” . Figure 1 illustrates the general model structure including the following five Markov states of the disease and an individual's possible transition between states . All clinically important events are modeled as transitions from one state to another using a transition probability [22] . Each cycle length is one month in the acute phase and then it switches to a year for the rest of disease phases . All Markov states are mutually exclusive . Transition can occur from one state to another during each cycle ( Figure 1 ) . Patients are absorbed into the death state , where they remain , not being allowed to transition to another state . The simulation is run until the entire cohort dies . We compared three detection and treatment scenarios: ( 1 ) 100% individuals are detected and treated early ( who are diagnosed and treated during the acute phase of the disease ) , ( 2 ) 100% of individuals are detected but only 80% are given treatment ( the latter scenario was developed to include patient refusal and/or consideration for those patients not clinically capable of treatment for concomitant health reasons ) , and ( 3 ) no one is diagnosed or treated . The comparative performance was assessed by summing direct costs for medical treatment and indirect costs . We used a modified social perspective , in that costs of patients' time and travel were not included . Future costs were discounted at 5% per year . The discount rate is a financial adjustment which is applied to determine the present value of a future payment and differs from the rate of interest , in that it applies to the original amount for the increase . A second order Monte Carlo simulation was used in which disease progression in an individual is characterized as a sequence of transitions between health states . One million patients were simulated , one at a time , in order to provide stable estimates of long-term outcomes for each strategy . All the parameters used to feed the model were introduced as statistical distributions: costs inputs are set as gamma distributions and probabilities of transition are beta distributed . Because in a second order calculation all these distributions are sampled , no sensitivity analysis is necessary . Baseline estimates for selected variables were developed from information provided from published studies and an expert panel ( Table 1 to Table 3 ) . Given the lack of information regarding medical care consumption by CD patients in Mexico ( direct costs ) , we consulted an expert panel of four experts . All of them with at least five years of experience in CD in Mexico . The first expert is a physician with experience in CD patient care in the state of Morelos , at the moment we consulted him he was vector control manager , which includes CD disease . The second expert is a health researcher and physician with experience in CD patient care in the state of Jalisco . The third expert is a health researcher and epidemiologist in the state of Jalisco , her main research line is CD . The fourth expert is a physician with experience in CD patient care in the state of Veracruz . All participants were sent instructions with a set of three ( one per CD phase ) forms to complete . Medical care and procedures for all phases were obtained by a panel of three clinical experts ( none of the experts are also authors of this manuscript ) . Costs include both direct and indirect costs . Direct medical costs for CD include those for hospitalization , outpatient consultations , laboratory tests , annual screening , clinical procedures , and medications . Undiagnosed patients have the national average medical care consumption . If a patient develops meningoencephalitis , myocarditis and/or megasyndromes ( megaeshophagus , megacolon ) , medical attention is calculated for these specific symptoms . Average medical care consumption was estimated using the National Survey of Health and Nutrition 2012 ( ENSANUT for its acronym in Spanish ) [23] . Costs of medical care were built using pricing from the Mexican Social Security Institute ( IMSS ) [24] . Total costs were obtained by multiplying the quantity of services consumed by the unit costs . The indirect costs , were calculated considering the IMSS average daily wage ( US$ 18 . 9 ) and the average working days lost due to illness [25] . The initial age of the cohort is 10 year old and we assume that children do not have remunerated work , so we did not consider working days lost due CD in the acute phase . All costs are expressed as 2012 value of the US dollar or after foreign currency conversion , using average annual exchange rates provided from the International Monetary Fund [26] . Once converted into US dollar , costs were adjusted for inflation using the US Consumer Price Index [27] . The average cost per patient for each phase of the disease was calculated considering the entire cohort of patients , regardless of the phase of the disease that the patient had reached . Subsequently , we calculated the average cost per patient for each phase of the disease , and considered only patients who achieved the phase .
The average cost per patient , considering the entire cohort of patients , for each of the scenarios is included in Table 4 . In the acute phase , the greatest cost per patient occurs for those in the 80% diagnosed and treated scenario ( US$ 234 ) , with little difference compared to the scenario where 100% of patients are diagnosed and treated ( US$ 232 ) . The category costs for timely diagnosis and treatment in the former group is US$ 31 for medical counseling , US$ 83 for hospitalization , US$ 66 for laboratory tests ( blood chemistry , urine test , complete blood count , urea , creatinine , indirect haemagglutination test , etc ) , US$3 for radiology and imaging , and US$ 51 for drugs ( benznidazole for CD treatment and other drugs ) . In the indeterminate phase of the disease , the timely diagnosis and treatment in 80% of patients generates an average US$ 6 , 505 cost per patient , while the timely diagnosis and treatment of 100% of patients generates an average of US$ 5 , 641 . The average cost per patient not receiving diagnosis or treatment is US$ 3 , 309 , since they do not have CD specific medical care . The most expensive chronic phase scenario occurs due to undiagnosed patients , US$ 8 , 449 , which includes working days lost . The cost per patient for the diagnosis and treatment of 80% of patients is US$ 4 , 819 and the least expensive scenario is where all patients are diagnosed and treated ( US$ 4 , 287 ) . If all costs per patient are compared among the three program scenarios , early diagnosis and treatment of 100% of CD cases results in a lifetime costs US$ 10 , 160 ( US$ 232+US$ 5 , 641+US$ 4 , 287 ) . The lifetime cost per patient if only 80% are diagnosed and treated is US$ 11 , 558 ( US$ 234+US$ 6 , 505+US$ 4 , 819 ) , and if no patient is diagnosed or treated , the cost is US$ 11 , 877 ( US$ 120+US$ 3 , 309+US$ 8 , 449 ) . The major cost components for the 100% and 80% scenarios are working days lost ( 44% ) , followed by hospitalization ( 23% ) , drugs ( 15% ) , laboratory ( 14% ) , and medical counseling ( 3% ) . However , for the undiagnosed scenario , the major cost component of working days lost rises to 75% , followed by hospitalization ( 12% ) , drugs ( 6% ) , laboratory test and diagnosis ( 6% ) , and medical counseling ( 1% ) . The cost of a diagnosed and treated case increases 24-fold from early acute to indeterminate stage ( 100% scenario ) . The cost per patient in the indeterminate stage is 1 . 32 fold , more than the cost in the chronic stage ( 100% scenario ) . The costs for the undiagnosed patient scenario are systematically lower than either of the 100% treated for acute and indeterminate phases ( 1 . 93 and 1 . 70 times , respectively ) , due to treatment-specific costs . However , in the chronic phase , the undiagnosed patient scenario incurs most costs , being 1 . 97 times greater than in the 100% treatment scenario . The phase specific cost per patient per year is summarized in Table 5 . While the results for the acute phase are the same as shown in the previous table , in the indeterminate phase , the average cost per patient is greater for the 80% diagnosed and treated early alternative ( US$ 12 , 772 ) . In the chronic phase , the average cost is greater for the alternative where 100% of patients are diagnosed and treated , with a total of US$ 24 , 588 .
The costs of a chronic CD case detected and treated vs an undetected case has been analyzed herein from a modified social perspective which allows us to take into account the value of working days lost . We calculated two different types of costs: costs per patient per lifetime according to disease stage ( or cohort cost , discounted at a rate of 5% ) and costs per patient per year . The lowest lifetime cost is estimated from the 100% early diagnosis and treatment scenario , due to the fact that in this scenario , less of the cohort reaches the expensive chronic phase of the illness . It is important to stress that the parallel costs between the 100% and the 80% treated scenarios is because both populations are diagnosed and have similar medical management . The phase specific costs increase accordingly with each progressive phase for all scenarios , although the cost estimated for the undiagnosed category is less in chronic phase than that for either 80% or 100% scenarios ( US$ 16 , 630 vs . US$ 23 , 929 or US$ 24 , 588 , respectively ) . Although surprising , this result is considered real , since there is high mortality in the undiagnosed patient group , and since costs are calculated for a complete cohort , they will be proportionally reduced due to patients lost to the cohort ( and hence reduced average patient cost ) . Vallejo et al . [16] reported the cost of medical treatment of 13 clinical cases for CD in a specialized hospital setting ( third level ) in Mexico . The annual cost for medical care for patients in this outpatient setting was estimated between US$ 4 , 463 and US$ 9 , 601 , and annual costs for patients admitted via an emergency care unit was between US$ 6 , 700 and US$ 11 , 838 . If we assume that these patients were in the chronic phase of the disease , the costs we calculated herein are similar to their higher costs . Contrary to our findings , Vallejo et al . conclude that highest cost components were radiology and imaging ( 63% ) and hospitalization ( 26% ) , while the component with least contribution to cost was drug treatment ( 3% ) . The limitations of the previous study are the few patients used to estimate costs , the costs of surgeries ( pacemaker placement were not considered ) , and the bias for disease phase , since in order to be attended in a specialized hospital , patients must have economic capacity ( to afford out-of-pocket expenses to go to Mexico City for varying periods ) , and must be in the chronic phase with cardiomyopathy . The present study uses second and third level social security ( IMSS ) costs as an alternative and complementary perspective for opportunity costs of diagnosis and treatment , since the IMSS system currently covers 47% of the Mexican population , and in the poorest states ( Chiapas , Campeche , Yucatan ) , the IMSS Oportunidades subsystem still covers at least half of the rural population [17] . Future studies should also focus on the Seguro Popular and the primary healthcare system , Secretariate's second level hospitals and costs generated in these systems . Until there is a more robust estimate of patient population seroprevalence in all Mexican healthcare groups and regions , the costs calculated herein may be considered pertinent particularly for populations with fixed incomes . In other countries such as Colombia , chronic CD cost has been calculated from the payer's perspective ( review of 63 clinical records ) [28] . Castillo-Riquelme et al . concluded that cost per patient per year for clinical management at the basic care level was US$ 46 to US$ 51 , the cost in the intermediate level of care was US$ 188 to US$ 259 , and in a specialized setting the cost per patient per year was US$ 3 , 652 to US$ 7 , 981 . If we compare the costs per patient per year of the acute phase of the 100% and 80% scenarios they are similar to those reported by Castillo-Riquelme et al . for their intermediate level of care . The costs for chronic phase reported in the present study is between 3 and 6 . 7 times greater than costs reported by Castillo-Riquelme et al . The difference may be due to the methodological perspective which in the latter was based on the provider and the present study based a modified social perspective . The value for worked days lost was 91 . 4% of the total costs in the 100% scenario in the chronic phase in the present study . Castillo-Riquelme et al estimated for the intermediate level of care , hospitalization contributes between 25% and 49% , drugs contribute 31% to 42% and for specialized care , the surgical procedures were the largest cost component , contributing between 41% and 55% , while the second largest was drugs ( 10% and 24% ) . The distribution of the total costs reported by the present study are similar to those reported by Castillo-Riquelme et al . except for the fact that the major cost component in the present study was the value of working days lost ( 44% ) , followed by hospitalization ( 23% ) and drugs ( 15% ) . Basombrio et al . [29] reported the direct and indirect ( value of working days lost ) cost of CD in Argentina and concluded that acute phase costs were US$ 591 per patient per year , of which 34% corresponded to medical counseling and 27% to labor loss . The indeterminate phase cost was US$ 174 , of which 30% corresponded to labor loss and 28% to laboratory tests , similar to that reported herein . Chronic phase costs were between US$603 and US$ 736 , of which 27% to 37% corresponded to medical counseling and 21% to 23% to labor loss , a significant difference with the proportion estimated with the present study . Contrary to data reported by Castillo-Riquelme , and similar to the present study , the contribution for surgery was between 1% and 8% [28] , [29] . Hence , labor loss costs in Argentina represent approximately 25% of total costs across all disease stages , while in the present findings , labor loss in indeterminate is higher than this , but becomes the largest cost component in the chronic stage . Schenone reported that average annual patient costs for chronic chagasic cardiopathy in Chile is between US$ 439 and US$ 584 , while we estimated a cost between US$ 16 , 630 and US$ 24 , 588 [30] . The previous study did not consider labor loss , which may in part account for these differences . Based on information gathered from the literature review and expert panel , Akhavan estimated in Brazil that the lifetime medical care cost of a chagasic patient in the indeterminate phase is US$ 1 , 140 , the cost for a patient with digestive complication was between US$ 4 , 510 and US$ 9 , 890 , while the cost for a patient with cardiac complications was US$ 4 , 075 to US $55 , 159 . Unfortunately , that study does not provide the distribution of the cost components [31] . The lifetime cost per patient in the indeterminate phase estimated by the present study ranges from US$ 6 , 488 to US$ 7 , 481 , which is 5 . 7 to 6 . 6 times greater than the costs estimated by Akhavan . In addition , the costs estimated by Akhavan do not include the value of working days lost . The lower limit cost estimated by Akhavan for the digestive and cardiac complications , which both occur in the chronic phase , are similar to the cost estimated by the present study . The upper limit costs from Akhaven vary between 1 . 8 to 11 . 2 times greater than the costs estimated in the present study . This difference can be explained due to the fact that costs in the previous study were calculated based on medical care consumption , and in the present study a cohort was used . Using a methodological approach similar to the present study , Lee et al . [32] estimated the global economic cost of CD from a societal perspective even though they do not report cost specifically for Mexico and they do not consider the only other CD cost study from Mexico for their analysis [16] . For Latin America , the annual health-care cost per patient was US$ 383 ( range: US$ 207–US$ 636 ) , annual cost per patient due to productivity losses was US$ 3 , 676 ( range: US$ 3 , 362–US$ 3 , 798 ) , that is to say that productivity losses were estimated at 9 . 6 times greater than the medical care cost ( direct cost ) . The lifetime cost per patient for an individual with CD was estimated at US$ 2 , 600 ( range: US$ 1 , 966–US$ 3 , 034 ) . The lifetime cost estimated in the present study is 3 . 9 times greater than that calculated by these authors . Present data suggest that in the long term , it is cheaper to appropriately diagnose and treat chagasic patients instead of doing nothing . This finding by itself should motivate public policy to attend and appropriately manage exposed and potentially infected populations and establish public health interventions for this disease in Mexico , which has been neglected by health authorities [33] , [34] . This finding is even more convincing if one considers that appropriate anti-parasitic treatment can arrest further progression of disease and enhance , in the case of chronic cases , the patient's quality of life . The short and long term labor context and impact of the disease should be more carefully analyzed and considered by labor management and economic strategists , as in the case of other neglected tropical diseases , especially when public policy prioritized evidence-based social en economic development . One of the important limitations of the present study , a reflection regarding the almost complete absence of this disease in the medical care and public health community in Mexico , was the reduced pool of clinical experts in order to construct more robust clinical care models . Once Mexico publishes a clinical guideline for CD , and if there is a decision to revert the neglect for the disease at all levels of health care and preventive public health programs , more complete analysis can consider the heterogeneity and real costs for all sectors of the Mexican population . Chagas disease is a neglected tropical disease , internationally , and particularly in Mexican public health policy . The implications of continued abandonment to prevent and attend exposed population should be evaluated from both individual and collective perspectives and from all sectors , so that its impact at all levels of the Mexican economy can be considered for evidence-based policy decisions . | Chagas disease is caused by the flagellated protozoan parasite Trypanosoma cruzi , vectored in Mexico in both rural and urban areas via one of 18 triatomine bug species . Despite ample morbidity and mortality evidence , however , health policy managers in Mexico have continued to neglect prevention , control and clinical attention for the disease . A computer simulation Markov model was programmed and fed with information from published evidence and an expert panel . The lifetime cost for a timely diagnosed and treated Chagas disease patient is US$ 10 , 160 , while the cost for an undiagnosed individual is US$ 11 , 877 . The cost of a diagnosed and treated case increases 24-fold from early acute to indeterminate stage . The major cost component for lifetime cost was working days lost , between 44% and 75% , depending on the program scenario for timely diagnosis and treatment . Timely medical attention for infected individuals is cheaper than doing nothing , especially if life and labor costs are included . The evidence provided , essential for decision-making , should be used to develop disease-specific prevention , control and patient clinical diagnosis and treatment policies for Chagas disease in Mexico . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"health",
"economics",
"social",
"sciences",
"economics",
"health",
"care"
] | 2014 | Opportunity Cost for Early Treatment of Chagas Disease in Mexico |
Humans and dogs are both affected by the allergic skin disease atopic dermatitis ( AD ) , caused by an interaction between genetic and environmental factors . The German shepherd dog ( GSD ) is a high-risk breed for canine AD ( CAD ) . In this study , we used a Swedish cohort of GSDs as a model for human AD . Serum IgA levels are known to be lower in GSDs compared to other breeds . We detected significantly lower IgA levels in the CAD cases compared to controls ( p = 1 . 1×10−5 ) in our study population . We also detected a separation within the GSD cohort , where dogs could be grouped into two different subpopulations . Disease prevalence differed significantly between the subpopulations contributing to population stratification ( λ = 1 . 3 ) , which was successfully corrected for using a mixed model approach . A genome-wide association analysis of CAD was performed ( ncases = 91 , ncontrols = 88 ) . IgA levels were included in the model , due to the high correlation between CAD and low IgA levels . In addition , we detected a correlation between IgA levels and the age at the time of sampling ( corr = 0 . 42 , p = 3 . 0×10−9 ) , thus age was included in the model . A genome-wide significant association was detected on chromosome 27 ( praw = 3 . 1×10−7 , pgenome = 0 . 03 ) . The total associated region was defined as a ∼1 . 5-Mb-long haplotype including eight genes . Through targeted re-sequencing and additional genotyping of a subset of identified SNPs , we defined 11 smaller haplotype blocks within the associated region . Two blocks showed the strongest association to CAD . The ∼209-kb region , defined by the two blocks , harbors only the PKP2 gene , encoding Plakophilin 2 expressed in the desmosomes and important for skin structure . Our results may yield further insight into the genetics behind both canine and human AD .
The domestic dog ( Canis familiaris ) has been bred for different purposes and characteristics for thousands of years [1] . The creation of modern dog breeds started around 200 years ago and was based on few founders and breeding strategies such as strong selection for certain traits , popular sires and inbreeding/backcrossing . This has led to enrichment of disease mutations in different breeds . The German shepherd dog ( GSD ) breed has an exceptionally high susceptibility to immunological diseases or immune-related disorders including skin as well as gastrointestinal problems . Inflammatory and immune-related diseases that have been reported with high incidence in GSDs are , for example exocrine pancreas insufficiency due to atrophy [2] , [3] , canine atopic dermatitis ( CAD ) [4] , [5] , anal furunculosis [6] , [7] and disseminated aspergillosis [8] . A predisposition for food hypersensitivity and bacterial folliculitis [9] as well as low serum IgA levels [10]–[12] have also been reported in the GSD breed . CAD is defined as an inflammatory and pruritic allergic skin disease caused by an interaction between genetic and environmental factors [13] , [14] . The characteristic clinical features are most commonly associated with IgE antibodies directed towards environmental allergens [15] . In dogs , the allergic symptoms appear as eczematous skin but do not show the sequential development called atopic march ( eczema in a child being often followed by asthma and allergic rhinitis in the adult patient ) as described in humans [16] , [17] . Clinical signs usually develop at a young age in both humans [16] and dogs . In dogs the disease onset is typically between six months and three years of age [18] . The initial signs of CAD can either be seasonal or non-seasonal , depending on the allergens involved . Face , ears , paws , extremities , ventrum and flex-zones are typically affected by pruritus and erythema [18] in a pattern similar to that observed in human AD [19] . To establish the diagnosis of CAD an extensive work-up is required [20] , where conditions with similar clinical presentations must be ruled out . These include: scabies or other pruritic ectoparasite infestations , pruritic bacterial skin infections , Malassezia dermatitis , flea allergy dermatitis and , less commonly , cornification disorders and contact dermatitis . Cutaneous adverse food reactions ( CAFR ) can present similarly or contribute to clinical signs of CAD , but can be mediated by either hypersensitivity or non-immunological reactions . Thus , ideally the presence of CAFR should be evaluated before making the diagnosis . Also scabies could satisfy many of the inclusion criteria [21] and therefore has to be excluded as possible differential diagnosis . A positive allergen-specific IgE test ( serology or intradermal test ) is needed for final diagnosis and aids in defining offending allergens . In humans , mutations in the gene filaggrin ( FLG ) increase the risk of several complex diseases , including AD . Altogether 42% of AD-affected individuals carry FLG mutations , which is considerably higher than the carrier frequency of 10% observed in Europeans [22] . The aetiology of Filaggrin deficiency in AD is characterized by a cutaneous barrier defect , which enhances allergen penetration , bacterial colonisation and infection and cutaneous inflammation driven by type 2 helper T cells [23] . Filaggrin mutations are also known to cause asthma regardless of atopic phenotype [24] and ichtyosis vulgaris [25] in humans . Asthma-like symptoms are rarely reported in dogs: in a multi-centre study including ∼800 CAD dogs only 0 . 07% had any respiratory signs in the form of sneezing/rhinitis [17] . Different types of ichtyosis have been described in various breeds such as Golden retriever [26] , Cavalier King Charles spaniel [27] and Soft Coated Wheaten terrier [28] , however , to our knowledge , not in GSDs . Alopecia areata in humans has been correlated to filaggrin mutations and development of atopic dermatitis [29] . Canine models have previously been suggested for Alopecia areata [30] , however this condition has not been reported in any dogs within our studied GSD population . Immunoglobulin A ( IgA ) consists of two different forms , secretory IgA and serum IgA . In humans , serum concentrations of IgA are normally around 2–3 g/l , which makes it the second most prevalent antibody in serum after IgG [31] . IgA deficiency ( IgAD ) is the most common primary immunodeficiency in Caucasians with an estimated frequency of 1/600 . IgA levels <0 . 07 g/l together with normal levels of IgG and IgM define IgAD in humans [32] . Compared to other dog breeds , very low IgA levels are known to be overrepresented in GSDs [33]–[37] Low serum IgA levels have also been reported in Shar-Pei [38] and Beagle [39] . Moreover , low levels of secretory-IgA in mucosa , tears [11] , [40] and faecal extracts [41] have been reported in GSDs . Human studies show that children tend to have lower serum IgA levels than adults [42] . This is in concordance to the lower serum and secretory ( tear ) IgA levels being described in one year old or younger dogs compared to older dogs [43] . While increased incidence of upper respiratory tract infections , allergies and autoimmune diseases are observed in IgA-deficient human patients; more often humans show no symptoms at low levels of IgA [44] . Similarly , dogs with low IgA levels can either be asymptomatic or affected with recurrent upper respiratory infections and chronic dermatitis [39] . Due to the similarities between human patients and GSDs affected by AD and low IgA levels , we decided to study these two traits in a cohort of GSDs . Our aim was to detect loci associated with CAD and evaluate whether IgA levels in serum are correlated with the CAD phenotype in GSDs . We found a strong correlation between low serum IgA levels and CAD and could identify a genome-wide significant association of a locus with CAD using serum IgA levels and age at sampling as covariates . In addition to reaching our primary aim , we could also present characteristics specific to our sample cohort , including the detection of subpopulations with diverse predisposition of the studied phenotypes resulting in pronounced population stratification .
We investigated the diagnostic features CAD and low IgA levels , in a Swedish population of GSDs . The total number of dogs included in the study is presented in Table 1 . When considering the CAD phenotype we first evaluated the relationship of the following parameters; CAD status , IgA levels and gender . 40 . 7% ( n = 37 ) of the CAD cases had IgA-levels ≤0 . 10 g/l compared to 5 . 4% ( n = 5 ) of the CAD controls . The IgA levels were significantly lower in CAD cases versus controls p = 1 . 1×10−5 ( Figure 1A ) , mean IgA level in cases was 0 . 16 g/l and 0 . 26 g/l in controls ( before excluding the 5 CAD controls with low IgA levels from the final association analysis , see Materials and Methods ) . We detected no gender bias in cases versus controls for CAD ( p = 0 . 88 ) . When considering whether IgA levels were related to age , we determined regression coefficient of 0 . 42 in all dogs together ( p = 3 . 0×10−9 ) , 0 . 37 in cases ( p = 3 . 6×10−4 ) and 0 . 28 in controls ( p = 8 . 5×10−3 ) . We added the age at sampling as a covariate in the association analyses in order to remove any confounding effects of the IgA measurements' dependency of age . We performed genotyping of ∼170 , 000 SNP markers of the entire GSD cohort ( n = 207 ) . We excluded non-informative markers and markers with low call rate and 114 , 348 markers remained for the final analysis . We performed an association analysis of CAD using IgA levels and age at sampling as covariates . The initial association analysis for CAD with IgA levels and age at sampling as covariates revealed that the GSD sample set was highly stratified with λ ( genomic inflation factor ) of λno correction = 1 . 3 . The GSD population is clearly formed into two subpopulations ( Figure 1B ) defined using K-means clustering as described in Materials and Methods . The major cause of the high inflation factor , i . e . stratification , is the uneven distribution of cases and controls across the subpopulations visualized as a multi-dimensional scaling ( MDS ) plot ( Figure 1C ) . In addition , the IgA levels followed a similar pattern , being unevenly distributed across the two subpopulations ( Figure 1D ) . We found a pronounced difference in disease risk between subpopulations ( p = 1 . 7×10−6 , odds ratio OR = 4 . 4 , CI95 = 2 . 3–8 . 8 ) . The subpopulation counts are presented in Table 2 . We used the mixed model approach to account for the observed population structure and cryptic relatedness between the individuals , which is common in dog breeds . After fitting the mixed model we observed no inflation ( λ = 1 . 0 ) as presented in quantile-quantile ( QQ ) plot ( Figure 2 ) . In the association analysis of CAD we found a significant association to chromosome 27 where 19 SNPs between 17 , 814 , 493–19 , 262 , 027 ( CanFam 2 . 0 ) showed association p<2 . 8×10−5 . The top two SNPs are located at canine chromosome 27 ( CFA 27 ) : 19 , 140 , 837 bp ( praw = 3 . 1×10−7 and pgenome = 0 . 03 ) and 18 , 861 , 228 bp ( praw = 6 . 7×10−7and pgenome = 0 . 07 ) ( Figure 3A–3C ) . To define the associated haplotype we performed clumping using r2 = 0 . 8 , and identified a 21 SNP haplotype spanning from 17 , 814 , 493 to 19 , 262 , 027 . This haplotype region contains eight genes ( CPNE8 , MRPC37 , ALG10B , NAP1L1 , SYT10 , PKP2 , YARS2 and DNM1L ) where the two top SNPs surround the PKP2 gene as indicated in Figure 3C . The haplotype corresponds to the region identified by the 19 associated SNPs and covers a region of ∼1 . 5 Mb . The haplotype region shows a mosaic pattern of association typical for purebred dogs [45] , thus it is not possible from this data to define a shorter associated haplotype . Using Haploview we detected lower association to CAD when considering the ∼1 . 5 Mb haplotype compared to the single top SNPs ( phaplotype = 2 . 6×10−5 ) . The observed minor allele frequency ( MAF ) of the top SNP ( CFA 27: 19 , 140 , 837 bp ) was 0 . 29 across all samples , and 0 . 40 and 0 . 16 in cases and controls , respectively . The minor allele ( G ) conferred an OR = 1 . 28 for CAD . We observed a two-fold difference in MAF between the two detected subpopulations ( MAFsubpopulation 1 = 0 . 40 , MAFsubpopulation 2 = 0 . 20 ) . We performed targeted re-sequencing ( Roche NimbleGen sequence capture array ) of the locus on CFA 27 spanning 16 . 8–19 . 6 Mb ( CanFam 2 . 0 ) i . e . including the associated haplotype located at ∼17 . 8–19 . 3 Mb . In total , three dogs homozygous for the control haplotype , one dog homozygous for the case haplotype and three dogs heterozygous for the case and control haplotypes were sequenced ( Figure 4A ) . In total , 2 , 587 SNPs of all the identified SNPs ( n = 8 , 765 ) followed the case and control haplotype pattern ( see Materials and Methods ) . We used SEQScoring [46] , ( see Materials and Methods ) to prioritize potentially causal variants . As expected , the majority of the SNPs detected to correlate with the case/control haplotypes ( 86% ) were located within the associated ( 17 . 8–19 . 3 Mb ) region . No structural variants were detected . In total , 54 SNPs were included on an iPLEX array for further genotyping in the same cohort used for the GWAS . These SNPs were concordant with the risk haplotype and considered functional candidates based on their location in conserved elements or in genes . In addition the top GWAS SNPs were included . For the final analysis , 42 SNPs and 84 controls and 91 cases remained after quality control ( see Materials and Methods ) . Using Haploview , we defined haplotypes based on r2≥0 . 9 between neighbouring SNPs . The risk alleles of block 11 and 7 ( GCCA and AGG , respectively ) had a frequency of 40 . 1% in the cases versus 16 . 7% in the controls ( praw = 1 . 3×10−6 , p1 , 000 , 000perm = 4 . 0×10−6 ) . The common control allele TTT of block 11 had the same p-value as the risk allele and a frequency of 83 . 3 % in controls versus 59 . 9% in cases . Considering single SNPs; the top associated were the risk alleles of 18 , 934 , 038 bp and 18 , 934 , 219 bp ( part of block 7 ) , and 19 , 140 , 837 bp ( part of block 11 and also the top GWAS SNP ) . They had the same frequency as the risk alleles of the corresponding haplotypes and were associated to the same extent ( praw = 1 . 3×10−6 ) but with a slightly less significant p-value after permutations ( p1 , 000 , 000perm = 3 . 1×10−5 ) due to the larger number of SNPs compared to haplotypes . See the association analysis results of haplotypes and SNPs in Table 3 and Table 4 , respectively ( see also Tables S1 and S2 ) . The association of SNPs and haplotypes ( p-value after 1 , 000 , 000 permutations ) as well as the defined haplotypes and the LD plot are visualized in Figure 4B–4E . These results indicate that the region; 18 , 934 , 038 – 19 , 142 , 893 Mb harbours the causative mutation predisposing for CAD in the studied GSD population . This is in concordance with the genome-wide association results where the top associated SNP is located at 19 , 140 , 837 bp . Only one gene , PKP2 , falls within the top region ( defined by block 7–11 ) . The PKP2 gene , encoding the protein Plakophilin 2 , a central component of desmosomes [47] , is an excellent candidate gene for CAD .
We detected a significant difference in IgA levels in CAD cases compared to CAD controls ( Figure 1A ) , using a mixed-model approach . This suggests a functional role of IgA in the aetiology of CAD . The overall low IgA levels seen in the GSD breed might contribute to its predisposition for CAD: among the CAD cases 40 . 7% had low IgA-levels compared to only 5 . 4% of the CAD controls . The associated haplotype on chromosome 27 from the genome-wide association analysis of CAD includes eight genes; CPNE8 , MRPC37 , ALG10B , NAP1L1 , SYT10 , PKP2 , YARS2 and DNM1L . The first German sheepdogs were exhibited in 1882 at a dog show in Hannover , Germany . These dogs were the ancestors to what became the German shepherd dog ( GSD ) breed formed in 1899 . The way breeding has been performed led to a split into two variants in the end of the 1970s [48] . The Swedish GSD population used in this study was highly stratified primarily due to the formation of two subpopulations . We found a significant difference between subpopulations regarding both phenotypes in the study ( IgA levels and CAD ) where subpopulation 2 harbours more CAD cases and dogs with low IgA levels than subpopulation 1 . When comparing the merits of the dogs included in the CAD association analysis , we noted that GSDs in subpopulation 1 were more often of working type compared to subpopulation 2 . Moreover , fewer dogs in subpopulation 1 had documented show results compared to subpopulation 2 . Thus , the risk of CAD and low IgA levels seems lower in the GSD population bred for working capacity . The stratification was successfully corrected for by using the mixed model approach within the GenABEL software . Not only does it correct for the formation of two clusters and the uneven distribution of cases and controls across the clusters , but also for cryptic relatedness typical for dog breeds . Despite the identified subpopulations , there is no apparent discontinuity between them in terms of gene flow ( Figure 1B ) . Therefore , a mixed model approach was sufficient to remove the effect of stratification . Simpler approaches , such as genomic control or PCA-based corrections , were not capable of correcting the observed stratification ( data not shown ) . In addition , we used IgA levels and age at sampling as covariates in order to account for their effect on the observed phenotypes . The sequencing data generated in the 2 . 8 Mb region on CFA 27 verified the ∼1 . 5 Mb long associated haplotype showing 86% of the 2 , 587 SNPs following the case and control haplotype pattern located at ∼17 . 8–19 . 3 Mb . Based on further genotyping of 42 SNPs within the region there is clear indication that the region 18 . 94–19 . 14 Mb , based on both haplotypes and single SNPs , harbours the mutation predisposing for CAD in GSDs . By performing targeted re-sequencing of the associated region we attempted to identify all variants concordant with the phenotype and then evaluate their potential as risk variants . Here we identified two haplotypes with multiple SNPs with equally strong association and a potential for function . While one or several of these variants may be the causative variant , it is also possible that actual mutation may have been missed in the targeted re-sequencing process or in the genotyping process as several SNPs failed genotyping for technical reasons . Furthermore , our ability to predict functionality is not comprehensive as functional variants may be located in non-conserved elements or in complicated regions with low sequence coverage . The actual functional variant may also be an indel or CNV not identified in this analysis . Further analysis should reveal the exact causative mutation . The gene PKP2 , encoding Plakophilin 2 , is the only gene located within the associated 200 kb region . Plakophilin proteins are localized in the desmosomal plaque and cell nucleus and participate in linking cadherins to intermediate filaments in the cytoskeleton [49] . Plakophilin 2 takes part in pathways that drive actin reorganization and regulation of desmoplakin-intermediate filament interactions required for normal desmosome assembly [50] . Changes in the corneodesmosomes ( modified desmosomes in the epidermis ) degradation process influence the thickness of the stratum corneum and surface of the skin and abnormal corneodesmosome degradation has been found in common skin diseases including atopic dermatitis [51] . A recent small study in dogs showed statistically significant altered mRNA expression of PKP2 between atopic and healthy skin ( 20 cases and 17 controls of various breeds and mongrels ) . In addition , the expression correlated with clinical severity in atopic skin [52] . Defective permeability barrier function enables enhanced infiltration of environmental allergens into the skin , which in turn triggers immunological reactions and inflammation . [53] . Based on the increasing evidence of the skin barrier being a crucial component in the development of human and canine atopic dermatitis [54]–[55] , PKP2 serves as an excellent candidate gene . Furthermore , Filaggrin is known as a filament-aggregating protein and it is important for the formation of the stratum corneum , the outermost layer of epidermis [25] . Since the desmosome is one of the best characterized components of the stratum corneum [56] the importance of Filaggrin and Plakophilin 2 for skin structure in the aetiology of AD may be very similar . Further studies are necessary to conclusively define how CAD and low IgA levels are correlated . Low IgA levels may also affect other immune-related diseases that occur in the GSD breed . The results presented here set a starting point for further studies of susceptibility to immune diseases within the GSD breed . Even more importantly a novel gene , PKP2 , is indicated to be involved in the development of CAD in GSDs . This may be of significance also in other dog breeds and in human AD .
We collected blood samples ( EDTA for DNA extraction and serum for IgA measurements ) from 207 German shepherd pet dogs in collaboration with veterinary clinics throughout Sweden . Owner consent was collected for each dog . The majority of dogs included in the study were registered in the Swedish Kennel club ( 180 out of 207 ) . We conformed the sampling to the approval of the Swedish Animal Ethical Committee ( no . C62/10 ) and the Swedish Animal Welfare Agency ( no . 31-1711/10 ) . We extracted genomic DNA from the EDTA blood samples using the Qiagen mini- and/or midiprep extraction kit ( Qiagen , Hilden , Germany ) . DNA samples were diluted in de-ionized water and stored at −20°C . Serum was separated from the red blood cells by centrifugation and then stored at −20/−80°C . The CAD cases were dogs of all ages with positive reactions on allergen-specific IgE test ( intradermal test or IgE serology test ) , either with or without concurrent cutaneous adverse food reactions ( CAFR ) . Clinical diagnoses were established by first ruling out other causes of pruritus such as ectoparasite infestation , staphylococcal pyoderma and Malassezia dermatitis . A hypoallergenic dietary trial ( at least 6–8 weeks followed by a challenge period ) was then conducted in order to evaluate the potential contribution of CAFR . Atopic reactions were concluded if the dog was not adequately controlled on hypoallergenic diet and had positive reactions on intradermal allergy tests ( skin prick test ) or IgE serology tests . All CAD controls were over five years of age and had never suffered from pruritus , repeated ear inflammations or skin lesions compatible with CAD , neither prior to nor at the time of sampling . The age cut-off for CAD controls was set at five since affected dogs rarely debut at ages older than three years of age [17] , [18] . The information was based on either owner questionnaire and/or clinical examination . In addition , we excluded dogs with low IgA levels ( IgA≤0 . 10 g/l ) as CAD controls . We measured serum IgA concentrations with enzyme-linked immunosorbent assay ( ELISA ) using polyclonal goat anti-dog IgA antibodies ( AbD Serotec , Oxford , UK ) , polyclonal mouse anti-dog IgA antibodies ( AbD Serotec ) and polyclonal , AP-conjugated goat anti-mouse IgG ( Jackson Immunoresearch , West Grove , PA ) . All antibodies were diluted 1∶2 , 000 in PBS and the serum samples were diluted 1∶25 , 000; 1∶50 , 000 and 1∶100 , 000 in PBS . All samples were measured at least twice . The coefficient of variation ( CV ) was calculated . Samples with a CV value ≥15% were measured again . Before the average concentration was calculated , potentially outlying concentrations were excluded . With a maximal variation of 15% the reproducibility of our measurements are in the lower range of ELISA measurements which can be as high as 25% . Dogs with serum IgA levels ≤0 . 10 g/l were considered to be IgA-deficient and thus not deemed appropriate controls for CAD . All the dogs were sampled at the age of more than one year except for one individual that was 11 months and 13 days at the time of sampling . We examined the relationships between measured phenotypes and other possible covariates . We used Fisher's exact test for count data to determine whether CAD-gender relationships were significant . Similarly we used the Welch two-sample t-test for determining the CAD-IgA levels relationship . We used the same approaches to check if there were any significant differences in CAD status or IgA levels between subpopulations . As IgA levels can vary with age , we fitted a linear model to determine the age effect on the IgA levels , and used Pearson's correlation coefficient to measure the strength of the relationship . We considered CAD cases and controls separately and together . The age at the time of sampling was defined at 0 . 1-year resolution for most individuals and estimated at a year resolution for 10 dogs ( ncontrols = 7 , ncases = 3 ) . The initial data set consisted of 207 individuals genotyped using the Illumina 170K CanineHD BeadChip ( Illumina , San Diego , CA ) . Summary of individuals in each trait class is presented in Table 1 , before and after quality control ( QC ) . Prior to principal GWAS , we performed iterative QC to remove poorly genotyped and noisy data . Out of the initial number of 174 , 376 SNP markers , we excluded 55 , 399 ( 31 . 77% ) non-informative markers ( minor allele frequency below 1% ) , 2 , 537 ( 1 . 45% ) due to call rate below 0 . 95 and 2 , 722 ( 1 . 56% ) markers due to the departure from Hardy-Weinberg equilibrium ( first p<1×10−8 and then FDR<0 . 2 in CAD controls only ) . In total , 114 , 348 markers ( 65 . 57% ) were included in both analyses . Considering the entire dataset consisting of 207 individuals , we excluded two individuals due to exceptionally high identity-by-state , IBS>0 . 95 ( the one with lowest call rate was excluded in each pair - all were CAD cases ) and two apparent outliers on the multidimensional scaling ( MDS ) plot resulting in 203 individuals passing QC . After QC , 25 individuals in total were excluded from the association analysis; five were missing CAD status , five CAD controls had low IgA levels and 15 CAD controls were missing IgA levels ( Table 1 ) . The initial association ( with IgA levels and age at sampling as covariates ) indicated population stratification ( λ = 1 . 3 , λse = 1 . 5×10−3 ) . Hence , we decided to perform a closer examination of the genetic structure of our GSD population by computing autosomal genomic kinship matrix and performing standard K-means clustering . In order to determine the number of clusters ( subpopulations ) , we performed a number of K-means clustering with K = {1 , 2 , … , 10} . At each iteration , we were computed and stored the sum of within-cluster sums of squares ( ΣWCSS ) . Subsequently , we used the so-called scree test by plotting ΣWCSS vs . K and choosing the number of clusters ( K = 2 ) corresponding to the first inflection point ( for details see: [57] ) . The clusters define our subpopulations . Using MDS , we present visualisation of the genomic-kinship matrix and subpopulations in Figure 1B–1C , and subpopulation statistics are shown in Table 2 . We performed association analysis of CAD ( 91 cases and 88 controls ) with IgA levels and age at sampling as covariates . We used the GenABEL package ver . 1 . 7-0 [58] , a part of R statistical suite/software , ver . 2 . 14 . 2 [59] for the genome-wide association analyses . We used the mixed model approach for all the final analysis presented in this paper . Mixed models were fitted using polygenic_hglm function from the hglm package ver . 1 . 2–2 [60] . All parameters used for functional calls are discussed in the paragraphs describing particular steps of the previous sections . We considered p-values below 0 . 05 ( praw ) as significant and after 100 , 000 permutations as genome-wide significant p-values ( pgenome ) . For haplotype definitions we performed LD-clumping ( settings; r2 = 0 . 8 , p1 = 0 . 0001 , p2 = 0 . 001 , distance d = 3 Mb ) using our own R implementation of the algorithm described in the PLINK documentation ( PLINK v1 . 07 , [61] ) and Haploview 4 . 2 ( version 1 . 0 ) . We selected five individuals for targeted re-sequencing of the CFA 27 locus . A single case was homozygous for the risk haplotype and two were heterozygous , whereas two controls lacked the risk haplotype . Targeted capture of in total 6 . 5 Mb out of which 2 . 8 Mb spanning CFA 27:16 . 8–19 . 6 Mb ( CanFam 2 . 0 ) including the ∼1 . 5 Mb associated haplotype , was performed using a 385K custom-designed sequence capture array from Roche NimbleGen , WI . Hybridization library preparation was performed as described by Olsson et al . [62] . Captured enriched libraries were sequenced with a read length of 100 bp ( paired-end reads ) , using HiSeq 2000 ( Illumina sequencing technology ) . Sequencing was performed by the SNP&SEQ Technology Platform at SciLifeLab Uppsala . Obtained reads were mapped to CanFam 2 . 0 [45] using Burrows-Wheeler Aligner ( BWA ) [63] . The Genome Analysis Toolkit ( GATK ) ( http://www . broadinstitute . org/gatk , all web resources used in this study are also summarized in Text S1 ) was used for base quality recalibration and local realignment and the tool picard ( hosted by SAMtools [64] ) for removing PCR duplicates . For variant calling SAMtools/0 . 1 . 18 was applied using mpileup format and bcftools . Maximum read depth to call a SNP ( -D ) was set to 300 and the function -C50 was applied to reduce the effect of reads with excessive mismatches ( http://samtools . sourceforge . net ) . Mean coverage in the five analyzed individuals was 66 . 9 reads and mean share of positions covered by at least 10 reads was 87% ( Table S3 ) . We searched for structural variants by performing depth of coverage analyses using average coverage for controls as a reference . Coverage was calculated using every 20-th position in the raw pileup files and then normalized for every individual . Next , the coverage was averaged within a 100 positions-wide window separately for controls and cases . The average cases/controls ratio was then computed and used as indicator of a copy-number variation . In regions with reduced ( <−1 . 0 ) or elevated ( >1 . 0 ) relative coverage , we additionally examined the length of inferred insert size using the integrative genomics viewer ( IGV ) [65] . We used SEQScoring [46] ( http://www . seqscoring . net ) to score the SNPs by conservation and haplotype pattern; and the integrative genomics viewer ( IGV ) was used for manual visualization of SNPs , individual coverage and indels . In total , 8 , 765 SNPs were identified in the chromosome 27 region . Out of these , 2 , 587 SNPs followed the pattern of the case and control haplotypes defined by the top GWAS SNPs . The pattern was based on three dogs homozygous for the control haplotype , one dog homozygous for the case haplotype and three dogs carrying the case and control haplotype ( i . e . carriers of the case haplotype ) . Out of the 2 , 587 SNPs only 46 SNPs were located within conserved elements ( +/−5 bp ) scored by SEQscoring according to SiPhy constraint elements detected by the alignment of 29 eutherian mammals [66] . We picked out 60 SNPs for designing a genotyping array . The selection was based on the following criteria; 40 SNPs out of the 46 SNPs stated above ( SNPs too close to each other and located in repeated sequences were excluded ) , SNPs from the genome-wide array for comparison , manually picked SNPs within the PKP2 gene ( not conserved ) and SNPs in gaps in order to cover the entire associated region . Out of these , 54 SNPs were successfully pooled for additional genotyping in all dogs . The 54 SNPs were genotyped using iPLEX Sequenom MassARRAY platform ( http://www . sequenom . com/iplex ) in 185 GSD dogs . After analyzing the quality of the SNP genotyping 12 SNPs were excluded due to bad calling; nine due to heterozygotes were incorrectly called as homozygous and two due to one of the homozygous genotypes was falsely called as heterozygous and one due to MAF = 0 . In total , 42 SNPs remained for the analysis . For the association analysis of the genotyped SNPs and for defining haplotypes we used Haploview 4 . 2 ( version 1 . 0 ) . In total , 84 controls and 91 cases were included in the analysis – the same set as in the genome-wide association analysis of CAD except for four excluded controls ( two were not included due to missing DNA and two were excluded due to low call rate = 48% ) . | Humans and dogs are both affected by the allergic skin disease atopic dermatitis ( AD ) , caused by an interaction between genetic and environmental factors . The German shepherd dog ( GSD ) is a high-risk breed for canine AD ( CAD ) , also affected by low serum IgA levels . A Swedish cohort of GSDs was used as a model for human AD in this study . We performed a genome-wide association analysis where a region associated with CAD was identified . IgA levels were included in the model due to strong correlation with CAD . Also , age at sampling was included in the model due to correlation with IgA levels . The associated region , consisting of eight genes , was further fine-mapped with sequencing and additional genotyping . Haplotype association analysis from the fine-mapping data indicates association of the gene , plakophilin 2 ( PKP2 ) , known to be important for skin structure . We detected a division of the GSD breed into two subpopulations where one is more prone to develop CAD and to have lower serum IgA levels compared with the other . Here , we present methods for performing genome-wide association analyses when the study population is complex and when the trait is affected by additional parameters . The PKP2 gene found within the associated region became an interesting target for further study of its importance both in canine and human AD . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genome-wide",
"association",
"studies",
"genome",
"analysis",
"tools",
"veterinary",
"immunology",
"animal",
"genetics",
"veterinary",
"diagnostics",
"genetics",
"biology",
"genomics",
"genetics",
"of",
"disease",
"computational",
"biology",
"small",
"animal",
"care",
... | 2013 | Genome-Wide Analysis in German Shepherd Dogs Reveals Association of a Locus on CFA 27 with Atopic Dermatitis |
The Toxoplasma gondii genome contains two aromatic amino acid hydroxylase genes , AAH1 and AAH2 encode proteins that produce L-DOPA , which can serve as a precursor of catecholamine neurotransmitters . It has been suggested that this pathway elevates host dopamine levels thus making infected rodents less fearful of their definitive Felidae hosts . However , L-DOPA is also a structural precursor of melanins , secondary quinones , and dityrosine protein crosslinks , which are produced by many species . For example , dityrosine crosslinks are abundant in the oocyst walls of Eimeria and T . gondii , although their structural role has not been demonstrated , Here , we investigated the biology of AAH knockout parasites in the sexual reproductive cycle within cats . We found that ablation of the AAH genes resulted in reduced infection in the cat , lower oocyst yields , and decreased rates of sporulation . Our findings suggest that the AAH genes play a predominant role during infection in the gut of the definitive feline host .
Toxoplasma gondii is an obligate intracellular parasite and a member of the phylum Apicomplexa . It is related to Plasmodium spp . , the causative agents of malaria , as well as parasites of human and veterinary importance including Cryptosporidium spp . , Eimeria spp . , and Neospora spp . T . gondii is one of the most widely distributed parasites in the world , and can be found on every continent and in virtually every species of warm-blooded animal investigated [1] . The definitive host of T . gondii is the cat , including all members of the family Felidae [2] . Within enterocytes of the cat intestine , T . gondii is capable of producing oocysts that are shed in the feces [3] . Oocysts are spheroid , 10–12 μm in size , and are comprised of an outer wall encapsulating two sporocysts that each contain four infectious sporozoites [4] . Oocysts are structurally robust with an elasticity and strength similar to common plastics [5] . They are very environmentally resilient , able to withstand a wide range of physical and chemical challenges including bleach , ethanol , acids , and bases [6] , can stay infectious for years in the environment [7] , and represent a significant source of dissemination for the parasite [8] . Omnivorous and herbivorous animals such as livestock can become infected by eating oocysts that contaminate rangeland , or by ingestion of contaminated water supplies [1] . Humans can also be infected by accidental ingestion of oocysts in contaminated food sources such as vegetables [9] , or by ingestion of oocysts in water [10] . The walls of T . gondii oocysts are highly proteinaceous , composed of >90% protein [6] , as well as β 1–3 glucan carbohydrates [11] , and acid-fast lipids [12] . Large-scale proteomic analyses have identified 1 , 031 [13] or 1 , 304 [14] individual , non-redundant proteins associated with the oocyst . Although the function and localization of many remain unknown , two classes of oocyst wall structural proteins have been identified in other apicomplexans . In Cryptosporidium parvum , cysteine-rich COWPs ( Cryptosporidium oocyst wall proteins ) form a proteinaceous structure through extensive disulfide bridges [15] . Alternatively , tyrosine-rich EmGam ( Eimeria gametocyte ) proteins form a proteinaceous structure through extensive dityrosine linkages in the oocyst walls of Eimeria maxima [16–18] . The T . gondii genome contains seven cysteine-rich TgOWP proteins that are thought to be homologous to the COWPs . TgOWP proteins TgOWP1-3 , were characterized and described in the outer oocyst walls but not the inner sporocyst walls [19] . Although T . gondii does not contain clear homologues of Eimeria’s EmGam proteins , many tyrosine-rich proteins have been identified in both outer oocyst wall and inner sporocyst wall fractions by mass spectrometry [13 , 14] , although they have not been definitively identified as structural components in the oocyst wall [5] . The genome of T . gondii contains two genes encoding aromatic amino acid hydroxylases referred to as AAH1 and AAH2 [20] . These genes encode predicted secretory proteins that catalyze conversion of phenylalanine to tyrosine , and tyrosine to 3 , 4 dihydroxyphenylalanine ( L- DOPA ) [20] . Conversion of tyrosine to L-DOPA is the rate-limiting step of dopamine synthesis in metazoans [21] . Although initial studies suggested that these enzymes are involved in modulating dopamine production in mammalian hosts [20 , 22 , 23] , we were unable to replicate these findings in our previous work that focused on generating a knockout of AHH2 [24] . Moreover , our findings failed to reveal an elevated level of dopamine in chronically infected animals or in dopaminergic cells infected in vitro [24] , consistent with recent reports by other authors [25 , 26] . Hence , we sought to investigate other pathways that could require aromatic amino acid hydroxylase activity by T . gondii . L-DOPA serves as a precursor to many structural components across other branches of eukaryotes , including helminths , molluscs , annelids , ascidians [27] , and insects [28] , and in coccidian apicomplexan parasites . In E . maxima , L-DOPA has been identified in the oocyst , where conversion of tyrosine to 3 , 4 dihydroxyphenylalanine ( i . e . L-DOPA ) on the tyrosine-rich EmGam precursor glycoproteins is an intermediate step in the formation of dityrosine crosslinks that provide structural strength to the Eimeria oocyst wall [16 , 17] . Dityrosine has a strong blue auto fluorescence under UV light , a fluorescence observed in the oocysts of both Eimeria and T . gondii [29] . Furthermore , microarray data indicates that the AAH genes are upregulated during oocyst development in T . gondii [30] , and protein mass spectrometry identifies both tyrosine-rich proteins and the tyrosine hydroxylases AAH1 and AAH2 in the oocyst of T . gondii [13 , 14] . In contrast , these hydroxylases are not found in similar mass spectrometric analyses of tachyzoites or bradyzoites [14] . Here we sought to investigate the role of the T . gondii AAH genes in oocyst development using a combination of genetic , cellular , and biochemical studies . Although deletion of AAH2 alone caused a mild defect , ablation of AAH1 , or loss of both genes , caused a severe defect in infection of the intestine and oocyst yield . Together , our results show that the AAH genes play an important role in parasite development during the sexual cycle in the intestinal epithelium of the cat .
Animal studies on mice were approved by the Institutional Animal Studies Committee ( School of Medicine , Washington University in St . Louis ) . All procedures on cats were carried out in accordance with relevant guidelines and regulations following a protocol approved by the Beltsville Area Animal Care and Use Committee ( BAACUC ) , United States Department of Agriculture , Beltsville , MD , USA . Parasites were propagated by serial passage in human foreskin fibroblast ( HFF ( obtained from the laboratory of Dr . John Boothroyd , Stanford University School of Medicine ) ) cells grown in Dulbecco’s Modified Eagle Medium ( DMEM ) ( Life Technologies , Carlsbad , CA ) containing 10% fetal bovine serum ( FBS ) ( Hyclone , Logan , UT ) 10mM HEPES , pH 7 . 4 , 1mM glutamine , 10 μg/mL gentamycin , under 5% CO2 at 37°C ( D10 media ) . The parental ME49Δhxg::Luc strain was obtained from Laura Knoll ( University of Wisconsin , Madison ) [31] . A complete list of strains and clones used or generated in this study is provided in S1 Table . Tachyzoites were maintained by serial passage in HFF cells , grown as above . For induction of bradyzoites , cultures were switched to Roswell Park Memorial Institute 1640 medium ( RPMI 1640 ) , 50 mM HEPES pH 8 . 2 ( Thermo Fisher Scientific , Grand Island , NY ) and grown at 37°C without CO2 , as described previously [32] . Cultures were determined to be free of mycoplasma using the e-Myco plus kit ( iNtron Biotechnology ) . Parasites were harvested for experiments by scraping infected HFF monolayers into suspension , lysing HFFs and liberating tachyzoites by passage through a 20 g needle , and purifying tachyzoites with a 3 . 0 micron polycarbonate filter . CNV data was obtained from an Illumina sequencing dataset of sixteen T . gondii reference strains and 46 non-reference strains , aligned using Bowtie2 using the end-to-end option [40] . A complete list of plasmids used or generated in this study is provided in S2 Table . CRISPR/Cas9 plasmids were adapted from the pSAG1:CAS9 , U6:sgUPRT plasmid previously generated by our lab [33] . The guide RNA of the plasmid was modified to target the AAH2 5’ UTR by Q5 mutagenesis ( New England Biolabs , Ipswich , MA ) , creating the plasmid pSAG1:CAS9 , U6:sgAAH2 . A second guide RNA expression cassette targeting the AAH2 3’ UTR was inserted into the same plasmid backbone by traditional cloning steps to create the CRISPR/Cas9 AAH2 double cut plasmid pSAG1:CAS9 , U6:dgAAH2 . The same plasmid backbone was similarly adapted to target the AAH1 5’ and 3’ UTRs ( pSAG1:CAS9 , U6:sgAAH1 and pSAG1:CAS9 , U6:dgAAH1 ) . The pSAG1:CAS9 , U6:sgUPRT plasmid described previously [33] was also modified to create a double-cutting CRISPR/Cas9 plasmid targeting the HXGPRT gene pSAG1:CAS9 , U6:dgHXGPRT [34] . Plasmids used to generate the Δaah2 knockout using the HXGPRT selectable marker to replace the gene in the ME49Δhxg::Luc strain [31] , and to restore expression of AAH2 were described previously [24] . Plasmids used to generate the Δaah1 mutant by replacement with the selectable marker DHFR-Ts , and to complement expression with a cDNA construct targeted to the uracil phosphoribosyl transferase ( UPRT ) locus , were created using Gibson assembly ( New England Biolabs ) . To generate transgenic ME49 parasites , 107 tachyzoites , harvested as described above , were mixed with 5μg of CRISPR plasmids and 15μg of the appropriate homologous repair construct as plasmids linearized by restriction digest . Parasites were transfected by electroporation , and allowed to recover on HFF monolayers for 24 h . Positive selection for the HXGPRT cassette was done with 25 μg/mL mycophenolic acid ( Sigma-Aldrich , St . Louis , MO ) supplemented with 50 μg/mL xanthine ( Sigma-Aldrich ) [34] . Negative selection against the HXGPRT cassette was done with 340 μg/mL 6-Thioxanthine ( Toronto Research Chemicals , Toronto , ON ) [35] . Positive selection for the DHFR-Ts construct was done with 5μM pyrimethamine ( Sigma-Aldrich ) [36] . Negative selection against the UPRT locus was done with 10μM 5-fluorodeoxyuracil ( FUDR ) ( Sigma-Aldrich ) [37] . Clones were isolated by limiting dilution in 96-well plates containing HFF monolayers , grown as above . Clones were screened by PCR against the selectable marker and the AAH genes ( S3 Table ) . Parasites were seeded into T-25s containing monolayers of HFF cells and allowed to invade and grow for 24 h . Infected T-25s were then rinsed three times with PBS to remove any extracellular parasites . Intracellular parasites were harvested as previously described , counted by hemocytometer and seeded into 96-well plates containing monolayers of HFFs with fresh D10 media at a concentration of 105 parasites per well . Plates were allowed to grow for 24 h before being lysed with 30uL of 1x Cell Culture Lysis Reagent ( Promega , Madison , WI ) . Luminescence was developed with the Luciferase Assay Kit ( Promega ) , and imaged on a Cytation 3 imaging system ( Biotek , Winooski , VT ) . Parasites were harvested as previously described , counted by hemocytometer , and diluted into PBS . Eight-week old female CD1 mice ( Charles River Laboratories , Wilmington , MA ) were injected i . p . in a volume of 200 μL PBS containing 103 parasites and monitored daily . One month post-infection , mice were euthanized by CO2 asphyxiation followed by cervical dislocation . Brains were removed , homogenized by passage through a 20 g needle , and stained with Dolicos biflorus lectin ( DBL ) as previously described [38] . Fifteen μL of stained homogenate was examined using a Zeiss wide-field epifluorescence microscope . Three separate aliquots were counted per brain sample , and total brain cyst load was determined based on the total volume of the brain homogenate and the average count per 15 μL . All procedures described here were carried out in accordance with relevant guidelines and regulations following a protocol approved by the Beltsville Area Animal Care and Use Committee ( BAACUC ) , United States Department of Agriculture , Beltsville , MD , USA . T . gondii -free kittens ( 10- to 12-week old ) were used to study T . gondii infections . Briefly , T . gondii infected mouse brains were homogenized by syringe and fed to the cats by placing them at the back of the tongue . All feces for each cat were collected daily after feeding infected mouse brains , and examined for T . gondii oocysts . The screening and harvesting of T . gondii oocysts were done between 3 to 21 days after infection by following procedures as described previously [1] . Cats were euthanized on day 21 post infection and blood was collected to do modified agglutination tests ( MAT ) to test for immunological reactivity to T . gondii antigens . Oocysts were collected by floatation methods using sucrose solution with a specific gravity of 1 . 15 or higher . Concentrated oocyst pellets were suspended in an aqueous solution containing 2% H2SO4 , and aerated on the shaker for 7 days at room temperature ( 20–22°C ) to allow for oocyst sporulation . Oocysts were counted using a disposable hemocytometer . Total oocysts shed by individual cats were calculated based on total counts , dilution factor , and total volume . For histological studies , infected cats were euthanized at day 6/7 and portions of intestinal ileum were fixed in 10% buffered neutral formalin . Fixed tissues were cut into sections ( 2 . 5 x 0 . 7 cm ) , placed in cassettes , embedded in paraffin , and sectioned 4–5 μm thick . Slides were deparaffinized , rehydrated , and stained with hematoxylin and eosin ( Leica Microsystems , Buffalo Grove , IL ) , or by immunohistochemistry with Rabbit anti-RH polyclonal antibody [39] and Streptavidin-HRP ( Jackson Labs , West Grove , PA ) , according to standard protocols [1] . Images were taken on a Zeiss AxioSkop wide field epifluorescence microscope equipped with AxioCam CCD camera and images were captured using AxioVision v3 . 1 ( Carl Zeiss Inc . , Thornwood , NY ) . For each image , 10 μL of oocyst-laden cat fecal suspension were placed on a slide and imaged with a DAPI filter ( 300–390 nm excitation , 420 nm emission ) . Statistical analysis was done in Prism 6 for Mac OSX ( GraphPad Software , La Jolla , CA ) . One-way and two-way ANOVAs for parametric data sets and Kruskal-Wallis tests for nonparametric data sets were conducted with a threshold of P ≤ 0 . 05 considered significant .
Previous studies have described two genes AAH1 and AAH2 that are very closely related and located on chromosome V ( ToxoDB ver . 8 ME49 genome ) [40] . Analysis of copy number variation ( CNV ) of AAH2 TgME49_212740 showed approximately two copies in the type 1 strain GT1 , the type 2 strain Pru , the type 3 strain VEG and the type 10 strain VAND ( Fig 1A ) . In contrast , the type 2 strain ME49 had a CNV level consistent with three copies ( Fig 1A ) . Although AAH1 and AAH2 genes appeared as tandem loci in ToxoDBv8 , v9 and subsequent assemblies placed AAH1 ( TgME49_087510 ) on an unassembled contig ( KE139705 ) , and contains an additional third gene consistent with AAH2 ( TgME49_212710 ) on another unassembled contig ( KE139818 ) , while recognizing only one tyrosine hydroxylase AAH2 within the parasite genome itself , located on chromosome V . Mapping reads across each base pair of the AAH2 locus showed a consistent CNV of approximately 3 across the coding region of AAH2 ( Fig 1B ) . To further examine the nature of the predicted third copy , we amplified the 3’ region of AAH1/ AAH2 using primers common to both genes ( Fig 1C ) . We then interrogated the nature of the alleles present in the ME49 strain using Sanger sequencing . Inspection of the chromatographs from Sanger sequencing indicated a 2:1 ratio of AAH2 to AAH1 single nucleotide polymorphisms ( SNPs ) , consistent with a duplication of AAH2 in ME49 ( Fig 1D ) . These sequencing results also confirmed the ToxoDB ver . 8 arrangement of flanking regions for AAH1 and AAH2 . We previously reported that deletion of AAH2 in the type 2 Pru strain has no effect on growth in vitro or development of bradyzoites in vivo [24] . To examine the ability of Δaah2 mutants to be passaged through cats , we decided to generate a similar Δaah2 deletion in the type 2 ME49 strain , which has a high capacity for oocyst generation . We targeted the AAH2 gene for replacement with the HXGPRT selectable marker in the ME49Δhxg::Luc strain ( referred to as wild type ( WT ) ) , which has a deletion on the hxgprt locus and is also tagged with firefly luciferase . To efficiently delete the AAH2 gene , a CRISPR/Cas9 plasmid containing two guide RNAs targeting the 5’ and 3’ UTRs of AAH2 was created ( Fig 2A ) ( S2 Table ) . This double-cutting plasmid was co-transfected into the parental WT strain with an HXGPRT drug resistance cassette targeted to the AAH2 locus to create the clone Δaah2::HXG ( S1 Table ) . Sanger sequencing of this clone revealed that both copies of AHH2 had been removed , while the AHH1 gene remained intact ( Fig 1D ) . To remove the HXGPRT selectable marker , a CRISPR/Cas9 double-cutter of HXGPRT was co-transfected with an aah2-null fusion construct of the AAH2 5’ and 3’ UTRs ( pΔaah2 ) or a complement construct of its 5’ and 3’ UTRs appended to a cDNA copy of AAH2 ( pAAH2 ) to make the clean knockout clone Δaah2 ( referred to as Δh2 ) and the complement clone Δaah2::AAH2 ( referred to as Δh2-H2 ) , which restores expression of AAH2 ( Fig 2A ) . Subsequently , to knock out AAH1 , we created a double-cutting CRISPR/Cas9 construct targeted to the UTRs of the AAH1 gene , and co-transfected it with a Δaah1::DHFR-Ts construct ( pΔaah1::DHFR-Ts ) ( S2 Table ) into WT or Δh2 strains to make the clones Δaah1 ( referred to as Δh1 ) and Δaah1Δaah2 ( referred to as Δh1Δh2 ) ( Fig 2B ) ( S1 Table ) . To restore AAH1 , we co-transfected the pSAG1:CAS9 , U6:sgUPRT CRISPR plasmid with a repair construct containing HXGPRT and a cDNA copy of AAH1 to create the clones Δaah1-AAH1 ( referred to as Δh1-H1 ) and Δaah1Δaah2-AAH1 ( referred to as Δh1Δh2-H1 ) ( S1 Table ) . Having generated a single knockout of each of the Δaah1 and Δaah2 , as well as the double Δaah1Δaah2 knockout and several complemented strains , we decided to compare their growth and differentiation abilities in vitro and in vivo . Consistent with the fact that we were able to obtain the mutants readily in culture without any apparent growth defect , their growth as tachyzoites was similar when compared using a highly quantitative luciferase assay ( Fig 2D ) . We also tested their ability to differentiate to bradyzoites in vitro under conditions of pH 8 . 2 stress , as assessed by staining with Dolichos biflorus lectin , which stains carbohydrates in the cyst wall . We observed that the ability of the knockout and complemented strains to differentiate into bradyzoites was unaffected ( Fig 3A and 3B ) . Additionally , these strains were injected into mice in order to assess their ability to form cysts in the brains of chronically infected mice . Loss of the AAH1 or AAH2 genes did not affect the ability to produce cysts in the mouse brain , and although the complementation of the double Δaah1Δaah2 ( Δh1Δh2 ) with the AAH1 gene showed slightly higher cyst burdens , this was not significant ( Fig 3C ) . The lack of a discernable phenotype on the development of bradyzoites is consistent with our previous studies in the Pru strain , albeit this was previously only tested with the Δaah2 mutant [24] . To investigate development during the sexual cycle , tissue cysts contained in mouse brain homogenate were fed orally to cats and oocyst shedding was monitored . The normal prepatent period for oocyst shedding following infection with bradyzoites is 3–5 days with peak shedding from 5–8 days [2] . Consistent with this , cats that showed oocysts shedding commenced within the first week . However , to be sure we collected all of the oocysts produced , we extended the observation period to 21 days . Infection with the WT strain consistently yielded around 106−107 total oocysts shed during this time period ( Fig 4A ) . Although the Δaah2 ( Δh2 ) mutant yielded much lower levels of oocyst in two of three cats , a third animal showed only ~ 10 fold reduction to ~105 total oocysts ( Fig 4A ) . In contrast , the Δaah1 mutant ( Δh1 ) and Δaah1Δaah2 double mutant ( Δh1Δh2 ) showed a severe defect in oocyst yield in two of two cats tested , leading to only ~103 total oocysts per animal ( Fig 4A ) . The differences observed in these animals were significant when the knockout strains were compared as a whole to the wild type ( Fig 4A ) . However , they did not reach statistical significance when compared individually to the wild type ( Fig 4A ) , due to the low sample sizes used . Given the magnitude of the phenotype , and the consistency among mutants , we did not feel it was worthwhile to use more animals simply to achieve an arbitrary level of statistical significance . The moderate defect in the Δaah2 ( Δh2 ) , and the very severe defect in both the Δaah1 ( Δh1 ) and the double Δaah1Δaah2 ( Δh1Δh2 ) knockouts , were fully restored in the respective complemented strains ( Fig 4A ) . We also tested the ability of shed oocysts to undergo sporulation , since meiosis occurs after oocyst shedding . The sporulation rate is a measure of viability as unless oocyst mature to form sporozoites , they remain non-infectious [4] . Wild type oocysts showed a successful sporulation rate of 75–80% and this dropped significantly to ~ 60% in the Δaah2 ( Δh2 ) ( Fig 4B ) . Oocyst shedding was so low that we were not able to adequately quantify the efficiency of sporulation in the single Δaah1 ( Δh1 ) and double Δaah1Δaah2 ( Δh1Δh2 ) mutants ( Fig 4B ) ; however , based upon very limited counts , the sporulation success rate of these strains varied from 10–50% across samples . Complementation of AAH1 to the Δaah1 ( Δh1 ) single knockout or the Δaah1Δaah2 ( Δh1Δh2 ) double knockout partially rescued sporulation efficiency ( Fig 4B ) . Dityrosine fluorescence is normally much stronger on the inner sporocyst walls , and consequently the intensity of fluorescence under UV illumination was lower in unsporulated oocysts ( Fig 5 ) . Although the single and double mutants showed variable defects in the extent of sporulation ( Fig 4B ) , when oocyst sporulation was normal , the resulting fluorescence of the inner sporocyst walls was similar among all the strains tested ( Fig 5 ) . We successfully hatched Δaah2 oocysts and recovered them back into in vitro culture as tachyzoites , indicating that the oocysts that appeared to develop successfully were viable . However , the yield of the Δaah1 and Δaah1Δaah2 knockouts was too low to allow for this method of recovery . We reasoned that any defect during asexual expansion in the cat intestine or during the sexual cycle could cause a block that resulted in fewer oocysts being formed . Infection in the cat intestine initially proceeds though asexual expansion , termed A-E forms , which divide by endodyogeny and schizogony , before sexual development commences with the formation of macro and microgamocytes [2] . This process culminates with the exflaggelation of microgametes followed by fertilization of the macrogamete to yield a zygote that matures into an oocyst [2] . To examine the parasite infectivity and development of stages that occur in the cat intestine , we euthanized animals during the initial phase of oocyst shedding and examined tissue sections by conventional histology . In tissue sections from cats infected with the wild type ( WT ) , Δaah1 ( Δh1 ) and Δaah2 ( Δh2 ) parasites taken at 6–7 days post-infection , parasite infection of the intestinal ileum was readily seen ( Fig 6A–6C ) . However , the Δaah1 parasites showed a significant defect in overall density of infection ( Fig 6D ) . We were readily able to recognize merozoites , schizonts , microgamonts and macrogamonts , indicating that these lines grow well in the gut ( Fig 7 ) . Although the density of infection was lower in the Δaah1 mutant ( Fig 8A ) , the relative distribution of parasite stages was not significantly different ( Fig 8B , S4 Table ) , ruling out the possibility of a defect in any specific stage of parasite sexual development inside the intestinal ileum . Collectively , these findings indicate that AAH1 plays a role in infection in the cat intestine , and that both AAH1 and AAH2 affect the efficiency of oocyst formation in vivo , and to a lesser extent the sporulation efficiency , and that these phenotypes are partially penetrant .
Previous studies have suggested that the presence of aromatic amino acid hydroxylase genes AAH1 and AAH2 in T . gondii may be an adaptation for altering host dopamine levels and thereby affecting behavior [20 , 22 , 23] . However , in prior studies [24] we were not able to replicate the association between T . gondii infection and elevated dopamine that was seen in mice [41] or in dopaminergic cell lines [22] . Additionally , alternative explanations for the AAH genes are provided by studies showing that oocyst walls of E . maxima [16 , 17] contain dityrosine crosslinks , and fluorescence under UV illumination suggests similar modifications exist in T . gondii oocyst walls [29] . To resolve the potential role of the T . gondii AAH genes in oocyst formation , we disrupted one or both genes using CRIPSR-based genome editing [33] . Our findings reveal that AAH2 plays a moderate role , while AAH1 plays a much stronger role in formation of oocysts during infection in the cat . Additionally , AAH1 may play a role in parasite survival inside the cat intestinal epithelium as it showed a defect in infectivity even at early stages of merogony and schizogony . It is possible that dityrosine or other L-DOPA derived products produced by these AAH genes play a protective role in shielding or cloaking the parasite from the host’s innate immune response in a manner analogous to the role of melanin in Cryptococcus neoformans [42] , or the AAH genes may play an additional role in nutrient availability for the parasite , converting scavenged phenylalanine to tyrosine or vice-versa . Although these findings do not rule out a CNS role for the AAH genes , they suggest that one primary function is during infection in the cat intestine , leading to formation of mature oocysts . Although the AAH genes of T . gondii have been proposed as candidate effectors for the parasite’s ability to manipulate host behavior via manipulating dopamine in the host [23 , 25 , 41 , 43–48] , our previous work failed to reproduce the parasite’s described ability to exert effects upon host dopamine levels [24] , consistent with other reports [25 , 26] . Further , inconsistencies in cat-aversive behavior and other reported behavioral changes including anxiety , activity level , learning , memory , and more , challenge the robustness of this behavioral manipulation [25 , 26 , 46 , 49–53] . Finally , the hypothesis that tissue cysts of brain-resident parasites actively alter host dopamine to exert behavioral control faces exceptional challenge from the observation that parasites defective in their ability to establish lifelong residency in the brain still result in abnormal cat attraction [51] . Additionally , the expression of the AAH genes is relatively low in both the lytic and chronic asexual stages [24] and is only upregulated in the sexual stages [30] and mass spectrometry has failed to find evidence of these proteins in tachyzoite or bradyzoite stages but identified them in the oocyst [14] . Hence if the AAH gene products are involved in altering dopamine levels in the CNS of infected rodents , they would need to do so based on exceedingly low expression levels , and in a localized region . We are presently examining neurotransmitter levels and behavioral change in mice infected with AAH mutants describe here , and such studies could potentially resolve the role of these genes in such pathways . Because of the high variability in findings regarding the effects of T . gondii infection on brain neurotransmitters and behaviors , we sought to explore alternate roles for these genes in the parasite life cycle . One obvious candidate would be the contribution of L-DOPA to the formation of protein-protein dityrosine crosslinks in the proteinaceous oocyst wall , analogous to what has been described in E . maxima [16 , 17] . Recently , the oocyst wall proteins TgOWP1-7 , which are cysteine-rich structural proteins analogous to the Cryptosporidium oocyst wall proteins , were characterized and shown to localize to the outer oocyst wall , but not the inner sporocyst walls [19] . Mass spectrometry data also reveal that tyrosine rich proteins are found in oocysts [13 , 14] , but as yet there is not direct biochemical evidence for dityrosine cross linked proteins in the oocyst wall . However , consistent with the presence of such crosslinks , both the outer oocyst wall and inner sporocyst walls show dityrosine fluorescence , although the signal is significantly brighter in the sporocyst walls . Using the efficiency of CRISPR/Cas9 to direct genetic disruption , we demonstrated that ablation of AAH1 or both AAH1 and AAH2 causes a severe defect in oocyst yield , as well as a maturation defect in the oocysts that do emerge . Parasites ablated for AAH1 were compromised in replication and development during growth in the cat intestine , and parasites ablated for AAH2 were able to develop normally within the cat intestine but were compromised in their yield and maturation efficiency after shedding into the environment . One potential function for the AAH genes is in generating modified tyrosine residues ( i . e . 3 , 4 dihydroxyphenylalanine ) that are the precursor for dityrosine crosslinks in oocyst wall proteins . This modification is expected to increase oocyst resistance to environmental conditions . The observed decrease in oocyst yield from the aah mutants following purification from cat feces is consistent with them being more fragile and prone to loss during the intensive process of osmolar , physical , and chemical treatments that are used during isolation . Although we were able to recover a small number of oocysts from the mutants , they underwent sporulation less efficiently . Since sporulation is associated with increased levels of UV fluorescence , the reduced rate of sporulation in the aah mutants is consistent with formation of fewer dityrosine crosslinks . However , some oocysts shed by the mutants were able to undergo sporulation and form oocysts with normal UV fluorescence , although at a much lower total numbers than the wild type . This suggests that if the AAH enzymes normally participate in dityrosine crosslinks , this function can be rescued in the absence of the parasite genes , albeit inefficiently . In this regard , there are at least two other potential sources for 3 , 4 dihydroxyphenylalanine that serves as a precursor for this reaction: the host cell and the microbiome . Hence , it is possible that salvage from these other sources may enable T . gondii to generate dityrosine crosslinks at a lower frequency in the absence of AAH genes . Combined with previous findings , our results suggest that T . gondii builds its oocyst walls using a hybrid strategy combining features of Cryptosporidium’s cysteine-cross-linked walls and Eimeria’s dityrosine-cross-linked walls . We hypothesize that the proteinaceous part of the outer oocyst wall of T . gondii is predominantly Cryptosporidium-like , composed of TgOWPs cross-linked by disulfides . A secondary Eimeria-like component of tyrosine-rich proteins cross-linked by dityrosines comprises the proteinaceous inner sporocyst walls in T . gondii oocysts . In this model , the aromatic amino acid hydroxylases AAH1 and AAH2 are expected to catalyze the conversion of tyrosine residues on wall proteins into 3 , 4 dihydroxyphenylalanine residues for subsequent dityrosine bond formation . The final conversion of these residues into cross-linked proteins is also likely to require a peroxidase , and a putative oxidoreductase that reliably emerges as the most abundant protein in mass spectrometry analyses provides a candidate for this activity [13 , 14 , 30] . The reduction in infectivity in the AAH1 mutant suggests that dityrosine or secondary quinones may also play a role as a virulence factor throughout earlier stages of development , analogous to the role of melanin in the neurotropic yeast Cryptococcus neoformans [42] . Alternately , the AAH genes may be involved in the conversion of phenylalanine to tyrosine to cope with nutrient limitations for growth in vivo . To test these models , further studies would be needed to define the localization of the putative tyrosine-rich protein precursors , confirm the presence of dityrosine crosslinks , and investigate the interaction of the AAH enzymes with such substrates during sexual stage and oocyst development . However , at present such studies are hindered by the necessity for sexual development to take place in the complex environment of the cat intestine . However , further exploration of these pathways may also be of value for defining attenuated mutants of T . gondii that are unable to yield infectious oocysts and yet which may induce protective immunity in the cat , thus potentially breaking transmission of the life cycle . | Toxoplasma gondii is an intracellular parasite that infects up to one-quarter of humans worldwide . Although it can infect virtually any warm-blooded animal , its definitive host is the cat where the sexual cycle occurs in enterocytes of the small intestine , producing microscopic , durable oocysts that are shed in feces and can remain infectious for extended periods of time in the environment . Two parasite genes , AAH1 and AAH2 , code for aromatic amino acid hydroxylases , which produce L-DOPA , the precursor to dopamine . However , L-DOPA is also a precursor of other structural molecules including dityrosine , which may play a role in the wall of the oocyst . We investigated the effect of AAH deletion on the ability of the parasites to undergo sexual reproduction in cats , and found that AAH-deficient parasites were defective in their ability to produce oocysts , and those oocysts were partially defective in their ability to undergo maturation once produced . Collectively , these results suggest that the AAH genes play their primary role in transmission through the definitive host . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"parasite",
"groups",
"medicine",
"and",
"health",
"sciences",
"oocysts",
"pathology",
"and",
"laboratory",
"medicine",
"toxoplasma",
"gondii",
"vertebrates",
"cloning",
"parasitic",
"diseases",
"parasitic",
"protozoans",
"animals",
"mammals",
"parasitology",
"apicomplexa... | 2017 | The aromatic amino acid hydroxylase genes AAH1 and AAH2 in Toxoplasma gondii contribute to transmission in the cat |
Release factors ( RFs ) govern the termination phase of protein synthesis . Human mitochondria harbor four different members of the class 1 RF family: RF1Lmt/mtRF1a , RF1mt , C12orf65 and ICT1 . The homolog of the essential ICT1 factor is widely distributed in bacteria and organelles and has the peculiar feature in human mitochondria to be part of the ribosome as a ribosomal protein of the large subunit . The factor has been suggested to rescue stalled ribosomes in a codon-independent manner . The mechanism of action of this factor was obscure and is addressed here . Using a homologous mitochondria system of purified components , we demonstrate that the integrated ICT1 has no rescue activity . Rather , purified ICT1 binds stoichiometrically to mitochondrial ribosomes in addition to the integrated copy and functions as a general rescue factor , i . e . it releases the polypeptide from the peptidyl tRNA from ribosomes stalled at the end or in the middle of an mRNA or even from non-programmed ribosomes . The data suggest that the unusual termination at a sense codon ( AGA/G ) of the oxidative-phosphorylation enzymes CO1 and ND6 is also performed by ICT1 challenging a previous model , according to which RF1Lmt/mtRF1a is responsible for the translation termination at non-standard stop codons . We also demonstrate by mutational analyses that the unique insertion sequence present in the N-terminal domain of ICT1 is essential for peptide release rather than for ribosome binding . The function of RF1mt , another member of the class1 RFs in mammalian mitochondria , was also examined and is discussed .
Stalled ribosomes must be rescued for productive cycles of cellular protein synthesis . So far , three rescue systems for ribosomes stalled at the ends of nonstop mRNAs have been identified in bacteria . The most thoroughly characterized pathway is the trans-translation system involving tmRNA [1] . The second pathway engages ArfA ( formerly YhdL ) [2] , [3] , a small protein composed of approximately 55 residues . ArfA was originally identified as a factor essential for the viability of E . coli in the absence of tmRNA , and accordingly was renamed ArfA , for alternative ribosome-rescue factor . ArfA , in concert with RF2 , takes over the rescue of stalled ribosomes , where RF2 hydrolyzes peptidyl-tRNA in a GGQ motif-dependent but codon-independent manner . The third pathway is mediated by ArfB ( formerly YaeJ ) [4] . This factor is a reduced paralogue of the bacterial class 1 release factors ( RFs ) that retains domain 3 but lacks domains 1 , 2 and 4 . Domains 2 and 4 are essential for stop-codon recognition , whereas domain 3 contains the GGQ-motif . In agreement with this , the ArfB in E . coli hydrolyzes the peptidyl-tRNAs of ribosomes stalled at the 3′-ends of nonstop mRNAs , and functions codon-independently . The deletion of the genes of both tmRNA and ArfB is not lethal in contrast to that of tmRNA and ArfA , but the overexpression of ArfB alone rescues the double depletion of tmRNA and ArfA . Accordingly , it was suggested that this factor should be renamed ArfB , for alternative ribosome rescue factor B . The crystal structure of ArfB , bound to the Thermus thermophilus 70S ribosome in complex with the initiator tRNAifMet and a short mRNA , has been determined [5] . The N-terminal globular domain of ArfB occupies the A site of the 50S subunit next to the P-site tRNA , and its C-terminal tail occludes the mRNA tunnel downstream of the 30S A-site . The latter is thought to function as a sensor to detect a stalled ribosome , based on the occupancy of the mRNA entry channel . Indeed , ArfB prefers a ribosomal complex with a nonstop mRNA , where the A-site is vacant on the ribosome [3] . Subsequently , the binding of the tail within the mRNA entry channel allows the N-terminal globular domain to optimally position its GGQ motif in the peptidyl-transferase center ( PTC ) and to catalyze the hydrolysis of peptidyl-tRNA . Unlike the other class 1 RFs , ArfB contains a 25 residue insertion sequence in its N-terminal globular domain . The function of this sequence in a codon-independent peptide release factor is unknown . We have only scarce knowledge about the ribosome rescue system in mitochondria . The so-called “immature colon carcinoma transcript-1” ( ICT1 ) is a bacterial ArfB homolog in mammalian mitochondria; members of this factor family are widely distributed in bacteria and organelles of all eukaryotic phyla [6] . No homologs for either tmRNA or ArfA have been found in mammalian mitochondria . ICT1 catalyzes the release of formylmethionine from its P-site fMet-tRNA in E . coli 70S ribosomes , in a codon-independent manner [7] . ICT1 is an essential mitochondrial protein , and a mutation in the GGQ-motif of ICT1 causes loss of cell viability [7] . These observations suggested that ICT1 may be responsible for the ribosome rescue in mammalian mitochondria . Strangely , ICT1 is an integral component of the mitochondrial ribosome ( mitoribosome [7] , [8] ) . ICT1 is a member of the large mitoribosomal subunit ( mitochondrial ribosomal protein L58 , MRPL58 ) , and a crucial component for its assembly [7] . No other RF family protein has been shown to be an integral ribosomal component . It is unclear , how the integrated ICT1 functions to rescue stalled ribosomes in the mammalian mitochondrial translation system . In the present study , we have investigated the function of human mitochondrial ICT1 , utilizing 55S mitoribosomes purified from pig liver mitochondria . We demonstrate that the codon-independent peptide-release activity on 55S ribosomes is only observed , when we add ICT1 to the system containing the 55S-integrated ICT1 ( or to a system containing bacterial 70S ribosomes lacking integrated ICT1 ) . These results suggest that the 55S-integrated ICT1 lacks peptide-release activity , as opposed to a previous model [7] . We further show that ICT1 can rescue ribosomal complexes not only at the ends of mRNAs , but also in the middle of mRNAs , and even without mRNAs . Our data suggest that ICT1 is a versatile ribosome rescue factor , which is also involved in the translation termination at non-standard stop codons AGG and AGA in mammalian mitochondria . The unique insertion sequence in the N-terminal domain of ICT1 is required for peptide release rather than involved in binding to the ribosome as demonstrated by a mutational analysis .
ICT1 is a component of the 55S mitochondrial ribosome ( mitochondrial ribosomal protein L58 , MRPL58 ) [7] , [8] . However , it was not clear whether the 55S ribosome itself with the integrated ICT1 exhibits the basal peptide-release activity , or whether an exogenous ICT1 can function on the 55S ribosome . To address these questions , we prepared 55S mitoribosomes from pig liver mitochondria as a model for human mitochondria , and tested them in a peptide-release assay . Pig mitochondrial ribosomes share high homology with those from human mitochondria and are easier to prepare . As shown in Fig . 1A ( upper panel ) , 55S ribosomes contained a full complement of ICT1 in a 1∶1 stoichiometry ( Fig . 1A , upper panel ) , which was a component of the large 39S mitoribosomal subunit ( Fig . 1A , lower panel ) , consistent with the previous reports [7] , [8] . In the peptide-release assay , a ribosomal complex was constructed using the 55S mitoribosome , mRNA encoding either MF-stop ( UAA ) or MFV , and N-acetyl-[3H]Phe-tRNAPhe ( ac[3H]Phe-tRNAPhe ) as a model of peptidyl-tRNA . Accordingly , ac[3H]Phe-tRNAPhe was present at the P-site , and the stop codon ( UAA ) or sense codon for Val ( GUU ) was at the A-site in the complex . The released ac[3H]Phe from tRNA was assessed in the presence of peptide release factors; values in the absence of release factors were taken as background and subtracted from the measured values ( Fig . 1B ) . The mitochondrial termination factor RF1Lmt/mtRF1a showed efficient peptide-release activity with MF-stop ( UAA ) mRNA , but not with MFV mRNA , as expected ( Fig . 1B , RF1Lmt ) [9] , [10] . Approximately 90% of the ac[3H]Phe-tRNAPhe on the ribosome programmed by MF-stop ( UAA ) mRNA was hydrolyzed by RF1Lmt/mtRF1a . RF1mt , a class 1 RF family protein in mammalian mitochondria with an unknown function , did not show peptide-release activity for both mRNAs ( Fig . 1B , RF1mt ) , even using an excess amount ( Fig . S1 ) . All the results with RF1mt hereafter are discussed later . ICT1 showed the ac[3H]Phe release activity for both MF-stop ( UAA ) mRNA and MFV mRNA on 55S ribosomes ( Fig . 1B , ICT1 ) , consistent with its codon-independent peptide-release activity on 70S ribosomes [7] . Unexpectedly , the activity of ICT1 greatly exceeded the 100% value . We suspected that ICT1 performs multiple reaction cycles on the ribosome until the acPhe-tRNA input has been consumed , but perhaps with that fraction of ribosomes , which does not carry mRNA . Therefore , we performed peptide-release assays in the absence of mRNA; Fig . 1C exhibits ICT1 peptide-release activity under this condition , but only in the presence of 55S ribosomes ( compare 55S[+] and 55S[−] ) , and even Phe is quantitatively released from Phe-tRNA when bound to non-programmed 55S ribosomes ( Fig . S2A ) . Accordingly , most of the ICT1 activity observed in the presence of mRNA actually originated from multiple reactions on mRNA-free ribosomal complexes; ICT1 releases AcPhe in the absence of mRNA also with bacterial 70S ribosomes as does ArfB ( Fig . S3 ) . Both RF1Lmt and RF1mt are not able to release peptides from peptidyl-tRNA bound to non-programmed mitoribosomes ( Fig . 1C , RF1Lmt and RF1mt ) . The 55S ribosomes alone ( containing stoichiometric amounts of ICT1 ) did not show peptide-release activity in the peptide release experiments ( for example , see Fig . 1C , compare RF[−] of 55S[+] and 55S[−] ) . These results indicated that the ICT1 integrated in the ribosome does not exhibit the peptide-release activity . We have shown in this section that ICT1 shows codon-independent peptide release activity when added to 55S mitoribosomes; the activity is not caused by the 55S-integrated ICT1 . Added ICT1 is able to exhibit peptide release activity even in the absence of mRNA . These results suggest that the 55S-integrated ICT1 is not present at the active site , where a release factor has to bind in order to trigger the release of the peptide . In the next experiment we performed a multi-round translation assay . We utilized an Escherichia coli-based reconstituted in vitro system of coupled transcription-translation [11] , which has the advantage that the 70S ribosomes do not contain integrated ICT1 ( ArfB ) factors . The peptide release factors were omitted from the system , in order to allow a specific analysis of the effects of ICT1 or other release factors . The ribosomes were programmed with mRNAs encoding a short polypeptide ( MFFLF ) . We applied three different mRNAs: a nonstop mRNA without a 3′-end , and the stop ( UAA ) and stall ( AGA ) mRNAs , both of which have a 3′-UTR with 14 nucleotides ( Fig . 2A , upper sketch in the upper panel , and Materials and Methods ) . The translation efficiencies were assessed by the incorporation of f[14C]Met . The ribosome recycling factor cannot hydrolyze a peptidyl-tRNA bound to ribosomes . Accordingly , the yield of the isolated polypeptides reflects the peptide-release activity of ICT1 with ribosomes stalled at the 3′ end of the mRNA or in the middle of the mRNA . ICT1 does not interfere with normal translation , because even excess of ICT1 in addition of the mito-release factor RF1Lmt does not affect the kinetics of the peptide release during the termination phase after the synthesis of the pentapeptide ( Fig . S2B ) . Interestingly , ICT1 could release the oligo-peptide from stalled ribosomes programmed with the stop ( UAA ) or the stall ( AGA ) mRNAs , although to a lesser extent than with the nonstop ( − ) mRNA ( Fig . 2A , lower panels ) . ICT1 functioned with the stop ( UAA ) and stall ( AGA ) mRNAs almost as efficiently as RF1Lmt did with the stop ( UAA ) . After 30 minutes of synthesis , ICT1 yielded a release of ∼0 . 3 pmol polypeptides with the stop ( UAA ) and stall ( AGA ) mRNAs , and of ∼2 . 0 pmol with the nonstop ( − ) mRNA . RF1Lmt released ∼0 . 5 pmol polypeptides with the stop ( UAA ) . These results suggest that ICT1 can rescue a stalled ribosome in the middle of the mRNA . Next , we analysed , whether ICT1 can also release polypeptides from native polysomes , in which most of the ribosomes are in the middle of the mRNA . The control is a treatment with the antibiotic puromycin , which is an analogue of the 3′-end of an aminoacyl-tRNA , binds to the A-site region of the PTC , receives the polypeptide from the peptidyl-tRNA in the P site via a peptide-bond and leaves the ribosome as peptidyl-puromycin [12] . The action of puromycin on polysomes is indicated by a polysome-breakdown and an increase of the ribosomal subunits , ( Fig . 2B , compare first two panels ) . The next three panels demonstrate that ICT1 is as effective as puromycin in breaking down the polysomes in contrast to RF1Lmt and RF1mt . ICT1 showed significant puromycin-like peptide-release activity on the polysomes , even though the A-sites on the polysomes are occupied with mRNA ( the rightmost panel ) . RF1Lmt did not exhibit this activity , since polysomes do not carry a stop codon in their A-site ( the third panel from the left ) . Collectively , our results demonstrated that ICT1 can rescue stalled ribosomes in the middle of mRNA occupying the mRNA tunnel . These observations support the recent structural model of ICT1 , according to which the C-terminal tail of ICT1 does not enter the mRNA path downstream of A-site of the ribosomal small subunit , at least as long as the ribosome stalling occurs in the middle of an mRNA [13] . Our findings also suggest that ICT1 is involved in the translation termination at non-standard stop codons AGA and AGG in mammalian mitochondria . In all assays of both the multi-round translation and the polysome breakdown , RF1mt did not show peptide-release activity with any mRNA construct examined ( Fig . 2A and B , RF1mt ) , consistent with the results of the peptide-release assay ( Figs . 1B and C , RF1mt ) . The classical class 1 RF proteins need to recognize the stop codon and simultaneously to place their GGQ-motif in the peptidyl-transferase center ( PTC ) of the ribosome . Bacterial RF1/RF2 has the “switch loop” between domains 3 and 4 ( Figs . 3A and B ) [14] , [15] . Stop codon recognition by domains 2 and 4 induces the rearrangement of this loop structure , thus positioning the GGQ-motif of domain 3 in the PTC ( Fig . 3B , see the sketch of RF2 on the ribosome ) . On the other hand , the codon-independent peptide release factors , such as ArfB and ICT1 , possess only domain 3 and lack the switch loop: domain 3 is connected by the flexible linker to its C-terminal tail that is thought to occupy the mRNA channel of the ribosome [5] . It is not clear how the GGQ-motif of ICT1 is optimally positioned in the PTC . In this regard , it is interesting that both ArfB and ICT1 uniquely contain an insertion sequence of 25 residues in the globular domain 3 , in contrast to the classical class 1 RFs ( Fig . 3B and C ) [5] , [7] . Therefore , we examined the function of the insertion sequence . We first applied the peptide-release assay with 55S ribosomes in the absence of mRNA . There are five basic amino acid residues in the insertion sequence: R116 , K118 , K124 , K126 and R129 ( Fig . 3C , highlighted with red letters; and Fig . 3D ) . Assuming that the basic amino acid residues are involved in the interaction with ribosomal RNA , alanine substitutions were introduced at these positions . The peptide-release analyses revealed that the ICT1[α2] mutant , in which K124 , K126 and R129 are simultaneously mutated to alanine , had little activity ( Fig . 4A , ICT1[α2] ) . The activity of ICT1[α2] was almost as low as those of the ICT1[GSQ] and ICT1[ΔC] mutants , in which the GGQ-motif was changed to GSQ and the C-terminal 14 amino acids were truncated , respectively ( Fig . 4A , ICT1[GSQ] and ICT1[ΔC] ) . The GGQ motif is critical for the peptide-release activity of ICT1 on the 70S ribosome [7] . The C-terminal 10 amino acids of ArfB , which correspond to the C-terminal 14 amino acids of ICT1 , are essential for its ribosome binding capacity [4] . We also performed the multi-round translation assay to assess the peptide release activities of the ICT1 mutants . The ribosomes were programmed by a nonstop mRNA encoding a short polypeptide ( MFFLF ) without a stop codon ( Fig . 4B , left ) . When wild type ICT1 was included in the system , the translation efficiency was increased , reflecting the enhanced multi-round translation ( Fig . 4B , right , red closed squares ) . An alanine substitution of any one of the basic resides in the insertion sequence showed various defects ( Fig . S4 ) . However , when ICT1[α2] was included in the system , peptide release was abolished ( Fig . 4B; green open squares ) . Neither ICT1[GSQ] nor ICT1[ΔC] enhanced the release ( Fig . 4B; red open squares and green closed squares ) . These results suggest that the peptide-release activity of ICT1 is dependent on the insertion sequence in the N-terminal globular domain . The mutations in the insertion sequence of ICT1 led to a loss of peptide-release activity ( Fig . 4 ) . Whether or not this loss was caused by a loss of binding to the ribosome was tested in the next experiment . Excess amounts of ICT1 proteins were incubated with 55S mitoribosomes or 70S E . coli ribosomes . The mixtures were then fractionated on a sucrose density gradient , and the ICT1 proteins were detected by immunoblotting of the fractions ( Fig . 5A ) . For the analysis with 55S mitoribosomes , N-terminal his-tagged ICT1 proteins were used to discriminate the exogenous ICT1 from the endogenous ICT1; the exogenous his-tagged ICT1 could be separated from the endogenous ICT1 in SDS-PAGE , according to the increased molecular weight due to the his-tag . Both ICT1 and ICT1[α2] bound similarly to the 55S mitoribosomes ( Fig . 5A , upper panels ) and 70S ribosomes ( Fig . 5A , lower panels ) . It is of note here that ICT1 proteins were incubated with 55S or 70S ribosomes in a 17-fold or 10-fold excess relative to the ribosomes , and approximately 5% and 10% of the input ICT1 was bound to the 55S and 70S ribosomes , respectively . This means that both 55S and 70S bound exogenous ICT1 with nearly a 1∶1 stoichiometry . These results clearly indicate that 55S ribosomes , already bearing the integrated ICT1 , possess two ICT1 binding sites . In Fig . 5B , the ribosome binding of wild type ICT1 or ICT1[α2] was analyzed more quantitatively . The 70S ribosomes were incubated with various amounts of ICT1 proteins , in up to 4-fold excess relative to 70S ribosomes . The ribosome-bound ICT1 proteins were then recovered by a filtering technique , and quantified by Western blotting against ICT1 . The amounts of ribosome-bound ICT1 per 70S ribosome were plotted against the input-amounts of ICT1 . The results confirmed that ICT1 and ICT1[α2] exhibit similar ribosome binding abilities . The ribosome binding of wild type ICT1 or ICT1[α2] was further analyzed with 55S mitoribosomes in the presence of the crosslinking agent BS3 , as shown in Fig . 5C . In this experiment , we aimed to obtain information about the binding site of exogenous ICT1 as well as that of the integrated ICT1 on the 55S ribosome . The 55S ribosomes were incubated with either wild type ICT1 or ICT1[α2] in the presence of BS3 . The cross-linked products were fractionated by SDS-PAGE , and analyzed by Western blotting with an ICT1 antibody . Similar cross-linking products of wild type ICT1 and ICT1[α2] were observed ( two asterisks ) , which was different when no ICT1 was added ( one asterisk ) . These observations indicate that the binding sites of the 55S-integrated ICT1 and that of the exogenous ICT1 on the 55S ribosome are different . Taken together , the results suggested that the insertion sequence of ICT1 does not affect the ribosome binding , which depends on the C-terminal tail of ICT1 . We propose that the interaction of the insertion sequence with the ribosome is responsible for an optimal positioning of the GGQ-motif of ICT1 in the peptidyl-transferase center .
A previous study addressing the function of ICT1 used a heterologous combination of E . coli 70S ribosomes and human mitochondrial ICT1 [7] . In the present study , we employed homologous systems in most experiments: 55S mitochondrial ribosomes and purified mitochondrial factors including human mitochondrial ICT1 . Our analyses revealed that exogenous ICT1 functions independently of the 55S-integrated ICT1; the 55S-integrated ICT1 shows no peptide-release activity . This is in agreement with a recent cryo-EM analysis at 4 . 9 Å resolution of the large subunit of porcine mitochondrial ribosomes , according to which ICT1 is located at the base of the central protuberance >70 Å away from the A site [16] . A conformational change of the large ribosomal subunit bringing the integrated ICT1 to the A-site is hardly conceivable . So far , ribosome-free ICT1 has not been yet detected in mitochondria [7] . Thus , a possible explanation for our observation is that the ribosome-integrated ICT1 may be released from ribosomes , which are stalled under certain cellular conditions; the liberated ICT1 then binds to the ribosomal A-site to exert its peptide-release activity ( Fig . 6 ) . In mammalian mitochondria , all of the mtDNA-encoded proteins are membrane proteins , which are probably co-translationally inserted into the mitochondrial inner membrane [17] . Therefore , the substrates of ICT1 could be ribosomes that become stalled during the insertion of the nascent peptide into the membrane . Defects in polypeptide insertion into the membrane would cause the nascent polypeptide to become stuffed in the tunnel , and eventually stall the ribosome . It is possible that the stuffed nascent polypeptide in the tunnel of the stalled ribosome causes a structural change in the large ribosomal subunit , which might be important for recognition of ICT1 and might even elicit the release of ICT1 from the ribosome . Further studies , such as search for the ribosome-free ICT1 , are required to support this hypothesis . It is also possible that ICT1 is overexpressed in response to a ribosome stalling in order to produce ribosome-free ICT1 , which acts on the stalled ribosome independently of the integrated ICT1 . Our analyses revealed that ICT1 functions as a versatile rescue factor , i . e . it releases the polypeptide from the peptidyl-tRNA from ribosomes stalled at the end or in the middle of an mRNA or even from non-programmed ribosomes ( Fig . 6 ) . The ability of ICT1 to hydrolyze the peptidyl-tRNA on the non-programmed ribosome , i . e . in the absence of mRNA , may have some advantages in mitochondrial translation . For example , ICT1 may rescue a defective initiation complex , e . g . a ribosomal complex that binds fMet-tRNA in the absence of mRNA . The ability of ICT1 to hydrolyze the peptidyl-tRNA on the ribosome in the middle of an mRNA would imply that ICT1 functions as a termination factor as well as a rescue factor . This assumption is supported by observations in S . pombe , according to which the growth defects observed in a Mrf1 ( mitochondrial release factor ) -deficient strain are compensated by overproduction of Pth4 , which is a counterpart of human ICT1 and is also a component of the large subunit of the mitoribosome [18] . In human , mtDNA-encoded cytochrome c oxidase subunit I ( CO1 ) and NADH dehydrogenase 6 ( ND6 ) carry AGA and AGG codons at the end of their mRNAs , respectively . There are no tRNAs in mammalian mitochondria that decode AGA/G codons , viz . AGA/G codons are unassigned codons in mammalian mitochondria . Ribosomes stalled at these codons might be recognized by ICT1 , i . e . it would function as a specific translation termination factor for CO1 and ND6 in human mitochondria . It has been shown that AGA/G codons in CO1 and ND6 , presumably in association with other cis elements , promote −1 frameshifting in human mitoribosomes providing an UAG stop codon at the ribosomal A-site for both CO1 and ND6 mRNA . Accordingly , it was proposed that the translation of CO1 and ND6 are terminated by RF1Lmt [19] . However , a U preceding AGA/G is rare in mitochondria of most vertebrates [13] . Moreover , the peptidyl-tRNA at the P-site can no longer interact with the mRNA via codon-anticodon interactions after the frameshift , and thus the peptidyl-tRNA would probably not be able to keep the P/P state , which is a requirement for the peptide release by RF1Lmt . Therefore , it is still unclear whether RF1Lmt or ICT1 catalyze the peptide release of CO1 and ND6 . The mutations of basic residues in the insertion sequence of ICT1 caused a loss of peptide release activity , but did not affect its ribosome binding ability ( Figs . 4 and 5 , respectively ) . The crystal structure of ArfB/YaeJ on the 70S ribosome revealed that the insertion sequence in the N-terminal globular domain interacts with 23S rRNA [5] . The phosphate backbones of U1946 and C1947 in 23S rRNA form hydrogen-bonds with the main chain N atom of His62 and the Oγ atom of Ser60 of ArfB , respectively ( Fig . S5 ) . Although these residues are not conserved in ICT1 , it is plausible that other residues in the insertion region interact with rRNA in the large mitochondrial ribosomal subunit . The mutations of basic residues in the insertion region of ICT1 might disrupt the proper interaction between the insertion sequence and rRNA , without affecting the ribosome binding ability of ICT1 , which is governed by the C-terminal tail of ICT1 . The role of the insertion sequence of ICT1 or ArfB , the codon-independent peptide release factors , could be to place the GGQ-motif in the appropriate orientation for peptide release . Very recently , effects of the point mutations in the insertion sequence of ArfB have been studied [20] . Decreased peptide release activities were observed for some mutants , probably due to the inappropriate orientation of the GGQ-motif . In bacteria , ArfA assists RF2 for the codon-independent peptide-release activity [2] , [3] . Although a structural study is required to verify how ArfA functions with RF2 , it is possible that ArfA plays the same role as the insertion sequence of ICT1/ArfB . ICT1 is one of four proteins of the class 1 RF-family in mammalian mitochondria: RF1Lmt/mtRF1a , RF1mt , C12orf65 and ICT1 [21] . Despite great efforts , the roles of RF1mt and C12orf65 are not known . A previous structural simulation study proposed that the RF1mt targets the ribosomal complex with a vacant A-site , and thus is a candidate for the ribosome rescue factor in mammalian mitochondria [21] . Another simulation study proposed that the RF1mt would display peptide release activity for UAA and UAG stop codons , but exclusively on 55S mitoribosome [13] . However , using such ribosomal complexes , we could not detect any peptide release activity of RF1mt by itself ( Figs . 1B , 1C and 3B ) . It is possible that RF1mt shows peptide release activity through its GGQ motif in collaboration with unknown factors , as bacterial RF2 shows codon-independent peptide release activity in concert with ArfA [2] , [3] . In spite of some answers concerning translation termination in mitochondria given in this study , still important questions remain , e . g . under which conditions ribosome-free ICT1 is produced to act on the ribosomal A-site .
55S ribosomes were prepared from pig liver mitochondria , as described previously [22] . The ribosome concentrations were calculated assuming 32 pmol/A260 for 55S ribosomes . HMICT1/pET15b was used as the E . coli expression vector for the N-terminally histidine-tagged ICT1 . The EST clone ( IOH11951 ) was purchased from Invitrogen . The coding sequence was amplified by PCR , using the primers 5′-CGGTGCCCACGCCATATGCTGCACAAGCAGAAAGACGGCACTG-3′ and 5′-GAGGCTCGAGTCAGTCCATGTCGACCCTCCTGCTT-3′ . The obtained coding sequences were cloned between the NdeI and XhoI sites of a modified pET15b vector ( Novagen ) , in which the original multi-cloning sites were modified in our laboratory . HMICT1/pET15b was transformed into E . coli Rosetta ( DE3 ) /pLysS . Cultures were induced with 100 µM isopropyl-1-thio-D-galactopyranoside at 18°C overnight . The protein was purified by Ni-NTA ( QIAGEN ) column chromatography . After histidine tag digestion using thrombin protease ( GE Healthcare ) , the protein was further purified by Q-Sepharose column chromatography ( GE Healthcare ) . The protein was concentrated to 5 mg/ml , divided into aliquots , flash-frozen , and stored at −80°C . The E . coli expression vectors for the ICT1 mutants were prepared by PCR , using HMICT1/pET15b . The primers for PCR were as follows: 5′-GTGGTCCTGGGTCGCAGAATGTGAAC-3′ and 5′-GTTCACATTCTGCGACCCAGGACCAC-3′ for GSQ , 5′-TGACTCGAGGGTACCCCGCGGGCGG-3′ and 5′-TCTCTTTTGTCTCAGCCTTTCCCGATTC-3′ for ΔC , 5′-ATCGCGGAGCCCGTGGCGCAGAAGATAGCCAT-3′ and 5′-ATGGCTATCTTCTGCGCCACGGGCTCCGCGAT-3′ for R116A , 5′-GAGCCCGTGCGGCAGGCGATAGCCATCACGCA-3′ and 5′-TGCGTGATGGCTATCGCCTGCCGCACGGGCTC-3′ for K118A , 5′-ATAGCCATCACGCATGCAAACAAGATCAACAG-3′ and 5′-CTGTTGATCTTGTTTGCATGCGTGATGGCTAT-3′ for K124A , 5′-ATCACGCATAAAAACGCGATCAACAGGTTAGG-3′ and 5′-CCTAACCTGTTGATCGCGTTTTTATGCGTGAT-3′ for K126A , 5′-AAAAACAAGATCAACGCGTTAGGAGAGTTGAT-3′ and 5′-ATCAACTCTCCTAACGCGTTGATCTTGTTTTT-3′ for R129A , and 5′-ATAGCCATCACGCATGCAAACGCGATCAACGCGTTAGGAGAGTTGATCC-3′ and 5′-GGATCAACTCTCCTAACGCGTTGATCGCGTTTGCATGCGTGATGGCTAT-3′ for α2 ( 124-126-129A ) . These mutants were expressed and purified as described above . The heteropolymeric MFV-mRNA and the MF-stop mRNA 5′-GGGAAAAGAAAAGAAAAGAAA-AUG-UUC-GUU/UAA-AAAAGAAAAGAAAAGAAAAUAUUGAAUU-3′ , containing three codons ( Met-Phe-Val or Met-Phe-stop , indicated with bold letters in mRNA sequence ) in the middle , respectively , were prepared by run-off transcription , as reported [23] . E . coli tRNAPhe was purchased from Sigma-Aldrich . The ac[3H]Phe-tRNAPhe ( N-acetyl-[3H]Phe-tRNAPhe ) was purified by reverse-phase HPLC on a Nucleosil 300-5 C4 column using a methanol gradient , as described previously [24] . Human RF1Lmt and RF1mt were expressed in E . coli and purified as described previously [9] . All complexes were prepared in a buffer containing 20 mM HEPES-KOH ( pH 7 . 6 ) , 4 . 5 mM Mg ( OAc ) 2 , 150 mM KOAc , 0 . 05 mM spermine , 2 . 0 mM spermidine , and 4 mM 2-mercaptoethanol . Ternary complexes were formed in a reaction volume of 50 µl , containing 20 pmol 55S ribosomes , 200 pmol MFV/MFstop mRNA and 20 pmol ac[3H]Phe-tRNAPhe . The reaction mix was incubated for 15 min at 37°C . A binding assay confirmed that approximately 0 . 9 pmol of ac[3H]Phe-tRNAPhe was bound to the ribosomes , at this point . The reactions were further incubated ( 75 µl total volume ) for 45 min at 25°C with 80 pmol of ICT1 , RF1Lmt or RF1mt . To stop the reaction , an equal volume of 1 N HCl was added to the reaction . The released ac[3H]Phe was extracted by ethyl acetate , and the amount of [3H]Phe incorporated into polypeptides was determined using a scintillation counter . The results were evaluated relative to the 100% value when all of the ac[3H]Phe-tRNAPhe was bound to the ribosome; i . e . , 0 . 9 pmol was hydrolyzed . The reactions in the absence of mRNA were also performed as above , with the initial incubation containing 55S ribosomes and ac[3H]Phe-tRNAPhe , except for the second incubation for 30 min at 25°C , where 16 pmol of ICT1 , RF1Lmt or RF1mt were added . Fig . 1B confirmed that acPhe-tRNA is programmed to P-site , and the stop codon is properly positioned to A-site on the ribosome , since RF1Lmt showed peptide-release activity with MF-stop ( UAA ) mRNA , but not with MFV mRNA . When the puromycin assay was performed using the ribosomal complex prepared in Fig . 1B , approximately 9 . 6 pmol of acPhe ( >650% ) was transferred to puromycin , while the binding assay confirmed that only 0 . 9 pmol of acPhe-tRNA ( 100% ) is bound to ribosome in the presence of mRNA . This indicates that acPhe-tRNA goes to P-site on the mRNA-containing ribosome as well as on the mRNA-free ribosome . Note that acPhe-tRNA on the latter ribosome can react with puromycin but not with RFs ( RF binds to ribosome in a stop-codon dependent manner . ) . Polysomes were prepared from the E . coli A19 strain , according to the procedures described previously [25] . The standard reaction mixtures ( 250 µl ) contained E . coli polysomes ( 2 . 0 A260 ) , RRFmt ( 15 µg ) , EF-G2mt ( 30 µg ) and peptide release factors ( RF1Lmt or RF1mt , 60 µg; ICT1 , 50 µg; 10 µM puromycin ) . The mixtures were incubated at 30°C for 20 min in buffer ( 10 mM Tris-HCl [pH 7 . 5] , 80 mM NH4Cl , 8 . 2 mM MgSO4 , 1 mM DTT , and 0 . 5 mM GTP ) and fractionated on 15%–30% ( w/v ) sucrose gradients ( containing 10 mM Tris-HCl [pH 7 . 4] , 80 mM NH4Cl , 8 . 2 mM MgSO4 , and 1 mM DTT ) by centrifugation at 39 , 000 rpm for 2 hour , using an SW41Ti rotor ( Beckman Coulter ) . The gradients were recovered from top to bottom , using a density gradient fractionator ( Towa Labo , Model 152–001 ) while monitoring the absorbance at 260 nm . DNA templates were prepared as follows . Using the plasmid pURE1 ( Post Genome Institute Co . , Ltd . ) , DNA fragments were amplified by PCR with the T7 promoter primer 5′-GCGCGTAATACGACTCACTATAG-3′ and 3′ primer 5′-GATCCCTAGAACAGTTAGAACAGGAAGAACATATGATATCTCCTTCTTAAAGTT-3′ ( stop ) , 5′-GATCCCTAGAACAGTCTGAACAGGAAGAACATATGATATCTCCTTCTTAAAGTT-3′ ( stall ) and 3′ primer 5′-GAACAGGAAGAACATATGATATCTCCTTCTTAAAGTT-3′ ( nonstop ) . The corresponding mRNAs used in Fig . 2 have the following sequences . Nonstop mRNA: 5′-GGGAGACCACAACGGUUUCCCUCUAGAAAUAAUUUUGUUUAACUUUAAGAAGGAGAUAUCAUAUGUUCUUCCUGUUC-3′ ; stop ( UAA ) : 5′-GGGAGACCACAACGGUUUCCCUCUAGAAAUAAUUUUGUUUAACUUUAAGAAGGAGAUAUCAUAUGUUCUUCCUGUUCUAACUGUUCUAGGGAUC-3′ ; and stall ( AGA ) : 5′-GGGAGACCACAACGGUUUCCCUCUAGAAAUAAUUUUGUUUAACUUUAAGAAGGAGAUAUCAUAUGUUCUUCCUGUUCAGACUGUUCUAGGGAUC-3′ . ( the underlined sequence encodes the MFFLF short polypeptide ) . [14C]Methionine-labeled fMet-tRNA was prepared as described previously [25] . The system was reconstituted as reported previously [11] , [25] , with slight modifications . Briefly , RRF was omitted from the standard system , and EF-G2mt ( 0 . 8 µg ) and RRFmt ( 0 . 2 µg ) were included as the recycling factors . E . coli RFs were omitted , and RF1Lmt ( 0 . 5 µg ) , RF1mt ( 2 µg ) and ICT1 ( 1 . 1 µg ) were included . Only phenylalanyl-tRNA synthetase and leucyl-tRNA synthetase were included as aminoacyl-tRNA synthetase sources . Instead of amino acid mixtures , phenylalanine and leucine were used , and [35S]methionine was omitted from the system . The other components were the same as those described previously . The reactions ( 50 µL ) were started by the addition of 0 . 2 pmol of PCR-amplified DNA template and 100 pmol of f[14C]Met-tRNA . After an incubation at 37°C for the indicated time periods , aliquots were withdrawn and added to an equal volume of 1 N HCl , to stop the reaction . The translated polypeptides were extracted with ethyl acetate , and the incorporation of [14C] methionine into polypeptides was determined using a scintillation counter . For the SDG analysis with E . coli 70S ribosomes , reaction mixtures ( 250 µl ) containing 25 pmol 70S ribosomes and 250 pmol ICT1 , in buffer A ( 10 mM Tris-HCl [pH 7 . 5] , 80 mM NH4Cl , 8 . 2 mM MgSO4 and 1 mM DTT ) , were incubated at 30°C for 20 min and fractionated on 15%–30% ( w/v ) sucrose gradients , in the same manner as the polysome breakdown assay . In the analysis with 55S mitoribosomes , reaction mixtures ( 250 µl ) containing 75 pmol 55S ribosomes and 1 , 250 pmol N-terminal histidine tagged ICT1 were incubated as in the assay with 70S ribosomes and fractionated on 15%–30% ( w/v ) sucrose gradients by centrifugation at 39 , 000 rpm for 5 . 5 hour , using an SW41Ti rotor ( Beckman Coulter ) . The fractions were analyzed by Western blotting . For the dot-blot analysis , reaction mixtures ( 50 µl ) containing 12 . 5 pmol E . coli ribosome and the indicated amount of ICT1 in buffer A were incubated at 30°C for 20 min , loaded on a Microcon-YM100 column ( Millipore ) , and centrifuged at 3 , 000× g at 4°C for 5 minutes . The columns were washed twice with 100 µl of buffer A . For sample recovery , the columns were incubated with 50 µl buffer A at room temperature for 1 minute . The samples were collected in fresh tubes from the inverted columns by centrifugation at 10 , 000× g at 4°C for 5 minutes , with rinses of 100 µl of buffer A . After the recovered samples were brought up to a volume of 500 µl with buffer A , 50 µl aliquots containing EDTA ( f . c . 20 mM ) were used as the dot-blot loading samples . For quantification , purified recombinant ICT1 was used as loading samples for a standard curve . 55S ribosomes ( 46 pmol in 100 µl final volume ) were mixed with a 5-fold excess of ICT1 , in buffer containing 20 mM Hepes-KOH ( pH 7 . 6 ) , 8 . 2 mM MgSO4 , and 80 mM NH4Cl . The mixtures were incubated at 30°C for 20 min . The reactions were supplemented with the amino-reactive crosslinking agent BS3 ( Bis[sulfosuccinimidyl]suberate , Thermo Scientific ) to a final concentration of 4 mM , and were incubated at 25°C for 30 min . BS3 crosslinking was stopped by the addition of SDS-PAGE loading dye . The samples were analyzed by Western blotting . The polyclonal antisera for ICT1 were generated in our laboratory , by injecting rabbits with the purified recombinant proteins . | Mammalian mitochondrial ICT1 , a bacterial ArfB homolog , is interestingly an integral component of the mitoribosome ( MRPL58 ) . The mechanism of ribosome rescue by this factor was obscure and is addressed here . Utilizing a homologous mitochondria system of purified components we demonstrate that the integrated ICT1 has no rescue activity , as opposed to a previous model . Rather , purified ICT1 added to mitoribosomes has a general rescue activity; it recycles ribosomes stalled at the end or in the middle of mRNAs and can even hydrolyze peptidyl-tRNA bound to non-programmed ribosomes . These results further imply that ICT1 can function in the translation termination at non-standard stop codons AGA/G in mammalian mitochondria . Our data challenge a previous model claiming that RF1Lmt/mtRF1a is responsible for the translation termination at non-standard stop codons . A mutational study indicates that the unique insertion sequence in ICT1 is essential for peptide release . The function of RF1mt , another member of the class1 RFs in mammalian mitochondria , was also examined and is discussed . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"molecular",
"cell",
"biology",
"biochemistry",
"biology",
"and",
"life",
"sciences",
"cell",
"biology"
] | 2014 | Ribosome Rescue and Translation Termination at Non-Standard Stop Codons by ICT1 in Mammalian Mitochondria |
Enhancers and promoters often contain multiple binding sites for the same transcription factor , suggesting that homotypic clustering of binding sites may serve a role in transcription regulation . Here we show that clustering of binding sites for the transcription termination factor TTF-I downstream of the pre-rRNA coding region specifies transcription termination , increases the efficiency of transcription initiation and affects the three-dimensional structure of rRNA genes . On chromatin templates , but not on free rDNA , clustered binding sites promote cooperative binding of TTF-I , loading TTF-I to the downstream terminators before it binds to the rDNA promoter . Interaction of TTF-I with target sites upstream and downstream of the rDNA transcription unit connects these distal DNA elements by forming a chromatin loop between the rDNA promoter and the terminators . The results imply that clustered binding sites increase the binding affinity of transcription factors in chromatin , thus influencing the timing and strength of DNA-dependent processes .
An intriguing question for understanding protein-DNA recognition is how low-abundant transcription factors recognize their target sites in genomic DNA [1] , [2] . Empirical studies revealed that regulatory regions , such as enhancers and promoters , comprise modular units of a few hundred base pairs that harbour multiple binding sites for the same transcription factor . Such ‘homotypic clustering sites’ ( HTCs ) have been identified in 2% of the human genome , being enriched at promoters and enhancers [3] . HTCs have been shown to play a role in Drosophila development , regulating early patterning genes [4]–[6] . Genome-wide binding analyses in yeast have demonstrated that the occupancy of transcription factors is higher at clustered binding sites compared to single ones [7] . Studies in mammalian cells have shown that clustering of binding sites facilitate the cooperative binding of nuclear receptors to their target sites in vivo , suggesting that HCTs coordinate the recruitment of transcription initiation factors [8]–[10] . Alternatively , cooperative binding could arise through indirect effects , e . g . by changing the accessibility of neighbouring binding sites in chromatin [11] . To assess the functional relevance of homotypic clustering of transcription factor binding sites , we studied the 3′-terminal region of murine rRNA genes ( rDNA ) , which contains ten binding sites ( T1–T10 ) for the transcription termination factor TTF-I . Binding of TTF-I to the terminator elements is required to stop elongating RNA polymerase I ( Pol I ) and termination of pre-rRNA synthesis occurring predominantly at the first terminator T1 [12]–[15] . In addition to the downstream terminators , there is a single TTF-I binding site , termed T0 , located 170 bp upstream of the transcription start site [16] . Binding of TTF-I to this site is required for efficient transcription initiation and for the recruitment of chromatin remodelling complexes that establish distinct epigenetic states of rRNA genes . The interaction of TTF-I with CSB ( Cockayne Syndrome protein B ) , NoRC ( Nucleolar Remodeling Complex ) , or NuRD ( Nucleosome Remodeling and Deacetylation complex ) , respectively , has been shown to recruit histone modifying enzymes which lead to the establishment of a specific epigenetic signature that characterizes active , silent or poised rRNA genes [17]–[20] . TTF-I has been shown to oligomerize in vitro and to link two DNA fragments in trans [21] . These characteristics enable TTF-I bound to the upstream binding site T0 and the downstream terminators T1–T10 to loop out of the pre-rRNA coding region [22] , [23] . Formation of a chromatin loop facilitates re-initiation and increases transcription initiation rates at the rRNA gene [22] , [24] . TTF-I is a multifunctional protein that is not only essential for transcription termination , but also directs efficient rDNA transcription , mediates replication fork arrest [25] , establishes specific epigenetic features and determines the topology of rDNA . The conservation of multiple TTF-I binding sites downstream of the pre-rRNA coding region raises the question whether homotypic clustering of terminator elements is functionally relevant . Here we demonstrate that HTCs serve a chromatin-specific function . Packaging into chromatin increases the binding affinity of TTF-I to clustered terminator elements , augments the efficiency of transcription termination , enhances transcription initiation , and changes the higher-order structure of rRNA genes . The homotypic clusters at the rRNA gene coordinate the timing of molecular events , coordinating transcription termination and initiation and the occurrence of higher-order chromatin domains , suggesting an important chromatin-dependent role for clustered binding sites in the genome .
The rDNA terminators in human and mouse exhibit an overall similar structure , containing 10 to 11 TTF-I binding sites in close proximity ( Fig . S1A ) . We focused on the murine rDNA terminator , which comprises 10 termination sites ( T1–T10 ) spaced by 18–123 bp , preventing the accommodation of nucleosomes in between the TTF-I binding sites . The consensus sequence of these TTF-I binding sites share almost perfect sequence identity within the core motif GGTCGACCAG , while the surrounding nucleotides vary slightly ( Fig . 1A ) . In electrophoretic mobility shift assays ( EMSAs ) recombinant TTF-I bound with comparable affinity to all terminators assayed ( data not shown ) . The DNA binding affinity of TTF-I was quantified by microscale thermophoresis , recording changes of nucleoprotein complex mobility in a small temperature gradient [26] . By titrating a wide range of TTF-I∶DNA ratios , the binding constant of TTF-I to free Sal-box DNA was determined to be 0 . 5 µM ( Fig . 1B ) , a relatively low DNA binding affinity which is one order of magnitude lower than the KD of other transcription factors [27]–[29] . In vitro transcription assays on a circular minigene comprising the rDNA promoter fused to a single termination site ( pMrSB ) yielded long read-through transcripts in the absence of TTF-I . The addition of recombinant TTF-I led to the synthesis of terminated transcripts whose lengths correspond to the distance from the transcription start site to the terminator T1 ( Fig . 1C ) . If the template contained all ten terminators ( pMrT1-T10 ) , both read-through transcripts and a heterogeneous population of transcripts randomly terminated at any of the TTF-I binding sites were synthesized due to sub-saturating TTF-I levels in the extract ( Fig . 1D ) . In the presence of increasing concentrations of recombinant TTF-I the amount of transcripts stopped at terminator T1 progressively increased ( Fig . 1D , lanes 1–8 and Fig . S2 ) . Thus , TTF-I binds to all sites with similar affinity and randomly terminates transcription until at saturating concentrations TTF-I occupies all ten terminators . A strikingly different result was obtained on rDNA templates assembled into chromatin with an extract from Drosophila embryos [30] ( Figure S1B ) . Consistent with Pol I transcription on chromatin requiring binding of TTF-I to the promoter-proximal terminator T0 and ATP-dependent chromatin remodelling [31] , [32] , transcription was repressed in the absence of TTF-I ( Fig . 1D , lane 9 ) . The addition of TTF-I relieved transcriptional repression , yielding only a single RNA species of 686 nt . On chromatin templates , already lowest TTF-I concentrations terminated transcription specifically at T1 ( Fig . 1D , lanes 10–16 and Fig . S2 ) . The result suggests that transcription in chromatin is only initiated when the termination sites are set , meaning that the TTF-I binding site at the promoter is only bound after sequestering TTF-I at the terminator . The qualitative difference between transcription on free DNA and chromatin templates indicates that on chromatin templates TTF-I either binds preferentially to T1 or the overall binding affinity of TTF-I to all terminator sites is increased in chromatin . Next , we performed electrophoretic mobility shift assays ( EMSAs ) and DNase I footprinting experiments to compare TTF-I binding to free DNA and chromatin . Consistent with the transcription data on free DNA , EMSAs on terminator DNA fragments containing more than one TTF-I binding sites yielded heterogeneous nucleoprotein complexes , reflecting binding to each binding site with similar affinity ( Fig . 2A ) . On chromatin templates , DNase I footprinting experiments demonstrate that TTF-I simultaneously bound to all terminator binding sites ( Fig . 2B ) . Together with the transcription results on chromatin templates , this suggests that homotypic clustering of target sites increases the binding affinity of TTF-I to chromatin . To compare the binding affinity of TTF-I to free DNA and reconstituted chromatin , we performed DNase I footprinting assays , monitoring DNase I cleavage sites by primer extension which allows simultaneous analysis of TTF-I occupancy at the promoter and terminator ( s ) ( Fig . 2C ) . TTF-I binding can be observed by the disappearance of a DNase I sensitive site that is apparent within the TTF-I binding sites of free DNA and reconstituted chromatin ( Fig . 2B and C ) . In agreement with the binding studies and the in vitro transcription experiments , TTF-I binds on free DNA to the promoter-proximal terminator T0 and the downstream terminators with similar affinity ( Fig . 2C , compare lanes 2–4 and lanes 9–11 ) . On chromatin templates , TTF-I binding to the upstream site T0 is comparable to its binding to free DNA ( Fig . 2C , upper panel ) . However , on chromatin templates TTF-I binds with higher affinity to the clustered sites , fully occupying all terminator sites at low protein concentrations ( Fig . 2C , lower panel ) . Significantly , TTF-I occupied the binding sites at the terminators prior to the promoter-proximal site ( compare lanes 5–7 and 12–14 ) , showing a specific role of chromatin and binding site clustering for increasing the binding affinity of TTF-I . The sequential binding of TTF-I , first to the terminators and then to the gene promoter in chromatin was also confirmed using a different method . Affinity purification of either TTF-I bound free DNA or chromatin revealed binding of TTF-I to the gene terminators reconstituted into chromatin already at concentrations one order of magnitude lower than with the gene promoter ( Fig . S3 ) . Like in the footprinting assay , this effect was not detectable using free DNA , where both TTF-I binding regions were occupied with similar affinity . Apparently , the clustered arrangement of binding sites increases the affinity of TTF-I , thus promoting the association of TTF-I with the downstream terminators T1–10 prior to the upstream site T0 , a process that appears to be essential for both TTF-I dependent transcription activation and transcription termination . To study the functional relevance of clustered sites in vivo , we transfected CHO cells with reporter plasmids containing the murine Pol I promoter , an internal ribosomal entry site ( IRES ) , Firefly luciferase cDNA and either no terminator ( pTΔ ) or one ( pT1 ) , two ( pT2 ) or ten ( pT10 ) termination sites . As shown in Figure 3A , the presence of one or two terminators ( pT1 and pT2 ) enhanced transcription of the luciferase reporter 8- to 12-fold compared to the terminator-deficient vector . The presence of ten termination sites ( pT10 ) decreased luciferase activity , presumably due to squelching of endogenous TTF-I . In support of this view , transient overexpression of TTF-I ( pTTFΔN348 ) revealed a linear correlation between the number of terminators and reporter gene activity , showing further transcriptional enhancement by the pT10 construct ( Fig . 3B and Fig . S4B ) . Additional controls revealed that the stimulatory effect depends on the TTF-I binding sites at the promoter and the terminator ( Fig . S4A ) and co-transfection of pTTFΔN470 revealed that the chromatin-binding domain of TTF-I is required . pTTFΔN470 represents a deletion mutant that is capable of binding to its target sites on free DNA but not on chromatin [31] . Therefore , pTTFΔN470 cannot activate transcription in a chromatin context [31] and transfection of this construct did not further activate transcription of the pT10 construct ( Fig . 3C and Fig . S4B ) . The control shows that chromatin-specific activities of TTF-I are required for efficient transcriptional activation . Notably , there was no luciferase expression using reporters with TTF-I binding site ( s ) in the reverse orientation ( pT1R , pT10R ) , supporting the importance of the topological arrangement of the HCTs for efficient Pol I transcription . To examine whether the number of terminators affects gene activity and/or the spatial organization of rDNA in a genomic context , we generated stable cell lines that harbour a single copy of mouse rRNA minigenes , either containing only T1 ( CHO-pT1 ) or all ten terminators ( CHO-pT10 ) ( Fig . S5 ) . Using the Flip-In system we generated comparable rRNA minigene lines , integrated at the same genomic site of CHO cells . This strategy allows us to rule out effects of inefficient chromatin packaging and minigene dosage in transfection experiments . The nuclear localisation of the ectopic rDNA was not affected , as 3D immuno-FISH experiments revealed that comparable number of rDNA was associated with the nucleoli in the stable cell lines ( 33 of 104 alleles were associated with nucleoli in CHO-pT1 and 50 of 160 alleles in CHO-pT10 cells; Fig . S5A , B ) . RNA FISH experiments confirmed that all cell lines were transcriptionally active ( Fig . 3D ) . Expression analysis of the rRNA minigene by qRT-PCR and reporter assays revealed that both the level of the ectopic pre-rRNA and the Pol I-driven luciferase activity were increased in CHO-pT10 compared to CHO-pT1 cells ( Fig . 3E and Fig . S5C ) , reinforcing the activating role of clustered termination sites in rDNA transcription . To decipher the molecular mechanism underlying HTC-driven transcriptional activation , we compared transcription factor occupancy within the stable cell lines , containing single rDNA minigenes with either one ( CHO-pT1 ) or ten terminators ( CHO-pT10 , Fig . 4A ) . As shown in Figure 4B , binding of Pol I and UBF was enhanced at the promoter , the transcribed region and the terminators of CHO-pT10 compared to CHO-pT1 cells . In addition , we observed increased binding of TBP to the promoter of CHO-pT10 cells , demonstrating that augmented rDNA transcription is a direct consequence of enhanced transcription initiation and polymerase occupancy . Pol I enrichment downstream of the terminator region was reduced in CHO-pT10 cells , consistent with clustered TTF-I binding sites promoting efficient termination . Similar results were obtained with the transient transfection of the constructs ( Fig . S6 ) . Active rRNA genes are known to form chromatin loops , connecting the promoter with the terminator to promote recycling of Pol I [22] , [23] , [33] . To examine whether multiple terminators facilitate loop formation , we determined the occupancy of TBP at the terminator in the stable cell lines CHO-pT1 and CHO-pT10 ( Fig . 4B lower panel and 4C ) . The close proximity of a protein to DNA results in crosslinking and co-purification of the DNA , even though the factor does not directly contact the DNA at this site . Such binding events indicate the close spatial proximity of distant DNA sites , comparable to 3C assays [34] . Obviously , TBP was found to be associated with the promoter of CHO-pT1 and CHO-pT10 as part of the initiation complex , while no binding was observed in the transcribed region ( Fig . 4B , TBP panel ) . Strikingly , TBP was also enriched at the terminator of CHO-pT10 but not CHO-pT1 cells , suggesting that clustered TTF-I binding sites are in close proximity with the gene promoter . Consistent with multiple terminators facilitating initiation of transcription , TBP and Pol I occupancy was about 4-fold higher in CHO-pT10 compared to CHO-pT1 ( Fig . 4B , TBP panel ) . To exclude the possibility that clustered TTF-I binding sites on their own recruit TBP to the 3′-end of rRNA genes , we examined TBP occupancy on a reporter plasmid in which the ten terminators were fused to a Pol II promoter . TBP was enriched at the Pol II promoter but close to background at the terminator ( Fig . 4C ) , emphasizing the importance of TTF-I binding sites at both elements , the promoter and the terminators , to form chromatin loops . Homo- and heterotypic clusters of transcription factor binding sites were shown to mark potential regulatory regions with enhancer function [35]–[39] characterized by eukaryotic histone marks like H3K27ac , which is involved in long-range chromatin interactions [40] . As the repetitive rDNA is left out of standard ChIP-Seq analyses , we artificially added a single mouse rDNA repeat to the current mouse genome version mm9 and mapped ChIP-Seq data of H3K27ac , H3K27me3 , H3K4me1 , H3K4me2 and H3K4me3 to this expanded reference genome ( Fig . 5A ) . We observed enrichment of H3K27ac and H3K4me2 in the terminator and promoter region of murine rDNA , enforcing our previous results and confirming that the homotypic cluster of TTF-I binding sites represents an active enhancer element . In contrast , H3K27me3 , commonly associated to repressed genes , is depleted at the terminator compared to the rDNA gene body . Therefore , the mouse rDNA terminator exhibits a histone modification profile typical for enhancer elements involved in Pol II transcription .
Clustering of transcription factor binding sites , comprising either multiple binding sites for the same factor ( homotypic clustering ) or different DNA binding motifs ( heterotypic clustering ) , is an important regulatory feature of eukaryotic gene expression , about 62% of transcription factor genes and 66% of developmentally regulated genes comprising clustered binding sites in vertebrates [3] . Therefore , this feature has been widely used for computational prediction . In Drosophila , predicted HTCs are present in more than 70% of regulatory regions and have been suggested to function as developmental enhancers [6] , [41] . Clustered binding sites are suggested to exert a positive effect on transcription by either of the following mechanisms . They could increase the local concentration of transcription factors or facilitate multiple interactions with components of the transcription machinery . Alternatively , they could provide functional redundancy [37] , [42] , allowing cooperative binding of the factors through interactions among the multiple binding sites or indirectly through multiple interactions with the transcriptional machinery [10] , [43]–[46] . Here , we have uncovered a novel chromatin-based mechanism underlying HTC-directed transcriptional activation . We show that packaging into chromatin converts multiple low-affinity terminators downstream of the rDNA transcription unit into a high-affinity binding platform for TTF-I . This preferential binding of TTF-I to the downstream terminators is a prerequisite for TTF-I binding to the promoter-proximal binding site , connecting the promoter with the terminator to allow efficient recycling of Pol I . Cooperative binding of proteins has been shown to disrupt nucleosomes , thereby increasing the accessibility of transcription factors to regulatory sites [47] , [48] . Our data reveal an alternative mechanism that increases the affinity of transcription factors . We show that binding of TTF-I to its target sites in chromatin is higher than to free DNA , suggesting that a specific nucleosomal arrangement or interactions with histones may trigger cooperative binding of TTF-I . Thus , HTCs attract transcription factors to functionally relevant sites , avoiding binding to single target sites in the genome . High-affinity binding of TTF-I to clustered termination sites will ensure loading of the downstream terminators ( T1–T10 ) prior to TTF-I binding to the promoter-proximal binding site T0 in vivo . Sequential binding of TTF-I to the 3′- and 5′-end of the rDNA transcription unit will ensure that transcription initiation will take place exclusively at rRNA genes that are associated with TTF-I and will be properly terminated . In addition , a direct interaction between the promoter and the terminator is only established when the terminator comprises several TTF-I binding sites . This mode of binding and the formation of an intragenic loop may serve two functions . First , it links the terminator with the respective transcription unit to be activated . Second , it enhances transcription at genes associated with TTF-I by forming a highly active ribomotor structure [22] , [49] . Thus , homotypic clustering of TTF-I binding sites coordinates transcription initiation and termination , thereby affecting both the timing and the efficiency of rDNA transcription . It is well established that gene activation by a distal regulatory element correlates with long-range interactions between enhancer ( s ) and gene promoters by factor-mediated formation of chromatin loops [50] . With regard to human and rat rRNA genes , previous studies suggested a role for TBP and c-Myc in loop formation at active rRNA genes [23] , [33] . However , genome-wide ChIP-Seq data did not reveal significant enrichment of c-Myc- and TBP at the terminator ( Fig . S7B ) . Moreover , murine rRNA genes lack clustered E-boxes ( Fig . S7A ) , and therefore the participation of c-Myc in loop formation is not very likely . Similar loop mechanisms were shown for RNA polymerase II transcribed genes , suggesting a common theme involving the interaction of promoters with transcription termination regions that enhance the transcriptional activity and gene regulation [51] . Active enhancer elements are characterized by eukaryotic histone marks , e . g . H3K27ac or H3K4me1 , which are involved in long-range chromatin interactions [40] . Notably , our integrative genomic analysis revealed characteristic enrichment of histone marks at the terminator , which can be observed in human as well [52] . The results support our finding that the homotypic cluster of TTF-I binding sites displays all hallmarks of a functional enhancer , such as distal location , presence of HTCs , regulatory histone marks and the potential to exert gene activation by direct , protein-mediated DNA loops . Chromatin-dependent high-affinity binding of TTF-I to the clustered binding sites adds a further regulatory level on the enhancer function , i . e . , coordination of transcription termination and initiation .
Histidine-tagged full-length TTF-I and the deletion mutants TTFΔN210 and TTFΔN348 were purified on a heparin column ( Bio-Rad ) , followed by purification with Ni-NTA agarose according to the manufacturer's instructions ( Qiagen ) . For microscale thermophoresis experiments , 50 nM of fluorescently labelled DNA oligonucleotides were incubated with 5 nM–2 . 4 µM of protein for 10 min at 30°C in 80 mM Tris-HCl ( pH 7 . 6 ) , 80 mM KCl , 0 . 2 mM EDTA , 5 mM MgCl2 , 10% glycerol and 0 . 05% IGEPAL CA-630 . Affinity measurements were carried out in a Monolith NT . 015T ( NanoTemper Technologies ) as described [26] . 300 ng of chromatin reconstituted with Drosophila extract was digested with 1 . 5 U of MNase ( Sigma ) for 40 s in 10 mM Tris-HCl ( pH 7 . 6 ) , 80 mM KCl , 1 . 5 mM MgCl2 , 10% glycerol , 0 . 5 mM ATP , and 200 ng/µl BSA . Reactions were then stopped by the addition of 0 . 2 volumes of 4% SDS , 100 mM EDTA , 1 µg of glycogen , 10 µg of proteinase K . Purified DNA was analysed by a single round of PCR ( denaturation , 5 min at 95°C; annealing , 2 min at 56°C; extension , 1 min at 72°C ) using radioactively labelled oligonucleotides that hybridize to the rDNA promoter or terminator . Primer extension fragments were resolved on 8% sequencing gels and visualized by autoradiography . Transcription experiments were performed on pMrWT-T , a template comprising the murine rDNA promoter ( from −170 to +155 with regard to the transcription start site ) fused to a 3 . 5 kb 3′-terminal rDNA fragment ( BamHI/EcoRV Fragment ) harbouring all ten terminators . ( T1–T10 ) . The promoter and the terminator elements are separated by 686 bp . Transcription reactions were performed as described [53] . CHO and CHO Flp-In cells ( Invitrogen ) were grown in DMEM ( GIBCO ) supplemented with 10% FBS , 100 U/ml penicillin and 100 µg/ml streptomycin . For transient transfections , 200 . 000 cells were transfected with 1 µg of plasmid DNA . Prior to transfection of the CHO Flp-In cells , 100 µg/ml zeocin ( Invitrogen ) was added to the medium and for transfection 0 . 25 , 0 . 5 or 1 . 0 µg of the rRNA reporter construct and the flipase encoding plasmid pOG44 ( Invitrogen ) in a ratio of 1∶9 were used . During the selection process , 500 µg/ml of hygromycin ( PAA ) was added to the medium; afterwards the stable cell lines were passaged with 250 µg/ml of hygromycin . Transiently transfected rRNA minigenes [22] contain mouse rDNA ( BK000964 ) sequences from position −1932 to +181 , an IRES , the firefly luciferase gene , and rDNA terminator regions from position +13169 to +15278 ( T10 constructs ) in a pGL3-Basic vector ( Promega ) . Plasmids for genomic integration contain in addition the enhancer/promoter regions from position −2148 to +181 cloned into a pcDNA5-FRT vector ( Invitrogen ) . Cells were transfected with Pol I driven firefly luciferase reporter constructs and a Pol II renilla luciferase control plasmid , pRL-TK ( Promega ) . TTF-I co-transfections were performed with the expression vectors TTFΔN348-EGFP or TTF-IΔN470-EGFP in a TTF-I∶reporter ratio of 10∶1 . Protein expression was monitored by Western Blot analysis . Reporter gene measurements were performed using the Dual Luciferase Reporter Assay System ( Promega ) according to the manufacturer's instructions using a single-tube luminometer ( Stratec Biomedical Systems ) . RNA isolation was performed with the NucleoSpin RNA II kit ( Macherey-Nagel ) . Purified RNA ( 500 ng ) was used for cDNA preparation with the iScript Select kit ( Biorad ) . To determine the number of integration sites , genomic DNA was isolated by cells lysis ( 1% SDS , 50 mM Tris-HCl ( pH 8 . 0 ) , 20 mM EDTA and 250 µg of RNase A ) , the addition of proteinase K and incubation at 37°C o . n . The supernatant was precipitated with ethanol and ammonium acetate . Quantitative real-time PCR was performed in a Rotor-Gene cycler ( Qiagen ) using a HotStar master mix containing SYBR green ( Qiagen ) . Primer sequences and annealing temperatures are listed in the in Table S2 . Fold inductions were calculated using the comparative quantitation software ( Qiagen ) . Post-PCR melting curves and agarose gels of PCR products ( Fig . S3F ) were used to assess the quality of primer pairs . Cells were transfected with 10 µg of DNA and cross-linked with 1% formaldehyde for 10 min ( α-Pol I and α-UBF ChIPs ) or 10 mM DMA for 30 min +1% formaldehyde for 10 min ( α-TBP ) at RT . The reactions were quenched with 125 mM glycine . Cells were washed twice in ice-cold PBS and the cell pellets were lysed in SDS lysis buffer ( 1% SDS , 50 mM Tris-HCl pH 8 . 0 , 20 mM EDTA , protease inhibitors ) . Chromatin was sheared in a Biorupter sonicator ( Diagenode ) to fragments of 400–1000 bp in length . The samples were diluted in IP dilution buffer ( 20 mM Tris-HCl , 2 mM EDTA , 1% Triton X-100 , 150 mM NaCl , pH 8 . 0 , protease inhibitors ) . Paf53 antibody for Pol I detection and the pre-serum were obtained from the Grummt lab [54] . Antibodies targeting RPA194 ( sc-28714 ) , UBF ( sc-9131 ) , TBP ( sc-273 ) and normal rabbit IgG ( sc-2027 ) were purchased from Santa Cruz . Antibodies ( 5 µg ) and chromatin were incubated on a rotating wheel at 4°C o . n . Pre-blocked Protein-G sepharose ( 500 µg/ml sonicated salmon sperm DNA and 100 µg/ml BSA in IP dilution buffer ) was added to isolate the immune-complexes and incubated for 2 h at 4°C . Beads were washed twice with IP dilution buffer , once with high salt buffer ( 20 mM Tris-HCl , 2 mM EDTA , 1% Triton X-100 , 150 mM NaCl , pH 8 . 0 ) , LiCl buffer ( 0 . 25 M LiCl , 1% NP40 , 1% Deoxycholate , 1 mM EDTA , 10 mM Tris-HCl , pH 8 . 0 ) and twice with TE buffer ( 10 mM Tris-HCl , 1 mM EDTA pH 8 . 0 ) . Elution was performed using 250 µl of 1% SDS , 0 . 1 M NaHCO3 . RNase A was added to a concentration of 100 µg/ml and incubated for 2 . 5 h at 37°C . Following the Proteinase K digestion ( 100 µg/ml , 2 . 5 h at 37°C ) , reverse crosslinking was carried out at 64°C o . n . DNA was isolated by phenol/chloroform/isoamylalcohol extraction and precipitated with ethanol and sodium acetate . Fluorescence in situ hybridizations on metaphase chromosome spreads and on interphase nuclei combined with nucleolar immunostaining were performed as described [55] . For RNA FISH , cells grown on coverslips were fixed for 10 min at room temperature with 3 . 7% formaldehyde/5% acetic acid/0 . 5% ( w/v ) NaCl , washed twice with 1× PBS , once in 50 mM NH4Cl/1× PBS pH 7 . 4 , and once in 1× PBS . Coverslips were then transferred to 70% ethanol and incubated o . n . at 4°C . Before hybridization , coverslips were rehydrated in 2× SSC/50% formamide for 15 min at RT . Hybridization mixtures were added for o . n . incubation at 37°C . Post-hybridization washes were carried out as follows: 2×25 min at 37°C in 50% formamide/2× SSC and 2×5 min in 2× SSC at RT . The subsequent immunostaining , DNA staining and mounting was performed as in interphase DNA FISH experiments . Nick-translated , biotin-labeled pcDNA5-FRT-rRNA reporter served as hybridization probe in all experiments . A custom build of the mm9 assembly was generated by replacing unsequenced bases at the 5′-end of chromosome 18 with a murine rDNA repeat ( GenBank accession no . BK000964 ) . We used Bowtie [56] to align published ChIP-seq data sets of 3T3L1 and MEL cells ( for details see Table S1 ) to the custom assembly using ‘–best -k 1’ settings . Input-normalized bedGraph files were generated using the makeUCSCfile . pl script contained in the HOMER software suite ( http://biowhat . ucsd . edu/homer/ , [57] ) using standard settings . | The sequence-specific binding of proteins to regulatory regions controls gene expression . Binding sites for transcription factors are rather short and present several million times in large genomes . However , only a small number of these binding sites are functionally important . How proteins can discriminate and select their functional regions is not clear , to date . Regulatory loci like gene promoters and enhancers commonly comprise multiple binding sites for either one factor or a combination of several DNA binding proteins , allowing efficient factor recruitment . We studied the cluster of TTF-I binding sites downstream of the rRNA gene and identified that cooperative binding to the multimeric termination sites in combination with low-affinity binding of TTF-I to individual sites upstream of the gene serves multiple regulatory functions . Packaging of the clustered sites into chromatin is a prerequisite for high-affinity binding , coordinated activation of transcription and the formation of a chromatin loop between the promoter and the terminator . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2013 | Chromatin-Specific Regulation of Mammalian rDNA Transcription by Clustered TTF-I Binding Sites |
Transcription factor binding site ( s ) ( TFBS ) gain and loss ( i . e . , turnover ) is a well-documented feature of cis-regulatory module ( CRM ) evolution , yet little attention has been paid to the evolutionary force ( s ) driving this turnover process . The predominant view , motivated by its widespread occurrence , emphasizes the importance of compensatory mutation and genetic drift . Positive selection , in contrast , although it has been invoked in specific instances of adaptive gene expression evolution , has not been considered as a general alternative to neutral compensatory evolution . In this study we evaluate the two hypotheses by analyzing patterns of single nucleotide polymorphism in the TFBS of well-characterized CRM in two closely related Drosophila species , Drosophila melanogaster and Drosophila simulans . An important feature of the analysis is classification of TFBS mutations according to the direction of their predicted effect on binding affinity , which allows gains and losses to be evaluated independently along the two phylogenetic lineages . The observed patterns of polymorphism and divergence are not compatible with neutral evolution for either class of mutations . Instead , multiple lines of evidence are consistent with contributions of positive selection to TFBS gain and loss as well as purifying selection in its maintenance . In discussion , we propose a model to reconcile the finding of selection driving TFBS turnover with constrained CRM function over long evolutionary time .
Gene expression in eukaryotes is generally controlled by transcriptional enhancers , also called cis-regulatory modules ( CRM ) , which are short regions in the genome consisting of a cluster of transcription factor binding sites ( TFBS ) spaced by intervening sequences ( spacers ) . Individual TFBS have been shown repeatedly to be required for CRM function , yet surprisingly they evolve rapidly and are frequently gained and lost in evolution , attributes that have been demonstrated for a large number of CRM and transcription factors [1]–[5] . These observations pose a challenge to understanding the forces driving the process , especially in cases where CRM function has been preserved despite sequence and structural divergence [6]–[8] . The gain or loss of a TFBS is unlikely to be functionally irrelevant , as repeatedly shown in TFBS knockout experiments [9]–[11] , and also demonstrated for the evolved differences between two species by a chimeric enhancer study [12] . One possibility for reconciling conservation of CRM function with rapid TFBS turnover is to assume that each loss of a TFBS is precisely balanced by the simultaneous gain of a cognate TFBS elsewhere in the CRM , a process we will call compensatory evolution [13] . The idea draws on a model first proposed by Kimura [14] , where he considers a pair of tightly linked mutant genes that are individually deleterious but in combination restore wildtype function . As applied to TFBS , the gain of a novel site on an allele carrying a mutation that decreases the quality of an existing binding site can offset the mutants fitness cost , creating a selectively neutral double-mutant allele . Binding site turnover - fixation of the double mutant allele - is achieved entirely by genetic drift , thus preserving both CRM function and population fitness . Recently , a theoretical model of this compensatory turnover process was developed to ask about the feasibility of compensatory evolution for TFBS [15] . With plausible assumptions about the mutation rate , population size and selection coefficient on the individual mutations , a completely neutral model cannot achieve a high enough level of turnover to explain Drosophila CRM evolution ( as exemplified by eve stripe 2 enhancer ) , whereas a model that assumes the double mutant to be more fit than the wildtype does . This theoretical finding raises the prospects for positive selection being an important driving force of TFBS gain and loss . Instances of directional selection have been documented in cases where a novel regulatory regime is favored [16] . Functional evolution of a transcription factor ( TF ) can also drive adaptive co-evolution of its TFBS [17]–[19] . Broad-scale studies in noncoding regions and promoters of genes have identified signatures of both selective constraint and positive selection in fruitfly and human [20]–[24] . However , only a small number of population genetics studies have been carried out to specifically test this hypothesis with TFBS or CRM , and because they focus on a single TF or CRM , they have low statistical power to distinguish between neutrality and selection [13] . The generality of the conclusions reached in these studies is also not established [25] , [26] . Several different approaches have been designed to detect and quantify selection in the system . One of them has been to consider the genome-wide ensemble of TFBS as evolving at mutation-selection balance , with the fitness of each instance of TFBS being strictly determined by its binding energy [4] , [27] , [28] . This approach proves useful in studying the strength of selective constraints on functional TFBS . However , the assumption of a unidirectional fitness function , i . e . selection always favors affinity-increasing mutations and against affinity-decreasing ones , could be violated if the loss of a TFBS were favored or gain ( or strengthening ) of a TFBS is deleterious . Another approach calculates the sum of mutational effects in TFBS on binding affinity and compares it to the expectation under a no-selection model [29] . A higher than expected sum could imply selective removal of affinity-decreasing mutations and therefore the action of purifying selection . Applying this approach to two of the CRM also included in this study , the author provided evidence for purifying selection acting to preserve the functional TFBS in the anterior Bicoid-dependent hunchback enhancer and the even-skipped stripe 2 enhancer . This test can also be used to detect positive selection , although its power is limited due to the mixed signal with purifying selection , which is expected to be dominant in most cases . In this study , patterns of polymorphism and divergence are investigated in a pair of closely related Drosophila species , D . melanogaster ( mel ) and D . simulans ( sim ) . The short evolutionary distance between the two species ensures unambiguous alignment for noncoding sequences and also allows one to capture the potentially rapid dynamics of TFBS gain and loss . A notable challenge in studying TFBS turnover is assembling a high quality set of TFBS that are precisely defined and contain few false positives . Large numbers of potential TFBS can be identified by methods involving genome-wide scans , such as computational prediction or ChIP , but these approaches generally include a large fraction of false positives , thus reducing their attractiveness for investigating the mechanisms of binding site turnover ( see Discussion ) . Instead , we chose to investigate a curated set of high-confidence TFBS identified by DNaseI footprint in well-studied D . melanogaster CRM . Short footprint regions usually contain only a single TFBS motif , which , in most cases , could be perfectly aligned with the other species to allow identification of single nucleotide differences within and between the species . Each of these differences , in turn , was evaluated for the predicted magnitude and direction of effect on TF binding energy . The neutral and selection models generate distinguishable predictions in both divergence to polymorphism ratios and in the site frequency spectra . Analysis of these patterns reveal evidence for purifying selection against affinity-decreasing mutations segregating in the population , while multiple lines of evidence indicate positive selection for both gains and losses of TFBS . These empirical findings challenge the prevailing view of neutral compensatory turnover , and have important implications for understanding CRM functional evolution . In the course of the analysis , we also identified and modeled a potential ascertainment that can impact population genetics studies of genomic features that have been identified only in a reference sequence such as TFBS .
Binding sites for an individual TF or a single CRM usually had too few counts of single nucleotide polymorphism or fixed differences to allow informative statistical analysis . Furthermore , the breadth of the turnover phenomenon across almost all investigated TF and CRM suggests a common underlying evolutionary mechanism [5] , [7] , [8] , [18] , [31] . We therefore considered pooling observations from across TFs and CRM . To see if the evolutionary rates in different TFs binding sites are sufficiently uniform , we measured sequence divergence between mel and sim for the 30 TF . After accounting for sample sizes , no significant departure from the average rate is detected by a binomial test ( Figure 1 ) . Moreover , the pooling approach should be conservative in deriving a general pattern with respect to among TF variations . We then estimated percent loss and gain of TFBS on the mel and sim lineages . For each of the 645 footprint TFBS , a PWM score was calculated for each occurrence ( ) in the alignment of mel , sim and the inferred mel-sim ancestor , by taking the log2 ratio of the probability of a sequence under the functional motif distribution versus that under the genomic background distribution ( Material and Methods ) . Using as a cutoff , approximately 2% of all footprint sites were found to be present in mel only and may represent mel specific gains; and about 2 . 5% were present in the inferred ancestor ( and mel ) but lost in sim . A set of empirical cutoffs were determined for each TF based on the range of PWM scores among its footprint sites , which produced similar results ( Table S2 ) . Consistent with the sequence divergence patterns , gain and loss of TFBS appear to be a general pattern across TF and CRM . A total turnover rate of 4 . 5% between mel and sim is similar to a previous finding of 5% for a single TF Zeste [5] . We observed approximately equal numbers of gains versus losses in our dataset , although the distribution of these events is asymmetric on the two lineages ( 16 losses , 0 gain along the sim lineage versus 12 gains , 0 losses along the mel lineage ) . This is not unexpected , given that all footprint TFBS were identified as being present in mel and the dataset doesn't include sim-specific TFBS . We predicted that identification of TFBS by computational methods would produce a more even pattern of gains and losses in both lineages . We tested this prediction for three TF ( Hb , Bcd , Kr ) using a stringent cutoff procedure and for each TF we found a similar total number of predicted binding sites in the two lineages ( Text S1; Figure S1 ) . We thus rejected the ( unlikely ) possibility that there has been a large-scale evolutionary gain of TFBS in mel and loss in sim . Gain and loss of TFBS may be subject to distinct evolutionary forces . To investigate them separately , we assigned each mutation within a footprint TFBS in mel or sim to either affinity-increasing or affinity-decreasing group based on PWM score difference between the ancestral and the derived mutation ( Materials and Methods ) . Bioinformatic and experimental investigation showed that this PWM-based procedure for inferring the direction of binding affinity change is reliable when PWM predicted magnitude of change is not too small ( Materials and Methods , Figure S2 and Figure S3 ) . We established a threshold corresponding to a PWM score difference of one , i . e . at least two-fold change in the likelihood ratio between a motif or background distribution , in order to minimize the chance for mis-assignment . Varying this threshold between zero and two do not affect the results qualitatively . We employed two approaches to investigate evolutionary forces acting on affinity increasing and decreasing changes . One approach is based on contrasting polymorphism and divergence patterns in a McDonald-Kreitman ( MK ) test framework [32] . Positive selection is expected to inflate substitution relative to polymorphism while negative selection will have the reverse but weaker effect [33] . We used synonymous changes in the target genes for the CRM as a proxy for a neutrally evolving class . Following established practices , we further classified each synonymous change as according to its expected impact on codon bias – No-Change , Preferred-to-Unpreferred , or Unpreferred-to-Preferred – and used the No-Change class as the neutral reference . The second approach investigates the site frequency spectrum of TFBS polymorphism to make inferences about selective pressures acting more recently on binding sites . The fact that all footprints were identified in mel impacts the analysis in two ways . First , gains of TFBS can be observed in mel but not losses , while the reverse is true in sim . Therefore , even though similar processes are most likely operating in both species , our evolutionary analysis of binding site gain will focus on changes in the mel lineage , whereas losses will be restricted to changes in the sim lineage . Second , affinity-decreasing and affinity-increasing mutations have the potential to differ in detectability as a footprint site in mel . This arises because mutations in TFBS were sampled conditioned on the TFBS being detected in mel and affinity-changing mutations in mel , in turn , have the potential to affect the detectability of the TFBS . Depending on whether the derived mutation is affinity-increasing or affinity-decreasing , two distinct biases are introduced in the expected neutral frequency spectrum ( Figure S4 ) . Given that the dataset consists only of TFBS that are detectable by footprinting , we assume that the high-affinity allele will always be detectable . Consider the possible situation in which the low-affinity allele is not detectable as a footprint: if the derived mutation is affinity-decreasing , the probability of detecting the TFBS will change inversely with the mutant allele frequency; conversely , if the derived mutation is affinity-increasing , the probability of detection will increase with the mutant allele frequency . Substitutions may be viewed as a special instance of a segregating mutation and treated similarly . This effect of ascertainment on neutral expectations for the MK test and the site frequency spectrum can be modeled analytically ( Text S2 ) ; there is no ascertainment if both alleles are equally detectable as footprints . To incorporate uncertainty in the detectability of the low-affinity allele , the model incorporates a parameter , f , which specifies the probability that the weaker affinity allele will not be detected in the footprint assay . While f is likely to be greater than 0 , it is unlikely to be close to 1 because footprint sites are degenerate and span a range of affinities . Under the conservative assumption that the lowest affinity among the footprint sites is the detection limit , we estimate for the 30 TF ( Text S2 ) , indicating that the majority of TFBS changes will be detectable . In the following sections , we first present our analysis of polymorphism and divergence in mel , focusing on the forces acting to either maintain functional TFBS or to create new ones . We then turn to sim , focusing on TFBS loss . Finally , we analyze the spacer sequences between TFBS in both species . For each class of change we summarized the data in the MK table by calculating the ratio , . The presence of weakly deleterious mutations can mask signatures of positive selection , and if removed can improve the power of the test [34] . Since most deleterious mutations will be at low frequencies , using 15% as a frequency cutoff has been shown to achieve most of the benefits of a more sophisticated model incorporating the distribution of deleterious effects [35] . We applied this cutoff and denote the ratio of substitutions to common polymorphism by . Under this procedure , is significantly higher for nonsynonymous changes than for the synonymous No-Change class ( Figure 2A ) , consistent with previous findings of positive selection driving amino acid substitutions in Drosophila [36] . To delineate the effect of ascertainment from that of selection for the affinity-increasing and affinity-decreasing mutations , we compared the observed to the expected neutral ratios under the ascertainment with different values ( Text S2 ) . For affinity-decreasing mutations in mel , the difference from the synonymous No-Change class is not statistically significant , even in the absence of ascertainment bias ( Figure 3A green , Figure 2A ) . This seems to suggest only neutral or deleterious mutations are present for this class and therefore no positive selection is involved . The validity of this conclusion can be questioned , however , because any affinity decreasing substitutions in mel that led to the loss of a site will not be included in the data while our correction for the ascertainment only accounts for neutral changes but not a potential adaptive excess . Thus , rejection of the neutral model in favor of positive selection is not possible for affinity-decreasing mutations in the mel lineage . However , this test is possible for the sim lineage ( reported in the next section ) , where the loss of a TFBS is observable . For affinity-increasing mutations no amount of ascertainment under our model can account for the observed relative excess of substitutions ( Figure 3A red ) . We further reasoned that the ascertainment effect should be weaker or non-existent for TFBS with an ancestrally strong binding affinity , which would be identified with or without the affinity-increasing mutations . We therefore investigated whether the excess of affinity-increasing substitutions differed if TFBS changes were grouped according to the strength of the inferred ancestral binding affinity . We found a consistently larger ratio , i . e . an excess of substitutions , across the entire range of inferred ancestral binding affinity classes compared to the No-Change class , including binding sites with the strongest ancestral binding affinity ( Figure 3B ) . These results collectively suggested that positive selection has contributed to the fixation of affinity-increasing changes . To further investigate evolutionary forces acting on the segregating mutations in TFBS in the population , we utilized the site frequency spectrum , for which we generated the neutral expectations for affinity-increasing and affinity-decreasing mutations separately under ascertainment , with or ( corresponding to no bias or complete bias , respectively ) . For affinity-decreasing mutations , with the ascertainment expected to shift the frequency spectrum to lower frequency classes ( Figure 4A , blue versus grey bar ) , the observed spectrum is shifted in that direction but is even more extremely so than the complete bias expectation ( Figure 4A , orange versus blue ) . Since is clearly an overestimate ( compared to our estimate of ) , this strongly suggests that forces other than ascertainment must have shaped this pattern . Both a recent selective sweep and population growth can produce an excess of rare variants and one or both mechanisms may be acting in this system , as is suggested by our finding that synonymous changes also show a relative excess of low frequency mutations ( Figure S5B ) . However , as we compared the site frequency spectrum of the affinity-decreasing mutations to that of synonymous sites ( corrected for ascertainment ) , we found the former is again more significantly shifted than the latter ( Figure S6 ) . Thus we suggest that the observed frequency spectrum is consistent with on-going purifying selection against affinity decrease in functional TFBS . The observed frequency spectrum for affinity-increasing mutations lies between the two expectations and the differences are not significant from either one , a possible consequence of the small sample size ( 15 observed affinity-increasing polymorphisms ) ( Figure 4B ) . Thus , while positive selection is indicated on the basis of the MK test , inference cannot be made about on-going selection for affinity-increasing mutations . Patterns of polymorphism and divergence in sim are not influenced by the ascertainment because the identification of TFBS in mel is independent of the effect of mutations fixed or segregating in sim . However , the inclusion of binding sites gained in mel may confound the analysis as their orthologous sequences in sim may have evolved under less or different kinds of selective constraints . We thus restricted the analysis to footprint TFBS predicted to be present in the mel-sim common ancestor , where we found a significant excess of substitutions for the affinity-decreasing mutations compared to the synonymous No-Change class ( Figure 2B , Fisher's Exact Test ) . Statistical significance of this pattern is robust to the cutoff for excluding binding sites gained in mel ( Table S3 ) . A relative excess of substitutions might also be a consequence of factors other than selection , such as systematic differences in the genealogical histories of CRM versus synonymous sites . However , these factors seem unlikely to be the cause of this type of departure from neutrality in these two species ( Kohn and Wu 2004 ) . Therefore we consider positive selection a more plausible explanation . We also compared the ratio between affinity-decreasing and affinity-increasing mutations in polymorphism to the expected ratio of the two classes in the mutational input , i . e . the probability for a new mutation to be one of the two classes ( Materials and Methods ) . Briefly , the expected ratio was obtained by considering all possible mutations in each of the 645 footprint TFBS and their predicted effects on binding affinity the same way as we did before . Assuming polymorphism for both classes were neutral , we expected similar ratios , whereas the observed results showed a significant deficit of affinity-decreasing polymorphism relative to affinity-increasing polymorphism ( Table 1 ) , which may suggest that among new mutations , affinity-decreasing ones are more likely to be deleterious , a result consistent with our finding based on frequency spectrum in mel . A similar approach has been applied before , using the sum of ( individual mutation's effect on binding affinity predicted by PWM ) within a CRM instead of counts of mutations in binary classes [29] . There the author also found evidence for purifying selection against affinity-decreasing mutations . The finding of both on-going purifying selection and potentially positive selection acting is not dissimilar to patterns found in nonsynonymous changes [36] . We reserve for the Discussion section the attempt to reconcile the adaptive loss of TFBS , as observed between the two species , with on-going purifying selection against affinity-decreasing new mutations . In both mel and sim we found a significant excess of substitutions in spacer sequences , indicative of positive selection in these intervals ( Figure 2 ) . Also , the frequency spectrum for this class is strongly shifted towards lower frequencies ( Figure S5E , Tajima's D = −1 . 09 ) , indicative of on-going purifying selection . The implication of these results is that spacer sequences might contain many unidentified functional elements , for example , TFBS for known or uncharacterized transcription factors , or perhaps other structural features not yet understood . To summarize , analysis of TFBS changes in mel indicates on-going purifying selection against affinity-decreasing polymorphism in the population , and positive selection for affinity-increasing substitutions . In sim , the analysis of affinity-decreasing changes indicates a significant , and potentially adaptive excess of substitutions that contributes to binding site loss . Spacer sequences between footprint TFBS in these well-characterized CRM also exhibit patterns of polymorphism and divergence consistent with both functional constraint and adaptive evolution .
Natural selection , both positive and negative , has been shown to act throughout noncoding regions of the Drosophila genome [21] , [22] , albeit with varying intensities [23] . Against this backdrop of ubiquitous selection in noncoding DNA , should it be surprising to find signatures of positive selection in Drosophila TFBS ? We think not . More surprising perhaps is the incompatibility of this finding with the model of neutral compensatory binding site turnover , a simple and appealing mechanism that allows for both rapid binding site turnover and functional stasis of CRM activity . But as explained below , there are good reasons to doubt whether a strictly neutral compensatory process can actually generate rapid TFBS turnover in Drosophila , even with its favorably large population size . Positive selection , in contrast , can drive arbitrarily fast rates of binding site turnover; the question is whether it can also allow for functional stasis of CRM activity . Below , we first discuss the strengths and limits of our analysis and then we describe properties of gene regulatory networks that can promote adaptive binding site turnover and yet also constrain the function of CRM . One challenge in investigating cis-evolution is the proper alignment of noncoding sequences . To minimize this potential problem , we specifically selected a pair of closely related sibling species , D . melanogaster and D . simulans for investigation . Sequence divergence between the two species in noncoding regions ranges only between 5% and 8% [37] , which allowed us to accurately identify single nucleotide differences from unambiguous alignments of binding sites ( those with alignment gaps were excluded from the analysis ) . Working with closely related sequences also provided accurate inference of ancestral states , and thus the direction of mutational change along the phylogeny , as well as minimized trans-cis co-evolution . Independently , Bradley et al also recommended mel and sim for measuring binding site divergence based on these same issues arising in their analysis of divergence between two more distantly related species [31] . Another challenge in studying TFBS turnover is the establishment of a TFBS dataset consisting of biologically functional sites , a difficult task due to both the high false positive rate in binding site prediction ( even in ChIP bound regions ) and the difficulty in validating the biological functionality of individual binding sites . While many genome-wide datasets for TFBS are becoming available , several properties of the Drosophila DNase I footprint dataset made it the one of choice for use in this study . First , the in vitro footprint experiments were applied not to anonymous noncoding regions but rather to specific sequences that had been identified with in vivo reporter assays as containing a CRM . Furthermore , the transcription factors assayed for each CRM were also chosen based on prior genetic evidence for their involvement in the regulation of the CRM . For both of these reasons , subsequent experimental analysis of Drosophila footprint sites has invariably validated their functionality [38]–[43] . This experimental sampling of footprint site functionality is unique among available TFBS datasets , and provides evidence for a low false positive rate . In contrast , a recent attempt to combine known CRM , ChIP bound regions , and PWM prediction to obtain a genome-wide TFBS dataset estimated false positive rate [4] . Although the footprint sites were identified in lab strains particular to each individual experiment , we provided reasonings and evidence why the annotation is applicable to natural populations ( Text S3 ) . In particular , we constructed phylogenetic trees based on the genomic sequences containing the CRM we studied for natural population lines as well as a representative lab strain ( the genome sequence reference strain ) , which shows that the later is indistinguishable from the rest ( Figure S8 ) . This also suggests the lab strains were not genetically divergent from the natural population . Genome-wide studies have identified signals of both positive and negative selection in noncoding sequences in Drosophila , but not the biological or functional basis for this selection . In this study , we distinguished mutations in the footprint sites by their functional impact – either increase or decrease the binding affinity of the corresponding TF – and observed different patterns of polymorphism and divergence between the two classes . For example , we found that affinity-decreasing mutations are on average more deleterious among new mutations than affinity-increasing ones , as revealed by a comparison of the ratio between the two classes in polymorphism with the expectation from mutational input . Such distinctions were not observed when mutations were grouped in other ways irrelevant to the function of TFBS ( for example , mutations in the first half of the motif versus the second half ) . For these reasons we think the evidence supports our specific model of selection acting on binding site gain and loss as opposed to an unidentified functionality in noncoding sequences in general . The mechanism of selection we described here for well-annotated TFBS could in principle be acting more broadly across noncoding regions inasmuch as noncoding DNA is often associated with proteins binding . Our ability to correctly categorize mutations into affinity-increasing or affinity-decreasing categories hinges on the accuracy of PWM predicted affinity differences . To investigate this issue , we employed a state-of-the-art microfluidics technique , MITOMI [44] , to experimentally measure the binding affinity differences for naturally occurring mutations in hunchback and bcd binding sites . To our knowledge , this is the first time that accurate measurements have been made on population-level variants in TFBS . We found that PWM scores correctly predicted the measured direction of affinity change for 21/25 mutations investigated . Of the four mutations that PWM predicted the wrong direction , three have effect sizes predicted to be close to zero . The PWM-based procedure , therefore , may not be accurate for small predicted differences in binding affinity . Taking these results into consideration , we employed a binary classification of mutations with PWM differences exceeding a threshold requirement rather than using quantitative predictions of all PWM score differences as a basis for our analysis . Another potential issue concerns applying mel derived PWM to score sim TFBS binding affinity . Transcription factor protein evolution between the two species , if it occurred , could lead to underestimation of binding affinity in sim , although the effect should be similarly applied to both substitutions and polymorphism and thus is not expected to cause a relative excess of the former as observed in the sim data . Nevertheless , we show two lines of arguments that suggest this is not the case in our study: first , for the 30 TF whose binding sites we investigated , the DNA bindings domains and other functionally annotated domains are completely conserved except for one biochemically conservative amino acid difference ( Asp/Glu ) in Dorsals RHD domain ( Table S4 ) . Although differences exist in other parts of the proteins , it has been shown that DNA binding domain may singly determine the sequence specificity of the protein [44] , [45] . Second , if what we identified as affinity-decreasing mutations in sim reflected on-going adaptations to a slightly different motif , we would expect , but did not find , a consistent pattern in the position and kind of nucleotide changes for a TF ( data not shown ) . To further support this argument , we derived PWM using MEME from the mel footprint sites as well as their aligned sequences in sim . As shown in Figure S7 , our classification of binding site differences did not differ between using either the mel PWM or the sim PWM , contrary to what would be expected if TF sequence specificity had evolved between the two species . Therefore we consider it very unlikely for the 30 TF included in this study to have undergone significant evolution in their sequence specificity . In addition , because the SELEX derived PWM produce consistent results with the footprint derived ones ( Figure S3 ) , we can also rule out the possibility of over-optimization of the PWM inducing a sequence preference for mel over sim . Finally , in the course of the analysis , we identified and modeled an ascertainment bias caused by the identification of footprint sites exclusively in a single strain of mel , and the possibility that sequence changes in the same species can lead to creation or destruction of the footprint feature ( as described in the Results section ) . Many other genomic features such as miRNA binding sites and recombination hotspots can also satisfy these two criteria . As new studies attempt comparative evolutionary studies of genomic features often identified in a single reference sequence , we expect this problem to become more common and , therefore , to require greater attention . If not properly accounted for , this form of ascertainment can lead to false rejection of the neutral hypothesis . The analytical model of ascertainment under neutrality we developed here should be applicable to population genetic and evolutionary analysis of many different structural features of genomes . Our population genetics analysis identified three major forces in TFBS evolution . First , we found functional TFBS were selectively maintained in the population by purifying selection , as revealed by a frequency spectrum skewed towards rare variants for affinity-decreasing polymorphism in mel and a significantly reduced proportion of affinity-decreasing polymorphism compared to mutational input in sim . These results are consistent with previous findings of selective constraints on functional TFBS . Mustonen and Lässig estimated that the average selection coefficient to maintain TFBS in bacteria and yeast genomes are on the order of [28] , [46] , and a similar estimate has been obtained for Drosophila [4] . The substitution rate with is expected to be less than 0 . 05% of the neutral rate in a population with a size as large as Drosophila ( Equation B6 . 4 . 1 , [47] ) . This means TFBS loss is unlikely to happen through fixation of deleterious mutations ( 0 . 2 losses expected for 645 footprint TFBS versus 16 inferred in sim ) . We can think of only three mechanisms by which TFBS loss can occur at an appreciable rate: ( 1 ) there is loss of constraint; ( 2 ) a pair of tightly linked compensatory mutations creates an effectively neutral allele; or 3 ) positive selection drives the loss of TFBS . Our second finding – a significant excess of substitutions compared to the neutral class for affinity-decreasing mutations in sim – is consistent only with positive selection for TFBS loss . Occasional adaptive loss of a TFBS is not inconsistent with more ubiquitous selection to maintain binding sites [28] , and has been suggested to account for the evolution of fermentation pathways in yeast [16] . Our third finding is positive selection contributing to the gain of TFBS , as revealed by a significant excess of substitutions for affinity-increasing mutations in mel . Collectively , the three findings indicate that natural selection is extensively involved in the maintenance , gain , and loss of TFBS . This conclusion challenges the prevailing view of a neutral TFBS turnover process [4] , [13] . We think that a selectionist interpretation of the turnover process is plausible for several reasons . First , the assumption of CRM functional stasis , which is the main argument for the neutral ( i . e . , compensatory ) view , is not well supported experimentally . Reporter transgene assays , in particular , are limited in their quantitative resolution , and yet even in these studies , repeatable differences were found between orthologous CRM [7] . A functional rescue experiment is potentially more sensitive than a reporter transgene assay . As applied to the Drosophila even-skipped stripe 2 enhancer , it demonstrated clear functional differences between CRM that were previously believed to have the same spatial pattern of expression [48] . Second , compensatory neutral evolution cannot account for the patterns of variation observed in this study . According to this model , affinity-decreasing mutations should in general be deleterious but occasionally become “effectively” neutral when a second compensatory mutation occurs in the CRM of the mutant allele . A mixture of deleterious and compensatory mutations , even if the latter is common , may bring patterns of polymorphism and divergence close to a neutral scenario , but cannot produce a signature of positive selection as observed for both classes of mutations in our analysis . In addition , analytical modeling of the compensatory evolution of TFBS finds that the waiting time for a turnover event is long if complete neutrality of the compensating mutations is assumed [15] . To shorten the waiting time to be compatible with the Drosophila TFBS turnover rate , the parameterization of the model requires that the double mutant allele have higher fitness than the non-mutant allele , making it a directional selection model . This supercompensatory scenario could produce signatures of positive selection both for binding site gain and loss , the latter occurring because the fixation of a deleterious mutation in an existing TFBS will have the appearance of being positively selected as it hitchhikes to fixation on the selectively favored allele . However , this scenario is biologically unrealistic , as it requires the second mutation ( the gain of a TFBS ) to be positively selected only on the background of the first mutation . As an alternative , consider the following model of positive selection on CRM structure/function . We propose that for CRM with large numbers of interacting partners , the network of cis- and trans-factors will inevitably be constantly evolving – due to both direct selective pressures imposed on the CRM or indirect effects caused by adaptations in other components of the network . For example , egg length variations between and within Drosophila species have been studied as potentially adaptive traits; if egg length evolves , genes such as eve whose expression pattern need to scale with the embryo may need to change its CRM to adapt to the new context [49] . This constant flux of change , we propose , imposes continual selection pressure for CRM function within the network to co-evolve and change . This “moving target” hypothesis finds support in an analytical study , which shows that fluctuating selection may be common in Drosophila , with changes in the sign of selection coefficient occurring at nearly the rate of neutral evolution [50] . Adaptive substitutions could therefore occur before selection switches its sign again , since positively selected mutations fix at rates much higher than the neutral mutation rate . At the same time , the high connectivity in the regulatory network implies pleiotropic effects while the essentiality of genes controlled by the network may call for accurate regulation , both suggesting that the net change in CRM function will be highly constrained ( Figure 5A ) . Under this conceptual model , functionally significant change will be possible on short evolutionary timescales , but will remain within constrained bounds over longer timescales . This feature of the model would account for adaptive gain and loss of TFBS in CRM , and could explain the strongly non-linear relationship between function and sequence evolution as exemplified by the Drosophila eve stripe 2 enhancer [7] , [8] . Moreover , it provides an explanation for the finding of a non-clocklike evolutionary pattern: sequences from D . pseudoobscura rescues a mel eve stripe 2 enhancer deficiency almost as well as the native mel enhancer and substantially better than ones from much more closely related species ( [48] , Figure 5B ) . In conclusion , our findings provide empirical evidence for positive natural selection acting in CRM and TFBS evolution . We suggest that CRM are not as functionally static as commonly believed , but rather may experience frequent adaptation through binding site turnover , even though there may be constraints on net change over longer evolutionary time .
REDfly [30] is a database of manually curated CRM and TFBS obtained from the literature from which we chose 118 non-overlapping autosomal CRM for investigation ( Table S1 ) . They regulate 81 target genes and contain binding sites for 82 TF . The 118 CRM range in size from 65 bp to 4 . 3 kb ( median = 515 bp ) and contain between 1 to 64 DNase I footprint sites ( median = 4 ) . From the set of 82 TF , we identified a subset of 30 with more than 10 footprint sites represented in the dataset and with carefully constructed Position Weight Matrices [51] . In each footprint region plus five flanking bases on each end , we applied the appropriate position weight matrix to identify the highest scoring match as the core motif for the TFBS ( referred to as TFBS in the text ) . We only included those TFBS for which the alignment between mel and sim sequences contain no insertions or deletions ( including both fixed or polymorphic sites ) . As a result , a total of 645 TFBS for these 30 TF were included for analysis . For each of the 118 CRM ( coordinates in dm3 of D . melanogaster reference genome listed in Table S1 ) , we downloaded pre-aligned MAF blocks from UCSC genome browser for D . melanogaster ( mel ) , D . simulans ( sim ) , D . sechellia ( sec ) , and two outgroup species , D . yakuba ( yak ) and D . erecta ( ere ) . D . sechellia is a sister species to D . simulans and is included to compensate for the low sequence completeness in the reference sim genome . We then used the baseml module in PAML 4 . 4c [52] to reconstruct the ancestral sequences from the alignments . Following analysis involving polarized changes were done either using a single ancestral sequence for mel and sim determined by the most probable ancestral state ( A , C , G or T ) at each position , or summing over the posterior probabilities of all four possible states ( full Bayesian approach ) . The two methods produced essentially the same results and therefore we only presented results using the most probable ancestral state . A maximum parsimony method was also investigated and was found to produce consistent results . For polymorphism analysis , alignments for the same 118 CRM regions were obtained of a population sample of 162 D . melanogaster lines ( http://www . hgsc . bcm . tmc . edu/projects/dgrp/ ) and six D . simulans lines ( http://www . dpgp . org/ ) . We also compiled the genome sequences of 150 coding regions corresponding to the target genes of the CRM listed in REDfly , for the purpose of compiling synonymous and nonsynonymous changes . For these data , we used codeml module in PAML 4 . 4c to reconstruct the ancestral sequence states following otherwise the same procedure as described above for CRM regions . PWM for 30 TF ( Antp , Deaf1 , Dfd , Kr , Mad , Trl , Ubx , Abd-A , Ap , Bcd , Br-Z1 , Br-Z2 , Br-Z3 , Brk , Cad , Dl , En , Eve , Hb , Kni , Ovo , Pan , Prd , Slbo , Tin , Tll , Twi , Vvl , Z , Zen ) were obtained from [51] . This set represents all the TF for which Down et al . identified a single best motif for the REDfly footprint sites . For comparison , we also constructed five PWM ( Hb , Bcd , Kr , Prd , Twi ) from SELEX ( Systematic Evolution of Ligands by EXponential enrichment ) data ( kindly provided by Mark Biggin ) . We ran MEME [53] with parameters “-evt 0 . 01 -dna -nmotifs 3 -minw A -maxw B -nostatus -mod zoops -revcomp text” on different selection rounds of the SELEX data . The best PWM was chosen based on the MEME score , percentage of footprint sites recovered and a penalty for the number of additional matches predicted in addition to the footprint sites ( Table S5 ) . Consider a mutation at the position in a binding site motif involving a change from nucleotide to ( take values 1–4 , corresponding to the nucleotides ACGT ) . We calculated , where is the PWM matrix of size . According to previous theories , the PWM score is proportional to the physical discrimination energy of the protein to the sequence and therefore the above calculation may be used to infer the direction and magnitude of binding energy change due to a mutation [54] . To evaluate the accuracy of the PWM-based inference , we experimentally measured the binding energy change of observed mutations in Hb binding sites , using a state-of-the-art microfluidics device that has high sensitivity for relatively weak molecular interactions ( MITOMI , [44] ) . The experiments were performed as described in Maerkl et al . [44] . Sixty-four oligonucleotides were synthesized to test 25 SNP in Hb footprint sites and their combination in cases of multiple SNPs in a single TFBS between mel and sim . Data were analyzed in GenePix 6 . 0 , R , and Prism 5 . 0 . We found that the PWM we used correctly predicted the direction of change in 21/25 cases ( Figure S2 ) . Three of the four disagreements had a predicted PWM score change close to or smaller than one , which indicates that PWM may not be accurate when its predicted binding energy differences are small . To minimize the chance of misassigning the direction of binding energy change to a mutation , we set a threshold corresponding to a PWM score difference of one , and classified mutations within ( smaller in absolute value ) that bound as uncertain . The conclusions are robust to the setpoint of the threshold ( for example , Table S3 ) . We also compared the PWM derived by Down et al . to the five PWM derived from SELEX data: 97% ( 33/34 ) of mutations in the TFBS were consistently classified after excluding nine mutations with small predicted effects by either PWM ( Figure S3 ) . To examine the extent of binding sites gain and loss between the two species , we calculated PWM scores for each of the 645 footprint TFBS ( from 1 to 645 ) in orthologous sequences in mel , sim or the inferred mel-sim ancestor ( j from 1 to 3 ) , using patser v3e ( by Gerald Z . Hertz , 2002 ) . To determine whether a sequence is a binding site or not , we established two sets of cutoffs for PWM scores . First , we used PWM score , corresponding to the sequence being more likely from a binding site distribution than from a background distribution . For the second we used a set of TF-specific cutoff values chosen by first ranking all footprint sites of a TF by their PWM scores in descending order and then taking the 80% quantile value . The two cutoff set produced similar results ( Table S2 ) . To test whether the mel-derived PWM might be over-optimized so that they would favor mel over sim sequences independent of the binding affinity differences , we ran MEME on both mel footprint sites for three TF ( Hb , Bcd , Trl ) and their sim orthologous sequences with the same parameters . The two set of ÒorthologousÓ PWM were then applied to score the observed variations in the TFBS of the three TF for comparison ( Figure S7 ) . We attempted to estimate the probability for a random new mutation to be affinity-increasing ( ) or affinity-decreasing ( ) by examining all possible mutations that can occur on the inferred ancestral sequence of mel and sim for the 645 footprint TFBS . At the site in a TFBS for TF x , the probabilities are calculated as: ( 1 ) ( 2 ) ( 3 ) where is the original nucleotide and varies among the three possible mutations . is the position weight matrix for TF x of size . These values were then summed across all 645 TFBS and divided by the total number of nucleotides involved . Mutation matrix is derived from polymorphism of the 4-fold degenerate sites of 9 , 628 genes in D . simulans [55] . For the generalized MK test , we counted the number of fixed and segregating sites for different functional categories in both mel and sim lineages . In sim , we required at least two of the six alleles to be non-missing for a site to be included in the analysis . For coding regions , synonymous sites were further classified into No-Change , Preferred-to-Unpreferred and Unpreferred-to-Preferred , following [22] . Polymorphism and divergence sites in both coding and CRM regions were counted using perl scripts adapted from Polymorphorama ( Peter Andolfatto , Doris Bachtrog , 2009 ) . Following the suggestion of [34] , we considered only common polymorphism ( derived allele frequency 15% ) in the generalized MK test to alleviate the problem caused by negatively selected mutations in detecting positive selection . For each mutation category , we compared the substitution-to-polymorphism ratio to the synonymous No-Change class using Fisher's Exact Test . Two-sided p-values are reported . Site frequency spectrum ( mel only ) : Next-generation sequencing data produce variable coverage . To estimate the site frequency spectrum , for each variable site ( TFBS , coding and spacers ) with a coverage greater than or equal to 150 ( maximum is 162 ) we randomly chose 150 and combined the counts for each frequency class ( from 1/150 to 149/150 ) . | Transcription factor binding sites ( TFBS ) turnover ( i . e . lineage-specific gain and loss ) is a well-documented phenomenon in eukaryote cis-regulatory modules ( CRM ) . The wide spread of the phenomenon and the appearance of conserved expression patterns for diverged orthologous CRM led to the standing view that the observed gain and loss of TFBS were functionally and selectively neutral . To the contrary , genome-wide population genetics analyses have unequivocally identified signatures of positive selection acting in noncoding regions in general , and particularly in 5′ and 3′ untranscribed regions of genes . To specifically test the neutral versus selection hypotheses for the TFBS turnover process , we analyzed natural variation patterns within and between two closely related Drosophila species . We found the patterns of divergence and polymorphism for two types of mutations—those inferred to increase or decrease the binding affinity respectively—are not compatible with a neutral hypothesis . Instead , multiple lines of evidence suggested that positive selection has contributed to gain as well as loss of TFBS in the two lineages , with purifying selection maintaining existing TFBS in the population . Spacer sequences also showed signatures of negative and positive selection . We proposed a model of CRM evolution to reconcile the finding of frequent adaptive changes with constraints on long-term evolution . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"animal",
"models",
"drosophila",
"melanogaster",
"model",
"organisms",
"natural",
"selection",
"neutral",
"theory",
"population",
"genetics",
"biology",
"evolutionary",
"biology"
] | 2011 | Does Positive Selection Drive Transcription Factor Binding Site
Turnover? A Test with Drosophila Cis-Regulatory Modules |
Merkel cell carcinoma ( MCC ) is an aggressive skin cancer with a high propensity for recurrence and metastasis . Merkel cell polyomavirus ( MCPyV ) is recognised as the causative factor in the majority of MCC cases . The MCPyV small tumour antigen ( ST ) is considered to be the main viral transforming factor , however potential mechanisms linking ST expression to the highly metastatic nature of MCC are yet to be fully elucidated . Metastasis is a complex process , with several discrete steps required for the formation of secondary tumour sites . One essential trait that underpins the ability of cancer cells to metastasise is how they interact with adjoining tumour cells and the surrounding extracellular matrix . Here we demonstrate that MCPyV ST expression disrupts the integrity of cell-cell junctions , thereby enhancing cell dissociation and implicate the cellular sheddases , A disintegrin and metalloproteinase ( ADAM ) 10 and 17 proteins in this process . Inhibition of ADAM 10 and 17 activity reduced MCPyV ST-induced cell dissociation and motility , attributing their function as critical to the MCPyV-induced metastatic processes . Consistent with these data , we confirm that ADAM 10 and 17 are upregulated in MCPyV-positive primary MCC tumours . These novel findings implicate cellular sheddases as key host cell factors contributing to virus-mediated cellular transformation and metastasis . Notably , ADAM protein expression may be a novel biomarker of MCC prognosis and given the current interest in cellular sheddase inhibitors for cancer therapeutics , it highlights ADAM 10 and 17 activity as a novel opportunity for targeted interventions for disseminated MCC .
Merkel cell carcinoma ( MCC ) is a highly aggressive neuroendocrine cancer of the skin [1] . Although rare , the incidence of MCC has increased over the past twenty years in both Europe and the United States of America [2] , attributed to advances in reporting , diagnostic improvements and known risk factors . UV light appears to be an important factor in MCC , with a positive correlation between geographic UVB radiation indices and age-adjusted MCC amongst Caucasians [1 , 3] . The predominance of MCC in elderly persons also highlights immunosuppression as an important risk factor , supported by disproportionally higher rates of MCC in patients on long-term iatrogenic immunosuppression , in addition to patients with lymphoproliferative disorders and HIV/AIDs [2] . Due to its aggressive nature MCC carries a high risk of local , regional and distant recurrence [4] . As such , the 5-year survival rates range from 60–87% for local disease to 11–20% for metastatic disease [5–7] . The majority of MCC cases , ~80% , are associated with Merkel cell polyomavirus ( MCPyV ) [8] , whilst the remaining cases contain a high degree of single nucleotide polymorphisms consistent with UV-mediated mutations [9 , 10] . MCPyV is a common skin commensal causing an asymptomatic infection usually acquired in childhood . Like other polyomaviruses , MCPyV expresses a variety of early spliced variant regulatory proteins required for viral replication and pathogenesis , including the small and large tumour antigens ( ST and LT , respectively ) [11] . Upon loss of immunosurveillance , the MCPyV genome integrates into the host genome prior to clonal expansion of tumour cells [12 , 13] . A further prerequisite for MCPyV-mediated tumourigenesis is the truncation of the LT antigen rendering the virus replication defective [13] . These truncations lead to the loss of functional LT domains associated with virus replication , although all preserve the LXCXE Retinoblastoma ( Rb ) protein-binding domain , which alters cell cycle progression contributing to increased cell proliferation [14 , 15] . Both MCPyV ST and truncated LT antigens are essential for MCC cell survival and proliferation , exemplified by siRNA-mediated depletion of either protein leading to cell cycle arrest and apoptosis [16] . Moreover , genetically engineered mice expressing MCPyV T antigens in the stratified epithelium display signs of neoplastic progression [17] . However , in contrast to the prototype polyomavirus , simian virus 40 ( SV40 ) , MCPyV truncated LT forms cannot initiate cellular transformation alone and function in an accessory role by binding host factors which regulate cellular proliferation , such as Rb and Hsc70 [18 , 19] . Conversely , MCPyV ST expression is sufficient to transform rodent cells to anchorage- and contact-independent growth and induce serum-free proliferation of human cells [18] . In addition , preterm transgenic mice co-expressing epidermis-tagged MCPyV ST and the cell fate determinant atonal bHLH transcription factor 1 developed widespread cellular aggregates representative of human intraepidermal MCC [20] . Together these observations show that MCPyV ST is the major oncogenic driver of MCC . Several MCPyV ST-mediated mechanisms contribute to MCC development and proliferation . ST expression leads to the hyperphosphorylation of the translation regulatory protein , 4E-BP1 , resulting in dysregulation of cap-dependent translation [18] and prevents SCFFwb7-mediated degradation of MCPyV LT and several cellular oncoproteins [21] . It induces centrosome overduplication , aneuploidy , chromosome breakage and the formation of micronuclei by targeting cellular E3 ubiquitin ligases [22] . MCPyV ST also functions as an inhibitor of NF-κB-mediated transcription [23 , 24] . Moreover , ST activates gene expression by associating with MYCL and the EP400 histone and chromatin remodelling complex [25] , inducing transcriptional changes effecting for example glycolytic metabolic pathways [26] . The poor survival rates of MCC strongly correlate to the high dissemination rates and metastatic nature of MCC [5] . Whether MCPyV T antigens contribute to MCC metastasis is yet to be fully elucidated . Metastasis is a complex process , with several discrete steps required for the formation of secondary tumour sites [27] . These metastatic hallmarks include loss of cell adhesion , gain of cell motility , dissemination via the vasculature , and colonisation of distant sites [28 , 29] . Recent quantitative proteomic studies suggest MCPyV ST expression can promote cell motility and migration [30–32] by inducing differential expression of cellular proteins involved in microtubule [30] and actin-associated cytoskeletal organization and dynamics [31] , leading to microtubule destabilization and filopodium formation . These results suggest that MCPyV may be associated with the highly metastatic nature of MCC , and is supported by studies showing that engraftment of MCC cell lines into SCID mice results in circulating tumour cells and metastasis formation [33] . One key trait that underpins the ability of cancer cells to become invasive and metastasise is how they interact with the surrounding extracellular matrix ( ECM ) and adjoining tumour and stromal cells [34 , 35] . Cell–cell junctions are sites of intercellular adhesion that maintain the integrity of epithelial tissue and regulate signalling between cells [36] . The expression of cell adhesion molecules is tightly regulated , as dysregulation of cell adhesion between tumour cells and turnover of the surrounding ECM plays a critical role in malignant transformation and the initiation of the metastatic cascade [37] . A key mediator of cell adhesion in epithelial tissues is E-cadherin and its loss can promote invasive and metastatic behaviour in many epithelial tumours [38] . The cytoplasmic domain of E-cadherin binds to members of the catenin family , linking this multiple protein complex to the actin cytoskeleton through alpha-E-catenin . The clustering of cadherin-catenin complexes on adjacent cells leads to localised actin remodelling required for the formation of adheren junctions [39] . Notably , the loss of E-cadherin and associated cell adhesion molecules , results in the suppression or weakening of cell–cell adhesion which is regarded as a crucial step in the epithelial–mesenchymal transition ( EMT ) [40 , 41] , a process enabling a cell to acquire a more migratory and invasive mesenchymal phenotype . Loss of E-cadherin and associated cell adhesion molecules in human tumours is caused by multiple factors , including germline mutations , promoter methylation , downregulation of EMT-associated transcriptional repressor proteins and the upregulation of cellular proteinases causing proteolytic cleavage of cell adhesion molecules [42–44] . ADAMs ( a disintegrin and metalloproteinases ) , are a family of zinc-dependent transmembrane proteins implicated in the ectodomain shedding of various membrane-bound proteins [45] . Of the 21 human largely cell-membrane associated ADAMs , 13 have proteolytic sheddase capacities modulating the activity of membrane cytokines and growth factors , their receptors and cell adhesion molecules , including cadherins , selectins and integrins [46] . ADAM sheddase activities have been implicated in several physiological and pathological processes including inflammation , tumour growth and metastatic progression [47] , reinforced by upregulation of proteolytic ADAMs in both tumour tissues and cancer cell lines [48–50] . Correlations exist between levels of specific ADAMs and parameters of tumour progression , implying that these sheddases are implicated in the process of cancer development and the dissemination of metastatic tumour cells [51] . ADAMs are now emerging as potential cancer biomarkers for aiding cancer diagnoses and predicting patient outcome [52] . In addition , selective ADAM inhibitors have promising anti-tumourigenic effects in in vitro and in vivo studies and are progressing into clinical trials [53] . Here we demonstrate that the cellular sheddases , ADAM 10 and 17 , are upregulated in a MCPyV ST-dependent manner . Work highlights the essential role of ADAM sheddases in MCPyV ST-mediated disruption of cell adhesion leading to enhanced cell dissociation and motility . This suggests that ADAM protein expression may be a novel biomarker of MCC prognosis and inhibiting ADAM activity may provide a novel opportunity for targeted interventions for disseminated MCC .
Cell-cell adhesion and cell interaction to the extracellular matrix is required for tissue integrity [54] . Disrupting cell-cell adhesion enhances cell scattering , which is essential to initiate cell migration and metastatic spread [55] . To determine whether MCPyV ST expression affects the integrity of cell junctions , EGFP and EGFP-ST transfected HEK 293 cells were stained with an Alpha-E-catenin-specific antibody . Alpha-E-catenin , which is predominantly expressed at the plasma membrane mediating cell adhesion and its breakdown impliess a loss of structural integrity at cell junctions [56] . Results demonstrate that Alpha-E-catenin in control EGFP-expressing cells primarily localised to the plasma membrane , in contrast a reduced and incomplete plasma membrane localisation is observed in EGFP-ST-expressing cells , indicative of diminished cell-cell adhesion ( Fig 1A ) . A similar result was also observed upon inducible MCPyV ST expression in a HEK 293 FlpIn-derived cell line ( i293-ST ) [30] ( S1 Fig ) . In addition , immunoblotting these cell lysates showed a decrease in Alpha-E-catenin protein levels ( S1 Fig ) . Quantification of Alpha-E-catenin levels at the plasma membrane in EGFP and EGFP-ST-expressing cells was then performed using flow cytometry . Results validated the immunofluorescence data demonstrating a reduction in Alpha-E-catenin levels upon MCPyV ST expression ( Fig 1B and 1C ) . To confirm the disruption of cell junctions , the levels of a second cell adhesion-associated protein , Zona occludin 1 ( ZO-1 ) [57] , was compared in EGFP versus EGFP-ST-expressing cells . Consistent with the reduction in Alpha-E-catenin levels , immunoblot analysis showed a significant decrease in ZO-1 expression upon MCPyV ST expression ( Fig 1D and 1E ) . Together , these results provide the first indication that MCPyV ST dysregulates cell-cell adhesion . Loss of cell junction integrity enhances the ability of a cell to migrate and dissociate from its primary site . To assess whether MCPyV ST induces cell dissociation and scatter , a cell scatter assay was performed as previously described [58] . Here EGFP and EGFP-ST transfected HEK 293 cells were incubated in low serum to induce aggregation , upon reintroduction of serum cells were fixed and stained with DAPI at 6 hourly intervals and clusters of cells were analysed to quantify the distance between each cell nucleus ( Fig 1F ) . Results show that EGFP control cells scarcely dissociate , instead remaining in cell clusters . In contrast , MCPyV ST-expressing cells dissociated significantly from their initial cell clusters . Similar results were also observed in the MCPyV negative cell line MCC13 , transfected with either EGFP or EGFP-ST expression constructs ( S1 Fig ) , although results in MCC13 cells were less pronounced than in HEK 293 cells . These results suggest that MCPyV ST expression can lead to the breakdown of cell junctions enhancing cell dissociation . Cellular sheddases function predominantly in the ectodomain cleavage of various membrane-bound proteins , including cell adhesion molecules . Therefore , to identify potential cellular sheddases induced upon MCPyV ST expression , we re-analysed a previously published SILAC-based quantitative proteomic dataset which determined alterations in the host cell proteome upon inducible MCPyV ST expression in a HEK 293 FlpIn-derived cell line ( i293-ST ) [30] . MCPyV ST expression led to an increase in the levels of two specific cellular sheddases , namely ADAM 10 and 17 proteins by 7 . 6 and 4 . 3 fold , respectively ( S1 Fig ) . To confirm an increase in ADAM protein levels upon MCPyV ST expression , cell lysates of uninduced and induced i293-ST cells were analysed by immunoblotting . Results demonstrated a significant increase in ADAM 10 and 17 mature protein levels , compared to ADAM TS1 ( Fig 2A ) . Densitometry-based quantification of the immunoblot analysis showed an increase in the mature forms of ADAM 10 and 17 expression of 6 and 4 fold , respectively ( Fig 2B ) . A similar fold increase was also observed in MCC13 cells , transfected with either EGFP or EGFP-ST expression constructs ( Fig 2C and 2D ) . The increase observed in ADAM protein levels occurs at the transcriptional level , as RT-qPCR showed significant changes in the mRNA levels of both ADAM proteins upon MCPyV ST expression in both HEK 293 and MCC13 cells ( Fig 2E ) , correlating with recent results showing MCPyV ST can dynamically alter the transcriptome of human cells [26] . To further investigate the differential expression of ADAM 10 and 17 proteins in the context of MCC , multicolour immunochemistry analysis was performed on formalin-fixed , paraffin-embedded ( FFPE ) sections of primary MCC tumours . Sections were stained with ADAM 10 and 17 , cytokeratin 20 ( CK20 ) ( a marker widely used to distinguish MCC ) and MCPyV LT specific antibodies . An isotyped-matched control was also used as a negative control . CK20 staining confirmed MCC status of the sections and results show increased levels of ADAM 10 and 17 expression coincident with LT staining in regions of both MCPyV-positive MCC tumours ( Fig 3A ) . Moreover , immunoblot analysis was performed on cell lysates of two unrelated MCPyV-positive MCC tumour samples comparing protein levels against a negative control non-tumour cadaveric skin sample . Results again demonstrated a similar increase in both ADAM 10 and ADAM 17 protein levels in MCC tumour samples compared to control , which was MCPyV negative as indicated by the lack of ST and LT expression ( Fig 3B and 3C ) . Moreover , we compared the MCPyV-negative MCC13 cell line versus two MCPyV-positive cells lines , WAGA and PeTa . Similar results were observed showing that the presence of MCPyV ST increases ADAM 10 and 17 protein levels ( S1 Fig ) . Immunoblot analysis was also performed on cellular lysates of the MCPyV-positive MCC cell line , WAGA , transduced with lentiviruses containing a shRNA scrambled control or shRNA targeting ST , as previously described [31] . Results demonstrated that MCPyV ST depletion did not affect MCPyV LT levels but led to a reduction in ADAM 10 and ADAM 17 protein levels . Conversely , ST depletion leads to increased Alpha-E-catenin levels ( Fig 3D ) . To confirm these observations and determine if ADAM 10 transcripts are significantly increased in MCPyV-positive MCC compared with MCPyV-negative MCC , gene expression profiles for a total of ninety-four patients were obtained from a publicly available dataset ( accession number GSE39612 [9] ) . Bioinformatic analysis identified a significant increase ( 2 . 5 fold , p = 0 . 03 ) in ADAM 10 expression in MCPyV-positive MCC compared with MCPyV-negative MCC control samples . Moreover , a similar analysis was performed to analyse ADAM protein expression in control GFP versus MCPyV ST expressing cell datasets ( accession number GSE79968 ) [26] . A significant increase in both ADAM 10 ( p = <0 . 0001 ) and ADAM 17 ( p = <0 . 0001 ) was observed upon 48 hours MCPyV ST expression . Together these data suggest that ADAM 10 and 17 protein levels are increased upon MCPyV ST expression and in MCPyV-positive MCC tumour samples . For active ADAM proteins to cleave their chosen substrate , they are required to be present at the same subcellular location [59] . As adhesion molecule receptors are localised at the plasma membrane , we next determined whether MCPyV ST enhancement of ADAM 10 and 17 protein levels led to their accumulation at the plasma membrane [60] . HEK 293 cells transfected with EGFP or EGFP-ST were fixed and stained for endogenous ADAM 10 and ADAM 17 in non-permeabilised cells . MCPyV ST-expressing cells showed increased levels of both ADAM 10 and 17 proteins at the plasma membrane , in comparison to the EGFP control cells ( Fig 4A ) . To confirm these results , cell surface accumulation of ADAM proteins was measured by surface biotinylation assays in EGFP versus EGFP-ST expressing HEK 293 cells . Immunoblotting of surface biotinylated proteins confirmed that MCPyV ST expression specifically increased the plasma membrane levels of ADAM 10 and 17 proteins , in contrast the control cell surface protein , CD71 , showed no such increase ( Fig 4B ) . Densitometry-based quantification of the immunoblot analysis showed a significant increase in both ADAM 10 and 17 accumulation at the plasma membrane by 5 fold and 2 . 5 fold , respectively ( Fig 4C ) . Further validation was performed using flow cytometry with ADAM 10- and ADAM 17-specific antibodies ( Fig 4D and 4E ) . Notably however , both assays showed a greater accumulation of ADAM 10 compared to ADAM 17 at the cell surface . Together , these results suggest that MCPyV ST expression results in the accumulation of cellular sheddases , primarily ADAM 10 , at the plasma membrane . To determine whether ADAM protein accumulation at the plasma membrane is implicated in the observed disruption of cell junctions upon MCPyV ST expression , EGFP and EGFP-ST HEK 293-expressing cells were incubated in the absence or presence of two distinct ADAM protease inhibitors . MTS assays identified non-cytotoxic concentrations of an ADAM 10-specific inhibitor ( GI254023X ) and dual ADAM 10/17 inhibitor ( TAPI-2 ) ( S2 Fig ) , no specific ADAM 17 inhibitor is commercially available . Following a 24 hour incubation period , cells were fixed and non-permeabilised cells stained with an Alpha-E-catenin-specific antibody . As previously shown in Fig 1 , incomplete staining of the cell junctions was observed in MCPyV ST-expressing cells , compared to control EGFP cells . However , retention of the cell junctions was observed in the presence of both the ADAM 10-specific and dual ADAM 10/17 inhibitors , implying that inhibition of ADAM sheddase activity , and specifically ADAM 10 , is sufficient to prevent MCPyV ST-induced breakdown of cell-cell junctions ( Fig 5A ) . Importantly , there was no observed change in the cell junction staining in EGFP control cells after incubation with either inhibitor . The inhibition of MCPyV ST-induced cell junction breakdown was also confirmed by quantifying the cell surface levels of Alpha-E-catenin using flow cytometry in EGFP versus EGFP-ST-expressing cells . Results demonstrated increased levels of Alpha-E-catenin expression at the cell surface upon addition of the inhibitors ( Fig 5B ) . Notably , taking into consideration the greater accumulation of ADAM 10 over ADAM 17 at the plasma membrane in MCPyV ST-expressing cells and no enhancement of Alpha-E-catenin expression at cell junctions in the presence of the dual ADAM10/17 inhibitor over the ADAM 10 inhibitor alone , these results suggest that ADAM 10 may be the main cellular sheddase required for MCPyV ST-induced cell junction disruption . To confirm that ADAM 10 was required for the enhanced cell dissociation observed in MCPyV ST-expressing cells , the cell scatter assay was repeated in EGFP control and MCPyV ST-expressing cells , in the absence and presence of the ADAM 10 specific inhibitor , GI254023X , at non-cytotoxic concentrations . Addition of GI254023X resulted in little change in the EGFP-expressing control cells . However , a significant decrease in cell dissociation , over the course of 48 hours , was observed in the presence of GI254023X compared to DMSO-treated MCPyV ST-expressing cells ( Fig 6A ) . A similar level of cell dissociation inhibition was also observed using the ADAM10/17 dual inhibitor , TAPI-2 ( S3 Fig ) , showing that no enhancement of inhibition is seen by targeting both ADAM 10 and 17 . To confirm the specific role of ADAM 10 in MCPyV ST-induced cell dissociation , siRNA-mediated depletion of ADAM 10 was performed in EGFP and EGFP-ST-expressing HEK 293 cells ( Fig 6B ) . Immunoblotting confirmed that MCPyV ST depletion led to Alpha-E-catenin protein levels comparable to EGFP control cells ( Fig 6B and 6C ) . Cell scatter assays were then repeated in EGFP control or MCPyV ST-expressing cells , in the presence of either scrambled or ADAM 10-specific siRNAs . Depletion of ADAM 10 resulted in a similar reduction in cell dissociation levels observed with the specific ADAM 10 inhibitor ( Fig 6D ) . These data therefore suggest that ADAM 10 is required for the increased ability of cells to dissociate upon MCPyV ST expression . ADAM-mediated shedding of cell adhesion molecules may also stimulate cell signalling pathways to induce cell motility [30 , 31] . Therefore , we next examined if ADAM proteins have any downstream impact on the motility and migratory potential of MCPyV ST-expressing cells . Here , the migrating potential of EGFP control and EGFP-ST HEK 293 and MCC13-expressing cells were assessed using Incucyte kinetic live cell imaging , in the absence or presence of non-cytotoxic concentrations of the ADAM 10-specific ( GI254023X ) and dual ADAM 10/17 ( TAPI-2 ) inhibitors . Incubation of the ADAM 10 ( GI254023X ) inhibitor showed a slight but insignificant decrease in the motility of EGFP control cells , implying that any changes observed in migratory rates of MCPyV ST expression cells is not due to changes in cell viability or cytotoxicity . In contrast , ADAM 10 inhibition resulted in a significant decrease in the distance travelled of MCPyV ST-expressing cells , reminiscent of control cell migration ( Fig 7A ) . A similar trend was also observed with the dual ADAM 10/17 ( TAPI-2 ) inhibitor ( Fig 7B ) , suggesting that inhibition of ADAM 10 alone was sufficient to repress the MCPyV ST-induced cell migratory phenotype . To validate the use of ADAM-specific inhibitors , similar live cell imaging motility assays were also performed in ADAM 10-depleted EGFP and MCPyV ST-expressing HEK 293 cells , which resulted in a reduction in the motility of MCPyV ST-expressing cells , to levels similar to control EGFP-expressing cells ( Fig 7C ) . To demonstrate that ADAM 10 is required for cell motility and migration of MCPyV-positive MCC cell lines , haptotaxis migration assays were performed . This assay investigates the three-dimensional migration of cells towards a chemoattractant across a permeable chamber . Two MCPyV-positive MCC cell lines , WAGA and PeTa , were incubated in the absence or presence of the ADAM 10 inhibitor ( GI254023X ) at non-toxic concentrations assessed by MTS assay ( S4 Fig ) or upon siRNA-mediated scramble or ADAM 10-specific depletion . After treatment , cells were allowed to migrate for 24 h before migration was assessed by immunofluorescent staining of cells that had migrated into the chambers . Results showed that migration of MCPyV positive MCC cell lines were significantly reduced compared to control , upon treatment with GI254023X ( Fig 8A ) or upon ADAM 10 depletion ( Fig 8B ) , suggesting that MCPyV positive MCC cell line migration is ADAM 10 dependent . Together , these results suggest that ADAM 10 is required for MCPyV ST-mediated enhanced cell motility and migration .
MCPyV ST has emerged as the major transforming factor in MCPyV-positive MCC . Recently we reported a potential role for MCPyV ST in MCC metastasis , whereby ST cultivates a pro-migratory cell phenotype by destabilising microtubules [30] , inducing filopodia formation [31] and modulating cellular chloride channels [32] . Cancer metastasis occurs via a series of complex events that are collectively known as the invasion-metastasis cascade [61] . The apex event in the metastatic cascade is broadly accepted to be mediated by an EMT , providing tumour cells increased motility allowing invasion of the ECM . Most oncoviruses have been shown to manipulate the EMT axis , for example , human papillomavirus 16 , Epstein-Barr virus ( EBV ) , hepatitis B virus and the polyomavirus simian virus 40 have all been shown to induce metastasis , through a variety of mechanisms including; cellular adhesion complexes , cytoskeletal reorganisation and gene expression modulation [62–65] . EBV latent membrane protein-1 , for example orchestrates EMT via several different routes , including the transcriptional repression of E-cadherin via activation of DNA methyltransferases [66] and increased expression of the pleiotropic EMT transcription factors , Twist and Snail [67 , 68] . Here we expand on recent observations suggesting that MCPyV ST can trigger elements of the EMT and initiate the invasion-metastasis cascade , by demonstrating that MCPyV ST induces cell-surface expression of cellular sheddases , specifically ADAM 10 and 17 . Moreover , we show that MCPyV ST-mediated induction of ADAM 10 is required for MCPyV ST-induced cell-cell junction disruption which in turn enhances cell dissociation , migration and invasion . Although we focus herein on the link between MCPyV ST induction of ADAM proteins in metastatic spread , it must be noted that activation of ADAM10 may also serve in MCPyV fitness . Fibroblasts are a target of MCPyV infection [69] and is it known that MCPyV is shed from the surface of the skin , it is plausible therefore ADAM10 expression be a way for infected fibroblasts to migrate into the epidermis or hair follicle so the virus can be shed into the environment . How MCPyV ST regulates ADAM 10 expression is not yet clear , although results suggest this is likely to be at the transcriptional level . The ADAM 10 promoter contains functional binding sites for Sp1 and USF [70] and has been reported to be activated by numerous transcriptional activators including , XBP1 , JUN , ACAD8 , PPARG , SCAND1 and ITGB3BP [71 , 72] . Interestingly , ACAD8 , PPARG and ITGB3BP all appear in a recent RNA-seq data set of MCPyV ST-induced genes [26] , raising the possibility that these transcription factors may be responsible for MCPyV ST-mediated induction of ADAM 10 expression . There is a growing appreciation for the role played by ADAM proteins in numerous human diseases [73] , including Alzheimer’s disease , cardiovascular disease , rheumatoid arthritis and cancer [52] . The best characterised sheddase in terms of cancer aetiology is ADAM 17 , which is implicated in the development and progression of numerous neoplasms [74] . ADAM 17 came to prominence due to its ability to shed the soluble form of the inflammatory cytokine , TNFα from it precursor product [75 , 76] , however , despite TNFα being widely implicated in tumour development and progression , it is the ability of ADAM 17 to hydrolyse and promote the release of epidermal growth factor receptor ( EGFR ) /human EGFR ( HER ) precursor ligands that features most frequently in published studies . For example , ADAM 17-mediated shedding of TGFβ is implicated in breast [77 , 78] and renal [79] cancer progression . Moreover , release of the transmembrane protein with EGF and two follistatin motifs ( TMEFF2 ) increases prostate cancer cell motility [80] . We observed significant upregulation of ADAM 17 in response to MCPyV ST expression and in MCC tumours , however , comparison of ADAM 10 and ADAM 10/17 inhibitor experiments suggest that ADAM 17 is not required for the EMT-associated phenotypes observed following expression of MCPyV ST . This supposition is supported by bioinformatic analysis of MCPyV-positive MCC compared with MCPyV-negative MCC tumours , which identified significantly increased expression of ADAM 10 , but not ADAM 17 in 94 patient samples . The role of ADAM 10 in cancer metastasis is less clear , however emerging evidence suggests that ADAM 10 maybe cell-type specific , driving motility and invasion in breast [81] , pancreatic [82] , melanoma [83] and bladder [84] metastasis compared with primary tumours , but having alternative effects on proliferation in other tissue types . Interestingly , while HER ligand release is generally ADAM-specific , overexpression of individual ADAM proteins drives promiscuity in terms of ligand cleavage [85] . This raises the possibility that MCPyV ST-induced overexpression may enable ADAM 10 to cleave proteins ordinarily regulated by other sheddases , a scenario that needs to be considered when investigating downstream targets of ADAM 10 in MCC . Generally , metastasised MCC is treated with various regimens of broad-spectrum chemotherapy agents . However , metastatic MCC responses are not robust and often associated with high toxicity in elderly patients [86] . Response rates range from 52% to 61% in the distant metastatic setting , with progression-free survival ( PFS ) and overall survival typically measured in months [87–89] . One of the strongest predictors for survival is a high level of intratumoural CD8+ T cells most frequently observed in MCPyV-positive MCC [90 , 91] . MCPyV-specific CD8+ T cells express high levels of PD-1 and TIM-3 ( the T cell immunoglobulin and mucin domain-3 ) , which prompted immunotherapy-based clinical trials in MCC patients with the anti-PD-1 antibodies , pembrolizumab [92] and avelumab [93] . Both phase 2 trials reported encouraging and positive response rates with improved PFS , leading to pembrolizumab being listed as a treatment option for late-stage MCC in the National Comprehensive Cancer Network 2017 guidelines and avelumab being granted accelerated FDA approval as a first-line treatment for metastatic MCC . Whilst promising , around half of the patients involved in these clinical trials derived limited benefit from either drug [94] , indicating the importance of identifying additional agents to use in combination with anti-PD-1 antibodies . This approach may have exciting possibilities for ADAM 10/17 inhibitors , as TIM-3 is shed by both ADAM 10 and 17 and ADAM 10 cleaves MHC-I [95] . Notably , monoclonal antibody blocking of TIM-3 reduced PD-1 expression and increased cytokine production [96] , indicating that TIM-3 functions to dampen the immune system [97] . Therefore , ADAM 10 and 17 inhibitors may stimulate the immune system by reducing TIM-3 cleavage . One of the most widely characterised ADAM inhibitory compounds is INCB3619 ( Incyte ) , a dual ADAM 10 and 17 inhibitor which inhibits the catalytic activity of ADAM proteins by chelating zinc at the active site [53] . In vitro studies using breast and small cell lung cancer cell lines , have shown that INCB3619 reduced the cleavage of HER2 and amphiregulin , thereby sensitising cells to the EGFR tyrosine kinase inhibitor , gefintinib or a dual EGFR/HER2 inhibitor , GW2974 [98–100] . These observations have also been extended in animal models where INCB3619 shows anti-cancer activity against malignancies of the lung ( non-small cell ) , breast , head and neck [98 , 99] . Notably , a structurally similar compound with enhanced pharmokinetic properties , IMCB7839 ( Aderbasib ) , has undergone phase I/II clinical trials in patients with HER2-positive breast cancer , in combination with Herceptin ( trastuzumab ) . Results showed improved clinical responses in a subset of HER2-positive metastatic breast cancer patients , expressing the p95 form of HER2 [52 , 98] . At present , additional phase I/II clinical trials are ongoing , for example in patients with diffuse large B cell non-Hodgkin lymphoma using INCB7839 in combination with the monoclonal antibody rituximab [52] . Therefore , given our data showing a significant upregulation of ADAM 10/17 in MCC cell lines and tumours and the integral role played by ADAM 10 in MCPyV ST-mediated enhanced cell dissociation and invasion , selective inhibitors of ADAM 10 and 17 may prove to be potent novel therapeutics when given in combination with immune checkpoint inhibitors for the treatment of advanced MCC .
The expression vectors for EGFP-ST has been previously described [23 , 30 , 31] . MCPyV ST-tagging shRNA plasmids were kindly provided by Dr Masa Shuda , Pittsburgh . ADAM 10 and 17-specific siRNAs were purchased from Dharmacon . Antibodies against ADAM 10 , ADAM 17 , ADAM TS1 , and GAPDH were purchased form Abcam and used at a dilution range of 1:100–1:500 , the ZO-1 , CD71 and Alpha-E-catenin antibodies were purchased from Cell signalling and used at 1:100 dilution . The 2T2 hybridoma was provided by Dr Buck , National Cancer Institute , Bethesda , MD . All antibodies used for immunofluorescence were diluted 1:200 . ADAM 10 specific inhibitor , GI254023X and ADAM 10/17 dual inhibitor , TAPI-2 where purchased from TOCRIS and Merck Millipore , respectively . Cell toxicity was measured using a MTS-based CellTiter 96 AqueousOne Solution Proliferation assay ( Promega ) , as previously described [101] . HEK-293 Flip-In cell line was purchased from Invitrogen . i293-ST , i293-GFP , and i293-GFP-ST cell lines were derived from HEK-293 Flip-Ins using manufacturer’s protocol as previously described [23] . HEK-293 cells were obtained from ECACC and were maintained in Dulbecco’s modified Eagle’s medium ( DMEM ) containing 10% foetal bovine serum ( FBS ) and 1% penicillin/streptomycin as previously described [102] . The MCPyV negative cell line MCC13 ( ECACC ) and positive MCC cell lines , WAGA and PeTa ( ATCC ) , were grown in RPMI 1640 ( Sigma ) supplemented with 10% FBS . ST-FLAG , EGFP and EGFP-ST expression was induced from i293-ST , i293-GFP , and i293-GFP-ST cells respectively with 2 μg/ml Doxycycline hyclate for up to 48 hours . Cells were plated into 6-well plates and transfections routinely used 1 μg plasmid DNA and Lipofectamine 2000 ( Life Technologies ) or 5 μg plasmid DNA and nucleofection ( Lonza ) following the manufacturer’s instructions . Immunofluorescence was carried out as previously described [103] . If appropriate , cells were treated with inhibitors for24 hours prior to fixation . Cells were viewed on a Zeiss LSM880 confocal laser scanning microscope under an oil-immersion 63x objective lens . Images were analysed using the LSM imaging software as previously described [104] . EGFP and EGFP-ST-transfected cells were detached using Versene ( Sigma-Aldrich ) . The harvested cells were washed with ice-cold PBS and resuspended at 2x106 cells/ml in freshly made staining buffer ( PBS , 10% FCS , 3% BSA ) . Cells were then incubated with appropriate dilutions of primary antibody or staining buffer for 1 hour at room temperature in the dark , washed with staining buffer and then incubated with Alexa-Fluor-tagged secondary antibodies or staining buffer for 1 hour at room temperature . Cells were washed twice in PBS with centrifugation ( 350x g , 5 min ) and then analyzed by flow cytometry on a FACSCalibur , ( BD Bioscience , Wokingham , UK ) and the data analyzed using FlowJo software ( Tree Star , Ashland , OR , USA ) . Skin and MCC tumour biopsy samples were crushed using a pestle and mortar on dry ice , and homogenised by sonication prior to lysis in RIPA buffer ( 50 mM Tris-HCl pH 7 . 6 , 150 mM NaCl , 1% NP40 ) , supplemented with protease inhibitor cocktail ( Roche ) as previously described [105] . Proteins were separated by SDS-PAGE , transferred to nitrocellulose membranes and probed with the appropriate primary and HRP-conjugated secondary antibodies . Proteins were detected using EZ-ECL enhancer solution ( Geneflow ) as previously described [106] . Densitometry was performed using ImageJ software . RNA was extracted using TRIzol ( Invitrogen ) and DNase treated using the Ambion DNase-free kit , as per the manufacturer’s instructions , before RNA ( 1μg ) from each fraction was reverse transcribed with SuperScript II ( Invitrogen ) , as per the manufacturer’s instructions , using oligo ( dT ) primers ( Promega ) . 10ng of cDNA was used as template in SensiMixPlus SYBR qPCR reactions ( Quantace ) , as per manufacturer’s instructions , using a Rotor-Gene Q 5plex HRM Platform ( Qiagen ) , with a standard 3-step melt program ( 95 °C for 15 seconds , 60 °C for 30 seconds , 72 °C for 20 seconds ) as previously described [107] . With GAPDH as internal control mRNA , quantitative analysis was performed using the comparative ΔΔCt method as previously described [108] . EGFP and EGFP-ST-transfected HEK 293 cells were seeded in DMEM containing 10% FBS at a density of 2 × 104 per 35 mm culture dish . 18 hours later , cells were serum starved for 24 hours to induce aggregate formation . Upon reintroduction of serum , cells were fixed and stained with DAPI at 6 hourly intervals and clusters of cells were imaged using a Zeiss LSM880 confocal laser scanning microscope using a 10x objective lens . Images were analysed using the LSM imaging software to quantify the distance between each cell nucleus . Formalin-fixed , paraffin-embedded ( FFPE ) sections from primary MCC tumours were purchased from Origene and analysed as previously described [32] . Primary antibodies were: FITC-conjugated anti-CK20 ( Dako , dilution 1:50 ) , MCPyV LT CM2B4 ( Santa Cruz Biotechnology , dilution 1:125 ) and ADAM 10 and 17 ( Abcam , dilution 1:250 ) . An isotype-matched irrelevant antibody was used as a negative control on sections of tissues in parallel , a rabbit polyclonal isotype control antibody ( Abcam ) was used to match the ADAM 10 primary antibody . Sections were incubated with appropriate secondary antibodies labelled with different fluorochromes ( Alexa Fluor 546 IgG2B , 643 IgG2A , Invitrogen , and IgG ( H+L ) -TRITC , Jackson ImmunoResearch ) . All slides were mounted with Immuno-Mount and images were captured with a Zeiss LSM880 confocal laser scanning microscope . Metadata and pre-processed data ( FPKM ) were downloaded from Gene Expression Omnibus ( GSE79968 ) [26] and GSE39612 [9] . Data were normalised by the trimmed mean of M-values methods using edgeR package to account for batch effects and differences in sequencing depth among the samples using R/Bioconductor [109] . The differential expression analysis was performed using the R Bioconductor packages , voom and limma . Cell surface biotinylation was performed using the Pierce Cell Surface Protein Isolation kit ( Thermo Scientific ) according to the manufacturer’s protocol . Cells were incubated a cell-impermeable , cleavable biotinylation reagent , EZ-LINK Sulfo-NHS-SS-Biotin , to label exposed primary amines of proteins on the cell surface . After cell lysis , biotinylated cell surface proteins were affinity-purified using NeutrAvidin Agarose Resin ( Thermo Scientific ) . Precipitated proteins were then analysed using immunoblotting with ADAM 10- and ADAM 17- specific antibodies . A CD71-specific antibody was used as a suitable loading control . Cell motility was analysed using an Incucyte kinetic live cell imaging system as directed by the manufacturer . HEK293 cells or i293-GFP/i293-GFP-ST cells were seeded at a density of 25 , 000 cells per well of a 6 well plate , MCC13 cells were seeded at a density of 100 , 000 cells per well of a 6 well plate . After 12 hours , the cells were transfected with 1 μg of DNA per well and/or induced using doxycycline hyclate . For transfected cells , media was changed after 6 hours ( HEK-293 or derivatives ) or 12 hours ( MCC13 ) . If appropriate , cells were treated with inhibitors for 24h pre-imaging . Imaging was performed for a 24 hour period , with images taken every 30 minutes . Cell motility was then tracked and analysed using ImageJ software . Migration assays were performed using a CytoSelect 24-well Haptotaxis Assay Collagen coated plates ( Cell Biolabs , Inc ) , as directed by the manufacturer . All conditions were performed in triplicate . | The majority of cancer-related deaths occur due to metastatic disease . Therefore , understanding the molecular and cellular mechanisms underlying the process of metastasis is essential to developing new therapeutic interventions to improve cancer patient survival . Merkel cell carcinoma ( MCC ) is an aggressive and highly metastatic cancer . Merkel cell polyomavirus ( MCPyV ) has been implicated as the causative agent in the majority of MCC cases . The MCPyV small tumour antigen ( ST ) is believed to function as the major oncoprotein . However , little is known about the mechanisms through which MCPyV ST may be implicated in causing the high rates of metastatic spread observed in MCC tumours . Here we show that specific cellular sheddases , namely A disintegrin and metalloproteinase ( ADAM ) 10 and 17 protein levels are increased upon MCPyV ST expression . Moreover , we show that MCPyV ST-induced ADAM 10 and 17 are required to breakdown cell-cell junctions resulting in increased cell dissociation , migration and invasion . As such , ADAM protein expression may provide a novel biomarker of MCC prognosis . In addition , linking cellular sheddases to MCPyV-positive MCC metastasis may provide novel therapeutic interventions . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"cell",
"motility",
"medicine",
"and",
"health",
"sciences",
"molecular",
"probe",
"techniques",
"hek",
"293",
"cells",
"biological",
"cultures",
"immunoblotting",
"basic",
"cancer",
"research",
"membrane",
"proteins",
"oncology",
"developmental",
"biology",
"molecular"... | 2018 | Cellular sheddases are induced by Merkel cell polyomavirus small tumour antigen to mediate cell dissociation and invasiveness |
Pathogenic spirochetes are bacteria that cause a number of emerging and re-emerging diseases worldwide , including syphilis , leptospirosis , relapsing fever , and Lyme borreliosis . They navigate efficiently through dense extracellular matrix and cross the blood–brain barrier by unknown mechanisms . Due to their slender morphology , spirochetes are difficult to visualize by standard light microscopy , impeding studies of their behavior in situ . We engineered a fluorescent infectious strain of Borrelia burgdorferi , the Lyme disease pathogen , which expressed green fluorescent protein ( GFP ) . Real-time 3D and 4D quantitative analysis of fluorescent spirochete dissemination from the microvasculature of living mice at high resolution revealed that dissemination was a multi-stage process that included transient tethering-type associations , short-term dragging interactions , and stationary adhesion . Stationary adhesions and extravasating spirochetes were most commonly observed at endothelial junctions , and translational motility of spirochetes appeared to play an integral role in transendothelial migration . To our knowledge , this is the first report of high resolution 3D and 4D visualization of dissemination of a bacterial pathogen in a living mammalian host , and provides the first direct insight into spirochete dissemination in vivo .
Pathogenic spirochetes are bacteria that cause a number of emerging and re-emerging diseases worldwide , including syphilis , leptospirosis , relapsing fever and Lyme borreliosis [1]–[6] . Many clinically-important spirochetes cross the blood-brain barrier and exhibit an unusual form of motility that is predicted to permit efficient movement through dense extracellular matrix in host tissues [6]–[9] . Spirochetes of the Borrelia burgdorferi sensu lato species complex are the causative agents of Lyme borreliosis [1] , [10] . B . burgdorferi are transmitted to the skin of mammalian hosts through the bite of an infected tick . Subsequently they enter the vascular circulation and disseminate hematogenously to multiple tissues by unknown mechanisms . Untreated Lyme borreliosis can result in arthritis , carditis and neurological complications . B . burgdorferi and other spirochetes interact with endothelial cells under static conditions in vitro [11]–[13] . However , spirochete-vascular interactions have never been examined in the host itself , or under the fluid shear forces that are present at dissemination sites . Indeed , host-pathogen interactions under shear stress conditions are still poorly understood for most bacterial pathogens that invade or disseminate in the mucosa or blood vessels , despite the importance of shear forces present in these environments . Early studies with cultured endothelial cells found that treatments or mutations that rendered B . burgdorferi non-motile impaired invasion but not interaction [14]–[17] , suggesting that the spirochete's ability to bore through dense tissues using translational motility might be important for vascular invasion . However , all previous investigations of B . burgdorferi dissemination were performed ex vivo in the absence of shear stress , using endothelial cell monolayers incubated with B . burgdorferi for periods as long as 24 hours , and employed non-dynamic visualization techniques such as electron microscopy which precluded observation of spirochete movement [14] , [15] , [17]–[19] . Conflicting reports found that extravasating B . burgdorferi were localized exclusively in either endothelial junctions or cells [14] , [15] , [18] . The role of host cells in transmigration was also controversial , because electron microscopy studies revealed no unambiguous evidence of endocytosis , and since the host microfilament toxin cytochalasin D did not inhibit spirochete internalization [14] , [15] . Thus , the mechanism of B . burgdorferi dissemination in mammalian hosts remains a mystery . It has been challenging to study host-spirochete interactions in a living host because their slender ( <1 µm ) morphology makes them difficult to visualize by standard light microscopy . Direct observation of Lyme borreliosis spirochete interactions with mammalian cells has been limited to cell culture models or host tissues removed from their native context . Several green fluorescent protein ( GFP ) alleles have been expressed in B . burgdorferi , usually in the context of reporter constructs used to monitor gene transcription and plasmid maintenance in spirochetes grown in culture [20]–[23] . Recently , a GFP reporter construct was used to monitor gene expression during B . burgdorferi infection , in tissues that were excised from the host before visualization [24]; however , detection of the GFP allele used in this study required relatively long , two second exposure times [23] . Intravital microscopy ( IVM ) is a powerful tool for studying the cellular dynamics of the immune and cardiovascular systems and tumor metastasis in the context of a living organism [25] , [26] . It has also been used to visualize tissue localization dynamics of bacterial pathogens in vertebrate hosts [27] , [28] . However , the small size of most pathogens and the spatial resolution limits of conventional epifluorescence IVM have impeded analysis of live host-pathogen interactions at the single cell level . The application of spinning disk confocal microscopy in an intravital setting finally enabled real-time visualization of transmission of malaria parasites to a living host [29] , but unambiguous analysis of dissemination by smaller pathogens such as bacteria requires the ability to perform three-dimensional ( 3D ) microscopy in vivo . In the current study we report the first use of spinning disk confocal microscopy to visualize dynamic host-pathogen interactions in three and four dimensions , revealing many aspects of the B . burgdorferi dissemination process that have not been previously observed , in vitro or in vivo . The ability to examine bacterial pathogenesis over time and in the three-dimensional space of living hosts will greatly enhance our understanding of many infectious diseases .
B . burgdorferi are difficult to modify genetically , and transformation with recombinant constructs often results in the loss of plasmids that are required for infectivity in the mouse [30] . We were able to engineer infectious and non-infectious B . burgdorferi expressing a highly fluorescent GFP allele optimized for bacterial expression [20] , [31] ( Fig . 1A ) ; this allele is distinct from the egfp [21] , [23] and gfpmut1 [21] , [23] alleles that have also been expressed in B . burgdorferi . The resulting infectious ( GCB726 ) and non-infectious ( GCB705 ) strains displayed similar levels of GFP fluorescence , which could be detected with very short exposure times of less than 100 ms with a conventional epifluorescence microscope . The infectious strain contained the full complement of B . burgdorferi plasmids required for infectivity ( see Materials and Methods ) . To confirm that fluorescent strain GCB726 was infectious , and to determine if GFP expression could be stably maintained without antibiotic selection in infected murine hosts , two C3H/HeN and two C57 BL/6 mice were inoculated with 5 . 5×104 fluorescent spirochetes . Two ear punches were collected for each mouse 13 days ( C3H mice ) or 28 days post-infection ( C57 mice ) , and cultured in B . burgdorferi growth medium with or without gentamycin selection . Spirochetes were recovered from all ear punches in both the presence and absence of antibiotic , confirming that strain GCB726 was infectious in both C3H and C57 mice . Furthermore , 98 . 7 −/+ 1 . 3% of the B . burgdorferi cultivated in the absence of gentamycin retained robust levels of GFP expression ( Fig . 1B ) , indicating that the GFP-expressing plasmid was stably maintained in the context of the murine host . We next investigated whether fluorescent B . burgdorferi could be exploited for real-time studies of host-spirochete interactions using intravital microscopy ( IVM ) . Fluorescent infectious spirochetes were observed in situ in the ears of living C3H mice 20 and 27 days post-infection and in the ears of C57 mice 28 days after infection , using both conventional epifluorescence and spinning disk confocal IVM . Consistent with a previous report that B . burgdorferi localize to the perivascular connective tissue [32] , fluorescent B . burgdorferi were observed outside , but usually close to blood vessels , and frequently translated back and forth repetitively over a relatively lengthy distance ( Fig . 1C and Video S1 ) . In situ , B . burgdorferi exhibited all of the translational ( running ) and non-translational ( flexing ) modes of movement that are characteristically observed in culture medium [7] , [33] . Although spirochetes traveled more slowly and reversed directions more frequently when passing around visible obstacles such as blood vessels , they could achieve speeds of up to 4 . 0 µm/s in the ear , a speed which is very similar to the previously reported 4 . 25 µm/s in vitro rate [34] . Interestingly , although B . burgdorferi in situ often reversed their direction of movement every few seconds , which is typical of translational motility in vitro [35] , they also exhibited sustained unidirectional movement for periods as long as 40 seconds . In order to investigate the behavior of spirochetes in the host microvasculature , fluorescent B . burgdorferi were injected directly into the bloodstream of C57 mice via the jugular or femoral veins , and were visualized in real-time using both conventional epifluorescence and spinning disk confocal IVM . Prior to inoculation , fluorescent spirochetes were cultured for 48 hours in the presence of 1% mouse blood to promote adaptation to the host environment , since growth in blood is known to regulate the expression of many B . burgdorferi genes [36] . Vascular interactions were analyzed in flank skin , where the best optical clarity was obtained ( see Fig . 2 and Video S2 ) . The relevance of this site as a target for B . burgdorferi dissemination was confirmed by recovery of spirochetes from cultures of skin taken from mice 28 days post-infection . Fluorescent B . burgdorferi maintained a stable density in the bloodstream for longer than four hours after injection , but interactions with the microvasculature were analyzed between 5 and 45 minutes after injection during which time endothelial activation was not observed and spirochete titers and rates of interaction were stable ( see Text S1 for data and discussion related to the lack of endothelial activation ) . Examination of interactions between fluorescent B . burgdorferi and the microvasculature in more than 40 mice yielded several general observations . First , at similar blood densities , non-infectious fluorescent spirochetes did not associate with blood vessels , though they remained in the circulation , indicating that vascular interactions were not an artifact caused by high blood titers or by mechanical impediments to cell flow . Second , infectious spirochetes associated with capillaries , postcapillary venules and larger veins , but not with arterioles ( Fig . 2 , Videos S2 , S3 and S4 ) . Vessel identity was determined by measurement of vessel diameters and by observation of blood flow patterns in the immediate vascular network ( convergence indicates venules while divergence identifies arterioles ) . The observation that spirochetes did not interact with the lumenal surface of arterioles differs from the conclusions derived from a previous report of mice infected intradermally with B . burgdorferi , which found that after several weeks of infection spirochetes were preferentially localized to the walls of arterial vessels [32] . One possible explanation for this discrepancy is that in the previous study spirochetes might have migrated into the walls of arterial vessels from extravascular tissues , an event that is unlikely to have occurred during the short time frame of our experiments . Additionally , colonization of the connective tissue-rich walls of arteries could be promoted by bacterial adaptation to the host environment during longterm infections . It was unlikely that the inability of B . burgdorferi to interact with arterioles in the time frame of our experiments was due to differences in expression of host cell ligands in arterial and venous vessels , since spirochetes readily associated with the arterial endothelium under conditions of reduced blood flow . Therefore , reduced spirochete interactions in arterioles may have resulted from the elevated shear forces present in these vessels . Third , interacting spirochetes in capillaries sometimes moved back and forth with and against the direction of blood flow ( Fig . 2 , Video S2 ) . In contrast B . burgdorferi usually moved with the direction of flow in venules and veins where blood flow was more rapid , but moved freely in multiple directions under conditions of reduced blood flow . From these qualitative observations we infer that spirochete interactions with the host microvasculature are strongly affected by blood flow . Finally , B . burgdorferi grown in the presence or absence of blood did not exhibit significant differences in the total number or type of interactions . This suggested that blood-stimulated gene expression in B . burgdorferi [36] was not a requirement for microvascular interaction . Quantitative analysis of B . burgdorferi interactions in postcapillary venules ( where interaction rates could be most accurately quantified ) revealed two major types of associations: short-term interactions and stationary adhesions ( Fig . 3A ) . Associations were characterized and quantified in venules because interactions in larger veins were too numerous for accurate quantification and blood flow in capillaries could be blocked by trapped spirochetes . Interactions were quantified using conventional video-based epifluorescence IVM , which captures rapid adhesion events more effectively than spinning disk confocal IVM . All spirochetes that paused and associated , even briefly , with the vessel wall were counted , and the length of time required to travel 100 µm along the vessel wall was measured . B . burgdorferi that did not associate with vessels moved very rapidly , and were visible only as blurs; therefore , non-interacting spirochetes could not be quantified . However , the total number of B . burgdorferi in the bloodstream could be quantified by counting the number of spirochetes in blood samples using cell-counting chambers and was similar in all experiments . Short-term interactions included two sub-groups: transient and dragging interactions . Transient interactions , which constituted the majority of interactions , were defined as those where B . burgdorferi slowed , associated briefly with the endothelium then detached in a tethering-type interaction cycle ( for examples , see Videos S3 and S4 ) . Transiently associating spirochetes took less than 1 second to travel 100 µm in the vessel , but moved at least 8 times more slowly than the speed of blood flow , and frequently interacted only partially with the endothelium . Transient interactions could occur at the tip of the bacterium or elsewhere on the bacterial cell body , implying that the ends of B . burgdorferi are not the exclusive sites of tethering . Transiently associating spirochetes that interacted with the endothelium along much of their length often slowed further and began dragging or crawling along the vessel wall . Short- and long-drag interactions were those in which spirochetes took 1–3s and 3–20s , respectively , to travel 100 µm along the vessel wall ( for an example , see Video S4 ) . B . burgdorferi dragging along the endothelium frequently slowed and stopped before dragging further along the wall in the direction of blood flow . The crawling movement observed at this stage of interaction may be similar to the crawling motion described for Leptospira , which results from translational motility in the context of simultaneous tethering of the spirochete at multiple distinct interaction sites [37] . Spirochetes that remained stationary at a single position on the vessel wall and did not translate along the vessel for at least 20 seconds were defined as stationary adhesions ( for an example , see Video S3 ) . Once stationary , these bacteria usually remained in the same place for at least two minutes ( average: 10 minutes ) , and were aligned lengthwise along the vessel wall in the direction of blood flow . One end of stationary adhesions was usually less adherent and more mobile than the other , and sometimes exhibited a probing-type behavior that was more consistent with active gyration of the free spirochete end than with a passive rearrangement due to blood flow . The most stably adhered end was always pointing in the opposite direction to blood flow . To more closely examine B . burgdorferi interactions with the host microvasculature we performed 3D reconstruction on z-series micrographs of spirochetes and PECAM-1-stained vessels obtained used spinning disk confocal IVM . PECAM-1 is expressed on endothelial cells and concentrates at endothelial cell junctions [38] , making it useful for visualizing both the lumenal endothelial surface of vessels , as well as the more PECAM-1-intensive intercellular junctions . PECAM-1 was visualized using an Alexa Fluor 555-conjugated antibody to PECAM-1 that has been used previously to study junctional extravasation of leukocytes in vivo [39] . Three-dimensional visualization of short-term interactions and stationary adhesions in venules revealed that these two classes of associating spirochetes differed with respect to their position relative to the PECAM-1-stained endothelium ( Fig . 3B and Video S5 ) . Greater than 93% of short-term interactions were localized on the lumenal surface of the vessel wall and were not observed to project into the PECAM-1-stained endothelium in three dimensional reconstructions ( see positions 1 and 2 of the dragging spirochete in Fig . 3B lower and side panels ) . In contrast , a large majority of stationary adhesions ( 79% , Fig . 3C ) were embedded in the PECAM-1-stained endothelium , either partially ( 57% ) or along their entire length ( 21% ) ( Fig . 3C; see positions 3 and 4 of the stationary spirochete in Fig . 3B lower and side panels ) . Video S5 presents a reconstructed three-dimensional view of a typical short-term interaction and stationary adhesion in a venule , with the short-term interaction visible only on the lumenal surface of the endothelium , and the adhering spirochete visible in both the lumen and projecting through the PECAM-1 . The right hand panel of Fig . 3B shows lumenal and exterior views of these spirochetes from a 3D reconstruction . Because the PECAM-1 antibody stains the lumenal surfaces of endothelial cells , these data indicate that stationary adhesions are embedded more deeply in the endothelium than transiently interacting spirochetes , but do not imply that stationary adhesions project beyond the external boundary of vessels . Also of interest , in the majority of stationary adhesions ( 71% ) , one end of the bacterium projected further into the PECAM-1 than the other end . The most deeply embedded portion of the adhesion was usually the most stable , since more superficially attached regions of the spirochete exhibited a greater range of movement ( see Video S3 for an example ) . This observation suggested that stationary adhesions might be slowly extravasating through the endothelium via the more deeply embedded tip . However , we did not detect any consistent outward migration of stationary adhesions during the experimental time period studied ( up to 45 minutes ) , although it remains possible that such emigration might take much longer to occur . Finally , the localization of endothelium-interacting spirochetes was more precisely determined by examining the position of these interactions with respect to endothelial junctions ( which are stained more intensely by anti-PECAM-1 antibody than the non-junctional surface of endothelial cells ) ( Fig . 4 ) . To determine if PECAM-1 redistribution occurred in response to B . burgdorferi , we visualized junctions using PECAM-1 antibody before and after injection of infectious spirochetes , and examined junctional staining patterns from 5–45 minutes after spirochete injection . During this time frame , we observed no PECAM-1 redistribution in 95% of venules examined ( n = 20 venules in 5 mice ) . Since junctional staining was sometimes incomplete , localization of interacting spirochetes was assigned only when the junctional boundaries of endothelial cells in the area of interaction were clearly demarcated . For the majority of stationary adhesions ( ∼70% ) , the most stable , deeply embedded region of adhesion occurred at junctions . However , about 25% were adhered primarily to endothelial cells ( Fig . 4B ) . In contrast , the vast majority of transient and dragging short term interactions ( 93% ) occurred on the surface of cells , with only 7% being found at junctions . Since cell surfaces make the largest contribution to total endothelial surface area , short-term interaction with cells may be a stochastic event , whereas stationary adhesion to junctions is likely the result of preferential localization . B . burgdorferi escaped the microvasculature in an end-first fashion , and therefore projected out of the planes of view where most interactions were observed; thus , 2D visualization alone was insufficient for unambiguous analysis of this final stage of dissemination . Using 4D spinning disk confocal IVM ( 3D time courses ) , we measured the percentage of each spirochete's length that projected beyond the PECAM-1-labeled endothelium in successive z-series , and calculated the time span and speed of escape ( Fig . 5A–C ) . As for stationary adhesions , most escaping spirochetes ( 83% ) extravasated through endothelial junctions . Transmigrating spirochetes preceded by stationary adhesion were not observed in this study , although the time necessary to acquire successive z-series in 4D IVM might have precluded detection of short-lived adhesions that began extravasating . Escape took an average of 10 . 8 minutes , at an average net displacement velocity of 3 . 4 µm/min ( Fig . 5c ) . The initial and final stages of the escape process were too rapid to capture visually in 4D , since they were faster than the 1–2 minutes necessary to acquire individual z-series ( for sample 2D footage of the final stage of escape , see Video S6 ) . Little net displacement occurred during the longer middle phase of escape , even though many bacteria in this phase exhibited obvious reciprocal translational motility ( Fig . 5D and Video S7 ) . Reciprocally translating spirochetes could move in either direction as quickly as 624 µm/min , a speed which greatly exceeded their net displacement velocity . The great speed of these bacteria in situ might thus have accounted for our difficulty in capturing the initial and final stages of extravasation . The abridged timelapse shown in Fig . 5A illustrates the typical triphasic escape dynamic . In this case , 41% and 43% of the spirochete length passed out of the PECAM-1-stained endothelium in the first and last 2 minutes of extravasation , respectively , whereas only 16% of the spirochete traversed the PECAM-1 layer in the intervening 16 minutes . It was unlikely that the speeds of the initial and final stages of extravasation were the result of passive drifting of the bacteria through the endothelium , leading us to conclude that transmigration was largely driven by spirochete motility . Furthermore , the speed of the final escape phase , in which spirochetes appeared to burst away from the vessel ( Video S6 ) , suggested that the reciprocal translational motility observed in the middle phase was the result of partial adhesion of either the middle or the lagging portion of the spirochete to the endothelium [37] . Together , these observations suggested a prominent role for spirochete motility in the final stage of dissemination .
In this work technological advances in confocal microscopy have been coupled with intravital imaging methodologies to allow for the first time , high resolution , three and four dimensional , real-time visualization of the interaction of a bacterial pathogen with its living host . We have used this technology to study the interaction of the Lyme borreliosis spirochete B . burgdorferi with the microvasculature of one of its natural hosts . One of the central events in the development of spirochetal diseases is hematogenous dissemination [40] . Previous investigations of dissemination by pathogenic B . burgdorferi were performed in a static environment using endothelial cell monolayers incubated with B . burgdorferi for several hours or longer , and methodologies that precluded direct observation of spirochete behavior [14] , [15] , [17]–[19] . In contrast , dynamic , 3D and 4D analyses of interactions in a living host under shear stress conditions indicate that B . burgdorferi escape from the microvasculature is a multi-stage process ( as summarized in Fig . 6 ) . Spirochetes first transiently tether to the endothelium , usually at cell surfaces and not intercellular junctions ( Fig . 6A ) , then drag and crawl along the vessel wall while interacting with the endothelium along much of their length ( Fig . 6B ) . In contrast , stationary adhesions are usually established , at intercellular junctions ( Fig . 6C ) , which also appear to be the major site for B . burgdorferi extravasation ( Fig . 6D ) . Both stationary adhesion and extravasation may , therefore , be mediated by host and spirochete molecules distinct from those involved in short-term interactions . It remains unclear if stationary adhesions represent an obligate step in the progression toward vascular escape , or if they act as facilitators of this event by modifying the endothelium ( see below ) . Similarly , multiple types of interactions are observed during leukocyte trafficking under the shear stress conditions of blood flow , which depends on a progressive association between different classes of endothelial and leukocyte molecules as the interacting cell slows down and locates an extravasation site [41] . It is probable that B . burgdorferi interactions with , and escape from the endothelium entail a similar progression . Previous reports indicate that B . burgdorferi invade cultured endothelial cells by both intracellular and intercellular routes [14] , [15] , [18] . Treponema pallidum generally migrate through endothelial monolayers via intercellular junctions , whereas Leptospira primarily invade endothelial cells themselves; however , these spirochetes have also been observed in endothelial cells and junctions , respectively [12] , [13] . Our results indicate that although the major extravasation route of B . burgdorferi in vivo is the intercellular junctions , a small percentage can also emigrate through endothelial cells . This conclusion raises the interesting possibility that other spirochetes could also exhibit the same versatile invasive capacity in vivo . B . burgdorferi are known to interact with multiple host molecules that could mediate interaction with and invasion of the host endothelium in vivo; these include fibronectin , plasminogen , glycosaminoglycans , and integrins such as the vitronectin and fibronectin receptors [18] , [42]–[47] . Indeed , we are currently using the technology described here to further study B . burgdorferi adhesion and have thus far identified several of the host and bacterial molecules involved at specific steps in the adhesion process ( manuscript in preparation ) . These observations support a progressive model of spirochete adhesion under shear stress conditions in which different classes of host and B . burgdorferi proteins mediate distinct phases of interaction . One of the most interesting , unprecedented and difficult findings to interpret in our study was that stationary adhesions projected deep into and sometimes through the PECAM-1-stained region of vessels , a phenomenon we refer to as “embedding . ” Embedding could occur along the entire length of the spirochete , or at one end only . Interestingly , we found that B . burgdorferi embedded in the PECAM-1 region along their entire length adhered for much longer periods than partially embedded bacteria , and were frequently observed protruding through both sides of the PECAM-1 signal ( e . g . see the stationary spirochete in Fig . 3B , lower panel ) , suggesting that they had migrated more deeply into junctions or endothelial cells than partially embedded adhesions . This observation may be consistent with the results of early electron microscopy studies demonstrating that B . burgdorferi can invade or be taken up by endothelial cells in monolayer cultures [15] , [19] , and is intriguing in light of previous proposals that spirochete evasion of the host immune system is mediated by “seeding” bacteria that escape immune surveillance in physically protected sites ( reviewed in [48] ) . Endothelial cells can be as thin as 0 . 1 µm [49] and the PECAM-1 antibody used in this study stained a 3 µm-thick region of the vessel wall . The observation that stationary adhesions often project beyond the PECAM-1-stained region suggests the possibility that these spirochetes are invading junctions or endothelial cells . However , the measured thickness of the PECAM-1 signal may overestimate the dimensions of the endothelium due to motion artifacts caused by respiration of the immobilized mouse . Therefore , we can only conclude that the apparently embedded state of stationary adhesions results from more intimate adhesion to the endothelium than that observed for short-term interactions . Additional higher resolution studies of stationary adhesions , performed under shear stress conditions , will be required to shed light on the true position of the spirochetes relative to the endothelium and the intriguing possibility that stationary adhesion of spirochetes might provide a protective mechanism for evasion of the immune response . Spirochete escape from the microvasculature was a rare event , even after intravenous inoculation with large doses of B . burgdorferi . Three dimensional timelapse data captured for 30 emigrating spirochetes revealed that B . burgdorferi escaping the microvasculature traversed the vessel wall end-first ( Fig . 6D ) . Interestingly , multiple emigrating spirochetes were sometimes observed in the same vessel ( data not shown ) . It appears unlikely that cases of multiple escape were the result of endothelial activation in response to B . burgdorferi , since these could be observed immediately after intravenous injection of spirochetes . Another possibility is that the presence of nearby stationary adhesions facilitated transmigration , since these adhesions were more abundant in vessels with escaping spirochetes , and as the site of transmigration was frequently in close proximity to a stationary adhesion ( data not shown ) . Stationary adhesions adjacent to escape sites might modify their immediate vascular environment to promote emigration of other spirochetes . Spirochetes emigrating end-first frequently exhibited a reciprocal translational form of movement that might drive much of the escape process . The average displacement velocity of emigrating spirochetes was 4-fold less than the average displacement velocity of B . burgdorferi translating in the extravascular tissues of the ear , suggesting that the endothelium presents significant physical barriers to transmigration . This conclusion is supported by the observation that the middle phase of escape was very slow relative to early and late stages . Previous work with Leptospira in vitro indicates that even cells adhered to immobile surfaces at a single point can move in a reciprocating fashion referred to as “staple movement” , likely as a result of a rapid ( 11 µm/s ) lateral displacement of the adhesion site within the spirochete outer membrane [37] , [50] . Such a model predicts that disruption of the adhesion site would cause a rapid change in the motile behavior of spirochetes , which is consistent with our observation that escaping B . burgdorferi “burst” out of the endothelium after a protracted period of reciprocal motility . Early studies of B . burgdorferi , T . pallidum and Leptospira transmigration performed with endothelial monolayers found that non-motile spirochetes could not invade endothelium [12]–[17] . This conclusion is supported by the real-time imaging data reported in this study . It will , therefore , be important to directly examine the role of spirochete motility in emigration by coupling the use of the well-characterized B . burgdorferi motility and chemotaxis mutants [35] , [51]–[53] with the technology reported here . We found that live fluorescent spirochetes can easily be observed in situ in living mice one month after subcutaneous and peritoneal inoculation . Furthermore , intravital microscopy can be performed in many of the tissues targeted by B . burgdorferi and other spirochetes , including the brain , liver , lung and joint cartilage [25] . It is , therefore , clear that the methodology described here could be a powerful tool for addressing a broad range of questions about host-pathogen interactions . In addition to the types of experiments reported here , this methodology could be exploited for the study of a variety of bacterial pathogens in terms of their invasion of the vascular system , interactions with cellular components of the innate and acquired immune responses , for monitoring gene expression and migration patterns in different tissues over the course of infection and for analysis of chemotactic behavior in the host . The methodology may also be useful for monitoring events that occur immediately after needle or tick bite inoculation , routes of spirochete entry that more closely recapitulate the natural infection process than the intravenous injection of high numbers of blood adapted spirochetes . In summary , dynamic and high resolution three-dimensional analyses of B . burgdorferi behavior in a living host have revealed numerous previously unobserved aspects of spirochete interaction with , and escape from , the host vasculature . The application of this powerful approach to the study of other micro-organisms is certain to enhance our understanding of the broad and always unpredictable repertoire of pathogenic agents and their interactions with their living hosts .
The terminator sequences ( T1 x 4 ) , rbs , B . burgdorferi flaB promoter and GFP coding sequences from pCE320 ( gfp ) -PflaB [20] were PCR-amplified with flanking SacI and KpnI sites , using primers B696 ( 5′-ccggagctcatgataagctgtcaaacatgag-3′ ) and B697 ( 5′-ccggtacctcagatctatttgtatagttcatc-3′ ) , and cloned into pCR Blunt II-TOPO ( Invitrogen Canada , Burlington , ON ) with the insert SacI site proximal to the vector PstI site , to make plasmid pTM41 . This insert could not be cloned into the gentamycin-resistant version of the pBSV2 shuttle vector ( pBSV2G ) [54] , presumably because replication origins and copy number sometimes affect the expression and toxicity of fluorescent proteins in E . coli . Therefore , a modified shuttle vector , pTM49 , was constructed , in which the colEI ori of pBSV2G was removed by restriction digestion with enzymes MluI and SnaBI , and replaced with an MluI/SnaBI fragment from pCR Blunt II-TOPO containing the pUC ori . The ( T1 x 4 ) -PflaB-gfp cassette from pTM41 was cloned into the SacI/KpnI sites of pTM49 to generate pTM61 . All strains were grown in BSK-II medium prepared in-house [55] . Electrocompetent infectious B . burgdorferi strain B31 5A4 NP1 [56] and non-infectious strain B31-A [57] ( both B31-derived ) were prepared as described [58] . Liquid plating transformations were performed with 50 µg pTM61 in the presence of 100 µg/ml gentamycin as described [59] , [60] . Gentamycin-resistant B . burgdorferi clones were screened for: 1 ) the presence of aacC1 sequences by colony screening PCR performed with primers B348 and B349 as described [61]; and 2 ) GFP expression by conventional epifluorescence microscopy . The presence of the pTM61 plasmid in non-integrated form in fluorescent strains was confirmed by agarose gel electrophoresis of total genomic DNA prepared on a small scale as described [62] . PCR screening for native plasmid content was performed as described [61] , [63] and indicated that one fluorescent infectious B . burgdorferi clone ( GCB726 ) contained all endogenous plasmids except cp9 , which was displaced by the cp9-based pTM61 construct . Non-infectious strain GCB705 was used for experiments with non-infectious B . burgdorferi . PCR screening for native plasmid content indicated that GCB705 contained the same plasmids as the B31-A parent [61] ( lp17 , lp28-2 , lp28-3 , lp38 , lp54 , lp56 , cp26 , cp32-1 , cp32-2/7 , cp32-3 and cp32-9 , but not lp21 , lp25 , lp28-1 , lp28-4 , lp36 , cp9 , cp32-6 or cp32-8 ) . Plasmids lp25 , lp28-1 and lp36 are known to be essential for infectivity [63] , [64] . All animal studies were carried out in accordance with approved protocols from the University of Calgary Animal Research Centre . C3H/HeN ( Harlan , Indianapolis , IN ) and C57 BL/6 ( Jackson Laboratory , Bar Harbor , ME ) mice were infected by both intraperitoneal ( 5×104 cells/ml ) and subcutaneous ( 5×103 cells/ml ) needle inoculation . Ear punches and flank skin samples were cultured in Barbour-Stoenner-Kelly II ( BSK-II ) medium supplemented with 6% rabbit serum ( Cedarlane Laboratories Ltd . , Burlington , ON ) with or without 100 µg/ml gentamycin . The percentage of ex vivo spirochetes that continued to express robust levels of GFP was calculated by counting the number of fluorescent spirochetes at 100 ms exposures compared to the number detected by phase contrast visualization . For each experiment , infectious or non-infectious strains expressing GFP were freshly inoculated from glycerol stocks into 15 ml BSK-II medium containing 6% rabbit serum and 100 µg/ml gentamycin . B . burgdorferi were grown to 5×107/ml , then diluted to 1–2×106/ml in BSK-II medium containing 6% rabbit serum , 100 µg/ml gentamycin , 1× Borrelia antibiotic mixture ( 20 µg/ml phosphomycin , 50 µg/ml rifampicin and 2 . 5 µg/ml amphotericin B , prepared from individual antibiotics obtained from Sigma ) and 1% C57 BL/6 mouse blood . Spirochetes were grown in the mouse blood for 48 hours at 35°C to a final density of ∼5×107/ml . B . burgdorferi were pelleted ( 6 , 000×g for 15 min at 4°C ) , washed twice in PBS ( Invitrogen Canada , Burlington , ON ) , and resuspended to 2×109 B . burgdorferi/ml in PBS . All experiments were performed at a final density of ∼1×107 spirochetes/ml of blood to facilitate quantitative analysis of interactions . Animals were anaesthetized by intraperitoneal injection of a mixture of 10 mg/kg xylazine hydrochloride ( MTC Pharmaceuticals , Cambridge , ON ) and 200 mg/kg ketamine hydrochloride ( Rogar/STB , London , ON ) . As previously described [65] , a depilatory solution ( Nair; Armkel LLC ) was applied to the dorsal and ventral surfaces of the ear . After 10 min , the solution was gently removed using 0 . 9% normal saline and cotton swabs . The ear was mounted against the adjustable plexiglass microscope pedestal and held in place under a coverslip . Mouse rectal temperature was monitored via rectal thermometer and maintained at 37°C using a self-regulating heating mat . The microcirculation of the ventral abdominal skin was prepared for microscopy as previously described [66] . Mice were anaesthetized and body temperature was monitored as described above . Briefly , after shaving a midline abdominal incision was made extending from the pelvic region up to the level of the clavicle . The skin was separated from the underlying tissue , remaining attached laterally to ensure the blood supply remained intact . The area of skin was then extended over a viewing pedestal and secured along the edges using 4 . 0 sutures . The loose connective tissue lying on top of the dermal microvasculature was carefully removed by dissection under an operating microscope . The exposed dermal microvasculature was immersed in isotonic saline and covered with a coverslip held in place with vacuum grease . The right jugular vein was cannulated to administer additional anaesthetic and fluorescent dyes . To visualize B . burgdorferi-endothelial interactions , 4×108 spirochetes in 200 µl of PBS were injected directly into the jugular or femoral veins of anaesthetized mice . Three to six dermal venules ( 15–45 µm in diameter ) were selected in each experiment . Conventional epifluorescence microscopy was performed with a Leica DM IRE2 inverted microscope ( Leica Microsystems , Frankfurt , Germany ) equipped with an Orca ER cooled CCD camera ( Hamamatsu , McHenry IL ) , using a 63× oil immersion objective , a narrow band GFP filter ( 480 −/+ 10 nm excitation wavelength; 510 −/+ nm emission wavelength: Chroma Technology Corp , Rockingham , VT ) and exposure times of 100 ms . Sixteen-bit images were acquired using OpenLab 5 . 0 . 2 ( Improvision Inc . , Lexington , MA ) , and exported images in . tiff format were converted to 8-bit , colorized using indexed color and cropped in Adobe Photoshop CS prior to export and conversion to CYMK mode in Adobe Illustrator CS ( Adobe Systems Inc . , San Jose , CA ) . Identical image capture and adjustment settings were used for all images . Conventional epifluorescence intravital microscopy was performed using a Zeiss Axioskop microscope equipped with a 40× Wetzlar water immersion lens ( Carl Zeiss Canada Ltd . , Toronto , ON ) . Manual focusing was used to ensure that spirochetes remained in the focal plane throughout recording . A video camera ( HS model 5100; Panasonic , Osaka , Japan ) was used to project the images onto a monitor , and the images were recorded at 29 . 97 fps for off-line video playback analysis using a videocassette recorder . VHS analogue videos of conventional IVM experiments were converted to digital format using Windows Movie Maker ( Microsoft Corporation , Redmond WA ) , and converted to . swf format using Macromedia Flash Professional 8 ( Macromedia Inc . , San Francisco , CA ) without altering frame rate or editing frame sequence . Leukocyte recruitment was monitored by rhodamine staining of leukocytes , as previously described [67] . Spinning disk confocal intravital microscopy [68] was performed using an Olympus BX51 ( Olympus , Center Valley , PA ) upright microscope equipped with a 20×/0 . 95 XLUM Plan Fl water immersion objective . The microscope was equipped with a confocal light path ( WaveFx , Quorum , Guelph , ON ) based on a modified Yokogawa CSU-10 head ( Yokogawa Electric Corporation , Tokyo , Japan ) . Endothelial cells and junctions were labeled with a monoclonal anti-PECAM-1 antibody ( Fitzgerald Industries International , Inc . , Concord , MA ) , conjugated to Alexa Fluor 555 ( Molecular Probes , Invitrogen Canada , Burlington , ON ) . One hundred µl of Alexa-conjugated anti-PECAM-1 were injected per mouse ( 50 µg/mouse ) . In some experiments , 50 µl of 5 mg/ml FITC-albumin in normal saline ( Sigma-Aldrich Canada Ltd . , Oakville , ON ) was injected to visualize blood vessels ( 250 µg/mouse ) . Laser excitation at 488 and 561 nm ( Cobalt , Stockholm , Sweden ) , was used in rapid succession and fluorescence in red and green channels was visualized with the appropriate long pass filters ( Semrock , Rochester , NY ) . Emission wavelengths for red and green channels were 593 nm and 520 nm , respectively , and no overlapping signal was detected in either channel . Exposure time for both wavelengths was 168 ms . A 512×512 pixels back-thinned EMCCD camera ( C9100-13 , Hamamatsu , Bridgewater , NJ ) was used for fluorescence detection . Volocity Acquisition software ( Improvision Inc . , Lexington , MA ) was used to drive the confocal microscope . Sensitivity settings were 255 and 251 for red and green , respectively , and autocontrast was used . Images were captured at 16 bits/channel in RGB . For timelapse series , manual focusing was used to ensure that spirochetes remained in the focal plane throughout recording . Red and green channels were overlaid using brightest point settings before export in . tiff or . mov format . Overlaid GFP and Alexa Fluor 555 . tiff images exported from Volocity were cropped in Adobe Photoshop CS without manipulation of signal levels or contrast prior to export and conversion to CYMK mode in Adobe Illustrator CS . Exported . mov files were imported without editing directly into Macromedia Flash Professional 8 for labeling and export as . swf files . Z-series were collected using spinning disk confocal IVM , with images captured in both red and green channels for each slice . All image acquisition settings were as described above , except where noted in the Specific Image Acquisition Settings section , below . The localization of B . burgdorferi relative to the lumen , endothelium , endothelial junctions and extravascular tissue was scored for each z-slice in the series , using xy , xz and yz images constructed from the z-section series using Volocity 4 . 0 . 2 . Scoring was performed independently by two individuals . Three-dimensional volume rendering ( voltex ) reconstruction of spirochetes in venules was performed in Amira 4 . 1 . 1 ( Mercury Computer Systems , Chelmsford , MA ) using series of GFP and Alexa Fluor 555 . tiff images exported separately from Volocity . Alpha and gamma settings were 1 , and GFP and Alexa Fluor sensitivities were , respectively , 30–170 and 30–225 . Animated rotation views of 3D volume rendering were exported as . mpeg files prior to import and labeling in Macromedia Flash Professional 8 . All images were acquired and processed as described in the preceding sections . All timelapse series were captured at 0 . 94 fps , and exported at 5 fps , except the timelapse presented in Video S1 , which was captured at 5 . 9 fps and exported at 50 fps . Other parameters: Fig . 1C: 0 . 485 µm/pixel ( x and y ) ; Fig . 2: 0 . 485 µm/pixel ( x and y ) ; Fig . 3B: 37 z-slices ( 45 . 3 sec/series ) , 5 µm step size , 0 . 485 µm/pixel ( x and y ) , 1 µm/pixel ( z ) ; Fig . 4A: 19 z-slices ( 23 sec/series ) , 1 µm step size , 0 . 485 µm/pixel ( x and y ) , 1 µm/pixel ( z ) ; Fig . 5A: 81 z-slices/time point ( 71 sec/stack ) , 0 . 5 µm step size , 0 . 485 µm/pixel ( x and y ) , 0 . 5 µm/pixel ( z ) ; Video S1: settings same as in Fig . 1C; Video S2: settings same as in Fig . 2; Video S3: 0 . 485 µm/pixel ( x and y ) ; Video S5: settings same as in Fig . 3b; Video S6: 0 . 485 µm/pixel ( x and y ) . For quantitative analysis , average and standard error values for different variables were calculated and plotted graphically for all vessels from all mice using GraphPad Prism 4 . 03 ( GraphPad Software , Inc . , San Diego , CA ) . Statistical significance was calculated in GraphPad Prism using a two-tailed non-parametric t-test with a 95% confidence interval . | Pathogenic spirochetes are bacteria that cause a number of emerging and re-emerging diseases worldwide , including syphilis , leptospirosis , relapsing fever , and Lyme disease . They exhibit an unusual form of motility and can infect many different tissues; however , the mechanism by which they disseminate from the blood to target sites is unknown . Direct visualization of bacterial pathogens at the single cell level in living hosts is an important goal of microbiology , since this approach is likely to yield critical insight into disease processes . We engineered a fluorescent strain of Borrelia burgdorferi , a Lyme disease pathogen , and used conventional and spinning disk confocal intravital microscopy to directly visualize these bacteria in real time and 3D in living mice . We found that spirochete interaction with and dissemination out of the vasculature was a multi-stage process of unexpected complexity and that spirochete movement appeared to play an integral role in dissemination . To our knowledge , this is the first report of high resolution 3D visualization of dissemination of a bacterial pathogen in a living mammalian host , and provides the first direct insight into spirochete dissemination in vivo . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"infectious",
"diseases/bacterial",
"infections",
"microbiology"
] | 2008 | Real-Time High Resolution 3D Imaging of the Lyme Disease Spirochete Adhering to and Escaping from the Vasculature of a Living Host |
The molecular mechanisms that underlie asymmetric PTEN distribution at the posterior of polarized motile cells and regulate anterior pseudopod formation were addressed by novel single-molecule tracking analysis . Heterogeneity in the lateral mobility of PTEN on a membrane indicated the existence of three membrane-binding states with different diffusion coefficients and membrane-binding lifetimes . The stochastic state transition kinetics of PTEN among these three states were suggested to be regulated spatially along the cell polarity such that only the stable binding state is selectively suppressed at the anterior membrane to cause local PTEN depletion . By incorporating experimentally observed kinetic parameters into a simple mathematical model , the asymmetric PTEN distribution can be explained quantitatively to illustrate the regulatory mechanisms for cellular asymmetry based on an essential causal link between individual stochastic reactions and stable localizations of the ensemble .
Intracellular signal transduction at the cell membrane mediates various extracellular signals inside the cell for proper environmental adaptation . Signaling molecules achieve their function in large part by translocating between the membrane and cytoplasm to mediate a variety of signaling systems including Raf , PKB and PLC in growth factor signaling , Zap-70 in T-cell receptor signaling , Rap1 in integrin signaling and heterotrimeric G protein , Crac and PTEN in Dictyostelium chemotactic signaling [1] , [2] , [3] , [4] , [5] , [6] . The intracellular distribution of signaling molecules is regulated dynamically via their transient and repetitive associations with the membrane , a phenomenon that can be viewed as dynamic shuttling between the membrane and cytoplasm . Upon environmental changes , the shuttling is modulated spatially and temporally in response to the multiple membrane-binding states that arise with changes in a signaling molecule's interactions with other molecules such as membrane receptors , effectors , and lipids . Spatial and temporal heterogeneities in the molecular states , the shuttling itself and finally the number of molecules interacting with the membrane arise inevitably during the signal transduction processes , which then act as a basis for cellular responses . Thus , signal transduction can be regarded as a molecular process that regulates the state transition . Nevertheless , few studies have described membrane localization based on the molecular reactions of membrane associations and dissociations or state transitions . One reason is the difficulty of directly observing and measuring signaling reactions on the membranes of living cells , although this has been growing increasingly more feasible with gains in single-molecule imaging techniques . For example , using total internal reflection fluorescence microscopy ( TIRFM ) , we have been able to trace the behavior of a single molecule while it is bound to a cell membrane [7] , [8] , [9] , [10] , allowing us to follow a series of signaling reactions while the signaling molecules associate with , move laterally along and dissociate from the membrane . The membrane-binding state can be characterized by the lateral mobility of diffusion and/or membrane-binding lifetimes , which are quantified by statistically describing the single-molecule trajectories . For example , diffusion analysis considers the spatial distance between two positions of a molecule over a unit time interval of the trajectory [11] , [12] , [13] , [14] . In contrast , lifetime analysis considers the time duration of individual trajectories from the onset to completion of a membrane association [15] , [16] . Since these two analyses focus exclusively on the spatial or temporal aspects of single-molecular behavior , respectively , heterogeneity in one cannot be easily correlated with the other , which complicates our understanding of the details of the state transition kinetics and the relevant signal transduction mechanisms . An ideal analysis method , therefore , will unify the spatial and temporal information of single-molecule trajectories . In the present study , we propose a novel and general statistical method for the single-molecule tracking analysis of signaling molecules on the membrane of living cells , which we name lifetime-diffusion analysis . The method estimates state transition kinetics and membrane-binding lifetimes from single-molecule trajectories by correlating each membrane-binding state to the characteristic lateral mobility . The method is here shown valid for the PtdIns ( 3 , 4 , 5 ) P3 phosphatase PTEN ( a phosphatase and tensin homologue deleted on chromosome 10 ) . PTEN , which was first identified as a tumor suppressor in mammalian cells , is involved in chemotactic signaling in Dictyostelium discoideum [6] , [17] . It has been observed that PTEN is excluded from the anterior membrane of polarized Dictyostelium cells that undergo chemotaxis in response to a chemical gradient and that a PtdIns ( 3 , 4 , 5 ) P3-enriched domain arises at the cell end with the higher gradient concentration [18] , [19] . The domain is generated in an ultrasensitive and self-organizing manner and serves as a signal to activate pseudopod formation concerting with other signals in parallel signaling pathways , with the posterior localization of PTEN being critical for the anterior confinement of the patch and efficient directed migration [6] , [19] , [20] , [21] , [22] , [23] . Various mathematical models have been proposed to understand the underlying mechanism for the domain formation assuming essential molecular reactions and movements involving catalytic processes by PI3K and PTEN , lateral diffusion of PtdIns lipids and interactions between the enzymes and lipid molecules [24] , [25] , [26] . To discriminate the correct model , we require information on local differences in the reactions and movements , as these explain the global domain generation , and also need to observe and analyze the behaviors of PtdIns lipids , PI3K and PTEN on the membrane of living cells according to the structures of the molecular ensemble . By using our lifetime–diffusion analysis , we here investigated the molecular mechanism driving asymmetric PTEN distribution in migrating Dictyostelium cells . A kinetic model was obtained and suggests that PTEN molecules exhibit stochastic transitions among three states with different diffusion coefficients and membrane-binding lifetimes . The asymmetric distributions were explained quantitatively by the spatially regulated heterogeneity of the state transition kinetics , illustrating the regulatory mechanisms of PTEN distribution with single-molecular resolution . Our lifetime-diffusion analysis described here can be applied in general to other membrane-bound signaling molecules on the membrane .
In order to understand the molecular mechanisms of PTEN asymmetric localization , we first examined the intracellular localization of the mutant PTENG129E , which has a G129E substitution that leads to no substrate binding or phosphatase activity against PtdIns ( 3 , 4 , 5 ) P3 [27] , [28] . PTENG129E was found localized on the membrane except in the pseudopod , which is similar to wild-type PTEN properties ( Figure 1A ) . Asymmetric PTENG129E distribution on the membrane was induced by applying concentration gradients of the chemoattractant cAMP to cells lacking functional actin cytoskeletons , which resulted in stimulation-induced local depletion of PTENG129E ( Figure 1B ) . Therefore , neither PtdIns ( 3 , 4 , 5 ) P3 binding nor enzymatic activity is a prerequisite for the asymmetric distribution of PTEN on the membrane of Dictyostelium cells that respond to cAMP gradients . Thus , because these properties simplify the analysis of kinetics before membrane dissociation , we used PTENG129E for the subsequent analysis . Single molecules of PTENG129E-Halo conjugated with tetramethylrhodamine ( TMR ) were visualized under TIRFM ( Figure 1C and Movie S1 ) . Visualized fluorescent spots showed quantized photobleaching , fluorescence intensities and spot sizes typical of single fluorophores ( Figures S1A , see Text S1 for detail ) . The region of the membrane corresponding to the pseudopod could be clearly distinguished from the rest of the membrane , as the number of PTENG129E molecules there was minimal ( Figure 1C ) . The trajectories of single PTENG129E molecules showed different behaviors in time and space between the pseudopod and elsewhere ( Figure 1D ) . Membrane dissociation occurred faster on the membrane at the pseudopod than at the tail ( Figure 1E ) . The dissociation curves had decay rates faster than that of photobleaching , indicating that PTENG129E molecules shuttle between the membrane and cytoplasm . The photobleaching rate constant was 0 . 1 s−1 when immobilized PTENG129E molecules on a membrane of fixed cells were visualized . The lateral diffusion was found to be slightly faster in the pseudopod than in the tail ( Figure 1F ) . Furthermore , mean square displacement ( MSD ) increased linearly with time , suggesting that individual molecules showed normal diffusion , not confined diffusion due to a compartmentalization ( see , for example , [29] ) or super diffusion due to a directional motility ( see , for example , [12] ) ( Figure S1B ) . Therefore , on average , PTENG129E molecules exhibited faster membrane dissociation and faster lateral diffusion at the pseudopod than at the rest of the polarized cell . In addition , the temporal correlation of the diffusion mobility was examined by using a time series of the displacements made from each trajectory ( Figure 1G ) . The autocorrelation function exhibits an exponential decay with a rate constant of 1 . 64 s−1 , indicating an alternation in the diffusion mobility ( Figure 1H ) [13] . To identify the possible multiple states that a membrane-bound molecule adopts and to characterize the corresponding state transition kinetics , we developed a novel single-molecule tracking analysis method , lifetime-diffusion analysis . We previously reported a method which can be used for analyzing the multistate kinetics of membrane-integrated molecules based on the simplest of three theoretical models [13] , which are only applicable to molecules like receptors , channels and adhesion proteins that are incorporated into the membrane lipid bilayer . Here , we extend these models to cases when molecules are shuttling between the membrane and cytoplasm ( Models S1–S3 in Figure 2A ) . The models are appealing , because they describe the essential behaviors of signaling molecules that show simple diffusion along a membrane , and therefore provide a theoretical basis for the analysis method . Should a molecule show confined or super diffusion , our method can still be applied but requires amendments to the diffusion equations to account . In order to clarify the multistate kinetics of a molecule , lifetime-diffusion analysis ( Figure 3A ) , which consists of the following series of steps , is done . First , we determine whether the molecule exhibits membrane dissociation or not . Second , we count the number of membrane-binding states with different diffusion coefficients . Third , we observe whether the molecule exhibits state transitions . Finally , we construct a model consistent with the experimental results and estimate the kinetic parameters by fitting the data to theoretical functions derived from the model . In the first step , membrane dissociation is examined by comparing the disappearance rate of single-molecule fluorescence with the photobleaching rate of the fluorophore itself ( Figure 1E ) . When the fluorophore is conjugated to membrane-integrated molecules or immobilized onto a glass surface , the fluorescence will disappear with the photobleaching rate . When it is conjugated to molecules shuttling between the membrane and cytoplasm , however , the disappearance rate becomes faster than the photobleaching rate due to the additional membrane dissociation rate . The number of fluorophores undergoing photobleaching will decay with time as follows , ( 10 ) where λb is the photobleaching rate constant and can be assumed constant throughout the analysis when excitation conditions are unchanged ( Text S1 ) . In the second step , the number of states with different diffusion coefficients is determined by statistical analysis of the displacements [13] . In the trajectories , displacement , Δr ( t ) = ( ( x′ ( t+Δt ) −x′ ( t ) ) 2+ ( y′ ( t+Δt ) −y′ ( t ) ) 2 ) 1/2 , is calculated at an arbitrary t with a unit time interval Δt . Under ideal conditions where the molecules adopt one of multiple membrane-binding states and do not change their state during Δt , the displacement distribution can be regarded as a mixture of distributions with different diffusion coefficients and described by the PDF , ( 11 ) where . i indicates the state number , and Dj and pj represent the diffusion coefficient of the j-th state and its proportion relative to all states , respectively . ε is the SD of the position error of the fluorescence spots . The minimum state number is defined as that when the theoretical PDF can fit well the experimental distribution and is determined by using Akaike Information Criterion ( AIC ) values calculated after the maximum likelihood estimation ( MLE ) for each state number [13] , [30] ( see Text S1 for details ) . Figure 3B shows a displacement distribution obtained from 1000 trajectories generated by numerical simulations using Model S3 . Two or three states were sufficient to fit the distribution to Eq . 11 well . Based on the AIC2 value , we concluded that the molecule adopts two states . The estimated diffusion coefficients were D1 = 0 . 011 and D2 = 0 . 100 µm2s−1 , which are almost the same as those used for the numerical simulation of the trajectories . For an accurate estimate of the diffusion coefficient , Eq . 11 should incorporate an ε value that is quantified prior to the MLE , which can be done by fitting MSD ( Δt ) = 4D*Δt+4ε2 to the calculated MSD ( Figure S1B , inset , Table 1 ) . Assuming a mean diffusion coefficient D* = 0 , the MSD ( Δt ) of immobilized molecules corresponds to 4ε2 ( see [13] for methods ) . In the third step , state transitions are determined by examining alternations in the diffusion mobility of single-molecule trajectories . In order to detect stochastic alternations , the displacement is statistically analyzed according to time after membrane association . We studied trajectories in the time interval between t and t+τ with τ> = Δt , and calculated the displacement , Δr , within the interval . The displacement distribution for a given interval is a mixture of distributions with diffusion coefficients that are determined by MLE as described above . We then estimated the ratio , pj , in Eq . 11 for every interval by using the diffusion coefficients . The ratio at the time interval between t and t+τ is represented by pj ( t ) , which demonstrates the ratio is time dependent ( Figure 3C , inset ) . The decay profiles of the subpopulations are obtained by multiplying the dissociation curve with pj ( t ) and are expected to follow the subpopulation probabilities Qj ( t ) obtained theoretically ( Figure 3C ) . In the absence of a state transition , each subpopulation decreases monotonically according to Eq . 6 ( Figure 2E ) . On the other hand , in the presence of a state transition , the probability of each subpopulation shows biphasic decay and finally decreases at the same rate after reaching a steady state with respect to the state transition described by Eq . 9 ( Figure 2G ) and shown in the analysis of the simulated trajectories ( Figure 3C ) . Finally , kinetic parameters such as dissociation rate constants , transition rate constants and initial probabilities are quantified by fitting the decay profiles to Qj ( t ) . The parameter values were estimated from the trajectories as λ1 = 0 . 11 , λ2 = 0 . 94 , k12 = 0 . 068 , k21 = 0 . 47 and q1 = 0 . 20 ( Figure 3C ) , which are in good agreement with the values used in the simulation: λ1 = 0 . 10 , λ2 = 1 . 00 , k12 = 0 . 10 , k21 = 0 . 50 and q1 = 0 . 20 . The PDF of the molecular positions was obtained and compared with the distribution obtained from the molecular trajectories ( Figure 3D ) . When the model PDF fits the experimental distribution well , the estimation is deemed to be successful . According to the same procedure as above , other models could also make successful estimates of the diffusion coefficients and kinetic parameters , including simple models like Models S1 and S2 ( Figure S2 ) as well as more complicated ones like that for PTENG129E ( described below ) . The estimation accuracy is dependent on the number of trajectories to be analyzed and the magnitude of the measurement error ( see Text S1 and Figures S3 and S4 ) . Using lifetime-diffusion analysis , we analyzed the trajectories of single PTENG129E molecules . To elucidate the general properties of membrane interactions that occur independently of cell polarity , molecular behaviors were analyzed in non-polarized cells . The membrane binding duration of each PTENG129E molecule was measured and the dissociation curve was obtained ( Figure 4A , open circles ) . The decay was faster than the fluorescent photobleaching , indicating transient association of PTENG129E to the membrane and is consistent with a previous report that examined mammalian and Dictyostelium cells [31] . To estimate the state number , the molecular displacement within a 33 ms window was measured to obtain the displacement distribution ( Figure 4B ) . In AIC analysis , the minimum value was obtained by assuming that PTENG129E adopts three states with different diffusion coefficients ( see Text S1 for details ) . This enabled us to fit well the displacement distribution using Eq . 11 ( Figure 4B ) . Thus , PTENG129E on the membrane of non-polarized cells likely adopts three states with different diffusion coefficients , D1 = 0 . 04 , D2 = 0 . 13 and D3 = 0 . 69 µm2s−1 . Whether a state transition occurs was determined by quantifying the decay profiles . In Figure 4C , the ratio of each subpopulation , which was obtained using the same manner as that for the inset in Figure 3C , was plotted as a function of time after the onset of membrane association . At the initial moment of membrane association , state 2 dominated the three states . All three ratios changed initially but eventually reached steady state after 500 ms when the ratios became constant . By multiplying the dissociation curve by the ratios , the decay profiles were obtained , showing the same decay rate after 500 ms of membrane association ( Figure 4A , crosses ) . Thus , all three states are likely to be involved in state transitions . The minimum model assumes three states with different diffusion coefficients and at least two state transitions from a given state ( Figure 4E ) . The diffusion equations for the model are , ( 12 ) The initial probabilities that the molecule adopts states 1 , 2 or 3 are represented by q1 , q2 and q3 , respectively , with q1+q2+q3 = 1 . These equations were solved after taking the Fourier transform , and the inverse transformation was performed to obtain the PDF by numerical integration ( Figure 4D ) . The derivations of R ( t ) and Qj ( t ) are described in Text S1 . In Figure 4A , the three decay profiles were well fitted by the theoretical functions with a single set of parameter values , indicating that this model can explain the membrane dissociation and state transition kinetics . The rate constants estimated by the fitting are summarized in Figure 4E and Table 1 . By using these values and diffusion coefficients in the theoretical PDF , we could explain the distributions of position ( Figure 4D ) . The estimated kinetic model can explain the shuttling of PTENG129E as follows ( Figure 4E ) . The major state that the molecule adopts initially on the membrane is state 2 , indicating that state 2 is the primary binding state responsible for recruitment of PTENG129E to the membrane . When the molecule adopts state 1 by transition from state 2 , the rate constant of membrane dissociation decreases by a factor of 100 , indicating that state 1 is a stably binding state that extends the membrane association duration . When the molecule adopts state 3 instead , the membrane dissociation rate increases by a factor of 3 compared with state 2 . A state 3 to state 2 transition is sufficiently rare such that almost all molecules are returned to the cytoplasm upon adopting state 3 , indicating this state is a weakly binding state that accelerates membrane dissociation . Therefore , most PTENG129E molecules were suggested to be recruited to the membrane primarily via state 2 , fluctuate between states 1 and 2 , but return to the cytoplasm at the highest probability when they take state 3 . The membrane dissociation kinetics are sensitive to changes in λ2 , k12 , k21 and q1/q3 such that the membrane binding lifetime can be modulated through these parameters ( Text S1 , Figure S5 ) . It has been suggested that PTEN membrane localization requires the PtdIns ( 4 , 5 ) P2-binding motif located at its N-terminus [28] , [31] , [32] . We therefore considered whether the membrane binding states are dependent on this motif . PTENG129E;Δ15-Halo , in which 15 amino acids constituting the motif were deleted , exhibited more localization to the cytoplasm ( Figure 5A ) . The fluorescence intensity of PTENG129E-Halo showed a 160% increase on the membrane compared to the cytoplasm . On the other hand , PTENG129E;Δ15-Halo on the membrane was 50% that in the cytoplasm ( Figure 5A ) . Consistent with this , the number of PTENG129E;Δ15 molecules observed on the membrane was largely decreased . The dissociation curve was not altered by deletion of the motif , indicating that the cytoplasmic localization of PTENG129E;Δ15-Halo was mainly due to a decrease in the membrane association rate rather than an increase in the membrane dissociation rate ( Figures 4A and 5B ) . Lifetime-diffusion analysis suggested the most likely state number for PTENG129E;Δ15-Halo is 3 , and diffusion coefficients and kinetic parameters similar to those of PTENG129E were estimated ( Figures 5C and 5D , Table 1 ) . Furthermore , the initial probabilities also resembled those of PTENG129E . These results suggest that the PtdIns ( 4 , 5 ) P2-binding motif is a prerequisite for membrane association independent of the state . Based on the model for non-polarized cells , stochastic trajectories in polarized cells were analyzed to reveal the mechanism for how heterogeneity in molecular behavior is established . AIC analysis was performed using the displacement distributions obtained from the pseudopod and tail membrane , with each location showing three states ( Figure 1F ) . The diffusion coefficients were estimated to be 0 . 04 , 0 . 11 and 0 . 56 µm2s−1 at the pseudopod , and 0 . 03 , 0 . 15 and 0 . 72 µm2s−1 at the tail . These values were similar to those estimated in non-polarized cells , indicating that the diffusion coefficient of each state was not affected significantly by cell polarity . We did observe that the two regions showed differences in the relative amount of each subpopulation on the membrane: PTENG129E adopts state 1 , the stably binding state , less frequently at the pseudopod ( 34% ) than at the tail ( 64% ) , which leads to an increase in average diffusion mobility ( Figure 1F ) . The decay profiles showed that PTENG129E on both regions exhibits state transitions ( Figures 6A and 6B ) . All subpopulations decayed at the same rate after reaching the stationary state . At the onset of membrane association , the major state was state 2 irrespective of cell polarity . After 500 ms of membrane association , state 2 remained the major state in the pseudopod , whereas it was state 1 in the tail ( Figures 6A and 6B ) . These results indicate that the state transition kinetics of PTENG129E molecules is different between the pseudopod and tail of polarized cells . We estimated the rate constants and initial probabilities by fitting the decay profiles ( Figure 6C , Table 1 ) . The dissociation rate constants for both regions were almost equivalent to the respective rate constants of non-polarized cells , indicating that cell polarity does not significantly affect the membrane-binding ability of each state . However , a remarkable change was found in the ratios of the states that the molecule adopts on its initial moment of membrane association ( Figure 6C ) . Initially at the pseudopod , state 1 PTENG129E molecules were few ( 4% ) , while at the tail they exceeded those seen in non-polarized cells ( 30% vs . 24% ) . Consistent with this , the transition from state 2 to 1 was decelerated at the pseudopod ( k21 = 1 . 93 ) , but accelerated at the tail ( k21 = 4 . 19 ) relative to non-polarized cells ( k21 = 2 . 90 ) . On the other hand , the transition from state 1 to 2 occurred slightly faster at the pseudopod ( k12 = 5 . 43 ) than at the tail ( k12 = 4 . 66 ) . As a result , the equilibrium between states 1 and 2 is biased toward state 2 at the pseudopod and toward state 1 at the tail compared to non-polarized cells . These results indicate that the membrane binding of PTENG129E is suppressed at the pseudopod , but stabilized at the tail , a property that can explain , on average , the spatial heterogeneities in membrane-binding lifetime and lateral mobility . Because the transition rates between state 2 and 3 are constant and independent of cell polarity , state 1 , the stably binding state , should play a major role in regulating the membrane-binding properties of a polarized cell . In order to examine whether differences in the kinetic parameters sufficiently explain the depletion of PTEN from the anterior pseudopod , we performed numerical simulations of the intracellular distribution based on the above model using experimentally obtained kinetic parameters . For this purpose , the membrane association frequency , which represents the frequency that a single molecule in the cytoplasm associates with the membrane in 1 sec , was estimated ( Figure 7A ) . Here we assumed that the membrane localization is in steady state in non-polarized cells of a constant volume and reaches a constant membrane PTEN to cytoplasm PTEN ratio in the molecular number . The ratio was quantified biochemically from the ensemble fluorescence intensities of TMR-labeled PTENG129E-Halo in insoluble and soluble fractions of the cell lysate [33] . The ratio was 0 . 76 and gave rise to three membrane association frequencies: μ1 = 0 . 54 , μ2 = 1 . 38 and μ3 = 0 . 36 s−1 , since the association frequencies are proportional to the respective initial probabilities , q1 , q2 and q3 ( Figure 7A , Text S1 ) . In the simulation where the number of cytosolic molecules was 20000 and the radius of the cell was spherical and 5 µm , the total number of membrane-bound molecules was 15200 and the density on the membrane was 48 . 5 molecules/µm2 ( time<0 s in Figure 7B ) . Next , the association frequencies at the pseudopod and tail were estimated ( see Text S1 for details ) . By directly counting the number of molecules that began their membrane association within a unit time interval in a unit area at the pseudopod or tail of a single cell , we found the overall association frequency via all three states , μ = μ1+μ2+μ3 , at the pseudopod to be on average approximately 60% that of the tail ( n = 5 cells ) . When μ at the tail was assumed to equal the μ of non-polarized cells , the association frequencies were μ1 = 0 . 68 , μ2 = 1 . 32 and μ3 = 0 . 27 s−1 at the tail and μ1 = 0 . 06 , μ2 = 1 . 21 and μ3 = 0 . 16 s−1 at the pseudopod ( Figure 7A , see Text S1 for detail ) . At time 0 in Figure 7B , when the kinetic parameters obtained in the polarized cells were incorporated , the simulated PTEN density on the anterior and posterior membrane diverged to show remarkable depletion from the anterior membrane , which recapitulates the asymmetry that arose with a cAMP gradient ( Figure 1B ) . The calculated posterior-to-anterior ratio of the PTEN density reached 2 . 5 , which is comparable to the ratio measured experimentally by the fluorescent imaging of PTEN-Halo ( 2 . 1 ) . Thus , PTEN is likely to be excluded from the pseudopod membrane of chemotaxing cells , probably due to the suppression of molecules adopting state 1 , the stable binding state . Consistently , when membrane recruitment via state 1 was temporally inhibited in a step-wise manner in a non-polarized cell , the simulated PTEN density on the membrane decreased quickly within 1 . 5 s after the change ( Figure 7C ) , demonstrating the importance of state 1 in regulating the amount of membrane-bound PTEN molecules . The suppression of state 1 then is likely to be a key regulator for the asymmetric PTEN distribution . We also performed lifetime-diffusion analysis assuming PTENG129E molecules adopt two states , because the PDF of displacement with two diffusion coefficients was similar to that with three diffusion coefficients when fitting the experimental data ( Figure 4B ) . The kinetics of the state transitions and membrane dissociation were analyzed from the experimental trajectories . In non-polarized cells , the estimated Ds were 0 . 07 and 0 . 43 µm2/sec for states 1 and 2 , respectively . The probabilities of adopting these states decayed at the same rate after 300 to 500 msec of membrane association , suggesting the molecules exhibit state transitions ( Figure S6 ) . By fitting the decay profiles to Eq . 9 , the kinetic parameters were estimated as k12 = 2 . 59 , k21 = 8 . 15 , λ1 = 1 . 51 , λ2 = 10 . 00 and q1 = 0 . 65 ( Figure S6 ) . The same analyses were performed on the trajectories obtained for polarized cells ( Table 2 , Figure 8 ) . The two models did share some essential kinetic features . The state with slower lateral diffusion exhibited slower membrane dissociation , and transition from the faster state to slower state was inhibited at the pseudopod . As a result , PTEN molecules that once adopted the faster state readily dissociated from the membranes at the pseudopod , resulting in shorter membrane binding lifetimes and slower lateral diffusion there ( Figures 1E and 1F ) . However , while a low initial probability of state 1 at the pseudopod was suggested in the three-state model , the initial probability in the two-state model was relatively unaffected by cell polarity . It is possible that states 1 and 2 in the two-state model behave as averages of states 1 and 2 and states 2 and 3 , respectively , in the three-state model . Therefore , the two-state model may serve as an average model , while the three-state model , which seems better able to statistically explain the trajectories obtained experimentally ( Text S1 , Figure S7 ) , is more accurate .
Here we describe a statistical model that uses single-molecule imaging analysis to explain the molecular mechanism responsible for the asymmetric membrane localization of PTEN . By developing and using a novel method , lifetime-diffusion analysis , we propose a multistate kinetics model in which PTENG129E on the membrane has three membrane-binding states that transit between themselves . The model can explain the heterogeneity of PTEN molecular behavior as the result of suppressing the stable binding state at the pseudopod . A simulation based on the model demonstrated that locally applied modulation on the stable binding state gives rise to an asymmetric distribution of PTEN on the membrane , suggesting an essential causal relationship between molecular membrane-binding kinetics and the PTEN distribution on the membrane of polarized cells . We compare the results obtained by the lifetime-diffusion analysis with those from lifetime or diffusion analysis . The dissociation curve of PTENG129E in non-polarized cells was fitted to a sum of three exponential functions , Eq . S11 , with rate constants of 2 . 5 ( 95% in the relative amount ) , 11 . 1 ( 3% ) and 12 . 6 s−1 ( 2% ) . The distribution of the displacement obtained from the same trajectories was fitted to a sum of three PDFs , Eq . 11 , with diffusion coefficients of 0 . 04 ( 48% ) , 0 . 13 ( 48% ) and 0 . 69 ( 4% ) µm2s−1 . Thus , analysis of membrane-binding and diffusion mobility led to different proportions of the subpopulations of the three states . In other words , temporal and spatial analysis gives different results for the three states . In contrast , our lifetime-diffusion analysis method offers a quantitative estimate of the kinetic parameter for each state simultaneously with the motility parameter . Applying the method to the trajectories of PTENG129E successfully provides a multistate kinetics model by which the spatiotemporal properties of the single-molecule trajectories can be quantitatively explained . Three membrane-binding states were found to have different roles for regulating the shuttling of PTENG129E molecules between the membrane and cytoplasm . The majority of molecules adopt state 2 at the onset of membrane association . Molecules that adopt this state can proceed in one of three ways: dissociation from the membrane ( highest probability ) , transition into state 1 or transition into state 3 ( lowest probability ) . By transforming into state 1 , the membrane association lifetime increased by about 100-fold . Transition into state 3 had the reverse effect , speeding dissociation . Thus , the intracellular distribution of PTENG129E is regulated via the state transition that occurs on the membrane , and the asymmetric distribution of PTEN located there is dependent on local differences in the state transition kinetics . The lifetime of intact PTEN was also shorter at the pseudopod than at the tail , suggesting that the same mechanism should principally work in localizing intact PTEN molecules [34] . The diffusion coefficients of molecules on the membrane depend mainly on their interactions , with the coefficients generally becoming smaller as the interaction interface becomes larger . For example , proteins associating only with the inner leaflet usually have a larger diffusion coefficient than those integrated into the lipid bilayer [35] . In the case of Dictyostelium cells , CRAC , a pleckstrin homology ( PH ) domain-containing protein which can bind to PtdIns ( 3 , 4 , 5 ) P3 on the inner leaflet , has a diffusion coefficient of 0 . 14 µm2s−1 [9] , while the diffusion coefficient of Gβγ molecules that bind to the membrane via isoprenylation is 0 . 096 µm2s−1 [36] . On the other hand , the cAMP receptor cAR1 , a seven transmembrane receptor , has a diffusion coefficient of 0 . 022 µm2s−1 [8] , [34] . Referring to these known values , we propose that states 2 and 3 only occur on the inner leaflet . Also of note is that state 2 shows a diffusion coefficient ( 0 . 11 to 0 . 15 µm2s−1 ) which is similar to that of the PtdIns ( 3 , 4 , 5 ) P3-binding protein CRAC . Since the membrane association of PTEN requires a N-terminal PtdIns ( 4 , 5 ) P2-binding motif ( Figure 5 ) [28] , [31] , [32] , state 2 likely involves binding to PtdIns ( 4 , 5 ) P2 . On the other hand , state 1 presumably depends on some membrane-integral protein in addition to PtdIns ( 4 , 5 ) P2 , although such a binding partner for PTEN has not been identified in Dictyostelium to date . This putative binding partner seems to be a key regulator for PTEN dynamics and thus for asymmetry generation in chemotactic signaling . We have previously reported that PTEN and PtdIns ( 3 , 4 , 5 ) P3 exhibit mutually exclusive membrane localizations in the absence of a cAMP gradient and in the presence of Latrunculin A and caffeine , and proposed a reaction-diffusion model assuming PtdIns ( 3 , 4 , 5 ) P3 negatively regulates PTEN membrane localization [22] , [37] . To account for such a regulation , the high PtdIns ( 3 , 4 , 5 ) P3 concentration on the membrane should change the membrane interaction kinetics of PTEN , leading to one or both of a decrease in the membrane association rate and an increase in the membrane dissociation rate . We found that the association was less frequent and the dissociation was faster at the pseudopod than at the tail by single molecule imaging , and it was suggested by lifetime-diffusion analysis that both are due at least in part to the inhibition of stably binding state . It is plausible that the aforementioned putative binding partner is inhibited by PtdIns ( 3 , 4 , 5 ) P3 . Therefore , the three-state kinetics should be examined after modulating PtdIns ( 3 , 4 , 5 ) P3 levels on the membrane . In addition , the negative regulation requires phosphatase activity of PTEN in the spontaneous formation of the PtdIns ( 3 , 4 , 5 ) P3 domain , which could be further examined by observing intact PTEN molecules . In conclusion , lifetime-diffusion analysis provides a multistate kinetic description of intracellular signaling without a priori information of the interaction partner molecules , and can therefore be readily applied to any type of single-molecule trajectory .
Dictyostelium discoideum wild-type strain Ax2 was used as the parental strain . The plasmid was generated from the extrachromosomal expression of HaloTag-fusion PTEN proteins ( PTEN-Halo ) ( Promega , Japan ) and introduced into Ax2 cells by electroporation [38] . A point mutation substituting E for G at amino acid 129 ( G129E ) was introduced according to the protocol of the QuikChange site-directed mutagenesis kit ( Agilent Technologies ) . The observation of single PTEN molecules was performed using cells prepared as follows . Cultured cells were suspended at 3×106 cells/ml in development buffer ( DB; 5 mM Na2HPO4 , 5 mM NaH2PO4 , 2 mM MgSO4 , 0 . 2 mM CaCl2 , pH 6 . 2 ) . 1 mL of the cell suspension was transferred to a 35-mm culture dish and kept still for about 7 hours at 21°C . When cells began to aggregate , HaloTag tetramethylrhodamine ( TMR ) ligand ( Promega , Japan ) was added to the cell suspension at a final concentration of 50 nM . After 10 minutes of incubation , DB was replaced with new DB twice to rinse out the residual ligands . The cells were washed twice with DB by centrifugation and suspended in DB at around 5×106 cells/ml . 10 µL of the cell suspension was placed on a glass coverslip and the cells were left to settle in a moist chamber . After 10-minute incubation , the cells were overlaid with a 1 cm square sheet of agarose ( 2% Agarose M in DB ) and excess DB was removed [39] . After 20 to 30 min and recommencement of cell aggregation , cells were observed under TIRFM . Single PTEN-Halo molecules labeled with TMR were observed under an objective type TIRFM constructed on an inverted fluorescence microscope ( IX70 , Olympus , Japan ) . The detailed configuration of the microscope system is described elsewhere [16] . The trajectories of single molecules exhibiting lateral diffusion on the membrane were obtained semi-automatically from the onset of movement until their completion using laboratory-made software . Briefly , the position of a single molecule was determined in each frame of a movie by fitting the fluorescence intensity profile to a two-dimensional Gaussian distribution . The positions of neighboring frames were regarded as part of a single trajectory when the distance was within a value typical for diffusing molecules [35] . Trajectories of single molecules were made by numerical simulations previously described [13] . Assuming a Brownian particle , a time series of the position , ( x ( t ) , y ( t ) ) , beginning at the origin was generated from t = 0 to 60 s . The simulated time series consisted of 18001 time steps with an interval of 1/300 s . A trajectory of 1801 time steps was obtained by extracting the data from the time series at a time interval of 1/30 s . In the trajectory , a Gaussian SD of 40 nm was added to the position at every time point . Photobleaching of the fluorophore was not taken into account . 3000 trajectories were used for each analysis unless noted otherwise . | Signaling molecules can adopt multiple states that are characterized by intermolecular interactions and chemical modifications , and can thus describe a mechanistic basis for signal transduction . To understand the molecular mechanisms driving signaling reactions , the kinetics and spatial regulation of the state transitions are indispensable . However , using data from single-molecule imaging observations , state transition behaviors of only signaling molecules integrated into the plasma membrane have been analyzed . Here , we propose a novel analysis method which estimates the kinetics of the state transitions and membrane dissociation of signaling molecules that shuttle between the cytoplasm and membrane , a method based on the statistics of the diffusive mobilities of individual molecules . Applying the method to single-molecule tracking analysis of PTEN , a key player in the cellular polarity of Dictyostelium discoideum and mammalian neutrophils , we demonstrate the validity of our method for investigating the mechanism of asymmetric membrane localization . The method enables a quantitative prediction of signaling activities at local membrane regions and therefore provides a powerful tool for understanding the essential mechanisms of a signaling system . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"microbiology",
"phosphoinositide",
"signal",
"transduction",
"model",
"organisms",
"cell",
"movement",
"signaling",
"signaling",
"pathways",
"biophysics",
"simulations",
"dictyostelium",
"discoideum",
"signaling",
"in",
"cellular",
"processes",
"biophysics",
"theory",
"pro... | 2013 | Asymmetric PTEN Distribution Regulated by Spatial Heterogeneity in Membrane-Binding State Transitions |
Both type I interferon ( IFN-I ) and CD40 play a significant role in various infectious diseases , including malaria and autoimmune disorders . CD40 is mostly known to function in adaptive immunity , but previous observations of elevated CD40 levels early after malaria infection of mice led us to investigate its roles in innate IFN-I responses and disease control . Using a Plasmodium yoelii nigeriensis N67 and C57BL/6 mouse model , we showed that infected CD40-/- mice had reduced STING and serum IFN-β levels day-2 post infection , higher day-4 parasitemia , and earlier deaths . CD40 could greatly enhance STING-stimulated luciferase signals driven by the IFN-β promoter in vitro , which was mediated by increased STING protein levels . The ability of CD40 to influence STING expression was confirmed in CD40-/- mice after malaria infection . Substitutions at CD40 TRAF binding domains significantly decreased the IFN-β signals and STING protein level , which was likely mediated by changes in STING ubiquitination and degradation . Increased levels of CD40 , STING , and ISRE driven luciferase signal in RAW Lucia were observed after phagocytosis of N67-infected red blood cells ( iRBCs ) , stimulation with parasite DNA/RNA , or with selected TLR ligands [LPS , poly ( I:C ) , and Pam3CSK4] . The results suggest stimulation of CD40 expression by parasite materials through TLR signaling pathways , which was further confirmed in bone marrow derived dendritic cells/macrophages ( BMDCs/BMDMs ) and splenic DCs from CD40-/- , TLR3-/- TLR4-/- , TRIF-/- , and MyD88-/- mice after iRBC stimulation or parasite infection . Our data connect several signaling pathways consisting of phagocytosis of iRBCs , recognition of parasite DNA/RNA ( possibly GPI ) by TLRs , elevated levels of CD40 and STING proteins , increased IFN-I production , and longer host survival time . This study reveals previously unrecognized CD40 function in innate IFN-I responses and protective pathways in infections with malaria strains that induce a strong IFN-I response , which may provide important information for better understanding and management of malaria .
Malaria infection triggers strong host immune responses , which may lead to parasite elimination and/or severe pathology [1–5] . Among the immune mechanisms , the role of type I interferon ( IFN-I ) and interferon-stimulated genes ( ISG ) in protection against malaria infection has been controversial , but is gaining more attention lately [5 , 6] . Recent studies have shown that IFN-I responses involving pathways mediated by cytosolic DNA and RNA sensors/adaptors such as STING ( Stimulator of interferon genes ) , MDA5 ( Melanoma differentiation-associated protein 5 ) , and MAVS ( Mitochondrial antiviral-signaling protein ) can suppress early parasitemia and control liver stage development [7–10] , although the mechanism of the IFN-I response in malaria control and pathology is far from clear . STING recognizes AT-rich DNA motifs from malaria parasites [7] , whereas MDA5/MAVS can detect parasite RNAs from both sporozoite and blood stages [8–10] . However , whether cyclic-GMP-AMP ( cGAMP ) synthase ( cGAS ) plays a role in sensing malaria parasite DNA remains to be determined . STING is an endoplasmic reticulum ( ER ) adaptor that can activate TBK1 ( TANK-binding kinase 1 ) /IRF3 ( IFN regulatory factor 3 ) pathway to induce expression of IFN-I and ISG genes [11 , 12] by directly recognizing dsDNA or cyclic dinucleotides ( CDNs ) generated by pathogens or metabolized from dsDNA by cGAS [13–15] . STING can also signal through TRAF3 ( TNF receptor associated factor 3 ) or TRAF6 and activate canonical and non-canonical NFκB ( Nuclear factor-kappa B ) pathways [16] . Another important molecule in host immune response to infections is CD40 ( or TNF receptor superfamily member 5 , TNFRSF5 ) that is a receptor expressed on the surfaces of many cell types , including B-cells , monocytes , dendritic cells , endothelial cells , and epithelial cells [17] . CD40 is a type I transmembrane protein with four extracellular cysteine-rich domains that are important for interaction with its ligand CD40L ( CD154 ) on T-cells . The intracellular region of CD40 includes binding domains to TRAF molecules ( TRAF1 , 2 , 3 , 5 and 6 ) [17] . CD40-mediated signal transduction activates multiple pathways , including those signaling through NFκB , STAT3 ( Signal transducers and activators of transcription-3 ) , MAPK ( Mitogen-activated protein kinase ) , and other kinases [17–19] . CD40 is mostly known to function in the initiation and progression of cellular and humoral adaptive immunity , including T cell-dependent immunoglobulin class switching , memory B cell development , and germinal center formation [17 , 18 , 20] . Engagement of CD40 on macrophages and CD40L on T-cells has been shown to promote inflammatory responses in many neurologic diseases [21] . CD40-CD40L interaction between the CD4+ T cell and dendritic cells ( DCs ) primes the DCs to activate CD8+ cytotoxic cells [22–24] . The process of CD8+ T cell activation is associated with induction of CD70 expression in activated DCs after combined TLR/CD40 stimulations [25] . More recently , ligation of CD40 by CD40L was shown to stimulate IFN-β expression through TRAF2/3 mediated NFκB pathways and sequential de novo synthesis of IRF1 [26 , 27] . However , the increase in IFN-β expression by the CD40-mediated NFκB pathway was generally low , compared with those induced by other molecules such as STING or TLR7 . CD40 has been associated with many diseases or disorders [21 , 28 , 29] . In addition to protection against viral and bacterial infections [30–32] , CD40 also contributes to protection against parasitic infections such as Trypanosoma cruzi , Leishmania amazonensis , and Plasmodium yoelii [33–35] . Production of nitric oxide ( NO ) , IL-12 , and IFN-γ after ligation of CD40 is critical for controlling T . cruzi parasitemia [33] . CD40 is required for the maturation of liver dendritic cells , accumulation of CD8+ T cells in the liver , and effective APC licensing during P . yoelii sporozoite infection [34] . Activation and ligation of CD40 and CD40L are also associated with many neurologic and autoimmune diseases that are characterized by elevated levels of IFN-I [28 , 29 , 36 , 37] . Considering the potential role of CD40 in IFN-I response and our observation of up-regulation of IFN-I and CD40 expression in P . yoelii nigeriensis N67 ( N67 ) infection [8] , we investigated the functional roles and the relationship of CD40 and STING in host response to N67 parasite infection and showed that the serum level of IFN-β was significantly reduced in CD40 knockout ( KO ) mice day 2 after N67 infection , leading to early host death . We further showed that CD40 could greatly enhance STING protein level and STING-mediated IFN-I responses . The effect of CD40 on STING and the IFN-I response was mediated through TRAF2/3 and/or TRAF6 binding domains , leading to changes in STING ubiquitination and protein level . We also showed that various TLR ligands , infected red blood cells ( iRBCs ) , and parasite DNA/RNA could stimulate CD40 expression , establishing a signaling axis of TLR recognition and signaling , increased CD40 and STING levels , elevated IFN-I production , and longer host survival time .
C57BL/6 mice infected with N67 parasite induced a strong IFN-I response , including increased expression of genes such as Cd40 , Isg15 , Isg20 , Ifit2 , Mx1 , Mx2 , Ddx58 , and Usp18 genes ( S1A and S1B Fig ) , leading to suppression of N67 parasitemia [8 , 38] . Ligation of CD40 with CD40L was recently shown to activate the NFκB pathway resulting in enhanced IFN-I response in carcinoma cells [27] . The elevated levels of Cd40 transcripts in the N67 infected mice suggest that it may play a role in regulating host IFN-I response to malaria infection . To investigate the role of CD40 in IFN-I mediated protection against malaria infection , we infected wild type ( WT ) and CD40-/- mice with N67 and monitored parasitemia and host mortality . Compared with WT mice , both infected male ( n = 9 ) and female ( n = 3 ) CD40-/- mice had significantly higher day 4 parasitemia ( P = 0 . 0078 for male; P = 0 . 0137 for female ) ( Fig 1A and 1B ) and died earlier ( Ave survival days , WT over 25 days vs KO 15 days for male , P<0 . 001; 15 days vs 10 days for female , P = 0 . 0623 ) than the WT mice ( Fig 1C ) . As expected , the day 2 ( 24 h after parasite injection ) serum levels of IFN-β were significantly lower in the CD40-/- mice than those of WT mice after parasite infection; however , the IFN-β level quickly returned to low levels despite continuous increase of parasitemia ( Fig 1D ) . These observations were similar to our previous measurements of IFN-I in splenic cDCs after N67 infection [8] . These results suggest that CD40 played a role , despite not a major role , in the IFN-β response and in controlling early parasitemia during N67 parasite infection , e . g . , lower day 2 IFN-β level in the CD40-/- mice resulted in higher day 4 parasitemia and earlier host death . The decline of IFN-β after day 2 also suggests active regulation of IFN-I levels during the infection . To further investigate the effects of CD40 on IFN-I response in vitro , we obtained bone marrow derived macrophages ( BMDM ) from WT and CD40-/- mice , stimulated the cells with cGAMP or poly ( I:C ) in vitro and measured IFN-β protein levels in the culture media . Significant reduction of IFN-β protein was observed in CD40-/- BMDM relative to those from WT mice when stimulated with 250 ng or 50 ng cGAMP , respectively ( Fig 1E ) . Similarly , significantly lower levels of IFN-β were detected in the cells from the CD40-/- mice when stimulated with poly ( I:C ) ( Fig 1F ) . These results show that CD40 plays a role in cGAMP and poly ( I:C ) stimulated IFN-β responses , likely mediated by STING/MAVS/TLR mediated pathways . To investigate the mechanism as to how CD40 regulates IFN-I response , we transfected 293T cells with plasmids containing genes encoding CD40 , a luciferase reporter plasmid driven by the IFN-β promoter ( IFN-β-luciferase ) , and a plasmid containing renilla luciferase as control and measured luciferase activities as described previously [8] . Compared with cells transfected with control pCMV plasmid vector , introduction of the plasmid containing CD40 gene significantly increased luciferase signals with or without poly ( dAdT ) stimulation 18 h after transfection ( Fig 2A ) . However , the increases in luciferase signals were not significant when stimulated with poly ( I:C ) or N67 parasite gDNA ( Fig 2A ) . We next co-transfected the cells with plasmids containing CD40 plus those carrying genes encoding STING , MAVS , TRIF , or TBK1 and showed that co-transfection with genes encoding STING , MAVS or TBK1 could significantly increase luciferase signals , but not those encoding TRIF ( Fig 2B ) . The effects of CD40 on enhancing the STING and MAVS activities were further confirmed in dose-response experiments ( Fig 2C and 2D ) . Consistently , no differences in dose-response were observed when the cells were co-transfected with genes encoding CD40 and TRIF ( Fig 2E ) . Similar effects of CD40 on the STING pathway were observed in RAW Lucia cells; significantly higher IFN-β protein level was observed when RAW Lucia cells were transfected with a plasmid containing CD40 gene and then stimulated with cGAMP ( for ISD , P = 0 . 052 ) that targets the STING pathway ( Fig 2F ) . The effect of CD40 on enhancing IFN-I production could be mediated by direct stimulation of STING expression and/or through protein modifications . To investigate the mechanisms of how CD40 enhances STING-mediated IFN-I production , we determined protein expression at different time points after introduction of the CD40 and/or STING plasmids into 293T cells and found that the levels of STING expression were greatly elevated 24–48 h post co-transfection with the CD40 expression vector , with the highest STING level at 48 h ( Fig 3A and 3B ) . We also used various antibodies to detect CD40 protein expression . Protein bands could be detected from 12–48 h using antibodies against DDK-tag fused to CD40 and against N’- or C’-termini of CD40 ( a small band could be seen 12 h post infection using anti-C-terminus and anti DDK tag ) . Interestingly , the presence of STING decreased CD40 levels between 12–48 h ( Fig 3C and 3D ) . The decrease in CD40 protein appeared to be due to cleavage of CD40 N’-terminus because of the presence of lower molecular weight bands and the reduction in protein signals detected by anti-C’- and N’-terminal antibodies , respectively ( Fig 3C ) . The higher STING protein levels in cells co-transfected with CD40 likely contributed to the elevated luciferase signals ( Fig 3E ) . Similarly , the CD40 protein degradation when co-transfected with STING could also explain the reduced NFκB signaling in cells co-transfected with STING ( Fig 3F ) . Both STING and CD40 have been shown to activate the canonical and non-canonical NFκB pathways and IFN-I production by recruiting various TRAF molecules [16 , 27] . However , STING is a stronger IFN-I inducer than CD40 , whereas CD40 is better at activating the NFκB pathway ( Fig 3E and 3F ) . These results show that CD40 can greatly enhance the STING-mediated IFN-I response , whereas STING can inhibit the CD40-mediated NFκB pathway . We also investigated the effects of CD40 on endogenous STING expression in different cell lines and in mice after infection with N67 parasite . Introduction of CD40 expressing plasmid could increase STING levels in DC2 . 4 and J774A . 1 cells , but not obvious in the RAW Lucia cells ( Fig 3G ) . In the spleen of uninfected mice , both CD40 and STING were expressed at low levels; however , both protein levels were increased in the spleen of N67-infected WT mice day-2 and -5 p . i . ( Fig 3H–3J and S2 Fig ) . Compared with WT mice , STING expression was significantly reduced in infected CD40-/- mice on both days , whereas CD40 levels were significantly increased in the infected STINGgt/gt mice ( Fig 3H–3J ) . We also investigated CD40 and STING expression in TRIF-/- , TLR3-/- , TLR4-/- , and MyD88-/- mice infected with N67 and showed that CD40 protein level was decreased in all the KO mice , ( S2A and S2B Fig ) , although the reductions in STING in the TLR3-/- , TLR4-/- , and MyD88-/- mice were minimum . These results suggest that TRIF can positively regulate CD40 expression in vivo , whereas STING negatively affects CD40 protein level in some cell lines ( dendritic cells and monocytes ) and in vivo during malaria infection . CD40 has intracellular domains that are known to bind TRAF2 , TRAF3 , and TRAF6 in canonical or non-canonical NFκB pathways [18] . Recently , the TRAF2 and TRAF3 binding domains of CD40 were also found to be responsible for IFN-β gene transactivation through NFκB pathways [27] . Similarly , STING can also function by signaling through TRAF3 and TRAF6 to modulate canonical and non-canonical dsDNA-mediated NFκB activation [16] . To investigate the functional role of CD40 TRAF domains in the IFN-I signaling , we first introduced amino acid substitutions in the CD40 TRAF2/3 and TRAF6 binding domains based on known mutations that are critical for its binding to TRAF molecules ( Fig 4A ) [16 , 17 , 39] . Plasmids containing one or more amino acid substitutions in the TRAF binding domains were generated using a site mutagenesis kit ( Quick Change II , Agilent Technologies ) and the expected substitutions were confirmed after DNA sequencing of the mutagenized clones ( S3 Fig ) . Substitutions in the binding domains of TRAF2/3 ( T255A ) , TRAF6 ( Q238A; E239A ) , TRAF2/3/box2 ( T255A; Q264A ) , TRAF6/box2 ( Q238A; E239A; Q264A ) , TRAF2/3/6 ( T255A; Q238A; E239A ) , and TRAF2/3/6/box2 ( T255A , Q238A , E239A , and Q264A ) significantly reduced luciferase signals driven by the IFN-I promoter in 293T cells , whereas the mutation in box2 ( Q264A ) alone did not alter the luciferase signals ( Fig 4A and 4B ) . Generally , simultaneous mutations affecting two TRAF domains reduced luciferase signals more significantly than those affecting one single TRAF domain or box . These results show that TRAF2/3 and TRAF6 binding domains play a role in stimulating STING-mediated IFN-I response . The luciferase signals were much higher when CD40 and STING were co-transfected into the 293T cells compared with expression of CD40 alone ( Fig 3E ) , suggesting that the impact of CD40 on IFN-I production is more significant when acting as regulator of STING than signaling through its own NFκB pathways . At the presence of a plasmid encoding STING , only the T255A mutation affecting TRAF2/3 could significantly reduce the luciferase signal , although all the constructs with mutations in more than one domain also had significant effects ( Fig 4C ) . The results again suggest that TRAF2/3 binding domain is the key domain affecting CD40 enhancement of the STING-mediated IFN-I response . As previously reported [18] , mutations in the TRAF2/3 and TRAF6 domains also significantly affect NFκB activation with or without STING expression , although the effect of TRAF6 mutations on NFκB activation was less than that of TRAF2/3 mutation in the presence of STING ( Fig 4D and 4E ) . Mutations in the TRAF binding domains could indeed affect the ability of CD40 to enhance STING protein levels ( Fig 4F and 4G ) . These results suggest that the stimulation of STING expression and therefore IFN-I response is mainly mediated through the TRAF binding domains of CD40 . We also investigated whether overexpression of TRAF2 , TRAF3 , or TRAF6 affected IFN-β and NFκB signaling with or without the presence of CD40 , STING or both . As expected , overexpression of TRAF3 , but not TRAF2 or TRAF6 , greatly suppressed CD40 mediated NFκB activation ( S4A–S4C Fig ) , suggesting inhibition of NIK disassociation from cIAP1/2 and of NFκB activation [40 , 41] . The patterns of TRAF2 and TRAF3 in affecting IFN-β responses were similar; both enhanced IFN-β signal when transfected with STING alone , but decreased IFN-I production when co-transfected with CD40 and STING ( S4D and S4E Fig ) . TRAF6 increased IFN-β signals in all situations ( S4F Fig ) . These results suggest that TRAF2/3 and TRAF6 play a role in the CD40 and/or STING signaling pathways leading to IFN-I production , but they may have different roles at high levels of CD40 and STING; TRAF2 and TRAF3 could affect the degradation of CD40 , STING , or proteins in pathways leading to increased STING level or other IFN-I signaling . Because CD40 and STING regulate each other’s protein levels and molecular signaling , we next investigated whether there was any physical interaction between the two molecules by performing co-immunoprecipitation ( co-IP ) experiments , although CD40 and STING are known to localize at the plasma membrane and ER , respectively [12 , 17] . Protein complexes were pulled down using anti-HA antibody after co-transfection of 293T cells with plasmids containing HA-tagged STING and/or DDK/Myc-tagged CD40 , and the bound/unbound proteins were detected using anti-DDK ( or anti-CD40 ) and anti-HA antibodies . CD40-DDK was detected in the unbound fraction , but not in the fraction pulled down by anti-HA antibody ( Fig 5A ) . As expected , these results suggest that CD40 does not bind to STING directly . The lack of direct interaction is consistent with IFA results showing lack of co-localization of the two proteins in both 293T cells ( S5A Fig ) and in Hela cells ( S5B Fig ) . Next we investigated how TRAFs interact and regulate CD40 and STING protein interaction . We transfected 293T cells with V5-tagged TRAF2 , TRAF3 , or TRAF6 and CD40-DDK/Myc , STING-HA , or both , used anti-Myc or anti-HA to pull-down protein complexes , and detected the pulled-down proteins using anti-TRAFs ( V5 ) , anti-STING ( HA ) , or anti-CD40 ( Myc ) , respectively . We found that both CD40 and STING were directly associated with TRAF3 , TRAF2 , and TRAF6 , and CD40 and STING could be pulled down together , at least with overexpression of the TRAFs ( Fig 5B–5D ) , suggesting the presence of CD40/TRAF/STING complex . Compared with TRAF2 and TRAF6 , overexpression of TRAF3 decreased CD40 protein level at 24 h ( Fig 5E ) and 48 h ( Fig 5F ) post transfection . Similarly , overexpression of TRAF3 greatly degraded STING , and co-expression of CD40 could protect some STING from degradation ( Fig 5E and 5F ) . These observations can be explained by competition for binding the TRAFs between CD40 and STING . Increased binding of TRAF3 to CD40 decreased its availability to STING , leading to reduced STING degradation . CD40 stimulation has been shown to facilitate K48-linked polyubiquitination and degradation of TRAF3 , causing disassociation of NIK from cIAP1/2p100 processing and NFκB activation [40] . To investigate the mechanism of how increased CD40 expression affects STING protein levels , we assayed for STING ubiquitination with or without overexpression of CD40 in 293T cells . Our results showed that addition of CD40 could reduce STING ubiquitination , including at position K48 and K63 ( Fig 5G and 5H ) . The TRAF molecules , particularly TRAF2 and TRAF3 , could play an important role in STING ubiquitination and degradation . Many ligands are known to stimulate CD40 expression; CD40L , TNF-α , type I and II interferons , and LPS are known to induce or modulate CD40 expression [21 , 42] . LPS was shown to induce IFN-β that in turn stimulated CD40 expression through TRAM/TRIF/IRF3 signaling [42] . To investigate possible mechanisms that stimulate CD40 expression and IFN-I production during malaria infection , we stimulated RAW Lucia cells with iRBCs , poly ( I:C ) ( TLR3 ligand ) , LPS ( TLR4 ligand ) , Pam3CSK4 ( TLR1/2 ligand ) , and DMXAA ( mouse STING ligand ) and detected CD40 and STING expressions on Western blot 1 , 8 , and 24 h post stimulation . In all cases , higher CD40 expression levels were observed with increased incubation time; for STING , increased protein levels over time were observed only in cells stimulated with iRBCs and poly ( I:C ) ( Fig 6A ) . We also measured luciferase activities driven by ISRE promoter and plotted the luciferase signals against CD40 and STING protein band intensities after stimulation of RAW Lucia cells for 1 , 8 , and 24 h . Interestingly , there were similar trends of protein intensities of CD40 and STING and luciferase signals after stimulations of the cells with poly ( I:C ) , LPS , and iRBCs ( Fig 6B–6D ) , suggesting direct relationships between CD40/STING levels and IFN-I response , but not those stimulated with Pam3CSK4 and DMXAA ( S6A and S6B Fig ) . We next stimulated the RAW Lucia cells with additional ligands for 24 h and measured luciferase signals driven by ISRE promoter and CD40 protein expression with or without lipofectamine treatment on Western blot ( S6C and S6D Fig ) . Compared with un-stimulated cells , iRBCs , DMXAA , cGAMP , poly ( I:C ) , LPS , and Pam3CSK4 again induced higher ISRE luciferase signals without lipofectamine , suggesting that TLR1/2 , TLR3 , and TLR4 may play a role in enhancing CD40 levels . However , the higher luciferase signals induced by poly ( dAdT ) and cGAMP at the presence of lipofectamine indicated that some of the signals could be from mechanisms bypassing signaling by receptors on the cell surface . CD40L , IFN-γ or both did not induce significant amounts of luciferase signals or CD40 protein expression . Stimulation of 293T cells with CD40L did not increase CD40 protein level either ( S6E Fig ) . To confirm that TLRs play a role in CD40 expression , we transfected 293T cells with TLR1 , TLR2 , TLR3 , TLR4 and TLR9 for 18 h and then detected CD40 expression with or without TLR ligand [Pam3CSK4 , CpG , poly ( I:C ) , and LPS] stimulation for another 12 h . Incubation with individual TLR ligands alone or transfection of TLRs into the 293T without ligand stimulation did not increase CD40 levels ( Fig 6E and 6F ) ; however , the presence of TLR3 and TLR4 plus their ligands , respectively , greatly increased CD40 expression ( Fig 6F ) . These results further confirm the role of TLR3 and TLR4 in stimulation of CD40 expression . To investigate parasite ligands that may play a role in this process , we assayed various parasite materials for stimulation of CD40 expression and IFN-I responses . Because malaria parasites reside within RBCs , one potential mechanism for the host cells to detect parasite molecules is through phagocytosis and digestion of iRBCs . We therefore incubated iRBCs ( schizonts ) with RAW Lucia cells and measured CD40 levels and luciferase activities driven by the ISRE promoter . More than twice the numbers of iRBCs were phagocytized than those of uninfected RBCs ( Fig 7A ) and higher amounts of CD40 protein were detected 8 h post co-incubation in RAW Lucia cells ( Fig 7B ) and slightly higher amounts in DC2 . 4 cells ( Fig 7C ) . The higher level of CD40 level also led to increased luciferase signals driven by ISRE promoter ( Fig 7D ) , although the luciferase signals could also include signals from other IFN-I signaling pathways . Phagocytosis of RBCs did not induce CD40 expression or luciferase signal ( Fig 7B and 7D ) . The lack of CD40 expression and luciferase signal after phagocytosis of RBCs and the dose-dependent signals suggest that the CD40 expression and luciferase signals detected were parasite specific . We next used parasite DNA and RNA to stimulate CD40 and ISRE luciferase expression in RAW Lucia cells and showed higher levels of CD40 protein and luciferase signals in the presence of lipofectamine ( Fig 7E and 7F ) . The increased CD40 levels stimulated by parasite DNA/RNA disappeared or were greatly reduced after treatment with nucleases ( Fig 7G and 7H ) . The results are consistent with the observation of reduced CD40 and STING levels in the spleen of TRIF-/- mice infected with N67 parasites ( S2 Fig ) , suggesting involvement of TRIF ( at least ) signaling in regulating CD40 expression . We also used Plasmodium falciparum GPI ( glycophosphatidylinositol ) to stimulate RAW Lucia cells at the presence of lipofectamine . Slightly increased CD40 protein level was detected when the cells were stimulated with 1 ug GPI , but not with 0 . 1 ug GPI ( Fig 7I and 7J ) , although a small but significant increase in ISRE-driven luciferase signal was detected in RAW Lucia cells at the GPI concentrations used ( Fig 7K ) . These results are also consistent with our previous observation on TLR2-/- , TLR3-/- and TLR4-/- mice after infection with N67; parasitemia from these TLR KO mice were different from those of WT on one or more days after infection with N67 parasite [8] , suggesting that these receptors play a role in host response to N67 infection . We next used bone marrow derived dendritic cells ( BMDCs ) and purified splenic DCs from CD40-/- , STING-/- , TLR3-/- , TLR4-/- , TRIF-/- , and MyD88-/- mice to evaluate the responses to iRBC and parasite DNA/RNA . Reduction in CD40 , STING and IFN-β protein levels were detected in TLR3-/- ( Fig 8A–8D ) and/or TLR4-/- ( Fig 8E–8H ) BMDCs after stimulation with iRBC and parasite DNA/DNA . Similarly , stimulation of purified splenic DCs from CD40-/- , TLR3-/- , TLR4-/- , TRIF-/- , and MyD88-/- mice had reduced CD40 and/or STING protein levels ( Fig 8I and 8J ) . Measurement of protein levels in splenic DCs of N67-infected mice day 4 p . i . ( in vivo stimulation ) showed significant reduction of CD40 and STING in DCs from TLR3-/- , TLR4-/- , TRIF-/- , and MyD88-/- mice ( Fig 8K and 8L ) . As expected , CD40 protein was increased in STING-/- mice , whereas STING protein was decreased in CD40-/- . These results confirmed the observations in RAW Lucia cells and further support the involvement of the TLRs and their adaptors in CD40 and STING expression in vitro and in vivo after N67 infection . We also performed phagocytosis of RBCs or iRBCs by bone marrow derived macrophages ( BMDMs ) from uninfected and day 5 infected mice in vitro and showed that iRBC ( compared with RBC ) triggered higher phagocytosis by BMDMs from both uninfected and infected mice ( S7A Fig ) . Approximately 67% of BMDMs from infected mice phagocytized iRBCs ( 4 . 7% for RBC; t-test , P<0 . 001 ) , whereas ~38% of BMDMs from uninfected mice contained iRBCs ( 14 . 0% for RBCs; t-test , P<0 . 001 ) 4 h after incubation ( S7A Fig ) . Stimulation of BMDMs from uninfected mice with iRBCs , Poly ( I:C ) , LPS , Pam3CSK4 , and parasite DNA/RNA also induced CD40 and STING expression ( S7B Fig ) . Similar results were also obtained from N67-infected mice ( S7C Fig ) . Higher IFN-β levels were also detected after stimulation of BMDMs from both infected and uninfected mice with iRBC , parasite DNA/RNA and other ligands ( S7D Fig ) , although some of the IFN- β measurements could be from other IFN-I signaling pathways . These results were consistent with those of RAW Lucia and DC cells described above .
One of the important discoveries of this study is linking the CD40 protective effect to elevated IFN-I responses in this N67 malaria parasite and C57BL/6 mouse model . CD40 is a TNF receptor superfamily member that provides critical activation signals in antigen presenting cells ( APC ) such as dendritic cells , macrophages , and B cells . When bound by its ligand CD40L that is transiently expressed on T cells and other non-immune cells , CD40 can activate a wide spectrum of molecular and cellular processes including the initiation and progression of cellular and humoral adaptive immunity [17 , 18] . In malaria infection , CD40 was reported to be necessary for the maturation of liver DCs and for the accumulation of CD8+ T cells in the liver in response to the invading P . yoelii sporozoites [34] . Mice without CD40 were not able to withstand infectious challenge after immunization with a P . yoelii mutant parasite ( fabb/f- ) , suggesting that CD40 signaling is a key requirement for host immunity in a lethal P . yoelii/C57BL/6 model . Additionally , CD40 expression was elevated in DCs in the immunized mice . In a non-lethal model of Plasmodium berghei XAT and WT C57BL/6 mice , limited protective effects were observed after administration of an agonistic anti-CD40 mAb to activate CD40 signaling , although administration of the anti-CD40 antibody to γδ T cell-deficient mice 3–10 days p . i . could help eliminate the parasites [43] . In another lethal model of P . berghei ANKA/C57BL/6 mice , host mortality was greatly decreased in CD40-/- and CD40L-/- mice after infection with the parasite , even though parasitemia were similar in the WT , CD40-/- or CD40L-/- mice [44] . In this model , mortality was attributed to the breakdown of the blood-brain barrier , macrophage sequestration , and platelet consumption . Therefore , the protective effects of CD40 KO depend on infections of specific malaria strains: Larger improvement in host survival was observed in models of early lethal infections ( references 34 and 43; mice died on day 7–8 p . i . ) than those of moderate or non-lethal infections ( mice died ~day 15 p . i . in our N67 model and the non-lethal model in reference 44 ) . The mechanisms of CD40-mediated protection are likely different among the models of various disease severities because dramatically different host responses were observed in infections with different parasite strains [8] . A strong day-2 IFN-I response was induced after N67 infection , leading to suppression of early parasitemia , whereas infection with N67C parasites stimulated a lethal inflammatory response with little IFN-I production [8] . In this regard , we should emphasize that the CD40/STING/IFN-I protective mechanism we observed in this N67/C57BL/6 mouse model may only apply to infections with parasite strains that can simulate strong IFN-I responses . Although the CD40/STING/IFN-I mechanism may not play an important role in some infections , it likely represents one of the possible protective mechanisms in human infections considering the potentially large number of P . falciparum strains in the field . Up-regulation of genes in IFN-I responses ( ISG genes ) has been associated with mild human P . falciparum malaria following an episode of severe malaria [45] , suggesting that CD40/STING/IFN-I and other IFN-I protective mechanisms may play a role in infections of some human parasite strains . DCs are the major sources of IFN-I during early malaria infection , and CD40 and CD86 were up-regulated in splenic pDCs that were dependent on TLR7 activation [8 , 46 , 47] . These data point to a potential link between up-regulation of CD40 expression and IFN-I production during malaria infection . Our current study showed that CD40 not only played a protective role during P . yoelii infection of C57BL/6 mice , but also regulated early IFN-I production for the first time . In our model , mice without CD40 had higher parasitemia on day 4 p . i . , died earlier , and had lower levels of IFN-β than the WT mice day 2 p . i . Although there was no correlation between elevated expressions of CD40/STING and IFN-β level at day 5 p . i . in vivo , the down-regulation of IFN-I after day 2 p . i . was likely due to activation of negative regulators in some signaling pathways downstream of STING . Indeed , our recent study identified a large number of negative IFN-I response regulators , including several molecules ( for example , Fosl1 , Fcgr1 , and selenbp2 ) that could down-regulate STING signaling during malaria infection [38] . However , the spike of day-2 IFN-I could affect the expression of ISGs and activate other immune pathways that could have some impact on host immune responses and survival in later phases of infection . The WT and CD40 KO mice also had similar parasitemia after day 8 , but the CD40 KO mice died earlier . In the P . berghei ANKA/C57BL/6 model , better survival rates were found in CD40-/- and CD40L-/- mice after infection with the parasite , despite similar parasitemia in the WT , CD40-/- or CD40L-/- mice [44] . Similarly , mice infected with N67 and N67C parasites had almost identical parasitemia before day 5 p . i . ; mice infected with N67C died on day 7 , but those infected with N67 suppressed parasitemia to below 10% [8] . These observations suggested that the main cause of host death was likely due to over-responses of host immunity , not parasitemia . Our results are consistent with most of the previous studies suggesting a protective role of CD40 during malaria infections , although its functional roles in different malaria models could be different , and provide a new mechanism on CD40 mediated protection through IFN-I production during early phase of malaria infection . The second important discovery of this study is linking the increased CD40 expression to higher STING protein level that greatly elevates the IFN-I response . A large number of studies have been done on CD40; however , the previous studies on CD40 have focused on inflammatory and adaptive immune responses to infection , not stimulation of IFN-I in early innate response . CD40 is best known for its interaction with CD40L expressed on CD4+ T cells and the subsequent signaling events leading to activation of NFκB and NFκB-like transcription factors and secretion of Th1 cytokines such as IL-12 and IFN-γ and to recruitment and priming of CD8+ T cells [17 , 18] . CD40-CD40L ligation regulates production of various inflammatory cytokines such as MIP-1α , TNF-α , IL-8 , and IL-12 by DCs and IL-1α , IL-1β , TNF-α , IL-6 , and IL-8 by monocytes . CD40-mediated signaling leads to the transcription of host defense genes against pathogens through activation of NFκB , MAPK ( Mitogen-Activated Protein Kinase ) and STAT3 ( Signal Transducers and Activators of Transcription-3 ) pathways [19 , 48] . Recently , a role of CD40 in regulating IFN-β expression through canonical NFκB and non-canonical NFκB signaling pathways leading to recruitment of p52 to the IFN-β promoter was described , although the fold changes in IFN-β level mediated through the NFκB signaling were relatively small [27] . Using in vitro transfection of cell lines , here we showed that over expression of CD40 in 293T or RAW Lucia cells increased STING protein levels , leading to highly elevated luciferase signals driven by ISRE and IFN-β promoters . The fold changes in luciferase signals in CD40+STING co-transfected cells were much higher than those transfected with CD40 or STING alone , suggesting that CD40 can greatly enhance STING-mediated IFN-I responses to infections . The role of CD40 in enhancing STING protein level was also confirmed using purified splenic DCs or total splenic cells from WT and CD40-/- mice after N67 infection . Additionally , phagocytosis of iRBCs or stimulations of BMDMs and BMDCs from infected or uninfected mice with parasite DNA/RNA also increased CD40 and STING protein levels . These in vitro and in vivo data establish a relationship of increased CD40 and STING expression . The protective roles of IFN-I in malaria infections have been reported [8–10] , including injection of recombinant IFN-α [49] . Our observations of CD40 affecting STING expression and IFN-I production in the early hours of infection add a new function to CD40 , in addition to its known functional roles in CD8+ T-cell activation and IgG isotype switching during host adaptive response . At day 11–15 p . i . , CD40-CD40L ligation , activation of NFκB in APCs , and recruitment of CD8+ T-cells , as reported previously [34] , were likely to play a role in host survival . It would be interesting to investigate how the spike of day 2 IFN-I affects the adaptive immune response , including the effects of higher CD40 level on T-cell activation and isotype switching . Proper activations of T-cells and B-cells for antibody production are critical for immune protection and host survival in later phases of infection . Interestingly , whereas CD40 could increase STING expression and greatly enhance IFN-I production , the presence of high levels of STING appeared to decrease CD40 protein levels both in transfected cells and in mice infected with malaria parasite N67 , possibly through degradation of the N-terminus of CD40 . Therefore , increased STING expression may dampen CD40 mediated inflammatory responses and damage to the host , while promoting a strong IFN-I response to control parasitemia . The negative regulatory role of STING in suppressing inflammation in systemic lupus erythematosus ( SLE ) was reported recently [50]; our observations here add another mechanism explaining the role of STING in regulating inflammatory responses . Many autoimmune diseases are associated with high levels of IFN-I and interferon induced genes ( ISGs ) , particularly IFN-I produced by over-activation of STING signaling pathways [36 , 37] . Interestingly , high levels of CD40 and CD40L or engagement of CD40/CD40L are also associated with the same sets of autoimmune diseases such as SLE and Aicardi-Goutieres syndrome ( AGS ) [28 , 51] . However , the functional roles of CD40 in autoimmune and inflammatory diseases have been largely attributed to CD40-CD40L interaction that mediates T-dependent B cell responses and T cell priming [29] . Our observation of CD40 enhancement of STING mediated IFN-I production provides an alternative mechanism that may unite these two hallmarks of autoimmune diseases . Higher expression of CD40 could increase STING level in DCs and greatly elevate IFN-I production , which may influence the susceptibility to autoimmune and inflammatory diseases in addition to traditionally known pathways of T- and B-cell activation . To investigate how the signal from CD40 was relayed to the downstream pathways leading to higher STING level , we constructed plasmids with various mutations in the cytoplasmic TRAF binding domains of CD40 and showed that the TRAF binding domains played a role in the increased STING level and IFN-I activity . Mutations in both TRAF2/3 and TRAF6 binding domains could affect luciferase activities driven by IFN-β promoter; however , when co-transfected with STING , the effect of mutations in the TRAF6 binding domain disappeared . The mutations in the TRAF binding domains also affect STING protein level , consistent with the observations showing increased CD40 and STING proteins after co-transfection of plasmids containing genes encoding both molecules and decreased STING protein in CD40-/- mice after parasite infection . As expected , the mutations in the TRAF binding domains also influenced NFκB mediated signaling . In particular , introduction of TRAF3 greatly reduced CD40-mediated NFκB signaling , which was consistent with a role of TRAF3 in regulating NIK protein level . TRAF3 degradation is necessary for the cell to activate the alternative NFκB pathway [40 , 41] , and higher expression of TRAF3 will inhibit the alternative NFκB signaling . Overexpression of TRAF2 and TRAF3 also inhibited IFN-β production mediated by CD40 directly or by CD40-enhanced STING signaling . These observations suggest a role of TRAF2 and TRAF3 in regulating levels of STING protein and IFN-I . Both mutations in CD40 TRAF2/3 binding domains and overexpression of TRAF2/TRAF3 may lead to more ‘free’ TRAF2 and TRAF3 not associated with CD40 , resulting in the observed lower levels of STING and IFN-β . On the other hand , increased CD40 levels will bind more TRAF2/TRAF3 and reduce the levels of ‘free’ TRAF2/TRAF3 , resulting in higher STING and IFN-β levels because overexpression of TRAF3 led to STING degradation ( Fig 5 ) . It has been shown that Sendai virus ( SeV ) infection increased K48 and K63 ubiquitination of TRAF3/6 after recruitment of cIAP1/2 to mitochondrial associated TRAF3 , which triggered IFN-I induction mediated through MAVS/VISA signaling [52] . However , viral infection or cIAP1/2 overexpression did not cause noticeable degradation of TRAF3/6 , and activation of downstream kinases such as TBK1 and TAK1 were proposed . Our data suggest a mechanism of reduced availability of TRAF2 and TRAF3 after binding to elevated level of CD40 and decreased STING ubiquitination and degradation , leading to higher STING level and IFN-I production . We showed that CD40 and STING could be co-IP when cells were transfected with TRAF2 , TRAF3 or TRAF6 , suggesting the presence of a CD40/TRAF/STING complex . This direct contact of the molecules may explain the reduced STING ubiquitination after CD40 expression , which could be mediated by TRAF2 and/or TRAF3 directly or by recruiting other enzymes with ubiquitinase activities . Reduced ubiquitination of STING at a site such as K48 may lead to higher protein levels . Our data suggest that TRAF3 plays an important role in STING stability; however , TRAF molecules are known to play a role in many signaling pathways [53] , and any change in the TRAF protein levels in the cell will likely affect more than one pathways , which will require additional investigations . Another interesting question is how CD40 is activated by malaria ligands . Some TLRs such as TLR2 and TLR4 have MyD88-dependent and MyD88–independent pathways , depending on cellular locations of the TLRs and adaptors recruited [54 , 55] . When TLR4 is located at the plasma membrane , activation of the MyD88-dependent pathway leads to production of proinflammatory cytokines; whereas the MyD88-independent signaling pathway is involved when TLR4 is at the endosome [54] . The MyD88-independent pathway signals through a TRIF-related adaptor molecule ( TRAM ) , TRIF and IRF3 , which leads to IFN-I production [55–57] and/or activation of CD40 , CD54 ( ICAM1 ) and CD86 [58] . Additionally , LPS has been shown to stimulate CD40 expression through TLR4 signaling that involved activation of both NFκB and signal transducer and activator of transcription1α ( STAT-1α ) in macrophages and microglia [42] . Various TLRs are known to play a role in host response to malaria infections [59 , 60] . Consistent with these reports , we showed that over expressions of TLR1/2 , TLR3 , and TLR4 followed by stimulations with their ligands [Pam3CSK4 , poly ( I:C ) , and LPS , respectively] could increase CD40 protein levels as well as luciferase signals driven by ISRE promoter in 293T cells . Consistently , stimulations of BMDMs and BMDCs with TLR ligand poly ( I:C ) and LPS , parasite DNA/RNA , and iRBCs increased CD40/STING levels ( S7 Fig ) , and compared with cells from WT , stimulations of BMDCs from TLR3-/- and TLR4-/- mice with parasite DNA/RNA and iRBCs had reduced CD40/STING proteins and IFN-β ( Fig 8 ) . Additionally , splenic DCs from TLR3-/- , TLR4-/- , TRIF-/- , and MyD88-/- had reduced CD40 and/or STING protein levels after in vitro stimulations with iRBC or in vivo N67 infection . These data support the involvement of TLR3 and TLR4 in CD40/STING expressions , although we still do not understand how TLR4 KO affects CD40/STING/IFN-β expression . TLR2 and to a lesser extent TLR4 were shown to mediate P . falciparum GPI recognition and signaling leading to proinflammatory responses [61] . Patients with severe and mild malaria also showed increased surface expression of TLR2 and TLR4 on CD14+ monocytes and cDCs and decreased intracellular expression of TLR9 on pDCs [59 , 62] . However , these reported TLR signaling pathways may or may not lead to increased expression of CD40 , STING , and IFN-I . TLR9 recognizes unmethylated CpG dinucleotides and signals mostly through the MyD88/NFκB pathway . TLR9 was reported to play an important role in the regulation and development of protective immunity to malaria [63 , 64] . Although our data showed that stimulation of TLR9 only slightly increased CD40 expression in 293T cells in vitro , and we did not have TLR9-/- mice to confirm the role of CD40 expression in vivo , the effects of MyD88 KO on CD40/STING expression in the infected splenic DCs suggested a minor role of TLR9 in parasite DNA recognition and CD40 expression . Parasite DNA can be converted into 5’ppp-RNA and recognized by cytosolic RNA sensors [8] . An unknown DNA sensor was reported to recognize AT-rich parasite DNA [7] . This receptor could be the recently identified cGAS [65] or an additional unknown receptor that may stimulate CD40 expression . Additional investigations are necessary to identify parasite DNA receptor ( s ) that can stimulate CD40/STING expression during malaria infection . Importantly , we also showed that iRBCs and parasite-derived materials including nucleic acids and possibly GPI could also increase CD40 levels and/or luciferase signals driven by ISRE promoter , although potential DNA contamination in GPI preparations might contribute to the observed activity [46 , 64] . Furthermore , there were good correlations between CD40/STING protein levels and ISRE driven luciferase signals after poly ( I:C ) , LPS , and iRBC stimulations , indicating a causative effect of CD40/STING protein levels and IFN-I/ISG responses . These data suggest that these parasite molecules can first activate CD40 expression through TLR/TRIF and other signaling pathways , leading to increased STING and IFN-I expression . Although the observations of changes in CD40 expression after LPS and iRBC stimulations or TLR4 KO suggested potential involvement of TLR4 ( maybe through TLR4/TRAM/TRIF/IRF3 signaling ) in CD40 expression , we are not sure of the specific parasite ligands for TLR4 . Parasite GPI could be a potential TLR2/4 ligand [61] . Interestingly , we also showed that P . falciparum GPI could stimulate low level , but significant , ISRE derived luciferase activity in RAW Lucia cell , although we could detect a small difference in CD40 protein level only at the 1 μM GPI group ( Fig 7J ) . The change in luciferase signal at the 0 . 1 μM GPI group was very small , and Western blot might not be sensitive enough to detect the small difference in protein band intensity . The luciferase signals could also be produced by pathways other than CD40/STING signaling . Our data also suggest that parasite nucleic acids can act as TLR3 ligands because TLR3 KO significantly affects CD40 and STING protein levels after parasite DNA/RNA stimulations . Based on the observations in this study , we can construct some preliminary signaling pathways for the elevated IFN-I response mediated by the interaction of CD40 and STING ( S8 Fig ) during early malaria infection , at least for the N67/C57BL/6 model . First , phagocytosis of iRBCs releases parasite materials such as nucleic acids and possibly lipids/GPI that are recognized by TLRs ( 2 , 3 and 4 ) , which triggers signaling mostly through TRIF dependent pathways leading to increased CD40 protein level . The increased CD40 expression leads to reduced STING ubiquitination and enhanced-levels of STING through binding additional TRAFs , particularly TRAF3 . Increased CD40 expression binds more TRAF2 and TRAF3 and diminishes the pools of ‘free’ TRAF2 and TRAF3 , leading to reduction of STING ubiquitination and degradation . Increased STING level can then greatly facilitate IFN-I production leading to suppression of early parasitemia and better host survival . In addition to demonstrating the protective role of CD40 in malaria infection by increased IFN-I production , this study reveals an unknown function of CD40 and TRAF2/TRAF3 in regulating STING and IFN-I expression and connects several innate signaling pathways involving TLRs , CD40 , STING and TRAFs . Of course , this CD40/STING/IFN-I signaling pathway is just one of many protective mechanisms the host mounts against the parasites , and a complete resolution of malaria infection will require a well-balanced response involving a large number of molecules . Our observations may also provide important information for better understanding of the molecular mechanisms underlying many autoimmune diseases such as SLE and AGS that have been associated with STING-mediated high IFN-I levels and/or CD40-CD40L meditated T- and B-cell activation and inflammation . Our data suggest an alternative mechanism that CD40 may contribute to autoimmune diseases through elevated activities of STING and IFN-I production .
All animal procedures were performed in accordance with the approved protocol ( approval #LMVR11E ) by Institutional Animal Care and Use Committees ( IACUC ) at the National Institute of Allergy and Infectious Diseases ( NIAID ) DIR ACUC following the guidelines of the Public Health Service Policy on Humane Care and Use of Laboratory Animals and AAALAC . The P . y . nigeriensis used in this study was described previously [66] . Parasitemia were counted from Giemsa stained thin blood smears using a light microscope . C57BL/6 mice , 5- to 8-weeks-old , from Charles River Laboratories were injected with 1×106 iRBCs as described [66 , 67] . The genetic KO mice ( CD40-/- , TLR3-/- , TLR4-/- , TRIF-/- , MyD88-/- , STINGgt/gt ) were from the Jackson Laboratory . Anti-CD40 N-terminus ( H-120; sc-9096 ) and anti-CD40 C-terminus ( C-20; sc-975 ) were purchased from Santa Cruz Biotechnology ( Dallas , TX ) ; Anti-β-actin ( A2228 ) was from Sigma ( St . Louis , MO ) ; anti-DDK mouse mAb ( 8146 ) , anti-HA rabbit mAb ( 3724 ) , and anti-STING ( 3647 ) were from Cell Signaling Technology ( Danvers , MA ) . Primary antibody binding was visualized using horseradish peroxidase ( HRP ) -conjugated secondary antibodies specific for mouse or rabbit IgG ( Sigma ) . The anti-Myc ( mouse mAb ) and anti-HA ( rabbit mAb ) were used in immunofluorescence microscopy ( IFA ) . Secondary antibodies used for IFA were Alexa Fluor 488 goat anti-mouse IgG and Alexa Fluor 594 goat anti-rabbit IgG ( Abcam , Cambridge , MA ) . DMXAA , 2’3’-cGAMP , Pam3CSK4 , CpG ( ODN 2395 ) were from InvivoGen . Poly ( I:C ) , LPS , Poly ( dAdT ) were from Sigma . P . falciparum GPI was prepared as described [68] . Human embryonic kidney 293T ( ATCC , Manassas , VA ) , Hela ( ATCC , Manassas , VA ) , RAW Lucia ISG cells ( InvivoGen , Carlsbad , CA ) and DC2 . 4 ( ATCC , Manassas , VA ) were maintained in DMEM containing 10% heat-inactivated FBS , 50 units/ml penicillin and 50 mg/ml streptomycin . BMDMs were collected from mouse femurs and tibiae . Cells were cultured in DMEM containing 10% FBS , antibiotics , and 30% L929 supernatant . The cells were grown for 7 days with addition of 10 ml fresh DMEM/L929 on day 4 . Mature cells were harvested using Nunc Cell Scrapers and plated into 24-well plates for further experiments . BMDCs were obtained by flushing bone-marrow cells from the marrow cavities of femurs and tibiae of C57BL/6 WT or different KO mice . Erythrocytes were depleted with ACK lysis buffer ( Thermo Fisher Scientific ) , and the remaining cells were cultured ( as day 0 ) at 1 × 107 cells/petri dish in the presence of murine recombinant GM-CSF at 20 ng/mL ( PeproTech , Rocky Hill , NJ ) in complete medium ( DMEM with 10% heat-inactivated FBS and penicillin/streptomycin antibiotics ) . The cultures were replenished with fresh medium on day 3 . At day 6 , primary/immature BMDCs were harvested and 2 x 105 cells per well ( of 24-well plate ) were seeded for subsequent ligand or iRBCs stimulation . DCs from spleen were purified using a EasySep Mouse Pan-DC Enrichment Kit ( STEMCELL Technologies , Cambridge , MA ) . Mouse IFN-β was detected using an ELISA kit according to the manufacturer's protocols ( PBL Biomedical Laboratories , Piscataway , NJ ) . Cells cultured in glass bottom dishes were fixed with 4% formaldehyde in PBS , pH = 7 . 3 for 15 min , followed by permeabilization with 0 . 5% Triton X-100 for 15 min at room temperature ( RT ) . Primary anti-DDK mouse mAb for CD40-DDK staining and anti-HA rabbit mAb for STING-HA staining were added ( 1:2000 dilution ) for 1 h at RT after blocking with washing buffer ( 3% BSA in PBS ) for 1 hour . The samples were further stained with Alexa Fluor 488 goat anti-mouse IgG or Alexa-594 goat anti-rabbit IgG , respectively , followed by 3X washes , and were mounted using ProLong Gold anti-fade reagent with DAPI ( Life technologies , Carlsbad , CA ) . Images were captured using Zeiss LSM 780 confocal microscope . 293T cells ( 2 × 105 ) were plated in 24-well plates and transfected using lipofectamine 2000 ( Thermo Fisher Scientific ) . Twenty ng of renilla luciferase reporter plasmid and 50 ng of firefly luciferase reporter plasmids were transfected together with indicated expression plasmids . Luciferase activity was measured 24 h ( or as indicated ) after transfection using the Dual-Glo Luciferase Assay System ( Promega , Madison , WI ) with a BMG FLUOstar OPTIMA microplate reader ( BMG LABTECH , Cary , NC ) . An Amaxa nucleofector kit V was also used for transfection according to the manufacturer's protocols ( Lonza , Walkersville , MD ) The STING-HA expression plasmids were kindly provided by Dr . Russell E . Vance1 ( University of California , Berkeley , California ) and have been previously described [69] . CD40-Myc was cloned into pCMV6-entry vector as described below . For immunoprecipitation , whole-cell extracts were prepared after transfection or stimulation with appropriate ligands , followed by incubation with anti-HA or anti-Myc magnetic beads ( Thermo Fisher Scientific , Rockville , MD ) . Co-IP was performed according to the manufacturer’s instructions and subsequently analyzed on Western blot . Cell lysates or immunoprecipitated complexes were separated on SDS-PAGE gels and transferred to PVDF membranes ( Thermo Fisher Scientific ) . Membranes were incubated with primary antibodies followed by HRP-conjugated secondary antibodies . Bound HRP-labeled antibodies were visualized using SuperSignal West Dura Extended Duration Substrate ( Thermo Fisher Scientific ) . Mouse cd40 sequence was PCR-amplified using a MGC Mouse cDNA library ( CloneId:4018221 ) as template . PCR primers were synthesized based on GenBank accession number BC029254 . 1 . The cd40 cDNA was cloned into the SgfI and MluI sites of pCMV6-entry vector ( Origene , Rockville , MD ) after PCR amplification using primers: 5’atctgccgccgcgatcgcCATGGTGTCTTTGCCTCGGC 3’ and 5’ tcgagcggccgcgtacgcgtGACCAGGGGCCTCAAGGC 3’ . A C-terminal DDK/Myc tagged cd40 was expressed under the control of CMV promoter . Point mutations in the cytoplasmic tail of CD40 were generated from this CD40-DDK/Myc expression vector using QuikChange II Site-Directed Mutagenesis Kit ( Agilent Technologies , Santa Clara , CA ) according to the manufacturer’s instructions . The following primers were utilized for the site-directed mutagenesis reactions with the desired mutations underlined: 5' CCAGTGCAGGAGGCGCTGCACGGGT 3' and its complementary oligonucleotide 5' ACCCGTGCAGCGCCTCCTGCACTGG 3' to generate T255A for mT2/3; 5' CGGGTGTCAGCCTGTCACAGCGGAGGATGGT 3' and its complementary oligonucleotide 5' ACCATCCTCCGCTGTGACAGGCTGACACCCG 3' to generate Q264A for mbx2; 5' CTCGACGGCAAGATCCCGCGGCGATGGAAGATTATCCCG 3' and its complementary oligonucleotide 5'CGGGATAATCTTCCATCGCCGCGGGATCTTGCCGTCGAG 3' to generate Q238A and E239A for mT6; Blood samples of 1 . 5–2 ml were collected from two N67- infected mice ( 20–40% total parasitemia ) and synchronized as described [70] . Briefly , collected blood was immediately added to 5 ml anticoagulant solution ( 1 . 5% sodium citrate , 0 . 9% NaCl ) . The blood sample was passed through a NWF filters ( Zhixing Bio , Bengbu , China ) to remove white blood cells [71] . The flow-through containing RBCs were centrifuged at 2000 rpm for 5min , and the pellet was suspended in RPMI 1640 culture medium [RPMI 1640 1000 ml , NaHCO3 ( 7 . 5% ) 30 ml , gentamycin ( 10 mg/ml ) 1ml , 20% ( vol/vol ) FBS] and distributed into culture flasks . The flasks were incubated at 37°C for 16 h in a humidified CO2/air incubator with shaking ( 60 rpm ) . The cultures were centrifuged at 2 , 500 rpm for 5 min , and the pelleted RBCs were re-suspended in 2 ml of PBS with 5% FBS . Two ml culture suspension was carefully placed on top of 5 ml 72% Percoll/NaCl solution ( 6 . 5ml Percoll , 0 . 72 ml of 1 . 5 M NaCl , 2 . 8 ml of 0 . 15 M NaCl ) and centrifuged at 450 g in a swing-out rotor at RT for 20 min . Schizonts at the interface were transferred to a 50-ml tube . Control RBCs were incubated and treated in the same way without schizont purification . Approximately 2×106 RAW Lucia cells were plated into wells and grown to 70–80% confluence before addition of 2×107 iRBCs or RBCs , respectively . The cells were incubated at 37°C and 5% CO2 , and samples were collected for the flow cytometry after 1 h or 8 h incubation . Flow cytometry analysis was performed on a BD FACSCalibur equipped with a 488 nm laser and three detection channels ( FL-1: 515–550 nm , FL-2: 575–620 nm , FL-3: >630 nm ) . Data acquisition and analysis was done using FlowJo . RAW Lucia cells were gated and plotted based on the forward scatter ( FSC ) versus side scatter ( SSC ) . For staining of RBCs , 5×107 RBCs in 1 ml PBS were incubated with 5 μM CFDA-SE ( Thermo Fisher Scientific ) at RT for 5 min before washes and flow cytometry counting . Counts from phagocytosis of uninfected RBCs were excluded by gating the same areas from RBC controls . For analysis of STING ubiquitination , 293T cells were transfected with plasmids expressing STING with C-terminal Myc tag and CD40 or empty vector as well as HA-tagged WT ubiquitin or ubiquitin mutants containing only one lysine at specific position ( K6 , K11 , K27 , K29 , K33 , K48 , K63 ) . The mutant ubiquitin plasmids were obtained from Addgene originally deposited by Sandra Weller as described [72] . Twenty hrs after transfection , cell lysates were immunoprecipitated using anti-Myc beads , followed by immunoblot analysis with anti-HA to detect ubiquitinated STING . | Malaria parasites infect millions of people and can cause severe disease with diverse symptoms . No highly effective vaccine is available to prevent malaria infection , largely due to the lack of understanding of the molecular mechanisms of host-parasite interactions . Here we investigated the interaction of two important host molecules , CD40 and STING , in host type I interferon ( IFN-I ) responses to malaria infection using a mouse malaria model . For the first time , we showed that malaria infection could increase CD40 expression , which in turn enhanced the expression of STING , leading to increased production of IFN-I during early infection and better host survival . The increased CD40 expression could be stimulated by phagocytosis of infected red blood cells ( iRBCs ) , parasite DNA/RNA and specific ligands of Toll-like receptors ( TLRs ) , suggesting recognition of parasite materials and signaling by TLRs . This study reveals previously unrecognized protective pathways during malaria infection and provides important information for designing better strategies for malaria control and vaccine development . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"luciferase",
"pathology",
"and",
"laboratory",
"medicine",
"enzymes",
"293t",
"cells",
"biological",
"cultures",
"tropical",
"diseases",
"enzymology",
"parasitic",
"diseases",
"parasitic",
"protozoans",
"immune",
"receptor",
"si... | 2016 | Increased CD40 Expression Enhances Early STING-Mediated Type I Interferon Response and Host Survival in a Rodent Malaria Model |
The bacterial flagellar type III export apparatus utilizes ATP and proton motive force ( PMF ) to transport flagellar proteins to the distal end of the growing flagellar structure for self-assembly . The transmembrane export gate complex is a H+–protein antiporter , of which activity is greatly augmented by an associated cytoplasmic ATPase complex . Here , we report that the export gate complex can use sodium motive force ( SMF ) in addition to PMF across the cytoplasmic membrane to drive protein export . Protein export was considerably reduced in the absence of the ATPase complex and a pH gradient across the membrane , but Na+ increased it dramatically . Phenamil , a blocker of Na+ translocation , inhibited protein export . Overexpression of FlhA increased the intracellular Na+ concentration in the presence of 100 mM NaCl but not in its absence , suggesting that FlhA acts as a Na+ channel . In wild-type cells , however , neither Na+ nor phenamil affected protein export , indicating that the Na+ channel activity of FlhA is suppressed by the ATPase complex . We propose that the export gate by itself is a dual fuel engine that uses both PMF and SMF for protein export and that the ATPase complex switches this dual fuel engine into a PMF-driven export machinery to become much more robust against environmental changes in external pH and Na+ concentration .
Many membrane-embedded biological nanomachines utilize proton motive force ( PMF ) across the membrane for their biological activities . In Escherichia coli and Salmonella enterica , PMF is utilized as the energy source for ATP synthesis , solute transport , nutrient uptake , protein transport , multidrug efflux pump and flagellar motility [1] . Alkaliphilic bacteria and hyperthermophilic bacteria utilize sodium motive force ( SMF ) instead of PMF [1] . The flagellar motor of E . coli and Salmonella uses H+ as the coupling ion to power flagellar motor rotation . In contrast , the flagellar motor of marine Vibrio and extremely alkalophilic Bacillus utilizes Na+ as the coupling ion instead of H+ [2] . It has been reported that some systems such as the melibiose permease of E . coli [3] and the flagellar motor of alkalophilic Bacillus clausii [4] can utilize both H+ and Na+ as their coupling ion . Interestingly , the flagellar motor of Bacillus alcalophilus Vedder 1934 can conduct K+ as well as Na+ [5] . Each biological system appears to have been optimized for the best use of specific ions according to the environmental conditions . The bacterial flagellum , which is responsible for motility , is a macromolecular assembly made of about 30 different proteins and consists of the basal body rings and a tubular axial structure [6–8] . Fourteen flagellar proteins are transported through these structures by its specific export apparatus for their incorporation at the distal end of the growing flagellar structure . The export apparatus consists of a PMF-driven transmembrane export gate complex made of FlhA , FlhB , FliO , FliP , FliQ and FliR and a cytoplasmic ATPase complex consisting of FliH , FliI ATPase and FliJ [6–8] . Because the flagellar export apparatus is evolutionally related to the injectisome of pathogenic bacteria , which inject virulence effector proteins into their eukaryotic host cells for invasion , these two systems are categorized to type III secretion systems [9] . The flagellar and non-flagellar type III export apparatuses require ATP and PMF as the energy source for efficient and rapid protein export [10–15] . Because the chemical energy derived from ATP hydrolysis by the ATPase is not essential for flagellar and non-flagellar type III protein export [11 , 12 , 15] , PMF is the primary fuel for unfolding and translocation of export substrates [10] . Since the flagellar type III export apparatus processively transports flagellar proteins to grow flagella even in the presence of the extremely low ATPase activity of FliI carrying the E211D substitution , relatively infrequent ATP hydrolysis by the cytoplasmic ATPase complex is sufficient for gate activation to start processive translocation of export substrates for efficient flagellar assembly [16] . PMF consists of two components: the electric potential difference ( Δ ) and the proton concentration difference ( ΔpH ) . Δψ alone is sufficient for flagellar protein export [12] but the export gate alone , in the absence of FliH and FliI , requires the ΔpH component of PMF in addition to Δψ [13] . An increase in the ΔpH component enhances flagellar protein export in the absence of FliH and FliI [13] . D2O significantly reduces the rate of protein export in the absence of the FliH and FliI , also indicating that H+ translocation through the export gate is directly coupled with protein translocation [13] . A specific interaction between FliJ and FlhA brought about by FliH and FliI switches the export gate into a highly efficient Δψ-driven export engine [13 , 17] . However , it remains unknown how and why the ΔpH component is required for the export gate to act as a H+–protein antiporter in the absence of the cytoplasmic ATPase complex . To clarify the role of H+ in flagellar protein export , we diminished the ΔpH component of PMF and investigated the export properties of a ΔfliH-fliI flhB ( P28T ) bypass mutant whose second-site FlhB ( P28T ) mutation increases the export efficiency of some substrates to wild-type levels and thereby restores flagellar formation in the absence of FliH and FliI [11] . We show that the ΔfliH-fliI flhB ( P28T ) bypass mutant can use Na+ as the coupling ion to assemble flagella in the absence of the ΔpH component , indicating that , in addition to PMF , the export gate is powered by SMF in the absence of the cytoplasmic ATPase . We also show that FlhA has both H+ and Na+ channel activities .
Our first step was to define whether the export gate utilizes only H+ as the coupling ion for flagellar protein export . Our assays used a wild-type strain in which Δψ alone is sufficient for protein export and a ΔfliH-fliI flhB ( P28T ) bypass mutant that can form flagella in the absence of FliI ATPase and is known to require both the Δψ and ΔpH components for the protein export activity [11–13] . We also used an external pH of 7 . 5 to diminish ΔpH of the energy source because the intracellular pH is maintained at around 7 . 5 [13] . The growth rate of Salmonella cells was not affected under our experimental conditions except in no salt condition , under which it was slightly reduced compared to the presence of 100 mM NaCl ( S1 Fig ) . In wild-type cells , neither Na+ , Li+ , K+ nor Mg2+ affected the secretion level of FlgD ( hook cap protein ) ( Fig 1A , left panel ) . In the ΔfliH-fliI flhB ( P28T ) ΔflhA mutant as a negative control , no FlgD was detected in the culture supernatants ( right panel ) . In the ΔfliH-fliI flhB ( P28T ) bypass mutant , Na+ dramatically enhanced FlgD secretion ( middle panel , lane 7 ) whereas neither of Li+ , K+ and Mg2+ did so ( middle panel , lanes 8–10 ) . The intracellular level of FlgD was not changed by these treatments ( middle panel , lanes 1–5 ) . There was no significant difference in PMF under these experimental conditions , either ( S2 Fig ) . Consistently , the free-swimming speed , which is proportional to PMF [18] , was not affected by the presence or absence of NaCl up to 100 mM ( S3 Fig ) . The levels of FlgD secreted by ΔfliH-fliI flhB ( P28T ) showed NaCl concentration dependence at external pH 7 . 5 ( Fig 1B , middle panel ) . We obtained the same results with FlgE ( hook protein ) , FliK ( hook-length control protein ) , FlgK ( first hook-filament junction protein ) and FlgL ( second hook-filament junction protein ) ( S4 Fig ) . In agreement with this , more than 95% of the ΔfliH-fliI flhB ( P28T ) cells had a couple of flagellar filaments in the presence of 100 mM NaCl whereas almost no flagella were observed in the absence of NaCl ( Fig 1C , middle panel ) . We also obtained essentially the same results with an alternative ΔfliH-fliI flhA ( V404M ) bypass mutant ( S5A Fig ) . In contrast , both the secretion levels ( Fig 1B , left panel ) and flagellar formation ( Fig 1C , right panel ) by the wild-type showed no Na+ dependence . These increased levels of protein secretion and flagellar assembly with an increase in external Na+ concentration in the ΔfliH-fliI flhB ( P28T ) bypass mutant could be an indirect result of increased flagellar gene expression [19] . On testing flagellar promoter activities , however , the flagellar gene expression levels were slightly higher in the absence of NaCl than in its presence ( S6 Fig ) . It has been shown that increased ionic strength facilitates the export of a flagellum-specific anti-sigma factor , FlgM , by wild-type cells , enhancing motility in soft agar [20] . Because neither Li+ , K+ nor Mg2+ affected flagellar protein export by the ΔfliH-fliI flhB ( P28T ) bypass mutant ( Fig 1A , middle panel , lanes 8–10 ) , we suggest that Na+ is specific for this positive impact on flagellar protein export by the bypass mutant . To test whether Na+ directly facilitates flagellar protein export by the transmembrane export gate complex in the absence of FliH and FliI , we analyzed the effect of depletion of Na+ ions on protein export by the ΔfliH-fliI flhB ( P28T ) bypass mutant . We chose FlgD as a representative export substrate because the level of FlgD secretion by the bypass mutant is even higher than the wild-type level due to its poor ability to form the hook structure [11] . Since the flagellar type III export apparatus switches its export specificity from hook-type ( FlgE , FlgD and FliK ) to filament-type proteins ( FlgM , FlgK , FlgL , FliD and FliC ) upon completion of hook assembly [6–8] , we used a flgE null mutant ( ΔflgE ) as a control; this strain continues to secrete FlgD because hook assembly does not occur and hence the export apparatus remains in the hook-type substrate specificity state . The cells were grown exponentially in T-broth ( pH 7 . 5 ) containing 100 mM NaCl to produce the basal bodies with the functional type III export apparatus associated . After washing twice with T-broth ( pH 7 . 5 ) , the cells were resuspended in T-broth ( pH 7 . 5 ) with or without 100 mM NaCl , and incubation was continued at 30°C for 1 hour . Cellular and culture supernatant fractions were prepared and analyzed by immunoblotting with polyclonal anti-FlgD antibody ( Fig 2 ) . Removal of Na+ ions considerably reduced the secretion level of FlgD by the ΔfliH-fliI flhB ( P28T ) bypass mutant ( right panel , lane 4 ) but not by the ΔflgE mutant ( left panel , lane 4 ) . These results suggest that Na+ is directly involved in flagellar protein export by the export gate in the absence of FliH and FliI but not in their presence . To test whether the Na+-dependent protein export results from these bypass mutations , we analyzed the effect of Na+ concentration on the levels of FlgD secreted by ΔfliH and ΔfliH-fliI mutants . The FlgD secretion levels by these two mutants showed a clear dependence on external Na+ concentration at external pH 7 . 5 ( S5B and S5C Fig ) , indicating that the flhB ( P28T ) and flhA ( V404M ) bypass mutations do not change the ion selectivity of the export gate complex . Therefore , we suggest that the gate can intrinsically utilize SMF in addition to PMF . Phenamil is known to inhibit Na+ channel activity without affecting cell growth [21] . The polar flagellar motor of marine Vibrio is powered by SMF , and the motor speed is decreased with an increase in the concentration of phenamil , showing a complete stop by 50 μM phenamil [22 , 23] . To investigate whether the export gate directly utilizes Na+ to drive flagellar protein export , we analyzed the effect of phenamil on flagellar protein export by wild-type cells and the ΔfliH-fliI flhB ( P28T ) bypass mutant . The levels of FlgD secreted by the ΔfliH-fliI flhB ( P28T ) bypass mutant cells were markedly reduced with increasing concentrations of phenamil up to 200 μM , which was 4-fold higher than the phenamil concentration that totally inhibits the swimming motility of Vibrio cells ( Fig 3A , right panel ) . The intracellular levels of FlgD were maintained . We obtained the same results with ethylisopropylamiloride ( EIPA ) ( Fig 3B , right panel ) , which acts not only as an inhibitor of Na+/H+ exchange but also as a Na+ ion channel blocker [4 , 5] . Interestingly , neither phenamil nor EIPA inhibited FlgD secretion by the wild-type ( left panels ) , indicating that the export apparatus does not use Na+ as the coupling ion in the presence of FliH and FliI . These treatments did not affect the swimming speeds of wild-type and fliH-fliI bypass mutant cells ( S7 Fig ) , indicating that PMF was not changed at all . Therefore , we suggest that the export gate is intrinsically a dual fuel engine that can use both H+ and Na+ as the coupling ion and that the ATPase complex switches this dual fuel engine into a PMF-driven export machinery . It has been reported that the secretion level by the ΔfliH-fliI flhB ( P28T ) bypass mutant is remarkably dependent on the ΔpH component of PMF in 10 mM potassium buffer , namely in the absence of NaCl; it increases on a downward pH shift from 7 . 0 to 6 . 0 and almost diminished by an upward shift to 7 . 5 . Since external pH change affects the ion selectivity of the stator complex of the flagellar motor of alkalophilic Bacillus clausii , which utilizes both H+ and Na+ as the coupling ion [4] , we investigated whether external pH change influences Na+-dependent protein export by the ΔfliH-fliI flhB ( P28T ) bypass mutant . We varied the external pH over a range of 6 . 0 to 8 . 0 in the presence of 100 mM NaCl ( Fig 4A ) . The level of FlgD secreted by the ΔfliH-fliI flhB ( P28T ) bypass mutant gradually increased on an upward pH shift from 6 . 0 to 7 . 0 ( right panel , lanes 6–8 ) and then was almost constant over a range of 7 . 0–8 . 0 ( lanes 8–10 ) although the cellular level of FlgD was not changed significantly ( lanes 1–5 ) . In wild-type cells , the secretion level of FlgD was almost constant over this pH range ( left panel , lanes 6–10 ) . We next investigated the effect of Na+ concentration on FlgD secretion at external pH 6 . 0 ( Fig 4B ) . The secretion level of FlgD by the ΔfliH-fliI flhB ( P28T ) bypass mutant was significantly increased by adding of 100 mM NaCl ( right panel , lanes 3 and 4 ) , indicating that Na+ still enhances FlgD secretion by this bypass mutant at external pH 6 . 0 . This suggests that the transmembrane export gate complex still utilizes Na+ to drive flagellar protein export even when a significant pH gradient is present across the cell membrane . This raises the possibility that without FliH and FliI the export gate prefers to utilize Na+ rather than H+ . In contrast , the secretion level of FlgD by the wild-type showed no Na+ dependence even at external pH 6 . 0 ( left panel , lanes 3 and 4 ) . Therefore , we suggest that FliH and FliI allow the transmembrane export gate complex to become a much more robust export engine against environmental changes . An interaction between FliJ and FlhA brought about by FliH and FliI is responsible for efficient PMF-driven protein export [13 , 17] . Therefore , we investigated the effect of FliJ deletion on Na+-dependent flagellar protein export . The Na+ dependence of the protein export in a ΔfliH-fliI-fliJ flhB ( P28T ) mutant was not different from the ΔfliH-fliI flhB ( P28T ) strain , i . e . FlgD secretion levels increased with increasing external Na+ concentrations ( Fig 5A ) . Interestingly , the Na+ dependence of protein export in the absence of FliJ still remained even in the presence of FliH and FliI ( Fig 5B , right panel ) . In contrast , when FliH and FliI were expressed in the ΔfliH-fliI flhB ( P28T ) bypass mutant , there was no Na+ dependence ( Fig 5B , left panel ) . This analysis confirmed that the export apparatus does not use Na+ for flagellar protein export in the presence of the entire ATPase complex and that FliJ is the key factor for this mechanism . FlhA plays an important role in the energy transduction mechanism along with FliH , FliI and FliJ [13] . To test whether FlhA acts as an ion channel to conduct H+ and Na+ , we expressed a ratiometric pH indicator probe , pHluorin [24 , 25] , in E . coli cells to study multicopy effect of FlhA on intracellular pH change at an external pH value of 5 . 5 ( Fig 6A ) . The MotAB complex acts as a proton channel of the H+-driven flagellar motor , and Asp-33 of MotB is a critical proton-binding site [2] . Because a plug segment of the MotAB proton channel , consisting of residues 53 to 66 of MotB , suppresses premature proton leakage when MotAB is not assembled into the motor [26 , 27] , we used MotABΔplug and MotAB ( D33N ) Δplug as the positive and negative controls , respectively . In agreement with previous data [26 , 27] , the intracellular pH of the cells over-expressing MotABΔplug dropped by ca . 1 . 2 units in 60 min after induction with arabinose , and this intracellular pH value showed a statistically significant difference compared to that of the vector control ( P < 0 . 001 ) using two-tailed t-test . The intracellular pH of the MotAB ( D33N ) Δplug-expressing cells was measured to be 6 . 77 ± 0 . 07 , which was almost the same as the intracellular pH value of the vector control ( 6 . 80 ± 0 . 07 ) . Two-tailed t-test revealed no significant difference between these two intracellular pH values ( P = 0 . 51 ) . Intracellular pH of the FlhA-expressing cells was 6 . 66 ± 0 . 07 , which was ca . 0 . 1 pH unit lower than that of the vector control . This small pH drop showed a statistically significant difference compared to the vector control ( P = 0 . 02 ) . It has been shown that a well-conserved Asp-208 of FlhA , which is located in the cytoplasmic juxtamembrane region , is essential for FlhA function . Only the conservative D208E replacement permits any function , indicating that the important feature of this residue appears to be either the negative charge of the side-chain or the ability to bind proton [28] . To test whether the FlhA ( D208A ) substitution suppresses such a very small decrease in the intracellular pH by over-produced FlhA , we measured the intracellular pH of the FlhA ( D208A ) -expressing cells . Surprisingly , the intracellular pH value dropped by ca . 0 . 34 units in 60 min after induction of FlhA ( D208A ) with arabinose , and this intracellular pH value showed a statistically significant difference compared to that of the vector control ( P < 0 . 001 ) . The expression level of FlhA ( D208A ) was almost the same as that of wild-type FlhA ( S8 Fig ) . These results suggest that FlhA has an intrinsic H+ channel activity and that a highly conserved Asp-208 residue suppresses massive proton flow through the FlhA channel . To test if FlhA exhibits the Na+ channel activity , we analyzed the effect of overproduced FlhA on intracellular Na+ concentration change of FlhA-expressing E . coli cells using a fluorescent Na+ indicator dye , CoroNa Green ( Fig 6B ) . Because the PomAB stator complex of the marine Vibrio Na+-driven flagellar motor acts as a Na+ channel [2] , we used PomABΔplug as a positive control . The intracellular Na+ concentrations of the vector control were measured to be 4 . 21 ± 0 . 04 mM and 8 . 03 ± 1 . 21 mM in the absence and presence of 100 mM NaCl , respectively . The intracellular Na+ concentration of the PomABΔplug-expressing cells was increased from 12 . 3 ± 1 . 0 mM to 105 . 7 ± 6 . 8 mM by adding 100 mM NaCl . These results were in good agreement with previous reports [29 , 30] . Overexpression of FlhA caused a significant increment in the intracellular Na+ concentration in the presence of 100 mM NaCl but not in its absence . The intracellular Na+ concentration of the FlhA-expressing cells reached to 97 . 9 ± 14 . 7 mM , indicating that FlhA has the Na+ channel activity ( Fig 6B ) . Therefore , we propose that FlhA acts as a Na+ channel of the export gate complex . Interestingly , the FlhA ( D208A ) substitution did not affect the Na+ channel activity of FlhA at all ( Fig 6B ) . This raises the possibility that Asp-208 is not involved in the Na+ channel activity of FlhA . The PomA ( D148Y ) and PomB ( P16S ) mutations confer the phenamil-resistant motility phenotype on Vibrio cells , suggesting that the phenamil-binding sites are located in both PomA and PomB [23] . We found that the level of FlgD secreted by the ΔfliH-fliI flhB ( P28T ) bypass mutant was significantly reduced by 200 μM phenamil ( Fig 3 ) , raising the possibility that the phenamil-binding site could be located in FlhA . Therefore , we analyzed the effect of phenamil on the Na+ channel activity of FlhA ( Fig 6B ) . Addition of 200 μM phenamil to the PomABΔplug-expressing cells reduced the intracellular Na+ concentration by only about 2-fold . Since the swimming motility of Vibrio cells were totally inhibited by 50 μM phenamil [22 , 23] , the binding affinity of phenamil for the PomABΔplug complex not incorporated into the Vibrio motor appears to be much lower than that for the PomAB complex incorporated in the motor . In contrast to the PomABΔplug complex , 200 μM phenamil did not inhibit the Na+ channel activity of FlhA at all . It has been shown that phenamil dissociates from the Na+-driven Vibrio motor much faster in the presence of the PomA ( D148Y ) and PomB ( P16S ) mutations than in their absence , thereby conferring the resistance to phenamil [22] . Interestingly , these two mutations are predicted to be located in the cytoplasmic juxtamembrane regions of PomA and PomB [23] . Since 200 μM phenamil did not completely inhibited the Na+ channel activity of the PomABΔplug complex , we suggest that the inhibitory effect of phenamil is not a direct one to the Na+ channel of the PomAB complex . Therefore , we propose that phenamil may not directly bind to the Na+ channel of FlhA to reduce the secretion activity of the export gate complex or that the binding affinity of phenamil for free FlhA may be much lower than that for FlhA incorporated into the export gate complex as seen in freely diffused PomABΔplug complex .
PMF is the primary driving force for the flagellar and non-flagellar type III export apparatus [10] . The flagellar export gate of S . enterica is intrinsically a H+–protein antiporter that requires both the Δψ and ΔpH components to couple the energy of proton influx with protein export in the absence of the ATPase complex [13] . The cytoplasmic ATPase complex switches the export gate into a highly efficient , Δψ-driven protein export apparatus , and an interaction between FliJ and FlhA is key in driving this switch [13] . In this study , we showed that , in addition to PMF , the export gate can use SMF to drive flagellar protein export over an external pH range of 6 . 0–8 . 0 in the absence of FliH , FliI and FliJ ( Figs 1 , 4 and 5 ) . This suggests that without FliH , FliI and FliJ the export gate alone is a dual fuel export engine that can exploit both H+ and Na+ as the coupling ion ( Fig 7 ) . Interestingly , environmental changes significantly affected flagellar protein export by the ΔfliH-fliI flhB ( P28T ) but not that by wild-type cells ( Figs 1 and 4 ) . Therefore , we propose that the export apparatus is robust and has evolved to be able to maintain protein export activity against internal or external , genetic or environmental perturbations . To achieve this level of robustness the export gate has evolved to exploit both H+ and Na+ as the coupling ion rather than becoming an exclusive PMF or SMF dependent machine . FlhA , which consists of an N-terminal integral membrane domain with eight predicted transmembrane helices ( FlhATM ) and a C-terminal cytoplasmic domain ( FlhAC ) [31] , forms a nonameric ring structure in the export apparatus [32 , 33] . FlhAC not only acts as a docking platform for FliH , FliI , FliJ , export substrates and chaperone-export substrate complexes [13 , 34–38] but also plays important roles in the energy coupling mechanism of flagellar type III protein export [13 , 17] . In this study , we showed that overexpression of FlhA resulted in a significant increment in the intracellular Na+ concentrations as seen in the PomAB Na+ ion channel complex , which works as the stator of the Na+-driven flagellar motor of marine Vibrio ( Fig 6B ) . However , when FlhA was overproduced , only a very small decrease in intracellular pH was observed in the FlhA-overexpressing cells ( Fig 6A ) . If overexpression of FlhA non-specifically perturbed the cell membrane , both H+ and Na+ would have leaked into the cell through the membrane , thereby increasing the intracellular concentrations of both H+ and Na+ considerably . Therefore , we conclude that FlhA has an intrinsic Na+ channel activity . Interestingly , neither Na+ nor Na+ channel blockers affected protein export by wild-type cells ( Figs 2 and 3 ) , indicating that the Na+ channel of FlhA is kept in a closed state in the presence of FliH , FliI and FliJ . Therefore , we propose that the intrinsic Na+ channel activity of FlhA may provide the cell with a genetic backup to rapidly compensate the occasional loss or inactivation of the ATPase complex during flagellar assembly . A highly conserved Asp-208 of FlhA is essential for PMF-driven flagellar protein export [28] . The FlhA ( D208A ) substitution results in a loss-of-function phenotype [28] . Here , we found that the intracellular pH decreased by about 0 . 34 units in 60 min after induction of FlhA ( D208A ) with arabinose whereas the intracellular pH of the cells expressing wild-type FlhA decreased by about 0 . 1 unit ( Fig 6A ) . The D208A mutation did not affect the expression level of FlhA at all ( S8 Fig ) . These results indicate that overexpression of FlhA ( D208A ) causes massive proton leakage through its proton channel , thereby inhibiting cell growth . Therefore , we propose that FlhA also has the intrinsic ability to conduct H+ . Since Asp-208 of FlhA is predicted to be located in the cytoplasmic juxtamembrane region [28] , we propose that this Asp residue plays a regulatory role in coordinated proton flow through the FlhA proton channel coupled with protein export . Interestingly , the D208A did not affect the Na+ channel activity of FlhA at all ( Fig 6B ) , raising the possibility that the Na+ pathway in FlhA could be distinct from the H+ pathway . Based on all available information , we propose that FlhA is an energy transducer of the export apparatus for flagellar protein export . In the absence of FliH , FliI and FliJ , Na+ ions still showed a positive impact on flagellar protein export by the export gate even at an external pH value as low as 6 . 0 ( Fig 4B ) . Although there is a significant pH gradient across the cytoplasmic membrane under this condition , the export gate prefers to use the Na+ gradient over the H+ gradient . This could explain why the ΔfliH-fliI flhB ( P28T ) bypass mutant requires the ΔpH component for flagellar protein export in addition to Δψ and why depletion of the ΔpH component and D2O significantly reduce the rate of protein export by this bypass mutant [13] . In the presence of FliH , FliI and FliJ , the export gate used only PMF , suggesting that the Na+ channel of FlhA is closed by the binding of the cytoplasmic ATPase complex to the gate . Because the intrinsic H+ channel activity of FlhA is quite low ( Fig 6A ) , we propose that the cytoplasmic ATPase complex may allow FlhA to conduct H+ more efficiently so that proton influx is not limiting the rate of protein export . FliI is the ATPase of the export apparatus [39] and forms a homo-hexamer to exert its ATPase activity [40] . FliJ binds to the center of the FliI6 ring to form the FliI6FliJ ring , which is structurally similar to F-type and V-type ATPases [41] . FliH connects the FliI6FliJ ring with the export gate complex through an interaction of FliH and FlhA [42] . ATP hydrolysis by FliI ATPase activates the export gate through an interaction between FliJ and FlhA , allowing the gate to transport flagellar proteins in a PMF-dependent manner [13 , 16 , 17] . Therefore , we propose that FliJ acts as a switch of the energy transducer to change the ion channel properties of FlhA from a dual ion channel mode to a H+ channel mode ( Fig 7 ) .
Salmonella strains and plasmids used in this study are listed in Table 1 . T-broth ( TB ) contained 1% Bacto tryptone , 10 mM potassium phosphate pH 7 . 5 . Ampicillin and chloramphenicol were added at a final concentration of 100 μg/ml and 30 μg/ml , respectively , if needed . The cells were grown with shaking in 5 ml of TB with or without various concentrations of NaCl , LiCl , KCl or MgCl2 at 30°C until the cell density had reached an OD600 of ca . 1 . 4–1 . 6 . To see the effect of removal of Na+ on Na+-dependent protein export by the ΔfliH-fliI flhB ( P28T ) mutant cells , the cells were grown with shaking in 3 ml of TB ( pH 7 . 5 ) with or without 100 mM NaCl at 30°C until the cell density had reached an OD600 of ca . 0 . 8–1 . 0 . After washing twice with TB ( pH 7 . 5 ) , the cells were resuspended in 3 ml TB with or without 100 mM NaCl and then incubated at 30°C for 1 hour . To test the effects of phenamil and EIPA on flagellar protein export , the cells were grown with shaking in 5 ml of TB containing 100 mM NaCl at 30°C until the cell density had reached an OD600 of ca . 1 . 0–1 . 2 . After washing the cells twice with TB containing 100 mM NaCl , the cells were resuspended in the 5 ml TB with 100 mM NaCl in the presence of various concentrations of phenamil or EIPA and incubated at 30°C for 1 hour . Cultures were centrifuged to obtain cell pellets and culture supernatants . Cell pellets were resuspended in the SDS-loading buffer , normalized to a cell density to give a constant amount of cells . Proteins in the culture supernatants were precipitated by 10% trichloroacetic acid , suspended in the Tris/SDS loading buffer and heated at 95°C for 3 min . After SDS-PAGE , immunoblotting with polyclonal anti-FlgD , anti-FlgE , anti-FliK , anti-FlgK or anti-FlgL antibody was carried out as described before [43] . Detection was performed with an ECL plus immunoblotting detection kit ( GE Healthcare ) . At least three independent experiments were carried out . Overnight culture of Salmonella cells was inoculated into fresh TB with 100 mM NaCl and incubated at 30°C with shaking for 4 hours . The cells were washed twice with TB and resuspended in TB with or without various concentrations of NaCl , LiCl , KCl or MgCl2 . To test the effects of phenamil and EIPA on free-swimming motility , the cells were resuspended in TB containing 100 mM NaCl in the presence of various concentrations of phenamil or EIPA . The swimming speed of individual motile cells was measured under a phase contrast microscopy at room temperature as described before [44] . The flagellar filaments produced by Salmonella cells were labelled using polyclonal anti-FliC antibody and anti-rabbit IgG conjugated with Alexa Fluor 594 ( Invitrogen ) as described [16] . The cells were observed by fluorescence microscopy as described previously [45] . Fluorescence images were analysed using ImageJ software version 1 . 48 ( National Institutes of Health ) . The membrane potential was measured using tetramethylrhodamine methyl ester ( Invitrogen ) as described before [13] . Intracellular pH measurements with a ratiometric fluorescent pH indicator protein , pHluorin [24 , 25] , were carried out as described before [27] . Salmonella SJW1103 and MMHI0117 strains were transformed with the pRGXX::cat series [46] . The cells were grown with shaking in 5 ml of T-broth with or without 100 mM NaCl at 30°C until the cell density had reached an OD600 of ca . 1 . 0–1 . 2 . The cultures were then pipetted ( 200 μl ) into a 96–well microplate ( Greiner Bio-One ) . Bioluminescence and absorbance of cultures were measured using 2030 ARVO X microplate reader ( Perkin Elmer ) at 30°C . All microplate assays were repeated four times . Promoter activities were calculated as the value for bioluminescence intensities divided by absorbance value after background correction . The E . coli BL21 ( DE3 ) strain was transformed with a pBAD24-based plasmid . The resulting transformants were grown in TB ( pH 7 . 0 ) at 30°C for 4 hours . The protein expression was induced by addition of 0 . 2% arabinose . After 1 h , the cells were washed three times with TB , resuspended in TB ( pH 7 . 0 ) containing 40 μM CoroNa Green ( Invitrogen ) and 10 mM EDTA and incubated in the dark room for 60 min at room temperature . Then , the cells were washed three times with TB to remove excess CoroNa Green and resuspended in TB with or without 100 mM NaCl . To observe epi-fluorescence images , we used an inverted fluorescence microscope ( IX-73 , Olympus ) with a 100× oil immersion objective lens ( UPLSAPO100XO , NA 1 . 4 , Olympus ) and an sCMOS camera ( Zyla4 . 2 , Andor Technology ) . Epi-fluorescence of CoroNa Green was excited by a 130 W mercury light source system ( U-HGLGPS , Olympus ) with a fluorescence mirror unit U-FGFP ( Excitation BP 460–480; Emission BP 495–540 , Olympus ) . Fluorescence images of CoroNa Green were captured at every 100 msec exposure . Fluorescence image processing was performed with the ImageJ version 1 . 48 software ( National Institutes of Health ) . To quantify the fluorescence intensity of each cell , integral fluorescence of CoroNa Green was measured and then the intensity of a nearby cell-less region was subtracted as the background intensity . To calibrate the intracellular sodium concentration , fluorescence intensity of the cells with CoroNa Green were measured at various sodium concentrations in TB containing 20 μM gramicidin and 5 μM carbonyl cyanide 3-chlorophenylhydrazone ( CCCP ) as described before [30] . All experiments were performed at 23°C . Statistical analyses were done using StatPlus::mac software ( AnalystSoft ) . Comparisons were performed using a two-tailed Student’s t-test . A P value of < 0 . 05 was considered to be statistically significant difference . * , P < 0 . 05; ** , P < 0 . 01; *** , P < 0 . 001 . | For construction of the bacterial flagellum beyond the inner and outer membranes , the flagellar type III export apparatus transports fourteen flagellar proteins with their copy numbers ranging from a few to tens of thousands to the distal growing end of the flagellar structure . The export apparatus consists of a transmembrane export gate complex and a cytoplasmic ATPase complex . Here , we show that the export engine of the flagellar type III export apparatus is robust in maintaining its export activity against internal and external perturbations arising from genetic variations and/or environmental changes . When the cytoplasmic ATPase complex is absent , the export gate complex is able to utilize sodium motive force ( SMF ) across the cytoplasmic membrane as a fuel in addition to proton motive force ( PMF ) . However , the export gate utilizes only PMF as the energy source when the ATPase complex is active . An export gate protein FlhA shows an intrinsic ion channel activity . These observations suggest that the export gate intrinsically uses both PMF and SMF for protein export and that the ATPase complex switches the export gate into a highly efficient PMF-driven export engine to become much more robust against environmental perturbations . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"protein",
"transport",
"protons",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"engineering",
"and",
"technology",
"molecular",
"probe",
"techniques",
"pathogens",
"vibrio",
"cell",
"processes",
"immunoblotting",
"microbiology",
... | 2016 | The Bacterial Flagellar Type III Export Gate Complex Is a Dual Fuel Engine That Can Use Both H+ and Na+ for Flagellar Protein Export |
Despite the current debate about the computational role of experimentally observed precise spike patterns it is still theoretically unclear under which conditions and how they may emerge in neural circuits . Here , we study spiking neural networks with non-additive dendritic interactions that were recently uncovered in single-neuron experiments . We show that supra-additive dendritic interactions enable the persistent propagation of synchronous activity already in purely random networks without superimposed structures and explain the mechanism underlying it . This study adds a novel perspective on the dynamics of networks with nonlinear interactions in general and presents a new viable mechanism for the occurrence of patterns of precisely timed spikes in recurrent networks .
Patterns of spikes that are precisely timed within the millisecond range have been investigated and observed in a series of neurophysiological studies [1]–[9] . This supports the ongoing debate whether cortical neurons are capable of precisely coordinating the timing of their action potentials across recurrent networks and whether only the neurons' firing rate or also the precise timing of their spikes encode key information that is intimately related to external stimuli and internal events [2] , [3] , [10]–[14] . During the last two decades , a branch of theoretical research has focused on the question how such precise timing could emerge . One prominent , possible explanation for the occurrence of precisely coordinated spiking is the existence of excitatorily coupled feed-forward structures , ‘synfire-chains’ , which are superimposed on a network of otherwise random connectivity , e . g . through strongly enhanced synaptic connectivity [10] , [15]–[18] . Under certain conditions , these additional feed-forward structures enable the persistent propagation of groups of spiking activity that is synchronous on a time scale of down to one millisecond [17] , [19]–[24] . So far , however , experimental research did not provide anatomical evidence for such structures . Other studies proposed that asynchronous propagation along paths with matching inhomogeneous delays [25] or the dynamics of local recurrent networks [26] , [27] might underlie precisely timed spike patterns . Here we show that nonlinear dendritic interactions , recently uncovered in neurophysiological experiments , offer a viable mechanism to support stable propagation of synchrony through random cortical circuits without additionally superimposed structures: Excitatory synaptic stimuli may not only superimpose linearly or sublinearly [28] , [29] , but may also induce strongly nonlinear , supra-additive coupling enhancement due to dendritic spikes [30]–[32] . Fast dendritic sodium spikes strongly enhance the effects of stimulus-evoked post-synaptic potentials in a supra-additive way and induce precisely timed and sharply peaked depolarizations in the somatic membrane potential . Remarkably , this enhancement occurs reliably only if the stimuli are synchronous in time with temporal difference of less than [33]–[36] , cf . also [37] . If the resulting depolarization triggers an action potential , it is highly precise in time up to less than ms [33]–[35] . Other types of much slower dendritic spikes are mediated by voltage gated or NMDA channels . They have longer time courses up to several hundreds of milliseconds and do not depend on synchronous stimulation ( see , e . g . , [38] , [39] , and , for reviews , [32] , [40] ) . In the following , we study consequences of coupling nonlinearities that are due to fast dendritic spikes onto the collective dynamics of recurrent neural networks . We find that , in contrast to linearly coupled networks , propagating synchronous activity may persist already in networks of simple neurons that have purely random connectivity and exhibit no additional structures . We conclude that the characteristic features of dendritic nonlinearity , in particular the amplification of ( only ) synchronous input and the induction of temporally precise output , predestine them to support the generation and propagation of persistent , highly synchronous spiking activity .
We investigate networks of integrate-and-fire neurons in the limit of fast response to incoming spikes and with nonlinear interactions ( see Methods ) . Similar models with linear interactions are widely used for studying the dynamics of networks of spiking neurons ( see , e . g . , [16] , [41]–[43] , [44] , [45] for recent reviews ) because they capture essential features of cortical neurons and at the same time allow to investigate the mechanisms underlying the dynamics of networks without obscuring them by a many-parameter , many-variable single neuron description ( see , e . g . , [44] , [46]–[48] ) . In this study they allow to interpret the dynamical regimes of the network activity qualitatively and to analytically assess them quantitatively . We assume that the delay between sending of a spike by a presynaptic neuron and postsynaptic ( somatic ) response is identical for all neurons . This is appropriate for the description of responses mediated by fast dendritic spikes because these evoke a fast and precise rise with sub-millisecond rise time constant in the somatic potential [33] , [36] . Moreover , if a somatic action potential is generated by fast dendritic spikes as observed in [33] , this occurs after presynaptic axonal stimulation with only sub-millisecond inter- and intra-neuronal jitter , while the action potential timing strongly varies in time if no dendritic spike is elicited . This is well resembled by our model dynamics where nonlinearly enhanced inputs yield fast , jump-like responses in the membrane potential and firing due to supra-threshold excitation occurs precisely after the delay time . For simplicity , we further assume that all postsynaptic responses to spikes occur after this delay time . ‘Imprecise’ spiking is generated due to a constant supra-threshold input current . To account for nonlinear enhancement and saturation of synchronous excitatory inputs , we modulate the linear sum of the amplitudes of excitatory post-synaptic potentials ( EPSPs ) that arise simultaneously from different synapses by a nonlinear function . This covers the main features of experimentally found nonlinear dendritic amplification ( cf . [33] , [36] , [38]–[40] ) , thus effectively modeling a neuron with one , nonlinear dendrite . For the neuron model considered , has a straightforward interpretation: It maps the peak EPSP amplitude expected from linearly adding the coupling strengths of synchronously received excitatory signals to the actual value ( cf . Fig . 1b ) . Such a modulation function has been directly [39] and indirectly [33] measured in experiments . It has a sigmoid shape , with linear summation for small summed amplitudes and saturation at high . We thus model the non-additive coupling using a function that is the identity at low values , has a constant saturation at high values , and linearly interpolates in between , cf . Fig . 1b . Inhibitory post-synaptic potentials ( IPSPs ) at the same neuron are linearly summed , independent on whether or not the synaptic signals are simultaneous , because there is no experimental evidence for supra-linear enhancement . If is the identity function ( Fig . 1a ) , the same holds for excitatory coupling and we recover a “conventional” network of linearly coupled neurons . In both additively and non-additively coupled sparse random recurrent networks , asynchronous irregular spiking activity constitutes a dynamical state typical for a wide range of parameters [42] , [43] , [49] , [50] . Sequences of groups of synchronously spiking neurons may spontaneously occur starting with a single neuron , or they can be initiated by a group of neurons that was excited to synchronous spiking by external input . If a single neuron or a group of neurons send spikes at one given time , a subset of neurons in the network will receive a synchronous pulse of spikes a delay time thereafter . All neurons for which the induced postsynaptic response leads to a supra-threshold depolarization in turn spike simultaneously so that another synchronous pulse of spikes is generated which can excite a further group of neurons and so on . Spontaneous chains are part of the background activity . They usually involve only small numbers of synchronously spiking neurons and quickly extinguish , cf . supporting Fig . S1 . How does a sparse random network respond to induced synchronous activity , initiated , e . g . , by external stimuli ? We compared the responses in networks with purely linear , additive coupling to those where the excitatory inputs cooperate supra-additively . For linearly coupled networks we find that pulse sizes in chains of synchronous spiking activity quickly reduce to the level of spontaneous synchronization and the chains rapidly die out ( cf . Fig . 2a ) . Propagation of synchrony is therefore short-lived in linearly coupled networks , consistent with previous studies [16] , [17] , [51] . In contrast , for nonlinearly coupled networks , in a wide range of parameters ( cf . Fig . 3 ) , a chain initiated by a large enough , but not too large synchronous group after a few steps reaches pulse-sizes that fluctuate around some typical value , Fig . 2b . These sizes are substantially larger than the sizes of synchronous pulses occurring in the background activity ( cf . Fig . S1b ) , which persists while synchrony is propagating on top of it . Only if the initial group size is too large , the chain of synchronous activity is again short-lived . Taken together , we find persistent propagation of synchrony in non-linearly coupled networks . Persistent propagation of synchrony is robust against parameter changes . We estimate a range of coupling strengths where persistent propagation of synchrony occurs in linearly and in nonlinearly coupled networks in Fig . 3 . Background activity is here considered stable if it contains at no time any synchronous pulse of more than of the network size ( red coloring if it became unstable spontaneously , i . e . before initiation of synchronous activity , yellow coloring if it became unstable thereafter , cf . also supporting Fig . S1 ) . Propagation of synchrony is considered persistent if background activity is stable and if at least synchronized groups within the chain are distinguishable from background activity , i . e . the minimal group size , , is larger than the largest group size occurring in background activity ( green coloring for stable background activity but short-lived propagation of synchrony , blue coloring for stable background activity and persistent propagation of synchrony ) . In nonlinearly coupled networks , propagation of synchrony is persistent in a wide range of parameters , while it is usually short-lived in linearly coupled networks . The mechanisms underlying this persistent propagation of synchrony can be intuitively understood . Sequences with small groups of synchronized neurons behave as for linear , additive coupling , i . e . they usually extinguish after a few steps , so there is no persistent spontaneous propagation and irregular background dynamics for the entire network is stable . If larger groups of neurons send spikes simultaneously , their postsynaptic neurons receive sufficiently many excitatory inputs so that the nonlinearities become effective . Since the inhibitory couplings add only linearly , excitatory input surpasses inhibitory input for a larger fraction of postsynaptic neurons than in a linearly coupled network . This causes more neurons to fire in response to the synchronous pulse; the number of neurons synchronized in each step of the chain grows . If synchronous pulses become too large , saturation becomes important and excitation becomes less efficient compared to inhibition . Further , many neurons are refractory . This implies that less neurons are excited in response to overly large groups of synchronously spiking neurons; consequently the group size is reduced . In addition , fluctuations in groups sizes occur due to the randomness of the network connectivity and the distribution of membrane potentials during pulse reception . These qualitative mechanisms keep the group sizes substantially large and fluctuating within a certain range . To quantitatively understand the mechanisms underlying persistent propagation of synchrony and to determine the group sizes which initiate and take part in persistent propagation , we studied the evolution of propagating synchrony both analytically and numerically ( see Methods and Fig . 4 ) . Approximating the dynamics of group sizes by a Markov process , we derived the transition probabilities for the transitions from the sizes of the th pulse to those of the . Here , , are random variables that assume values in , where is the number of neurons in the network . Accordingly , is the probability that the th pulse generated by simultaneously spiking neurons causes a group of neurons to spike simultaneously in response . From the conditional ( transition ) probabilities , we derived the conditional expectation , i . e . the average size of a pulse following a pulse of size . Since the distributions are similar to also for later stages , we assume stationarity and approximate and for all stages of propagation . The points , , where for and , , determine the range of typical group sizes occurring in the networks ( Fig . 4 ) . The analytical predictions agree well with the numerical results . The quantities and yield a quantitative explanation of the mechanisms that lead to persistent propagation of synchrony: For networks of linearly coupled neurons , each synchronous group with neurons ( small , e . g . in Fig . 4a ) on average excites synchronous groups with less neurons . The smaller groups in turn excite even smaller groups so that synchronous activity rapidly decays to the level of a few synchronized neurons and fluctuates near . Thereafter , due to the fluctuations from the already small group size , propagating synchronous activity rapidly extinguishes completely ( group size zero ) . So the theory predicts that in networks of linearly coupled neurons the chain of synchronous activity quickly extinguishes even if excited by external synchronous input , consistent with the above observations ( Fig . 2a ) . Since the shape of the transition matrix stays invariant when network parameters like the coupling strengths are changed , such a change will not lead to persistent propagation of synchrony . If , e . g . , the size of excitatory coupling strength is increased , only the slope of the curve is increased . This predicts the transition to unstable background activity shown in Fig . 3 . In contrast , nonlinear supra-additive excitatory coupling enables persistent propagation of activity with a substantial number of neurons synchronized . The sizes of the propagating synchronous pulses are of the order of a typical size and range between and , all of which are substantially larger than ( cf . Fig . 4b ) . Pulses of sizes between and usually evoke pulses of sizes in the same range , i . e . between and again . Only rarely , propagating synchronized activity becomes smaller than or larger than ; if so , the pulse size is likely to stay smaller than for longer , decay even further as for linearly coupled networks , and the chain may cease to exist . A steeper and narrower peak can lead to transiently increased activity and short-lived propagation of larger synchronous groups [51] . The different dynamics for linearly and nonlinearly coupled networks can also be understood by approximating the stochastic dynamics by a deterministic iterative map derived from interpolating between the values of . For networks of linearly coupled neurons , the map has only one stable fixed point which is at small pulse sizes of the order of spontaneous synchronization; it may be distinct from the trivial fixed point zero . Any larger initial pulse size will thus lead to a chain decaying to the level of spontaneous synchronization . If coupling is non-additive , there can be two stable fixed points and , and an unstable fixed point in between . Chains starting with sizes in the basin of between and then evolve towards stable propagation with pulse-size . For different parameter settings , stable propagation of synchrony is supported by a stable periodic orbit close to an unstable fixed point . Taken together , the theory for nonlinearly coupled networks predicts persistent propagation of synchronous activity in a typical range of pulse sizes and a decay that is possible only due to fluctuations . This agrees with the numerical observations ( Fig . 2b ) .
In summary , we presented a theoretical analysis and numerical simulations of recurrent networks of spiking neurons with nonlinear dendritic interactions . The results indicate that networks with nonlinear dendritic interactions are capable of generating persistent propagation of synchronous spiking activity even if the network is purely randomly connected and has no additional structural features . Theoretical studies on active dendrites mainly considered single neurons . Simulations of neuron models with detailed channel density and morphology showed dendritic spike generation in agreement with neurobiological experiments [33] , [34] , [36] , [38] . For neurons with slow dendritic spikes , which are largely insensitive to temporal coincidence of inputs , firing rate models have been developed [52] . They reproduce the response properties of detailed models to diverse stimuli and possess computational capabilities comparable to multi-layered feed-forward networks of simple rate neurons [38] , [39] . Based on this result , the computational abilities of simple circuits have been considered , also with other types of neuron models ( e . g . [32] , [53] , [54] ) . Refs . [55] , [56] studied propagation of bursts in networks where the bursts can be explained by slow dendritic spikes , and slow nonlinear dendrites were suggested to underlie the persistent activity observed in working memory tasks [57] . Active dendrites generating fast dendritic sodium spikes were studied in a two-neuron circuit and in a simple feed-forward structure [58] , and model neurons incorporating such dendritic spikes were used as an output layer in simulations of hippocampal network models [59] . Very recently , ref . [51] has shown that fast dendritic spikes can lead to intermittent , transiently increased propagation of synchrony and it was suggested that they underlie hippocampal sharp wave/ripples characteristic for slow wave sleep . The present study now shows that fast dendritic spikes can lead to persistent propagation of synchrony in random neural networks . In particular , feed-forward structures based on large-scale additional couplings [10] , [15] , [16] or strongly and systematically adapted strengths of specific synapses and neuron properties [17] may not be needed . As such , our results suggest an alternative mechanism and a potential complementary explanation for the occurrence of patterns of precisely timed spikes [1]–[5] , [7]–[9] . Our study uses a model that is appropriate for quantitative numerical analysis of larger networks and at the same time allows analytical predictions that yield further insights into the dynamics of recurrent networks . The theoretical predictions made are based on mean field arguments , strictly valid only in the limit of infinite network size [42] , [49] , [60] . As our results indicate , these predictions are in good agreement with simulation data already for networks of finite size . The number of neurons participating in pulses of synchronous activity as well as their number relative to the total number of neurons may vary strongly with network features such as the connectivity and the effective total input coupling strengths . Additional external noise , e . g . due to further random spiking inputs , is expected to be beneficial because it stabilizes background activity and leads to a fast equilibration of the neurons' potentials after a synchronous pulse . Both facts support dynamical mixing and thus are in favor of our approximation that the propagation of synchronous activity does not further influence the statistics of the background . We have demonstrated that nonlinear dendritic interactions enable persistent propagation of synchrony even in random neural networks . The results show that the nonlinear interactions are in fact the main ingredient controlling the mechanism underlying the transition to persistent propagation ( Fig . 4a vs . 4b ) , so that the phenomenon is insensitive against variations in parameters such as details of the individual neuron dynamics , the exact form of nonlinearly modulated interactions ( Fig . 1 ) , and the coupling strengths ( see Fig . 3 ) . The current study contributes to a new field of research that focuses on neural networks with supra-additive coupling . The influence of different levels of individual neuron reliability , of recurrent and feed-forward network topologies , of dynamic connectivity ( learning ) and of slow dendritic spikes have to be reconsidered in this context . Our study also suggests future experiments on the propagation of synchrony due to nonlinear dendritic interactions e . g . in cultured neurons [61] . Interestingly , the propagation of synchrony found here for nonlinearly interacting neurons does not follow any specific , predefined propagation paths of synchronous activity across the network; the propagation path will depend not only on the currently excited group but also on which neurons in the background activity are sufficiently depolarized when they receive synchronous spikes from the current group . In a random network , the propagation of synchrony will thus resemble reverberating high-frequency oscillations involving highly synchronous spiking activity . The network structure might shape the activity and lead to a significantly enhanced occurrence of specific sequences of synchronous groups . These spike patterns , however , are noisy and less obvious than those in synfire-chains [10] , [15]–[17] , [19]–[21] , [23] , [24] , where the propagation paths of synchronous activity are predefined by the embedded feed-forward networks . These different dynamics may provide an experimentally testable distinction between synchronous events created by synfire chains via additional feed-forward structures and those created by nonlinear dendritic interactions in largely or purely random networks . Of course , a more specifically structured network connectivity [62]–[64] , the effects of synaptic location on different dendritic branches [39] , specific distributions of transmission delays [25] , [65]–[67] as well as strongly heterogeneous synaptic strengths [17] will further influence pulse propagation . As an example , nonlinear interactions may facilitate or enable localized persistent synchrony in Hebbian cell assemblies [18] , [68] , [69] . It will thus be important to extensively investigate to which degree nonlinear interactions as well as non-random network structure are contributing to creating collectively coordinated spiking dynamics , in order to understand the computational capabilities of cortical networks .
We considered networks of leaky integrate-and-fire neurons connected to form an Erdös-Rényi random graph [70] where each directed synaptic connection between two neurons is present independently with probability . For each connection , the probabilities and specify whether the coupling is excitatory or inhibitory . The dynamics of the membrane potential of neuron obeys ( 1 ) where denotes times at which spike are sent within the network , the inverse membrane time constant measures the dissipation of the neuron and is the transmission delay . We further introduced the set ( 2 ) of neurons sending at time an excitatory spike to neuron , where is the th spike time of neuron and is the coupling strength from neuron to neuron . The set ( 3 ) lists the neurons sending at time an inhibitory spike to neuron . is the possibly nonlinear dendritic modulation function mapping the input strength expected from linear addition of excitatory inputs to the actual input strength . Each neuron receives some constant external input . When the membrane potential reaches or exceeds the threshold , , where is the possibly arriving total input at time , it is reset to and a spike is emitted . See supporting Table S1 for a tabular description of our model following ref . [71] . The parameters used in the given examples are for the onset of supra-additivity , for the onset of saturation and for the level of saturation , in agreement with a direct experimental measurement of given in [39] for slow nonlinear interactions . In [33] , the onset of nonlinearity and the level of saturation lie higher . For comparison with linearly coupled networks , we take an identity modulating function , effectively choosing , i . e . there is no supra-additivity and no saturation . The analytical methods presented below and the theory presented in the main text are valid for arbitrary parameter choices and hold as long as the background activity stays asynchronous , irregular and sufficiently uncorrelated . In the simulations , the remaining network parameters are , , , , , , , . If not stated otherwise , , if the coupling strength from neuron to neuron is excitatory and , if it is inhibitory . Network simulations were done in phase representation [72] . For this , the membrane potential and its threshold are mapped one-to-one to a phase and a phase-threshold using the inverse of the transfer function of the leaky integrate-and-fire neuron , as elaborated in ref . [27] . evolves linearly with slope between spike sendings and spike receivings . Spike sendings occur when the phase reaches or exceeds its threshold . When neuron receives input of total strength at time , its phase is updated according to , where is the response function of the leaky integrate-and-fire neuron , for subthreshold total inputs and for suprathreshold ones which evoke spike sending . The numerical simulations were implemented using an event based algorithm which may be outlined as follows [41] , [50] , [73] , [74]: We keep track of the “pseudo-spike time” [75] of each neuron , i . e . of the time remaining to the next hypothetical spike of the neuron without interaction . Further , we keep track of the spike arrival times together with the neurons that sent the spikes . In each step , the smallest pseudo-spike time is compared with the time remaining until the next spikes arrive . If the next event is ( i ) a spike sending event , the dynamics is linearly evolved to this event and the pseudo-spike time of each sending neuron is reset to . The newly sent spikes are stored in the spike list . If the next event is ( ii ) a spike receiving event , the dynamics is linearly evolved to this event and the excitatory and inhibitory input strengths to each neuron are determined . We apply to the excitatory input strength and add the inhibition . The resulting total input strength determines the update of the phase via and therewith the new pseudo-spike time as well as immediate spiking responses . For the spike-train analysis , propagating chains initiated at some time can be separated from background activity because synchronized groups which are part of the chain by construction send spikes precisely at , , while spikes which are part of background activity are sent at times which are at least slightly different . Fig . 4 shows the numerically derived frequency of occurrence of a group size when the initial group had size and its mean value , which are approximations to the conditional probability and the conditional expectations , respectively . For the numerical measurements , synchronous pulses of size were initiated twice after equilibration of the dynamics ( initial phases were randomly drawn from a uniform distribution on where is the phase threshold , and random initial spikes were added ) in different random networks and the size of the subsequent pulse was measured . Fig . 2 shows two single simulations with . For Fig . 3 , the mean total input strengths of the excitatory and the inhibitory input were varied in steps of by changing and , from ( corresponding to ) to ( ) and from ( ) to ( ) . For each data point , the stability of background activity and the persistence of propagating synchrony was checked in different random networks with different random initial conditions , initial phases were drawn from a uniform distribution on where is the phase threshold , and random spikes initially in transit were added . The stability of background activity without propagating synchrony was checked for simulated time , where . At , synchronous activity was initiated by external stimulation of a group of neurons . Stability of propagating synchrony was checked for steps after initiation ( corresponding to of propagation ) and stability of background activity after was checked for an interval of after pulse initiation . We note that for stable irregular background activity finally ( for time tending to infinity ) every chain will die out with probability one , because the group size has finite probability to leave the zone of propagation and to reach the absorbing fixed point zero . We implemented the network dynamics simulations in C and embedded them with MathLink into Mathematica . We used Mathematica to implement user interfaces , control programs and data analysis . We computed the transition probabilities for the group-sizes analytically and semi-analytically . In the analytical approach , the probability distribution for the membrane potentials was derived in diffusion approximation , also approximating the actual number of synaptic connections by its mean and describing the background activity as consisting of independent Poissonian spike trains [42] , [44] . To eliminate errors due to these approximations in a semi-analytical approach , was derived by direct measurements of the relative frequency of occurrences of membrane potentials at different times in numerical network simulations , simulations in different random networks with different random initial conditions as described above . In both approaches , we computed from the cumulative probability distribution from the right , ( 4 ) which yields the average probability that a neuron is excited above threshold when it receives an input of strength . We further assumed ( a ) that previous groups with do not influence , i . e . the sequence of group sizes is a realization of a Markov chain , ( b ) that the propagating synchrony does not change the statistics of the background dynamics of the non-participating neurons , and ( c ) that neurons which spiked in the th step are refractory while the other neurons are equilibrated at the time of the pulse . The validity of the approximations depends on the network parameters and was checked by numerical simulations . Under these assumptions , the statistical properties of the neural network topology allow to compute the probabilities that a neuron receives an input of strength at time under the condition that a synchronized group of size has sent spikes simultaneously at time . Together with , the conditional probability distributions and the conditional expectations can be derived . , the probability that a group size occurs in response to a group size , follows a binomial distribution , ( 5 ) where ( 6 ) is the probability that a neuron spikes in response to a synchronous pulse of spikes . is the total input strength due to excitatory and inhibitory inputs and and are the strengths of excitatory and inhibitory connections . According to Eq . ( 5 ) , , the average next group size given a current group size of , is ( 7 ) as derived from the diffusion approximation and from the semi-analytical approach is illustrated in Fig . 4 for linearly and nonlinearly coupled networks . The values agree well with the results of the explicit numerical measurements , deviations are due to the specified approximations . The critical pulse-sizes , and are the intersection points of the interpolated -values with the diagonal , denotes the size , where the interpolated -values equal . If present , and roughly bound the pulse-sizes in persistently propagating chains of synchronous activity . | Most nerve cells in neural circuits communicate by sending and receiving short stereotyped electrical pulses called action potentials or spikes . Recent neurophysiological experiments found that under certain conditions the neuronal dendrites ( branched projections of the neuron that transmit inputs from other neurons to the cell body ( soma ) ) process input spikes in a nonlinear way: If the inputs arrive within a time window of a few milliseconds , the dendrite can actively generate a dendritic spike that propagates to the neuronal soma and leads to a nonlinearly amplified response . This response is temporally highly precise . Here we consider an analytically tractable model of spiking neural circuits and study the impact of such dendritic nonlinearities on network activity . We find that synchronous spiking activity may robustly propagate through the network , even if it exhibits purely random connectivity without additionally superimposed structures . Such propagation may contribute to the generation of spike patterns that are currently discussed to encode information about internal states and external stimuli in neural circuits . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"computational",
"neuroscience",
"biology",
"neuroscience"
] | 2012 | Non-Additive Coupling Enables Propagation of Synchronous Spiking Activity in Purely Random Networks |
Microglia are the immune cells of the brain . In the absence of pathological insult , their highly motile processes continually survey the brain parenchyma and transiently contact synaptic elements . Aside from monitoring , their physiological roles at synapses are not known . To gain insight into possible roles of microglia in the modification of synaptic structures , we used immunocytochemical electron microscopy , serial section electron microscopy with three-dimensional reconstructions , and two-photon in vivo imaging to characterize microglial interactions with synapses during normal and altered sensory experience , in the visual cortex of juvenile mice . During normal visual experience , most microglial processes displayed direct apposition with multiple synapse-associated elements , including synaptic clefts . Microglial processes were also distinctively surrounded by pockets of extracellular space . In terms of dynamics , microglial processes localized to the vicinity of small and transiently growing dendritic spines , which were typically lost over 2 d . When experience was manipulated through light deprivation and reexposure , microglial processes changed their morphology , showed altered distributions of extracellular space , displayed phagocytic structures , apposed synaptic clefts more frequently , and enveloped synapse-associated elements more extensively . While light deprivation induced microglia to become less motile and changed their preference of localization to the vicinity of a subset of larger dendritic spines that persistently shrank , light reexposure reversed these behaviors . Taken together , these findings reveal different modalities of microglial interactions with synapses that are subtly altered by sensory experience . These findings suggest that microglia may actively contribute to the experience-dependent modification or elimination of a specific subset of synapses in the healthy brain .
Upon invasion of the central nervous system during embryonic and early postnatal development , bone-marrow-derived microglia become involved in apoptosis and phagocytic elimination of supernumerary neurons [1]–[3] . As they complete their differentiation , microglia change their morphology from amoeboid to ramified and are thought to become quiescent [4] . In the event of pathological insult , microglia rapidly become activated , thicken and retract their processes , migrate to the site of injury , proliferate , and participate in the presentation of antigens , phagocytosis of cellular debris , and secretion of proteases that promote microglial motility , as well as myelin and extracellular matrix degradation [5]–[7] . Additionally , activated microglia can separate presynaptic axon terminals from postsynaptic neuronal perikarya or dendrites in a process known as synaptic stripping [8] . Even though microglia are quiescent under non-pathological conditions , their highly motile processes continually survey the local environment and make transient contacts with astrocytes , neuronal perikarya , axon terminals , and dendritic spines in vivo [9]–[11] . Microglial apposition with astrocytic and neuronal elements has also been observed with electron microscopy ( EM ) in situ [9] , but a detailed analysis of microglial ultrastructural relationships is still lacking . Reports of spontaneous engulfment of cellular debris [10] suggest that resting microglia may exert phagocytic roles in the healthy brain . Because changes in the level of neuronal activity can also modify the volume of neuropil that microglia sample [10] , as well as their frequency of contact with axon terminals [9] , Wake et al . [9] proposed that resting microglia could monitor the functional state of synapses . However , aside from immune surveillance , the fate of synaptic architecture under the care of microglia remains poorly understood . The dynamic nature of microglial processes and their interaction with synapses suggest that microglia could effect structural changes at synapses , which are crucial to circuit remodeling and brain plasticity . To begin to investigate this possible task of quiescent microglia at synapses , we verified whether microglial interactions with synapses occur at random or coincide with structural synaptic changes and alterations in sensory experience . Specifically , we characterized the ultrastructural and structural/dynamic interactions between microglia and synapse-associated elements during normal sensory experience , sensory deprivation , and subsequent light exposure in the primary visual cortex ( V1 ) of juvenile mice . In addition to revealing the three-dimensional ( 3-D ) geometry of cell–cell contacts between microglia and all synapse-associated elements ( dendritic spines , axon terminals , perisynaptic astrocytic processes , and synaptic clefts ) , our observations uncovered new modes of microglia–synapse interaction under non-pathological conditions , particularly the regulation of the perisynaptic extracellular space and the phagocytosis of synaptic elements . Moreover , we found that microglia specifically localize to the vicinity of a subset of synaptic elements in vivo , in particular the structurally dynamic and transient dendritic spines . Lastly , we demonstrate that several modalities of microglia–synapse interactions are regulated by sensory experience . Thus , our findings indicate that microglia are not activated only during early brain development or pathological conditions; rather , they also subtly change their behavior toward synapses in correspondence with sensory experience . This raises the intriguing possibility that microglia may contribute to fine-tuning the plastic capacities of individual synapses in the healthy brain .
To provide a detailed view of the modes of interaction between microglia and excitatory synapses , we analyzed their ultrastructural relationships in layer II of mouse V1 on postnatal day ( P ) 28 , around the peak of the critical period for experience-dependent plasticity [12] . Using immunocytochemical EM with an antibody against the microglia-specific marker IBA1 [13] ( see Figure S1 for IBA1 immunostaining at the light microscopic level ) , we found that microglial cell bodies , as well as proximal and distal processes , juxtaposed synapse-associated elements including synaptic clefts ( Figures 1A–1C ) , an area generally thought to be exclusively reserved for astrocytic processes . Quantitative analysis revealed that the vast majority of microglial process profiles directly contacted at least one of the synapse-associated elements ( synaptic index: 94%±0 . 6%; ∼1 , 000 µm2 of neuropil in each of three animals ) . Axon terminals , dendritic spines , perisynaptic astrocytic processes , and synaptic clefts were contacted by microglial processes , in decreasing order of frequency ( n = 150 IBA1-positive microglial processes; three animals; see Table S1 for detailed analysis ) , and more than one synapse-associated element was generally contacted by each process ( 68%±4%; see Table S2 ) . To uncover the 3-D relationships between microglia and synapse-associated elements , we used serial section EM ( SSEM ) with 3-D reconstructions ( Figures 2 and S2 ) in layer II of mouse V1 at P28 . The 3-D reconstructions showed that proximal and distal microglial processes simultaneously contacted synapses of different shape and size , with typically more than one subcellular element contacted at individual synapses ( Figures 2B , 2D , 2E , S2A , and S2B ) . Whereas most contacts between microglia and synapses occurred en passant along microglial processes , without morphologic evidence of specialization , strikingly , the reconstructed distal microglial process displayed finger-like protrusions that wrapped around a dendritic spine making a synapse with an axon terminal ( Figure 2B , 2D , and 2E ) . Interestingly , the proximal microglial process was found to engulf cellular components ( see SSEM images in Figure S4 , but also the reconstructed cellular inclusion in Figures 2B , 2C , and S2A , S2B , S2C ) , suggesting microglial involvement in phagocytosis during normal sensory experience , along with our immunocytochemical EM observations that microglial perikarya and large processes sometimes contained cellular inclusions . Lastly , analysis of the series also showed occasional coated pits at the sites of cell–cell contact between microglia and dendritic spines , axon terminals , or astrocytic processes , either inside microglia or inside synapse-associated elements ( see Figures 1D and S3 for examples ) . Immunocytochemical EM and SSEM also revealed the appearance of large electron-lucent extracellular spaces nearby microglia , with both acrolein ( Figures 1A–1C and 2A ) and glutaraldehyde ( Figure 1D ) fixatives . Pockets of extracellular space , which consists of interstitial fluid supplemented with various extracellular matrix proteins , were otherwise rarely observed around any other structural elements in juvenile mice , in clear contrast with early postnatal stages of development [14]–[16] . Areas of extracellular space apposing IBA1-positive microglia were found to be significantly larger ( 985±41 nm2; n = 3 animals; ∼500 µm2 of neuropil in each ) than areas not associated with IBA1-positive processes ( 382±18 nm2; p<0 . 0002; Figure 1E ) . This analysis likely underestimated the association of extracellular space with microglial processes because of the partial penetration of antibodies under these stringent immunocytochemical conditions . In fact , most extracellular spaces were associated with unlabeled structural elements that resembled microglia morphologically ( see Materials and Methods for identification criteria ) . In addition , areas of a microglial process and associated extracellular space were tightly correlated ( r = 0 . 48; p<0 . 0001; n = 150 processes; three animals; Figure 1F ) , suggesting that microglia may exert a large influence in the creation of such space . SSEM with 3-D reconstructions also showed that microglia-associated spaces exhibited highly complex and varied morphologies ( Figures 2B , 2C , and S2C ) . Quantitative analysis of the series revealed that the 3-D pockets of extracellular space varied in volume by two orders of magnitude: from 20 , 223 to 7 , 048 , 921 nm3 ( mean: 905 , 834 nm3; median: 150 , 225 nm3; n = 15 extracellular spaces; Table S3 ) . In light of these results , when considering the interactions of synapses and microglia , it is important to take into account the complex organization of extensive contacts between a single microglia and every synapse-associated element at multiple synapses , interrupted by many pockets of geometrically complex microglia-specific extracellular spaces , simultaneously throughout the neuropil . Taken together , our findings from immunocytochemical EM and SSEM with 3-D reconstructions indicate that microglia are uniquely positioned to play several physiological roles at synapses: through cell–cell communication with perisynaptic astrocytic processes , dendritic spines , and axon terminals simultaneously , as well as through the regulation of the extracellular environment intervening between them . To characterize the structural dynamics of microglia–synapse interactions , we used two-photon in vivo imaging of layers I/II of V1 in juvenile CX3CR1-GFP/Thy1-YFP mice [17] , [18] , in which both microglia and layer V neurons are fluorescently labeled . We utilized a thinned-skull preparation , which minimizes brain injury and allows long-term tracking of microglial dynamics without causing microglial activation [19] ( Figure S5 ) . Even though the resolution of two-photon microscopy ( see Materials and Methods for details on measurement of the experimental point spread function ) prevents visualization of direct contacts between fluorescently labeled elements , it enables the study of their structural dynamics and determination of close proximity ( the apparent colocalization of fluorescence for microglial and neuronal elements was considered putative contact ) . A recent study demonstrated a relatively constant duration ( 4 . 60±0 . 08 min ) of putative contacts between microglia and synaptic elements in layer II/III of juvenile mouse V1 [9] . In contrast , our time-lapse imaging ( every 5 min for 30–120 min ) revealed similar contacts between microglial processes and a subset of the YFP-labeled dendritic spines and axon terminals ( n = 37 putative contacts with spines and 29 putative contacts with terminals in eight animals; Figures S7A and 3A; see Videos S1 and S4 ) that varied between 5 and 50 min . The cortical layers examined or the identity of synapses imaged ( GFP-M and YFP-H mice label different subsets of pyramidal cells ) [20]–[23] might explain this discrepancy . Nevertheless , these differences in contact duration reveal a new dimension in microglial interactions with synapses in the healthy brain . To verify whether microglial processes target specific subsets of synapses , we measured the size of dendritic spines and axon terminals in the presence and absence of a putative microglial contact . In this analysis , we noticed that dendritic spines that were in close proximity to microglial processes at any point during the imaging session were generally smaller than the rest of the spine population . We quantitatively recorded spine size at the beginning of imaging and found that dendritic spines receiving putative contact during imaging were significantly smaller than spines remaining non-contacted ( p<0 . 001; n = 31 contacted spines in five animals and 45 non-contacted spines in three animals , with infrequent stubby spines excluded; see Materials and Methods ) . When microglia came into close proximity with these synaptic structures , individual pre- and postsynaptic elements both expanded and shrank: 38% of axon terminals grew , 55% shrank , and 7% remained stable ( Figure S7B and S7C ) . Average axon terminal sizes were not significantly different with and without putative microglial contact ( size differential: −1%±3%; p>0 . 9; n = 24 terminals and 29 putative contacts in three animals; Figure S7B and S7C ) , nor correlated to initial terminal size or putative microglial contact duration ( p>0 . 5 and p>0 . 2 , respectively; Figure S7D and S7E ) . In contrast , 62% of dendritic spines grew , 32% shrank , and 6% remained stable during putative contact; additionally , we observed a significant increase in average spine sizes in the presence versus in the absence of putative microglial contact ( size differential: 9%±3%; p<0 . 03; n = 31 spines and 37 contacts in five animals; Figure 3B ) . These changes were generally transient , since dendritic spine size was not significantly different between before and after the contact ( p>0 . 9; Figure 3C ) . No correlation between size change and putative contact duration was noted ( p>0 . 9; Figure S8A ) , but the size change and initial spine size were significantly correlated ( r = 0 . 28; p<0 . 001; Figure 3D ) , with the smallest spines undergoing the largest size changes during contact . Indeed , the comparison of small and large dendritic spines revealed a significant difference in their structural changes during putative microglial contact ( average size differential for large spines: 1%±3%; for small spines: 17%±5%; p<0 . 01; Figures 3E , 3F , S8B , and S8C ) . Therefore , these observations indicate that microglial processes preferentially localize to small and structurally dynamic dendritic spines . To determine whether spines targeted by microglia exhibit a different long-term fate with respect to their longevity , we carried out chronic in vivo imaging experiments tracking individual dendritic spines over a period of 2 d ( Figure 3G and 3H ) . Surprisingly , we found that dendritic spines that received putative microglial contact during the first imaging session were more frequently eliminated ( 24%±6%; n = 30 spines in four animals ) than non-contacted spines ( 7%±3%; p<0 . 05; n = 56 spines in four animals ) . Interestingly , among contacted dendritic spines , only small spines were lost ( 8/18 small spines and 0/12 large spines lost ) . We conclude that the subset of small and dynamic dendritic spines contacted by microglia also have an increased rate of elimination over a span of 2 d . Our two-photon imaging results raise the intriguing possibility that microglia may not only monitor the functional status of synapses , but also exert control on structural changes or spine elimination either through direct contact or indirect signaling that requires close microglia–synapse proximity . To investigate whether microglial behavior towards synapses is regulated by sensory experience , we altered visual experience by housing juvenile mice in complete darkness ( dark adaptation [DA] ) for 6 d , from the beginning to the peak of the critical period [12] . Binocular deprivation increases dendritic spine motility and turnover in mouse V1 in vivo [24] , [25]; this provides an excellent model to correlate microglial behavior with the experience-dependent modification and elimination of synapses under non-pathological conditions . While microglial processes were generally larger in layer II of V1 , they exhibited two morphological phenotypes at the ultrastructural level: some processes appeared very large and thick ( “bulky”; Figure S10D and S10E ) , while others exhibited many short , thin fingers and appeared “spindly” ( Figures 4C and S10F ) . Both types of microglial processes made multiple contacts with synapse-associated elements , including synaptic clefts ( Figures 4C and S10D , S10E , S10F ) . Microglial perikarya and bulky processes often contained cellular inclusions ( p<0 . 0001; n = 3 control and 3 DA animals; 50 IBA1-immunopositive microglial processes each; Figures 4A , 4B , 4E , S9A , S10D , and S10E ) , which sometimes resembled dendritic spines or axon terminals , suggesting their phagocytic engulfment by microglia . Spindly processes typically ensheathed dendritic spines or axon terminals and displayed extended extracellular space areas ( Figures 4C and S10F ) . Microglial process area ( p<0 . 01; n = 3 animals per condition; 50 IBA1-immunopositive processes each; Figure 4F; Table S1 ) and extracellular space area ( p<0 . 05; n = 3 animals; Figure 4G; Table S1 ) were significantly increased during DA , with the correlation between microglial process area and extracellular space area remaining significant ( r = 0 . 44; p<0 . 0001; n = 150 processes in three animals; Figure S9B ) . In contrast , microglial process density ( DA: 131±6; control: 142±6 processes per 1 , 000 µm2 of neuropil; n = 3 animals; 1 , 000 µm2 of neuropil each; Figure S9C ) and synaptic index ( DA: 95%±0 . 6%; control: 94%±0 . 6%; n = 3 animals; Figure S9D ) were unchanged by sensory experience ( p>0 . 3 and p>0 . 1 , respectively ) . Surprisingly , contacts with synaptic clefts were more frequent in DA animals ( p<0 . 05; n = 3 animals per condition; total for 50 processes each; Figure 4H; Table S1 ) than controls , while contact frequency with dendritic spines , axon terminals , and astrocytic processes was unchanged ( Figures 4H and S9E; Tables S1 and S2 ) . Furthermore , as expected from extended microglial processes , their average perimeter of contact with every synapse-associated element was also increased during visual deprivation ( spine: p<0 . 01; astrocyte: p<0 . 02; terminal: p<0 . 05; n = 3 animals per condition; 50 processes each; Figure 4I; Table S1 ) . Thus , our ultrastructural observations reveal subtle experience-dependent changes in the behavior of microglia , most notably an expansion of their processes and associated extracellular space , an increased occurrence of cellular inclusions , an increased frequency of contact with synaptic clefts , and an increased apposition with every synapse-associated element . To determine whether these changes in behavior could be reversed by reexposure to daylight , we subjected mice to DA for 6 d followed by a fixed 12-h light/dark cycle for 2 d ( DA+light ) , as a few hours of reexposure to light after dark rearing ( i . e . , housing in complete darkness from birth ) can cause rapid molecular changes at synapses , cause structural changes of dendritic spines , and reverse the effects of dark rearing on synaptic transmission and plasticity [26]–[29] . At the ultrastructural level in layer II of V1 , most microglial process profiles appeared bulky during light reexposure , contained many inclusions , and were surrounded by small pockets of extracellular space ( see Figures 4D and S10G , S10H , S10I for examples ) . Quantitative analysis revealed that microglial process area returned to control levels ( p>0 . 2 versus control; p>0 . 09 versus DA; n = 3 animals per condition; Figure 4F ) while microglia-associated extracellular space area was significantly reduced following light reexposure ( p<0 . 004 versus control; p<0 . 02 versus DA; n = 3 animals per condition; Figure 4G ) . In contrast , the number of cellular inclusions , many of which contained elements resembling synaptic profiles ( Figure S10G , S10H , S10I ) , remained significantly higher than in control animals ( p<0 . 008 versus control; p>0 . 2 versus DA; n = 3 animals per condition; Figure 4E ) . Similarly , contacts with synaptic clefts ( p<0 . 03 versus control; p>0 . 9 versus DA; n = 3 animals per condition; Figure 4H ) and perimeters of contact with dendritic spines ( p<0 . 01 versus control; p>0 . 4 versus DA; Figure 4I ) and axon terminals ( p<0 . 02 versus control; p>0 . 6 versus DA ) , but not with perisynaptic astrocytes ( p>0 . 2 versus control and DA ) , remained extended . Future experiments with longer light exposures after deprivation will be needed to determine whether these phenomena can be reversed with further light reexposure . Taken together , these observations reveal a complex interaction between sensory-driven activity and microglial behavior . While the expansion of microglial processes and associated extracellular spaces reversed after brief reexposure to daylight , microglial ensheathment of dendritic spines and axon terminals , as well as their phagocytic inclusion , were still increased . To assess the dynamic changes in microglia–synapse interactions during visual deprivation , we used two-photon imaging of layers I/II of V1 in juvenile mice that were subjected to DA for 8–10 d , from the beginning to the peak of the critical period [12] . Microglial processes appeared thickened and sparse , and more often terminated into crown-like structures resembling phagocytic cups than in control animals [30] ( Figure 5A and 5B; Video S2 , as well as Figure S12 and Videos S4 and S5 for comparison of microglial morphology in control and DA animals ) . We also found that the average motility of microglial processes was significantly reduced ( Figure 5C and 5D ) when assayed in two ways: comparing morphology over a 5-min interval ( motility index; control: 8%±0 . 8%; DA: 6%±0 . 5%; p<0 . 05; n = 10 microglia in four control animals and 8 microglia in four DA animals ) and over a 25-min interval , where the difference between control and DA animals was more pronounced ( control: 11%±0 . 8%; DA: 7%±0 . 6%; p<0 . 01 ) . Quantitative analysis of dendritic spines showed that spines not receiving putative microglial contact were significantly smaller in DA animals than in non-light-deprived animals ( p<0 . 001; n = 45 spines in three animals per experimental condition ) , whereas contacted spines were equally small in both DA and control animals ( p>0 . 5; Figure 5E ) , in agreement with an observed reduction in synaptic strength during binocular deprivation [31] , [32] and the smaller sizes of dendritic spines during synaptic depression [33] , [34] . Surprisingly , dendritic spines putatively contacted by microglia in DA animals were significantly bigger than non-contacted spines ( p<0 . 05 ) , revealing that microglia no longer localize to smaller spines in this condition . The average duration of putative microglial contact with dendritic spines was slightly increased in DA animals ( 10±2 min; n = 13 contacts in three animals ) compared with controls ( 9±1 min; n = 37 contacts in five animals; Figure S11A ) , but the difference was not significant ( p>0 . 8 ) , suggesting that increased coverage of synaptic elements by microglia is not a result of longer duration of contact . Similarly , the frequency of putative contacts with individual dendritic spines was unchanged by sensory experience ( DA: 1±0 . 08; control: 1±0 . 07 contacts per 40 min; p>0 . 6; Figure S11B ) . Lastly , most dendritic spines shrank during microglial contact ( 29% grew , 57% shrank , and 14% remained stable; n = 12 spines and 14 putative contacts in three animals ) , unlike in the control condition , where most spines grew . While average size changes during contact were not significant ( size differential: −4%±4%; p>0 . 3; Figure S11C ) , interestingly , dendritic spine shrinkage persisted after microglial contact , with a significant difference in size between before and after the contact ( p<0 . 02; Figure 5F ) , a phenomenon which may contribute to the reduction in size of the spine population . These results indicate that sensory-deprived microglia undergo subtle behavioral changes reminiscent of activation , including reduced motility , thickened processes , and phagocytic specializations . Intriguingly , despite unaltered duration and frequency of microglial contacts with synaptic elements , their preference of localization did change , from a subset of smaller dendritic spines that transiently grow to a subset of bigger spines that persistently shrink . To further investigate these experience-dependent changes in microglial behavior , juvenile animals that were subjected to DA for 6–8 d were reexposed to daylight for 2 d before two-photon imaging of layers I/II of V1 . Microglial morphologies resembled those in control animals , with generally thinner and more abundant processes within the neuropil ( Figure S12; Video S6 ) , but microglial processes still displayed phagocytic structures ( Video S3 ) . Microglial motility in animals reexposed to light was similar to that in control animals and was significantly increased compared with DA animals , when assessed during a 5-min interval ( motility index = 10%±0 . 8%; p>0 . 2 versus control; p<0 . 001 versus DA; eight microglia in three DA+light animals; Figure 5D ) and a 25-min interval ( motility index = 13%±1 . 3%; p>0 . 1 versus control; p<0 . 0007 versus DA ) . Quantitative analysis of dendritic spines showed that spines receiving putative microglial contacts were of similar sizes to those in non-light-deprived and light-deprived animals ( p>0 . 1 versus control and DA; n = 14 contacted spines in three DA+light animals; Figure 5E ) , while non-contacted spines were significantly bigger in animals reexposed to light ( p<0 . 01 versus control; p<0 . 0001 versus DA; n = 45 spines in three animals per condition ) . The sizes of contacted and non-contacted spines were not significantly different in animals reexposed to light ( p>0 . 9; Figure 5E ) , indicating that microglia no longer localize to specific spine types as in control or DA animals . Lastly , microglia–synapse interactions showed similar structural effects on dendritic spines as in control conditions . Most dendritic spines increased in size during putative microglial contact ( 67% grew , 20% shrank , and 13% remained stable; average size differential: 13%±5%; p<0 . 01 comparing with/without and before/during contact; n = 14 spines and 15 putative contacts in three animals; Figures 5F and S11C ) , and this growth was transient ( p>0 . 8 comparing size differential before/after contact; Figure 5F ) . These results reveal additional changes in microglial behavior that can be reversed by brief reexposure to daylight , particularly their motility and preference of contact for subsets of dendritic spines .
One of the most striking findings of our study was the demonstration of distinctive extracellular spaces closely correlated with the presence of microglial processes . Our EM observations revealed large electron-lucent pockets of extracellular space surrounding microglia . To our further surprise , we also observed changes in the distribution of these microglia-associated extracellular spaces during alterations in visual experience: an expansion during light deprivation and shrinkage during light reexposure . Although alteration of these spaces by brain fixation and embedding for EM may warrant further investigation , we observed them under all conditions tested . In future experiments , it will be important to determine whether microglial processes create this extracellular space themselves or move in to fill space that is created by an unknown mechanism . In any case , our findings of microglia-specific extracellular spaces suggest that microglia have an intimate relationship with their extracellular milieu and may even regulate their surrounding environment in a unique and specific way that is determined by physiological conditions . If microglia directly modulate the extracellular space , they may do this through the secretion of various proteases that degrade extracellular matrix proteins , including cathepsins , metalloproteases , and tissue-type plasminogen activator [35] . This proteolytic degradation of specific matrix proteins could , in turn , facilitate microglial motility , as suggested by the finding that the migratory behavior of cathepsin S–deficient microglia is severely impaired in vitro [35] . In line with this , the volume of extracellular space greatly decreases during postnatal cortical development , concomitant with changes in extracellular matrix composition and reductions in cell migration and process elongation [14]–[16] , [36] , [37] . The appearance of extracellular spaces specifically associated with microglial processes could therefore reflect their highly motile behavior , relative to other structural elements of neuropil in juvenile cortex . Additionally , regulation of the extracellular matrix composition by microglia-derived proteases could contribute to dendritic spine motility and pruning , as well as activity-dependent and experience-dependent plasticity , which are profoundly affected in vitro and in vivo by treatments with proteases that degrade extracellular matrix proteins [38]–[43] . Immunocytochemical EM and SSEM with 3-D reconstructions enabled us to analyze the morphology of microglial processes and their ultrastructural relationships with the other subcellular compartments of neuropil—astrocytic processes , axon terminals , and dendritic spines—in situ at high spatial resolution . Building on previous EM observations that microglia contact axon terminals and dendritic spines [9] , [44] , [45] , our quantitative analysis revealed that most microglial processes directly appose not only axon terminals and dendritic spines , but also perisynaptic astrocytic processes and synaptic clefts . Our SSEM with 3-D reconstructions also uncovered the 3-D relationships between microglia and synapses , revealing that microglial processes contact multiple synapse-associated elements at multiple synapses simultaneously . Additionally , we found clathrin-coated pits at interfaces between microglia and dendritic spines , axon terminals , or perisynaptic astrocytic processes , suggesting clathrin-mediated endocytosis of membrane-bound receptors and their ligands , a phenomenon known to initiate various cellular signaling events [46] . Since clathrin-coated pits also occur at the tips of most spinules undergoing invagination , as previously observed in small dendritic spines , axon terminals , and perisynaptic astrocytic processes of mouse hippocampus [47] , this may suggest trans-endocytosis of membrane-bound receptors and their ligands , in addition to a direct exchange of cytoplasm , between microglia and synapse-associated elements . To better understand the functional significance of these forms of molecular communication between microglia and synapse-associated elements , it will be important to identify the molecules that are being internalized during their dynamic interactions . Following visual deprivation and reexposure to daylight , our EM results revealed that microglial processes change their morphology , appose synaptic clefts more frequently , and envelop synapse-associated elements more extensively . Since glial presence at the synaptic cleft was classically restricted to astrocytic processes regulating synaptic function through modification of the extracellular space geometry and bidirectional communication with synaptic elements [48] , [49] , this novel finding suggests that microglia may also contribute uniquely to synaptic transmission and plasticity in the healthy brain . Additionally , the ensheathment of synaptic elements and synaptic clefts by microglial processes may imply their participation in activity-dependent synapse elimination [50] , through a separation of pre- and postsynaptic elements reminiscent of synaptic stripping , as previously reported between axon terminals and neuronal cell bodies during immune responses [8] . Lastly , our EM and two-photon in vivo imaging observations revealed that a subpopulation of microglial processes displays phagocytic structures , with an increasing prevalence during alterations in visual experience , thus providing evidence for a microglial role in phagocytic engulfment under non-pathological conditions . This is supported by observations that quiescent microglia can spontaneously engulf tissue components in vivo [10] and that activated microglia play an essential role in phagocytosis of cellular debris [1]–[3] , [5] , [6] . At the ultrastructural level , we also found cellular inclusions that resembled dendritic spines or axon terminals , suggesting that quiescent microglia can phagocytose synaptic elements in juvenile cortex . In line with this , classical complement proteins C1q and C3 were recently shown to be involved in the pruning of inappropriate retino-geniculate connections during early postnatal development [51] . Since C1q and downstream complement protein C3 can trigger a proteolytic cascade leading to microglial phagocytosis , this finding supports a model in which microglia may contribute to synaptic pruning under non-pathological conditions [52] . Taken together , our findings indicate that distinct modes of microglial interactions with synapses , most notably apposition , ensheathment , and phagocytosis , are subtly regulated by sensory experience . Two-photon visualization of microglia and synaptic elements with two different colors in CX3CR1-GFP/Thy1-YFP mice enabled clear distinction of the separate structures , which facilitated identification of their putative contacts . This approach revealed transient localization of microglial processes to the vicinity of dendritic spines and axon terminals , in agreement with observations from a recent study [9] . In our study , clear distinction of the separate structures also allowed , for the first time , measurement of dendritic spine and axon terminal morphological changes during episodes of proximity with microglial processes . Our results revealed a specificity of these putative microglial contacts for a subset of small , transiently growing , and frequently eliminated dendritic spines . This is consistent with previous findings showing that in mouse visual and somatosensory cortical areas in vivo , small dendritic spines are more motile and more frequently eliminated than their larger counterparts [20] , [22] , [24] , [53] , [54] . Dendritic spines undergoing long-term potentiation transiently or persistently enlarge in vitro [33] , [34] , [55] , [56] , but since dendritic spine structural changes can occur independently of changes in synaptic strength [57] , a connection between putative microglial contact and synaptic plasticity remains to be determined . During light deprivation , microglia preferentially localized to larger dendritic spines that persistently shrank , akin to spines undergoing long-term depression in vitro [33] , [34] or shrinking before elimination in vivo [58] , while during reexposure , microglia reversed to contacting spines that transiently grew , as in control animals . This uncovers a specificity of microglial interaction for subsets of structurally dynamic and transient dendritic spines , a specificity that , surprisingly , changes with sensory experience . In future experiments , correlating the temporal dynamics of microglial contact with dendritic spine activity could be achieved by in vivo labeling of neurons with electroporated calcium indicators [59] or viral delivery of genetic calcium indicators [60] . A challenge with such techniques will be their invasive nature; the mechanical disruption of the brain alone during indicator delivery can result in microglial activation . Additionally , it will be important to determine whether microglial contacts occur in response to structural changes in dendritic spines and whether such contacts instruct subsequent spine elimination . This will , however , require new technical advances enabling interference with microglial contacts while preserving physiological conditions . Importantly , whether direct microglial contacts , such as observed with EM , are required for these structural changes or whether microglia can exert effects on synapses without close apposition will need to be explored . Several lines of evidence , including those presented here , suggest that microglial motility may be regulated by neuronal activity . A pioneer study reported an increase in the volume sampled by quiescent microglia in vivo over a period of 1 h after application of the ionotropic GABA receptor blocker bicuculline , whereas the sodium channel blocker tetrodotoxin had no significant effects [10] . More recently , microglia were shown to retract their processes and reduce their frequency of contact with axon terminals in vivo , over a period of 4–6 h after binocular enucleation or intraocular injection of tetrodotoxin [9] . In our study , two-photon analysis revealed a global decrease in microglial motility during light deprivation without correlated changes in the duration or frequency of microglial contact with individual dendritic spines . This may highlight differences between short-term ( 1–6 h ) and long-term ( 8–10 d ) microglial responses to sensory deprivation or distinguish between more invasive manipulations , which approximate nervous system injury , and more physiological paradigms such as DA . Our observations further suggest that subsets of microglial processes may have different behavior: those in contact with dendritic spines may be highly motile , while others ( for example bulky microglial processes displaying phagocytic structures ) may become less motile . It is also possible that spindly microglial processes ( <100 nm ) , which were typically devoid of cellular inclusions and surrounded by extended extracellular space , were undetected with two-photon imaging and therefore not accounted for in our motility analysis . This could explain the finding that microglial motility decreased during light deprivation while microglia-associated extracellular spaces expanded . However , as we found an increase in microglial motility during light reexposure that was accompanied by a decrease in microglia-associated extracellular space areas , it is likely that microglial motility and extracellular space areas are regulated independently . Using several technical approaches , we were able to provide a qualitative and quantitative characterization of the ultrastructural and structural/dynamic interactions between microglia and synapses under non-pathological conditions . This characterization revealed specific modalities of microglia–synapse interactions that are subtly altered by sensory experience , supporting the exciting possibility that microglial influence on synaptic plasticity is not restricted physiologically to an immune response to brain injury and disease .
Animals were treated in strict accordance with the University of Rochester Committee on Animal Resources and the United States National Institutes of Health standards . Light-reared animals were housed under a fixed 12-h light/dark cycle; DA animals were placed in complete darkness for 6–10 d , from P20–P22 until P28–P32 ( the onset of experimentation ) ; DA+light animals were housed under a fixed 12-h light/dark cycle for 2 d following DA until P29–P32 ( the onset of experimentation ) . For immunocytochemical EM and SSEM , eight C57Bl/6 mice ( P28 ) were anesthetized with sodium pentobarbital ( 80 mg/kg , i . p . ) and perfused through the aortic arch with 3 . 5% acrolein followed by 4% paraformaldehyde as previously described [14] , [61] or with 2 . 75% glutaraldehyde in 2% paraformaldehyde to compare extracellular space areas surrounding microglia between both types of fixatives . For two-photon imaging [17] , 21 CX3CR1-GFP/Thy1-YFP mice ( P28–P39 ) , in which microglia and cortical layer V neurons are respectively GFP- and YFP-labeled [17] , [18] , were anesthetized with a mixture of fentanyl ( 0 . 05 mg/kg , i . p . ) , midazolam ( 5 . 0 mg/kg ) , and metatomadin ( 0 . 5 mg/kg ) [62] and kept at 37°C with a heating pad . Stereotaxical coordinates encompassing a 2- to 3-mm-diameter region ( A +0 . 16 to A +0 . 64 , between 2 and 3 mm from the midline ) [63] were used to identify V1 . In all cases , DA animals were anesthetized in the dark using infrared goggles or a darkroom red light before undergoing perfusion or surgery in the light . After acute ( 30 min to 2 h ) and chronic ( two imaging sessions over 2 d , 30 min to 2 h each ) two-photon imaging , eight mice were perfused with 4% paraformaldehyde to confirm that microglia are not activated by the surgical procedure and imaging paradigm ( Figure S4 ) . Transverse sections of the brain ( 50 µm thick ) were cut in ice-cooled PBS ( 0 . 9% NaCl in 50 mM phosphate buffer [pH 7 . 4] ) with a vibratome . Sections were immersed in 0 . 1% sodium borohydride for 30 min at room temperature ( RT ) , washed in PBS , and processed freely floating following a pre-embedding immunoperoxidase protocol previously described [14] , [61] , [64] , [65] . Briefly , sections were rinsed in PBS , followed by a 2-h pre-incubation at RT in a blocking solution of PBS containing 5% normal goat serum and 0 . 5% gelatin . They were incubated for 48 h at RT in rabbit anti-IBA1 antibody ( 1∶1 , 000 in blocking solution; Wako Pure Chemical Industries ) and rinsed in PBS . After incubation for 2 h at RT in goat anti-rabbit IgGs conjugated to biotin ( Jackson Immunoresearch ) and with streptavidin-horseradish peroxidase ( Jackson Immunoresearch ) for 1 h at RT in blocking solution , labeling was revealed with diaminobenzidine ( 0 . 05 mg/ml ) and hydrogen peroxide ( 0 . 03% ) in buffer solution ( DAB Peroxidase Substrate Kit; Vector Laboratories ) . Sections for light microscopy were mounted onto microscope slides , dehydrated in ascending concentrations of ethanol , cleared in xylene , and coverslipped with DPX ( Electron Microscopy Sciences ) . Sections for EM were post-fixed flat in 1% osmium tetroxide and dehydrated in ascending concentrations of ethanol . They were treated with propylene oxide , impregnated in Durcupan ( Electron Microscopy Sciences ) overnight at RT , mounted between ACLAR embedding films ( Electron Microscopy Sciences ) , and cured at 55°C for 48 h . Areas of V1 , at a level approximating the transverse planes A +0 . 16 to A +0 . 72 [63] , were excised from the embedding films and re-embedded at the tip of resin blocks . Ultrathin ( 65–80 nm ) sections were cut with an ultramicrotome ( Reichert Ultracut E ) , collected on bare square-mesh grids , stained with lead citrate , and examined with a Hitachi 7650 electron microscope . Light microscope pictures of IBA1 immunostaining were taken at 20× in layer II of V1 , using a Spot RT color digital camera ( Diagnostic Instruments ) . Eighty pictures were randomly taken at 40 , 000× in layer II of V1 in each animal ( n = 3 control , 3 DA , and 3 DA+light ) , corresponding to a total surface of ∼1 , 000 µm2 of neuropil per animal . Cellular profiles were identified according to criteria previously defined [14] , [44] , [66] , [67] . In addition to their IBA1 immunoreactivity , microglial cell bodies were recognized by their small size and the clumps of chromatin beneath their nuclear envelope and throughout their nucleoplasm . Microglial processes displayed irregular contours with obtuse angles , dense cytoplasm , numerous large vesicles , occasional multivesicular bodies , vacuoles or cellular inclusions ( large lipidic vesicles , profiles of cellular membranes , and profiles of other structural elements including dendritic spines and axon terminals ) , and distinctive long stretches of endoplasmic reticulum . These morphological characteristics enabled identification of microglial processes on an ultrastructural level in non-immunocytochemical SSEM and non-immunocytochemical glutaraldehyde-fixed material . In each of three control and three DA animals ( ∼1 , 000 µm2 of neuropil per animal ) , IBA1-immunopositive microglial process profiles were counted and their contacting structural elements identified . A synaptic index was calculated based on the percentage of IBA1-positive microglial processes contacting synapse-associated elements divided by the total number of IBA1-positive microglial processes ( Figure S9 ) . It is important to note that since analysis was performed on single sections , different microglial process profiles could be part of a single microglial process , which may lead to an overestimation of the synaptic index in this analysis . In 39 pictures in each of three control animals ( ∼500 µm2 of neuropil per animal ) , all extracellular space areas wider than synaptic clefts ( 10–20 nm ) [66] were measured and determined to contact or not contact IBA1-stained microglial processes ( Figure 1; Table S3 ) . To characterize the ultrastructural relationships between microglia and synapse-associated elements , 50 randomly selected IBA1-positive microglial processes per animal were analyzed in more detail with Image J software ( United States National Institutes of Health ) . For measurement of microglial process and extracellular space areas , individual microglial processes and all their adjacent extracellular spaces wider than synaptic clefts were traced with the freehand line tool . Their area was quantified in pixels and converted into nanometers ( Figures 1 , 4 , and S9; Table S1 ) . For quantification of contacts between microglial processes and synapse-associated elements , we first counted all direct juxtapositions between individual microglial processes and dendritic spines , axon terminals , synaptic clefts , and perisynaptic astrocytic processes ( Figure 4; Table S1 ) . For example , microglial processes contacting synaptic clefts were also contacting dendritic spines and axon terminals . Since most microglial processes contacted multiple synapse-associated elements simultaneously , we performed a second analysis where we categorized individual microglial processes based on their contacting partners: dendritic spine only , axon terminal only , perisynaptic astrocytic process only , spine+terminal , spine+astrocytic process , terminal+astrocytic process , or spine+astrocytic process+terminal . Microglial contacts with synaptic clefts were included in the spine+terminal or spine+astrocytic process+terminal categories ( Table S2 ) . For measurement of perimeters of contact between microglia and dendritic spines , axon terminals , or perisynaptic astrocytic processes , all microglial cell membranes in direct juxtaposition with these structural elements were traced with the freehand line tool ( Figure 4; Table S1 ) . Their length was quantified in pixels and converted into nanometers . A series of 100 ultrathin sections ( 65 nm ) was cut , collected on pioloform-coated grids , stained with lead citrate , and examined with a JEOL JEM-1230 electron microscope . Fifty serial images of a microglial process were taken at 10 , 000× . The alignment of images , tracing of structural elements , and 3-D reconstructions were performed with the Reconstruct software [68] . To calculate the volume of 15 extracellular spaces , we measured their area in all sections in which they appeared and multiplied their total area by the number and thickness of sections . The skull above V1 was exposed , cleaned , glued to a thin metal plate , and carefully thinned to an approximately 20- to 30-µm thickness , using a high-speed dental drill ( Fine Science Tools ) and a microsurgical blade [19] , [24] , [69] . Drilling was interrupted periodically , and sterile saline was applied on the skull to prevent heat-induced damage . A custom-made two-photon microscope [70] with a Ti∶Sapphire laser ( Mai Tai; Spectra Physics ) tuned to 920 nm was used for transcranial imaging . Fluorescence was detected using two photomultiplier tubes in whole-field detection mode and a 506-nm dichroic mirror with 580/180 ( green channel: GFP ) and 534/34 ( yellow channel: GFP and YFP ) emission filters . A 20× water-immersion lens ( 0 . 95 N . A . ; Olympus ) was used throughout the imaging session . Dendrites and axons near microglia , located at least 50 µm below the pial surface , were imaged under a digital zoom of 4–6× , using the FluoView software . Z stacks taken 1 µm apart were acquired every 5 min for 30 min to 2 h . To measure the resolution of two-photon imaging under our experimental conditions , we imaged subresolution fluorescent beads ( 0 . 1–0 . 2 µm diameter; Molecular Probes ) to obtain the experimental point spread function . The measured 1/e radius was 0 . 35 µm radially and 1 . 3 µm axially , corresponding to the X/Y and Z resolution , respectively . Analysis of microglial contacts with dendritic spines ( n = 5 control animals , 3 DA animals , and 3 DA+light animals ) and axon terminals ( n = 3 control animals ) , changes in synaptic structures ( n = 3 and 5 control animals with terminals and spines , respectively , 3 DA animals with spines , and 3 DA+light animals with spines ) , dendritic spine turnover ( n = 4 control animals ) , and microglial motility ( n = 4 control animals , 4 DA animals , and 3 DA+light animals ) was performed with Image J software . Axon terminals were identified as finger-like protrusions ( >0 . 2 µm ) from the axon or bead-like structures along the axon ( at least two times the axon diameter ) , as previously described [54] . While dendritic spine morphology exists as a continuum , different behaviors have been assigned to spines with different structures [71] . Therefore we classified dendritic spines into mushroom , thin , and stubby types based on their length and spine head volume , as previously described [24] . The only three stubby spines contacted by microglia were excluded from the analysis of spine size because of their rarity and the difficulty of measuring their fluorescence intensity . Consequently , we also removed stubby spines from the analysis of size for non-contacted and contacted dendritic spines . Among thin and mushroom spines , we determined a range of normalized dendritic spine sizes . The spines belonging to the first quarter of this range were considered “small , ” and those in the other three-quarters were identified as “large . ” Filopodia were rarely observed at these ages and were excluded from the analysis . For visualization of microglial contacts with dendritic spines and axon terminals , the green channel was arbitrarily assigned the color red and the yellow channel assigned the color green , enabling the visualization of microglia in yellow and neuronal elements in green ( see Figure S6 ) . Microglial contacts were identified manually by stepping through the Z stack without projection . All microglial contacts ( colocalization of fluorescence for microglia and synaptic elements ) that started and ended during imaging were included in the analysis . For analysis of changes in synaptic structures , the background was subtracted from each of the two channels and the green channel bleedthrough was subtracted from the yellow channel ( see Figure S6 for uncorrected and corrected images ) . In the stacks unadjusted for brightness or contrast , dendritic spines and axon terminals were analyzed for fluorescence at the Z level where they appeared brightest . A line was traced through each element , and a fluorescence plot profile was created , which was then fitted to a Gaussian . Because the majority of dendritic spines are below the resolution of our two-photon microscope [72] , the maximal fluorescence ( amplitude of the Gaussian fit ) was used to assess the relative spine size ( see Figure S8C for sizes of large spines assessed with the width of the fluorescence profile [1/e1/2 radius of the Gaussian fit] to rule out underestimation of their size changes during putative contact ) . Since axon terminals are generally much larger than dendritic spines , the width of the Gaussian fit to the fluorescence profile was used to determine their relative size . To rule out contamination of neuronal measurements by any unsubtracted bleedthrough of microglial fluorescence , we analyzed thin fluorescent axons ( n = 1 axon in each of 14 animals ) that were contacted by microglia , using the height of the Gaussian fit to assess relative size . Axon size measured in this manner was relatively unchanged between populations with and without microglial contact ( average size differential of −0 . 1%±1%; see Figure S6D and S6E ) , as expected for these structurally stable elements . Size differentials of terminals and spines were calculated as the ratio of size difference with and without contact over the size without contact . In order to compare spine size between animals , we normalized spine fluorescence by the maximal fluorescence in the adjacent dendrite . Axon terminals and dendritic spines with size differentials under 1% were considered stable . For presentation purposes , we normalized the size of terminals and spines by dividing each individual structure's average size in the presence of microglial contact ( with or during ) by its average size in the absence of microglial contact ( without , or before or after ) . We also normalized the size of terminals and spines by dividing each individual structure's average size by the size of the largest terminal or spine . To determine the turnover of dendritic spines that were or were not contacted by microglia in the first imaging session , the position of individual dendritic spines was compared between time points separated by 2 d . The proportion of eliminated dendritic spines was defined as the proportion of spines from the original population not observed on the second day of imaging . Spines located more than 0 . 8 µm laterally from their previous location were considered to be new spines . For measurement of microglial motility , images centered on the cell body from five consecutive Z levels were projected into two dimensions , for each microglia and each time point analyzed ( 0 , 5 , and 25 min ) . For each microglia , the images were aligned and grouped into a stack . Stacks were adjusted for brightness and contrast , and then binarized . For each microglia , the difference between images taken at the 0- and 5- or 25-min time points was calculated . A motility index was determined , as the proportion of the pixels that differed between the two images . Analyses were performed with Prism 5 software ( GraphPad Software ) . All values reported in the text are mean ± standard error of the mean ( SEM ) . For all statistical tests , significance was set to p<0 . 05 . Two-tailed unpaired Student's t tests and linear regressions were used for both EM and two-photon analyses . For two-photon analyses , two-tailed paired Student's t tests were also used to compare the size of the same dendritic spines or axon terminals in the presence versus in the absence of microglial contact . Sample size ( n ) represents individual animals for EM ( except for correlation analysis , where n represents individual extracellular space and microglial process areas ) and synaptic elements or microglial contacts for two-photon analyses ( except for analysis of dendritic spine turnover , where n represents individual animals ) . | Microglia are important players in immune responses to brain injury . In the event of pathological insults , microglia rapidly become activated and acquire the ability to release various inflammatory molecules that influence neuronal survival as well as synaptic function and plasticity . Similarly to macrophages in other areas of the body , activated microglia can engulf , or phagocytose , cellular debris and are believed to eliminate synapses . In the absence of pathological insult , microglia are more quiescent , but still , these immune surveillants continually sample their surrounding environment and contact neighboring cells and synapses . To further explore the roles of microglia at synapses under non-pathological conditions , we used quantitative electron microscopy and two-photon in vivo imaging to characterize the interactions between quiescent microglia and synaptic elements in the visual cortex of juvenile mice . We also examined the “activity-dependent” processes involved by preventing light exposure in a group of mice . We show surprising changes in microglial behavior during alterations in visual experience , such as increased phagocytosis of synaptic elements and interaction with subsets of structurally dynamic and transient synapses . These observations suggest that microglia may participate in the modification or elimination of synaptic structures , and therefore actively contribute to learning and memory in the healthy brain . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"cell",
"biology/neuronal",
"and",
"glial",
"cell",
"biology",
"neuroscience/sensory",
"systems",
"immunology/immune",
"response",
"neuroscience/neurodevelopment",
"cell",
"biology/extra-cellular",
"matrix",
"neuroscience/neuronal",
"and",
"glial",
"cell",
"biology",
"neuroscie... | 2010 | Microglial Interactions with Synapses Are Modulated by Visual Experience |
We undertook a population genetics analysis of the tsetse fly Glossina palpalis gambiensis , a major vector of sleeping sickness in West Africa , using microsatellite and mitochondrial DNA markers . Our aims were to estimate effective population size and the degree of isolation between coastal sites on the mainland of Guinea and Loos Islands . The sampling locations encompassed Dubréka , the area with the highest Human African Trypanosomosis ( HAT ) prevalence in West Africa , mangrove and savannah sites on the mainland , and two islands , Fotoba and Kassa , within the Loos archipelago . These data are discussed with respect to the feasibility and sustainability of control strategies in those sites currently experiencing , or at risk of , sleeping sickness . We found very low migration rates between sites except between those sampled around the Dubréka area that seems to contain a widely dispersed and panmictic population . In the Kassa island samples , various effective population size estimates all converged on surprisingly small values ( 10<Ne<30 ) that suggest either a recent bottleneck , and/or other biological or ecological factors such as strong variance in the reproductive success of individuals . Whatever their origin , the small effective population sizes suggest high levels of inbreeding in tsetse flies within the island samples in marked contrast to the large diffuse deme in Dubréka zones . We discuss how these genetic results suggest that different tsetse control strategies should be applied on the mainland and islands .
Mating pattern , population size and migration represent key factors determining the population genetic structure of organisms and shape the evolutionary of species [1]–[4] . Estimating these parameters is a major objective of population and conservation genetics [5]–[7] . Molecular markers are useful for estimating these parameters without the need for costly capture-mark-release-recapture ( MRR ) studies; particularly for organisms such as parasites and their vectors , where techniques such as MRR are difficult , impossible or unethical to apply [2] , [4] . Furthermore a detailed understanding of parasite and vector population dynamics is crucial for effective sustainable control [2] , [8]–[10] . The World Health Organisation recently launched a Human African Trypanosomosis ( HAT , or sleeping sickness ) elimination programme to counter the recent decline in case detection and treatment , notably in Central Africa [11]–[12] . However the situation in West Africa and the epidemiology of HAT is less well described . Guinea ( especially the coastal area ) is believed to be the country most affected by this disease [13] . Guinea has a long history of sleeping sickness , which was particularly prevalent in the years 1930–40 [14] . Current data show prevalences of between 2 and 5% in villages in the coastal mangrove area ( Dubreka focus ) [15] . This coastal area is composed of mangrove on the coastal margins and savannah inland . Offshore but in close proximity to the Dubreka focus lie the Loos Islands . These islands , physically separated from the mainland about 5000 years ago ( D . Bazzo , pers . com . ) are known to harbour tsetse flies ( Glossina palpalis gambiensis ( Diptera: Glossinidae ) , the main vector of HAT in West Africa and the Guinean National Control Programme against HAT recently launched an tsetse elimination programme on the archipelago . To facilitate the work of the elimination programme we used microsatellite and mitochondrial markers to address the following questions which are key to the successful control of tsetse: What is the effective population size in this tsetse species ? What is the extent of genetic differentiation between mainland sites , between the islands and the mainland , and between the different islands of Loos archipelago ? By answering these questions we hope to improve the design of control strategies in the region , especially with respect to designing and implementing area wide strategies which must target genetically isolated populations if elimination is the objective [16] . Our results suggest that tsetse elimination is a feasible strategy on the Loos islands given both the genetic isolation between island and mainland populations and the small total surface to be controlled , but transmission reduction rather than elimination is more advisable for mainland tsetse populations .
The Loos islands are a small archipelago of five islands separated from the mainland of Guinea and the capital Conakry , by 4 km of sea . Three of the islands are inhabited , Kassa , Fotoba and Room in order of decreasing population size , with a total of around 7 , 000 inhabitants . On Kassa Island , two areas were sampled for tsetse , one in the north and one in the south . The biggest focus of HAT in Guinea , Dubréka , is on the mainland in a mangrove some 30 km distant from the Loos Islands . The area around Dubreka is characterised by coastal mangrove , with anthropic Guinean savannah , and permanently or temporarily inundated areas . Near the town of Dubréka ( 25 , 000 inhabitants ) , people live in villages of between 300 to 2 , 000 inhabitants . The main economic activities include fishing , salt extraction , and agriculture ( palm and mango plantations , rice and food crops ) . In the Dubréka area , tsetse were sampled in 12 sites from two main areas: Touguissoury ( five sites ) , in the mangrove habitat and accessible only by boat , and Magnokhoun ( six sites ) which is at the boundary between mangrove and savannah . A savannah area comprising a forest gallery bordering a water course was also sampled: Falessadé ( one site ) , 30 km from the mangrove areas of Magnokhoun ( see Figure 1 ) . Tsetse collections were made at each location using Vavoua traps [17] . Collecting cages were changed daily over a period of two to four days , and tsetse were counted and separated by sex . Three legs were removed from each fly and put in individual , labelled , dry eppendorf tubes . All the continental samples and those from Fotoba Island were taken in 2005 . Temporal samples were taken in Loos islands: 2005 and 2006 for Fotoba , 2006 and 2007 for Kassa . A total of 195 individuals were used for the genetic analyses at microsatellite loci: 7 males ( M ) and 15 females ( F ) in Kassa 2006 , 14 M and 11 F in Fotoba 2005 , 7F and 14M in Fotoba 2006 , 18F and 12M in Kassa 2007 , 17M and 15F in Magnokhoun 2005 , 17M and 17F in Touguissoury 2005 , and 11M and 20F in Falessadé 2005 . Ten microsatellite loci were analysed: Gpg55 , 3 [18] , pGp24 , pGp 13 , pGp11 , pGp1 [19] , C102 , B104 , B110 ( kindly given by A . S . Robinson ) , GpCAG [20] , and A10 kindly provided by G . Caccone . Locus Gpg55 , 3 has been reported to be located on the X chromosome [21] , and given an absence of heterozygotes on a subsample of males ( data not shown ) , B104 , B110 , Pgp13 and pgp11 were also interpreted to be located on the X chromosome . For ease of reference we renamed X linked loci XGpg55 , 3 , XB104 , XB110 , XPgp13 and XpGp11 . Because loci A10 and pGp1 were unavailable before 2007 , these loci were only used for 2007 sample ( Kassa 2007 ) and thus only influenced local results ( Linkage disequilibrium and FIS analyses ) . In each tube containing three legs of the tsetse , 200 µl of 5% Chelex chelating resin was added [22]–[23] . After incubation at 56°C for one hour , DNA was denatured at 95°C for 30 min . The tubes were then centrifuged at 12 , 000 g for two min and frozen for later analysis . The PCR reactions were carried out in a thermocycler ( MJ Research , Cambridge , UK ) in 10 µl final volume , using 1 µl of the supernatant from the extraction step . After PCR amplification , allele bands were resolved on a 4300 DNA Analysis System ( LI-COR , Lincoln , NE ) after migration on 96-lane reloadable 6 . 5% denaturing polyacrylamide gels . This method allows multiplexing of loci by the use of two infrared dyes ( IRDye ) , separated by 100 nm ( 700 and 800 nm ) , and read by a two channel detection system that uses two separate lasers and detectors to eliminate errors due to fluorescence overlap . To determine the different allele sizes , a large panel of about 30 size markers was used . These size markers had been previously generated by cloning alleles from individual tsetse flies into pGEM-T Easy Vector ( Promega Corporation , Madison , WI , USA ) . Three clones of each allele were sequenced using the T7 primer and the Big Dye Terminator Cycle Sequencing Ready Reaction Kit ( PE Applied Biosystems , Foster City , CA , USA ) . Sequences were analysed on a PE Applied Biosystems 310 automatic DNA sequencer ( PE Applied Biosystems ) and the exact size of each cloned allele was determined . PCR products from these cloned alleles were run in the same acrylamide gel as the samples , allowing the allele size of the samples to be determined accurately . Linkage disequilibrium between pairs of loci was tested under Fstat 2 . 9 . 3 . 2 [24] , updated from [25] by randomising loci ( free recombination ) with a G ( log-likelihood ratio ) based test allowing to get , for each pair of loci , a global test across sub-samples . For this analysis , the sub-sample unit was the smallest available one ( e . g . the trap in Dubreka ) . Because this procedure involves multiple testing the P-values obtained were adjusted with a sequential Bonferroni procedure [26] ( see [2] and references therein for detailed information ) . A binomial test was used to check if the proportion of significant tests was significantly greater than expected based upon a 5% significance level ( see [27] ) . Wright's F-statistics , the parameters most widely used to describe population structure [28] , were initially defined for a three levels hierarchical population structure ( individuals , sub-populations and total ) . In such a structure , three fixation indices or F-statistics can be defined: FIS is a measure of the inbreeding of individuals ( hence I ) resulting from the deviation from panmixia ( random union of gametes ) within each sub-population ( hence S ) . FST is a measure of inbreeding of individuals due to the structure of the population ( non-random distribution of individuals among sub-populations ) ; FST also quantifies the differentiation between sub-populations in the total population ( hence S and T ) . FIT is a measure of the inbreeding of individuals resulting both from non-random union of gametes within sub-populations and from population structure ( deviation from panmixia of all individuals of the total population , hence I and T ) . These F-statistics are classically estimated by Weir and Cockerham's unbiased estimators f ( for FIS ) , θ ( for FST ) and F ( for FIT ) [29] . When appropriate , these statistics were estimated with Fstat 2 . 9 . 3 . 2 . Bilateral sex-biased dispersal tests were done in Fstat 2 . 9 . 3 . 2 with the FST based test and the mean assignment index ( the multilocus probability of belonging to the sampling site ) corrected for population effects and its variance ( AIc and vAIc ) [30] , as recommended in [24] . We used 10000 permutations of individuals within samples [24] and applied the tests in continental samples considering each trap containing more than three flies or all traps from the same village as the sub-population unit ( two analyses ) . More than three levels ( i . e . individuals , sub-populations and total ) exist in the tsetse samples near Dubreka . Here , individuals were sampled using traps , in sites that are located in different “districts” ( i . e . Magnokhoun , Touguissoury ) . HierFstat version 0 . 03–2 [31] is an analytical package written in R [32] . This package computes hierarchical F-statistics from any number of hierarchical levels . The significance of FT/S , the homozygosity due to subdivision into different traps within sites , was tested by randomising individuals among traps of the same site . The significance of FS/D , which measures the relative homozygosity due to the geographical separation between sites within districts , was tested by randomizing traps ( with all individuals it contains ) among the different sites in the same district . Finally , FD/T measures the relative homozygosity due to the subdivision into different districts , and was tested by randomising districts in the total sampling area . A gentle step by step description of how using HierFstat can be found in [2] . The significance of the F-statistics was tested by randomization ( 10000 permutations in each case ) . The significance of FIS was tested randomising alleles between individuals within sub-samples . The significance of FST was tested by randomising individuals among sub-samples . These tests were performed with Fstat 2 . 9 . 3 . 2 . For FIS the statistic used was directly the f ( FIS estimator ) . For FST ( and other differentiation tests ) , the statistic used was the maximum likelihood ratio G [33] . Differentiation between the northern and southern samples in Kassa were analysed by paired FST ( G based test ) and tested in each year ( 2006 , 2007 ) with Fstat 2 . 9 . 3 . 2 . The two FST were combined with an unweighted mean and the corresponding P-values with Fisher's procedure [34] as described in [2] . Non random association of alleles within individuals ( FIS>0 ) may be due to null alleles . We used Micro-Checker 2 . 2 . 3 [35] to detect null alleles and estimate their frequency pn at each locus according to Brookfield's second method [36] . We compared , at each locus , the expected frequencies of blanks ( i . e . null allele homozygotes ) under panmixia ( pn2 ) with the blank individuals observed using a binomial exact test . For X-linked loci we compared the number of blanks observed with the expected one as computed with null allele frequency found from female data with Micro-Checker . In that case a binomial test was also undertaken with the direct null allele estimates provided by the frequency of blanks in males at such loci . For the sake of power , all binomial tests were undertaken over all sub-samples ( with mean expected frequencies and total observed blanks ) and were one-tailed ( H1: there are fewer blanks than expected ) . Short allele dominance may also explain a significant part of heterozygote deficits . It was tested using a multiple regression approach of FIS observed at each allele at the locus of interest as a function of allele size and sub-sample , following the procedure of [37] . This was made under S-Plus 2000 professional release 2 ( MathSoft Inc . ) . For X linked loci , only females were considered . Because highly polymorphic microsatellite loci are used , the level of differentiation as measured by FST may be constrained ( e . g . [2] ) . We thus used a “corrected” version of this statistics FST′ = FST/ ( 1−Hs ) where Hs is Nei's unbiaised estimator of genetic diversity and 1−Hs corresponds to the maximum possible value for FST in a model with many completely isolated sub-populations ( see [38]–[39] , as in [40] ) . The effective population size , usually noted Ne , is a measure of the rate at which a population looses genetic diversity by drift and roughly represent the number of adults that effectively contribute to the next generation ( see [2] for a more precise definition and examples ) . Effective sub-population sizes could be estimated in each site using various methods: the linkage disequilibrium methods of Bartley et al . ( LDB ) [41] and of Waples and Do ( LDWD ) [42] , the temporal moment based method of Waples ( 1989 ) ( Temporal ) [43] for sites sampled at different times , joint estimation of migration and effective population size with the maximum likelihood ( ML ) , the moment based ( Moment ) methods of Wang and Whitlock [44] and the method of Vitalis and Couvet ( 2001 ) ( Estim ) [6] , [45]–[46] . Bartley's and Waples' methods were implemented with NeEstimator [47] , Waples and Do's method with LDNe [42] . Wang and Whitlock's methods were implemented with MLNE v 1 . 1 . , and Vitalis and Couvet's method was implemented by Estim 1 . 2 [45] . For temporal based methods six generations were assumed to separate tsetse flies in one year interval . We also estimated m ( migration rate ) from Nem with the formula Nem = ( 1−FST ) / ( 8FST ) that is appropriate for two populations and probably more appropriate here between the two Loos islands in 2006 , between North and South in Kassa ( 2006 and 2007 ) and between Fotoba island and the mainland samples in 2005 . We used Ne from the Temporal method to extract m . Implementing all these methods that work under more or less different assumptions allowed the comparison of the values obtained and gaining some confidence on the parameters' range . MLNE dataset was obtained using CREATE 1 . 0 [48] . For LDWD method , values obtained for alleles at least as frequent as 0 . 05 were chosen . Signatures of bottleneck events were investigated by comparing the expected heterozygosity for a sample ( HE ) with the heterozygosity that would be expected for a sample taken in a population at mutation/drift equilibrium with the same size and allele number ( HEQ ) : as allele number decreases faster than heterozygosity , bottlenecks are indicated by HE>HEQ in subsequent generations [49] . This analysis was performed using Bottleneck software [50] under an IAM ( infinite allele model ) , a SMM ( stepwise mutation model ) or a TPM ( two phase model ) , in the latter case we assumed that 70% of mutations consist of one step and 30% consist of multistep change with a variance of 30 ( default values ) . Significance was assigned using one-tailed Wilcoxon tests [49] . Global P-values , overall Fotoba 2005–2006 and overall Kassa-North 2006–2007 samples were obtained with the Fisher procedure [34] . Given it only had three individuals , the Kassa-South 2006 sub-sample was excluded from these analyses . Given our sample sizes and number of loci , bottleneck detection is only possible if it occurred between t1 = 0 . 025×2Ne and t2 = 2 . 5*2Ne generations ago [49] , where Ne represents the post-bottleneck effective population size . We compared these generation times to those believed to have occurred on Kassa Island since the 1960's ( i . e . 276 generations ago ) when an important bauxite mining activity is thought to have strongly altered ecological conditions ( see http://www . nationsencyclopedia . com/Africa/Guinea-MINING . html ) . This provided a possible Ne included between 55 and 5520 individuals . A portion of the 5′ end of the mitochondrial gene cytochrome oxidase 1 was amplified using the primers CI-J-2195 TTGATTTTTTGGTCATCCAGAAGT [51] and CULR TGAAGCTTAAATTCATTGCACTAATC . Double distilled water containing 10× PCR buffer ( Bioline ) , dNTP 0 . 8 mM , primers 0 . 5 µM each , MgCl2 3 mM was incubated with 0 . 25units of BIOTaq DNA polymerase and approximately 0 . 5 ng of template DNA in 25 µl reactions . Temperature cycles were 5 min 95°C , 35 cycles of 93°C for 1 min , 55°C for 1 min and 72°C for 2 min , then 72°C for 7 min . PCR products were purified using the Bioline SURECLEAN reagent ( BIO-37046 ) according to the manufacturer's instructions , and sequenced using an ABI3730XL sequencing machine ( Macrogen ) . Each template was sequenced bi-directionally . Sequence traces were checked using Codoncode Aligner ( CodonCode Corporation ) , and aligned using the ClustalW algorithm implemented in MEGA version 4 with the following settings: gap opening penalty15 , gap extension penalty 6 . 6 , IUB weight matrix , transition weight 0 . 5 , delay divergent cut-off 30 [52] . The PCR product size is 850 bp , but for analysis the alignment was trimmed to 723 bp of good quality sequence . The following statistics were calculated using DNAsp: FST_Seq was calculated according to equation 3 in [53] and is comparable to Weir and Cockerham's FST estimator for sequence data [29] . HST , an equivalent of Nei's estimator of FST ( GST ) was calculated according to equation 2–4 , and KST* according to equations 7–11 in [54] and is an equivalent of Nei's sequence statistic γST [55] . A permutation test , in which haplotypes or sequences were randomly assigned to the different localities 10000 times , was used to test the significance of HST and KST * [54] . Mitochondrial analysis could only be undertaken with 10 individuals from Fotoba 2005 , five individuals from Touguissoury and five individuals from Magnokhoun .
The only tsetse species caught was G . p . gambiensis . Entomological surveys gave mean catches of flies per trap per day ( FTD ) of 10 in Kassa and 1 in Fotoba for Loos islands . On the mainland , mean FTDs were 7 . 5 in Magnokhoun , 5 . 5 in Touguissoury , and 11 in Falessadé . No sex biased dispersal was found ( all P-values>0 . 05 ) in 2005 samples from continental sites . Therefore in all further analyses data from females and males is combined . Among the 36 tests of linkage disequilibrium between paired loci ( locus XB110 was excluded due to insufficient polymorphism ) , only two pairs were in significant linkage that did not stay significant after Bonferroni correction ( PBinomial , 2 , 36 , 0 . 05 = 0 . 5433 ) . HierFstat analyses gave no effect for district ( FD/T = 0 . 005 , P = 0 . 21 ) , site ( FS/D = 0 . 016 , P = 0 . 454 ) or traps ( FT/S = −0 . 006 , P = 0 . 90 ) . Thus , individual tsetse flies from Touguissoury and Magnokhoun ( Dubréka focus ) were considered to belong to the same population for the following analyses . In Kassa , over 2006 and 2007 a substantial differentiation could be seen between northern and southern samples ( FST = 0 . 095 , P = 0 . 018 ) . FIS analysis revealed a significant excess of homozygosity , variable across loci and significant for X55 . 3 , XpGp11 , pGp24 , XB110 , pGp1 and A10 ( Figure 2 ) . All but XB110 were reasonably explained by the presence of null alleles ( Table 1 ) . For XB110 , the binomial test was only significant with “Males” method , which could be explained by Type I error , given the number of tests undertaken . There was no evidence for short allele dominance at this locus . Excluding the six loci with null alleles provided a much smaller deviation of heterozygote frequency from panmictic expectation ( FIS = 0 . 04 , P = 0 . 069 ) ( Figure 2 ) . Differentiation was significantly positive between each continental site and Loos islands ( 2005 samples , P<0 . 0001 ) . In Table 2 are presented the paired FST , mean between paired samples and paired . The highest levels of differentiation were found between Loos islands and all the other sites . Geographical differentiation was highly significant for each pair of samples and the smallest ( although highly significant ) value was observed between continental sub-samples ( Falessadé and Dubréka ) . Within the Loos islands , there was a high and significant differentiation between Kassa and Fotoba , the two main islands . It is noteworthy that the same analysis undertaken with the four loci without null alleles ( XpGp13 , XB104 , C102 and GpCAG ) provided very similar results ( not shown ) , with the exception of the temporal analysis for which the weak differentiation observed in Kassa between 2006 and 2007 was no longer significant . Genotypes at microatellite loci of all the individuals analysed can be seen in Table S1 . A fragment of the mitochondrial Cytochrome Oydase I gene was amplified by PCR and sequenced ( see Table S2 ) . COI sequences were deposited in Genbank ( accession numbers FJ387505-FJ387524 ) . The statistics of genetic differentiation based on COI are presented in Table 3 . Within Dubréka , as with nuclear microsatellite markers , Touguissoury 2005 and Magnokhoun 2005 showed an absence of differentiation . Differentiation between Fotoba and Dubréka was strong and significant ( Table 3 ) and comparable to that observed using microsatellites ( Table 2 ) . Effective population size estimates with different methods are given in Table 4 . These results were obtained for all loci , including those with null alleles . Using the four loci without null alleles or even the two autosomal loci with no deviation from Hardy-Weinberg ( C102 and GpCAG ) greatly widened confidence intervals but provided mean estimates of the same order of magnitude as with the complete data set ( not shown ) . Table 5 gives migration rate ( m ) estimates for the three sites ( Fotoba , Kassa North and Kassa South ) for which it was possible . Except for LDWD and Estim for which most Ne were infinite or undefined , the different methods gave consistent Ne and m estimates , particularly for Fotoba where the four available methods converged to very similar estimates . On the mainland , Dubréka seemed to harbour a large effective population of tsetse flies ( 475<Ne<2016 ) while Falessadé showed a notably small estimate ( 25<Ne<63 ) . Loos islands displayed surprisingly small effective population sizes ( Table 4 , 2<Ne<145 ) . Bottleneck signatures were only detected with the IAM model in Kassa North sub-sample with a global P-value ( Fisher's procedure ) of 0 . 022 . Thus , bottleneck detection in Kassa North is suggestive of a post bottleneck effective population size range ( 55-5520 ) that overlaps with what was suggested by other methods , most particularly with temporal based methods ( Table 4 ) , but with of course a much higher upper bound .
The high and significant genetic differentiation found between the Loos islands and the mainland combined with the small total surface area to be treated ( around 15 sq km ) , has led the Guinean National Control Programme to launch a tsetse elimination campaign ( M . CAMARA , pers . com . ) . We show here that within the Loos islands , the high and significant differentiation found between Fotoba and Kassa suggests a low number of migrants between islands ( probably less than one per generation ) . This result , combined with morphometric data suggest that the elimination can be based on the implementation of an area wide approach using a sequential control strategy on each of the islands in turn [16] , [40] . On the mainland , all sites in Dubréka could be considered as within a single reproductive unit . These sites were significantly differentiated but not completely isolated , from Falessadé ( Savannah site ) . This suggests that , contrarily to what was observed on Loos islands , tsetse elimination in this mainland area should not yet be chosen as the ideal strategy so far because exchanges of flies between these sites may still occur through the dense hydrographic network present . An elimination programme in the area would then require to create artificial barriers around the area to be treated ( made of insecticide impregnated traps for instance , see [56] ) , which would prevent immigration into this area . Such a programme would require more detailed sampling . Alternatively , tsetse control with the participation of local communities to reduce ( but not eliminate ) tsetse densities may be advised since this has been shown to be technically feasible [57] . Over all sample sites and loci in our Guinean samples , the strong and significant FIS found was reasonably explained by null alleles , it was thus not necessary to invoke a Wahlund effect , in contrast to results obtained in Burkina Faso for the same tsetse species [22] , [58] or for the closely related subspecies G . p . palpalis in Côte d'Ivoire [59] . In Guinea , this could be explained by the high rainfall ( 3000 mm /annum ) and large numbers of river and stream habitats , combined with good host availability allowing good dispersal conditions and a less restricted distribution . The exception to this would be on Kassa , where a human settlement that led to habitat fragmentation may have caused a slight Wahlund effect ( in this island mean FIS = 0 . 092 , P = 0 . 058 ) , which may also explain why Ne estimated with LD methods were often lower than those estimated with temporal based methods . We found surprisingly low estimates of Ne . To our knowledge , such estimates were made only once in tsetse , for a savannah species belonging to the morsitans group in East Africa , G . swynnertoni [60] , which is phylogenetically and ecologically very far from G . palpalis ( riverine fly of the palpalis group ) [61] . Using mitochondrial markers , Marquez et al . , 2006 found very low estimates of population size , and attributed it to a recent bottleneck . In Kassa Island , tsetse seemed structured in fairly isolated populations that may explain why extremely small Ne were found . However , Kassa Island is the place where tsetse densities were the highest ( mean of 10 flies / trap / day , up to 100 in some traps ) . In North Kassa , the possible signature for a Bottleneck that occurred about 276 generations ago led to a higher estimate of post bottleneck effective population size ( between 55 and 5520 ) . The only explanation we can provide is that intense bauxite mining activity in Kassa caused a drastic reduction in G . p . gambiensis populations at that time ( year 1960 corresponds to 276 generations from 2006 sampling using 6 generations per year ) . Indeed it is very well known that bauxite mining is very destructive for the environment ( e . g . http://www . idrc . ca/en/ev-31010-201-1-DO_TOPIC . html ) . In contrast on Fotoba , which was not involved in bauxite mining activity , no bottleneck could be detected . After the end of these mining activities in Kassa , tsetse populations would have recovered , and reached high densities again , aided by the important pig rearing activity which began on Kassa ( but not in Fotoba ) since the 1980s . A significant correlation has been observed in Kassa between tsetse densities and pig rearing presence [62] . No signature of a bottleneck was found in any other site . Reduction in effective population size can occur in the case of variance in reproductive success . In the case of tsetse flies , where a modal number of four larval progeny per female can be assumed , there is more opportunity for variance in male mating success . In Appendix S1 we have derived a very simple model aimed at illustrating the drop in Ne as a function of the census size Nc and number of sired females per successful males nffec that would result when some males sire several females while the others do not . We used North Kassa estimate of “real” Ne , obtained from the bottleneck procedure , to estimate a possible range for nffec using equation ( 5 ) from Appendix S1 . This estimate ranged between four and 490 females mated per successful males . Table 6 provides the results obtained for Nc in each sub-sample when applying equation ( 6 ) of Appendix S1 . Apparently , variance in male reproductive success must be very high if this has to explain all small Ne . A variance in female reproductive success may also act but is less probable since when females were dissected , most of them were found to be pregnant ( tsetse are viviparous ) ( data not shown ) . Whatever the cause of these small effective population sizes , they are suggestive of significant ( very high in certain sites such as in Kassa Island ) levels of inbreeding in G . palpalis gambiensis populations from coastal Guinea . These results suggest that further analyses should be conducted on the potential of G . p . gambiensis to maintain genetic diversity at local and global scales in Guinea , in particular regarding interactions with the aetiological agent of HAT , Trypanosoma brucei gambiense , the epidemiology of which is known to vary substantially through West Africa and in Guinea in particular [13] , [15] . We also hope that , within the context of the Pan African eliminations programmes that have been launched by the WHO [11]–[12] and the African Union PATTEC ( Pan African Tsetse and Trypanosomosis Eradication Campaign , [63] ) , more studies will be conducted on tsetse population genetics . Indeed , control programmes are beginning to recognise the importance of such data in helping to choose specific control strategies . | Guinea is the country with the highest prevalence of sleeping sickness in West Africa , and we undertook a population genetics analysis there of the most dangerous tsetse fly species of West Africa , Glossina palpalis gambiensis . Our aims were to estimate effective population size and the degree of isolation between coastal sites on the mainland of Guinea ( including Dubréka , a highly prevalent sleeping sickness focus ) and Loos Islands in order to get the most possible accurate vision of feasibility and sustainability of anti-tsetse strategies of these sites . We found very low migration rates of tsetse between sites except between those situated in the Dubréka area , which seems to contain a widely distributed panmictic tsetse population ( i . e . a population where mating occurs at random ) . Effective population sizes on Loos islands estimated with various techniques all converged to surprisingly small values . These values might be explained by a recent decrease in tsetse numbers on Kassa Island due to bauxite mining activities . But on the other sites , other explanations have to be found , including possible variance in reproductive success . Our genetic results suggest that different control strategies should be advised on the mainland ( reduction in tsetse densities , no elimination ) compared to the islands ( total elimination feasible ) . This approach could be extended to many areas where vector control of Human and Animal Trypanosomoses is contemplated . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"infectious",
"diseases/neglected",
"tropical",
"diseases",
"public",
"health",
"and",
"epidemiology",
"genetics",
"and",
"genomics/population",
"genetics"
] | 2009 | The Population Structure of Glossina palpalis gambiensis from Island and Continental Locations in Coastal Guinea |
Individuals living in areas endemic for helminths are commonly infected with multiple species . Despite increasing emphasis given to the potential health impacts of polyparasitism , few studies have investigated the relative importance of household and environmental factors on the risk of helminth co-infection . Here , we present an investigation of exposure-related risk factors as sources of heterogeneity in the distribution of co-infection with Necator americanus and Schistosoma mansoni in a region of southeastern Brazil . Cross-sectional parasitological and socio-economic data from a community-based household survey were combined with remotely sensed environmental data using a geographical information system . Geo-statistical methods were used to explore patterns of mono- and co-infection with N . americanus and S . mansoni in the region . Bayesian hierarchical models were then developed to identify risk factors for mono- and co-infection in relation to community-based survey data to assess their roles in explaining observed heterogeneity in mono and co-infection with these two helminth species . The majority of individuals had N . americanus ( 71 . 1% ) and/or S . mansoni ( 50 . 3% ) infection; 41 . 0% of individuals were co-infected with both helminths . Prevalence of co-infection with these two species varied substantially across the study area , and there was strong evidence of household clustering . Hierarchical multinomial models demonstrated that relative socio-economic status , household crowding , living in the eastern watershed and high Normalized Difference Vegetation Index ( NDVI ) were significantly associated with N . americanus and S . mansoni co-infection . These risk factors could , however , only account for an estimated 32% of variability between households . Our results demonstrate that variability in risk of N . americanus and S . mansoni co-infection between households cannot be entirely explained by exposure-related risk factors , emphasizing the possible role of other household factors in the heterogeneous distribution of helminth co-infection . Untangling the relative contribution of intrinsic host factors from household and environmental determinants therefore remains critical to our understanding of helminth epidemiology .
People living in poor areas of the tropics commonly harbour multiple parasitic infections , including infection with multiple helminth species [1] , [2] . An increasing number of studies demonstrate that individuals infected with multiple helminth species tend to harbour the most intense infections [3]–[11] and can be at an increased risk of infection-related morbidity [12]–[15] . For example , a study of Brazilian school children showed those harbouring concomitant infection with Ascaris lumbricoides and Trichuris trichiura were at increased risk of stunting [16] , whilst another Brazilian study found the risk of anaemia among school children infected with Schistosoma mansoni and two or three soil-transmitted helminth ( STH ) infections was significantly higher that those harbouring single STH species [12] . The occurrence of extensive polyparasitism in human communities also has important implications for a multiple infection approach to control [17] . Recent interest in the scientific study of polyparasitism has given renewed prominence to some old epidemiological questions; in particular identifying factors governing patterns of infection . A wealth of epidemiological investigation across numerous ecological and socio-economic settings indicate that certain characteristics are common to the epidemiology of single helminth species in communities , including household clustering and spatial heterogeneity [18] . Such features most likely result from the combined effects of extrinsic ( exposure to infection ) and intrinsic ( host resistance ) factors [18] , [19] . However , our understanding of the determinants of multiple helminth species infection patterns within communities remains poorly defined . For example , while recent studies have documented the prevalence of multiple helminth infections and their patterns by age and sex [3]–[10] , little is known about spatial and household clustering of multiple helminth infection within communities or putative risk factors [20] . In the paper , we investigate the spatial patterns and household clustering of helminth co-infection and associated risk factors among individuals living in the state of Minas Gerais , Brazil . A previous analysis has already highlighted the high frequency of multiple helminth infection in the area [21] . Although A . lumbricoides is also endemic to the region , we focus specifically on co-infection with the hookworm Necator americanus and S . mansoni since these species both contribute to iron-deficiency anaemia ( via distinct mechanisms [22] , [23] ) but have dissimilar life cycles and modes of transmission . First , we explore spatial patterns of co-infection with N . americanus and S . mansoni using spatial statistics . We then investigate the role of individual , household and environmental risk factors in explaining the observed heterogeneities in infection patterns using a multi-level Bayesian multinomial approach , whereby individuals are assumed to be clustered within households . This approach permits robust , unbiased investigation of within-household clustering .
The study was conducted from June to September 2004 in Americaninhas , a region in the municipality of Nova Oriente in the northeast of Minas Gerias state , which is situated in southeast Brazil ( Figure 1 ) . Details of the study area , recruitment method , and cross-sectional parasitological and questionnaire surveys have been provided elsewhere [21] , [24] , [25] and only a summary is given here . The area is hilly and has an average temperature of 24°C , with a rainy season between November and March; annual rainfall is 1300–2000 mm . The study area is divided by a high ridge of land running north-south , separating the study area into two distinct zones or watersheds . The majority of inhabitants are involved in rural subsistence farming; cattle ranching is another important source of income . A series of meetings was held with community members to explain the purpose of the study . that participation was voluntary and that participants were able to withdraw from the study at any time . Written or oral consent was obtained from all adult subjects and from parents or guardians of minors . A pre-tested standardized questionnaire was administered to the head of each household to collect information on household socio-economic characteristics including house construction , water and sanitation , parental education , and ownership of selected household assets . During the parasitological survey , stool samples were collected over the course of two days ( if possible ) and were initially examined using the formalin-ether sedimentation technique for the presence of helminth eggs . Individuals positive for any helminth infection were subsequently examined by the Kato–Katz faecal thick smear technique to quantify the intensity of the infection expressed as eggs per gram of faeces ( epg ) . Two slides were taken from each day's faecal sample for a total of up to four slides from each individual . Morphological examination of expelled worms following treatment among a sub-sample of individuals showed that hookworm infection was exclusively of the species Necator americanus [25] . A polymerase-chain-reaction ( PCR ) test was performed on 20 of the above samples to confirm the morphological examination as Necator americanus [26] . N . americanus was found in 100% of these samples , no A . duodenale infection was found . Household locations were mapped using a hand-held Trimble GeoExplorer global positioning system ( GPS ) receiver ( Trimble Navigation , Sunnyvale , CA , USA ) and ArcPad 6 . 0 . 3 software ( Environmental Systems Research Institute Inc . , Redlands , CA , USA ) . Readings , with a resolution of 5 m , were taken at the front door , or as near as possible in order to receive a sufficient satellite reception and an average of 10 readings of the co-ordinates were taken . Remotely sensed proxy environmental data were extracted for May 2001 from the Advanced Spaceborne Thermal Emission and Reflection Radiometer ( ASTER ) satellite sensor at 30 m spatial resolution ( http://edcdaac . usgs . gov/aster/asterdataprod . asp ) . ASTER provides information on Normalized Difference Vegetation Index ( NDVI ) , a proxy of vegetation density and soil moisture , and digital elevation [27] . The GIS was compiled and all maps were created using ArcView 3 . 3 ( Environmental Systems Research Institute Inc . , Redlands , CA , USA ) . The study was reviewed and approved by the ethical committee of the Centro de Pesquisas René Rachou-FIOCRUZ and the Brazilian National Committee for Ethics in Research ( CONEP ) , and the ethical review boards of George Washington University ( USA ) and London School of Hygiene and Tropical Medicine ( UK ) . Individuals found to be infected with any soil transmitted helminth or with S . mansoni were treated with a single dose of 400 mg albendazole and 40 mg/kg praziquantel , respectively . Participants were recorded as positive for an infection with S . mansoni or N . americanus if at least one egg was detected by either formalin-ether sedimentation or Kato–Katz faecal thick smear . Participants were classified into five age-groups: under 5 years , younger children ( 5–9 years ) , older children ( 10–19 years ) , adults ( 20–59 years ) and over 60 years . Information on ownership of household assets was used to construct a wealth index using principal component analysis , using the method of Filmer and Pritchett [28] . Following this approach , households were divided into tertiles , to provide a categorical measure of relative socio-economic status . Household factors potentially directly associated with infection outcomes ( such as toilet facilities and household construction ) were not included in the wealth index to allow for independent assessment of their involvement: details of the derived wealth index are provided elsewhere [25] . Information from the digital elevation model was used to divide households into either the eastern or western watershed . Housing density calculated in ArcView 3 . 3 was used to categorise households as urban ( >55 households within 1 km of the household ) , rural ( 5–55 households within 1 km ) and isolated ( <5 households within 1 km ) , with cut-offs chosen to reflect the distribution of households within the study region . As an outcome measure a ( mutually exclusive ) multi-categorical response for infection status was constructed as follows: ( i ) no infection , ( ii ) mono-infection with N . americanus , ( iii ) mono-infection with S . mansoni and ( iv ) co-infection with N . americanus and S . mansoni . In order to assess the importance of demographic , socio-economic and environmental risk factors on the occurrence of mono- and co-infection simultaneously we used a multinomial modelling approach , which extends logistic regression by estimating the effects of explanatory variables on the probability that the outcome is in a particular category . Initially , for each covariate frequentist unadjusted multinomial models were fit on the outcome in Stata 9 . 1 ( College Station Texas , USA ) , and covariates with P>0 . 2 ( Wald test ) were excluded from further analysis . Standard errors were adjusted for dependence between individuals within households . Scatter-plots and the entry of categorised predictor variables were used to investigate non-linear relationships . Subsequently , the retained covariates were built into a Bayesian multinomial mixed effect model in WinBUGS Version 14 ( MRC Biostatistics Unit , Cambridge , UK ) . To account for dependence of individuals within households , household was included as a random effect . We employed a Bayesian Monte Carlo Markov Chain ( MCMC ) approach , which readily allows the development of complex random effects models [29] . Age and sex were retained in all models during the model identification process . Variables were added to the models in a forward stepwise fashion , comparing the statistical fits of alternative ( nested and un-nested ) models using both the residual deviance of the models and the Deviance Information Criteria ( DIC; where a lower value indicates a better compromise between model fit and parsimony ) . A hierarchical approach was adopted when entering collinear predictor variables , whereby distal determinants ( such as relative socio-economic status ) are included prior to more proximal determinants ( such as crowding and sanitation ) [30] . Detailed descriptions of the Bayesian hierarchical models and the process of model assessment are described in Appendix S1 . Spatial heterogeneity ( or structure ) refers to the spatially non-random distribution of infection across the study region , such that an individual's risk of infection may be more similar to those living close to them that those living farther away . Such spatial clustering is not necessarily synonymous with clustering within households , because , whilst individuals in the same household may have more similar risk than individuals in different households , household-level risk may or may not be spatially autocorrelated . In order to examine the spatial structure of co-infection with N . americanus and S . mansoni at the household level , semi-variograms were generated using the R module GeoR on the basis of household prevalence of mono-infection with N . americanus and S . mansoni and co-infection with both parasites . Before variography , the data was de-trended by regressing against longitude and latitude , in order to remove large-scale spatial trends . Semi-variograms present the semi-variance ( i . e . half the mean squared difference ) of pairs of observations that are separated by the same distance; thus , describing how similar observations are at different spatial distances [31] . If there is spatial autocorrelation in the data semi-variance increases with separation distance; levelling out of the semi-variogram indicates the distance beyond which spatial autocorrelation ceases to occur . When the semi-variogram appears to show little or no spatial autocorrelation , Monte Carlo envelopes ( computed from random permutations of the residuals from random permutations of the data holding the corresponding locations fixed ) can be used to assess more formally whether the data are compatible with spatial structure , under the assumption of no correlation [32] , [33] . If the variogram plot falls within the envelope , there is no evidence of spatial autocorrelation at that distance .
Of the 1687 residents of the mapped households , 1539 individuals provided stool samples . Sixteen households ( 59 residents ) in the far south-east of the study site were excluded from analysis because cloud-free satellite data were not available . Socio-economic data were unavailable for a further 275 individuals , who mainly lived in the urban municipality . As such , 1208 individuals living in 275 households had complete data . Households with GPS positions less than 10 m apart were treated as a single spatial unit , providing data for 230 locations for spatial analysis , the largest of which had 16 residents . The majority of individuals were infected with helminths: 71 . 1% were infected with N . americanus , 50 . 3% had S . mansoni and 41 . 0% of individuals were co-infected with both helminths ( co-infection ) . 30 . 1% were infected with only N . americanus , and only 9 . 4% of individuals were infected with only S . mansoni . The prevalence of co-infection was significantly higher among males than females ( p<0 . 001 ) and increased significantly with increasing age , peaking among persons aged 20–59 years ( p<0 . 001 ) ( Table 1 ) . The occurrence of co-infection also varied considerably by household , with prevalence varying from 0–100% ( interquartile range: 0–67% ) ; in 13 . 5% of households , all residents were co-infected with N . americanus and S . mansoni . Figure 2 shows the spatial distribution of mono-infection with either N . americanus or S . mansoni or co-infection with both . The highest frequencies of co-infection were observed in the east of the study area , with an overall prevalence of 86 . 3% compared to 13 . 7% in the western watershed . To investigate the global spatial structure of infection patterns semi-variograms were estimated on the basis of household prevalence of mono-infection with N . americanus and S . mansoni and co-infection with both parasites . After removal of the large-scale spatial trend ( by regressing against longitude and latitude ) there was an apparent lack of any spatial structure for both N . americanus and S . mansoni mono-infection across all separation distances ( not shown ) . Likewise , the semi-variogram for co-infection provides no evidence of spatial dependency , indicating that once the large-scale trends were removed there was no general spatial structure in the distribution of co-infection ( Figure 3 ) . Relative frequencies of household and environmental factors are shown in Table 1 according to infection status . Unadjusted results from fixed effects multinomial analyses showed that characteristics associated with lower socioeconomic status ( SES index , toilet facilities , household crowding , flooring material ) , and residential environment ( living in the eastern watershed , in more densely populated areas , or in areas with less vegetation ) were significantly associated with both mono-infections and with co-infection ( p<0 . 01 ) . Posterior estimates from the adjusted analysis using a hierarchical Bayesian multinomial mixed effects model confirm that the risk of co-infection relative to being uninfected was highest among males and adults aged 20–59 years ( Table 2 ) . There was also evidence of an increased risk of co-infection among individuals resident in households with a lower socio-economic index and in overcrowded households . After accounting for relative socio-economic status , toilet facilities and flooring material were no longer significant due to considerable co-linearity between these variables . Individuals living in the eastern watershed were 6 . 9 times more likely to harbour a co-infection than those living in the western watershed , while those living in areas with less vegetation cover ( NDVI<0 . 2 ) were at reduced risk of co-infection . Associations between risk of infection and characteristics relating to lower socioeconomic status were observed for mono-infection with N . americanus , but not for S . mansoni mono-infection , while residential environment was associated with both mono-infections . There was significant household clustering for all outcomes , as indicated by estimates for the household level random effects; the highest degree of unexplained household-level variation was observed for co-infection . Household-level variance was substantially higher when household and environmental risk factors were excluded from the model ( Table 3 ) . Whilst 40% of household-level variation could be explained by relative socio-economic status and household crowding for N . americanus mono-infection ( i . e . inclusion of these covariates reduced the household-level variance parameter ui by 40% ) , substantially less household heterogeneity was explained by these factors for S . mansoni mono-infection ( 8% ) and co-infection ( 10% ) . In contrast , environmental factors ( living in the eastern watershed and areas with low NDVI ) explained 36% of household-level variation in N . americanus mono-infection , 45% in S . mansoni mono-infection but only 19% in co-infection . Household and environmental factors jointly explained 54% of household variation in N . americanus mono-infection , but only 39 . 5% for S . mansoni mono-infection and 31 . 9% for co-infection .
We employed a combination of spatial statistics and hierarchical multinomial modelling to investigate spatial patterns and household and environmental factors influencing occurrence of mono- and co-infection by the helminths N . americanus and S . mansoni . Our multi-level approach has the advantage of taking into account household clustering of infection , a commonly observed feature of helminth epidemiology [24] , [34] , [35] . The results suggest that , in addition to age and sex , characteristics associated with lower socioeconomic status ( relative socio-economic status , household crowding ) and residential environment ( living in the eastern watershed or in areas with less vegetation ) were significantly associated with the risk of co-infection relative to being uninfected with either species . Risk factors for co-infection reflected those associated with mono-infection , with no identified risk factors specific to co-infection . The results presented in Table 2 and Figure 2 provide strong evidence of household clustering of co-infection with N . americanus and S . mansoni . While household clustering of single helminth infections is well-documented [24] , [36] , [37] , the factors potentially responsible for such patterns remain less clear . Our observation that co-infection with N . americanus and S . mansoni is more common in households with lower relative socio-economic status is consistent with a study among schoolchildren in rural Cote d'Ivoire , which investigated school-level patterns in co-infection with N . americanus and S . mansoni [20] . Together , these studies suggest that socio-economic status influences the risk of co-infection at both household and local levels . The mechanisms through which socio-economic status influences infection risk are likely to reflect exposure-related factors , including poor hygienic behaviour , lack of clean water and inadequate sanitation , household construction ( e . g . cement or dirt floors ) and access to effective anthelmintics [38]–[41] . The increased risk of co-infection in households located in areas of higher NDVI ( indicative of increased humidity and soil moisture ) and in overcrowded households are consistent with previous studies reporting associations between hookworm and NDVI [42] and between helminth infection and overcrowding [43]–[45] . Our data demonstrated a dominant spatial trend ( NE-SW ) in household prevalence of co-infection ( Figure 2 ) , but there was little evidence of a second order spatial structure once this has been removed by regressing the data against latitude and longitude and plotting a semi-variogram of the model residuals ( Figure 3 ) . We suggest therefore that previous observations of small-scale spatial structure [42] probably reflect a combination of spatial variation in household characteristics and environmental risk factors . The absence of second order spatial structure is likely to reflect the high spatial resolution of the study , and it is plausible that in larger study areas and in areas with different eco-epidemiological and socio-economic characteristics , clearer spatial patterns may emerge; it would therefore be useful to investigate these issues in different epidemiological settings and at varying spatial scales . The dominant NE-SW trend in co-infection observed reflects the distribution of S . mansoni rather than N . americanus , which is more homogeneously distributed across the study area . The high prevalence of S . mansoni in the east of the study region is likely to reflect the increased infectivity of water bodies in this area . It has been frequently demonstrated that within communities high intensity Schistosoma infections can be found clustered around water bodies such as rivers and lakes [36] , [46] , [47] . A limitation of our study is the lack of information regarding infectious water sources . We were unable to find recent and geo-referenced topographic maps from the area under investigation at the desired scale and quality , and it was not possible to delineate water-bodies from our remotely sensed images . Household water sources in this region of Brazil are typically small and private to each household , thus making them difficult to identify; this is reflected by the absence of large-scale spatial correlation between locations , suggesting that there are few large transmission sites ( such as large , communal water sources ) shared by many widely spaced households . In terms of extrapolation to other settings , Americaninhas is representative of areas of rural northeast Minas Gerias state where helminth infections are highly endemic . Factors which may vary in other settings include contrasting socio-economic and environmental conditions , giving rise to different patterns and risk factors . However , our adopted analytical approach provides a robust methodology to further investigate the epidemiology of polyparasitism in other settings . A final potential limitation of our study , which applies to all multinomial analyses , is the assumption of Independent Irrelevant Alternatives ( IIA ) , which essentially states that the risk associated with each outcome will not change if a new outcome is introduced . However , we believe that this analysis should not be restricted by IIA because our four choices exhaust the available responses ( there are no other possible outcomes involving these two infections ) [48] . A key finding of our study was that household and environmental risk factors could only account for an estimated 32% of variation between households in the risk of co-infection . Furthermore , unexplained household-level variation of co-infection with N . americanus and S . mansoni was considerably greater than for mono-infection with either N . americanus or S . mansoni . Although this may simply be a reflection of additive household variation associated with each species , it may alternatively be indicative of household factors specifically influencing risk of co-infection . For example , extrinsic factors such as location and infectivity of household water-sources [36] , [39] , or hygiene behaviours , health knowledge and water-contact patterns shared by members of the same households [49] , [50] may influence exposure to both infections . Unaccounted household-level variability may alternatively be explained by intrinsic host-related factors such as genetics [51] , [52] , nutrition [53] , immune response [54] , [55] or concomitant infection with other parasites [56] . Despite an increasing number of studies suggesting a genetic component to variation in intensity of helminth infection , the relative importance of host genetics and exposure remain unclear and vary considerably between the settings studied ( reviewed in [57] and [58] ) . For example , in Zimbabwe , 37% of the total variation in N . americanus infection intensity was attributed to genetic factors [59] , while in Brazil genetic factors only explained 21% of total variation in S . mansoni infection intensity [60] . To our knowledge , the genetic component of helminth co-infection has been investigated in only one study , conducted among residents of a rural community in Jiangxi Province , China [61] . The results of this study suggested that the risk of infection with multiple helminth species ( Schistosoma . japonicum , Trichuris trichuria and A . lumbricoides ) was in part explained by both genetic ( 16% of total variation ) and household ( 9% ) components . This was however a post-treatment study setting , hindering interpretation of results and preventing analysis of infection intensity . Household clustering of helminth infection may also be influenced by genetic heterogeneity in the parasite population . Numerous molecular studies have revealed allelic and nucleotide diversity in the genomes of human helminth parasite populations [62] , [63] , with genetic variation occurring even among parasites sampled at very fine spatial scales [64] . As such , similarities in infection status within households may be in part due to parasite-relatedness , rather than host-relatedness [18] . However , separating the effects of host and parasite genetics on the variation in helminth infection remains a formidable task . Whilst studies at micro-epidemiological scales are less useful for mapping and prediction of the distribution of co-infection , they are valuable in identifying why certain individuals within communities are at increased risk of multiple helminth infection , and as such have an increased risk of morbidity [15] . The balance between exposure and host-related factors such as genetics , nutrition , or the immune response as determinants of infection remains one of the most fundamental questions in parasite epidemiology and is a critical element in the rational development of control approaches [18] . The results presented here demonstrate considerable household clustering of co-infection , which could not be explained by a number of micro-climatic , socio-economic and other exposure-related factors . This further emphasises the role of the household in the heterogeneous distribution of helminth co-infection in human communities , pointing to the involvement of behavioural or genetic factors . Previous studies of household and familial clustering of single-species helminth infection have reached conflicting conclusions [34] , [35] , [65] , [66]; few have simultaneously estimated the influence of both genetic and environmental factors [51] , [60] , [67] , and only one has quantified influences on co-infection [61] . Future work is clearly needed to untangle the role of host factors such as genetic relatedness from household and environmental determinants of infection if we are to fully understand the basic epidemiology of human helminth infection at a community level . | Helminth species such as Necator americanus and Schistosoma mansoni are among the most prevalent of chronic human infections in the developing world . Individuals living in endemic areas are commonly infected with both species . Although the implications of being co-infected with helminths are increasingly recognized , factors influencing patterns of co-infection within human communities remain ill-defined . Here , we describe spatial patterns and risk factors for co-infection with N . americanus and S . mansoni in a co-endemic area in south-eastern Brazil . The prevalence of co-infection with these two helminths in this region was high ( 41% ) , varied across the study area and was clustered in high-risk households . We reveal that factors associated with lower socio-economic status ( relative socio-economic status , household crowding ) and residential environment ( living in the eastern watershed or in areas with less vegetation ) were significantly associated with the risk of co-infection relative to being uninfected with either species . Importantly , much of the variability in risk between households ( i . e . household clustering ) could not be readily explained by these risk factors . The results suggest that , whilst measures aimed at reducing exposure to infection may have an important impact on co-infection and its associated morbidity , untangling the relative contribution of intrinsic host factors ( e . g . immune response ) from household and environmental determinants remains critical to our understanding of helminth epidemiology . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"infectious",
"diseases/helminth",
"infections",
"infectious",
"diseases/epidemiology",
"and",
"control",
"of",
"infectious",
"diseases"
] | 2008 | Human Helminth Co-Infection: Analysis of Spatial Patterns and Risk Factors in a Brazilian Community |
Over the past ten years the incidence of pertussis in the United States ( U . S . ) has risen steadily , with 2012 seeing the highest case number since 1955 . There has also been a shift over the same time period in the age group reporting the largest number of cases ( aside from infants ) , from adolescents to 7–11 year olds . We use epidemiological modelling and a large case incidence dataset to explain the upsurge . We investigate several hypotheses for the upsurge in pertussis cases by fitting a suite of dynamic epidemiological models to incidence data from the National Notifiable Disease Surveillance System ( NNDSS ) between 1990–2009 , as well as incidence data from a variety of sources from 1950–1989 . We find that: the best-fitting model is one in which vaccine efficacy and duration of protection of the acellular pertussis ( aP ) vaccine is lower than that of the whole-cell ( wP ) vaccine , ( efficacy of the first three doses 80% [95% CI: 78% , 82%] versus 90% [95% CI: 87% , 94%] ) , increasing the rate at which disease is reported to NNDSS is not sufficient to explain the upsurge and 3 ) 2010–2012 disease incidence is predicted well . In this study , we use all available U . S . surveillance data to: 1 ) fit a set of mathematical models and determine which best explains these data and 2 ) determine the epidemiological and vaccine-related parameter values of this model . We find evidence of a difference in efficacy and duration of protection between the two vaccine types , wP and aP ( aP efficacy and duration lower than wP ) . Future refinement of the model presented here will allow for an exploration of alternative vaccination strategies such as different age-spacings , further booster doses , and cocooning .
In 2012 , the United States experienced the highest number of reported pertussis cases since 1955 , shortly after the introduction of vaccine in the 1940s [1–3] . This upsurge occurred in the context of steadily rising reported disease from the early-1980s [4] , despite the maintenance of high vaccination coverage levels ( >90% ) [5] . Pertussis remains a major cause of childhood mortality world-wide , responsible for 195 , 000 deaths in 2008 [6] . A vaccine for the disease was developed in 1942 in the U . S . , and was included and administered as a killed whole-cell component of the diphtheria/tetanus/pertussis ( DTP ) combination vaccine . Following introduction of vaccination , the reported disease incidence in the U . S . declined from 150 cases per 100 , 000 per annum prior to 1940 , to the point of near elimination in the mid-1970s ( 0 . 5 reported cases per 100 , 000 ) [4] ( Fig 1 ) . Due to the possible reactogenicity of the whole-cell pertussis component of the DTP vaccine , acellular pertussis ( aP ) vaccine was developed as a replacement in 1991 ( DTP became DTaP ) . This new vaccine was administered to children in 1992 [7 , 8] and then phased into the infant immunization schedule beginning in 1997 [9 , 10] , following approval by the U . S . Food and Drug Administration ( FDA ) and recommendations by the U . S . Advisory Committee on Immunization Practices ( ACIP ) . Over the past 30 years , reported pertussis incidence in the U . S . has been steadily increasing ( Fig 1 ) . Even accounting for this steady upward trend , which some have attributed to improved surveillance and diagnostics [11 , 12] , case numbers during 2004–2012 were notably high and many state public health departments reported outbreaks during these years . During the two most recent of these outbreaks , the worst-hit states were California ( in 2010 ) [2 , 3] and Washington ( in 2012 ) [13] with case counts unmatched since the mid-1940s . While the 2004–2005 outbreak was characterized by an increase in the number of cases among adolescents ( 10–20 year olds ) , as well as a less-pronounced increase across all ages , the most recent outbreaks have seen many more cases among 7–10 year olds and young adolescents ( 11–13 year olds ) along with an overall increase in case numbers [2] . A number of explanations have been put forward for the increase in disease that has been observed , particularly for the outbreaks of the past 10 years [11 , 14–19] . These include: 1 ) The evolution of the circulating bacteria away from the targets of vaccine antigens; 2 ) A decline in vaccine coverage levels; 3 ) a change in vaccine efficacy ( VE ) , and/or duration of protection due to the shift to the acellular vaccine; or a specific lot of manufactured vaccine with lower efficacy; 4 ) The decline of natural boosting , through reduced exposure to naturally circulating pertussis bacteria , as a result of years of successful mass-vaccination; 5 ) An increase in disease reporting rates ( e . g . due to improved diagnostics and surveillance ) [20 , 21] . We examine these possible explanations using a population dynamic model representing infection with Bordetella pertussis in an age-structured population . We fit the output of this model to 20 years of age-stratified incidence data from the National Notifiable Diseases Surveillance System ( NNDSS ) . The model includes both vaccine and naturally-induced immunity ( with both of these sources of protection waning over time ) , the current vaccination schedule administered to the population , an estimate of the underreporting of disease , and a decreased infectiousness of those who have experienced greater than one infection .
Eight different ( nested ) models were fitted to the NNDSS incidence data ( Table 1 ) . The most parsimonious ( best fitting ) model , Model 8 , has a difference in efficacy and duration of protection between the whole-cell and acellular vaccines , and a difference between the duration of protection of whole-cell vaccine and natural infection ( Table 1 ) . Model 5 , in which the duration of protection for whole-cell vaccine and natural infection is the same , has a slightly lower DIC value , but is very close to Model 8 , and these two models are almost interchangeable in terms of their parameter values . Further details of the candidate models are provided in the Supporting Information . The model reproduces the trends in the NNDSS incidence data up to 2009; running the model forward over another three years shows a continued correspondence with observed incidence trends ( Fig 2 ) . A large proportion of forward simulations ( >50% ) show an upsurge in incidence around 1990 , a phenomenon that is not recorded in the data . However , the mean of the model-predicted incidence remains much closer to the data , albeit with a large range of uncertainty . Fig 3 illustrates the age distribution of incident cases every two years from 1994–2012 , for Model 8 . Several features of the NNDSS data are reproduced by the model: 1 ) an overall rise in disease cases throughout the observed period; 2 ) an adolescent peak in the incidence curve for most of the years post-1995; 3 ) a gradual rise in cases among 5–10 year olds from 2006 , forming a peak which is comparable in size with the adolescent peak by 2008 . In 2002 , the dominant peak in incidence is among adolescents but another smaller peak emerges among infants . This smaller peak moves forward one year at a time , growing in magnitude until 2006 when the two peaks are of similar size . By 2008 , the cohort of individuals immediately following the shift in vaccine efficacy have advanced to become 10 years of age and they constitute the 'wavefront' of a swell of new disease . The two plots for 2010 and 2012 are included to illustrate that the model continues to correspond well with the age-distribution of disease beyond the time period for which it was fitted ( i . e . beyond 2009 ) . Note that the correspondence between these ‘predicted’ distributions of disease with age was not quantitatively determined , due to the complexity of the total likelihood used for model-fitting , but these distributions continued to capture the main features of the disease profile , namely: a ) a rising overall incidence; b ) the highest peak of disease among infants; c ) a growing adolescent peak moving up in age with each advancing year . The best-fitting model incorporates a drop in vaccine efficacy and a rise in the vaccine-protection waning rate from the whole cell to acellular vaccine . Both the change in the per dose efficacy and the duration of protection change appreciably between vaccine types . Table 2 shows the estimated mean and 95% confidence intervals ( CI ) of the efficacies of the first three doses , considered together , and the fourth and fifth doses ( the probability of fully protective seroconversion ) . Estimates of the epidemiological parameters are shown in Table 2 . The basic reproduction number , R0 , was estimated to be 11 [95% CI: 9 . 9 , 11 . 5] , slightly lower than some previous estimates [18] . Secondary and subsequent infections ( those which are experienced by individuals who retain some level of immune protection , whether due to vaccination or infection-induced immunity ) are found to be 32% [95% CI: 29% , 35%] as infectious as primary infections and these individuals are found to be 17% [95% CI: 14% , 23%] as susceptible to infection as infection-naïve individuals . The disease reporting rate is included in the model as a fraction of the infected cases experiencing primary infection who will be symptomatic and captured by NNDSS surveillance . We find that the best-fitting model does not require a change in the reporting rate to reproduce the major trends in the data . The reporting rate is estimated at 6% [95% CI: 0 . 1% , 22%] with the wide credible intervals suggesting that a gradual change in reporting over time may not be precluded , even if a step change is not supported . A gradual change in reporting is consistent with the uptake of more sensitive diagnostic methods as well as greater awareness of pertussis in recent years . Fig 4 shows the results obtained by simulating the case-control study conducted in 2010 by Misegades et al [22] . We demonstrate that , in successive ages beyond when the final vaccine dose is administered , there is a decline in the 95% credibility envelope of the model-based VE estimates ( gray-shaded region ) : the bottom of this envelope declines more rapidly than measured for the California study . The curve lying above the shaded region illustrates the results of a hypothetical VE study if it had been conducted in 1990 , the pre-acellular era . It can be seen that VE declines more slowly in this case .
We have shown that relatively small differences in the per-dose vaccine efficacy and duration of protection between acellular and whole-cell vaccines is sufficient to explain the recent upsurge in pertussis in the United States . These differences caused a shift in the age distribution of pertussis disease incidence: the adolescent peak in years prior to 2006 shifted to a younger age group ( 5–10 year olds ) post-2006 . This explanation for pertussis disease dynamics is in broad agreement with case-control and retrospective cohort studies showing that vaccine-induced immunity wanes faster than previously thought [22–24] . The average duration of protection we estimate for acellular vaccine is shorter than the essentially lifelong protection we estimate for both whole-cell vaccination and natural infection; however our estimate of the acellular protection duration is around 50 years , which is longer than might be expected from published VE studies [22 , 23] . Since our calculated waning rate is an average over the whole population , this means some individuals will re-enter the susceptible pool more quickly than others . By simulating a prospective VE study , we find that even a small increased flow into the susceptible pool is sufficient to result in a strong signal of declining VE following the final vaccine dose ( see Figs 4 and 5 and [22] ) . The efficacy values implied here for the full acellular vaccine regimen ( i . e . 5 doses ) are consistent with those estimated in the acellular pertussis vaccine ( APERT ) trial [25] , lending further weight to the correspondence of our model with the broad range of data . Furthermore , VE values for individuals with mixed acellular and whole-cell vaccine histories are consistent with a recent Australian report of disease rates and linked vaccination histories [26] . While our analysis does not preclude the possibility of a rise over time in the rate of reporting ( or detection ) of cases of disease , such a rise is not necessary to explain recent patterns in pertussis incidence . Our estimate of case reporting rates ( approximately 6% ) is in line with a recent multi-country analysis [20] , but has wide 95% credible intervals . Hence , progressive changes in disease reporting , such as those expected by more sensitive diagnostic testing ( e . g . PCR ) or a greater awareness of pertussis among patients and physicians , is within the scope of our model . Further work with this model could investigate published reporting trends more closely [11 , 20 , 21 , 27 , 28] . Our analysis sheds light on other , previously uncertain , aspects of pertussis epidemiology . First , we estimate R0 to be in the range of 9–12 which is closer to the values found by previous mathematical models [29 , 30] than the often-quoted range of 12–17 [31] . Such values do not rule out elimination at high enough vaccination coverage levels . Second , previous recipients of vaccine whose protection has waned are estimated to be 32% as susceptible to infection as infection-naïve individuals and , if infected , these individuals are estimated to be 17% as infectious as primary infections in immunologically naïve individuals . These numbers suggest that adults and adolescents may be an important reservoir of infection . Third , the most parsimonious model with a good fit to the data suggests that the duration of immunity generated by the whole-cell vaccine is nearly equivalent to that induced by natural infection . Our explanation for the upsurge and the age-shift in disease incidence differs from other recent model-based studies [14 , 15] . We find a 'hypersensitive' boosting model [14] to be too unstable to mimic the steady post-1970s rise in disease; the incidence of disease for that model shows a large amplitude oscillatory behaviour . Population age-dependent contact patterns alone also appear to be insufficient to capture recent changes in U . S . incidence of disease [15] . Declines in vaccination coverage and the evolution of the circulating bacteria would lead to rising disease but not the changing age-distribution , although the recent discovery of pertactin-negative ( or other genetic ) variants may be playing a role in the vaccine efficacy decline we have found [32 , 33] . Changes in age-dependent disease reporting could mimic observed incidence patterns , of course , but such models would lack the simplicity of ours . Future refinement of the model presented here will allow for an exploration of alternative vaccination age-spacings for the five-dose childhood schedule , investigation into the effects of further booster doses ( for example , decennially to adults ) and cocooning vaccination strategies designed to protect infants who are most vulnerable to severe disease [22] . Further data collection will help to narrow uncertainties in the model , for example , the range of R0 and , the infectiousness of those who have been previously infected , both relevant to control strategies . Alternative model structures capturing the development and waning of immunity and boosting of immunity may be contributing to discrepancies between the outputs of our best-fitting model and the available data , such as an overestimate of the incidence peak in adolescents after 2008 ( Fig 3 ) , and signs of a model underestimate of the overall incidence ( Fig 2 ) . Future work should attempt to incorporate the full range of possibilities for modeling boosting and immunity e . g . using the proposed model of Lavine et al . [14] , as well as mixing matrix data that are as locally-specific as possible , following the work of Rohani et al . and Riolo et al . [15 , 34] . Clearly , the use of a mixing matrix from Great Britain is not ideal for the U . S . but was all that was available at the time of writing . In addition , fitting our model to data from the U . S . aggregates the incidence from a large number of more spatially localized epidemics , which may not be completely synchronized; this means that , while our model is convincing , it tends to describe phenomena such as the ‘cohort effect’ and peaks in incidence more sharply than the data . Future work should focus on modeling local outbreaks and endemic pertussis with high spatial resolution , following Rohani et al . and Choisy et al . [35 , 36] . More intricate modeling frameworks , such as individual-based modeling , are also promising directions for future work , given their ability to represent human behavioural dynamics , which could be a major local contributor to vaccine uptake or refusal [37] . A further important step for this study is the analysis of pertussis incidence data—along with demography and vaccination histories—from around the world . Jackson et al . recently reviewed such data and showed that global pertussis epidemiology is complicated and by no means follows the same patterns and explanations as the U . S . [38] . However , the authors of the review do note that the issue of the switch from the wP to the aP vaccine may be sufficiently specific to allow investigation of its impact upon global trends . Transmission models with a realistic representation of the natural history of infection and immunity , and population contact processes , fitted to data with statistically rigorous methods , are useful tools to quickly investigate increases in disease incidence , whether they occur within the U . S . or globally [39–41] . Pertussis epidemiology is complex and disease incidence is non-linearly related to vaccine coverage . With doubt around the efficacy and duration of protection of the acellular vaccine , modeling is an essential tool to help us better understand the changing epidemiology of pertussis .
The mathematical model we constructed has an age-structured susceptible-infected-recovered ( Si , I1i , Ri ) structure ( the subscript i , corresponds to one of 35 age-groups ) , but with the addition of a second infected compartment ( I2i ) to account for those who have been previously infected . The mixing matrix β ( i , j ) represents the product of the contact rate ( obtained from the ‘POLYMOD’ diary study of contact patterns in Great Britain ( GB ) [43] ) and the transmission probability per contact between individuals in age-groups i and j . Individuals in state I2i have an infectiousness value of η compared with those in state I1i , and the relative susceptibility to infection of those in Ri ( the ‘recovered’ state ) compared with those completely susceptible . Vaccination occurs with the DTP vaccine with whole-cell pertussis component by shifting individuals into compartment V1i and into V2i for the DTaP and Tdap vaccines ( the main text simplifies this flow into a single compartment Vi so as not to clutter the exposition ) . The equations corresponding to the model flow diagram ( Fig 5 , Table 3 ) are given below: dSidt=−Si . ( λi+μi ) +α . RidI1idt=Si . λi− ( τ+μi ) . I1idI2idt=σλi . Ri− ( τ+μi ) . I2idRidt=τ . ( I1i+I2i ) − ( δ+σλi+μi ) . RidV1idt=− ( γ+μi ) . V1idV2idt=− ( γ+μi ) . V2iwhere the force of infection , λi=∑j=135β ( i , j ) . ( I1jNj+η . I2jNj ) Pertussis case count data from NNDSS were aggregated into annual counts for each age group so that yi ( t ) is the number of pertussis disease cases for age group i in year t . Our mathematical model outputs were also aggregated into annual counts for each age group i , so that xi ( t ) was the model-derived case count for age group i , during year t . These model-derived case counts are functions of the model structure and parameters , so that they might be better expressed as xi ( t|θ , M ) , where θ represents the parameter vector for model M . Since the pertussis cases counted by NNDSS may be a proportion of the true number ( and the output of the natural history epidemiological model is designed to simulate the true number ) we can say that the observed number of cases follows an observation model so that annual counts of pertussis cases ( obsi ( t ) ) are governed by a binomial distribution: obsi ( t ) =Binomial ( xi ( t|θ , M ) , ( t ) , pi ( t ) ) The proportion of individuals with disease who report it ( pi ( t ) ) may vary over time in a number of ways ( e . g . linear/nonlinear increase or decrease ) but here we account for this temporal uncertainty in the simplest way , by allowing the value of pi ( t ) to change once during the course of a simulation run ( i . e . it has one value prior to a time T and a different value after this time ) so that: pi ( t ) ={p1 , t≤Tp2 , t>T The binomial log-likelihood of the data given these considerations is: LogLikelihood=∑i∑tln ( xi ( t ) ! ) −ln ( yi ( t ) ! ) −ln ( ( xi ( t ) −yi ( t ) ! ) ) +yi ( t ) . ln ( pi ( t ) ) + ( xi ( t ) −yi ( t ) ) . ln ( 1−pi ( t ) ) This LL sum can now be used as an objective function for our Markov Chain Monte Carlo scheme , to fit our model parameter values so that model outputs match the NNDSS data . However , we find that it can be excessively difficult to explore the parameter space productively ( i . e . moving from regions of poorer to relatively better fits to the data ) since the binomial likelihood surface can take on high ( i . e . better fitting ) values in very small ( and therefore very hard to locate ) regions of parameter space . To attempt to remedy this problem ( as well as to account for a more general degree of uncertainty in the disease reporting rate ) , we introduce into our observation model a further layer of uncertainty such that the proportion of individuals reporting disease is drawn from a Beta distribution , which has two parameters , π ( pi ( t ) ) =Beta ( ∝ ( t ) , β ( t ) ) where the Beta distribution’s parameter values are time-dependent in the same way as we posited above i . e . : ∝ ( t ) , β ( t ) ={∝1 , β1 t≤T∝2 , β2 t>T Using this structure for the reporting rate component of the observation model , we can recalculate the likelihood of a particular dataset given the model ( removing the subscripts for simplicity of presentation ) : likehood=∫01Binomial ( x , p ) . π ( p ) . dp This integral allows us to compute an average likelihood allowing for the entire distribution of the reporting rate p , given its Beta distribution . Because the Beta and Binomial distributions are conjugate , the integral is straightforward . Where B ( ∝ , β ) is the normalizing constant for the Beta probability distribution . The integrand can be seen to be the probability density function of the Beta distribution Beta ( ∝′ , β′ ) with the adjusted ( or ‘renormalized’ ) parameter values , so that: ∝′=y+∝ , β′=x−y+β So the integral above is the normalization constant of the Beta distribution ( which is referred to as the Beta function , itself comprised of a set of gamma functions , see below ) with the adjusted parameters . The likelihood therefore becomes: likelihood= ( xy ) . B ( y+∝ , x−y+β ) B ( ∝ , β ) = ( xy ) . Γ ( y+∝ ) . Γ ( x−y+β ) . Γ ( ∝ . β ) Γ ( ∝ ) . Γ ( β ) . Γ ( ( y+∝ ) . ( x−y+β ) ) And for all of the data points , the log-likelihood becomes: LL=∑i∑t[ln ( xi ( t ) yi ( t ) ) +ln ( Γ ( yi ( t ) +∝ ( t ) ) ) +ln ( Γ ( xi ( t ) −yi ( t ) +β ( t ) ) ) +ln ( Γ ( ∝ ( t ) . β ( t ) ) ) −ln ( Γ ( ∝ ( t ) ) ) −ln ( Γ ( β ( t ) ) ) −ln ( Γ ( ( yi ( t ) +∝ ( t ) ) . ( xi ( t ) −yi ( t ) +β ( t ) ) ) ) ] | Over the past ten years the incidence of pertussis in the United States ( U . S . ) has risen steadily , with 2012 seeing the highest case number since 1955 . There has also been a shift over the same time period in the age group reporting the largest number of cases ( aside from infants ) , from adolescents to 7–11 year olds . We investigate several hypotheses for the upsurge in pertussis cases by fitting a suite of epidemiological models to incidence data from the National Notifiable Disease Surveillance System ( NNDSS ) between 1990–2009 . We find that: 1 ) the best-fitting model is one in which the vaccine efficacy and duration of protection of the acellular pertussis vaccine is lower than that of the whole-cell vaccine , 2 ) increasing the rate at which disease is reported to NNDSS is not sufficient to explain the upsurge and 3 ) 2010–2012 disease incidence is predicted well . These results demonstrate that the resurgence in pertussis in the U . S . can be explained by past changes in vaccination policy . However , our findings suggest that the efficacy of the currently-used acellular vaccine is not much lower than that of the whole-cell vaccine , and booster doses may be sufficient to curtail epidemics while vaccine research continues . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2015 | A Change in Vaccine Efficacy and Duration of Protection Explains Recent Rises in Pertussis Incidence in the United States |
Mathematical models that integrate multi-scale physiological data can offer insight into physiological and pathophysiological function , and may eventually assist in individualized predictive medicine . We present a methodology for performing systematic analyses of multi-parameter interactions in such complex , multi-scale models . Human physiology models are often based on or inspired by Arthur Guyton's whole-body circulatory regulation model . Despite the significance of this model , it has not been the subject of a systematic and comprehensive sensitivity study . Therefore , we use this model as a case study for our methodology . Our analysis of the Guyton model reveals how the multitude of model parameters combine to affect the model dynamics , and how interesting combinations of parameters may be identified . It also includes a “virtual population” from which “virtual individuals” can be chosen , on the basis of exhibiting conditions similar to those of a real-world patient . This lays the groundwork for using the Guyton model for in silico exploration of pathophysiological states and treatment strategies . The results presented here illustrate several potential uses for the entire dataset of sensitivity results and the “virtual individuals” that we have generated , which are included in the supplementary material . More generally , the presented methodology is applicable to modern , more complex multi-scale physiological models .
Global initiatives such as the IUPS Physiome project [1] , [2] and the Virtual Physiological Human ( VPH ) project [3] , [4] aim to quantitatively understand human physiology at all levels from gene to organism through the use of mathematical modelling . Beyond a certain degree of complexity , the combinatorial number of interactions between the parts of a system can defy intuition and present severe challenges [5] . Mathematical models are appropriate tools for developing our understanding of human physiology , since they can be used to represent and analyse the combinatorial number of interactions between parameters in a rigorous and systematic manner [6] . In short , computational models that integrate physiological data from multiple scales ( both physical and temporal ) provide a framework for understanding the maintenance of biological entities under physiological and pathological conditions . One significant application for such models is individualized predictive medicine; i . e . , tailoring models to the characteristics of an individual patient and predicting the outcomes of different treatment strategies , to help select the best strategy for that patient [3] . Many challenges must be overcome before a truly integrative model of human physiology can be constructed [6] , [7] . Gaining a real quantitative understanding of the phenotypic variation in humans as a function of genes and environment in a mechanistic sense ( i . e . , understanding the genotype-phenotype map in both the explanatory and predictive sense [8]–[10] ) is a tremendous challenge that awaits technological , conceptual and methodological breakthroughs [11] . A number of models have already been used to develop insight into aspects of human physiology [12]–[22] , many of which have their origin in the control-theory model of whole-body circulatory regulation introduced by Guyton et al . in 1972 [23] , [24] . Although it was published over 30 years ago , the Guyton model remains a landmark achievement , and with the rise in the last 10 years of systems physiology , it has attracted renewed attention [18] , [25]–[27] and even generated some recent controversy [24] , [28]–[30] . It was the first “whole-body” , integrated mathematical model of a physiological system; it was particularly instrumental in identifying and exploring the relationship between blood pressure and sodium balance , and in demonstrating the key role of the kidney in long-term regulation of blood pressure . It allows for the dynamic simulation of systemic circulation , arterial pressure , and body fluid regulation , including short- and long-term regulations . In previous work , the Guyton models were modularized and re-implemented in Fortran , C++ ( M2SL [31] ) , and Simulink [14] . Furthermore , since one of the main limitations of the early Guyton models is the low-resolution description of most of their constituting modules , a framework was built to allow replacement of the original sub-modules by new versions at a higher temporal or spatial resolution [32]; e . g . , a pulsatile heart was introduced to treat systolic and diastolic blood pressures instead of only mean blood pressure [33] , and a detailed model of the renin-angiotensin-aldosterone system ( RAAS ) has also been integrated [34] . That work was also linked to efforts in the European VPH via two Exemplar Projects , one of which used our modular reimplementation of the Guyton model as the basic set of “bricks” for a collaborative core-modeling environment for multi-organ physiology modeling [13] , [14] , and the other uses the Guyton model as a demonstrator for the tagging of parameters and variables with a set of reference ontologies common to databases of high-throughput genomic and proteomic data [35] . Collaborators in the Physiome/VPH community have also carried out XML markup of the individual modules of the Guyton model in CellML ( http://models . cellml . org/workspace/guyton_2008 ) , thus providing precious documentation of its structure and content . The analysis and results presented here arose naturally from this body of work . Our motivation was to develop a methodology for systematically exploring the ramified implications of multi-parameter interactions in multi-scale physiological models . We present such a methodology , which incorporates the elementary effects technique introduced by Morris [36] . As a case study , we present a sensitivity analysis of the 1992 version of the Guyton model [24] , [30] , [37] , with a focus on the multiple interactions involved in blood pressure regulation . This version was never published , but represents a more complete and modern understanding of the cardiovascular system [24] , [30] ( e . g . , the inclusion of ANP [37] ) , and it is the version that members of the Guyton group have continued to use . Indeed , such a model , grounded in decades of hands-on experimental work and built with an engineer's approach to control processes , should serve as a rigourous platform for discovery of non-intuitively obvious relationships . However , despite the significance of the Guyton model , the dynamics of the model have not yet been analysed in a systematic and comprehensive study . The results provide valuable information about the inter-dependencies of parameter effects on the model outputs , thus providing direction for future physiologically-applicable sensitivity studies of the effects of changes to multiple parameters . These results also lay the groundwork for the use of multi-parameter models such as the Guyton model in systematic in silico exploration of possible new drug effects , hypotheses about multiple perturbations leading to disease states , and alternative treatment strategies . An additional outcome is the production of a virtual population , where each virtual individual is characterized by its set of parameter values ( loosely analogous to genotypes ) and the associated outputs ( “phenotypes” ) . Note that the parameters of the Guyton model are in fact lower-level phenotypes , but as models continue to span larger physical and temporal scales , model parameters will approach the genotype level [38] , [39] . A given real-world patient can be associated with one or more of these virtual individuals on the basis of clinically identifiable parameters or dynamics ( e . g . , mean arterial pressure , serum total protein , cardiac output , heart rate ) . Searching an existing collection of simulations in this manner avoids the inherent pitfalls in solving the inverse problem of ( uniquely ) identifying unknown model parameters and states from clinical observations [40] . Thus , the construction of a comprehensive virtual population could prove a useful tool in future efforts to provide efficient , individualized health-care . Note that beyond the methodology itself , the results presented in this manuscript also serve to demonstrate some of the uses to which the complete set of elementary effects and virtual individuals may be applied . We provide tables of all of the resulting output in the supplementary material ( Dataset S1 ) , which we hope will be of use in physiological , pathophysiological and clinical settings .
The Guyton model comprises parameters and output variables . We restricted our analysis to parameters {} and output variables {} ( as indicated in Equation 1 , Table 1 , and documented in Tables S1 and S2 ) , focusing on those parameters with direct physiological relevance and ignoring parameters with no clear physiological interpretation ( such as curve-fitting coefficients ) . The distribution of these 96 parameters was: 32 cardiac , 21 renal , 16 autoregulation , 16 hormonal , 11 local circulation , and 4 thirst-related . To determine which parameters have significant effects on each of the model outputs , we computed the elementary effects of each parameter using a modification of the formula defined by Morris [36] , which we now detail . The influence of the parameter on some output is defined by Equation 2 . Assuming that each parameter is normalized to the unit interval that , the region of experimentation––the portion of the parameter space that will be explored––is a regular -dimensional -level grid , where each parameter may take on values from ( Equation 3 ) . For each parameter in turn , a perturbation is chosen ( Equation 4 ) . For positive perturbations , we restrict ( Equation 5 ) , and for negative perturbations , we restrict ( Equation 6 ) , so that . For any point ( where or ) , Morris defined the elementary effect of as per Equation 7 . In our analysis of the Guyton model , we chose to normalize the elementary effects with respect to ( Equation 8 ) rather than by , which is always a fixed percentage of the range of . Each elementary effect was calculated times , where each of the simulations was performed with randomized values for all parameters , in order to obtain a representative sample of the magnitude of the effect . Given a set of values for a single elementary effect , it is important to note that the mean and variance of this set provide different insights into the nature of the relationship between the parameter and the output . The mean indicates the sensitivity of to , while the variance indicates the influence of other parameters on this relationship or the non-linearity of the effect . For each random input vector , a simulation was started with the default initial state ( ) and progressed for four weeks of simulation ( ) , at which time a pseudo-steady state had either been reached , or a new random input vector was chosen and the simulation was restarted . The parameter under investigation ( ) was then incremented ( or decremented ) by and the simulation continued for another four weeks of simulation time , after which either a new pseudo-steady state had been reached , or a new random input vector was generated and the simulation was restarted . Throughout the simulations , a number of output variables were monitored to ensure that they remained within physiological bounds ( i . e . , that the virtual individuals remained “alive” , see Table 2 ) . If these bounds were violated during a simulation , the simulation was discarded and a new input vector was chosen . Since the system is highly non-linear , the effects of a perturbation in the parameter on the output variables vary over time , so elementary effects were calculated at times ( Equation 9 ) and the state of the model ( ) was recorded at times ( Equation 10 ) . The parameters for this mass-simulation process are given in Table 3 . This entailed simulations to obtain estimates ( with positive perturbations and with negative perturbations ) of the elementary effect of each parameter on each output . In each simulation , two distinct points in parameter space ( before and after the perturbation ) resulted in two steady states . Each input vector and steady state can be viewed as a virtual individual; that is , a virtual human whose “genotype” is described by the input vector and whose “phenotype” is described by the resulting steady-state outputs . Thus , the sensitivity analysis simulations also produced a virtual population of virtual individuals . We detail how this virtual population may be of use for diagnosis and exploration of treatment strategies for real-world patients in our discussion .
Given our interest in the development of hypertension , we focus the discussion here on variables directly related to blood pressure . For example , Figure 2a shows the most significant elementary effects ( at each time ) on three such variables: the mean arterial pressure ( MAP ) , the cardiac output ( QAO ) , and the rate of urine production ( VUD ) . The single largest effect on all three variables is that of HYL ( the quantity of interstitial hyaluronic acid ) , which affects the tissue hydrostatic and osmotic pressures . This effect is only observed one hour after the perturbation is made . That is , a change in hyaluronic acid takes more than one minute to have an effect , and the effect is no longer evident after 24 hours . The large deviations ( significantly larger than those of any other parameter ) demonstrate that the effects of HYL are highly non-linear . We will demonstrate how to identify interesting multi-parameter effects , using HYL as an example . To clearly depict the other elementary effects , they are shown in Figure 2b without the effects of HYL . The largest steady-state elementary effects at are shown in Figure 3 . The complete table of elementary effects is available in the supplementary material . Correlations were calculated between each parameter and each output variable at each time , using the Spearman rank-correlation [62] . A rank-correlation method was chosen because such methods are sensitive to any near-monotonic relationship and do not assume that the data is normally distributed . The correlations showed negligible variance ( ) over these times , in contrast to the elementary effects presented earlier . This is because the correlations are sensitive to the absolute value of the parameter , while the elementary effects are sensitive to the influence of a perturbation and not the absolute value . Significant correlations are shown in Figure 5 for the same three variables ( MAP , QAO and VUD ) whose elementary effects were presented in Figure 2 . Consider the correlations with MAP; the most-highly correlated parameters ( ) are CPR , AARK , EARK , GFLC and HM6 , all of which also exhibit significant elementary effects on MAP . As noted earlier , all of these parameters affect glomerular filtration: AARK , EARK and GFLC are all related to physical properties of the glomerulus , while CPR and HM6 affect the driving pressure gradient for ultrafiltration . In contrast , the parameters most-highly correlated with QAO ( ) are HM6 , OMM , CPR , EARK , NID , O2M and RTPPR ( the effect of glomerular oncotic pressure on renal tissue oncotic pressure ) . RTPPR was not seen to exert a significant elementary effect on QAO , but it shows a higher correlation with QAO than do AARK , ANUM , GFLC and LPPR , all of which exerted significant steady-state effects on QAO . Three of these parameters––HM6 , OMM and O2M––are directly related to oxygen supply and utilization in the body , whilst CPR and NID affect both the plasma volume and renal filtration , EARK also affects renal filtration , and RTPPR affects tubular reabsorption . The parameters most-highly correlated with VUD ( ) are NID , CPR , RTPPR , POR , AARK and KID . As was the case for QAO , RTPPR does not exert a significant elementary effect on VUD , but demonstrates higher correlation with VUD than do ANCSNS , ANUM , EARK and GFLC , all of which exhibit significant steady-state effects on VUD . All of these parameters , except for POR , are directly related to renal filtration and reabsorption , while POR modulates the vasoconstrictor effect on blood-flow autoregulation across rapid , intermediate and long-term timescales . One parameter , CPR , is notable for being highly correlated with all three output variables MAP , QAO and VUD . In particular , CPR has a correlation of with MAP; the only other correlation greater than is that between HM6 and QAO ( ) . This parameter is the critical plasma protein concentration for protein destruction in the liver , which affects the colloid oncotic pressure in the vasculature . The direct effects of this parameter include the rate of glomerular filtration and the rate of capillary leakage . These observations demonstrate that the Guyton model reflects the importance of renal filtration and colloid oncotic pressure to overall haemodynamic regulation [45] , [46] , [54] , [55] . The virtual individuals were divided into normotensive and hypertensive sub-populations based on their mean arterial pressure , as illustrated in Table 4 . The probability densities of each parameter and variable were compared across these populations , as were the correlations between the model parameters and the output variables . The probability densities revealed observable differences between the populations ( Figure 6 ) , both in the model parameters ( e . g . , CPR ) and output variables ( e . g . , AAR ) . Note that the two probability densities shown here for CPR are markedly more distinct than when CPR was classified based on the elementary effect of HYL ( not shown ) . However , obvious differences were observed for very few parameters , all of which had already been highlighted in the sensitivity and correlation analyses . Correlations between parameters and variables were then compared between the two populations; some results are shown in Figure 7 . The colour-coded regions of each graph represent different relationships between the correlations: green indicates a decreased correlation in the hypertensives; blue indicates an increased correlation in the hypertensives; and red indicates that the correlation has switched sign between the two populations . The correlations with MAP in the hypertensive population are systematically larger than those in the normotensive population ( Figure 7a ) , which supports the notion that arterial pressure regulation has been reduced in the hypertensive population . However , the correlations with QAO show no such relationship ( Figure 7b ) with the sole exception of EARK . This suggests that the regulation of cardiac output has not been reduced in the hypertensive population , and that a change in cardiac output is neither a cause nor symptom of the hypertension that is observed in the virtual population , which reflects Guyton's explanation of arterial hypertension being fundamentally a renal pathology [23] , [24] , [56] . When correlations with blood volume are considered ( Figure 7c ) , the parameters with the largest increases in correlation ( ANCSNS , ANUM , ANY ) are all related to the effects of angiotensin on arterial resistance and venous volume . Parameters with decreased correlation in the hypertensive population include NID , VV9 and CV ( venous compliance ) . The logical inference is that angiotensin is playing a more significant role in regulating the blood volume in the hypertensive individuals than in the normotensive individuals . Angiotensin plays a role in the activation of the RAAS [56] , [63] , [64] , which increases salt and water retention in the kidney [65]–[67] and raises the “set-point” arterial pressure that the kidney will maintain [50] , and these effects are incorporated into the Guyton model . More recent studies have also revealed angiotensin's roles in hypertension via oxidative stress [68]–[70] and inflammatory vascular injury [71] , [72] , but these phenomena are not included in the Guyton model . The correlations with urine production ( Figure 7d ) reveal changes in only a few parameters . The decreased correlation with RTPPR indicates that glomerular oncotic pressure has a smaller effect on tubular reabsorption in the hypertensive population . Of the parameters with increased correlations , AARK and POR are directly related to blood-flow autoregulation and vasoconstriction , and CPR affects the plasma colloid oncotic pressure , which affects the plasma volume and the driving pressure gradient for glomerular filtration . This leads us to conclude that the urine production in the hypertensive population is more sensitive to blood-flow autoregulation and plasma colloid oncotic pressure . The large virtual population that has been assembled here ( ) can be used not just to analyse relationships between model parameters and outputs , but also to derive and evaluate classifiers for predicting particular phenotypes in virtual individuals . Since hypertension places a heavy burden on health-care systems around the world , and blood pressure regulation is the chief focus of the Guyton model , the most obvious phenotype to predict is hypertension . The virtual population was divided in two: a randomly-chosen training set of the population size , and the remainder of the population served as an evaluation set . A generalized linear model ( GLM ) [43] , [44] with a binomial distribution function was fitted to the training set to predict whether or not each individual was hypertensive ( i . e . , ) . A minimal GLM was then selected by step-wise reduction of the original GLM with Akaike's information criterion ( AIC ) [73] , resulting in a 30-parameter classifier . This classifier was then evaluated on the evaluation set ( i . e . , the rest of the virtual population ) , shown in Figure 8a , and demonstrated a high degree of accuracy . The sensitivity of the classifier to each of the 30 parameters is shown in Figure 8b . This list of parameters closely resembles those parameters most-highly correlated with mean arterial pressure ( Figure 5 ) . But no matter how accurately this classifier can predict hypertension in the virtual population , one should not conclude that it will be of practical use for predicting hypertension in real-world patients . The classifier is a function of model parameters , many of which are not physiologically derived or measurable . In order to feasibly use such a classifier with real-world patients , the model parameters must be restricted to those that are readily identifiable and measurable in human beings . Of the parameters listed in Figure 8b , we assume that CPR and LPPR can be estimated from blood tests and that the values of the renal filtration parameters AARK , EARK and GFLC could possibly be estimated from whole-body glomerular filtration rate ( GFR ) ( or , more invasively , from a biopsy ) . NID can be estimated from the person's diet . The resulting classifier ( “Renal+Liver” in Figure 8a , coefficients given in Table 5 ) predicts hypertension on the basis of these parameters ( see Table 6 ) and suffers from a modest loss of predictive power in comparison to the optimal classifier . It can correctly identify of the hypertensive virtual individuals with a false-positive rate , in comparison to the optimal false-positive rate of . Further restricting the parameters to either solely liver-related or kidney-related ( Table 6 ) significantly reduces the predictive power of the classifiers .
The Guyton model was constructed and refined over many years , and has been validated against a large amount of experimental data [23] , [24] . However , many simplifications were necessary in order to permit simulated experiments under the computational resources that were available at the time [24] , and the model does not incorporate recent advances in our understanding of the cardiovascular system . Thus , our results will tend to highlight the underlying assumptions and limitations of the Guyton model , rather than physiological phenomena . Indeed , one of the goals of this study was to provide sufficient data ( in the supplementary material ) to allow researchers to identify whether the Guyton model is sufficiently detailed for specific physiological applications . More recent models have incorporated greater levels of detail for individual organs [12] , [74] or for the whole body [16] , [19] , and a comparison between the Guyton model and these newer models can illustrate the suitability of the Guyton model for clinical applications . Of course , the methodology we employed can be applied to these modern , more detailed models . Here we present a brief comparison of the Guyton model to the human renal/body fluid model of Uttamsingh et al . [74] , which was validated against several sets of experimental data . The result of ingestion of either hypotonic and hypertonic fluid in the Guyton model ( shown in Figure 9 ) produces similar effects on the urine flow rate to that seen in the model of Uttamsingh et al . However , in response to the infusion of hypertonic saline ( 0 . 91 g of sodium chloride per kg of body weight , over a period of 65 minutes for a “normal human of 70 kg” ) urine flow in the Guyton model increases at a slower rate , plateaus at a lower rate and eventually returns to the baseline level , while urine flow in [74] plateaus at twice the baseline and better matches the experimental data [75] . Larger variation between the two models is observed when aldosterone is increased four-fold , in order to simulate the administration of deoxycorticosterone acetate ( DOCA ) , a mineralocorticoid with similar effects to those of aldosterone [74] . The model of Uttamsingh et al . demonstrates gradual increases in extra-cellular fluid volume ( 1 L ) and mean arterial pressure ( 10 mmHg ) , and a rapid drop in sodium excretion in response to the elevated aldosterone level , followed by a slow rise to match the rate of intake . The Guyton model , as shown in Figure 10 , produces different behaviour . The extra-cellular fluid volume rises briefly and then gradually decreases until it is 0 . 1 L below the baseline ( Figure 10a ) and mean arterial pressure rapidly rises by 10 mmHg and then gradually increases by a further 2 mmHg ( Figure 10b ) . Sodium excretion ( Figure 10d ) drops rapidly in the first 2 hours , then rises rapidly and overshoots in the following 6 hours , before equilibrating after 24 hours have elapsed . The Uttamsingh et al . model again matches the experimental data [76] better than the Guyton model ( e . g . , it reproduces the “escape” phenomenon , where the rate of sodium excretion eventually rises to match the increased rate of intake ) . However , the limited time-resolution ( at most one data point every 24 hours ) makes a precise comparison impossible . Indeed , with the exception of the extra-cellular fluid volume , the behaviour of the Guyton model also provides a reasonable fit to the data . The differences highlighted here between the Guyton model and the model of Uttamsingh et al . are certainly due in part to the lower level of detail in the renal portion of the Guyton model , but the Guyton model also includes a more complete cardiovascular model , which would necessarily alter the dynamics produced in response to a chronic increase in aldosterone load . Thus , these observations may indicate a shortcoming in the Guyton model , but further analysis is required before a definitive statement can be made . These results highlight , however , the need to identify portions of the Guyton model that must be refined to replicate experimental data more recent than those used to originally validate the model . We discuss refinement of the Guyton model in the following section . In our analysis we perturbed a single parameter in each simulaton ( although each parameter was perturbed 1000 times , each simulation with a different set of randomly-selected parameter values ) . Perturbation of multiple parameters would yield a wealth of additional information , but without any guidance the only recourse would be to exhaustively search every combination of parameters , for perturbations . Instead , with the results presented here one can select one parameter ( ) for perturbation and additionally perturb only those parameters that are significantly correlated with the effect of ( as per our brief example: “Multi-parameter effects: accounting for the variance in HYL” ) . Given the population of virtual individuals that was presented here , an obvious and desirable application is to draw comparisons between subsets of this population and a given real-world patient . That is , given some observations of a real-world patient , we can select those virtual individuals who best match these observations and see whether one can draw conclusions about the condition of the real-world patient based on the long-term dynamics of the selected virtual individuals . Beyond using virtual populations merely as a reference for the current and ongoing condition of real-world patients who receive no intervention , ongoing refinements of the Guyton model may ultimately support individualized health-care and individualized medicine . The application of mathematical models to individualized medicine would necessarily involve integrating detailed models of physiology , pharmacokinetics and pharmacodynamics . Current efforts on this front include the BIMBO project [77] . Development of chronic diseases such as cardiovascular disease is a complex process that involves environmental and cultural factors shared by the individuals living in the same geographical area , as well as ageing , genetic and disease determinants . Hunter et al . [3] have emphasized the need for diagnostic workflows on the prediction of risk that integrates the influence of both population and patient-specific information in support of tailored interventions aiming at optimizing diagnosis and treatment planning and monitoring . Researchers of the BIMBO project have defined a modeling approach to estimate the public health impact , in terms of the reduction in the number of cardiovascular deaths ( CVD ) , of administering blood pressure lowering drugs to a virtual population of patients [77] . That virtual population [77] ( distinct from the virtual population presented here ) reproduces the demographic composition as well as the cardiovascular risk factor profiles of a country population , each virtual individual being characterized by a number of features allowing estimation of CVD risk and treatment efficacy . The individuals eligible for treatment could be selected from their computed CVD risk over a fixed threshold and by having blood pressure in excess of 140/90 mmHg . The authors used a simplified approach where treatment effect was represented by the relative risk , which was assumed to be constant over time and among different individuals , to estimate the public health impact of BP lowering drugs [77] . The work presented here illustrates the value of using population information to predict the success of treatment strategies , whilst also moving towards a more ambitious goal: taking into account the individual genetic backgrounds and pathophysiological profiles . This would contribute to the delivery of individualized healthcare , by optimizing the impact of treatments for both the individual patient and at the population level . Future challenges include the development of more sophisticated effect models [78] , such that relative-risks and odds ratios depend on individual characteristics which affect the pharmacokinetic and/or pharmacodynamic parts of the model [79] . Realization of these goals would represent a significant step towards personalizing anti-hypertensive treatment . The implications of pharmacogenetic parameters on drug efficacy have been explored in the context of diuretic treatment for blood pressure [80]–[82] . One candidate for the identification of responders to thiazide diuretics is the polymorphic gene coding the cytoskeleton protein -adducin , whose mutant form has been associated with an increased rate of sodium reabsorption [83] , [84] , elevated blood pressure [85] , [86] , salt-sensitivity [87] and increased risk of cardiovascular events [88] . The same associations first documented in Caucasian populations [84] , [87] have not been reported in all other populations , with contradictory evidence from studies in Chinese , African American and Japanese populations [89] , suggesting the role of additional factors in mediating the effects attributed to the -adducin polymorphism . But before rejecting the hypothesis of a pharmacogenetic effect of the -adducin variant , a number of epistatic interactions and environmental influences contained in the virtual population characteristics ( e . g . , different degrees of RAAS activation in response to salt consumption ) could be explored through physiological modeling . With regard to the diagnosis and treatment of hypertension , a practical model would predict the effects of the various diuretics and other drugs that are commonly administered to ameliorate hypertension . This would allow the model predictions to be directly compared to clinical studies such as INDANA [90] . To this end , refinements are being incorporated into the original Guyton model [32] as part of the SAPHIR project [13] , such as a detailed model of the RAAS [34] . The culmination of these efforts will result in a richly-detailed and more accurate model of renal autoregulation being incorporated into the Guyton model , providing a platform for pharmacological predictions that may assist in the diagnosis and treatment of hypertension [77] . We have presented a sensitivity analysis of the Guyton model of human physiology ( 1992 version ) , which examined the elementary effects of each parameter over a range of timescales and the correlations between model parameters and key output variables . We also demonstrated how interesting multi-parameter combinations can be identified , and how this can highlight shortcomings in the model . A pool of simulations with randomized parameters ( analagous to genetic variants ) was generated for this analysis , forming a diverse virtual population of virtual individuals from which representative subsets can be drawn to match characteristics of individual real-world patients . The population was divided into normotensive and hypertensive sub-populations , and a 6-parameter linear classifier was shown to have good predictive power for identifying hypertensive virtual individuals , based on parameters that are feasible to estimate in vivo . Work is currently underway on comparing these results to real-world patient data from clinical studies of the effect of Avastin on hypertension in cancer patients [91] , [92] . About half of the patients develop hypertension in response to Avastin , and are also the most likely to experience a remission . The analysis will aim to identify whether any of the elementary effects or correlations presented in this manuscript are evident in real-world patients , and to evaluate the use of the virtual population in selecting regions of the parameter space of the Guyton model that correspond to the characteristics of a real-world patient . This exploratory project is at a preliminary stage and no results can be presented at this time . The methodology we have presented here and applied to the Guyton model is generic in that it can be applied to any mathematical model of sufficient complexity . As physiological models encompass larger and larger scales , both spatially and temporally , this methodology should prove beneficial in elucidating the subtle interactions between model parameters in these complex models . Such an effort is required to evaluate the clinical suitability of using the Guyton model to assist in providing individualized predictive medicine , as per the goals of both the IUPS Physiome and the Virtual Physiological Human projects . | As our understanding of the human body at all scales increases , the construction of a “Virtual Physiological Human” is becoming more feasible and will be an important step towards individualized diagnosis and treatment . As computational models increase in complexity to reflect this growth in understanding , analysis of these models becomes ever more complex . We present a methodology for systematically analysing the interactions between parameters and outputs of such complicated models , using the Guyton model of circulatory regulation as a case study . This model remains a landmark achievement that contributed to the development of our current understanding of blood pressure control , and we present the first comprehensive sensitivity analysis of this model . Effects of varying each parameter are explored over randomized simulations; our analysis demonstrates how to use these results to infer relationships between model parameters and the predicted physiological behaviour . Understanding these relationships is of the utmost importance for developing an optimal treatment strategy for individual patients . These results provide new insight into the multi-level interactions in the cardiovascular-renal system and will be useful to researchers wishing to use the model in pathophysiological or pharmacological settings . This methodology is applicable to current and future physiological models of arbitrary complexity . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"systems",
"biology",
"physiological",
"processes",
"homeostasis",
"theoretical",
"biology",
"physiology",
"integrative",
"physiology",
"biology",
"anatomy",
"and",
"physiology",
"cardiovascular",
"system",
"computational",
"biology",
"circulatory",
"physiology"
] | 2012 | Virtual Patients and Sensitivity Analysis of the Guyton Model of Blood Pressure Regulation: Towards Individualized Models of Whole-Body Physiology |
Chikungunya virus ( CHIKV ) is a positive sense , single stranded RNA virus in the genus Alphavirus , and the etiologic agent of epidemics of severe arthralgia in Africa , Asia , Europe and , most recently , the Americas . CHIKV causes chikungunya fever ( CHIK ) , a syndrome characterized by rash , fever , and debilitating , often chronic arthritis . In recent outbreaks , CHIKV has been recognized to manifest more neurologic signs of illness in the elderly and those with co-morbidities . The syndrome caused by CHIKV is often self-limited; however , many patients develop persistent arthralgia that can last for months or years . These characteristics make CHIKV not only important from a human health standpoint , but also from an economic standpoint . Despite its importance as a reemerging disease , there is no licensed vaccine or specific treatment to prevent CHIK . Many studies have begun to elucidate the pathogenesis of CHIKF and the mechanism of persistent arthralgia , including the role of the adaptive immune response , which is still poorly understood . In addition , the lack of an animal model for chronic infection has limited studies of CHIKV pathogenesis as well as the ability to assess the safety of vaccine candidates currently under development . To address this deficiency , we used recombination activating gene 1 ( RAG1-/- ) knockout mice , which are deficient in both T and B lymphocytes , to develop a chronic CHIKV infection model . Here , we describe this model as well as its use in evaluating the safety of a live-attenuated vaccine candidate .
Chikungunya virus ( CHIKV ) is a positive sense , single-stranded RNA virus in the genus Alphavirus , and the etiologic agent of many epidemics in Africa , Asia , Europe and most recently the Americas [1–6] . CHIKV causes chikungunya fever ( CHIKF ) , a syndrome characterized by rash , fever , and debilitating arthralgia . In recent outbreaks , CHIKV has been recognized to manifest neurologic signs of illness in the young , elderly and in patients with co-morbidities [7] . The CHIKF syndrome is often self-limited; however many patients develop persistent arthralgia that can last months or years [8–10] . These characteristics make CHIKV not only important from a human health standpoint , but also from an economic standpoint . Currently , there are no licensed vaccines or specific treatments to prevent or control CHIKF . Many studies of CHIKF pathogenesis have focused on the role of interferon and macrophages . As with many alphaviruses , type I interferon plays a critical role in the host response to CHIKV infection [11] . A deficiency in type I interferon signaling has been shown to cause CHIKV infection to become lethal in the mouse model [12 , 13] . With immunocompetent mice , previous studies have reported that footpad swelling , myositis and tenosynivitis in adult C57BL/6J animals can be induced by footpad inoculation with wild-type ( wt ) CHIKV [14] . This research also demonstrated viral persistence , up to 21 days post-inoculation [15] . Results suggested that macrophages are involved in inducing footpad swelling and generating inflammatory lesions in this location [16] . Other studies have shown that CHIKV RNA can persist in the joint tissue of naturally infected humans [17] as well as experimentally infected non-human primates [18] . In mice and humans , there appears to be an overlap of up- and down-regulated genes during the arthralgic manifestations of CHIKV infection and rheumatoid arthritis , although these changes do not appear to be identical in the two diseases [19] . This raises the possibility of a derangement in the adaptive immune response to CHIKV in persons with persistent symptoms . However , the mechanism of arthralgia and persistence remain poorly understood . Because of the focus on type I interferon and macrophages , the role of the adaptive immune response in CHIK pathogenesis has garnered little attention , especially in the mouse model . However , some recent studies have begun to elucidate the role of the adaptive immune response [20 , 21] by demonstrating a possible role for CD4+ T cells in the pathogenesis of CHIK and persistence in young mice with deficits of the adaptive immune response . In addition to understanding fundamental aspects of CHIK pathogenesis and its persistence , murine models are needed to evaluate the safety of vaccines , especially live-attenuated candidates . One such strain , 181/clone 25 [22] , was tested through phase II human trials , where it proved highly immunogenic but mildly reactogenic [23] . We recently developed a live-attenuated vaccine candidate for CHIK based on the insertion of an internal ribosome entry site ( IRES ) sequence into the CHIKV genome . This vaccine candidate ( CHIKV/IRES ) is safe and immunogenic in type 1 interferon-deficient mice [12 , 13] as well as in cynomolgus macaques [24] . However , due to the prevalence of HIV infection as well as immunosuppressive therapies for cancer and other chronic diseases , live-attenuated vaccines should ideally be safe in persons with compromised adaptive immune responses . In addition , malnutrition in resource-limited areas where CHIK is endemic , which can lead to immune suppression , also underscores the need to evaluate live vaccine strains in adaptive immune-deficient models . To improve understanding of the role of the adaptive immune response in CHIK , especially in chronic arthritic manifestations , and to further evaluate the safety of our live-attenuated vaccine candidate , we used adult recombinase activating gene-1 ( RAG1-/- ) knockout mice , which are deficient in both T and B lymphocytes . We performed safety studies with the CHIKV/IRES vaccine candidate using RAG1-/- mice to evaluate the dependence on adaptive immunity of control of vaccine virus replication and pathogenesis . Also , because the roles of T and B cells in acute and chronic CHIKV infection have received limited attention , we sought to determine if the clinical manifestations ( i . e . footpad swelling ) of murine infections by CHIKV are T cell-driven . In addition , we wanted to determine if the adaptive response is necessary for viral clearance by determining whether its absence leads to viral persistence or death , and potential sites of persistent viral replication .
A wt CHIKV strain from La Reunion derived from a cDNA infectious clone was described previously [25] . The wt CHIKV as well as vaccine strain 181/clone 25 were passaged once on Vero cells . The CHIKV/IRES vaccine strain was generated by electroporation of Vero cells with RNA transcribed from a cDNA as described previously [13] and passaged once on 293 cells . Work with infected animals was carried out in either animal biosafety level ( ABSL ) -2 or -3 facilities under an approved UTMB Institutional Animal Care and Use Committee ( IACUC ) protocol 02-09-068 . All animals were cared for in accordance with the guidelines of the Committee on Care and Use of Laboratory Animals ( Institute of Laboratory Animal Resources , National Research Council ) . RAG1-/- mice on the C57BL/6J background were purchased from Jackson Laboratories ( Bar Harbor , ME ) and were 8–10 weeks of age at the time of inoculation . On day 0 , RAG1-/- and C57BL/6J mice were inoculated either subcutaneously ( SC ) or in the footpad ( FP ) with either wt CHIKV ( 3 or 5 log10 PFU ) , vaccine candidate CHIKV/IRES ( 3 or 5 log10 PFU ) , vaccine strain 181/clone 25 ( 3 or 5 log10 PFU ) , or PBS . On days 1–14 after infection , animals were weighed and FP height ( in mm ) was also measured . Serum was collected on days 1–7 , 14 , and 28 after infection to assay viremia . Two to 3 mice from each cohort were sacrificed on days 2 , 4 , 7 , 14 , 28 , 42 , 56 , and 70 after infection for collection of tissues for histopathologic evaluation and for measuring viral load by plaque assay . Frozen ( -80°C ) tissues were thawed in a 10X volume of DMEM and homogenized using a TissueLyserII ( Qiagen , Valenica , CA ) at 25 cycles per second for 4 minutes . The homogenate was clarified by centrifugation ( 2500 x g for 5 min ) and the supernatant was removed and stored at -80°C for plaque assay . Plaque assays were performed on Vero cells ( American Type Culture Collection , Manassas , VA ) in either 6 or 12 well plates as described previously [26] . Titrations were overlaid with 0 . 2% agarose in Dulbecco’s modified minimal essential medium ( DMEM ) . Two or 3 days later , cells were fixed with 10% formaldehyde for at least 30 minutes and stained with crystal violet to visualize plaques . Plaque assays were performed in duplicate or triplicate depending on the amount of serum available . Tissues were fixed in 10% neutral buffered formalin ( RICCA Chemical Co . , Arlington , TX ) , and bone tissue was decalcified overnight using Fixative/Decalcifier ( VWR International , Radnor , PA ) . Organs were embedded in paraffin and 5μm sections were cut for histopathological analysis . Sections for hematoxylin and eosin staining ( H&E ) were prepared as previously described [13] . Total RNA was extracted from 140 μl of viral stocks or 120 μl of 10% tissue homogenates using the Viral mRNA extraction kit ( Qiagen , Valencia , CA ) and the manufacturer's protocol . cDNA was synthesized using the SuperScriptIII First Strand kit ( Invitrogen ) with random hexamer primers . Then , 3 μl reverse transcription mixtures were used for PCRs , to generate overlapping amplicons covering the entire genome , with Phusion Hot Start II High-Fidelity DNA Polymerase ( New England BioLabs , Ipswich , MA ) and CHIKV-specific primers . ( sequences are available upon request ) . Amplicons were purified from agarose gels with the QIAquick Gel Extraction kit ( Qiagen , Valencia , CA ) and used for direct sequencing . Differences in animal weights and footpad swelling were analyzed using two-way ANOVA with a Tukey-Kramer post-hoc test . Significance was determined by a p-value of <0 . 05 .
To determine the effects of wt CHIKV , CHIKV/IRES and CHIKV vaccine strain 181/25 , mice were inoculated either SC or via the FP ( each route 103 PFU ) . There were no clinical signs of illness ( e . g . , lethargy , ruffled fur ) following SC or FP inoculation of RAG1-/- or congenic C57BL/6J mice with either vaccine strain or wt CHIKV , and no change in weight was identified after SC infection regardless of the virus strain ( S1 Dataset ) . This was in contrast to RAG1-/- and C57BL/6J mice infected with wt CHIKV via the FP , which showed mild but significant weight loss compared to sham-infected controls ( Fig 1 ) . Vaccine strain 181/25 was not tested via the FP route . Unlike C57BL/6J mice , RAG1-/- mice developed persistent infection when inoculated with wt CHIKV by either route ( Fig 2 ) . Viremia reached a peak titer of about 4 log10 PFU/ml on days 5–6 after SC infection , gradually decreasing to 2 log10 PFU/ml on days 14 and 28; no viremia was detected after day 28 post-infection ( Fig 2A ) . These findings were in contrast to the vaccine candidate strains , which never produced detectable viremia . C57BL/6J mice inoculated via the FP with wt CHIKV showed a peak viremia of about 10 PFU/ml on day 1 post-infection ( Fig 2B ) . RAG1-/- mice again showed persistent infection as evidenced by prolonged viremia ( Fig 2B ) . Like mice inoculated SC , they had a peak viremia of 4 log10 PFU/ml on day 5–6 post-infection , gradually decreasing to 2 log10 PFU/ml . However viremia persisted in these animals until at least day 56 post-inoculation . In contrast , the CHIKV/IRES vaccine strain never produced detectable viremia . RAG1-/- mice inoculated with 5 log10 PFU of wt CHIKV showed a similar pattern of viremia to those inoculated with 3 log10 PFU ( S2 Dataset ) . To assess tissue tropism and persistence , we harvested the organs of RAG1-/- and C57BL/6 mice inoculated with wt CHIKV , CHIKV/IRES or vaccine strain 181/25 . Both SC and FP routes of inoculation were tested ( Tables 1 and 2 ) . Neither CHIKV/IRES nor strain 181/25 was detected in RAG1-/- or C57BL/6J mice regardless of route of inoculation , and wt CHIKV was not detected in C57BL/6J mice regardless of the route . However , wt CHIKV predominantly persisted in the brain and kidney following SC inoculation of RAG1-/- mice regardless of the route . Other major organs ( heart , skeletal muscle ) were sporadically positive for persistent virus , mainly in those mice inoculated via the FP . Following FP inoculation with wt CHIKV , C57BL/6J mice developed a biphasic pattern of inflammation at the injection site that was not observed with RAG1-/- mice ( Fig 3 ) . C57BL/6J and RAG1-/- animals showed some footpad swelling on day 2 post-inoculation , which decreased to baseline levels by day 3 . C57BL/6J mice inoculated with wt CHIKV showed increased swelling at day 6 , peaking at day 7 , and resolving by day 10 post-inoculation . In contrast , RAG1-/- mice never showed footpad swelling after day 3 , and similarly , mice inoculated with vaccine strain CHIKV/IRES never exhibited footpad swelling . To evaluate tissue damage following each inoculation route ( SC or FP ) , the brain , heart , lung , kidney , spleen , stomach , intestines , liver , gonads and leg tissues ( including joints ) were collected on days 7 , 14 , 28 , 42 , 56 and 70 post-inoculation and evaluated histologically . Poorly formed granulomatous inflammation was identified in the brain , liver and lungs of RAG1-/- mice inoculated SC with wt CHIKV on days 28–56 ( Fig 4 ) . No inflammation was identified in the leg tissues of these animals . C57BL/6J animals showed poorly formed granulomatous inflammation only in the liver ( Fig 4A ) . The areas of granulomatous-type inflammation were identified in small foci . The inflammation in the brain of the RAG1-/- mouse was small and confined to the cerebrum , with no evidence of meningitis ( Fig 4B ) . The inflammation in the lung consisted of tiny foci ( Fig 4C ) . Several small poorly formed granulomas were identified in the livers of RAG1-/- and C57BL/6J mice , with no evidence of hepatocyte necrosis ( Fig 4A ) . Granulomatous inflammation in the RAG1-/- mice occurred only in those identified to have persistent CHIKV in either their serum or organs . Those RAG1-/- mice that were negative for persistent CHIKV were histologically normal . This was observed only in mice with persistent CHIKV infection and not in those without detectable CHIKV housed in the same cages . This finding suggests that the inflammation was caused by persistent CHIKV infection and not another pathogen infecting the colony or cages of immunodeficient mice . In similar fashion , some animals ( both RAG1-/- and C57BL/6J ) inoculated via the FP had very small granulomas in the liver only , similar to Fig 4B . These findings contrasted with those seen in the animals inoculated via the FP ( Fig 5 ) ; following this route , C57BL/6J mice exhibited severe myositis in the inoculated leg on day 7 ( Fig 5B ) . By day 14 post inoculation , inflammation was resolving and muscle regeneration had begun ( Fig 5E ) ; lymphocytes and neutrophils had left the lesions leaving only macrophages behind . Many of the cells present , which appeared large , were regenerating myocytes . By day 14 macrophages would be expected to be present to promote wound healing and muscle regeneration . By day 28 , the lesions were healed ( Fig 5H ) . The results in RAG1-/- mice inoculated via the FP were surprising because , as noted above , no footpad swelling was noted after day 2 in these mice following FP injection with wt CHIKV ( Fig 3 ) . Based on these results and previous studies , we did not expect to observe any inflammation or tissue damage in the RAG1-/- mice . At day 7 post-inoculation , no inflammation or tissue damage was noted in these animals ( Fig 5C ) . However , at day 14 post inoculation , severe muscle damage was observed along with a mild inflammatory infiltrate . The muscle damage observed at day 14 ( Fig 5F ) in RAG1-/- mice was comparable to damage seen in wt mice at day 7 post inoculation ( Fig 5B ) . By day 28 the lesions were healed ( Fig 5I ) . Sham-infected animals had no evidence of histologic damage in the FP ( Fig 5A , 5D and 5G ) . The PBS groups from RAG1-/- and C57BL/6J mice were identical . To determine if mutations could be involved in CHIKV persistence in RAG1-/- mice , we randomly selected four tissue samples collected on days 28 and 42 for complete genome sequencing ( Table 3 ) . When compared to the sequence of the inoculum strain , one sample from a day 28 kidney sample showed an identical consensus sequence . Virus from the other three samples had mutations only in the non-structural protein ( nsP ) genes; the majority were synonymous , while only two were non-synonymous . However , two identical synonymous mutations in nsP3 and nsP4 occurred in two independent samples from different animals on different days , suggesting positive selection at the nucleotide level .
Our study had three key findings: 1 ) the adaptive immune system is not only critical for clearance of CHIKV , but it plays a role in the inflammatory response to infection; 2 ) tissue damage occurs in the absence of an adaptive immune response , and; 3 ) the newly developed CHIKV/IRES vaccine candidate and strain 181/25 do not persist in mice , even in the absence of T/B cells . The latter point is very important when considering vaccine safety , because many people in developing countries who are exposed to CHIKV are also immunocompromised due to various conditions ( e . g . , HIV , malnourishment , etc . ) . Viral pathogenesis research has recently focused on the type I interferon response , which is also critical in controlling many alphavirus infections , including CHIKV . Also , macrophages/monocytes are known to precipitate tissue damage in mice infected with the related Ross River alphavirus , and have been implicated in CHIKV pathogenesis in mice and humans [16 , 18 , 27] . The role of the adaptive immune response in the clearance and pathogenesis of CHIKV infection is also beginning to be explored using mouse models . One study implicated a role for CD4+ T cells in the pathogenesis of CHIK , and another showed persistence in mice deficient in the adaptive immune response [20 , 21] . Our experiments , while confirming some findings of these previous studies , also generated contradictory findings . We found that T cells are not entirely responsible for the CHIK disease process . While T cells are involved in footpad swelling , disease , as indicated by histologic damage , occurs in the absence of T cells . This damage is delayed in RAG1-/- compared to C57BL/6J mice , although the muscle damage seen by day 14 post infection is comparable in the two strains . Also , the adaptive immune system is not necessary to clear footpad disease in the adult mouse . The muscle and footpad recover in RAG1-/- mice , just as in C57BL/6J mice , by day 28 post infection . The adaptive response though does appear to be necessary for complete viral clearance . We used adult 8-10-week-old RAG1-/- mice for several reasons . First , the live-attenuated CHIKV vaccine candidate CHIKV/IRES is progressing toward clinical trials and wanted to explore its safety under immunocompromised conditions because many people in CHIK-endemic regions are immunosuppressed ( HIV , malnutrition ) . Second , weanling C57BL/6J mice , another CHIK model used in previous studies [13 , 28] , are highly susceptible to alphaviral disease regardless of the status of the adaptive immune system . In fact , wt mice are susceptible to CHIK at this age . However , it is well known that older mice ( 8–10 wks ) are not nearly as susceptible to most alphaviral disease , including CHIK . In addition to this age-dependence , 8-10-week-old mice have a fully developed innate immune response compared to younger mice . Using older RAG1-/- mice for our studies overcame many of these confounding factors . Although others have found colocalized viral mRNA and tissue damage in younger RAG1-/- mice and have observed tenosynovitis for extended periods of time , we did not observe these findings using adult RAG1-/- mice [21] . Despite damage in the ipsilateral skeletal muscle of the leg at day 14 , no infectious virus was isolated beyond day 28 post-infection in RAG1-/- mice . Interestingly , the organ most consistently and persistently infected , regardless of the route of infection , was the kidney . Why the kidney is specifically targeted for persistence without apparent histologic damage remains a subject of further investigation . However , the kidney is not an unusual target for persistence , given the findings associated with West Nile virus infections in hamsters [29 , 30] and humans [31] . The changes in the CHIKV viral genome sequence that we observed during persistence do not appear to be random , but instead seem to focus on the non-structural protein ( nsP ) genes , with identical synonymous mutations occurring in independent samples from different animals . Because synonymous mutations are not typically associated with convergence reflecting positive selection , reverse genetic experiments beyond the scope of the current study are needed to determine whether they play a role in CHIKV persistence . Our results along with others point to an important role of the adaptive immune system in controlling persistent CHIKV infection and/or its sequelae . Despite its control of persistent infection , the adaptive immune response does not prevent acute disease or tissue damage . In fact , the presence or absence of the adaptive response does not appear to affect the amount of damage to the skeletal muscle of the injected FP; its absence only appears to delay the damage by a few days . Our data also indicate that FP swelling is not necessarily a good measure of CHIK . Although the absence of T-cells completely abrogated FP swelling , disease was still detected by histopathologic analysis . Overall , our results combined with those of previous studies of the adaptive immune response [20 , 21] , along with those focusing on macrophages/monocytes and interferons [9–18] , indicate that each part of the immune system plays an important , yet different and sometimes overlapping role in controlling CHIKV infection and disease . Our findings could have important implications for the treatment and understanding of CHIKV infections in humans . Individuals with HIV/AIDS may not always manifest overt signs and symptoms despite having disease . Immunocompromised individuals could be viremic for extended periods of time , like our RAG1-/- mice . Though the viremia seen in RAG1-/- mice is too low for mosquito transmission , more study needs to be done in individuals with HIV/AIDS to determine the levels of viremia and if sustained viremia occurs . Those with HIV/AIDS could also conceivably have the granulomatous-type inflammation seen in persistently infected RAG1-/- mice . Our results of persistent kidney infection also suggest that it might be useful to test the urine of acutely infected individuals and those suffering from persistent CHIK symptoms to determine if virus is shed there . Finally our study demonstrates the safety of the CHIKV/IRES vaccine candidate in RAG1-/- mice . Previous studies have demonstrated the safety of this vaccine in mice deficient in a type I interferon response ( A129 ) [13] , which is important given that this innate defense is critical for controlling the early stages of alphavirus infections . Because this live-attenuated vaccine candidate would ideally be deployed in areas of the world where many individuals are deficient in adaptive immune responses , ( i . e . HIV/AIDS and malnutrition ) its safety in this population is critical . Therefore , the lack of CHIKV/IRES detection in any tissues or in serum of RAG1-/- mice , as well as the lack of any signs of disease , including histologic lesions , suggests that it is safe for immunization of immunocompromised people . | Chikungunya virus ( CHIKV ) , the etiologic agent of chikungunya fever ( CHIK ) , is a positive sense , single-stranded RNA virus in the genus Alphavirus . Chikungunya fever begins as a flu-like illness , which progresses to severe arthralgia and debilitating arthritis . This syndrome is often self-limited and rarely fatal; however many patients develop persistent arthralgia that can last from months to years . Currently there is no licensed vaccine or specific treatment for CHIKF , leaving current treatment as purely supportive in nature . The role of the adaptive immune system in disease course and viral persistence is still poorly understood . The lack of an animal model of persistent CHIKF has hindered the study of the role of the adaptive immune response and safety testing of vaccine candidates , which are under development . Due to the fact that the vaccine candidate would be deployed in areas where numerous individuals have an impaired adaptive immune system due to malnutrition or disease ( HIV/AIDS ) ; it is important to study the safety of the vaccine candidate in immunodeficient animals with no adaptive immune system . In this study we present an animal model of persistent CHIKV in adult mice , which lack an adaptive immune system and demonstrate the safety of a live-attenuated vaccine candidate . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [] | 2015 | A Rodent Model of Chikungunya Virus Infection in RAG1 -/- Mice, with Features of Persistence, for Vaccine Safety Evaluation |
African trypanosomes are devastating human and animal pathogens that cause significant human mortality and limit economic development in sub-Saharan Africa . Studies of trypanosome biology generally consider these protozoan parasites as individual cells in suspension cultures or in animal models of infection . Here we report that the procyclic form of the African trypanosome Trypanosoma brucei engages in social behavior when cultivated on semisolid agarose surfaces . This behavior is characterized by trypanosomes assembling into multicellular communities that engage in polarized migrations across the agarose surface and cooperate to divert their movements in response to external signals . These cooperative movements are flagellum-mediated , since they do not occur in trypanin knockdown parasites that lack normal flagellum motility . We term this behavior social motility based on features shared with social motility and other types of surface-induced social behavior in bacteria . Social motility represents a novel and unexpected aspect of trypanosome biology and offers new paradigms for considering host-parasite interactions .
Studying microbial life under conditions that promote a single-cell lifestyle has proven very effective for uncovering important aspects of microbial physiology . However , microbes are social organisms , capable of communicating with one another and engaging in cooperative behavior [1]–[3] . Well-characterized social activities include biofilm formation , social motility , fruiting body development and quorum sensing [1] , [3]–[10] . Social interactions among cells in a population effectively give rise to multicellular communities having specialized functionalities and offering advantages over a unicellular lifestyle . Some of these advantages include increased protection from external antagonists , such as desiccation or host defenses , access to nutrients , exchange of genetic information and enhanced ability to colonize , penetrate and migrate on surfaces [1] , [2] , [11] . In bacterial and fungal pathogens , social interactions have major influences on microbial physiology and disease pathogenesis and considering multicellularity as a general property of bacteria has profoundly changed our understanding of microbiology [1]–[3] , [6] . Most microorganisms , particularly pathogens , are intimately associated with surfaces in their natural environments and preferentially engage in social behavior when exposed to semisolid surfaces [2] , [7] , [10] , [12]–[14] . Commonly , this is manifested as various forms of social motility , including swarming , gliding and twitching [12] , [15] . Each of these surface-induced motilities is influenced by environmental and genetic factors and driven by overlapping yet distinct mechanisms that are not completely understood . The defining feature is cooperative movement of groups of bacteria across a surface , requiring active motility and cell-cell communication among members of the group in response to external stimuli . Once studied only in a few bacteria , such as Proteus mirabilis and Serratia marcescens , surface-induced cooperative motilities are now known to be widespread among both Gram-negative and Gram-positive bacteria , including several important pathogens , such as Salmonella and Pseudomonas spp . [13] , [16]–[18] . Surface-induced social interactions have also been observed in yeasts and fungi , including the opportunistic pathogen Candida albicans [5] , [6] . Thus , various types of surface-induced social behavior are widespread among microorganisms and applying this conceptual framework to studies of bacterial biology has yielded many novel insights . Surprisingly , the paradigm of social behavior has not previously been applied to parasitic protozoa . African trypanosomes , i . e . Trypanosoma brucei and related species , are protozoan parasites that cause significant human mortality and limit economic development in sub-Saharan Africa [19] . T . brucei is transmitted to the bloodstream of a mammalian host through the bite of an infected tsetse fly vector . Parasite motility is important in both hosts and this is especially apparent in the tsetse , where parasites undergo an ordered series of directional migrations that are critical for parasite survival and completion of the life cycle [20]–[23] . Trypanosomes first colonize the midgut , then migrate into the ectoperitrophic space and advance back up the alimentary canal to the mouthparts and from there , to the salivary glands [21] , [23] . Throughout this process , parasites are in intimate contact with tissue surfaces of the tsetse fly . Once in the salivary glands , epimastigotes colonize the epithelial surface , stimulating the final stage of differentiation into mammalian-infective trypomastigotes [22]–[24] . Thus , throughout the tsetse stage of its life cycle T . brucei is in intimate contact with host tissue surfaces and exhibits an implicit requirement for sensing and signaling to guide parasite migration and differentiation . Currently , little is known about how surface contact modulates trypanosome behavior or motility [25] . Here we report that T . brucei engages in social motility when cultivated on semisolid agarose surfaces . This behavior is characterized by the formation of multicellular communities that sense external stimuli and communicate with one another to coordinate movement of the population . T . brucei social motility shares features with surface-induced social behavior in other microorganisms and represents a novel form of motility and intercellular communication not previously observed in these pathogens . As such , our findings present a novel and unprecedented feature of trypanosome biology and provide new concepts for considering development and pathogenesis of parasitic protozoa .
Surface-induced changes to microbial motility and behavior are common among diverse bacteria and protists [2] , [5] , [7] , [26] , [27] . T . brucei spends much of its life cycle in contact with host tissue surfaces and interaction between parasite and tsetse epithelia is well documented [22] , [23] , [28] , yet studies of T . brucei motility to date mainly utilize suspension cultures and do not provide information about how parasite behavior , e . g . motility , is affected by contact with surfaces . As part of our ongoing investigations into trypanosome motility , we thus cultivated procyclic form T . brucei on semisolid agarose plates [29] . We focused on procyclic forms because we know more about the motility apparatus and have more mutants available in this life cycle stage than in bloodstream forms and because the potential impact of parasite motility is most pronounced in this stage [30] . We found that procyclic trypanosomes formed groups of densely-packed cells within 24h post-plating ( Fig . 1 ) . The approximate doubling time on plates was 24 hours ( Fig . S3 ) , indicating that these groups did not arise simply through clonal expansion of single cells . Individuals within each group remained highly motile and actively moved out and back from the group . Interestingly , parasites were often arranged in distinct patterns on the agarose surface , with large , tightly packed groups surrounded by a zone of clearance and then a perimeter of smaller groups ( Fig . 1 ) . To investigate how these patterns arose , we established a system to monitor parasite movements over several hours using time-lapse and video microscopy ( Materials and Methods ) . Time-lapse imaging revealed a striking behavior in which groups of hundreds to thousands of parasites moved en masse across the agarose surface , recruiting neighboring cells and enabling mergers of large groups ( Fig . 1B–F , Video S1 ) . This confirmed that the assembly of large communities was an active process and not simply the result of clonal expansion . The en masse movement of large groups of trypanosomes across the agarose surface suggested some form of cooperation among individuals in the population . We therefore investigated this behavior more closely using increased magnification and greater time resolution ( Fig . 2 , Video S2 ) . These analyses showed that recruitment of individual parasites into a community followed a specific sequence of events as described here . Cells at the periphery of the group were highly motile and moved out and back from the community . We refer to these cells as “scouts” . When scouts came into contact with cells located adjacent to the community , they returned and induced polarized movement of the community outward at this position , forming a multicellular “pseudopod” that extended to recruit the external parasites ( Fig . 2 , Video S2 ) . Mergers of large groups of cells followed essentially the same sequence of events ( Fig . S1 , Video S3 ) . First , single trypanosomes advanced and returned randomly from the group periphery . Second , contact of scouts with an adjacent group biased their movement and initiated a period of reciprocal exchange . This led to formation of a multicellular “pseudopod” between the groups that intermittently broke down and reformed . Ultimately , stable contact was made and the groups merged along a path defined by the “pseudopod” . Thus , the arrangement of cells on the agarose surface resulted from the cooperative movement of parasites into groups , which then expand through recruitment of neighboring cells . Long-term cultivation of social bacteria on semisolid agarose gives rise to large macro-communities that form complex patterns on the agarose surface [2] , [15] , [17] . T . brucei formed macrocommunities within three to six days following inoculation ( Fig . 3A ) . A characteristic feature of this process is that parasites initially collected into small clusters that were distributed around the perimeter of the inoculation site . Parasites in these clusters then advanced outward from the site of inoculation , forming symetrical arrays of radial projections , with a median of 13 projections per inoculum . This pattern is similar to that produced during social motility in Pseudomonas aeruginosa , Myxococcus xanthus and Paenibacillus dendritiformis [7] , [13] , [18] , [31]–[33] , as shown by others ( Fig . 3C ) . Movement of trypanosome projections was polarized , with a single leading edge that advanced at a steady rate on the order of a few microns per minute ( Fig . S2 , Video S4 ) . The leading edge was characterized by a bulbous accumulation of densely packed cells , while the proximal region maintained a constant width ( Fig . 3B ) . Cells along the lateral edge readily moved out and back ( Fig . S2 , Video S5 ) , demonstrating they are not physically restrained . Therefore , polarized migration of projections is governed by parasite actions , rather than physical restrictions on parasite movement . To determine whether social motility requires directional motility , we employed a trypanin RNAi knockdown line that is incapable of directional motility [34] . Trypanin knockdown cells were evenly distributed at the site of inoculation and did not accumulate at the perimeter , as seen for cells having wild type motility ( Fig . 4A ) . Moreover trypanin knockdown cells did not form radial projections ( Fig . 4A–D ) . Cell doubling continued normally , as indicated by the roughly equivalent increase in cell density over time versus control cells ( Fig . S3 ) . Trypanin knockdown and control cells from the same plate were collected and assayed for ( i ) cell number , ( ii ) RNAi induction and ( iii ) motility . Both groups demonstrated an approximately equal cell doubling ( Fig . S3 ) and trypanin protein was undetectable in the knockdown cells ( Fig . 4F ) . Absence of directional motility in trypanin knockdowns was confirmed by direct microscopic examination ( data not shown ) . Therefore , social motility in trypanosomes requires directional motility and is an active process . Radial projections advanced in parallel and did not cross paths ( Fig . 4A–D ) . Moreover , when projections from a control group approached a non-motile group , their movement was either halted or was diverted so as to avoid contact ( Fig . 4A–D ) . When diverted , projections did not cross , rather they continued in parallel , implying that cells in each projection are capable of sensing each other and coordinating their movements . Avoidance of non-motile communities occurred within a radius of approximately 0 . 5–1 cm ( Fig . 4E ) . To determine if this avoidance was uniquely a response to non-motile cells , we inoculated two groups of control cells on opposing sides of a culture plate and followed their development and migration over the course of several days ( Fig . 5 ) . Opposing radial projections either halted advancement , or diverted paths so as to avoid contact with one another . As a negative control , radial projections from a single community of motile cells did not divert their path of migration over time ( Fig . 5F ) . The combined data thus indicate that T . brucei can sense and respond to external signals and that parasites in a community can sense other parasites and may choose to include them in the group or to avoid them .
The impact of cell-cell communication and a multicellular lifestyle on the physiology and pathogenesis of bacteria is now well-established and related phenomena operate in yeast and fungi [1] , [3] , [5] , [6] , [35] . To date however , the paradigm of microbial social interactions has not been applied to parasitic protozoa . We report here that T . brucei is capable of social behavior in which parasites communicate with one another and assemble into multicellular communities with emergent properties that are not evident in single cells . This behavior manifests as groups of parasites engaging in cooperative movement across the surface of semisolid agarose and altering course in response to an external stimulus . We term this behavior social motility , based on analogy to social motility in bacteria . These results demonstrate a novel feature of trypanosome biology and reveal a level of complexity and cooperativity to trypanosome behavior that was not previously recognized . Given the widespread distribution of social interactions among other microbes , we expect our findings to have broad relevance among parasitic protozoa . Social interactions among microbes are manifested in a variety of forms and represent complex behavioral responses for which the underlying molecular mechanisms are not well-understood . As is the case for bacteria [2] , social motility in T . brucei requires directional motility , involves some form of cellular differentiation upon exposure to a semisolid surface and culminates in cooperative cell migration in response to external signals . At early stages parasites merge into groups , while at later stages the behavior has the added feature of groups avoiding one another . This suggests some form of differentiation and is consistent with different stages of social motility observed in some bacteria , such as Paenibacillus spp . [33] . In most cases where it has been investigated , social motility requires a combination of external and internal , i . e . genetic , factors and it is likely that this is also the case in trypanosomes . Based on our observations and what is known in other organisms , a minimum requirement for social motility in T . brucei would be directional motility , the ability to sense an external signal and to transduce this signal into a cellular response and communication between parasites in a group . Trypanosomes are certainly capable of directional motility [36] , [37] and must integrate host-derived and parasite-derived signals to complete their life cycle [24] , [38]–[40] , although their signaling and sensory capacities are poorly understood . The trypanosome genome encodes several components of classical signal transduction pathways , as well as numerous predicted cell surface proteins of unknown function that might serve sensory and/or signaling roles [40]–[45] . The contribution of these proteins to cell-cell signaling or other sensory functions is not known and efforts to address this question have been limited by the lack of a defined in vitro assay for cell-cell signaling . Social motility assays therefore provide an opportunity to test the requirement of trypanosome signaling systems in social motility and overcome a major barrier to dissecting signaling and sensory mechanisms in trypanosomes . Within the tsetse , close contact between parasites , as well as intimate interactions with host tissue surfaces are readily observed [20] , [22] , [23] , indicating that surface-induced social behavior might operate in vivo . However , until appropriate mutants are available for direct investigation , we can only speculate on potential physiological roles for social motility . In this context it is informative to consider whether there are features of the parasite life cycle that might benefit from social motility or related behavior . T . brucei development within the tsetse fly requires parasite migration across and through a variety of host tissues . These migrations lead ultimately to colonization of the tsetse salivary gland epithelia , which the parasites must reach in order to complete development into mammalian-infective trypomastigotes . Trypanosomes progress through specific tsetse tissues in a well-defined order , but the mechanisms responsible for tissue tropisms are unknown . Social motility offers a system in which groups of cells coordinate their movements in response to an external stimulus and thus could provide a mechanism for parasite navigation through host tissues . In bacterial pathogens , cell-cell signaling , assembly into multicellular communities , social motility and other types of surface-induced behavior provide several advantages . Groups of bacteria feed cooperatively , resist hostile environments , prey on other microbes , exchange genetic information and develop functional specializations [1] , [2] . Quorum sensing and biofilm formation induce programs of virulence gene expression , facilitate colonization of host tissues and provide resistance to immune and physical defenses [3] . We speculate that trypanosome cell-cell communication and social behavior may have similar impacts on development and pathogenesis of T . brucei . For example , assembly into groups might facilitate resistance against host defenses in the tsetse [24] , as well as promote tissue colonization and invasion . Social motility might also provide a means to bring parasites together for genetic exchange [46] , [47] . Finally , signaling pathways required for social motility are expected to overlap with host-parasite signaling pathways , about which very little is known . In summary , the identification of social motility in T . brucei reveals a novel and unexpected aspect of parasite biology and provides entirely new conceptual approaches for considering host-parasite interactions .
Three procyclic T . brucei brucei cell lines , Antar 1 R5 Pro/G ITM [21] , 29-13 double marker [48] , and trypanin RNAi ( KHTb12 ) [34] , were used for these studies . While each experiment was not duplicated for each cell line , social motility was observed for both Antar and 29-13 lines . Suspension cultures were maintained using Cunningham's semi-defined medium ( SM ) , supplemented with 10% heat-inactivated fetal calf serum as described previously [49] . For 29-13 cells , the medium was further supplemented with 15µg/ml G418 ( Gibco ) and 50µg/ml Hygromycin ( Gibco ) . For the trypanin RNAi line , 2 . 5µg/ml Phleomycin , 15µg/ml G418 ( Gibco ) and 50µg/ml Hygromycin ( Gibco ) were included in the medium and RNAi was induced by adding 1µg/ml tetracycline . Cell doubling was monitored using a Z1 Coulter Particle Counter ( Beckman Coulter , USA ) . Cultivation on semi-solid agarose plates was adapted from [29] . Four percent ( w/v ) agarose ( SeaPlaque GTG Agarose , Cambrex-LONZA , ME , USA ) solution was made in MiliQ water , autoclaved and cooled to 65°C . A 1∶10 dilution of this 4% stock solution was prepared in pre-warmed ( 42°C for 20min ) SM culture medium supplemented with the appropriate antibiotics for selection . The resulting 0 . 4% agarose solution was cooled to 37°C for 1h . In most cases ethanol ( final concentration 1% ) was added to the medium . A 13ml aliquot was poured into Petri Dishes ( 100×15mm ) , which were then dried without lids for 1 . 5h in a laminar flow hood at room temperature . For inoculation onto the plate , 5µl of cells from a suspension culture at a density of 1 . 5×107cells/ml were added on the agarose surface . For the experiments in Fig . 2 and S1 , 50ul of cells were spread on the surface by gently rotating and rocking the plate . Trypanin RNAi lines were induced for 72h with 1µg/ml tetracycline in suspension culture prior to plating . Inoculated plates were dried for 3 min without lids , closed and sealed with parafilm and incubated as for suspension cultures at 27°C . For Fig . 1A , the plate was imaged using a Zeiss Axioskop II microscope with a 2 . 5×LD Plan NeoFluor objective and Zeiss Axiocam camera . For Fig . 1B–F ( Video S1 ) , the plate was imaged using a Zeiss Axiovert 200M microscope with a 2 . 5×LD Plan NeoFluor objective and a COHU RS-170 high performance CCD camera ( COHU , Inc . ) . Images were captured at 1 frame per 10min at room temperature using Adobe premiere Elements 1 . 0 ( Adobe Systems ) . Time stamps are indicated in the panels . For Video S1 , images were compiled into a movie using NIH-ImageJ ( http://rsbweb . nih . gov/ij ) . The playback speed is 5 frames per second ( 3000× original speed ) and elapsed time is 24h . For Fig . 2 ( Video S2 ) and Fig . S1 ( Video S3 ) , plates were maintained at 28°C , 5% CO2 in a CTI humidified live cell cultivation chamber equipped with heating insert and CTI 3700 controller from Zeiss , Inc . This chamber allows independent control of humidity , temperature and CO2 on the microscope stage . Plates were monitored on a Zeiss Axiovert 200M microscope , using a 10×LD Plan NeoFluor objective and a COHU RS-170 High performance CCD camera ( COHU , Inc . ) . For Fig . 2 ( Video S2 ) , images were captured once every 5 sec and played back at 10 frames per second ( fps ) , giving a final playback speed of 50× . For Fig . S1 ( Video S3 ) , the video was recorded in real-time using a VCR , then digitized in AVI format at 30 frames per second ( fps ) using an in-line Sony Handycam digital camera as an analog/digital converter and Adobe Premier Elements ( Adobe Systems ) . Individual images were extracted at 1 fps , exported into QuickTime video format using the Sorenson TM CODEC within Adobe Premier Elements , and played back at 30 fps . The final playback speed is thus 30× real speed and elapsed time is 10 minutes 57 seconds . For Fig . 3A , 4A–E and 5 , plates were imaged at the indicated times post plating using an Olympus Stylus 770 SW digital camera and processed using Adobe Photoshop 8 . 0 . For Fig . 3B , the plate was imaged as described above for Fig . 1B–F . For Fig . S2 ( Videos S4 and S5 ) , time-lapse images were captured and compiled into video as described above for Fig . 1B–F . The playback speed is 6429× and elapsed time is 21 . 43 hours for Video S4 and 8 . 9h for Video S5 . Cells were collected in PBS from the agarose plate , counted and washed two times in PBS . The equivalent increase in opaqueness of control and trypanin RNAi communities on plates , Fig . 4 , indicated that they continued doubling at equivalent rates and direct cell counting confirmed this . Protein samples were prepared and subjected to Western blot analysis as described [49] , using 1×106 cell equivalents per lane . Monoclonal anti-trypanin antibody [50] was used at 1∶5000 , and monoclonal anti-β-tubulin E7 hybridoma supernatant was used at 1∶5000 . The anti-β-tubulin antibody was developed by Michael Klymkowsky , University of Colorado and was obtained from the Developmental Studies hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa , Department of Biological Sciences , Iowa City , IA 52242 . Secondary antibody was horseradish peroxidase-coupled goat anti-mouse ( BioRad ) used at 1∶2500 . | African trypanosomes , e . g . Trypanosoma brucei , and related kinetoplastid parasites cause morbidity and mortality in several million people worldwide . Trypanosomes are protists and are thus generally considered to behave as single-celled microorganisms . In other microorganisms , social interactions among individuals lead to development of multicellular communities with specialized and advantageous capabilities versus single cells . The concept of bacteria acting as groups of cells communicating and cooperating with one another has had a major impact on our understanding of bacterial physiology and pathogenesis , but this paradigm has not been applied to parasitic protozoa . Here we report that T . brucei is capable of social behavior when exposed to semisolid surfaces . This behavior , termed social motility , is characterized by the assembly of parasites into multicellular communities with emergent properties that are not evident in single cells . Parasites within communities exhibit polarized movements and cooperate to coordinate their movements in response to an external stimulus . Social motility offers many potential advantages , such as facilitating colonization and navigation through host tissues . The identification of social behavior in T . brucei reveals a novel and unexpected aspect of parasite biology and provides new concepts for considering host-parasite interactions . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"infectious",
"diseases/neglected",
"tropical",
"diseases",
"cell",
"biology",
"microbiology/parasitology"
] | 2010 | Social Motility in African Trypanosomes |
Skeletal muscles provide metazoans with the ability to feed , reproduce and avoid predators . In humans , a heterogeneous group of genetic diseases , termed muscular dystrophies ( MD ) , lead to skeletal muscle dysfunction . Mutations in the gene encoding Caveolin-3 , a principal component of the membrane micro-domains known as caveolae , cause defects in muscle maintenance and function; however it remains unclear how caveolae dysfunction underlies MD pathology . The Cavin family of caveolar proteins can form membrane remodeling oligomers and thus may also impact skeletal muscle function . Changes in the distribution and function of Cavin4/Murc , which is predominantly expressed in striated muscles , have been reported to alter caveolae structure through interaction with Caveolin-3 . Here , we report the generation and phenotypic analysis of murcb mutant zebrafish , which display impaired swimming capacity , skeletal muscle fibrosis and T-tubule abnormalities during development . To understand the mechanistic importance of Murc loss of function , we assessed Caveolin-1 and 3 localization and found it to be abnormal . We further identified an in vivo function for Murc in Erk signaling . These data link Murc with developmental defects in T-tubule formation and progressive muscle dysfunction , thereby providing a new candidate for the etiology of muscular dystrophy .
Muscular dystrophies ( MD ) are a heterogeneous group of genetic diseases that result in progressive dysfunction of skeletal muscle [1] . Muscle weakness and loss of muscle mass can begin in childhood , while some forms of the disease manifest in adulthood . Different groups of muscles are affected depending on the type of MD and complications can include confinement to a wheelchair , scoliosis , difficulty swallowing , impaired breathing , cardiomyopathies , and early death [2 , 3] . Although MD can be caused by mutations in more than 30 different genes [4] , common phenotypic characteristics include disruption in the membrane network [5] and abnormal calcium flux in skeletal muscle [6 , 7] . Loss of membrane integrity has been shown by the intracellular accumulation of proteins normally absent in skeletal muscle [8 , 9] . Membrane stability requires functional costameres [10] , which link intracellular force generation with the extracellular matrix [11 , 12] via the dystrophin glycoprotein complex [13] . Cell membrane ( sarcolemma ) stability is critical to muscle cells , which must maintain a resting membrane potential in order to propagate action potentials via voltage-gated ion channels [14] . Excitation-contraction coupling requires specialized membrane structures , the transverse ( T ) -tubules and terminal cisternae , for calcium flux to trigger an actin/myosin dependent contraction . Two terminal cisternae are found on opposing sides of the T-tubules and form a structure known as the triad [15] . Disruption of calcium signaling also has long term consequences for skeletal muscle health by altering signaling pathways required for muscle remodeling [16] and may underlie the necrosis seen in MD [6] . Interestingly , in a zebrafish model of Duchenne muscular dystrophy , a serotonin reuptake inhibitor was shown to improve skeletal muscle health through improved calcium homeostasis [17] . In mouse , the development of the T-tubule network begins with invaginations in the sarcoplasmic reticulum at embryonic day ( E ) 15 followed by postnatal maturation of the network [15] . Thus far , a few proteins are known to be required for T-tubulogenesis: MTM1 , JPH1 , MG29 , DYSF , BIN1 , and Caveolin-3 ( CAV3 ) . Loss of MTM1 in humans results in a disease known as myotubular myopathy and is characterized by severe muscle atrophy at birth , and mice lacking MTM1 show T-tubule disorganization with disrupted calcium homeostasis and excitation-contraction coupling defects [18] . Bin1 , which possesses membrane deforming properties , has been shown to be enriched at T-tubules and induce tubular invaginations in the plasma membrane [19] . Loss of CAV3 can result in several skeletal muscle diseases including type 1C limb-girdle muscular dystrophy [20] . Knockout of Cav3 in mouse leads to disrupted dystrophin glycoprotein complex distribution as well as abnormal T-tubulogenesis [21] , and Cav3 is necessary for skeletal muscle differentiation in vitro [22 , 23] and in zebrafish [24] . Primarily expressed in striated muscle , Cav3 is an integral membrane protein required for the formation of caveolae , invaginations in the cell membrane that function in lipid storage , signaling and endocytosis [25] . Caveolin-1 ( Cav1 ) , a broadly expressed member of the Caveolin family , is required for lipid homeostasis and insulin sensitivity [26] . The Caveolin proteins interact with a class of membrane-associated proteins termed Cavins , which are enriched in caveolae and are critical for their formation . The Cavin family consists of 4 members , one of which is Cavin4 , also known as Murc ( Muscle-restricted coiled-coil protein ) . Cavin4/Murc was originally discovered through SAGE screening of a cDNA library prepared from adult mouse hearts [27] and expression analysis revealed high mRNA expression in both heart and skeletal muscles . In cultured C2C12 myoblasts , CAVIN4/MURC expression was found to increase during differentiation [28] . Additionally , shRNA-mediated knockdown of CAVIN4/MURC resulted in the inhibition of myotube formation . Concurrent with these initial studies , Cavin4/Murc was also identified by scanning the genome for homologs of three known Cavin genes and was thus termed “Cavin4" [29] . The four members of the Cavin family were found to localize to the cell membrane in a CAV1-dependent manner , and in skeletal muscle , CAVIN4/MURC co-localizes with CAV3 in putative caveolae [29] . In cardiomyocytes , CAVIN4/MURC localizes to caveolae and T-tubules and its over-expression results in disrupted caveolar shape [30] . Furthermore , in a patient with rippling muscle disease and mosaically expressed CAV3 , muscle fibers lacking CAV3 had a distinct absence of CAVIN4/MURC expression [29] . The Cavin proteins possess substantial homo- and hetero-oligomerization capacity via two helical repeat ( HR ) domains [31] . Although their functional importance for caveolar formation is unclear , Cavins have the capacity to remodel membranes and it has been proposed that they polymerize to form large assemblies of rod-like structures stabilized by interactions with membranes [31] . Here , we report the generation of a zebrafish line carrying a TALEN-induced murcb mutant allele and show that Cavin4b/Murcb is necessary for skeletal muscle function in both larval and adult fish . Cavin4b/Murcb deficiency results in progressive dysfunction and fibrosis of skeletal muscles . Additionally , we show that disruption of Cavin4b/Murcb is associated with changes in Erk signaling and Caveolin expression in adult zebrafish . Larvae lacking Cavin4b/Murcb exhibit an increased number of caveolae , mislocalization of Caveolin-1 expression , and T-tubule defects .
Using a combinatorial Translating Ribosome Affinity Purification ( TRAP ) approach , we previously found that murca and murcb are expressed in the somites of developing zebrafish [32] . To understand the function of the murc genes , we first examined their conservation and spatiotemporal expression pattern . We aligned the protein sequences of Cavin4a/Murca and Cavin4b/Murcb with mammalian and Xenopus paralogs ( S1A Fig ) and concluded that murca and murcb are indeed duplicated genes by examining their genomic structure ( S1B Fig ) and through phylogenetic analysis of the protein sequences ( S1C Fig ) . To determine when murca and murcb are expressed , we performed RT-PCR analysis on RNA isolated from developing zebrafish embryos and larvae at various times and found that mRNA for both murc paralogs are detectable beginning around 29 hours post fertilization ( hpf ) and expressed through 128 hpf ( Fig 1A ) . In adult tissues , we found murca and murcb to be highly expressed in skeletal muscle , and to a reduced extent , both murc paralogs are also expressed in the kidney ( Fig 1B ) . Interestingly , murcb , but not murca , is expressed in the adult zebrafish heart . To determine the spatial distribution of murca and murcb in zebrafish larvae , we performed in situ hybridization at 48 hpf and found expression in the somites with a higher signal at the somite boundaries ( Figs 1C and S1D ) . To test the function of murcb in vivo , we generated a mutant allele with TALE-nucleases designed to cut in the first exon . We used the TALEN Targeter 2 . 0 software from the Bogdanove laboratory to design the constructs and the Golden Gate cloning kit from Bogdanove and Voytas to generate the constructs from which mRNA was made and injected into one-cell stage embryos ( Fig 1D ) [33 , 34] . An allele of murcb with an 11 base pair deletion and 3 base pair insertion was identified and designated s983 . The putative protein product encoded by the s983 allele includes a frameshift in the 28th codon ( p . G28QfsX83 ) and a premature stop in the 110th codon . The mutant protein is predicted to lack the conserved HR1 and HR2 domains that mediate interactions with the Cavin proteins [31] . To determine whether the s983 lesion results in altered Cavin4b/Murcb protein expression , we performed antibody staining on transverse sections of skeletal muscle harvested from 10 wpf zebrafish and found a reduction in anti-Cavin4/Murc immunoreactivity ( Fig 1E ) . To confirm that the s983 homozygous mutant larvae lack Cavin4b/Murcb , we performed tandem mass spectrometry and found that peptides corresponding to Cavin4b/Murcb were not detected in lysates from mutant larvae but were clearly present in heterozygous controls ( S1 Table , S2 Fig ) . Altogether , these data indicate that s983 is a severe mutant allele . To characterize the phenotype of murcb mutant zebrafish , we examined 10 wpf mutant and siblings from in-crosses of murcbs983/+ animals and measured body mass and length and found that murcbs983/s983 fish were smaller with an average mass of 0 . 25 g compared to 0 . 46 g for murcb+/+ fish and 0 . 45 g for murcbs983/+ fish ( Fig 2A and 2B ) . The average length of murcbs983/s983 animals at 24 . 4 mm also differed significantly from the murcb wild-type and heterozygous animals with average lengths of 27 . 7 mm and 27 . 9 mm , respectively . To analyze skeletal muscle structure in murcb mutant fish , we harvested trunk tissue from animals sacrificed at 10 wpf and examined transverse sections histologically . Hematoxylin and eosin ( H&E ) staining revealed that the overall organization and symmetry of the trunk was not altered in murcbs983/s983 fish ( Fig 2C ) compared with sibling controls , though skeletal muscle fibers appeared smaller . F-actin visualization in these transverse sections revealed that skeletal muscle fibers were smaller and less uniform in size and shape in murcbs983/s983 fish compared to heterozygous siblings ( Fig 2D ) . To assess the skeletal muscle pathology of Cavin4b/Murcb deficient zebrafish , we performed trichrome staining of transverse sections made from 12 mpf trunk tissue and found evidence of fibrosis in mutant animals but not in sibling controls ( Fig 2E ) . To determine whether there is a role for Cavin4b/Murcb during the early development of skeletal muscle structure , we initially examined murcbs983/s983 larvae at 5 dpf and found no gross morphological differences with murcbs983/+ siblings ( Fig 2F ) . Additionally , whole mount phalloidin staining of skeletal muscle F-actin revealed no obvious differences between murcbs983/s983 larvae and murcbs983/+ siblings ( Fig 3 ) at 3 dpf . In contrast , at 6 dpf we found that skeletal muscle fibers were smaller and less consistent in size in murcbs983/s983 larvae compared with murcbs983/+ siblings ( Fig 3 ) . To determine whether the membranes of the skeletal muscle were intact in murcb mutant larvae , we performed Evans blue dye staining [35] , and found the somites of both heterozygous murcbs983/+ and murcbs983/s983 larvae to be impermeable to the dye ( S3A Fig ) . In order to understand whether apoptosis was contributing to the disruption in skeletal muscle fiber size in murcbs983/s983 larvae , we performed acridine orange staining on whole mount zebrafish and found no difference between mutants and murcbs983/+ siblings ( S3B Fig ) . Additionally , we assessed mRNA stability for both murca and murcb by RT-PCR in both murcbs983/+ and murcbs983/s983 larvae and found that transcripts for both murc paralogs were relatively unaffected ( S3C Fig ) . To determine whether structural defects in the skeletal muscle of murcb mutants affect skeletal muscle function , we quantified the maximum swimming velocity and acceleration following the startle response . At 10 wpf , murcbs983/s983 fish were significantly slower with a maximum velocity of 0 . 473 m/s compared to their heterozygous siblings that had a maximum velocity of 0 . 891 m/s ( Fig 4 ) . Acceleration was also reduced in murcbs983/s983 fish at 14 . 2 m/s2 compared with 26 . 1 m/s2 for heterozygous siblings . In order to determine whether there was a functional defect in Cavin4b/Murcb deficient skeletal muscle during development , we measured the maximum velocity and acceleration at 60 and 80 hpf . We found that at 60 hpf , mean maximum swimming velocity was not significantly different between murcbs983/+ and murcbs983/s983 embryos ( Fig 4 , 0 . 062 versus 0 . 056 m/s , respectively ) ; however , mean maximum acceleration was reduced in murcbs983/s983 embryos ( 1 . 9 m/s2 ) compared to heterozygous siblings ( 2 . 6 m/s2 ) . At 80 hpf , the difference in maximum velocity between murcbs983/+ and murcbs983/s983 larvae was significantly different ( 0 . 104 versus 0 . 077 m/s , respectively ) . The difference in maximum acceleration between murcbs983/+ and murcbs983/s983 larvae was also significantly different ( 3 . 69 versus 2 . 59 m/s2 , respectively ) . To rule out an effect of loss of Cavin4b/Murcb on the neuromuscular junctions , we performed whole mount α-bungarotoxin and Synaptic Vesicle glycoprotein 2 staining at 80 hpf . No obvious differences in post-synaptic structures could be observed between murcbs983/+ and murcbs983/s983 larvae ( S4 Fig ) . To investigate the underlying cause of the swimming defect in Cavin4b/Murcb deficient larvae , we isolated mRNA from mutant and sibling zebrafish at 72 hpf for microarray analysis . We examined the list of genes with an expression ratio of greater than 1 . 5 or less than 0 . 7 in murcbs983/s983 compared to murcbs983/+ samples by the PANTHER database protein class statistical overrepresentation test . 1309 transcripts were differentially expressed and recognized in the PANTHER database . Of these 55 are annotated in the ion channel protein class and 26 are annotated as voltage-gated ion channels while the expected number from 1309 input transcripts is 27 ( p = 0 . 000177 ) and 10 ( p = 0 . 00291 ) , respectively ( Fig 5 ) . The relative mRNA expression levels of the differentially expressed genes in the voltage-gated ion channel class are also shown ( Fig 5 ) . Because voltage-gated ion channels are critical to excitation-contraction coupling , we assessed the ultrastructure of the skeletal muscle in murcb mutant larvae to determine whether there were T-tubule abnormalities . Electron microscopy ( EM ) performed on 80 hpf larvae revealed that Cavin4b/Murcb deficiency was associated with a disruption of T-tubule development and maturation ( Fig 6A ) . Quantification of the electron micrograph data showed that sibling controls had 88 . 3% intact T-tubule triad structures while Cavin4b/Murcb deficient larvae had 10% intact triads ( S2 Table ) . Additionally , we found an increased number of caveolae in murcbs983/s983 larvae at 80 hpf compared with sibling controls ( Fig 6B ) . Because of the role of Cavin4/Murc in the localization of Caveolin proteins , we examined the localization of Cav1 in transverse sections from 10 wpf fish and found increased and mislocalized anti-Cav1 immunoreactivity ( Fig 7A ) . In sections from heterozygous siblings , Cav1 was predominantly localized to the surface of skeletal muscle fibers while in murcbs983/s983 skeletal muscle , there appeared to be additional Cav1 staining in the internal , putative membrane structures . In contrast to what has been reported in mammals [21 , 29] , we found that Cav3 does not localize to the surface of skeletal muscle fibers in 10 wpf zebrafish , but rather appears to be associated with internal structures and not with Cavin4/Murc ( S5 Fig ) . Interestingly , Cavin4b/Murcb is nonetheless required for Cav3 localization , which is disrupted in murcbs983/s983 samples ( Fig 7A ) . In order to determine if Murcb has an effect on Cav1 localization in larvae , we performed whole mount antibody staining on larvae at 80 hpf . We found that Cav1 localization was altered in murcbs983/s983 larvae compared with heterozygous siblings with increased localization in a striated pattern ( Fig 7B ) . To profile the staining of the striations , we measured the pixel density relative to the background of confocal images along a 3 μm distance perpendicular to the striations . We then plotted the pixel intensity on the y-axis versus the distance across the image on the x-axis and found that compared to heterozygous siblings , murcbs983/s983 larvae have a higher Cav1 antibody reactivity in these striations . Because caveolae are known to regulate growth factor signaling , we also assayed phosphorylated Erk1/2 in Cavin4b/Murcb deficient fish . We performed immunoblotting of skeletal muscle lysates from 3 murcbs983/s983 zebrafish and 3 murcbs983/+ siblings at 10 wpf and found in each mutant that phospho-Erk1/2 was increased relative to total Erk1/2 ( Fig 7C ) .
Here , we have shown a requirement for Cavin4b/Murcb in the development and function of skeletal muscle . Additionally , we have shown that Cavin4b/Murcb is necessary in vivo for caveolae function , Caveolin-1 and -3 localization , and Erk signaling in skeletal muscle . Our analyses of a targeted murcb mutation reveal the importance of caveolae for the formation of T-tubules during myogenesis and suggest an additional etiology for muscular dystrophy . The complex membrane structures necessary for skeletal muscle function , such as the T-tubule network , require specialized membrane microdomains and their associated proteins for formation and function . Disruption of these proteins can lead to human disease . The nature of the Cavin4b/Murcb dependent muscle pathology described here is likely due to defects in the formation of T-tubules as evidenced by immature triad structures at 80 hpf . Caveolae are known to be abundant in T-tubules during development [22]; Cav3 is required for T-tubule shape and organization , and Cav3 mutant mice exhibit similar myopathologies as patients with type 1C limb-girdle MD [23] . Skeletal muscle fibers lacking a functional T-tubule network likely have calcium flux and excitation-contraction coupling deficiencies that could result in ion channel gene expression changes initially , and subsequently result in muscle wasting [6] . Several genetic models have yielded insights into the pathology of other types of MD . For example ENU-induced mutations in zebrafish laminin β2 ( lamb2/softy ) cause embryonic muscle degeneration due to loss of the extracellular matrix supporting the sarcolemma during skeletal muscle contraction [36] . Zebrafish dystrophin ( sap/dmd ) mutants display skeletal muscle fiber detachment and membrane damage resulting in uptake of Evans blue dye [37] . These models of MD present phenotypes early in development and are therefore identifiable in a forward genetic screen . In contrast cavin4b/murcb , a much smaller gene than lamb2 or dmd , is less likely to be mutagenized by random means and is more likely to be determined to be phenotypically normal prior to 5 dpf . Therefore , a targeted genetic approach was critical to understand the function of murcb in zebrafish . The pathomechanics of Cavin4b/Murcb deficiency appear to be rather different than those of dystrophin deficiency as dystrophin mutant larvae show muscle detachment by 48 hpf and die around 31 dpf [37] . Therefore , cavin4b/murcb mutants represent a distinct tool to better understand processes such as T-tubulogenesis and how caveolinopathies lead to skeletal muscle diseases . Mutations in other Cavin genes have been linked to skeletal muscle disease . Mutations in human CAVIN1 ( PTRF ) have been reported to cause MD and generalized lipodystrophy , defects that are likely secondary to caveolae disruption [38] . Patients in this study had histopathologies in common with MD such as variable skeletal muscle fiber size , while other phenotypic changes were not consistently present , suggesting the influence of genetic modifiers . It will be interesting to examine CAVIN4 localization and expression in patients with CAVIN1 mutations , as there appears to be interdependence in their expression and regulation . Indeed , it was recently shown that in Cavin1 KO mice , CAV1 protein expression was reduced and CAVIN4 expression was not detectable [39] . This reduction in CAVIN4 expression was evident in both skeletal muscle and heart , and suggests that a homeostatic feedback mechanism exists that either stabilizes a Cavin hetero-oligomeric complex or regulates transactivation . Kovtun et al . have recently characterized Cavin interactions , finding that CAVIN1 can bind to other Cavin family members as well as homo-oligomerize [31] . This interaction is mediated through the Helical Region ( HR1 ) , which forms a trimeric coiled coil . Cavin oligomerization seems to be organ-specific [39] , and may depend on interactions with CAVIN4 or CAV3 , which have restricted expression patterns . In human skeletal muscle , CAVIN4 expression correlates very highly with mosaic expression of CAV3 [29] . While the function of Cavin4/Murc remains poorly understood , a role in signal transduction has emerged in recent studies . In cultured cardiomyocytes , hypoxia induces Cavin4/Murc expression in an Erk-dependent manner [40] . A Cavin4/Cav3 complex co-localizes with α1-adrenergic receptors ( α1-AR ) in cardiomyocytes and α1-AR-induced cardiac hypertrophy is attenuated in Cavin4/Murc mutant mice [30] . Apart from a direct upstream effect of Cavin4/Murc on Erk signaling , one explanation for the observed increase in phosphorylated Erk in Cavin4/Murc deficient zebrafish skeletal muscle is chronic injury [41] . Satellite cell exhaustion from continual repair is thought to contribute to the pathology of MD . Interestingly , Cavin4/Murc , but not Cav1 or Cav3 , is highly upregulated in “alert” satellite cells [42] ( see http://www . ncbi . nlm . nih . gov/geo/query/acc . cgi ? acc=GSE55490 ) raising the possibility that Cavin proteins are necessary for muscle membrane structures as well as for repair , perhaps through a role in cell fusion [43 , 44] . Further examination of the Cavin4/Murc KO mouse , which in contrast to zebrafish has not been reported to have a skeletal muscle phenotype [30] , should be directed toward muscle injury and repair . Caveolar proteins have been linked to cardiomyopathies as well as MD [45–47] . In mouse , mutations in Cav3 result in cardiomyopathy characterized by myocyte hypertrophy , dilation of the heart chambers , and a reduction in fractional shortening [48] . Given the interdependence between Cavin and Caveolin protein expression and distribution , the role of Cavin4 in cardiomyopathy should be studied further . CAVIN4 mutations have previously been linked to Dilated Cardiomyopathy ( DCM ) . Rodriguez et al found 6 heterozygous mutations ( N128K , R140W , L153P , S307T , P324L , and S364L ) in patients with DCM but none in patients with hypertrophic cardiomyopathy or in healthy controls[49] . Our Cavin4b/Murcb deficient fish line will be a useful tool to determine whether these mutations in human CAVIN4 indeed result in a loss of protein function . As there is an overexpression phenotype for Cavin4/Murc reported in mouse [30] , the cloning of the zebrafish cavin4b/murcb promoter/enhancer will likely be necessary for rescue experiments with the human mutant alleles . The pathomechanics of the various forms of MD have been of great interest due to the severity of the disease , limited treatment options , and the difficulty of implementing gene therapy [4] . Due to the accessibility of developing skeletal muscle in zebrafish larvae , zebrafish models of MD can be used to test how aberrant calcium flux , T-tubule development , and growth factor signaling can impact progressive skeletal muscle disease . A targeted mutation approach has revealed that Cavin4b/Murcb is necessary for the development of the T-tubule network during embryonic myogenesis and the proper localization of Caveolin-1 and -3 in adult muscle . These changes are associated with reduced skeletal muscle fiber size and uniformity , fibrosis , and reduced muscle function , suggesting that CAVIN4 may be important to the etiology of MD early in the development of skeletal muscle .
Antibodies used for immunoblotting were against Erk1/2 ( Cell Signaling , 9102 ) , phosphoThr202/Tyr204-Erk1/2 ( Cell Signaling , 4370 ) , Caveolin-1 ( BD Transduction Labs , 610059 ) , Caveolin-3 ( BD Transduction Labs , 610420 ) , and Cavin4/Murc ( Sigma-Aldrich , HPA021021 ) . Whole mount antibody and phalloidin stainings were performed on larvae fixed with 4% paraformaldehyde , 4% sucrose , 77 mM Na2HPO4 , 22 . 6 mM NaH2PO4 , 120 μM CaCl2 pH 7 . 35 overnight at 4°C , permeabilized with PBS , 0 . 5% Triton X-100 for 3 hours at room temperature , and stained with Alexa-488 conjugated phalloidin ( Cytoskeleton , PHDG1 ) at a 1:100 dilution in PBS , 0 . 5% Triton X-100 . Whole mount immunofluorescence staining was performed on fixed larvae that were dehydrated in 4 steps to 70% ethanol in PBS and stored at 4°C . Prior to staining , larvae were rehydrated in 4 steps to 1X PBS , digested with proteinase K at 30 μg/ml for 30 minutes at room temperature and blocked with PBS , 5% normal goat serum ( NGS ) , 0 . 5% Triton X-100 and incubated with indicated antibodies diluted 1:100 in PBS , 5% NGS , 0 . 5% Triton X-100 overnight at 4°C . Larvae were then washed 4 times with PBS with 0 . 5% Triton X-100 , incubated with secondary antibodies , mounted in agarose , and imaged on an LSM700 confocal microscope ( Zeiss ) . H&E and immunofluorescence stainings were performed on trunk tissue dissected from adult fish and fixed in 4% paraformaldehyde , 4% sucrose , 77 mM Na2HPO4 , 22 . 6 mM NaH2PO4 , 120 μM CaCl2 pH 7 . 35 overnight at 4°C , transferred to PBS with 20% sucrose for 24 hours and then frozen in OCT for cryosectioning . 10 μm transverse cryosections were stored at -80°C prior to staining . Sections were briefly dried , then permeabilized with PBS with 0 . 5% Triton X-100 . Alexa-488 conjugated phalloidin was diluted 1:100 in PBS with 0 . 5% Triton X-100 and incubated for 3 hours at room temperature followed by 5 washes with PBS with 0 . 5% Triton X-100 . Indicated antibodies were diluted in PBS , 5% NGS , 0 . 5% Triton X-100 and incubated overnight at 4°C . Sections were then washed 4 times with PBS with 0 . 5% Triton X-100 and incubated with Alexa-488 or Alexa-568 conjugated secondary antibodies and imaged on an LSM700 confocal microscope . Presynaptic structures were visualized in 80 hpf fish that were fixed in 4% paraformaldehyde for 4 hours at 4°C , washed 4 times with PBS , and permeabilized with protease K ( 10 μg/ml ) for 30 minutes at room temperature . Larvae were then washed in PBS , 0 . 5% Triton X-100 , 1% DMSO ( PBT-DMSO ) before blocking overnight at 4°C in PBT-DMSO with 1% BSA and 2% goat serum . Antibodies against Synaptic Vesicle 2 ( SV2 , Developmental Studies Hybridoma Bank ) were diluted 1:20 in blocking solution and incubated overnight at 4°C . After washing with PBT-DMSO , larvae were incubated with Alexa-568 conjugated donkey anti-mouse antibodies ( Molecular Probes ) for 2 hours at room temperature . Larvae were then washed , mounted , and imaged on an LSM 780 confocal microscope ( Zeiss ) . Postsynaptic structures were visualized in 80 hpf fish that were fixed in 4% paraformaldehyde for 4 hours at 4°C , washed 4 times with PBS , permeabilized with 0 . 2% collagenase P ( Roche , 11213805001 ) for 50 minutes then washed 4 times with PBS . Embryos and larvae were incubated with 20 μg/ml Tetramethylrodhamine ( TRITC ) conjugated bungarotoxin ( Sigma Aldrich , T0195 ) for 30 min at room temperature . Larvae were then washed , mounted , and imaged on an LSM 780 confocal microscope . Acridine orange ( Sigma , 235474 ) staining was used to identify apoptosis in live larvae at 80 hpf . Larvae were incubated in a solution of 10 μg/ml acridine orange for 30 minutes , then washed with PBS three times and imaged through a Fluorescein-Isothiocyanate ( FITC ) filter using an LSM780 confocal microscope . Evans Blue Dye ( Sigma Aldrich , E2129 ) was dissolved in egg water at 0 . 1% ( w/v ) . 4 dfp 1-phenyl-2-thiourea ( PTU ) -treated larvae were transferred to a nylon mesh basket and submerged in dye for 60 minutes . The basket was removed from the dye solution , washed with egg water , and larvae were transferred to a fresh dish for imaging and genotyping . RNA and gDNA were extracted from individual larvae at 72 hpf using Trizol ( Life Technologies ) . Following genotyping , the aqueous phases from at least 10 Trizol extractions were combined by genotype and purified over microspin columns ( ZymoResearch ) . RNA expression from heterozygous and mutant larvae was analyzed by single-color microarray ( 8x60K Zebrafish Array XS , Oaklabs , Germany ) . The arrays were imaged using a SureScan Microarray Scanner and Agilent’s Feature Extraction software version 11 was used to read and process the TIFF images . Data were normalized by using the ranked median quantiles [51] . Microarray data were archived in the NCBI database ( GSE70858 ) . PANTHER ( v 9 . 0 , release 20140124 ) database protein class statistical overrepresentation test ( release 20141219 , with Bonferroni correction ) was performed on a list of genes that expressed at a ratio of greater than or equal to 1 . 5 times or less than or equal to 0 . 7 times the expression found in heterozygous murcb versus homozygous murcb larvae and having a normalized signal with a value greater than 50 . 1309 differentially expressed genes were identified in the database . 25708 genes comprise the reference set for Danio rerio . Swimming speed and acceleration were measured in 10-week-old fish that were transferred to a 40 cm by 40 cm tray containing system water above which a digital camera was mounted . Movies were recorded following the startle response and frame-by-frame location within the tray over time was determined using PhysMo V2 software ( http://physmo . sf . net ) . Distance calibration was made using a ruler submerged in the tray . Measurements of swimming larvae were performed in a similar manner in a 10 cm dish . Deformed fish were not used in the swimming analysis . Statistical significance was determined using a two-tailed unequal variance t-test . TALENs were designed against murcb using the TALEN Targeter 2 . 0 software [33 , 34] . The plasmid kit used for generation of TALENs was a gift from Daniel Voytas and Adam Bogdanove ( Addgene kit # 1000000024 ) . Capped mRNA encoding the TALEN was generated using the mMESSAGE mMACHINE SP6 Transcription Kit ( Life Technologies , AM1340 ) and 200 pg of each arm were injected into one-cell stage embryos . TALENs were tested by HRMA analysis on an amplicon from gDNA harvested from individual embryos at 24 hpf . Injected F0 siblings from functional TALENs were raised to adulthood and outcrossed to identify founders . F1 fish were then raised to adulthood and gDNA was isolated from the caudal fin to identify F1 heterozygous fish by HRMA . PCR products from the HRMA were cloned into pGEM-T Easy ( Promega ) for sequencing . gDNA was isolated by digesting individual embryos or clipped fins in proteinase K at 65°C for 1 hour with occasional vortexing followed by inactivation at 95°C for 10 minutes . For screening of founders , HRMA primers were designed to amplify between 80 and 100 nucleotides . Upon identification of a suitable allele , HRMA primers were redesigned to amplify a region immediately surrounding the mutation ( Forward: 5’-TCTCTGCTGGAGAGGGTGTC , Reverse: 5’-CTGGCAGGCCTGAACATTGT , Annealing temperature: 65°C ) . RT-PCR primers for murca were: Forward: 5’- TGTGATCTATCAGGGTGAAAAAG , Reverse: 5’- TGGAGAACGCAGCCTTGATGTTC . RT-PCR primers for murcb were: Forward: 5’- GTCATCTACCAGGGAGATAATGAG , Reverse: 5’- GTCATGTTCTCGCGTGAGAAGGTC . Zebrafish larvae were fixed with 3% glutaraldehyde ( Merck ) , followed by 4% OsO4 in 0 . 1 mmol/L sodium cacodylate buffer ( pH 7 . 4 ) . After dehydration in ethanol and propylene oxide , they were embedded in Epon as described . Semi-thin ( 1 μm ) sections were stained with toluidine blue and viewed in a Leica DM microscope . Ultrathin sections were stained with uranyl acetate and lead citrate and viewed and photographically recorded under a Philips CM 10 electron microscope . Animal studies were carried out following guidelines and standard procedures set forth by the Regierungspräsidium Gießen Dezernat V 54 ( Protocol No . B2/342 ) . | Membrane structures are critical to skeletal muscle excitation-contraction coupling , and disruptions in key membrane proteins can lead to muscular dystrophy . Caveolae , micro-domains in the cell membrane , are known to be important in forming muscle-specific membrane structures such as the T-tubule network and play a role in some forms of muscular dystrophy . Two classes of proteins help form caveolae: Caveolins and Cavins . However , it is poorly understood how Cavins function in skeletal muscle development or whether Cavin dysfunction can lead to muscular dystrophy . To address these questions , we generated a Cavin4b/Murcb deficient zebrafish line by targeted TALEN mutagenesis and found that Cavin4b/Murcb is necessary for the formation and function of skeletal muscle . Cavin4b/Murcb mutant zebrafish exhibit progressive skeletal muscle deterioration , disrupted membrane structures early in development and aberrant growth factor signaling . Cavin4b/Murcb deficient fish have smaller and irregularly shaped muscle fibers with fibrosis , similar to what is observed in muscular dystrophy . Cavin4b/Murcb mutant fish also exhibit mislocalized Caveolin-1 and Caveolin-3 as well as disrupted T-tubules , suggesting that Cavin4b/Murcb regulates the distribution of caveolar proteins and that this misregulation impacts the development of the T-tubule network . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"swimming",
"skeletal",
"muscles",
"vertebrates",
"animals",
"biomechanics",
"biological",
"locomotion",
"animal",
"models",
"osteichthyes",
"developmental",
"biology",
"muscle",
"proteins",
"model",
"organisms",
"skeletal",
"muscl... | 2016 | Cavin4b/Murcb Is Required for Skeletal Muscle Development and Function in Zebrafish |
Traction forces exerted by adherent cells on their microenvironment can mediate many critical cellular functions . Accurate quantification of these forces is essential for mechanistic understanding of mechanotransduction . However , most existing methods of quantifying cellular forces are limited to single cells in isolation , whereas most physiological processes are inherently multi-cellular in nature where cell-cell and cell-microenvironment interactions determine the emergent properties of cell clusters . In the present study , a robust finite-element-method-based cell traction force microscopy technique is developed to estimate the traction forces produced by multiple isolated cells as well as cell clusters on soft substrates . The method accounts for the finite thickness of the substrate . Hence , cell cluster size can be larger than substrate thickness . The method allows computing the traction field from the substrate displacements within the cells' and clusters' boundaries . The displacement data outside these boundaries are not necessary . The utility of the method is demonstrated by computing the traction generated by multiple monkey kidney fibroblasts ( MKF ) and human colon cancerous ( HCT-8 ) cells in close proximity , as well as by large clusters . It is found that cells act as individual contractile groups within clusters for generating traction . There may be multiple of such groups in the cluster , or the entire cluster may behave a single group . Individual cells do not form dipoles , but serve as a conduit of force ( transmission lines ) over long distances in the cluster . The cell-cell force can be either tensile or compressive depending on the cell-microenvironment interactions .
Recent research has demonstrated that cells communicate with each other as well as with their microenvironments through mechanical signaling [1] , [2] , [3] , [4] , [5] , [6] , in addition to biochemical ones [7] , [8] , [9] , [10] , [11] , [12] , [13] , [14] . Many physiological processes , including cell adhesion [15] , [16] , [17] , cytoskeleton polarity [13] , [18] , cell proliferation [19] , [20] , cell differentiation [12] , [21] , [22] , embryogenesis [23] , [24] , cancer metastasis [7] , [25] , and wound-healing [26] , [27] , can be significantly influenced by the transmission and sensation of physical forces between the cells and their microenvironments . For example , exposure of HCT-8 human colon cancer cells to soft substrates results in a profound stable cell state transition from an epithelial phenotype to a metastasis-like phenotype ( MLP ) [7] , [8] , [28] , [29] , [30] , [31] . Adherent cells actively sense the local anisotropy of their microenvironment [2] , [18] , [32] , [33] as well as the forces applied by neighboring cells [1] , [4] , [11] , [34] , [35] , followed by polarization of stress-fibers and synergetic cell functions . Hence , accurate estimation of the traction forces exerted by the cells on their substrates under various physiological conditions can provide important insight on many fundamental questions regarding the mechanical interactions between various cell types and their microenvironment [36] , [37] , [38] . Over the past few decades , several seminal techniques to assess the cellular traction forces have been developed ( see reviews [14] , [39] , [40] , [41] , [42] , [43] , [44] ) . However , most of them are limited to computation of traction forces exerted by single , isolated cells . Efforts at visualizing cellular traction forces may be traced back to 1980s when Harris et al . used thin polymeric silicone substrates for cell culture , and observed the wrinkling phenomena caused by the traction of migrating cells [45] . However , quantitative estimation of the traction from the wrinkling of silicone substrates is challenging due to the inherent non-linearity of the problem . From 1995 on , Lee , Jacobsen and Dembo et al . , as well as other groups , developed several traction force microscopy techniques ( TFM ) to quantify the cellular traction produced by migrating or stationary cells on soft substrates [46] , [47] , [48] , [49] , [50] , [51] , [52] , [53] , [54] . TFM computes the cell traction forces from the deformation of a soft substrate with known elastic properties , such as polyacrylamide ( PA ) gel , on which cells are cultured . The deformation is measured from the displacements of micro-fluorescent markers embedded in the substrate . The motion is measured from two images . First image is taken with the cells adhered to the substrate . Here , the cells have generated traction force on the substrate , and the image gives the deformed configuration of the soft substrate . Then cells are removed from the substrate through trypsinzation , and a second image is taken . Subsequently , the substrate is relieved of cell traction , and the image shows the un-deformed configuration of the substrate . A comparison of the two images gives the displacement field of the substrate's top surface due to cell tractions . Digital image correlation method ( DICM ) is used to quantify the displacement field . The traction field is estimated from the displacement field . Several methods have been proposed for force estimation ranging from analytical methods , i . e . the Boussinesq formulation ( either using Bayesian likelihood regularization method [51] , [55] or Fourier transformed approach [49] ) , to computational methods like finite element analysis ( FEA ) [56] . The Boussinesq formulation approach , which assumes the substrate as a semi-infinite elastic half space [57] , was first adopted by Dembo and Wang , et al . , to compute the traction forces from the displacement fields followed by regularization [51] , [55] , [58] , [59] . Since the Boussinesq formulation involves solving an inverse problem , the solution demands computational regularization schemes to predict the approximate traction solutions . Importantly , Butler , Trepat and Fredberg , et al . [49] , [60] , [61] , [62] , [63] made significant progress in mitigating some pitfalls of the regularization scheme by solving the Boussinesq equation using Fourier transform . Later Schwarz et al . introduced a new method to compute traction forces only at the focal adhesion site of the cell by assuming that the cell force transfer occurs only through these sites[50] . Some novel platforms , such as the photobleaching-activated monolayer with adhesive micro-patterns developed by Scrimgeour et al . [64] and the elastic substrates with micro-contact printing demonstrated by Stricker et al . [65] , were also used to characterize the cell traction force . Furthermore , a FEA-based technique was also developed by Yang et al . to greatly improve the accuracy of traction force calculations [56] . The FEA method no longer depends on the Boussinesq formulation and thus is not limited by the semi-infinite elastic half space assumption [66] , [67] . Recently additional contribution has been made in traction force computation in three dimensions [19] , [68] , [69] , [70] , [71] , [72] , [73] . 3D TFM techniques compute the 3D traction force fields from the cell induced 3D displacement and strain fields obtained using laser scanning confocal microscopy ( LSCM ) and digital volume correlation ( DVC ) . However , it is challenging to obtain the Z-dimension displacement field and the technique can only be applied to single cell cases , rather than multiple cells or cell clusters . The above studies focused on traction force computation for single cells far from their neighbors , i . e . cells that do not interact mechanically with each other . However , live cells do interact with their neighbors chemo-mechanically and form cell clusters [7] , [29] , [37] , [74] , [75] , [76] . In this paper we present a novel finite-element-based TFM technique to compute the traction fields generated by multiple cells and clusters . We first present a theoretical proof showing that the 3D traction field computed from prescribed displacement field of the substrate is unique . We verify the uniqueness by considering a 2-cell case . We test the accuracy of the computational technique by applying a known force on PA gel substrate using a micro-needle , and by comparing the experimental force with the computed one . Finally , we compute the traction fields generated by multiple cancerous and fibroblast cell clusters , and reveal that cells might be under compression in such 2D clusters . We believe that the present technique may enable better examination and understanding of a variety of biological phenomena involving homotypic and heterotypic cells and cell cluster interactions [77] , [78] , [79] .
Consider a 3D linear elastic solid with volume V in static equilibrium . Its boundary , S , consists of Su and Sσ ( S = Su+Sσ ) where displacements and traction are prescribed respectively . Proposition: Given displacement field at Su and traction at , the corresponding traction at Su is unique . ( Note: indices i , j = 1 , 2 , 3 correspond to x , y , z Cartesian coordinates respectively; all equations follow standard tensor notation and summation convention ) . Supporting material Text S1 presents the proof of the proposition . We illustrate our computational scheme as follows . Consider two separate cells on a soft elastic substrate . The substrate is adhered to a rigid surface ( such as glass ) at the bottom . The lateral boundary of the substrate is far from the cells . In the finite element scheme , the substrate is modeled as a rectangular pyramidal solid body . It is discretized as a collection of small cubes with common nodes . We need to prescribe three boundary conditions , namely any combination of forces ( Fx , Fy , Fz ) and displacements ( ux , uy , uz ) , at each of the surface nodes . For example , ( Fx , uy , uz ) can be a boundary condition at a surface node . To ensure that the body is at rest ( no rigid body translation or rotation ) , at least two of the nodes are prescribed with ux , = uy , = uz = 0 . Given the boundary conditions , finite element scheme calculates the deformation of the solid body such that the total energy is minimized . Thus the displacements at each node within the body , and at the surface nodes where forces are prescribed are evaluated . This leads to the evaluation of strains and stresses using the elastic properties of the solid ( Young's modulus and Poisson's ratio for the isotropic gel ) . Surface traction is calculated from the stress near the surface and normal vector to the surface ( ) , as shown in Supplementary Materials S1 . Surface nodal forces are calculated from an area integral of traction at the vicinity of the node . Thus , the analysis provides the forces at nodes where displacement is prescribed , and displacements where forces are prescribed . If ( Fx , uy , uz ) is prescribed at a surface node for example , one gets ( ux , Fy , Fz ) at that node . Even though the solution is unique in principle , errors are introduced if the discretization is coarse . With finer discretization , the solution converges to the correct one . This convergence test is often employed to gage the accuracy of the solution . In our problem with two cells , we prescribe zero displacement boundary conditions at the bottom surface and at the four vertical sides of the body ( Fig . 2 ) . Thus all the nodes on the bottom and the vertical sides are fixed . For simplicity of illustration , consider that there are a few nodes on the top free surface outside the cell boundary , and a few nodes within ( Fig . 2 ) . Our objective is to calculate the traction on these nodes . We can experimentally measure displacements ( ux , uy , uz ) at all the nodes on the surface . They are generated by cell forces , although we do not know the precise locations of these forces . We also know that the surface outside the cells has no traction , and that each cell or cell cluster produces a traction field that is self-equilibrated , i . e . , the sum of forces applied by the cell or the cell cluster on the substrate is zero . Cell traction can be evaluated by prescribing either of the two boundary conditions: Remarks . ( 1 ) The mixed boundary scheme applies exact boundary condition ( zero force ) at nodes outside the cells . Hence none of the displacements ( ux , uy , uz ) need to be prescribed at these nodes . Thus , it is not necessary to measure the displacements of the beads outside the cells . Due to the exact boundary conditions outside the cells , the traction solution is expected to be more accurate . However , errors will be introduced if the cell boundary is incorrectly defined and there are nodes that fall outside the cell boundary where cells apply traction . In cases where the cell boundaries cannot be identified due to imaging conditions ( Fig . 3 ) , displacements should be prescribed for regions nearby the cells . ( 2 ) Displacement uz and Poisson's ratio: It is shown in the supplementary material ( Supplementary materials text S3 , Fig . S1b and c ) , that if the Poisson's ratio of the gel approaches 0 . 5 , then the in-plane displacements , ( uy , uz ) , on the surface of the gel are independent of the out-of-plane component of traction ( Fz ) . That is , ( ux , uy ) are determined by ( Fx , Fy ) on the surface . Similarly , uz is determined by Fz on the surface only . Thus , in order to evaluate the in-plane traction only , one needs to measure and prescribe in-plane displacements only at the surface nodes , and prescribe arbitrary boundary condition in z direction ( i . e Fz = 0 or uz = 0 ) at all surface nodes , when Poisson's ratio is close to 0 . 5 . We experimentally measured the Poisson's ratio of our gel as 0 . 47±0 . 02 ( Fig . S3b , n = 5 ) . In order to estimate the in-plane traction only , we have prescribed Fz = 0 for all nodes within the cells in the rest of the paper . This results in an error of less than 2% in the calculation of in- plane forces Fx and Fy ( Supplementary materials text S3 and Fig . S3b ) . If Fz is desired , one needs to measure and prescribe ( ux , uy , uz ) at the surface nodes . Also , if Poisson's ratio is much less than 0 . 5 ( e . g . , 0 . 35 ) , ( ux , uy , uz ) must be prescribed at the nodes within the cells even when only in-plane traction is desired . In this section , we demonstrate computationally that the traction solution from finite element simulation is unique as long as the full 3D boundary conditions are prescribed . We define two circular boundaries representing two cells with half-cell distance apart on a soft gel surface . The diameter of each boundary is chosen as 20 µm , close to real cell size . A three-dimensional finite-element ( FEM ) block model is generated ( ANSYS 12 . 0 Workbench Package ) to represent the PA gel substrate [79]–[98] . The gel is presumed linear elastic , isotropic , and homogeneous in their mechanical properties for a wide range of deformations [78] , [99] . The Elastic modulus , E , of the gel is 1KPa ( our experimental value is 1 . 05±0 . 17 kPa , measured by AFM indentation ( n = 15; Fig . S3a ) , [99]–[101]] ) . The model height is 70 µm , same as the thickness of PA gel used in experiments . We first apply an in-plane force field ( Fig . 4a ) within each boundary , and compute the corresponding displacement field , ux , uy , uz ( Fig . 4b ) . Second , we use the computed ux , uy and uz within the cell boundary on the surface ( Fig . 4c ) , and zero-traction conditions outside the boundaries to calculate the traction within the cells ( Fig . 5d ) . A comparison between the prescribed and the calculated forces from the two steps shows close quantitative agreement ( within 1% ) ( Fig . 4e-f ) . Note that individual cells or cell clusters generate self-equilibrated traction on the substrate . Hence , we use a measure of accuracy of the traction solution by defining the error ratio , ( 2 ) where Fxi and Fyi , are the nodal force components within the individual cells , and i = 1 , n , the number of nodes within the cell or cell cluster boundary . For exact solution , ε = 0 . In this section , we demonstrate the applicability of the method by evaluating the traction induced by two neighboring cells . Here , two monkey kidney fibroblasts were plated on PA gel ( 1 kPa ) with Poisson's ratio of 0 . 47 ( Fig . 5a ) . Two different regions ( two sets of Su and Sσ ) were selected to prescribe the displacement boundary conditions: ( 1 ) displacement field underneath the two cells were prescribed in the model ( the white parts in Fig . 5b ) , whereas the traction-free condition was applied outside the cells ( the black part in Fig . 5b ) ; ( 2 ) the displacement field within a region enclosing both cells was prescribed ( the white part in Fig . 5c ) , whereas the traction-free condition was applied outside this region ( the black part in Fig . 5c ) . The out-of-plane force , Fz , was prescribed as zero within the cellular regions in ( 1 ) and ( 2 ) . The traction fields were calculated for both cases ( Fig . 5d , e , g , h ) , and compared ( Fig . 5f and i ) . The RMS of node-by-node traction difference inside 2-cell region ( superscripts indicate regions 1 and 2 ) was 21 . 7 Pa , which shows close match with only 5 . 1% of maximum traction inside the cells ( 426 . 8 Pa ) . In this section , we compare our mixed-boundary condition method with traditional whole-field displacement boundary condition method , which requires iterative calculation and has been successfully used by Fredberg , et al [49] , [102] . Briefly , the iteration calculation proceeded as follows: ( a ) we assigned the complete 2D DICM ( digital image correlation method ) displacement data ( ux , uy ) for all nodes of the top surface of the gel ( both intracellular and extracellular regions; Fig 6a-b ) . We prescribe Fz = 0 within the cluster for both the mixed boundary condition and iterative methods . ( b ) The traction field was solved using FEM . Then all the forces in the extracellular region were replaced by Fx = Fy = Fz = 0 to satisfy the traction-free condition , while the forces in the intracellular region were retained intact . ( c ) The new traction field was used to generate a new displacement field using FEM . Thus a new displacement field was computed within the intracellular region . ( d ) The computed intracellular displacement field was replaced with the DICM displacement field ( ux and uy ) , while the computed extracellular ux , uy , and uz from previous step were retained intact . ( e ) The steps ( b ) , ( c ) , ( d ) were repeated until the solution converged , i . e . , the difference between the root mean square ( RMS ) of surface nodal forces in two consecutive cycles became less than 5% ( Fig . 6c-e ) . Our computational results showed that the solutions from mixed-boundary and iterative methods converge ( Fig . 6c-e ) . We found , the difference between the root mean square ( RMS ) value of traction from the two methods was 1 . 6×10−1 kPa ( Fig . 6f ) , less than 3 . 8% of the maximum computed cell traction . The difference between the RMS of the nodal forces was 0 . 2 nN , which is 0 . 25% of the maximum nodal force at cell cluster - substrate interface ( Fig . 6g ) . The distribution of traction |t| and forces at nodes ( Fig . 6h-6j ) shows good agreement between the two methods . We used ε ( Eqn 2 ) as a measure of accuracy of the traction solution . In FEM , convergence test is required to determine the optimal mesh size needed to obtain the accurate solution . Three mesh sizes , 3 . 23 µm , 4 . 84 µm , and 6 . 45 µm were tested , as shown in Fig . 7a-c , and used to calculate the traction field of the same cell cluster by mixed-boundary condition method . The distribution of nodal traction and forces showed minor difference between the three mesh sizes ( Fig . 7a-c and 7e ) . The values of ε were 4 . 74% , 6 . 69% , and 6 . 12% for mesh size of 3 . 23 µm , 4 . 84 µm , and 6 . 45 µm respectively ( Fig . 7d ) . Therefore , in the following computations , mesh size of 4 . 84 µm was used for analysis . The upper limit of mesh size is dependent on the specific cell size and the gradient of the traction field produced by the cell . A starting point on mesh size can be <20% of cells size . A key attribute of the present method is the computation of traction fields generated by multiple cell clusters interacting with each other . Each cluster may consist of multiple cells , and the cluster size might be similar to or larger than the thickness of the soft substrate . Hence the effect of the glass-gel interface needs to be considered , and the gel may not be treated as half space . In the following , we study several cell clusters ( Figs . 8–10 ) and outline the main biological findings . The mixed-boundary condition method was used to compute the traction fields .
The majority of fundamental physiological processes in tissue development , health , and disease are coordinated by the collective activities of multiple cells [60] , [62] , [76] , [102] , rather than single cells[10] , [103] . To understand how mechanical traction applied by neighboring cell cluster groups could specify or mediate the tissue functionalities [7] , [8] , [11] , [75] , [104] , [105] , [106] , robust cellular traction evaluation method is indispensable . In the present study , we developed a finite element element-based traction force microscopy ( TFM ) to accurately compute and visualize the traction maps resulting from multiple cell clusters . The uniqueness , convergence , and correctness of traction solutions are substantiated . We showed that as the gel Poisson's ratio >0 . 4 , the in-plane traction can be obtained with minimal error from the in-plane displacement field alone . For Poisson's ratio <0 . 4 , both in and out of plane traction depend on both in and out of plane displacement boundary conditions , and it is essential to measure these displacements to compute any of the traction components . The method presented is applicable to substrates with any value of the Poisson's ratio . It calculates the full 3D traction field given the 3D displacement boundary condition within cells or cell clusters . Moreover , unlike the classical TFM methods that are based on Boussinesq solutions [39] , [40] , [48] , [49] , the FEM takes into account the effect of substrate thickness and nearby environment . It is now known that cells can sense the substrate depth within the cellular length scales by showing distinct morphological variation on the gel substrate with same Elastic modulus but with varying thickness[22] , [107] . We applied the method to compute the traction generated by multiple cell clusters . Some of the clusters were more than 100 µm in size consisting of many cells , while others were in close proximity to each other . The computational scheme presented here is ideal for studying such clusters , since the domain of traction field is much larger than the thickness of the gel , and one needs to account for the finite thickness of the substrate . A few interesting biological insights emerge from these analyses . First , the cluster may behave as a single contractile unit where the peripheral cells serve as anchorage sites . Force is transmitted between distant peripheries by the cells inside the cluster . Thus the cells are subjected to tensile intercellular forces , as if the peripheral cells are pulling the interior cells outward . It needs to be seen whether there are specific cells within the cluster that generate the force , or all the cells behave as contractile actuators . In any case , the cells probably use cell-cell junctions and cytoskeleton to transmit the force through the cluster . We also found instances where traction is limited to small regions well within the clusters . These regions can have locally balanced traction ( forming dipoles ) , leaving the rest of the clusters nearly traction free and weakly adhered to the substrate . These clusters are spherical in morphology , as expected . The traction free regions tend to minimize the surface area by being circular , just as a free-standing cell cluster takes a spherical shape . It is plausible that the cells within the circular clusters are under compression due to the surface tension of the peripheral cells . In any case , the interior traction maps can be highly dynamic . When cell clusters merge , the traction map can change their orientations , and the net force can increase by an order of magnitude over short times . It is known that cells generate contractile forces . Thus , it is expected that the cells in a 2D cluster will be under intercellular tension . We found evidence to the contrary . If the cells are on soft substrates where they do not spread much , but they adhere to the substrate , then some of the cells in the cluster may be subjected to compression . We found regions within such clusters where the neighboring cells apply repulsive forces on the substrate , i . e . , the cells are pushing against each other while being adhered to the substrate . One possible explanation might be that the neighboring cells are growing , but their adhesion sites are stationary . In conclusion , we developed a robust FEM-based cell traction force microscopy technique to estimate the traction forces produced by multiple cells and clusters . The utility of the technique is exemplified by computing the traction force fields generated by multiple monkey kidney fibroblast ( MKF ) and pre-MLP human colon cell ( HCT-8 ) clusters in close proximity . The developed technique is user-friendly and computationally inexpensive . Our FEM-based traction force microscopy provides a powerful tool to probe multi-cell questions involving assembly/disassembly dynamics of cell ensembles , tissue network formation , and wound healing . Future work is needed to determine the subcellular processes involved in mechano-sensing and regulation , and their respective timescales .
Polyacrylamide ( PA ) gel substrates with 1 kPa stiffness used in present study were made by mixing 12 . 83% ( v/v ) of acrylamide ( Sigma-Aldrich , Inc . ) , 1 . 54% ( v/v ) of N , N-methylene-bisacrylamide ( Sigma-Aldrich , Inc . ) , 2% ( v/v ) of 1 µm diameter fluorescent micro-beads ( Invitrogen , Inc . ) and 10 mM Hepes ( Gibco . , Inc . ) [7] , [11] . Solution was vortexed thoroughly for 5 min to obtain uniform distribution of beads . TEMED and ammonium persulfate ( Fisher Scientific , Inc . ) were used to initiate PA gel crosslinking . Chemical modification of glass slides and preparation of PA gels were carried out following the procedures described previously [50] , [80] , [81] , [82] , [83] , [84] , [85] . Briefly , a circular glass coverslip ( Fisher Scientific , Inc . ) of 1 . 2 cm in diameter was placed on an acrylamide solution drop on activated coverslip and placed on the bottom of a petri dish . Capillarity spreads the drop and fills the space between the circular coverslip and the activated coverslip . The gel was cured at room temperature and reached to the stabilized thickness of 70 µm [82] , [85] , [86] . The circular glass coverslip was peeled off from the gel that remained on the activated cover slip . The surfaces of the air dried PA gels were activated by incubating in 97% hydrazine hydrate ( Acros Organics . ) for 12 h followed by a complete rinsing with DI water and 30 minutes incubation along with gentle shaking in 5% acetic acid ( Avantor Performance Material , Inc . ) [7] , [8] , [11] , [13] , [81] . Solution of human fibronectin ( 25 µg/ml , BD Biosciences ) was prepared by dissolving in phosphate buffer saline ( PBS ) and the carbohydrate groups of fibronectin were oxidized by sodium periodate ( Sigma-Aldrich , Inc . ) . To minimize the displacement noise and rigid body motion during imaging , the glass slides was firmly adhered to the bottom of 30 mm petri dish using adhesive glue ( Henker Consumer Adhesive , Inc . ) . Full experiment procedures and sample characterization are provided in Supporting Materials Text S4-S9 and Figures S1–S5 . | Adherent cells sense , transduce and respond to their microenvironment by generating traction forces on their surroundings . To accurately understand these mechanotransduction processes , it is critical to have a robust and reliable method for traction force visualization and quantification . However , most cell traction force microscopy methods are limited to only single cell traction force analysis . Considering that most physiological processes are essentially collective multi-cellular events , there is a need for traction force microscopy methods capable of analyzing traction forces resulting from multiple cells . We have developed a novel and robust multi-cellular traction force microscopy method for computing cell traction on soft substrates , and applied it to compute traction field generated by both multiple cells and cell clusters . We verified the accuracy , robustness , and efficiency of the method by theoretical , numerical and experimental approaches . Our method provides a powerful toolset to pursue the mechanistic understanding of collective biological activities , such as cancer metastasis and neuromuscular interactions . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biotechnology",
"physics",
"mechanical",
"engineering",
"nanoengineering",
"classical",
"mechanics",
"materials",
"science",
"engineering",
"and",
"technology",
"biology",
"and",
"life",
"sciences",
"biomaterials",
"computational",
"biology",
"physical",
"sciences",
"bioph... | 2014 | A Novel Cell Traction Force Microscopy to Study Multi-Cellular System |
CD4 T cells harboring HIV-1/SIV represent a formidable hurdle to eradicating infection , and yet their detailed phenotype remains unknown . Here we integrate two single-cell technologies , flow cytometry and highly multiplexed quantitative RT-PCR , to characterize SIV-infected CD4 T cells directly ex vivo . Within individual cells , we correlate the cellular phenotype , in terms of host protein and RNA expression , with stages of the viral life cycle defined by combinatorial expression of viral RNAs . Spliced RNA+ infected cells display multiple memory and activation phenotypes , indicating virus production by diverse CD4 T cell subsets . In most ( but not all ) cells , progressive infection accompanies post-transcriptional downregulation of CD4 protein , while surface MHC class I is largely retained . Interferon-stimulated genes were also commonly upregulated . Thus , we demonstrate that combined quantitation of transcriptional and post-transcriptional regulation at the single-cell level informs in vivo mechanisms of viral replication and immune evasion .
CD4 T lymphocytes that support HIV-1/SIV replication are central to the development of AIDS-defining illness as well as to the establishment of cell-associated viral reservoirs that persist despite years of antiretroviral therapy [1 , 2] . Despite the clinical importance of these infected T cells , their properties are poorly defined at the cellular level due to the difficulty of characterizing them in vivo or directly ex vivo [3] . Barriers include their low frequency , estimated at 10−3–10−6 during chronic untreated HIV-1 infection [2 , 4 , 5] , and lack of defining markers on their surface . Consequently , most data about infected cells is derived from either in vitro infection models or analysis of ex vivo bulk cell populations comprised mostly of uninfected cells . Ex vivo studies employing methodology to specifically identify and characterize rare in vivo HIV-1/SIV-infected cells , as defined by expression of viral RNA , DNA or protein , are essential to gaining a better understanding of cells harboring virus . To date , only a few studies have accomplished this feat , and typically only a small number of surface markers were measured . From these , there is compelling evidence for cell surface CD4 ( and CD3 for SIV ) downregulation , a hallmark of in vitro HIV-1/SIV infection , in vivo [6–10] , although an earlier report demonstrated CD4 retention [4] . MHC class I downregulation , another well-described in vitro phenomenon , has also been observed ex vivo , albeit subtly and not consistently in all hosts [7 , 8] . Markers of T cell exhaustion ( CTLA-4 , PD-1 , and TIGIT ) , peripheral follicular helper cells , Th17 cells [10] , T cell memory , and activation ( HLA-DR ) also appear elevated . These basic phenotyping findings warrant more extensive investigation examining a greater number of markers and including the application of more sensitive methodology not reliant on viral protein detection to identify infected cells . To further overcome this long-standing challenge to the field and establish a more detailed profile of elusive in vivo infected cells , we integrated two complementary approaches into a single technology , measuring the simultaneous expression of surface proteins ( by flow cytometry ) and over 90 host genes ( by highly multiplexed qPCR ) with single-cell resolution [11] . Using PCR assays specific for multiple forms of viral RNA , we identify SIV-infected cells directly ex vivo in different stages of the viral life cycle spanning early to highly productive states . Cell surface protein and transcriptional profile is compared across each infection stage to determine differential expression patterns associated with infection in individual cells . Moreover , we demonstrate post-transcriptional regulatory events in single infected host cells and correlate these events with viral gene expression .
Progression through the HIV/SIV life cycle is characterized by sequential accumulation of multiply-spliced , singly-spliced , and unspliced viral RNA ( vRNA ) , which thereby distinguish discrete infection stages . We used RT-qPCR assays to identify cells transcribing SIV by expression of spliced ( tat/rev , env ) , unspliced ( gag ) , and total ( LTR ) vRNA ( S1 Fig ) . In vitro , cell-associated spliced viral transcript expression followed expected kinetics during SIVmac239 infection of rhesus macaque PBMCs ( S1 Fig ) [12 , 13] , and reverse transcriptase inhibition blocked de novo tat/rev expression , confirming specificity for transcription from proviral DNA . In sum , the spliced vRNA assays identify active viral transcription , while gag and LTR detect more prevalent vRNA species not necessarily specific for gene expression . We determined the frequency of infected cells in vivo during acute and chronic SIV infection of 14 rhesus macaques ( Macaca mulatta; S1 Table ) , which reproduce most clinical and virological features of HIV-1 infection in humans [14] . Unmanipulated viable memory CD4 T cells from multiple tissues were sorted by flow cytometry at serial 3-fold dilutions in replicate to estimate the percent positive for spliced or unspliced vRNA ( Fig 1A , top ) . tat/rev ( multiply-spliced ) RNA+ cells ranged from <0 . 01 to 6 . 4% ( mean 2 . 0% ) of memory CD4 T cells at 9-14d post-infection ( Fig 1B , S2 Fig ) . On average , gag+ cells were present at ~10-fold higher frequency than tat/rev+ cells , comprising 0 . 2–80% ( mean 26% ) of memory CD4 T cells . Control experiments performed in the absence of RT for three lymph node samples yielded a two-fold reduction in gag+ cells ( Fig 1B; 3% , 20% , and 36% ) , similar to but slightly less than previously reported frequencies of viral DNA+ T cells during acute SIV infection [15] . We attribute the DNA signal in our assay primarily to cytoplasmic reverse transcription products rather than integrated provirus as the latter is not efficiently recovered by the cell lysis protocol ( S3 Fig ) , which may explain our lower DNA values . Not surprisingly , the frequencies of tat/rev+ and gag+ cells were strongly correlated with one another as well as with both total cell-associated proviral DNA measured in bulk memory CD4 T-cells and plasma viremia ( Fig 1C and 1D ) . These results are consistent with virus production by tat/rev+ cells detected by our assay and recapitulate similar correlations observed in HIV-1 infection [16 , 17] . To characterize viral gene expression in individual cells , we measured co-expression of vRNAs in FACS-sorted single cells directly ex vivo from six acutely SIV-infected macaque specimens ( Fig 1A , bottom ) . Notably , this quantitation is sensitive and linear at the single-copy per cell level [11] . PBMC , lymph node ( LN ) , and jejunum tissues were chosen for analysis based on predetermined infected cell frequencies ≥1% within memory CD4 T cells ( Fig 1B ) . Four distinct subsets of vRNA+ cells were apparent based on the quantity and identity of vRNA species within a cell: 1 ) gag+ and/or LTR+ , spliced vRNA–; 2 ) tat/rev+ only; 3 ) tat/rev+env− and gag+ and/or LTR+; and 4 ) tat/rev+env+ and gag+ and/or LTR+ ( Fig 1E and 1F ) . The profile of cells positive for gag , LTR , or both , in the absence of either of the spliced transcripts ( hereafter referred to as stage 1 ) is consistent with early , abortive , or latent infection [18 , 19] . In cells positive for multiple vRNAs , vRNA levels were highly correlated with one another ( S4 Fig ) . 102−105 vRNA copies were expressed per cell , similar to estimates from HIV-infected cells [17] . Compared to tat/rev− ( stage 1 ) infected cells , tat/rev+ cells ( stages 3–4 ) contained ~100-fold more gag RNA , indicating expression of large quantities of unspliced vRNA . Notably , the env−subset of tat/rev+ cells ( stage 3 ) expressed less tat/rev per cell than the env+ ( stage 4 ) cells , as would be expected early in the viral life cycle prior to nuclear export of partially processed vRNA and Tat-mediated transcriptional activation of the viral promoter . Thus the combination and quantitative expression of viral transcripts can be used to determine the stage of the viral life cycle in individual cells ( Fig 1F ) . Single-cell viral gene co-expression analysis among tat/rev+ cells revealed that the majority ( 40–70% ) of tat/rev+ cells also expressed gag , LTR , and env , while the remainder was largely env–gag+LTR+ ( 25–45% ) ( Fig 1G ) . An unusually low proportion of env+ cells ( 16% ) was present in animal 08D227 ( PBMC ) , despite abundant tat/rev RNA ( Fig 1E and 1G ) . This may reflect viral sequence divergence from consensus SIVsmE660 in this animal , limiting detection by the env assay . Jejunum contained a unique subset ( stage 2; 21% ) in which tat/rev was the only vRNA detected , and at very low copies per cell . Together , these data further support tat/rev expression as a marker of virus-producing cells [18 , 19] , and we therefore consider these cells productively infected . To identify cellular factors associated with viral infection at the single-cell level , differential host gene expression between uninfected cells and cells at each infection stage ( Fig 1F ) was assessed . Genes involved in T-cell activation , cell cycle regulation , signaling , viral restriction , and interferon response were selected for analysis ( S2 Table ) . Differential gene expression was performed as previously described [20] , with cell infection status modeled as a discrete covariate and each infection stage coefficient tested against uninfected cells ( stage 0 ) . Both the proportion of cells positive for a gene and the RNA copies per positive cell were considered . Among PBMC specimens , nine of the measured genes were altered in one or more infected cell populations , of which CD28 , ICOS , NKG7 , and TCF7 differed in multiple animals ( Fig 2A–2D , S5 Fig ) . Over 35 genes were differentially expressed by infected cells in lymph node and jejunum from animal AY69 , and several of these genes were common to both tissues ( Fig 2E–2H ) . As with PBMC , these included several activation markers and interferon-stimulated genes . Among productively infected ( stages 3–4 ) cells , BAX , CD28 , CTLA4 , FLIP , ICOS , CXCL10 , and OASL were upregulated . BAX and ICOS were differentially expressed in all three tissues form this animal . Genes encoding host proteins essential for viral infection and replication , the co-receptor CCR5 and vRNA nuclear export factor XPO1 , were also upregulated in lymph node infected cells . The largest magnitude differences between uninfected and productively infected cells were observed in the jejunum , with >2-fold increases in BAX , CTLA4 , ICOS , IFIT3 , IL2RG , IL6R , LAT , OAS2 , OASL , PRKACB , TNF , and USP18 . The considerable heterogeneity in host gene expression across animals , tissue type , and cell infection status indicates that viral expression occurs in a wide range of distinct subsets of CD4 T cells–making selective targeting of infected cells , necessary for cure modalities , a much greater hurdle . Combining flow cytometric single-cell immunophenotyping and RNA quantitation for each cell allows us to define post-transcriptional gene regulatory events within single cells . In in vitro models , HIV/SIV downregulate expression of several surface proteins on infected cells via putative post-translational mechanisms [21–29] , but the degree to which this occurs in vivo is largely unknown . By comparing surface CD4 protein levels on uninfected and stage 3–4 infected CD4 T cells , we confirmed CD4 protein downmodulation on in vivo infected cells , but only on a subset of cells in jejunum and lymph node ( Fig 3A ) . Indeed , the majority of tat/rev+ cells in PBMC ( >95% ) and lymph node ( >70% ) retained surface CD4 at levels comparable to uninfected cells . Expression was not reduced in stage 1 ( spliced vRNA- ) infected cells ( S6 Fig ) . Because SIV Nef also downregulates CD3 [22] , sorting from the three additional PBMC specimens included CD3-negative cells ( S2 Fig ) and surface CD4 was indeed diminished on 40–55% of stage 3–4 cells and downregulation correlated with decreased CD3 ( Fig 3B , S6 Fig ) . Overall , the decrease in surface CD4 on downmodulated cells was ~90% , indicating residual surface CD4 despite active SIV transcription . Remarkably , the nine stage 2 ( tat/rev+ only ) cells observed only in jejunum expressed significantly more surface CD4 ( and CD3 ) than uninfected cells , supporting the classification of this population as a unique subset of infected cells distinct from stages 3–4 . Taken together , CD4 downmodulation in vivo is heterogeneous , varying across anatomical sites and among infected cells within a specimen , and indicates that this process may not be critical to viral pathogenesis . To explore the mechanism of CD4 downmodulation within single cells infected in vivo , we quantified tat/rev and CD4 transcript levels . At the one-cell level , surface CD4 protein was inversely associated with tat/rev RNA ( Fig 3C and 3D ) . CD4 protein downmodulation was not due to decreased CD4 mRNA ( Fig 3D and 3E , S6 Fig ) ; in fact , we observed an association with higher CD4 mRNA levels . Moreover , we observed a progression of CD4 ( and CD3 ) protein downmodulation , and increased CD4 gene expression , in concert with viral transcription . These findings demonstrate CD4 regulation by SIV in vivo via either post-transcriptional or post-translational mechanisms . Of note , despite similar tat/rev expression ( >1000 copies/cell ) observed in AY69 PBMC , lymph node , and jejunum , surface CD4 was unchanged on tat/rev+ PBMC . Thus while high levels of tat/rev correlate with lower expression of CD4 protein , robust tat/rev expression is not sufficient for decreased expression , and therefore may require tissue-specific factors . Post-translational downregulation of MHC class I HLA-A/B/C proteins from the surface of in vitro HIV/SIV-infected lymphocytes suggests that this may be a mechanism for evading cytotoxic T cell recognition in vivo [30–32] . Surprisingly , we found no consistent evidence of MHC class I downregulation on productively infected tat/rev+ cells , with the vast majority positive for HLA-A/B/C surface staining , even in cells that substantially downregulated CD4 protein ( Fig 3F and 3G , S6 Fig ) . Decreased MHC class I surface expression was observed in tat/rev+ cells in two animals ( 08D108 , p = 0 . 006 and 8–116 , p = 0 . 04 ) , although the amount of protein expressed per cell remained within the range observed for uninfected cells . Sequence analysis of the nef coding region in the SIVsmE660 inoculum did not reveal evidence of mutations known to impair MHC downmodulatory activity ( S6 Fig ) [23 , 30] . Taken together , we find that MHC class I downregulation is limited in vivo and thus this proposed mechanism of immune evasion may not operate in pathogenic SIV infection . To define phenotypic traits that distinguish infected cells , we measured expression of activation and differentiation surface markers . Productively infected tat/rev+ cells were nearly exclusively CD28+ central memory ( CM ) in lymph node and jejunum ( Fig 4A , S4 Table ) . Among PBMC , 3–74% of tat/rev+ cells were effector memory ( EM ) , suggesting preferential infection of CM , EM or no bias across hosts . The activation state of tat/rev+ cells was also diverse . In AY69 , tat/rev+ jejunal cells largely expressed CD69 with variable CD38 , while tat/rev+ PBMCs were exclusively CD69– , and lymph node was mixed ( Fig 4B ) . tat/rev+ PBMCs from the other animals were also remarkably heterogeneous . Surprisingly , the presence of CD69–CD38– tat/rev+ cells in multiple animals and tissues suggests that cellular activation may not be required for abundant viral gene expression in vivo . In addition , the activation profiles of uninfected T cells varied markedly among animals , likely reflecting variable degrees of host immune activation during acute infection . These findings are summarized in S4 Table . These differential surface protein expression profiles were further quantified by comparing the staining distribution of uninfected , tat/rev− ( stage 1 ) , and tat/rev+ ( stage 3–4 ) infected cells . CD95 and ICOS showed higher expression on both tat/rev+ and tat/rev− infected cells compared to uninfected cells ( Fig 4C ) . Surface expression of CD38 , CD69 , and HLA-DR was also elevated on tat/rev+ cells in one or two specimens , while there was either no significant difference or diminished expression on tat/rev+ cells for other specimens . Of note , within animal AY69 , specific activation markers were upregulated dependent on the tissue: CD69 in lymph node versus CD38 in jejunum . Overall , the broad distribution of these activation markers on the surface of tat/rev+ cells indicates highly variable expression among virus-producing cells . In most cases , differentiation and activation marker expression did not differ between productive tat/rev+ and non-productive tat/rev− infected cells . Exceptions included elevated CD95 , ICOS , and CD38 ( jejunum ) or CD95 and CD69 ( lymph node ) by tat/rev+ cells . Conversely , the stage 1 infected cells , which presumably represent cells at the time of or shortly after infection , exhibited significantly higher surface CD4 and CD3 relative to uninfected ( and tat/rev+ ) cells in some specimens ( S6 Fig ) . The distinct protein phenotype of stage 1 cells relative to uninfected cells with respect to CD4 , CD3 , CD95 , and ICOS ( among others shown in Fig 4C ) provides independent validation of the gag and LTR assays to identify a discrete subset of infected cells .
We integrated single-cell transcriptomic and flow cytometric technologies and successfully quantified and characterized rare in vivo SIV-infected CD4 T cells , including demonstration of post-transcriptional gene regulation at the one-cell level . Within single cells , different combinations of viral RNA molecules were co-expressed in distinct patterns and quantities , consistent with a continuum of virus replication stages . Productively infected tat/rev+ cells commonly , but not universally , expressed elevated levels of host cell genes and proteins associated with T cell activation . Protein profiling revealed broad similarities to uninfected cells and , surprisingly , remarkable heterogeneity among tat/rev+ cells with respect to CD4 downregulation , memory phenotype , and activation state , along with a lack of consistent evidence of MHC class I downregulation , even within CD4dim cells . Ample viral gene expression in memory cells devoid of multiple T cell activation surface markers indicates that an activated state is not required for productive CCR5-tropic virus infection , corroborating several previous studies primarily investigating CXCR4-tropic virus [33–36] . Our finding that MHC class I downregulation was limited on the surface of productively in vivo infected CD4 T cells is at odds with several prior studies demonstrating the importance of this SIV Nef function in vivo [30 , 37] . In these studies , strong selective pressure resulted in either reversion of a Nef point mutation or compensatory mutations elsewhere in Nef that restored MHC-downregulating activity . However , it is also possible that other unknown Nef activities were affected by these mutations and contributed to the selective pressure . One limitation of our approach was the use of a pan-MHC class I antibody . Maintenance of nonclassical MHC class I molecules ( e . g . Mamu-E ) on the cell surface may mask downregulation of classical class I molecules and thus corroboration with additional antibody specificities is warranted . Nonetheless , our data calls into question the extent to which MHC class I is downregulated in vivo and the relevance of the hypothesis that the virus modulates the expression of MHC as a mechanism of immune escape . Paradoxically , several ISGs with known antiviral properties were upregulated in productively infected cells , suggesting ineffective inhibition of SIV replication within CD4 T cells by these ISGs . Increased ISG expression may reflect a more activated state that predisposes cells to express viral genes . Alternatively , productive infection may directly or indirectly trigger ISG expression following autocrine or paracrine interferon signaling , respectively . It should be emphasized that gene expression differences between infected and uninfected cells may indicate pre-infection profiles , post-infection virus-induced modulation , or a combination thereof . The phenotype and host/viral transcriptome of infected cells varied considerably across the different tissues analyzed . First , a rare subset of tat/rev+ infected cells lacking other vRNA molecules ( stage 2 ) was present only in the jejunum , and not in PBMC or LN ( Fig 1 ) . This may reflect greater basal T cell activation in jejunum fostering low-level viral transcription without progression to a more productive state . Indeed , CD69 expression was greatest in jejunum , both among vRNA− and vRNA+ cells ( Fig 4B ) . Second , the largest fold-changes in host gene expression by infected cells were observed in the jejunum; specifically among cells in the highly productive stage 4 ( Fig 2G ) . Third , some differentially expressed genes were upregulated among infected cells in one tissue , while downregulated in a different tissue ( e . g . IL6R , LEF1 , NKG7 , and TCF7 ) , even within the same animal . Fourth , CD4 downmodulation , while observed in all three tissues , was greater in jejunum than LN and PBMC . These inter-tissue differences do not appear to be associated with viral transcription levels as infected cells in each tissue expressed comparable quantities of the vRNAs measured . Rather , other factors inherent to the anatomical site , such as cellular activation , metabolism , cytokine milieu , and interferon-responsiveness may contribute to the divergent , tissue-specific properties of in vivo infected cells [38] . Given that the lymph node and jejunum analyses were derived from a single animal , these observations will need to be confirmed in additional animals in future studies . Some caveats to our findings should be considered . First , the SIV gene expression assays do not distinguish between viral DNA and RNA . As a result , the gag and LTR assays , neither of which relies on RNA splicing , measure the presence of RNA , DNA , or both . Therefore positivity for these viral genes in the absence of spliced RNA ( stage 1 ) is unlikely to reflect viral transcription and we refer these cells as most likely residing in early , latent , or abortive infection states . We cannot distinguish between these different infection stages using the current technology . For example , virion-derived genomic RNA , nascent cytoplasmic reverse transcribed proviral DNA , and silent integrated proviral DNA ( although inefficiently detected in our system ) would all be similarly characterized as gag and/or LTR positive . This distinction also has important implications for measuring infected cell frequencies . By capturing both genomic viral RNA and DNA , our approach may report higher values than previous studies reliant on DNA alone , particularly during acute viremia when virion RNA is pervasive . The reduction from 60% to 36% infection among memory CD4 T cells in one lymph node sample when vRNA was excluded is consistent with a substantial vRNA contribution . Second , the viral life cycle staging we assigned to single infected cells should be considered a theoretical framework rather than a definitive consecutive series . Productive infection ( stages 3 and 4 ) , for example , may be followed by transition to a quiescent state in which low levels of unspliced RNA but little to no spliced RNAs are expressed ( stage 1 ) ; a status that has been described in studies of treatment suppressed HIV-1 infection [18 , 39] . However , given the context of untreated acute infection from which all of our single-cell analyses were derived , we believe the ordered staging likely applies to most cells . Viral cytopathic effects and cell-mediated immunity are expected to rapidly eliminate most stage 3–4 infected memory CD4 T cells . Third , the number of host factors and cell types investigated in our single-cell approach was limited . Flow cytometric indexed single-cell sorting is currently limited to ~15 parameters , while the Fluidigm Biomark qPCR gene expression platform measures 96 user-determined genes . Furthermore , in the present study , for pragmatic reasons ( to enrich for vRNA+ cells ) , we examined memory CD4 T cells at peak viremia time points . The extent to which our findings apply to naïve T cells , non-lymphoid cells , and post-acute infection settings requires further investigation . Given the remarkable phenotypic and transcriptional diversity of in vivo infected cells observed here , therapeutic strategies that aim to target infected cells must address an eclectic mix of CD4 T cells . Future studies employing single-cell RNA-Seq technology and 35-40-parameter flow cytometry sorting will further increase the power of this approach , as will the incorporation of DNA assays to specifically identify latent infection . These data illustrate the increased information content from integrating these single-cell technologies for revealing immunobiology of viral infections , including , for the first time , the ability to quantify post-transcriptional regulation at the single-cell level . This technology will have broad applicability in defining regulatory normal and aberrant mechanisms in cell biology and pathogenic infections .
Rhesus macaque PBMC were stimulated with PHA-P ( Sigma L9017 ) for three days followed by spinoculation with SIVmac239 virus generated by plasmid transfection of 293T cells ( ATCC CRL-3216 ) . Infected cell cultures were maintained in recombinant human IL-2 ( 1 U/μl; R&D Systems 202-IL ) and indinavir ( 2 μm ) to limit virus replication to a single round . Lamivudine ( 3TC; 10μM ) was used to prevent productive infection . Cells were lysed for RNA extraction at serial time points ( RNaqueous kit , Ambion ) or stained for intracellular Gag expression ( BD Fix/Perm kit , Coulter KC57RD1 anti-p24 ) . cDNA was synthesized from RNA using the DyNAmo cDNA Synthesis Kit ( Thermo Scientific ) as per the manufacturer’s instructions . Reverse transcription was primed with random hexamers . The following reagents were obtained through the NIH AIDS Research and Reference Reagent Program , Division of AIDS , NIAID , NIH: Indinavir Sulfate , Lamivudine . Viral genomic sequences from thirteen SIVmac251 and SIVsmE660 isolates were used to identify conserved sequences within individual SIV tat/rev , env , and LTR for maximal cross-reactivity . Probes for spliced genes span exon-exon junctions to minimize amplification of unspliced cDNA . Primer/probe sequences were optimized on bulk RNA extracts from SIVmac239 and SIVsmE660 in vitro infected cells . qPCR was performed using Platinum Taq polymerase as per manufacturer’s instructions ( Life Technologies ) . Unspliced viral genomic RNA was measured using SIV gag qPCR primer and probe sequences previously described [40] . The SIV LTR . U3 qPCR primer probe set consisted of the following: forward: equimolar amounts of 5’-TAC CCA GAA GAG TTT GGA AGC AAG TCA-3’ and 5’-TAC CCA GAA GAG TTT GGT AGT AAG TCA-3’; reverse: equimolar amounts of 5’-TTG TCA GCC ATG TTA AGA AGG CCT CTT G-3’ and 5’-TTG TCA GCC ATT TTA WWA AKG CCT CTT G-3’; probe: equimolar amounts of 5’-CTG TCA GAG GAA GAG GTT AGA AGA AGG CTA AC-3’ and 5’-TTG TCA GAG GAA GAG GTA AAG AGA AGG CTA AC-3’ . The SIV env assay consisted of the following: forward: 5’-AGA GGC CTC CGG TTG CA-3’; reverse: equimolar amounts of 5’-CTT ACT TGT TTG ATG CAG AAG ATG-3’ and 5’-CTT ACT TGT TTG ATG CAG RAR RTG-3’; probe: 5’-TTA GYC TTA GYC TTY TTC GGA GTT CTT CTT-3’ . The SIV tat/rev assay: forward: equimolar amounts of 5’-GAA CTC CGA AAA AGG CTA AGG CTA ATA CA-3’ and 5’-GAA CTC CGA ARA AGR CTA AGR CTA ATM CA-3’; reverse: equimolar amounts of 5’-CCK TCT CCT TCT TCT CCT TCT TTG GTT-3’ and 5’-CCG TCT CRT TCT TTG CCT TCT CTG GTT-3’; probe: equimolar amounts of 5’-CTGCATCAAACAACCCATATCCAACAGGACC-3’ and 5’-CTG CAT CAA ACA A ATC CCT ATC CAC AAG GRC C-3’ . Fourteen colony-bred Indian-origin male and female rhesus macaques ( age 3 to 9 years old ) were infected with various SIV strains administered as follows: intravenous SIVmac251 ( 100 MID50; VRC animal protocols 150 . 1 , 356 , 417 ) , intrarectal SIVmac251 ( AID50; VRC 211 . 3 ) , and intrarectal SIVsmE660 ( AID30; VRC 332 ) ( S1 Table ) . Virus preparations of 1 . 0 ml were inoculated via the saphenous vein or a lubricated feeding catheter inserted into the rectum . Infection duration ranged from 4 days to 8 months . Viable mononuclear cell suspensions were stained with fluorochrome-conjugated monoclonal antibodies from BD Biosicences ( San Jose , CA , unless otherwise indicated ) to CD4 ( clone OKT4 , BioLegend #317434 ) , CD3 ( SP34-2 , #557757 ) , CD8 ( RPA-T8 , in-house Qdot655 conjugate ) , CD28 ( CD28 . 2 , #555730 ) , CD95 ( DX2 , #558814 ) . Dead cells and monocytes were excluded by: LIVE/DEAD Aqua ( ThermoFisher ) and CD14 ( M5E2 in-house Ax700PE ) . Memory ( CD95+ ) CD8- CD3+ T-cells were sorted on a BD FACSAria ( BD Biosciences ) as either serial limiting 3-fold dilutions ranging from 3–1000 cells per well for frequency calculations or as single cells ( n = ~1850 ) for Biomark transcriptomic analysis in 96-well plates ( Fig 1A , top and bottom , respectively ) . The total number of cells sorted for limiting dilution analysis is shown in S1 Table for each specimen . For example , a typical limiting dilution sort plate consisted of 6 replicates each of 500 and 150 cells/well and 12 replicates each of 50 , 15 , and 5 cells/well ( n = 4740 cells analyzed ) . For animals 08D108 , 08D227 , and 8–116 , CD3- cells were included to account for CD3 downmodulation by SIV Nef [22] while CD16 ( 3G8 , BioLegend #302048 ) and CD20 ( 2H7 , in-house Ax700PE ) were used to exclude NK and B cells . Additional markers recorded for indexed phenotyping of single cells mapped by well position were HLA-DR ( L243 , #339194 ) , CD69 ( FN50 , in-house Ax594 ) , ICOS ( C398 . 4A , BioLegend #313505 ) , CD38 ( OKT10 , NHP Reagent Resource , PE ) , CCR7 ( 150503 , in-house Ax680 ) , and MHC class I ( W6/32 , BioLegend #311430 ) [41] . All clones were previously determined to cross-react with rhesus macaques ( NHP Reagent Resource ) and lots were individually titrated to identify optimal concentration per test . Cells were deposited into 96-well PCR plates ( GeneMate , Bioexpress ) for immediate lysis ( within ~100 μsec from protein detection ) and RNA extraction . 3D8 cells containing a single copy of integrated SIV DNA [42] were FACS sorted in control experiments and as standard curves for absolute DNA quantitation . To minimize changes in cell RNA and protein expression prior to analysis , cell samples were maintained on ice at all times , with the exception of the surface stain ( 15 min ) and elapsed sorting time ( ~30 min ) . All samples were previously cryopreserved , which was determined not to impact viral gene expression ( S1 Table ) . 96-well PCR collection plates were maintained on pre-chilled aluminum blocks during and after the sort . FACS data was analyzed using FlowJo v9 . 8 . Cells unlikely to be CD4 T-cells were excluded from downstream FACS analyses if they met the following three criteria: CD4 fluorescence intensity below cut-off ( <175 for AY69 LN; <500 for 08D227 , 08D108 , 8–116 ) , undetected CD4 mRNA , and undetected CD40LG mRNA expression . Undetected CD3E mRNA was used as an additional exclusion criteria for animals 08D108 , 08D227 , and 8–116 . tat/rev+ cells with downregulated CD4 ( “dim” ) were defined by fluorescence staining intensity as follows: <1100 in lymph node and PBMC; <3000 in jejunum . RNA from sorted cells collected in 10 μl of SuperScript III-Platinum Taq One-step qRT-PCR mastermix ( Life Technologies ) was directly reverse transcribed and PCR pre-amplified , as previously described [11] , with the following thermocycling conditions: 50°C for 15 minutes , 95°C for 2 minutes , followed by 18 cycles of 95°C for 15 seconds and 60°C for 4 minutes . Gene-specific primers were used for priming both the RT and PCR pre-amplification reactions . cDNA was diluted 5-fold and subjected to either conventional qPCR on an ABI 7900 real-time PCR instrument for 40 cycles or multiplexed qPCR on a Fluidigm Biomark HD system . The frequency of cells expressing a given viral transcript was calculated by plotting the fraction of replicate wells positive by RT-qPCR at each limiting dilution versus the number of cells sorted per well , followed by Poisson distribution analysis of the cell dilution corresponding to one positive cell per well , based on the expected frequency of 63 . 2% of wells positive at that dilution ( Fig 1A ) . All qPCR was performed in dedicated plasmid-free workspaces . Multiple PCR experiments were negative for all viral genes , including animal AY69 pre-infection PBMC and day 4 post-infection AZ26 and ZC55 PBMC ( S1 Table ) , indicating minimal SIV RNA or DNA PCR contamination . Quantitative gene expression in single cells was measured using the Fluidigm Biomark microfluidic chip platform ( Fig 1A ) . TaqMan assays ( Life Technologies ) consisted primarily of FAM-MGB probes that span exon-exon junctions ( S2 Table ) and passed qualification tests to establish both efficient and linear amplification as well as multiplexing capability [11] . Assays not specific for exon-exon junctions or otherwise capable of detecting genomic DNA ( suffix “s1” and “g1” ) were considered unlikely to influence our gene expression results because: 1 ) genomic DNA is not readily accessed by our cell lysis protocol ( S3 Fig ) , and 2 ) all cells are expected to contain the same number of genomic DNA copies . Samples and assays were loaded on the 96 . 96 Biomark Dynamic Array Chip for Gene Expression following manufacturer’s instructions . Briefly , 3 . 6 ul of 1:5 diluted cDNA was mixed with 4 . 4 ul of a 1:10 mixture of Fluidigm Sample Loading Regent and Taq Universal PCR Master Mix to create the real-time reaction sample mix . Equal volume of 20X TaqMan assay and Fluidigm GE Assay Loading Reagent prior were combined to generate the 10X assay mix . 5 μl of each mix was loaded on the chip inlets . Biomark qPCR was performed using the GE 96 . 96 Standard V . 1 protocol with 40 cycles of PCR and analyzed using the auto initialized Ct thresholds for each detector . Relative qPCR values are reported as expression threshold ( Et , where Et = 40-Ct ) . Absolute RNA copies were calculated as 2 ( Et-13 ) , given Et = 13 corresponds to a single copy of RNA using this protocol [11] . Positivity for either gag or LTR within single cells likely reflects variable qPCR assay detection at ≤2 copies . Single cell gene expression data was analyzed using the hurdle model framework implemented in the MAST package [43] . Data was filtered for cells with low cellular detection rate ( CDR ) ( < 12 . 5% of genes expressed , 9 cells ) [43] . Additional cells not flagged by this method that expressed a frequency of genes similar to that of excluded events were also omitted from analysis to eliminate all potential outlier cells ( S7 Fig ) . For samples sorted to include CD3- cells , there was substantial variation that could not be explained by potential confounders such as animal or chip after filtering . A subset of cells from stage 0 ( viral RNA negative ) from different animals formed a distinct cluster in tSNE space ( S7B Fig , red , 55 cells ) that was not associated with cellular detection rate . Because we are interested in differences between cellular infection states within each animal , we assessed gene expression differences between this outlier cluster and cells not in the cluster ( S7C Fig ) . Monocyte genes , NLRP3 and NOD2 , were enriched in this cluster , while T cell genes , CD3 , CD28 , and CD40LG , were relatively absent , indicating that these cells are most likely not T cells . This provided biological justification for excluding this cell cluster from the downstream analysis . The number of cells in each infection state from each animal is shown in S3 Table . Differential gene expression analysis was performed as previously described [20 , 43] . Gene expression in infected cell subsets , defined by the number and type ( unspliced or spliced ) of viral genes expressed ( Fig 1E and 1F ) , was modeled using the hurdle generalized linear model implemented in MAST and differences between infected cell subsets in each animal and tissue were tested using the hurdle likelihood ratio test . The MAST hurdle model tests for a difference in proportion ( percent cells positive for a gene ) and a difference in conditional mean ( Et ) and provides a combined likelihood ratio test that includes both sources of information . Multiple testing adjustment was applied using the Benjamini Hochberg FDR method across all tests and significance was called at a false discovery rate threshold of 10% . Differentially expressed genes were visualized by plotting point estimates of group effects from the continuous and discrete parts of the model with simultaneous 90% bivariate confidence ellipses ( Chi-square , 2 d . f . ) around the estimates . These generally agree with the likelihood ratio test , but can be less conservative for small sample sizes in some instances . The estimated average expression level for cells from each subject and group was calculated from the hurdle model ( controlling for CDR ) fit to the single cell expression data , as described [43–45] . Briefly , the estimate combines the discrete and continuous estimates , with standard errors derived via the delta method , and can be interpreted as the average expression that would be expected to be observed in a bulk cell experiment . Significant differences in gene or surface protein ( mean fluorescence ) expression between two cell populations were determined by the Student’s t-test ( p<0 . 05 ) . For protein comparisons involving three or more populations , differences were first assessed by ANOVA ( p<0 . 05 ) followed by Student’s t-test . All single cell protein and gene expression experiments were performed on six biological replicates with multiple cells measured per animal . Sample size was determined by available resources , i . e . number of SIV-infected animals and viable cells from each specimen . Given previous experience analyzing Biomark single-cell data , the available number of cells was deemed sufficient . All animals were cared for in accordance with guidelines set by the NIH Guide for the Care and Use of Animals . The NIH Vaccine Research Center IACUC approved all animal protocols and procedures . Animal protocols included ASPs 150 , 356 , 417 , 211 , and 332 . All data for the single-cell gene expression analysis are available at https://zenodo . org/record/803385 . | HIV-1 , and its simian counterpart , SIV , infect and kill CD4 T cells , resulting in their massive depletion that ultimately leads to AIDS in the absence of antiretroviral therapy . With effective therapy , these cells are largely preserved , but a subset harbors latent virus that can persist for decades and reemerge upon therapy interruption , preventing HIV-1 cure . To prevent or eliminate productive cellular infection , there is tremendous demand to identify host factors expressed by these cells in vivo , which may serve as unique biomarkers or drug targets . Here we provide the first detailed combined transcriptomic and protein expression profile of SIV-infected cells directly ex vivo using novel single-cell technologies . Our survey of activation markers , interferon-stimulated genes , and viral restriction factors identified multiple host genes differentially expressed by SIV-infected cells and will inform future therapeutic strategies . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"blood",
"cells",
"t",
"helper",
"cells",
"mhc",
"class",
"i",
"genes",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"immunology",
"microbiology",
"retroviruses",
"viruses",
"immunodeficiency",
... | 2017 | Combined single-cell quantitation of host and SIV genes and proteins ex vivo reveals host-pathogen interactions in individual cells |
Innate immune responses in general , and type I interferons ( T1IFNs ) in particular , play an important and often essential role during primary viral infections , by directly combatting the virus and by maximizing the primary adaptive immune response . Several studies have suggested that T1IFNs also contribute very substantially to the secondary ( recall ) response; they are thought ( i ) to be required to drive the early attrition of memory T cells , ( ii ) to support the subsequent expansion of surviving virus-specific memory cells , and ( iii ) to assist in the suppression and clearance of the infectious agent . However , many of these observations were predicated upon models in which T1IFN signaling was interrupted prior to a primary immune response , raising the possibility that the resulting memory cells might be intrinsically abnormal . We have directly addressed this by using an inducible-Cre model system in which the host remains genetically-intact during the primary response to infection , and in which T1IFN signaling can be effectively ablated prior to secondary viral challenge . We report that , in stark contrast to primary infection , T1IFN signaling is not required during the recall response . IFNαβR-deficient memory CD8+ and CD4+ memory T cells undergo attrition and expansion with kinetics that are indistinguishable from those of receptor-sufficient cells . Moreover , even in the absence of functional T1IFN signaling , the host’s immune capacity to rapidly suppress , and then to eradicate , a secondary infection remains intact . Thus , this study shows that T1IFN signaling is dispensable during the recall response to a virus infection . Moreover , two broader implications may be drawn . First , a T cell’s requirement for a cytokine is highly dependent on the cell’s maturation / differentiation status . Consequently , second , these data underscore the importance of evaluating a gene’s impact by modulating its expression or function in a temporally-controllable manner .
The recognition of pathogen associated molecular patterns ( PAMPs ) , including single stranded endosomal RNA , by pattern recognition receptors such as TLR7 within virally infected cells and/or specialized secreting cells ( e . g . plasmacytoid dendritic cells ) induces the production of pro-inflammatory cytokines , including the type I interferons ( T1IFNs ) , IFNα and IFNβ [reviewed , 1] . Signaling through their common receptor IFNαβR to activate the Jak/Stat intracellular signaling cascade [reviewed , 2] , type I IFNs occupy an often critical lynchpin of productive anti-viral responses: ( i ) inducing a local anti-viral state through the upregulation of interferon stimulatory genes ( ISGs ) capable of limiting the replication of a broad range of viruses within infected and neighboring cells [3] and ( ii ) activating and , in general , promoting both innate and adaptive arms of the immune system [reviewed , 4 , 5] . The inability of IFNαβR-deficient mice to control a myriad of viral infections , including lymphocytic choriomeningitis virus ( LCMV ) [6–8] , highlights the importance of an intact type I IFN signaling system in vivo . T1IFNs directly promote dendritic cell maturation , antigen processing , and costimulation expression [9–12] , leading to greater and more potent in vitro [10 , 13] and in vivo [12 , 14] T cell expansion and effector function . In addition , during certain viral infections T1IFN can provide a potent “signal 3 , ” aiding in the activation of CD8+ T cells , even in the absence of CD4+ T cell help [15] . Further , T1IFN also directly regulates T lymphocyte expansion and differentiation during viral infection , as antigen-specific CD8+ T cells lacking T1IFN receptor fail to appropriately expand in vivo during certain primary viral infections , including LCMV [16–19] , in part , due to their inability to upregulate NK cell regulatory molecules ( e . g . MHC Class I , Qa-1 , NCR1 ) and escape NK cell dependent lysis [20 , 21] , thereby greatly limiting ( up to 100-fold ) the number of memory T cells generated . Notably , during other primary viral or intracellular bacterial infections , including vaccinia , VSV , or Listeria monocytogenes [18 , 22] , the requirement for functional IFNα/β signaling upon CD8+ T cells is not nearly as severe , although T1IFN still , broadly speaking , promotes CD8+ T cell expansion . Whether memory CD8+ T cell function in vivo is similarly predicated upon T1IFN signaling is not clear . Resting IFNαβR-deficient memory CD8+ T cells are capable of degranulation and cytokine production following in vitro restimulation [19 , 22] . Secondary expansion and in vivo effector function of IFNαβR-deficient memory CD8+ T cells has been shown to be curtailed during secondary infection with recombinant Vaccinia or LCMV [20 , 22] , whereas an additional study by Oexnius and colleagues detailed no similar deficits in secondary expansion during recombinant Vaccinia virus infection [19] . T1IFN can also prove to be detrimental towards the host or the immune system in general . For example , IFNα/β potentiates leukopenia [23 , 24] , and , more specifically , is a mediator of CD8+ T cell attrition [25 , 26] during viral infection . Pre-exposure to IFN limits CD8+ T cell expansion and proliferation [27] , and prolonged and aberrant IFN expression has been implicated in the dysregulated anti-viral responses observed during chronic , persistent viral infection [28 , 29] . Finally , T1IFN is also key to the development of immunopathology following influenza and SARS-CoV infection [30 , 31] and during models of viral-induced hemorrhage [32 , 33] . Since T1IFN exerts disparate and pleiotropic influences on both the immune system and most somatic cells , conventional methods including germline and/or conditional knockout models that would limit requisite IFN signaling during primary infection are inappropriate means to assess whether memory responses to virus are similarly dependent upon IFNα/β-signaling , since it would be impossible to separate whether an observed deficiency in secondary immune function was the consequence of an inherent and preexisting deficit or a true reflection for concurrent IFN signaling during secondary viral infection . Here we have utilized an inducible-Cre system to temporally delete IFNαβR from both total immune and parenchymal cells prior to secondary viral infection . Importantly , prior to gene deletion , maturation of naïve T cells , and their responses during and following primary viral infection , are allowed to proceed unhindered . Thus we are able to accurately , and without bias , simultaneously evaluate whether T1IFN influences immune responses and/or viral control during secondary viral infection . We demonstrate that immediately following secondary viral infection both type I and II IFNs are transiently produced , manifesting both type I and II IFN inducible gene expression within the spleen . Surprisingly , inducible , tamoxifen-mediated deletion of T1IFN receptor signaling during secondary viral infection did not affect the attrition , expansion , quality , or reestablishment of virus-specific memory CD8+ T cells . Furthermore , viral control was not abrogated in either the spleen or the liver , demonstrating that a functional T1IFN system is not required for either secondary immune responses or containment of viral replication .
We evaluated the production , and biological activities , of interferons in the hours immediately following secondary viral infection . Long-term LCMV immune mice ( >6 weeks post primary LCMV-Arm ) were rechallenged with LCMV , and plasma was collected prior to , and at the indicated times following , the secondary infection; IFNα and IFNγ levels were assessed via ELISA or multiplex assay . Within 12 hours of secondary viral infection , both IFNα and IFNγ became readily detectable within the plasma of infected mice . Thereafter , the expression of both of these cytokines waned dramatically , and within 48–60 hours of infection IFNγ and IFNα were undetectable within the plasma ( Fig 1A and 1B ) . We were unable to detect any IFNβ throughout the course of secondary viral infection . We next determined whether these transient pulses of type I & II IFNs triggered the expression of interferon-inducible genes . Spleens were taken from sham infected LCMV-immune mice , and from LCMV-infected mice at 12 hours p . i . , the peak of peripheral IFNα and IFNγ expression . RNA was extracted and subjected to PCR array analysis . Both type I ( e . g . MX1 , IFIT1-3 , Oas1a , & IRF7 ) and type II ( e . g . CXCL10 & IRF1 ) interferon-inducible genes were significantly upregulated within 12 hours of secondary LCMV infection ( Fig 1C ) . Therefore , during secondary viral infection , type I and II interferons are rapidly but transiently induced , and upregulate the expression of their target genes . T1IFN signaling is required to control many primary viral infections in vivo [6 , 7] , and the absence of IFNαβR from virus-specific T cells compromises their in vivo expansion; this is true for both naive and memory T cells during a number of viral infections [16 , 17 , 20 , 21] . However , the reported defects in memory cell expansion could have resulted from their having developed ( i ) from receptor-deficient naïve / primary cells and ( ii ) in an environment lacking T1IFN signaling . To better determine the impact of T1IFN signaling upon recall responses , we developed a model in which IFNαβR ablation could be induced in vivo at any time selected by the investigator . Using this approach , the mice are genetically intact prior to , and during , primary infection , ensuring that the resulting memory T cells are developmentally normal; then , IFNαβR ablation can be induced prior to secondary challenge . To this end , UBC-Cre-ERT2 mice , expressing a tamoxifen sensitive variant of cre recombinase under the control of the human ubiquitin C promoter , were bred with IFNαβRfl/fl mice , generating UBC-Cre-ERT2+IFNαβRfl/fl and control littermate UBC-Cre-ERT2-IFNαβRfl/fl mice . We considered it important to first test the validity of the model , and did so by evaluating the outcome of primary virus infection in naïve mice that had been treated with tamoxifen . In Cre+ recipients , tamoxifen exposure should ablate IFNαβR expression , and these mice should recapitulate the published findings with conventional germline IFNαβR knockout mice [6 , 7 , 16 , 17 , 20 , 21] , which are unable to control primary viral infection , and in which the expansion of virus-specific T cells is severely compromised . Therefore ( Fig 2A ) , naïve UBC-Cre-ERT2+IFNαβRfl/fl mice , and their Cre- littermates , were treated with tamoxifen and , several weeks later , were challenged with LCMV-Arm . Seven days p . i . , LCMV-specific T cells were enumerated by standard intracellular cytokine staining . UBC-Cre-ERT2+ mice were found to have markedly-diminished CD8+ T cell responses compared to their tamoxifen-treated Cre- counterparts . Representative responses to the viral GP33 epitope in individual Cre- and Cre+ mice are shown , demonstrating a marked reduction in the proportion of CD8+ T cells that have responded to this epitope ( Fig 2B ) , and cumulative data for this , and two other , LCMV CD8+ T cell epitopes revealed >10-fold lower numbers of cells in Cre+ animals ( Fig 2C ) . Similar data were observed for GP61-80 specific CD4+ T cells ( Fig 2D and 2E ) . Finally , these reductions in T cell responses were accompanied by greatly elevated ( ~100 fold ) levels of viral RNA in the spleens of tamoxifen-pretreated Cre+ mice , when compared to drug-treated Cre- littermates ( Fig 2F ) . The above data indicate that tamoxifen-driven IFNαβR deletion is sufficiently effective to functionally recapitulate the impact of a complete genetic knockout during primary viral infection . Thus , we began our evaluation of drug-induced IFNαβR deletion in long-term immune mice . As indicated in Fig 3A , genetically intact Cre+ and Cre- IFNαβRf/f mice were infected with LCMV , and were allowed to develop a normal T cell memory pool; then , these long-term LCMV-immune animals were injected with tamoxifen . Several weeks later we assessed the overall in vivo efficiency of tamoxifen-induced deletion of IFNαβR , by various criteria . First , genomic DNA was isolated from purified splenocytes , lymph nodes , liver , heart , kidney , lung , and brain . The status of the IFNαβRf/f locus was assessed using PCR primers that flank the loxP sites and , in all of the tissues analyzed , near-complete ( brain ) or complete ( other tissues ) deletion of IFNαβR was observed ( Fig 3B ) . Next we determined the efficacy of tamoxifen-mediated Cre activation in T lymphocytes . Cre+ and Cre- IFNαβRf/fzsGreen+/wt mice , generated as described in materials and methods , were infected with LCMV and allowed to develop into long-term immune mice . Tamoxifen was administered and , two weeks later , CD4+ and CD8+ T cells were evaluated for zsGreen expression . In Cre- mice , as expected , CD8+ memory T cells specific for each of the three indicated epitopes were identified , and the cells did not express zsGreen ( Fig 3C , left column ) . In contrast , in Cre+ animals ( right column ) >90% of CD8+ T cells expressed zsGreen , demonstrating that Cre activity is widespread in that cell population ( the percentage of zsGreen+ cells in each population is shown as a red numeral ) . Moreover , the ability of tamoxifen to drive Cre enzymatic activity on cellular DNA was not affected by the immunological status of the T cell . The proportion of reporter-expressing cells among all CD8+ T cells in the long-term immune Cre+ mice ( “no peptide” , which will include CD8+ T cells that are not LCMV-specific ) was ~95 . 3% [94 . 8/ ( 4 . 65 + 94 . 8 ) ] , and a very similar proportion of epitope-specific CD8+ T cells were reporter-positive: e . g . , 93 . 7% of GP33-responsive ( IFNγ+ ) cells [6 . 19/ ( 0 . 417 + 6 . 19 ) ] expressed zsGreen . CD4+ T cells presented a similar picture ( Fig 3D ) ; in Cre+ mice , the vast majority of cells expressed zsGreen , in both total cells ( 96 . 9% ) , and in cells specific for the LCMV GP61-80 MHC class II epitope ( 94 . 8% ) . Finally , and most importantly , we determined whether Cre activity , and deletion at the level of genomic DNA , resulted in the functional ablation of T1IFN signaling . Splenocytes were isolated , and stimulated in vitro with rIFNβ . As shown in Fig 3E , almost all CD8+ T cells from tamoxifen-treated Cre- mice responded to IFNβ stimulation , as judged by increased levels of phosphorylated Stat1 . In contrast , zsGreen-expressing CD8+ T cells from Cre+ mice failed to upregulate phospho-Stat1 following IFNβ exposure . In summary , tamoxifen treatment of UBC-Cre-ERT2+ mice results in the widespread deletion of IFNαβR , both within peripheral organs and LCMV-specific memory T cells . Moreover , the Cre-mediated removal of IFNαβR quickly renders the memory T cells incapable of responding to T1IFN . In the absence of T1IFN signaling , CD8+ T cells fail to sufficiently upregulate NK inhibitory receptors , such as MHC Class I and Qa-1 , during primary viral infections , and therefore become subject to NK cell mediated lysis , limiting their in vivo expansion [20 , 21] . In addition , T1IFN has been shown to promote the attrition of memory CD8+ T cells , in particular , during either viral infection or poly IC injection [26 , 34] . To determine whether the deletion of IFNαβR from normal memory T cells prior to secondary viral infection would limit T cell attrition and/or abrogate MHC class I or Qa-1 expression , LCMV-immune UBC-Cre-ERT2+ and UBC-Cre-ERT2- IFNαβRfl/fl mice were treated with tamoxifen , then rechallenged with LCMV-Arm , as outlined in Fig 4A , or given a sham infection . Splenocytes were harvested at the indicated hours p . i . , and LCMV specific CD8+ and CD4+ T cells were identified and enumerated using standard intracellular cytokine staining ( Fig 4B and 4C ) or Class I tetramers ( Fig 4D–4G ) . As we have previously observed [35] , immediately following viral infection the absolute number of LCMV specific CD8+ and CD4+ memory T cells within the spleen rapidly declines; after 24 hours , virus specific T cells were reduced approximately 4- to 10-fold in Cre- mice ( Fig 4B and 4C , white bars ) . Contrary to the predictions based on published studies using conventional KO mice , cells lacking the IFNαβR ( Cre+/black bars ) contracted to very similar extents , indicating that memory T cell attrition does not depend upon T1IFN signaling . We next assessed whether the absence of IFNαβR limited MHC Class I and/or Qa-1 expression upon virus specific CD8+ T cells during the first 24 hours of secondary viral infection . Resting IFNαβR sufficient ( Cre- ) and IFNαβR deficient ( Cre+ ) DbGP33-41+ or DbNP396-404+ specific memory CD8+ T cells expressed limited Qa-1 ( Fig 4D and 4E , 0 time point ) . Immediately following viral infection , at 12 and 24 hours p . i . , Qa-1 was significantly upregulated upon both LCMV specific memory CD8+ T cell populations , and , additionally , Qa-1 expression was not reliant upon IFNαβR signaling; both Cre- ( white bars ) and Cre+ ( black bars ) CD8+ T cells expressed similar levels of Qa-1 throughout the first 24 hours of infection ( Fig 4D & 4E ) . Similarly , MHC Class I ( H2-Kb ) expression was also comparably increased upon both Cre+ and Cre- DbGP33-41+ and DbNP396-404+ specific memory CD8+ T cells in the hours following viral infection ( Fig 4F & 4G ) . Therefore , deletion of IFNαβR from previously-normal virus specific memory T cells does not spare them from attrition during secondary viral infection , and the cells remain capable of upregulating Qa-1 and MHC Class I . Previous reports , using conventional IFNαβR knockout mice or adoptively-transferred IFNαβR-deficient CD8+ T cells , have documented severe deficits in the ability of IFNαβR deficient memory CD8+ T cells to undergo secondary expansion [20 , 22] , and this was attributed to their failure to escape NK cell mediated attenuation [20] . Since we observed no deficit in MHC Class I or Qa-1 expression on memory CD8+ T cells following secondary viral infection in our inducible model , we next evaluated the secondary expansion of these IFNαβR-deficient cells , determined their capacity to establish secondary memory T cells . LCMV immune Cre+ and Cre- IFNαβRf/f were generated , treated with tamoxifen , and exposed to secondary LCMV infection as in Fig 4A , and five timepoints thereafter , immunodominant virus-specific CD8+ ( Fig 5A–5C ) and CD4+ ( Fig 5D ) T cells were enumerated . At each time point , mice were sacrificed ( 4–11 per group per time point ) , and virus-specific cells were enumerated . No difference in attrition ( day 2 p . i . ) , secondary expansion ( days 5 & 10 p . i . ) or the establishment of long-term 2° memory ( days 14 & 23 p . i . ) was observed , for either CD8+ or CD4+ memory T cells ( Fig 5 ) . Although we observed no deficit in the total numbers of virus-specific T cells across the course of the secondary response , we reasoned that the quality of the virus specific T cells might nevertheless have been impacted in the absence of T1IFN signaling . We therefore assessed the proportion of LCMV-specific CD8+ and CD4+ T cells capable of producing IFNγ , TNF , and/or IL-2 following in vitro peptide stimulation . We have previously shown that virus specific memory T cells become predominantly monofunctional ( capable of producing only one cytokine ) soon after secondary viral infection , and their multifunctionality ( i . e . , their ability to rapidly produce 2–3 cytokines ) is restored over the following 2–3 weeks [35 , 36] . As expected , these findings are recapitulated , for both CD8+ and CD4+ T cells , in the genetically-intact mice in this study ( Cre- mice in Fig 5E and 5F ) . The increased monofunctionality at days 2 & 5 p . i . is demonstrated by the increasing size of the black segment , and the concomitant shrinkage of the grey and white areas; and the opposite occurs at days 14 and 23 p . i . , indicating the gradual reappearance of multifunctionality . More importantly , these qualitative changes were almost identical , both in kinetics and in extent , in the absence of functional IFNαβR ( Cre+ mice in Fig 5E and 5F ) . Thus , the overall kinetics and relative quality of memory T cell responses to secondary viral reactivation appear to be independent of T1IFN signaling . The above data show conclusively that , over the entire course of the recall response , the numbers and quality of virus-specific T cells present in Cre+ and Cre- mice are almost identical . However , it is formally possible that this outcome could have resulted from the selective expansion , in Cre+ mice , of the small proportion of virus-specific memory T cells ( ~5% , see Fig 3C and 3D ) that had escaped tamoxifen-driven Cre activity . In this scenario , these putative IFNαβR-expressing cells in Cre+ mice would , over time , come to predominate the pool of the virus-specific T cells . To assess this , we again exploited Cre reporter ( zsGreen ) mice . LCMV-immune Cre+ IFNαβRf/f mice were generated , treated with tamoxifen , and subjected to secondary LCMV infection ( as described in Fig 3A ) and , at the indicated time points p . i . , mice were sacrificed and zsGreen-positive and -negative virus-specific CD8+ and CD4+ T cells were enumerated using intracellular cytokine staining ( Fig 6 ) . Representative contour plots from day 23 p . i . data for CD8+ and for CD4+ T cells are shown ( Fig 6A and 6B respectively ) . For all epitopes , >90% of the responding ( i . e . , IFNγ+ ) cells were zsGreen+ , suggesting that zsGreen-negative cells ( which should presumably express functional IFNαβR ) did not have a marked selective advantage over the course of the recall response . This conclusion is strengthened by cumulative data , from multiple mice at various time points throughout secondary infection ( C-F ) ; all of the responding CD8+ and CD4+ populations maintained a high , and stable , proportion of zsGreen-expressing cells . Finally , we further considered the possibility that , in individual T cells , Cre-reporter activation may not always be accompanied by the deletion of IFNαβR . In that case , zsGreen expression would not invariably reflect the IFNαβR status of the T cell , which would complicate the interpretation of the data in panels A-F . However , if this were the case , and if IFNαβR provided a selective advantage , one would expect to see increasing numbers of zsGreen-positive T cells that remained functionally receptive to T1IFNs . We evaluated this over the course of the recall response ( Fig 6G ) , and saw no significant increase in the IFNβ-responsiveness of zsGreen+ T cells . Together , the data in Fig 5 and Fig 6 provide strong support for our assertion that T1IFN signaling plays little or no role in supporting the attrition , secondary expansion , or stable secondary memory formation of memory T cells . Lastly , we determined the ability of tamoxifen-treated Cre+ mice to control the secondary LCMV challenge . Published data suggest that their ability to do so may be compromised , for at least two reasons . First , T1IFNs are a key component of the innate response , and act by triggering the expression of a large number of interferon-stimulated genes ( ISGs ) , many of which are antiviral . However , in our model , tamoxifen-induced Cre activation of LCMV-immune mice leads to widespread ablation of IFNαβR ( Fig 3 ) , which should limit the capacity of these innate cytokines to stimulate ISG expression . Second , the IFNαβR-deficient memory cells–although exhibiting normal kinetics of attrition , expansion , and secondary memory formation ( Fig 5 ) –may themselves be dysfunctional: previous work has indicated that adoptively-transferred memory CD8+ T cells derived from naïve IFNαβR-deficient TCR transgenic cells were unable to control secondary LCMV infection [22] . We first compared the extent of ISG expression at day 1 following secondary infection of tamoxifen-treated LCMV-immune Cre+ and Cre- IFNαβRf/f mice ( Fig 7A ) . Type I dependent ISGs known to be capable of directly limiting viral replication [3 , 37] were significantly elevated in the spleens of Cre- mice , relative to Cre+ mice , with very high levels of statistical significance; type II interferon associated ISG expression–with the exception of Ifi35 –were similar in Cre+ and Cre- infected mice . To determine the impact of IFNαβR ablation on viral control , we used qPCR to quantitate LCMV viral RNA in the spleens ( Fig 7B ) and livers ( Fig 7C ) of LCMV infected Cre+ and Cre- mice at the indicated time points following secondary challenge . Regardless of IFNαβR status , LCMV vRNA was cleared from both organs at near-equivalent rates . Strikingly , no statistically-significant differences were observed even at early ( 0 . 5 and 1 day p . i . ) times , when T1IFNs would be expected to exert the greatest anti-viral effect . Moreover , complete viral clearance appeared to be achieved in the Cre+ mice; no viral recrudescence was observed as late as >3 weeks after secondary challenge . Thus , antigen-specific CD8+ and CD4+ T cells can mount protective antiviral recall responses that are independent of IFNα/β signaling; this is consistent with their similar kinetics of expansion ( Fig 5 ) and their ability to escape NK cell mediated lysis through the upregulation of MHC class I and Qa-1 ( Fig 4 ) .
T1IFN is a well-established mediator of anti-viral immunity during primary infection , directly limiting viral replication by upregulating ISGs in many somatic cells , and activating and regulating the innate and adaptive arms of the immune system [1–5] . The role of these cytokines during the recall response is far less well-characterized , but extant studies have proposed that T1IFNs also influence the responses of memory T cells [reviewed , 38] . However , those studies relied on memory cells that had been generated during a primary infection of T1IFN-deficient mice , or from receptor-deficient precursor T cells , and one can conceive of at least two ways in which these techniques might have affected the memory cells’ development: first , directly–the cells will not receive direct T1IFN signals; and , second , indirectly–virus and antigen load will be abnormally high , and prolonged , during primary infection in T1IFN-deficient animals , potentially modulating the maturation of virus-specific cells . Herein , we have employed an inducible deletion model in which both of these potential confounders are nullified . The primary viral infection , and the corresponding immune response , are entirely normal; only after memory cells are established ( >6 weeks following the primary infection ) is IFNαβR inducibly deleted . During secondary LCMV infection , IFNα and IFNγ are transiently produced during the first 24 hours of infection ( Fig 1 ) . We [35 , 39] and others [40] have previously reported the transient burst of IFNγ by CD8+ T cells in response to homologous and heterologous secondary viral infection; the focus of the present study was on T1IFN . Relative to primary LCMV infection [28] , peak levels of IFNα during secondary LCMV infection are substantially reduced but remain sufficient to drive the significant upregulation of a number of T1IFN dependent ISGs in the spleen at 12 hours p . i . ( Fig 1C ) ; these findings validated our investigation of the impact of these cytokines on memory T cells . We developed a tamoxifen-sensitive inducible system to delete IFNαβR from all cells and demonstrated that , during primary viral infection ( Fig 2 ) , T cells in mice that had been pre-treated with tamoxifen failed to appropriately expand , and the mice were unable to control the viral infection . These data largely recapitulated published studies using either conventional IFNαβR knockout mice , or recipients of transferred IFNαβR knockout CD8+ T cells [6 , 7 , 16 , 17 , 20 , 21] , providing additional validation for our model . The profound biological impact of tamoxifen administration also indicated not only that in vivo Cre-mediated deletion of floxed IFNαβR DNA must be efficient , but also that it must have led to the rapid and widespread loss of any existing functional IFNαβR . These observations were confirmed by our analyses of tamoxifen-treated LCMV-immune mice ( Fig 3 ) , in which IFNαβR deletion was near-complete in genomic DNA extracted from all tested tissues , and in which Cre activity was demonstrable in the vast majority of CD8+ and CD4+ T cells as shown by the use of a zsGreen reporter line . Most importantly , we confirmed that the genetic disruption of IFNαβR was accompanied by the rapid cessation of receptor function; Cre reporter+ CD8+ T cells were unable to phosphorylate Stat1 in response to in vitro IFNβ stimulation ( Fig 3E ) . Several studies have shown that T1IFNs can cause a loss of antigen-specific and nonspecific memory ( and to a lesser extent naïve ) CD8+ T cells during viral infection in vivo , in a process termed attrition [25 , 26 , 34] . We have previously confirmed that attrition occurs [35] , and Fig 4B shows that a large proportion of LCMV-specific memory CD8+ T cells are , indeed , lost from the spleen within the first 24 hours of secondary LCMV infection . However , and to our surprise , the ablation of IFNαβR from normal memory T cells prior to the secondary viral infection did not prevent this attrition , indicating that an alternative pathway ( s ) must be contribute to the process; we speculate that this may be driven by CD8+ T cell derived IFNγ , which is produced in the hours immediately following secondary viral infection ( Fig 1 ) [35] . In addition to being involved in attrition , direct IFNα/β signaling leads to the specific upregulation of MHC Class I and Qa-1 on CD8+ T cells during in vitro stimulation and/or primary viral infection in vivo , and the failure to appropriately upregulate these and other inhibitory molecules allows the cells to be attacked by NK cells , limiting the expansion of the virus-specific CD8+ T cells [21] . Qa-1 , the mouse analog of human HLA-E , acts as an external readout for intracellular MHC class I processing [41] , binding to the inhibitory receptor NKG2a [42] and suppressing NK cell activation [43] . LCMV-specific IFNαβR-deficient memory CD8+ T cells upregulated both MHC class I and Qa-1 in vivo ( Fig 4D–4G ) , suggesting that NK-mediated killing may not contribute to the cells’ rapid attrition; moreover , this upregulation occurred in both Cre- and Cre+ mice , indicating that the expression of MHC Class I and Qa-1 by memory CD8+ T cells also can occur independently of IFNα/β signaling . In summary , multiple studies have shown that both T cell attrition and the upregulation of MHC class I & Qa-1 can be triggered by T1IFN , but our data indicate that there is functional redundancy in memory T cells which allows these changes to take place even when IFNαβR signaling is abrogated . We infer , from the existence of this redundancy , that these biological events–attrition , and upregulation of protective molecules–may play a key role in ensuring a balanced and appropriate memory T cell response . More detailed analyses of the kinetics of the secondary T cell responses ( Fig 5 ) confirmed that attrition ( day 2 ) was unaffected by ablation of IFNαβR , and also showed that the subsequent expansion of the T cells , their ability to produce multiple cytokines , and their entry into secondary memory , were both entirely normal in IFNαβR-ablated animals . This is in stark contrast to the primary immune response , in which T1IFN signaling plays a key supporting role in T cell expansion , and provides a powerful selective pressure favoring effector CD8+ T cells that bear the receptor [16 , 17 , 20 , 21] . We considered the possibility that such pressures may have promoted the selective expansion of the few memory cells that had evaded tamoxifen-induced Cre activation prior to secondary LCMV infection; theoretically , this could have provided an alternative explanation for why IFNαβR-sufficient ( Cre- ) and IFNαβR-ablated ( Cre+ ) animals had near-identical numbers of virus-specific cells over the course of the recall response . However , the proportion of zsGreen-positive cells remained stable over time ( Fig 6 ) , furthering reinforcing our conclusion that all phases of the CD8+ and CD4+ virus specific recall response are independent of T1IFN signaling . These findings are discordant with published data using conventional knockout mice , in which memory cells were reliant on T1IFN signaling . We propose that those memory cells , which had developed in an environment devoid of T1IFN signaling , were intrinsically abnormal; indeed , it is possible that even their naïve precursors were , in some unknown way , defective . Finally , we evaluated the importance of T1IFN signaling during secondary virus infection . Compared to naïve mice , LCMV-immune mice rapidly control viral replication , and significant differences in viral RNA are detectable within as few as 6 hours of infection [35] . T1IFNs are rapidly expressed during the recall response ( Fig 1 ) [28] , so it was possible that these cytokines might have contributed to this antiviral effect . Conceivably , they could have done so in two ways , by ( i ) upregulating ISG expression in somatic cells and/or ( ii ) promoting the effector functions of virus-specific T cells , a suggestion supported by previous work showing that the antiviral effector function of IFNαβR-deficient memory CD8+ T cells is defective in vivo [22] . Therefore , we questioned whether the de novo ablation of IFNα/β signaling prior to secondary infection would inhibit the control of LCMV in the spleen and the liver . We found that the inducible deletion of IFNαβR had no significant impact on the containment and ultimate clearance of a secondary viral infection ( Fig 7B and 7C ) . Additionally , viral recrudescence was undetectable as late as 23 days p . i . Thus , we contend that–contrary to current thinking–T1IFN signaling is not required for the secondary response to viral infection , neither regulating the attrition , expansion , and secondary memory formation of memory CD8+ and CD4+ T cells , nor being required for establishing an antiviral state within host parenchymal cells . It is possible that , in the absence of T1IFN signaling , other inflammatory cytokines may contribute to supporting these key biological functions . Two additional points are worthy of consideration . First , it has been known for some time that memory T cells can very rapidly suppress viral replication . Viral RNAs are one of the major instigators of the T1IFN response , and we recently showed that–when compared to naïve mice–a reduction in RNA levels was detectable as early as 6 hours p . i . in LCMV-immune mice , and there was a >100-fold reduction in viral RNA by 12 hours after LCMV challenge [35] . Thus , the recall response very quickly prevents the accumulation of molecules that drive T1IFN production; in that light , it makes good evolutionary sense that the recall response should not rely on T1IFNs , and should be effective in its absence . Second , in recent years the concept of “innate memory” has come to the fore . Our findings suggest that T1IFNs are not required for the control or eradication of secondary viral challenge . How might these observations relate to human immune responses and viral diseases ? Genetic deficiencies in IFNγ signaling ( and related pathways ) are increasingly being identified; perhaps surprisingly , sufferers are not particularly susceptible to most viral infections , instead being vulnerable to infection by mycobacteria [44] . To date , fewer defects in T1IFN responses ( and related pathways ) have been reported in humans , and we speculate that such deficiencies may be extremely rare because they confer a dramatic evolutionary disadvantage , by rendering the victims open to severe primary viral infections . Nevertheless , there are some examples , in humans , of mutations that disrupt T1IFN responses . Homozygous mutation in IRF7 led to a near-fatal infection by influenza virus [45] , while mutations in TLR3 [46] and in several other T1IFN-related genes [47] markedly increase the risk of herpes simplex encephalitis ( HSE ) . This condition can be treated with acyclovir , and , although fatal in some individuals , the majority survive , albeit usually with neurological sequelae . However , and notably , it appears that the survivors are not at high risk for HSE recurrence; we speculate that the therapeutic intervention led to the rapid suppression of the primary infection , allowing these patients to develop memory T cell responses; thereafter , these patients are , in essence , phenotypically similar to our LCMV-immune , tamoxifen-treated Cre+ mice; despite being unable to respond to T1IFN signaling , they are protected against secondary exposure to the virus . In conclusion , we show here that T1IFNs are dispensable during a secondary viral infection . This finding contrasts with published studies , and we suggest that this discrepancy has broad implications . Many–indeed almost all–analyses of how individual genes contribute to memory T cell responses have relied on approaches in which the knockout is either present in the germline ( conventional KO ) , or is conditional ( e . g . , Cre is expressed from a cell-specific promoter ) , and , in almost all of those cases , the genetic defect is present prior to the primary infection . Using these approaches to study IFNαβR , defective expansion by memory T cells was observed; it was inferred , therefore , that the memory T cell response–like that of their primary counterparts–must benefit from T1IFN signaling . However , our data strongly suggest that the memory cells upon which those inferences depend are intrinsically abnormal , presumably because they arose and developed in the absence of T1IFN signaling , and , therefore , such memory cells cannot reliably be used to assess how individual genes affect the recall T cell response . Thus , we contend that it is vital that the evaluation of any gene’s contribution to secondary ( and subsequent ) immune responses should employ an approach that does not compromise the very response from which those memory cells arise .
All animal experiments were approved by The Scripps Research Institute ( TSRI ) Institutional Animal Care and Use Committee and were carried out in accordance with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals . C57BL/6J mice were purchased from the TSRI rodent breeding colony . Transgenic inducible Cre mice under the control of the human ubiquitin promoter ( B6-UBC-Cre-ERT2 , JAX 008085 ) [48] and IFNαβRfl/fl mice [49] were crossed together to generate UBC-Cre-ERT2+IFNαβRfl/fl ( referred to as Cre+IFNαβRfl/fl ) . To generate LCMV-immune mice , Cre+IFNαβRfl/fl and control littermate Cre-IFNαβRfl/fl mice were challenged with 2x105 PFU LCMV-Arm ( Armstrong strain ) intraperitoneally ( i . p . ) . >6 weeks following LCMV infection , Cre+ and Cre- mice were injected with 3 successive , daily i . p . doses of 2mg of tamoxifen dissolved in corn oil . At least two weeks were allowed to elapse after tamoxifen injection before the Cre+IFNαβRfl/fl and Cre-IFNαβRfl/fl mice were rechallenged with LCMV-Arm ( 2x106 PFU i . p . ) . For experiments using a Cre reporter , IFNαβRfl/fl mice were crossed and backcrossed with zsGreen Cre-reporter mice ( B6 . Cg-Gt ( ROSA ) 26Sortm6 ( CAG-ZsGreen1 ) Hze/J , JAX 007906 [50] ) to generate doubly-homozygous IFNαβRf/fzsGreen+/+ animals . These mice were crossed against Cre+IFNαβRfl/fl animals , providing F1 mice that were all IFNαβRf/fzsGreen+/wt , and were either Cre+ or Cre- . Two weeks following tamoxifen administration , flow cytometry was used to identify T cells in which Cre activity had occurred , indicated by zsGreen expression . Mice were genotyped with the following primers ( 5’-3’ ) : UBC-Cre-ERT2 ( F ) GCC AGC TAA ACA TGC TTC ATC GT & ( R ) CGC GGC AAC ACC ATT TTT TCT GA , IFNαβRfl/fl ( F ) AAG CTC CTT GCT GCT ATC TG & ( R ) CAC ACC AGG CTT CTA ATG TC , and Cre-Reporter ( WT F ) AAG GGA GCT GCA GTG GAG TA , ( WT R ) CCG AAA ATC TGT GGG AAG TC , ( Tg F ) AAC CAG AAG TGG CAC CTG AC , & ( Tg R ) GGC ATT AAA GCA GCG TAT CC . To assess UBC-Cre-ERT2 mediated deletion of IFNαβR , genomic DNA was extracted from various tissues using Qiagen DNeasy Blood & Tissue Kit , then used as a PCR template with the above IFNαβRfl/fl primer set . These primers flank both loxp sites and exon 10 of IFNαβR , thereby discriminating between WT , floxed , and deleted alleles . Analysis of IFNαβR deletion was routinely carried out for every mouse in this study . LCMV-immune ( >6 weeks p . i . ) C57BL/6 mice were rechallenged with 2x106 PFU LCMV-Arm ip , and blood was collected at the indicated times following secondary LCMV infection . Plasma was isolated using K3-EDTA coated microvette tubes ( Starstedt , Nümbrecht , DEU ) , aliquoted , and stored at -80°C . Interferon alpha levels within plasma were assessed using Verikine Mouse Interferon Alpha ELISA kit ( PBL Assay Science , Piscataway , NJ ) , and interferon gamma was determined by LEGENDplex Mouse Inflammation panel multiplex ( Biolegend , San Diego , CA ) according to manufacturer instructions . Interferon stimulatory gene expression within the spleens of wildtype C57BL/6 and Cre+ and Cre-IFNαβRfl/fl mice was quantified via Mouse Interferons & receptors PCR Array ( PAMM-064Z , SA Biosciences , Frederick , MD ) according to manufacturer instructions . Copies of LCMV vRNA within the spleens of Cre+ and Cre- were assessed using reverse transcriptase real time PCR ( qPCR ) as previously described [35 , 51] . Single cell suspensions of splenocytes were obtained from mechanically disrupted and red blood cell lysed spleens . Dead cells were identified and excluded using Zombie NIR Fixable Viability Kit ( Biolegend ) . After Fc-blocking with anti-CD16/32 ( BD Biosciences , San Diego , Ca ) , splenocytes were immunophenotyped with the following fluorescently conjugated antibodies to cell surface markers ( Biolegend ) CD8α ( 53–6 . 7 ) , CD4 ( RM4-5 ) , CD44 ( 1M7 ) , H2Kb ( AF6-88 . 5 ) , & Qa-1 ( 6A8 . 6F10 . 1A6 , Milteyi Biotec , Bergisch Gladbach , DEU ) . MHC-Class I tetramers specific for DbGP33-41 and DbNP396-404 were provided by NIH Tetramer Core Facility ( Emory University , Atlanta , GA ) . Appropriately conjugated isotype-control antibodies were used in all experiments . Samples were acquired on a BD Biosciences LSR-II and analyzed using FlowJo ( Treestar , Ashland , OR ) . As previously described [35] , 2x106 splenocytes were incubated for 6 hours directly ex vivo with GolgiPlug ( BD Biosciences ) and 1μM of the CD8+ epitopes GP33-41 , NP396-404 , or GP276-286 or 10µM of the CD4+ epitope GP61-80 . Stimulated splenocytes were subsequently washed , stained with viability dye , Fc-blocked , and surface stained with CD4 , CD8α , & CD44 as above . After surface staining , splenocytes were washed , fixed with Cytofix/Cytoperm ( BD Biosciences ) and resuspended in 1X PermWash ( BD Biosciences ) . Intracellular cytokines were identified with fluorescently conjugated antibodies to IFNγ ( XMG1 . 2 ) . Unstimulated ( no peptide ) cultured splenocytes were used to identify antigen-specific cytokine production . 4x106 splenocytes were stimulated for 25 minutes directly ex vivo with or without 104U/ml of recombinant mammalian IFNβ ( PBL Assay Science ) . After stimulation , splenocytes were washed in cold FACS Buffer ( 1x PBS pH 7 . 4 , 2% FBS , 0 . 1% Na Azide ) , fixed with Cytofix/Cytoperm ( BD Biosciences ) , Fc blocked , stained for CD8α as above , permeabilized with Perm Buffer III ( BD Biosciences ) , and phosphorylated Stat1 ( pY701 ) was identified using fluorescently conjugated antibodies ( 4a , BD Biosciences ) . The functional deletion of IFNαβR signaling was assessed for all mice presented in this study . Significant differences were assessed via one or two way ANOVA with Sidak correction where appropriate , and results were considered significant if P<0 . 05 ( Prism 7 , Graphpad , San Diego , CA ) . | Type I interferons ( T1IFNs ) are key pleiotropic anti-viral cytokines , necessary to effectively promote and coordinate immune responses and control a number of primary viral infections . In their absence , the expansion and later memory formation of anti-viral T cells is severely curtailed . T1IFNs are also thought to perform a similar role during a recall response , although whether this signaling axis is truly required is obscured by the fact that initial immune responses , and by extension any memory cells subsequently generated , in the absence of these cytokines are potentially defective . Here we have employed an inducible-deletion model , wherein normal responses to primary infections proceed unhindered , and only after the successful establishment of long-term memory T cells is the common receptor for T1IFNs , IFNαβR , deleted . We demonstrate that , unlike during primary infection , both the recall immune response and control of virus is independent of functional T1IFN signaling . Importantly , our results highlight a need to re-evaluate the critical and requisite determinants of a successful recall response in a temporally-controlled manner . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"blood",
"cells",
"innate",
"immune",
"system",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"immune",
"physiology",
"cytokines",
"viral",
"transmission",
"and",
"infection",
"spleen",
"immunology",
"microbiology",
"neuroscience",
"learning",
"and",
"me... | 2016 | Type I IFN Signaling Is Dispensable during Secondary Viral Infection |
Within an isogenic population , even in the same extracellular environment , individual cells can exhibit various phenotypic states . The exact role of stochastic gene-state switching regulating the transition among these phenotypic states in a single cell is not fully understood , especially in the presence of positive feedback . Recent high-precision single-cell measurements showed that , at least in bacteria , switching in gene states is slow relative to the typical rates of active transcription and translation . Hence using the lac operon as an archetype , in such a region of operon-state switching , we present a fluctuating-rate model for this classical gene regulation module , incorporating the more realistic operon-state switching mechanism that was recently elucidated . We found that the positive feedback mechanism induces bistability ( referred to as deterministic bistability ) , and that the parameter range for its occurrence is significantly broadened by stochastic operon-state switching . We further show that in the absence of positive feedback , operon-state switching must be extremely slow to trigger bistability by itself . However , in the presence of positive feedback , which stabilizes the induced state , the relatively slow operon-state switching kinetics within the physiological region are sufficient to stabilize the uninduced state , together generating a broadened parameter region of bistability ( referred to as stochastic bistability ) . We illustrate the opposite phenotype-transition rate dependence upon the operon-state switching rates in the two types of bistability , with the aid of a recently proposed rate formula for fluctuating-rate models . The rate formula also predicts a maximal transition rate in the intermediate region of operon-state switching , which is validated by numerical simulations in our model . Overall , our findings suggest a biological function of transcriptional “variations” among genetically identical cells , for the emergence of bistability and transition between phenotypic states .
Individual cells of a given genotype can exhibit various phenotypes . The phenotype of a cell usually refers to distinct characteristics ( static and dynamic , physical or chemical ) and the associative biological functions of the cell . Extending the central dogma of molecular biology , it is now accepted that the behavior of a single cell is determined by both the genomic polynucleic acid sequence and the dynamics of intracellular biochemical networks in space and time . The biochemical reactions inside cells serve as the immediate environment for the genome , where genotypic information resides . It is only through intracellular biochemistry that extracellular conditions can interact with genes . Based on this perspective , we propose the following: for a population of cells with identical genomes and extracellular conditions , each phenotype can be represented by a cluster of single-cell data defined as a peak ( e . g . , modal value ) in the multi-dimensional histogram of biomolecular copy-numbers measured at steady state . In general , the peak is a sizable region in the vast biochemical kinetic space , which is known as an attractor in chemical kinetics [1 , 2] . Multiple peaks naturally discretize the space; at a given instance in time , a single cell can reside in one of these discrete states . More interestingly , a homogeneous cell population responds to a varying environment through changes in the distribution among discrete phenotypic states , rather than through gradual adaptation to an intermediate state [3] . At a single-cell level , this observation is known as all-or-none [4] . Furthermore , it has recently been shown that a steady-state multi-modal distribution can be recovered after a subpopulation of cells under a peak is removed [5 , 6] , indicating that dynamic interconversion between phenotypic states occurs within a single cell . In a sufficiently long time , each single cell is considered ergodic among the different phenotypes . The coexistence of multiple phenotypic states diversifies clonal cells; and provides a non-genetic evolutionary advantage for survival in unpredictable environments [7–9] . Recent experiments have revealed that the dynamics of a single cell are essentially stochastic , as there is only a single copy of DNA inside a typical cell , which leads to stochastic mRNA and protein production [10 , 11] . Both transcriptional and translational events have been shown to occur in stochastic bursts [10 , 12–20] indicating that the gene state switches stochastically and is relatively slow compared with the typical rates of active transcription and translation . The stochastic operon-state switching of the lac operon has also been shown to be crucial for the change in a cell’s phenotype [3 , 13] , which highlights the importance of single-molecule events inside the cell . In terms of quantitative biochemical kinetics , the temporal evolution of the probability distribution of a well-mixed reaction system is governed by a Chemical Master Equation ( CME ) [21] , from which the corresponding stochastic trajectory of a single cell can be computationally simulated . We have recently shown , using a toy model of gene regulation , that when the rate of gene-state switching is low relative to the typical rates of active transcription and translation , the full CME can be reduced to a single-molecule fluctuating-rate model , in which the dynamics of mRNA and protein copy numbers at each given gene state follow deterministic dynamics while transcription rates fluctuate due to stochastic gene-state switching [22] , which is necessary for spontaneous phenotypic state transitions . In the full Chemical Master Equation , the copy-number fluctuations of mRNA and protein resulting from stochastic synthesis and degradation are present , which prevent us from studying the role of only the stochastic gene-state switching . However , in fluctuating-rate models , stochastic gene-state switching is the only source of randomness , the conclusions drawn from which are much more clean and unambiguous . On the other hand , although numerical simulations of full Chemical Master Equation can be practical , theoretical analysis is still difficult to implement; while solid theoretical foundations have already been proposed for fluctuating-rate models , which are also called piecewise deterministic Markov processes [22 , 23] . The fluctuating-rate model is easier to implement both theoretically and numerically . Therefore , it is a good candidate for studying single-cell dynamics , especially towards investigating the role of only the stochastic gene-state switching . So far , the exact role of stochastic gene-state switching that occurs during the transition between phenotypic states in a single cell is unclear , especially in the presence of positive feedback . In the present study , we address this problem using the lac operon as an archetype . Recent experiments have shown that the switching of operon states of the lac operon is slow compared with typical rates of active transcription and translation . Thus , in such a region of operon-state switching , we propose to explore the single-molecule fluctuating-rate model in quantitative detail , by incorporating the previous described operon-state switching mechanism [3] . This mathematical model illustrates the emergence of discrete phenotypic states from detailed nonlinear biochemical kinetics , and the robustness of such cellular states follows naturally . Although in general , positive feedback is necessary for bistability in a biochemical network , we show that the stochasticity in operon-state switching of an individual cell is able to not only trigger stochastic transitions between phenotypic states , but also significantly broaden the range of environmental parameters under which bistability occurs . The bistability that occurs in the absence of stochasticity is called deterministic bistability , while the bistability which occurs in the presence of stochasticity but beyond the parameter range of deterministic bistability is called stochastic bistability . We further show that stochastic operon-state switching must be extremely slow to trigger stochastic bistability ( bimodal distribution ) by itself in the absence of positive feedback . On the other hand , positive feedback is known to be able to maintain a stable state [24–26] , hence with the help of positive feedback , the induced state is stabilized beyond the range of deterministic bistability , even when the rates of stochastic operon-state switching is only within the physiological region . However , positive feedback is not able to stabilize the uninduced state within the same parameter region . We show that the uninduced state is instead stabilized by the relatively slow operon-state switching . Together , the mechanism of significantly broadened parameter range of bistability is explained . We further predict how the phenotype-transition rates vary with operon-state switching rates under each type of bistability . We also illustrate that the maximal transition rates between different phenotypic states are achieved with an intermediate rate of operon-state switching , which is a phenomenon that was predicted previously [27] and is explained using a recently proposed phenotype-transition rate formula . Finally but not the least , we not only explained the previously reported experimental discoveries in the present study , but also further refined some earlier conclusions that were not as precisely presented , such as the effect of DNA looping as well as the concept and quantification of thresholds of phenotype transitions .
Stochastic gene-state switching is a major source of stochasticity inside a single cell [10 , 13 , 28] , and even responsible for the phenotype transition [3] . By evoking the recently identified stochastic gene-state switching mechanism , we propose a single-molecule fluctuating-rate model for the lactose operon , introducing the fluctuating transcription rates into the deterministic dynamics described in previous studies [4 , 29 , 30] . The deterministic dynamics of all other chemical species under each operon state in the fluctuating-rate model consist of several differential equations representing the temporal evolution of mRNA , LacY polypeptides , and the intracellular inducer concentrations . The stochastic kinetics of the operon states are described in Fig 1B and 1C , and are modeled by a simple Markovian jumping process . The state O denotes the free operon; the state O*R denotes the operon with the repressor bound only at the auxiliary lac operator ( partial dissociation ) ; the state OR denotes the operon with the repressor bound at both the major and auxiliary lac operators; and the state O*RIm denotes the operon bound by both the repressor at the auxiliary lac operator and inducer molecules . While the inducer is unlikely to interact with the fully bound tetrameric repressor , it could conceivably bind to the inducer once a dimer head of the repressor dissociates . Traditionally , the O*RIm complex , which contains both the repressor and the inducers bound on the operon , is omitted in mathematical models . Such an over-simplified model cannot explain why the repressor binds stably to DNA in the absence of inducer , and is released rapidly in the presence of inducer . Previous models , assuming either O + R ⇋ OR or O + R ⇋ O*R ⇋ OR , imply that the rate of the complete dissociation of the repressor is independent of the intracellular inducer concentration . However , data show that when the intracellular inducer concentration is high , the frequency of complete dissociation can also be high ( 0 . 01 minute−1 ) ( Fig 2A in [3] ) . Alternatively , when the intracellular inducer concentration is low , the frequency of complete dissociation events is low and shows very weak concentration dependence ( Fig 3D in [3] ) . Therefore it appears that the repressor also binds the inducer when bound to the operon . In addition , the repressor has 2 different binding constants ( i . e . K and 1/K3 , see S1 Text for details ) for the inducer , depending on whether the repressor is already bound to DNA [31 , 32] , which are 10 to 100-fold apart . Accordingly , when inducer concentrations are below the lower binding constant , there is weak concentration dependence of the complete dissociation rate , whereas once the inducer concentration approaches the higher binding constant ( 100 μM ) , the complete dissociation rate increases dramatically via the OR → O*R → O*RIm → O pathway shown in Fig 1B and 1C . This is the basic type of lac operon induction with which most molecular biologists are familiar . However , the role of single-molecule fluctuations of DNA transcription under intermediate concentrations of inducer was unclear prior to the work of Choi , et al . [3] . The parameters of our model were obtained either directly from experimental measurements or through fitting the predictions of the model to experimental data , as explained in S1 Text . In addition to the mechanism of the emergence of bistability , we also sought to quantitatively investigate the transition rates between different phenotypic states and the more detailed molecular mechanisms that trigger them .
Only a single copy of a DNA molecule exists inside a typical cell . Hence , the stochastic dynamics of a single cell resulting from the fluctuating kinetics of single DNA molecule are a consequence of fundamental physical and chemical laws . Still , individual cells can control the stochastic kinetics of DNA molecules over a reasonable time scale and fluctuations due to biochemical reactions can be even advantageous . Recent high-precision measurements performed in single cells have revealed that stochastic gene-state switching is slow compared to typical rates of active transcription and translation . The fluctuating-rate model is a good candidate for the investigation of single-cell dynamics in this region because it only incorporates the gene-state switching mechanism . The present study reveals the power of this type of model , which will be used to investigate other questions regarding single-cell dynamics . We investigated the origin of the bimodal distribution of the lac operon in a realistic model incorporating the recently discovered mechanism of operon-state switching . It has been shown that either positive feedback or single-molecule fluctuations gives rise to bistability by its own . However , we show here that the interplay of these two mechanisms makes the bimodal distribution more realistic and reliable in the presence of environmental perturbations . Without positive feedback , the single-molecule kinetics of gene states are not sufficient slow , at least in E . coli , to induce bistability , and without fluctuations of single DNA molecules , positive feedback cannot stabilize the uninduced state when the extracellular inducer concentration is high . The physiological region for the gene-state switching rates is therefore favorable and balances the two contradictory purposes of controlling stochasticity within a certain magnitude and triggering phenotype transitions within a reasonable time scale . The stochastic model can quantify the relative stability ( fractions in a population ) of coexisting phenotypes , which cannot be achieved using a deterministic approach . However , the time scale of the stochastically triggered spontaneous phenotype transition is quite long , which prevents direct laboratory measurement of the relative stability ( given the time window of a typical experiment ) , due to inconsistencies between a quasi-steady state and the final steady state . Such inconsistencies also mean that the concept of the threshold is not well-defined in a stochastic scenario , which is considerably different from the deterministic threshold predicted from the deterministic mean-field model . Recently , Razooky , et al . also investigated the interplay between positive feedback and relatively slow gene-state switching kinetics in the transcriptional program controlling HIV’s fate decision between active replication and viral latency , and found out that the positive feedback shifts and expands the region of LTR bimodality [54] . However , the positive feedback in LTR dynamics lacks cooperativity and cannot produce deterministic bistability by itself , which is essentially different from the lac operon dynamics we studied here . Also the perspective we used to explain the broadened bistability is different from [54]: their explanation more focused on the mean-noise relation while ours more focus on the stability of each phenotypic states . In many experiments , people used minimal media for cells at 37°C , making the E . coli cells grow slowly ( doubling time is about one hour ) . In there experiments , a single copy or at most two copies are reasonable . However , in a more natural environment , E . coli grows at most commonly 20 − 30 minutes doubling time , which implies more copies of operons . Hence we also simulate the case with more than one copy of operons ( see Materials and methods ) , and find out that the qualitative results are exactly the same as the case with only a single copy of operon ( S13 Fig ) . On the other hand , in the main text , we model the lac operon under unnatural conditions , i . e . using the unnatural lactose-analogue TMG , which is used in most of the experiments . We also simulate the extended version of the model with lactose replacing TMG , in which an additional term representing the hydrolysis of allolactose is added ( see Materials and methods ) . Under steady-state condition of extracellular lactose , the results are quite the same as those from the main model in which we use TMG ( S15 Fig ) . Finally , the notion of cell diversification of genetically identical phenotypes in biological entities , due to stochastic gene expression , requires a mechanism for the inheritance of an “intercellular biochemical” state through cell division . This issue has been discussed previously [55 , 56] . Briefly , if the volume of a biochemical system doubles while maintaining the same internal concentrations , the phenotypic state of the cell is maintained . Therefore , the phenotype of a single cell can be preserved via growth and division into two daughter cells . This epigenetic inheritance mechanism is based on dynamic biochemical self-organization , which is fundamentally different from the Watson-Crick genetic template-copying mechanism .
Several extended versions of the fluctuating-rate model have also been investigated: ( 1 ) Without feedback: set α to be zero; ( 2 ) Without DNA loop: there is only two operon states O and OR; ( 3 ) In the case of multiple operons: independent n operons coupled only through the intracellular M , Y , I , and the corresponding cell division time is set to be 50/n minutes , which makes the parameters rI = 0 . 012nmin−1 and rY is equal to 0 . 1 + rI; ( 4 ) Lactose replacing TMG: a term - h y d · I I + K I · Y representing the hydrolysis of allolactose is added to the right-hand-side of d I d t . Simulated results from these extended versions are in the Supporting Information . We define the bimodal steady-state distribution as bistability in the presence of stochasticity . However , it is quite time-consuming to obtain the exact steady-state distribution in simulation , if it is bimodal . Luckily , if we only want to determine whether the steady-state distribution is bimodal or not , it is much easier . There is a fact that if the steady-state distribution is unimodal , the simulated distribution will rapidly converge to the steady-state distribution , while if the steady-state distribution is bimodal , the converging time is extremely long . Hence , we can use the quasi-steady-state distribution and hysteresis response curves to determine whether the system is bistable or not . We only need to simulate the system for a reasonably long time , which is enough for making the system converge into the unimodal steady-state distribution if it is not bistable , or into the bimodal quasi-steady-state distribution if it is bistable . Hysteresis response curve follows the same idea . After a reasonably long time , if the simulated distributions starting from induced state or uninduced state can not merge together , then it implies bistability . We used the standard exact method to simulate the dynamics of the operon developed by Doob , Bortz et al . , and Gillespie [58–61] . See the Supplementary Material in [22] for details . | Identifying the mechanism underlying the coexistence of multiple stable phenotypic states has been a challenging scientific problem for more than half a century , and an appropriate mathematical model at the single-cell level is also in high demand . Single-cell measurements conducted in the past ten years have shown that gene-state switching is slow relative to the typical rates of active transcription and translation; hence the recently proposed fluctuating-rate model is a good candidate for describing the single-cell dynamics . We use the classic gene regulation module of the lac operon as an archetype and build a specific fluctuating-rate model based on the recently identified operon-state switching mechanism . This model is analyzed to dissect the interplay between positive feedback and the stochastic switching of gene states in the emergence of bistability/multistablity and the transition between phenotypic states . We show that relatively slow operon-state switching stabilizes the uninduced state and that the positive feedback stabilizes the induced state . Thus , the parameter range for bistability is significantly broadened . In addition , recently proposed landscape theory and rate formula predict opposite phenotype-transition rate dependence on operon-state switching rates for the two types of bistability . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"chemical",
"compounds",
"disaccharides",
"gene",
"regulation",
"enzymology",
"operons",
"carbohydrates",
"organic",
"compounds",
"dna",
"transcription",
"probability",
"distribution",
"mathematics",
"dna",
"enzyme",
"kinetics",
"chemical",
"dissociation",
"lac",
"operon",
... | 2018 | Relatively slow stochastic gene-state switching in the presence of positive feedback significantly broadens the region of bimodality through stabilizing the uninduced phenotypic state |
The Sigatoka disease complex , caused by the closely-related Dothideomycete fungi Pseudocercospora musae ( yellow sigatoka ) , Pseudocercospora eumusae ( eumusae leaf spot ) , and Pseudocercospora fijiensis ( black sigatoka ) , is currently the most devastating disease on banana worldwide . The three species emerged on bananas from a recent common ancestor and show clear differences in virulence , with P . eumusae and P . fijiensis considered the most aggressive . In order to understand the genomic modifications associated with shifts in the species virulence spectra after speciation , and to identify their pathogenic core that can be exploited in disease management programs , we have sequenced and analyzed the genomes of P . eumusae and P . musae and compared them with the available genome sequence of P . fijiensis . Comparative analysis of genome architectures revealed significant differences in genome size , mainly due to different rates of LTR retrotransposon proliferation . Still , gene counts remained relatively equal and in the range of other Dothideomycetes . Phylogenetic reconstruction based on a set of 46 conserved single-copy genes strongly supported an earlier evolutionary radiation of P . fijiensis from P . musae and P . eumusae . However , pairwise analyses of gene content indicated that the more virulent P . eumusae and P . fijiensis share complementary patterns of expansions and contractions in core gene families related to metabolism and enzymatic degradation of plant cell walls , suggesting that the evolution of virulence in these two pathogens has , to some extent , been facilitated by convergent changes in metabolic pathways associated with nutrient acquisition and assimilation . In spite of their common ancestry and shared host-specificity , the three species retain fairly dissimilar repertoires of effector proteins , suggesting that they likely evolved different strategies for manipulating the host immune system . Finally , 234 gene families , including seven putative effectors , were exclusively present in the three Sigatoka species , and could thus be related to adaptation to the banana host .
Bananas and plantains ( Musa spp . ) are amongst the world's top five staple food crops , as approximately 100 million tons of bananas are produced annually in nearly 120 countries in tropical and subtropical regions [1] . However , bananas are prone to many diseases that can severely reduce production , and thus pose a threat to global food security . The problem is intensified by the very narrow genetic basis of currently cultivated banana varieties , as most are sterile triploid hybrids ( AAA , AAB , ABB ) between the wild species Musa acuminata ( A genome ) and Musa balbisiana ( B genome ) . This includes desert bananas ( AAA ) of the Cavendish-subgroup , cooking bananas ( AAA or ABB ) , and nearly all plantain landraces ( AAB ) [2] . Currently , the so-called “Sigatoka disease complex” is one of the most destructive diseases in banana worldwide , reducing yields by more than 50% [3 , 4] . The socio-economic impact of the disease is much higher in small farming communities in sub-Saharan Africa , Southeast Asia , and Latin America that depend almost exclusively on the banana crop for their survival . Therefore , managing this disease is of urgent importance and is currently under critical public review for humanitarian , biosafety , and environmental reasons [1 , 4] . Three phylogenetically closely related species of Pseudocercospora ( class Dothideomycetes , order Capnodiales , family Mycosphaerellaceae ) have been recognized as the primary constituents of the Sigatoka disease complex in banana , namely Pseudocercospora fijiensis ( Pf ) ( M . Morelet ) Deighton [sexual morph: Mycosphaerella fijiensis M . Morelet] , causal agent of black Sigatoka or black leaf streak disease , Pseudocercospora musae ( Pm ) ( Zimm . ) Deighton [sexual morph: Mycosphaerella musicola R . Leach ex J . L . Mulder] , causal agent of yellow Sigatoka disease , and Pseudocercospora eumusae ( Pe ) Crous & X . Mourichon ( sexual morph: Mycosphaerella eumusae Crous & X . Mourichon ) causal agent of eumusae leaf spot disease [3–5] . The host range of the three species is believed to be restricted to Musa spp . , although clear differences in virulence exist amongst them , with P . fijiensis considered as the most aggressive and P . musae the least damaging species [5–7] ( Fig 1 ) . Despite such differences in virulence , however , the three species share a common hemi-biotrophic lifestyle and disease-cycle on their host , often causing similar and easily confounded symptoms on infected leaves . More specifically , compatible interactions are characterized by a biotrophic latent phase of 3–4 weeks , depending on the specific species/isolate-host interaction , during which the pathogen colonizes the intercellular spaces before any necrotic symptoms appear on the infected leaves . On the other hand , incompatible interactions are expressed either in the form of partial resistance or bear the signatures of a hypersensitivity response ( HR ) , typical of gene-for-gene interactions [3 , 5 , 8] . In addition to a common phytopathogenic and infectious lifestyle , multilocus DNA analysis has also revealed that the three species form a monophyletic group , and thus are likely to have originated from a common ancestral species [9] . The common evolutionary history of these pathogens was also confirmed by characterization of their mating-type loci , which suggested a stepwise evolution from an heterothallic ancestor splitting first into P . fijiensis and subsequently into P . musae and P . eumusae [10] . Although not exclusively specified , the analysis also suggested that these events are likely to have taken place relatively recently in the evolutionary past of the three pathogens . Indeed , the disease chronology records suggest that all three pathogens emerged in Southeast Asia during the last century , with P . musae appearing first in the Indonesian island of Java in 1902 from where it rapidly expanded to all banana producing areas of the world , occasionally causing severe epidemics [3 , 4] . Nowadays , the pathogen has typically been displaced by the more aggressive P . fijiensis , which was first recorded in the Sigatoka district of Fiji in 1963 , and since then has become the dominant species in areas where the two pathogens co-exist [11] . Compared to P . musae , P . fijiensis is able to infect a wider range of cultivars , including ones with resistance to P . musae , and cause considerably more damage that can affect losses up to 76% , thus endangering food security . At present , P . fijiensis has spread to most parts of the world where bananas and plantains are grown , and continues to advance to new ecological niches [4 , 12 , 13] . The third species associated with the Sigatoka disease complex , P . eumusae , was first described in mid-1990s in Southern and Southeast Asia [6] and , although on the march , so far seems to be restricted to these parts of Asia and some parts of Africa . Notably , P . eumusae is able to infect banana and plantain cultivars that are resistant to both P . musae and P . fijiensis , causing yield losses of up to 40% [3 , 4 , 6] . Despite the fact that P . musae was the first pathogen to be described in the disease chronology records , in reality it is possible that the three species co-existed on banana until changes related to the genetics of the pathogens or/and exogenous factors , such as changes in cultural practices and environmental conditions , have prompted the observed alterations in their virulence spectra and the sudden flare-up and over-dominance of one species over the others . A recent study has described more than 20 Mycosphaerella species on banana , many of which can co-exist on the same leaf or even the same lesion with the three primary constituents of the Sigatoka disease complex [9] . Although most of these species are only mildly virulent on banana , it is possible that niche sharing by multiple closely related species on the same host could facilitate inter-species exchange of genetic material and result in new species with altered virulence patterns [9] . Understanding the evolutionary and genomic changes involved in the emergence of new pathogens and shifts in virulence spectra is critical . Such knowledge is beyond academic interest alone , as it is vital for deciphering the biological process of disease emergence and for designing new and effective disease control methods . In this study , we employed comparative and evolutionary genomics in order to understand the evolutionary trends and genomic modifications associated with shifts in virulence spectra among P . musae , P . eumusae , and P . fijiensis , the main constituents of the Sigatoka disease complex on banana , and to further identify their pathogenic core that can be exploited in disease management programs . Using next generation sequencing technologies , we have sequenced the genomes of P . eumusae and P . musae and compared them with the recently determined 74 . 1 Mb genome sequence of P . fijiensis [14] . Genome-wide molecular selection analysis was further used to estimate whether changes in virulence spectra are mainly facilitated by adaptive evolution of the core genome or through species-specific gene acquisitions and losses . Overall , our analysis identified a significant amount of species-specific adaptations , but also revealed convergent patterns of evolution in the two more aggressive pathogens , suggesting that the evolution of virulence traversed through key changes in specific molecular pathways . The results presented in this study enable a deeper understanding of the Sigatoka disease complex and the evolution of virulence in these pathogens and beyond .
Whole-genome shotgun sequencing of P . musae and P . eumusae on the Illumina Hiseq2500 platform generated a total of 22 . 6 and 38 . 9 millions of high quality pair-ended reads ( 150x150 bp ) for each species , respectively , that were first used for an assembly-independent estimation of their genomic characteristics , by k-mer analysis ( k = 17 ) ( Tables 1 and S1 , S1 Fig ) . Based on the total k-mer number and the volume peak , genome sizes were estimated to 82 . 8 Mb for P . musae and 53 . 8 Mb for P . eumusae , thus revealing that , as compared to P . fijiensis ( 74 . 1 Mb ) [14] , P . musae has the largest and P . eumusae the smallest genome size from the three species . Subsequent analysis of the single-copy and repeat regions , in which k-mer frequencies falling between the boundaries of the peak region were considered as single-copy regions , indicated that the differences in genome sizes are essentially due to differences in repeat content . Indeed , while 31 . 3 Mb ( 37 . 8% ) , 34 . 6 Mb ( 64 . 3% ) , and 36 . 4 Mb ( 49% ) of the genomes of P . musae , P . eumusae , and P . fijiensis , respectively , are classified as single-copy regions , in contrast , the amount of repetitive content and unassembled sequences ranges from 51 . 5 Mb ( 62 . 2% ) in P . musae , to 19 . 2 Mb ( 35 . 7% ) in P . eumusae , and 37 . 7 Mb ( 51 . 0% ) in P . fijiensis , thus showing that , as with other Dothideomycetes [15 , 16] , repeat content is highly variable and plays the largest role in determining genome sizes ( Tables 1 and S1 , Fig 2A ) . The high number of repetitive sequences also impeded assembly efforts , as de novo assembly of the NGS reads produced a highly scaffolded genome of 60 . 4 Mb with 3331 scaffolds for P . musae and 47 . 1 Mb with 2626 scaffolds for P . eumusae ( Tables 1 and S1 ) . Average contig length was 18 . 1 Kb and 17 . 9 Kb for P . musae and P . eumusae , respectively , while the N50 size of the genome scaffolds was 0 . 4 Mb for P . musae and 0 . 16 Mb for P . eumusae ( S2 Fig ) . Given the high sequencing depth of 100-150x that is considered sufficient to cover the breadth of protein-coding exons , the apparent discrepancy between the final assembly sizes and the genome sizes estimated by the k-mer distribution analysis , can be attributed to the high repeat content , which prevented the complete assembly of the repeat-rich regions and led to highly fragmented genome assemblies . In spite of the fragmented genome assemblies , analysis of syntenic relationships in scaffold alignments between pairs of the three species , revealed high levels of localized conservation of gene content , order and orientation in most identified syntenic blocks . In pairwise comparisons of scaffolds larger than 200 kb in size from P . musae and P . eumusae , such regions of co-linearity were occasionally extended along the length of entire scaffolds in the form of segmental and tandemly repeated blocks of synteny , indicating the presence of “broken” or “segmented” macrosynteny ( S3A Fig ) . For example , this was the case with scaffold number 2 ( 223 . 5 kb ) , 3 ( 225 . 7 kb ) , 6 ( 233 . 5 kb ) , and 9 ( 209 . 4 kb ) of P . musae that showed almost perfect but broken macrosynteny to scaffolds in P . eumusae . However , the signature of macrosynteny was eroded in other scaffold alignments between the two species , as synteny was restricted to interspersed and very short genomic segments , as in alignments of scaffolds number 5 , 7 and 11 from P . musae to those of P . eumusae . In a similar way , analysis of the pairwise syntenic relations between scaffolds in P . musae and P . eumusae larger than 200 kb in size , on one hand , and scaffolds in P . fijiensis , on the other , revealed an analogous pattern of broken macrosynteny , as stretches of interspersed co-linearity occasionally combined with intra-chromosomal inversions were frequently observed ( S3B and S3C Fig , S1 Text ) . Although difficult to infer with certainty , due to the highly fragmented genome assemblies , overall the scaffold alignments suggest a pattern of broken or segmented macrosynteny among the three primary agents of the Sigatoka disease complex , which could possibly be driven by the lineage-specific proliferation of repetitive elements in each species and other genomic rearrangements . This pattern is different from the mesosynteny that is usually observed in genome-wide synteny alignments between more distantly related species of Dothideomycetes [17] . To further investigate the impact of the fragmented assemblies on gene identification and to assess the completeness of the assemblies with regard to gene content , we used the CEGMA pipeline to match them against a set of 248 core eukaryotic gene ( CEG ) families that are highly conserved across nearly all eukaryotes [18 , 19] . For the analysis , the CEG families were classified into four groups ( Groups 1-to-4 ) based on the degree of protein sequence conservation across eukaryotes , ranging from low ( Group 1 ) , to high ( Group 4 ) ( S4 Fig ) . Overall , CEG completeness ratios were slightly higher for P . eumusae ( 96 . 2 , 95 . 5 , 95 . 1 , and 94 . 6% , respectively ) as compared to P . musae ( 96 . 2 , 90 . 1 , 87 . 7 , and 96 . 9% , respectively ) , and P . fijiensis ( 94 . 7 , 92 . 9 , 95 . 9 , and 97 . 7% , respectively ) for nearly all four groups . Nonetheless , all three species had completeness ratios comparable to those previously reported for other fungal genome sequencing projects [19] , and thus the produced genome assemblies for P . eumusae and P . musae should cover almost the entire gene space . The increase in genome size in species of Dothideomycetes has been frequently connected to an invasion of their genomes by TEs , consequently altering their genome structure and function and shaping their pathogenic life-styles [15 , 16 , 20] . Therefore , we classified and compared the diversity of TEs and other repeats present in P . musae , P . eumusae , and P . fijiensis ( S2 Table ) in order to understand their impact on genome organization and evolution of the three species . Overall , TEs comprise an estimated 73 . 6% ( 21 . 4/29 . 2 Mb ) , 65 . 8% ( 8 . 3/12 . 6 Mb ) , and 79 . 3% ( 29 . 9/37 . 73 Mb ) of the repetitive fractions in P . musae , P . eumusae , and P . fijiensis [14] , respectively , whilst the rest of the repeat sequences can be mainly attributed to satellites , simple repeat and low complexity sequences ( Pm: 1 . 6 Mb; Pe: 5 . 6 Mb; Pf: 0 . 43 Mb ) , unclassified repeats ( Pm: 6 . 1 Mb; Pe: 3 . 7 Mb; Pf: 7 . 4 Mb ) , and unassembled sequences ( Pm: 22 . 3 Mb; Pe: 6 . 7 Mb; Pf: 0 . 0 Mb ) ( Fig 1A ) . Class I TEs , in particular , account for the majority of the repetitive content in each genome , totaling 69 . 8% ( 20 . 4/29 . 2 Mb ) in P . musae , 61 . 1% ( 7 . 7/12 . 6 Mb ) in P . eumusae , and 62 . 3% ( 23 . 5/37 . 73 Mb ) in P . fijiensis ( Fig 2B , S2 Table ) . The high ratio of Class I elements in the genomes of the three species is in-between the ratio previously reported for other Dothideomycetes , such as Fulvia fulva ( syn . Cladosporium fulvum , syn . Passalora fulva ) ( 90 . 9% ) , Dothistroma septosporum ( 40 . 6% ) , Plenodomis lingam ( syn . Leptosphaeria maculans ) ( 83 . 3% ) , Zymoseptoria tritici ( syn . Mycosphaerella graminicola ) ( 54 . 4% ) , and others [15 , 16 , 21] . Within Class I TEs , LTR retrotransposons are the most numerous retroelements in all three genomes , but their fraction is much higher in P . fijiensis ( 21 . 5 Mb , 57 . 7% ) [14] as compared to P . musae ( 12 . 5 Mb , 42 . 8% ) and P . eumusae ( 5 . 2 Mb , 41 . 3% ) ( Fig 2B , S2 Table , S1 Text ) . In contrast to Class I TEs , Class II transposons are considerably less expanded in the genomes of P . musae and P . eumusae , occupying only a minor 3 . 8% ( 1 . 1/29 . 2 Mb ) and 4 . 7% ( 0 . 6/12 . 6 Mb ) of the repetitive fraction , respectively . In P . fijiensis , however , Class II elements are strikingly more abundant , tallying up to 17 . 2% ( 6 . 4/37 . 3 Mb ) of the total repetitive fraction in this species ( Fig 2B , S2 Table , S1 Text ) [14] . Overall , the marked differences in the repertoire of TEs among the three species suggest that they are major contributors to genome evolution , organization , and function , also conceivably affecting their pathogenic lifestyles and contributing to the generation of new virulence specificities . In addition , such differences may also imply differences in TE activity and possibly genome defenses against mobile genetic elements , such as those mediated by repeat-induced point mutation ( RIP ) [22 , 23] , [24] . In this respect , analysis by RIPCAL [25] indicated that a larger fraction of the P . fijiensis ( 60 . 2% , 44 . 58 Mb ) and P . musae ( 53 . 5% , 31 . 97 Mb ) genomic sequences are under RIP as compared to P . eumusae ( 37 . 2% , 17 . 06 Mb ) ( S3 Table , S1 Text ) . In all three genomes RIP occurred mainly on large repeat sequences ( > 500 bp ) as the vast majority ( ~98% on average ) shows signs of RIP . Such high levels of RIP in repeat sequences , although comparable to those reported for other Dothideomycetes [15 , 16 , 21] , are inconsistent with the high density of TEs in the genomes of the three Sigatoka complex species , suggesting that RIP cannot perhaps effectively defend against TE activity ( S1 Text ) . The disease chronology record suggests that P . musae was the first of the three pathogens to appear on the banana host , followed in quick succession by P . fijiensis and then P . eumusae . However , analysis of mating-type genes combined with multilocus sequence analysis of four housekeeping genes suggested a stepwise evolution from a common ancestor splitting first into P . fijiensis and then into P . musae and P . eumusae [10] . Although not conclusively determined , this analysis also suggested that these events are likely to have taken place relatively recently in the evolutionary past of the three pathogens . To discriminate between the two opposing hypotheses and obtain a deeper insight into the species history and divergence times , we reconstructed their phylogenetic relationships based on concatenated sequences of 46 single-copy genes that are conserved across Dothideomycetes [16] , and further used molecular clock analysis to obtain time estimates of their divergence [26] . In order to place the relationships among the three Pseudocercospora species in a broader context of other Dothideomycetes with sequenced genomes , we also incorporated 16 more species of Dothideomycetes in the analysis , including six species from the order Capnodiales , eight species from the order Pleosporales , and two species from the order Hysteriales [16] . Finally , the Eurotiomycete Aspergillus nidulans was used as an outgroup species for rooting the phylogenetic tree . In agreement with previous studies , Pleosporales , Hysteriales , and Capnodiales formed three tight clades within Dothideomycetes , whereas P . eumusae , P . musae , and P . fijiensis produced a highly supported ( bootstrap value of 100 ) monophyletic clade embedded within the Capnodiales , indicating shared ancestry and a close phylogenetic relationship ( Fig 3 ) . In the inferred topology , P . eumusae is sister to P . musae and these two species together are sister to P . fijiensis , thus , conforming with the scenario inferred by other molecular markers of an ancestor successively diversifying first into P . fijiensis and then into P . musae and P . eumusae [10] . To further resolve when these radiations might have taken place and the time interval between speciation events , we used molecular clock analysis [26] to obtain estimates of divergence times . The origin of the Dothideomycetes crown group has been previously estimated to be 394–284 million years ago ( MYA ) , during the Carboniferous period [27] . Using this time period as a calibration point and the highly-supported species tree obtained using the 46 selected genes , we estimated the divergence of the Capnodiales to be 234 . 2–180 . 2 MYA , while the radiation of Pleosporales likely took place much later at approximately 111 . 1–85 . 5 MYA and that of Hysteriales at 146 . 4–112 . 6 MYA ( S5 Fig ) . Within the Capnodiales clade , the emergence of Mycosphaerellaceae is estimated to have occurred between 186 . 7–143 . 6 MYA , thus placing it almost immediately after the appearance of the Capnodiales but considerably earlier than the previously estimated 120–87 MYA [16] . This may be the result of incorporating a higher number of genes in our analysis and/or the limited sampling . Despite such discrepancies in time estimations , our results are in agreement with previous reports that support an earlier origin for the Capnodiales as compared to Pleosporales and Hysteriales [16 , 28] . Within Mycosphaerellaceae , the last common ancestor of P . eumusae , P . musae , and P . fijiensis seems to have appeared at 146 . 6–112 . 8 MYA , splitting shortly after into P . fijiensis and the progenitor of P . eumusae and P . musae at 39 . 9–30 . 6 MYA . Finally , the split between P . eumusae and P . musae is estimated to 22 . 6–17 . 4 MYA ( S5 Fig ) . Combined , these results validate the recent evolutionary radiation of the three species associated with the Sigatoka disease complex and , given the relatively short time interval between speciation events , further suggest high rates of diversification and consequently speciation . De novo genome annotations yielded 10 632 gene models for P . musae of which 10 548 represented protein-coding genes , while the rest were classified as tRNA sequences and pseudogenes . Similarly , a total of 11 173 gene models were predicted for P . eumusae , of which 11 064 represented protein-coding genes ( Tables 1 and S1 ) . The predicted proteome of P . eumusae and P . musae is slightly smaller than that of P . fijiensis ( 13 107 ) but nonetheless within the range of the proteome size reported for other plant pathogenic Dothideomycetes [15 , 16] . Thus , despite the large differences in genome sizes , there is considerably less variation in protein-coding gene counts among the three species that constitute the Sigatoka disease complex . Further annotation of the species predicted proteomes by assignment into the four major functional categories of the eukaryotic orthologous groups ( KOG ) database [29] , indicated that a fairly similar percentage of each species proteome could be assigned to KOGs ( Pm: 59 . 7% , Pe: 61 . 3% , Pf: 55 . 9% ) , although the total number of proteins assigned to each main category of KOG could be different among the species ( S6 Fig , S1 Text ) . Similarly , proportionally to their proteome sizes the three species do not exhibit any significant differences in the percentage of proteins distributed across the 25 subcategories of KOG , indicating that , based on their KOG profiles , they execute a fairly similar spectrum of biological activities ( S6 Fig , S1 Text ) . We focused next on a comparative analysis of the protein-coding gene complements of the three species . For this purpose , reciprocal BLAST analysis ( e-value: 1e-5 , alignment coverage > 50% ) as implemented in OrthoMCL [30] was used to retrieve the set of orthologous protein-coding gene groups among the three species and consequently determine the core , lineage- and species-specific gene families and genes . We defined “core” as the gene families that are shared by all three species and “lineage-specific” as the subset of core gene families that are not present in any other fungus . We considered “species-specific” as genes that are found in only one of the three species that constitute the Sigatoka disease complex , while we classified “orphans” as the subcategory of species-specific genes that do not have homologs in the other fungal species . A total of 6307 protein-coding gene families shared by all three species were identified that represent their core proteome complement ( Fig 4A , S7 Fig , S1 Text ) , whereas a broader BLAST-based search ( e-value: 1e-5 , alignment coverage > 50% ) against all currently available fungal genomes in the JGI database revealed that 234 of the core families are lineage-specific to the Sigatoka species , which could facilitate virulence specifically to the banana host ( Fig 4B , S8 Fig , S1 Text ) . A larger number of species-specific protein-coding genes were retrieved from P . fijiensis ( 3442/13 107 , 26 . 2% ) as compared to P . eumusae ( 1759/11 064 , 15 . 9% ) and P . musae ( 1867/10 548 , 17 . 7% ) ( Fig 4A , S7 Fig , S1 Text ) , which is in line with the earlier branching of P . fijiensis from the last common ancestor of the three species [10] . Of the species-specific genes , 2176 , 1403 , and 1120 genes in P . fijiensis , P . musae , and P . eumusae , respectively , can be further classified as putative orphans , as no homologs could be identified in any other species ( Fig 4B , S8 Fig , S1 Text ) . Taken as a whole , it is perhaps surprising to see such diversity in the species’ gene complements given the common ancestry and relatively short evolutionary distance among the three species along with the fact that they have been co-evolving with their banana host . This , in turn , implies that the evolution of virulence in these pathogens has , to an extent , been facilitated by a number of species-specific adaptations . While species-specific acquisitions of new genes with novel functions have likely significantly contributed to the phenotypic variation among the three species , changes in gene family sizes as a result of gene duplication , loss , or elevated sequence diversification are also a major evolutionary force that could have further fostered the shifts in virulence spectra . The pairwise comparisons of gene content , for example , showed that the number of gene families shared exclusively between P . fijiensis and P . eumusae ( n = 1591 ) is much larger than the number of gene families shared only between P . fijiensis and P . musae ( n = 1021 ) ( Fig 4A , S7 Fig ) . This was rather surprising as it suggests that the evolutionary distance between P . fijiensis and P . eumusae is shorter than the one between P . fijiensis and P . musae . Alternatively , it could be that P . fijiensis and P . eumusae share more similar patterns of duplications and losses in gene families that were inherited from the common ancestor of the three species . If true , it is conceivable that the evolution of virulence in P . fijiensis and P . eumusae may have been additionally facilitated by parallel gains and losses in specific gene families , which in turn may underlay the molecular basis of virulence in these two pathogens . To investigate this possibility , we first examined whether specific gene categories are enriched for copy number variants ( CNVs ) among the three species and , subsequently , whether a pattern exists on the expansion and reduction in the size of gene families with CNV among P . musae , P . eumusae , and P . fijiensis that could be linked to changes in their virulence phenotypes . Of the 6307 core gene families shared by the three species , 5583 are single-copy families . The remaining 724 correspond to multi-copy gene families , of which 575 display copy number variations ( CNV ) among the three species . Functional annotations revealed that while the KOG-based distribution of the 5732 gene families without CNV follows a similar pattern to that obtained for the core proteome of the species ( Fig 5A ) , in contrast , gene families with CNV are significantly enriched in genes encoding for proteins that are involved in metabolism ( 211 KOG terms , 190/575 gene families , 33% ) rather than cellular processes and signaling ( 84 KOG terms , 76/575 gene families , 13 . 2% ) , or information storage and processing ( 36 KOG terms , 35/575 gene families , 5 . 7% ) ( Fig 5A ) . Further characterization of the gene families with CNV , according to the subcategories of KOG , showed that most could be classified in secondary metabolite biosynthesis transport and catabolism ( 46 gene families ) , followed by carbohydrate metabolism and transport ( 42 gene families ) , and finally lipid transport and metabolism ( 39 gene families ) ( Fig 5B ) . Taken together , the above results indicate that changes in gene family sizes across the three species are not selectively neutral and uniform for all biological processes , but largely affect genes involved in metabolic processes . Such a functional bias in gene categories enriched for CNVs implies an association of virulence with altered metabolism in the three pathogens , perhaps for enhanced uptake and utilization of the nutrients obtained from the host and/or production of certain secondary metabolites . To further elucidate whether a causal relationship exists between CNV in genes involved in metabolism and the species virulence phenotypes , we performed hierarchical clustering based on the KOG distribution profiles ( i . e . by enumerating the number of genes assigned to each category of KOG ) of the species entire proteomes and compared it with the species hierarchical clustering based on the KOG distribution profiles of their core gene families with CNV . When clustering was performed using the species entire proteomes , then the obtained tree topology was reflective of their evolutionary relationships , with P . musae and P . eumusae clustering together as a monophyletic group ( Fig 6A ) . In contrast , hierarchical clustering of the species based on the KOG distribution profiles of the 575 core gene families with CNV ( Fig 6B ) or the subset of 190 gene families with CNV that are predicted to be involved in metabolism ( Fig 6C ) , returned swapped topologies in which P . fijiensis clustered with P . eumusae as a monophyletic group with strong supporting bootstrap values ( 93 and 86 , respectively ) . These clustering patterns were consistent and irrespective of distance measure and clustering algorithm used , suggesting that P . fijiensis and P . eumusae share a more congruent pattern of gene family expansions and contractions as compared to P . eumusae and P . musae or P . fijiensis and P . musae . Similar results were also obtained when the above analysis was expanded to include gene families that are shared by at least two of the species but not necessarily the third one , in which case pairwise comparisons showed that a significantly higher number of the gene families had exactly the same copy number shared between P . eumusae and P . fijiensis ( 1742 gene families ) , rather than between P . musae and P . fijiensis ( 1127 gene families ) or between P . musae and P . eumusae ( 945 gene families ) ( S9 Fig , S1 Text ) . Moreover , the analysis of CNV in the metabolic gene families of the nine Capnodiales species that were previously used for phylogenetic reconstruction and estimation of divergence times ( Fig 3 , S5 Fig ) , further supported that the clustering of P . fijiensis together with P . eumusae , when considering changes in metabolism , is likely due to parallel expansions and contractions in these two species rather than changes that took place solely in P . musae ( S10 Fig , S1 Text ) . Although the analysis performed based on the KOG annotations of the species entire proteomes indicated that the more virulent P . eumusae and P . fijiensis share complementary patterns of expansions and contractions in core gene families related to metabolism , it does not provide any information regarding the metabolic pathways that these gene families are involved in . To investigate which pathways are likely to have been affected by parallel changes in the two more virulent species , we performed a genome-wide GO ( Gene Ontology ) -based analysis and identified GO terms that support the clustering of P . eumusae with P . fijiensis ( S11 Fig , S1 Text ) . The analysis indicated that GO terms associated with metabolic processes ( GO: 0008152 ) and particularly regulation of metabolic processes ( GO: 0019222 ) and cellular metabolic processes ( GO: 0044237 ) ( S12 Fig , S1 Text ) are those contributing the most to the clustering of P . eumusae together with P . fijiensis when considering changes in the species proteome , thus further corroborating the KOG-based analysis . Taken together , the above results indicate that changes in gene family sizes among the three species that constitute the Sigatoka disease complex have not been selectively neutral but are more respectful of the species virulence profiles rather than their evolutionary relationships . This implies that , next to species-specific evolutionary adaptations , the evolution of virulence in the three pathogens has also been driven by recurrent genomic changes on particular molecular pathways . Among the evolutionary mechanisms shared by the more virulent P . fijiensis and P . eumusae are matched changes in the size of families related to metabolism that could potentially translate into a higher efficiency of nutrient uptake and utilization . Although speculative , the annotation of the species-specific genes shows that they mostly encode for novel proteins with unknown function , suggesting that they might be virulence-associated genes with a role in overcoming or evading the host immune system . The fairly coordinated changes in the size and range of metabolic gene families shared between P . fijiensis and P . eumusae suggests that many of these families could have played a significant role in the evolution of virulence in these two pathogens . However , next to nutrient uptake and utilization , nutrient acquisition through the enzymatic degradation of plant polysaccharides is also an important aspect of pathogenesis that promotes host colonization and infection . To assess the ability of P . musae , P . eumusae , and P . fijiensis to degrade and metabolize different polysaccharides , we annotated and contrasted their repertoires of putative carbohydrate-active enzymes ( CAZymes ) , with an emphasis on characterizing enzymes that are involved in the breakdown of plant cell walls ( PCWs ) . In order to identify any features specific to the three banana pathogens we , additionally compared the CAZyomes of the three Sigatoka species to the ones of 16 other Dothideomycetous fungi with different nutritional lifestyles and host specificities [15 , 16] ( S1 Text ) . Our CAZy annotations identified a total of 490 , 501 , and 516 CAZyme modules from all six major CAzyme superfamilies in the predicted proteomes of P . musae , P . eumusae , and P . fijiensis , respectively ( S4 Table , S13 and S14 Figs , S1 Text ) . Plant cell wall degrading enzymes ( PCWDEs ) , in particular , are the most abundant in the three species , accounting approximately for a quarter of their CAZyomes ( Pm: 119/490 , 24 . 3%; Pe: 125/501 , 25 . 0%; Pf: 130/516 , 25 . 2% ) . The majority of PCWDEs are putatively directed towards the degradation of hemicellulose ( Pm: 54 . 6% , Pe: 55 . 2% , Pf: 53 . 1% ) , followed by the decomposition of hemicellulose-pectin complexes ( Pm: 21 . 0% , Pe: 22 . 4% , Pf: 21 . 5% ) , pectin ( Pm: 21 . 0% , Pe: 20 . 8% , Pf: 22 . 3% ) , and cellulose ( Pm: 3 . 4% , Pe: 1 . 6% , Pf: 3 . 1% ) ( S5 Table , S15 Fig , S1 Text ) . The higher number of hemicellulases in the three Sigatoka species is not unusual among plant pathogenic fungi [16 , 31] , whereas comparative analysis with the group of 16 Dothideomycetous fungi included in this study did not , based on Mann-Whitney U tests , identify any significant differences in the abundance of PCWDEs present in these groups . However , significant differences were detected at the individual CAZYme family level , including when the CAZyme distribution profiles of the three Sigatoka species were compared with the distribution profiles of five hemibiotrophic fungi from the Capnodiales clade that were included in the group of 16 Dothideomycetes ( S6 Table , S1 Text ) . Such differences could reflect an evolutionary adaptation of P . musae , P . eumusae , and P . fijiensis to their banana host and the fine-tuning of their CAZyme repertoire for a better exploitation of the polysaccharide resources available in this host . Although the three banana pathogens share similar overall numbers in PCWDEs , they do display some differences at the individual family level , perhaps as a result of the enzymatic redundancy exhibited among many of the CAZy families ( S6 Table , S16 Fig , S1 Text ) . Notably , hierarchical clustering of the species based on the distribution profiles in individual CAzyme families of their entire CAZYomes or arsenal of PCWDEs , resulted once more in P . eumusae grouping with P . fijiensis rather than P . musae , as expected based on the phylogenetic placement of the three species . This indicates that P . eumusae and P . fijiensis share complementary patterns of expansions and contractions in CAZymes ( Fig 7A ) and PCWDEs ( Fig 7B ) more specifically . Such coherent changes between P . eumusae and P . fijiensis in the size of gene families related to nutrient acquisition could reflect evolutionary changes that underlie a more effective exploitation of the banana host . Thus , in addition to parallel adaptations for nutrient utilization , P . eumusae and P . fijiensis seem to have evolved more similar mechanisms for nutrient acquisition as well . Taken together , based on their overall arsenal of CAZymes , the three species likely do not exhibit substantially large differences in their ability to break-down and metabolize different types of plant cell material , although , given their differences at the individual CAZy family level , they may differ in the efficiency by which they hydrolyze different types of polysaccharides . The production of phytotoxic metabolites by the three pathogens that constitute the Sigatoka disease complex has long been known , but whether these play a pivotal or rather secondary role in the interaction of the pathogens with their Musa host remains in question [32–37] . To obtain an insight into the commonalities and differences of the arsenal of phyto- and mycotoxins that may be produced by the three banana pathogens , we performed an inventory of the genes encoding the four core enzyme types that catalyze the first committed step in the biosynthesis of the major secondary metabolite ( SM ) classes found in fungi , namely the non-ribosomal peptide synthases ( NRPSs ) , the polyketide synthases ( PKSs ) , the terpene synthases ( TSs ) , and the dimethylallyl tryptophan synthases ( DMATSs ) ( S1 Text ) [38] . Despite their hemibiotrophic lifestyle , 28 , 27 , and 21 genes encoding core SM enzymes were identified in the genomes of P . musae , P . eumusae , and P . fijiensis , respectively , indicating that the three pathogens have the ability to produce diverse SMs . The majority of core enzymes in all three species are predicted as PKSs ( 7 in Pm: PksA-to-PksG , 10 in Pe: Pks1-to-Pks10 , and 7 in Pf: PksI-to-PksVII ) , followed by NRPSs ( 10 in Pm: NpsA-to-NpsK , 7 in Pe: Nps1-to-Nps6 , and 8 in Pf: NpsI-to-NpsVII ) or hybrid PKS-NRPSs ( 1 in Pm: PksNpsA , 2 in Pe: PksNps1 and PksNps2 , and 2 in Pf: PksNpsI and PksNpsII ) , and finally TSs ( 5 in Pm: TsA-to-TsG , 5 in Pe: Ts1-to-Ts5 , and 4 in Pf: TsI-to-TsIV ) ( S7 Table , S1 Text ) . No DMATs were detected in any of the three species . The number and type of core SM genes predicted in the genomes of the three banana pathogens are comparable to those reported previously for other species of Capnodiales , including the close-related tomato pathogen F . fulva , the wheat pathogen Z . tritici , and the poplar pathogen S . populicola [15 , 16] . Furthermore , phylogenetic analysis with other fungal core SM enzymes [39–40] , showed that most core enzymes from the three banana pathogens could be clustered with high support ( ML bootstrap values ≥80% ) with enzymes that are involved in the biosynthesis of known phyto- and mycotoxins in other fungi , and thus could be involved in the production of structural analogs with matching backbones . Among others , these include core enzymes that are involved in the biosynthesis of notorious mycotoxins , such as fumonisins , and light-activated phytotoxins , such as elsinochrome and cercosporin , thus corroborating earlier experimental findings suggesting the involvement of photoactivated toxins in the pathogenesis of the three Sigatoka species ( S17 , S18 and S19 Figs , S1 Text ) . Overall , the annotation and analysis of core SM enzymes in P . musae , P . eumusae , and P . fijiensis showed that although the three pathogens share some orthologous core enzymes , they differ in the arsenal of SMs that they potentially produce , some of which could bare structural similarity in their backbone structure to already characterized phyto- and mycotoxins ( S1 Text ) . To gain a deeper insight into the pathogenic potential of the three species that constitute the Sigatoka disease complex , we characterized their secretomes ( S20 Fig ) , placing an emphasis on identifying and comparing their repertoires of candidate effectors . A total of 612 , 638 , and 584 secreted proteins , of which 110 , 112 , and 105 represented putative effector proteins , were predicted in the genomes P . musae , P . eumusae , and P . fijiensis , respectively , indicating that the three species employ secretome and effector arsenals of comparable size to those of most other hemi-biotrophic fungi ( Mann-Whitney U test , P-value = 0 . 01 ) ( S8 Table , S1 Text ) . Clustering by OrthoMCL indicated that , on average , ~50% of the effectors in each species could be regarded as species-specific ( Pm: 48 effectors , Pe: 54 effectors , Pf: 61 effectors ) , while a broader BlastP-based search for homologs in the NCBI nr database and the JGI fungal genome database , suggested that a large number of the species-specific effectors can be further classified as orphans ( Pm: 30 effectors , Pe: 27 effectors , Pf: 39 effectors ) ( S9 Table; S21A Fig ) . To further confirm that some of the differences in the effector repertoires of the three species are species- rather than strain-specific , we randomly selected a set of 12 species-specific or orphan effectors from each of the three pathogens and used PCR , with primers designed within the effectors’ genes coding sequences , to amplify them from seven isolates of each species . PCR and subsequent sequencing analysis of the amplified products confirmed that the 12 randomly selected species-specific or orphan effectors were both conserved within their species of origin and absent in the other two species ( S10 Table; S22 Fig ) . The clustering analysis also indicated that more effector families are shared between P . eumusae and P . musae ( n = 23 ) as compared to P . fijiensis and P . eumusae ( n = 9 ) or P . fijiensis and P . musae ( n = 10 ) . Thus , unlike changes in the metabolome and CAZyome of the species , clustering of the species based on the effector repertoires is more respectful of their evolutionary relationships rather than their virulence on their Musa host . Moreover , 22 core effector families shared by all three pathogens were identified , seven of which can be regarded as lineage-specific , as they were only present in the three pathogens that constitute the Sigatoka disease complex and none of the other fungal species ( S9 Table , S21A Fig , S1 Text ) . Among the core effectors shared by the three banana pathogens and other fungi are three paralogs of Ecp2 ( i . e . Ecp2-1 , Ecp2-2 , and Ecp2-3 ) [41] and homologs of the F . fulva Ecp6 [42] and Avr4 [43] chitin-binding effectors ( S9 Table , S1 Text ) . Overall , the analysis suggests that the three banana pathogens , despite their very close evolutionary relationships , common host and infection biology , exhibit a considerably diverse arsenal of effector proteins that could have contributed to their differences in virulence ( S1 Text ) . Next to gross genomic changes in content and architecture , the identification of the genes and genetic pathways most affected by selection during speciation is essential for both understanding the evolutionary history of fungal plant pathogens , as well as for finding important traits that contribute to phenotypic diversity and disease [44] . Our previous analysis of gene content indicated a functional bias in the pattern of expansions and contractions in families related to metabolism and enzymatic degradation of PCWs . Here we examined whether , within the group of orthologous genes shared by the three species , similar patterns of elevated selection pressure could be observed among the different functional categories of gene families . Along the same lines , we also investigated whether putative effectors and other secreted proteins shared by the three pathogens show evidence of positive selection or higher evolutionary rates . If the case , such findings could suggest that next to changes in gene content , positive selection has also contributed to the phenotypic divergence of the three species . For the analysis of selection pressures , we used the maximum likelihood method implemented in the Codeml program of PAML [45] to calculate the ratio of non-synonymous ( dN ) to synonymous ( dS ) substitutions for all between species pairwise comparisons of the 6307 orthologous gene families shared by them . For any given pair dN/dS >1 is suggestive of positive selection , while dN/dS <1 indicates purifying selection . As different parts of the proteome and functional categories of genes could experience significant differences in selection pressure , dN/dS ratios were also examined separately for different gene families and subgroups of genes , including , for example , genes encoding secreted or non-secreted proteins and genes encoding putative effectors or secreted proteins excluding the effectors . dN/dS ratios for the entire set of orthologous genes shared by the three species ranged from 0 . 00–2 . 63 , while the median dN/dS value is very low ( 0 . 1 ) indicating that the vast majority of the orthologous gene pairs appear to be under purifying selection ( S23 Fig ) . The subgroup of genes encoding for secreted proteins displayed slightly higher evolutionary rates as compared to genes encoding non-secreted proteins , although median dN/dS values for each specific subgroup remained very low ( 0 . 128 and 0 . 099 , respectively ) . Also , within secreted proteins , putative effector encoding genes have experienced relatively higher levels of adaptive evolution ( median dN/dS value of 0 . 214 ) as compared to the pool of secreted but non-effector encoding genes ( median dN/dS value of 0 . 124 ) . However , caution is needed when comparing evolutionary rates among the different subgroups , as the sample sizes used in calculations of median dN/dS values varied considerably among them . In this respect , median dN/dS values were lower for all groups than mean values , suggesting a skewed distribution and an excess of proteins with evolutionary rates lower than the average . A search within each group for orthologous pairs with dN/dS >1 identified only a single core effector ( Avr4-2 ) , which however did not receive any statistical support ( P = 0 . 233 , Fisher’s exact test ) for being positively selected , and 27 non-secreted proteins of which only four received statistical support at the 0 . 05 level for being positively selected ( S11 Table ) . Of these four proteins , one could be annotated as a sulfatase based on Pfam and GO annotations , while none of the other three proteins received any functional annotations . In addition , examination of dN/dS rate ratios in orthologous pairs of protein-coding genes representing the different functional categories of KOG did not indicate any significant differences in evolutionary rates between the group of genes encoding for proteins that are involved in metabolism ( median dN/dS value of 0 . 07 ) as compared to the group of genes encoding for proteins that are involved in cellular processes and signaling ( median dN/dS value of 0 . 08 ) , information storage and processing ( median dN/dS value of 0 . 08 ) , or poorly characterized ones ( median dN/dS value of 0 . 09 ) ( S24 Fig ) . Also , among the group of orthologous CAZymes that are shared by all species , we could not identify any genes as being under positive selection or an elevated dN/dS ratio for the subgroup of genes encoding PCWDEs ( S25 Fig ) . Overall , based on a global analysis of dN/dS ratios , we identified only very few cases of positive selection in the group of orthologous genes shared by the three species . Instead , we observed abundant purifying selection , suggesting that the conserved between the species proteome has likely played a less significant role in the phenotypic diversification among the three species . Currently , the Sigatoka disease complex of banana , caused by the closely related Dothideomycetes ( Ascomycetes ) , P . musae , P . eumusae , and P . fijiensis , is the most devastating disease on bananas , reducing yields by more than 40% [3–5] . The three species have surfaced as destructive pathogens on bananas during the last century and although they have evolved from a recent common ancestor , clear differences in virulence exist amongst them that correlate with the time of their appearance [5–7] . Within this complex , P . musae was the first of the three pathogens to be recorded on banana , although black Sigatoka caused by P . fijiensis is currently the major agronomic constraint for banana production , necessitating over 50 contact fungicide applications per year for its control . It is also one of the most marked examples of a recent pandemic in the plant kingdom and , considering the importance of banana as a staple food crop , a serious threat to global food security . Despite its aggressiveness , over the last decade black Sigatoka is gradually replaced by P . eumusae , which appears to be equally , if not more , aggressive and resilient than P . fijiensis [5–7] . Thus , there is an urgent need to understand the pathobiology of these species in order to safeguard banana production for the future [1 , 4] . The relative short evolutionary distance of the three Sigatoka pathogens and their differences in virulence that broadly parallel their historical record of appearance , offer an excellent opportunity to examine the genomic changes associated with increased virulence , speciation , and specialization of parasites on their host . Evolution of microbial virulence and the genetics of host-adaptation is a highly active and competitive field but there is only a limited knowledge about these processes in plant pathogenic fungi , as compared to bacteria and oomycetes , Here , we have sequenced the genomes of P . musae and P . eumusae , and compared them with the available genome sequence of P . fijiensis [14] in order to first understand the nature , diversity and extent of genomic modifications associated with shifts in their virulence spectra on banana after speciation and second , to determine whether some of the changes and evolutionary processes are recurrent across the species , and thus predictable . A critical question in fungal evolutionary biology is whether speciation and diversification of virulence is mainly facilitated by adaptive evolution of the core genome or through species-specific gene acquisitions . Our analysis showed that speciation has largely altered both the genome architecture and composition of the three species . More specifically , comparative analysis of genome architectures revealed marked differences in genome sizes among the three species that positively correlate with different rates of TE , and especially LTR-retrotransposon , accumulation and retention . The three species also show marked differences in the type of TEs that they maintain in their genomes , including in the ratios of Class I and Class II TEs . As these two classes of transposons leave different imprints on coding and non-coding DNA sequences [46] , they may have also differentially impacted genome evolution and innovation in the species . The differential invasion of the genomes by TEs has also likely contributed to chromosomal rearrangements and the breakdown of macrosynteny among the three species , consequently accelerating the process of speciation and diversification . Analysis of gene content showed that although the three species retain a similar in size predicted arsenal of protein-coding genes , they exhibit considerable differences in their gene composition , suggesting that the evolution of virulence in these pathogens has , to an extent , been facilitated by a number of species-specific adaptations . This is particularly true for putative virulence associated genes , such as those encoding for effectors , as ~50% of the effectors in each species could be regarded as species-specific . Notably , of the core effectors , seven were found only in the three pathogens that constitute the Sigatoka disease complex and these might play an essential role in the interaction with the banana host . Next to overcoming the host immune system , the capacity for metabolic adaptation , in terms of acquiring and exploiting the host nutrient resources has also likely played a major role in the evolution of virulence in the three species . In this respect , metabolic streamlining in P . fijiensis and P . eumusae through independent but parallel expansions and contractions in gene families that are associated with metabolism and PCWDEs may have contributed to the increased virulence of these two species on the banana host . Such parallel changes in the two most aggressive species suggest that they may represent molecular fingerprints of adaptation to the banana host . Thus , next to species-specific adaptations , convergent evolution in specific molecular pathways seems to have facilitated the evolution of higher virulence in P . eumusae and P . fijiensis .
The genomes of P . musae ( strain CBS116634 ) and P . eumusae ( strain CBS114824 ) were sequenced by the UC Davis Genome Sequencing Core facility using the Illumina HiSeq technology ( 150 bp pair-end reads ) . A total of 22 . 7 and 26 . 0 million pair-end reads were obtained for P . musae and P . eumusae , respectively . The read quality was assessed by FastQC [47] and low quality reads and/or bases were trimmed using Trim Galore [48] . The high quality reads were assembled using different assembly software , including SoapDenovo2 [49] , SPAdes [50] , and ABySS [51] , and different k-mer sizes ( k = 55 , 77 , 99 , and 121 ) and the assembly with the highest assembly qualities in terms of N50 value and assembly size was selected and merged by GAM-NGS [52] to obtain a consensus assembly for each species . The consensus assembly was scaffolded by SSPACE [53] and the remaining gaps in the scaffolds were closed by GapFiller [54] . The estimated genome coverage is 112x in P . musae and 165x in P . eumusae . The P . musae ( GenBank: LFZO01000000 ) and P . eumusae ( GenBank: LFZN01000000 ) genomes have been deposited to DDBJ/EMBL/GenBank , whereas the genome of P . fijiensis was reported earlier ( GenBank: GCA_000340215 . 1 ) [14] . An estimation of the repeat content size was first performed through a calculation of k-mer occurrence by Jellyfish [55] using k = 17 bp and summarized as a histogram . The histogram was examined by custom R scripts to partition it into regions that corresponded to potential unique and repetitive fractions of the genomes , based on peak positions . The total number of k-mer in the unique and repetitive fractions was calculated as an estimate of the fraction size . RepeatModeler [56] incorporating RECON [57] , RepeatScout [58] , TRF [59] , and RepeatMasker [60] was used for de-novo identification and modeling of the different classes of repeat families . For each species , RepeatModeler produced a library of classified putative interspersed repeats . All repeat families were compared with Repbase sequences [61] for classification . The consensus repeat element library identified in each species was fed into the downstream annotation pipeline . The genomic regions subject to repeat induced mutation ( RIP ) were predicted following the composite RIP index ( CRI ) method as described in de Wit et al . ( 2012 ) [15] . RIPCAL [25] and custom Perl scripts were used to analyze and annotate the genomic regions under RIP mutations . The RIPed sequences were defined according to RIP product ( ≤ 1 . 2 ) , RIP substrate index ( ≤ 0 . 8 ) and composite RIP indices ( ≥ 1 . 0 ) . A genomics region was considered as a RIPed locus when its sequence length was larger than 750 nt along with a peak CRI ≥1 . 5 . The P . musae and P . eumusae genomes were both annotated using the Maker2 annotation pipeline [62] , which incorporated several gene model prediction programs and sequence analyses based on EST and transcriptome , to improve the quality of genome annotations . In the pipeline , RepeatMasker [60] was first used to mask the genome regions that were comprised of low-complexity repeats and interspersed repeats , based on the repeat element library produced from RepeatModeler . RepeatRunner [63] was then used to identify more divergent transposable elements and viral proteins that may have been missed by RepeatMasker . After masking the repeat elements , ab initio gene predictors such as SNAP [64] , Augustus [65] , and GeneMark-ES [66] were used for prediction of gene models in the genomes . To improve the quality of gene model prediction , we performed transcriptome sequencing ( RNA-seq ) of cDNA libraries representing two different in vitro growth conditions , i . e . growth in rich media ( 10 g/L Yeast extract , 30 g/L Glucose ) and growth in poor media ( 1 g/L KH2PO4 , 1g/L KNO3 , 0 . 5 g/L MgSO4x7H2O , 0 . 5 g/L KCl , 0 . 5 g/L Sucrose , 0 . 5 g/L Glucose ) , using Illuimina HiSeq platform ( PE100x100 ) in each species . The generated 26 . 3 and 23 . 6 M of pair-end reads for P . musae and P . eumusae , respectively , were de novo assembled by Trinity [67] and the resulting transcriptome shotgun assemblies have been deposited at DDBJ/EMBL/GenBank under the accession GDIK00000000 ( P . musae: PID PRJNA289098 and P . eumusae: PID PRJNA289096 ) . The resulted transcriptomes along with ESTs deposited in the NCBI dbEST database were used for training the gene prediction parameters . Maker2 merged all the predicted gene models from different gene predictors to generate a set of predicted gene models , which were further polished by EST and protein alignments by BLAST and Exonerate [68] to avoid spurious predicted gene models . To further improve the performance of the de novo gene prediction , a second round of gene predictions was conducted using the generated gene annotations as input for the training step in order to re-annotate the genomes using the Maker2 pipeline iteratively . The completeness of the genome assembly was assessed by the CEGMA pipeline [18] , as indicated elsewhere [19] . Gene families were predicted using the OrthoMCL pipeline [30] , which produces normalized score based on the E-values generated from an all-versus-all BLASTp analysis ( 1e-5 as the cutoff value ) for pairs of the compared genomes . The normalized scores were fed into the MCL algorithm to classify the genes into hypothesized orthologous and paralogous gene families using a default inflation parameter of 1 . 5 . Functional annotations were first performed using the InterProScan pipeline [69] , which compared encoded protein sequences against the PFAM [70] , PROSITE [71] , and ProDom [72] , to identify the domains and motifs present in each gene model . Meanwhile , the associated gene ontologies and pathways of each gene model were retrieved for the InterProScan hits , when available . The second layer of genome annotations was performed by sequence similarity , by comparing the protein sequences ( BLASTp ) against the non-redundant protein database in NCBI and SwissProt database . A hit was considered significant when the E-value was lower than 1e-4 and the coverage higher than 50% . The eukaryotic orthologous groups of proteins ( KOG ) was analyzed by RPSblast [73] against the KOG database deposited in the NCBI CDD database ( E-value < 1e-3 ) . The frequency of GO terms , as identified using InterProScan was also enumerated in each species . Based on the GO frequency , we implemented the random forest method ( REF ) to select the GO terms that may contribute to the observed switched topology [74] . A total of 5000 trees were generated and the classification was based on combining the entire generated trees using a majority rule . The mean decrease of the Gini index ( MDGI ) was used to select the important GO terms . A supervised hierarchical clustering was applied for the GO terms with MDGI value > 0 . 01 to produce the clustering topology and heatmap . The annotation of the carbohydrate-active enzymes was performed based on a sequence search against the CAZyme Hidden Markov Models ( HMM ) using the HMMER3 as implemented in the dbCAN annotation server ( E-value < 1e-4 ) [75] . The secondary metabolic genes were annotated by the AntiSMASH 2 . 0 pipeline [76] , using the HMMs of nonribosomal polypeptide synthetase ( NRPS ) , polyketide synthase ( PKS ) , and terpene synthase ( TPS ) . The prediction was further cross-validated by BLASTp analysis . The phylogenetic trees of NRPS and PKS were constructed based on the predicted NRPS and PKS sequences in P . musae , P . eumusae and P . fijiensis and an additional set of NRPS and PKS homologous sequences as described in Collemare , et al . [77] . For TPS , the sequences for phylogenetic tree construction was by blasting the TPS protein sequences in P . musae , P . eumusae and P . fijiensis against the SwissProt database ( E-value < 1e-4 and coverage > 50% ) . The clustering analyses of the annotation were performed by the R ggplot [78] package . The clustering procedure was performed with different distance measures ( Euclidian and Manhattan ) and linkage methods ( Ward , single and complete linkage methods ) , all of which produced a consistent clustering topology . The reliability of the topologies was assessed by multiscale bootstrap analyses by Pvclust [79] with 1000 bootstraps . The SignalP [80] , TMHMM [81] , TargetP [82] , Phobius [83] were incorporated in the InterProScan pipeline [69] to predict the presence of signal peptide , transmembrane ( TM ) domains and cellular localization for each protein sequence . The WoLF PSORT program [84] was used to refine the prediction result . This information was used for secretome and effector protein prediction . Briefly , proteins with a signal peptide and a signal peptide cleavage site were predicted by SignalP ( D-score > 0 . 5 ) , whereas those with no TM domains or with a single TM domain within the first 40 amino acid and overlapping with the signal peptide as predicted by TMHMM and Phobius were considered as candidates of secreted proteins . The candidate proteins that were predicted by TargetP as targeted to mitochondria were also discarded . The prediction was re-examined by WoLF PSORT and those consistently predicted as secreted proteins were considered as true candidates . Finally , PredGPI [85] was used to predict the presence of GPI-anchor signal in candidate proteins , in which those with a predicted GPI-anchor signal were removed to yield the final set of secreted proteins . An effector was defined as a secreted protein with a protein length < 250 aa and a high percentage of cysteine residues in the protein that was higher than two-fold of the average cysteine percentage in all predicted proteins of each species . The amplification of selected effectors from field isolates of the three species was performed by PCR , using primers designed at the beginning and the end of the effector’s coding sequence ( S12 Table ) . Genomic DNA of the isolates was kindly provided by Prof . Gert Kema ( Wageningen University—Plant Research International , The Netherlands ) , Pablo Chong Aguire ( Wageningen University—Plant Research International , The Netherlands ) , and Dr . Ewald Groenewald ( CBS-KNAW Fungal Biodiversity Centre , The Netherlands ) . PCR conditions included an initial 95°C denaturation step for 10 minutes followed by denaturation for 15 seconds at 95°C , annealing for 30 seconds at 50–60°C depending on the effector amplified , and extension for 30 seconds at 72°C for a total of 35 cycles . PCR products were directly sequenced using the Sanger technology and sequences were aligned to the original effector sequence using the MEGA6 software [86] . An orthoMCL [30] classification was performed on the three target species along with 17 additional species [16] to identify the homology between species following the approach as described above . A total of 46 single-copy orthologous genes were identified and used for the subsequent analysis . The amino acid sequences of these 46 genes were aligned using PRANK [87] . All the gaps present in the alignments were removed by Gblocks [88] prior to phylogenetic tree construction . Two different approaches were used for phylogenetic tree construction . First , all the genes were individually subjected to a tree construction using the maximum likelihood approach by RAxML ( 1000 bootstraps ) and a consensus tree was produced [89] . The best evolutionary model for each alignment was determined by ProtTest [90] . Second , a maximum likelihood ( ML ) tree was constructed based on a concatenated alignment using PROTGAMMAWAG model with 500 rapid bootstraps . Both tree topologies were found to be consistent with each other . The divergence time of the species was estimated using the phylogenetic tree , based on the concatenated alignment by the penalized likelihood analysis , as implemented in the r8s program [26] . Based on previously published data [27] , the upper and lower bound of the divergence time estimation of the root of tree ( the Dothideomycetes crown group ) was calibrated as 394 million years ago ( MYA ) and 284 MYA , respectively . The final chronogram was visualized by FigTree [91] . The syntenic relationships among the three species were calculated using SyMap 4 . 0 [92] . Since the P . eumusae and P . musae genomes are more fragmented than the P . fijiensis genome , the P . fijiensis genome was used as a reference in the analysis . SyMap first performed an alignment of the genomes using MUMmer [93] and identify the anchor hits clusters by clustering the MUMmer hits into gene or putative gene regions . The clustered anchor regions were filtered by a reciprocal-best filtering algorithm . Synteny blocks were then identified by searching collinear sequences of anchors in the compared genomes . Pairwise synonymous and non-synonymous substitution rates ( dN and dS ) were calculated for the gene families with one-to-one orthology relation in the proteomes of the three species . The sequences of each family were aligned using PRANK [87] based on protein sequences and back-translated into codon alignment . The alignments were trimmed by Gblocks [88] with stringent criteria that trimmed small alignment blocks , gaps from the alignments . The Codeml program of PAML [45] was used to calculate pairwise dN and dS ( mode = -2 ) , taking the transition and transversion bias and codon usage bias into consideration . Fisher’s Exact test ( FET ) was used to assess the significance level of selection . | Understanding the evolutionary and genomic changes involved in the emergence of new pathogens and shifts in virulence spectra is vital for deciphering the biological process of disease emergence and for designing new and effective disease control methods . In this study , we employed comparative genomics in order to examine the nature , diversity , and extent of genomic modifications associated with changes in virulence among Pseudocercospora musae , Pseudocercospora eumusae , and Pseudocercospora fijiensis , the main constituents of the Sigatoka disease complex on banana , currently one of the most destructive diseases on banana worldwide . Our comparative genome analyses have highlighted the role of pathoadaptive changes in virulence associated genes , such as those encoding for effectors , in shaping the underlying differences in virulence spectra among the three species , and also revealed that changes in the size of gene families associated with nutrient acquisition and assimilation are more respectful of the species virulence profiles rather than their evolutionary relationships . Thus , we posit that next to species-specific evolutionary adaptations in virulence-associated genes , the increase in virulence of P . eumusae and P . fijiensis has been driven by convergent evolution in metabolic pathways that likely facilitate a higher efficiency of nutrient acquisition , uptake , and utilization . | [
"Abstract",
"Introduction",
"Results",
"and",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"bananas",
"genome",
"evolution",
"fungal",
"genetics",
"plant",
"science",
"phylogenetic",
"analysis",
"crops",
"plant",
"pathology",
"molecular",
"biology",
"techniques",
"plants",
"research",
"and",
"analysis",
"methods",
"mycology",
"genome",
"complexity",
"crop",
... | 2016 | Comparative Genomics of the Sigatoka Disease Complex on Banana Suggests a Link between Parallel Evolutionary Changes in Pseudocercospora fijiensis and Pseudocercospora eumusae and Increased Virulence on the Banana Host |
The chiasma is a structure that forms between a pair of homologous chromosomes by crossover recombination and physically links the homologous chromosomes during meiosis . Chiasmata are essential for the attachment of the homologous chromosomes to opposite spindle poles ( bipolar attachment ) and their subsequent segregation to the opposite poles during meiosis I . However , the overall function of chiasmata during meiosis is not fully understood . Here , we show that chiasmata also play a crucial role in the attachment of sister chromatids to the same spindle pole and in their co-segregation during meiosis I in fission yeast . Analysis of cells lacking chiasmata and the cohesin protector Sgo1 showed that loss of chiasmata causes frequent bipolar attachment of sister chromatids during anaphase . Furthermore , high time-resolution analysis of centromere dynamics in various types of chiasmate and achiasmate cells , including those lacking the DNA replication checkpoint factor Mrc1 or the meiotic centromere protein Moa1 , showed the following three outcomes: ( i ) during the pre-anaphase stage , the bipolar attachment of sister chromatids occurs irrespective of chiasma formation; ( ii ) the chiasma contributes to the elimination of the pre-anaphase bipolar attachment; and ( iii ) when the bipolar attachment remains during anaphase , the chiasmata generate a bias toward the proper pole during poleward chromosome pulling that results in appropriate chromosome segregation . Based on these results , we propose that chiasmata play a pivotal role in the selection of proper attachments and provide a backup mechanism that promotes correct chromosome segregation when improper attachments remain during anaphase I .
During cell division , chromosomes that harbor genetic information are accurately segregated into daughter cells . Chromosome segregation depends on attachment of chromosomes to the spindle via chromosomal sites called kinetochores . The interaction between kinetochores and spindle microtubules , which extend from opposite spindle poles , generates pulling forces on the chromosomes from opposite directions , causing them to migrate toward opposite spindle poles . To understand the mechanisms underlying chromosome segregation , it is crucial to elucidate how chromosomes attach to the spindle . In mitosis , sister chromatids are segregated to opposite poles ( equational segregation; Figure 1 ) . The sister chromatids are associated until anaphase via a protein complex called cohesin [1] , [2] , which is required for the back-to-back arrangement of the kinetochores that permits their attachment to opposite spindle poles [3] . In addition , when sister chromatids are pulled from opposite directions , the cohesion generates tension at the kinetochore that leads to stabilization of the kinetochore–microtubule interaction , probably via inactivation of aurora kinase [4] . When the cohesion is compromised , sister chromatids fail to attach to the spindle properly and are mis-segregated [5]–[8] . During meiosis , on the other hand , a physical association between homologous chromosomes additionally contributes to proper spindle attachment of chromosomes [3] , [9] , [10] . Meiosis occurs during gamete formation , and during meiosis , two rounds of chromosome segregation follow a single round of DNA replication , resulting in the production of gametes with half the original number of chromosomes . Chromosome segregation during meiosis I is specific to meiosis: Homologous chromosomes attach to opposite spindle poles , with each pair of sister chromatids attaching to the same pole ( monopolar attachment ) , and are segregated to the opposite poles ( reductional segregation; Figure 1 ) . As in mitosis , sister chromatid cohesion is required for proper kinetochore arrangement during meiosis . However , a meiosis-specific type of cohesin mediates this cohesion [11]–[15] , and sister kinetochores are arranged side by side facing the same direction so that they become attached to the same pole [16] . Furthermore , shugoshin proteins maintain centromeric cohesion during anaphase I [17]–[21] . These proteins inhibit the removal of centromeric cohesin and regulate centromeric aurora kinase [17]–[19] , [21]–[24] . Elimination of both of these functions compromises sister chromatid segregation during meiosis I and II [17] , [18] , [22] , [25] . Further , elimination of the cohesin-retention function alone causes sister chromatid separation after anaphase I but has little if any effect on sister chromatid segregation toward the same pole during anaphase I [17] , [19] . Unlike the situation in mitosis , homologous chromosome association contributes to the generation of tension at the kinetochore in meiosis . Homologous chromosomes are physically associated with each other via the chiasmata that are formed by reciprocal recombination . When homologous chromosomes are pulled in opposite directions , the chiasmata generate tension at the kinetochore and stabilize the kinetochore–microtubule interaction . Elimination of chiasmata leads to non-disjunction of homologous chromosomes [26] . In addition to this widely accepted role , chiasmata appear to play additional roles in the attachment of chromosomes to the spindle . A lack of chiasmata results in the separation or fragmentation of sister chromatids during meiosis I in many species [27]–[29] , suggesting that chiasmata prevent the bipolar attachment of sister chromatids . Furthermore , chiasmata greatly alter meiotic sister chromatid segregation patterns in several different types of fission yeast cells . Fission yeast cells normally undergo meiosis after responding to the mating pheromone [30] , but meiosis can also be induced without mating pheromone response by inactivation of Pat1 kinase , a key negative regulator of meiosis [31] , [32] . We previously reported that when haploid fission yeast cells lacking homologous chromosomes were forced to enter meiosis by Pat1 inactivation after a mating pheromone response , sister chromatids were primarily segregated to the same pole at meiosis I , as seen in normal diploid meiosis [33] . However , when they were induced to enter meiosis without a mating pheromone response , sister chromatids primarily underwent equational segregation . By contrast , when Pat1 inactivation forced diploid cells to enter meiosis without a mating pheromone response , the sister chromatids were primarily segregated to the same pole in a recombination-dependent manner . Similar recombination-dependent co-segregation of sister chromatids has been observed in several cohesin-related mutants of fission yeast [34] . These findings suggest that chiasmata promote the monopolar attachment of sister chromatids; however , because a loss of recombination causes only a negligible level of equational segregation during normal diploid meiosis in fission yeast cells , chiasmata have previously been thought to be dispensable for monopolar attachment of sister chromatids [35] . The contribution of chiasmata to the monopolar attachment during meiosis I , therefore , remains elusive . To understand the mechanisms underlying meiotic chromosome segregation , we examined the functions of chiasmata in spindle attachment and segregation of sister chromatids during meiosis I in fission yeast . Our analysis of chromosome segregation and dynamics in several different types of achiasmate cells showed that in the absence of chiasmata , sister chromatids were frequently attached to opposite poles during anaphase I . High time-resolution analysis of centromere dynamics further showed that chiasmata contribute to the elimination of bipolar attachments during the pre-anaphase stage . Furthermore , when the bipolar attachments remain during anaphase I , chiasmata induce a bias toward the proper pole during poleward chromosome pulling from opposite directions that results in correct chromosome segregation . Based on our findings , we discuss how chiasmata contribute to spindle attachment and segregation of chromosomes and further extend our idea to include the general functions of chromosome association during mitotic and meiotic chromosome segregation .
Elimination of chiasmata induced by depletion of Rec12 , a recombination factor required for the formation of double-strand breaks [36] , causes occasional equational segregation of sister chromatids [33] and frequent non-disjunction of homologous chromosomes [37] . As a first step toward understanding the role of chiasmata in the spindle attachment of sister chromatids , we re-examined chromosome segregation during meiosis I in more detail in rec12 mutant cells . We examined chromosome segregation by visualizing centromere-linked loci of chromosome I ( the lys1 locus: cen1 ) and chromosome II ( the D107 locus: cen2 ) using green fluorescent protein ( GFP ) [33] . After the first division , homologous centromeres were partitioned into two nuclei and rarely into the same nucleus in wild-type cells ( Figure 2A ) . In contrast , homologous centromeres were frequently partitioned into the same nucleus in rec12 mutant cells ( Figure 2A ) . Furthermore , sister centromeres were partitioned into the same nucleus and rarely into two nuclei in wild-type cells but were occasionally partitioned into two nuclei in rec12 mutant cells ( Figure 2B , + , rec+ and rec12Δ ) . The ∼4% of wild-type cells that showed a partition of cen1 into the distinct nuclei was most likely the result of recombination between the centromere and the lys1 locus used for this analysis . These results confirmed the mis-segregation of both homologous chromosomes and sister chromatids during meiosis I in rec12 mutant cells . The same mis-segregation phenotypes were also observed in cells lacking the Rec14 recombination factor , which functions together with Rec12 and the depletion of which eliminates recombination ( Figure 2A and 2B ) [38] , [39] . Segregation analysis showed that the overall mis-segregation frequency of sister chromatids in recombination-deficient , chiasmata-lacking cells ( i . e . , achiasmate cells ) was small . However , live cell analysis of cen2 dynamics suggested that improper spindle attachment of sister centromeres occurs more frequently during anaphase I . Although the sister centromeres eventually moved to the pole in rec12 mutant cells , they frequently remained between the two spindle poles and were dissociated during anaphase I [observed for 7 out of 14 centromeres examined ( 50 . 0% ) ; Figure 2C , rec12Δ] . These centromeres are called lagging centromeres , and they were not observed in wild-type cells [observed for 0 of 12 centromeres examined ( 0% ) ; Figure 2C , Wt] . The chromosome lagging is most likely caused by a loss of chiasmata and not a loss of Rec12 function , because lagging chromosomes were also frequently observed when meiosis was induced in haploid cells [33] , which do not form chiasmata due to their lack of homologous chromosomes ( Figure S1 ) . These results suggest that sister centromeres are frequently attached to both poles and are pulled from opposite directions during anaphase I in achiasmate cells . To confirm the frequent bipolar attachment of sister chromatids in achiasmate cells , we depleted Sgo1 , which inhibits the removal of centromeric cohesin during anaphase I [17] , [19] . We hypothesized that although sister chromatids are frequently attached to both poles and are pulled from opposite directions , the centromere cohesion that persists until meiosis II should provide resistance against this force and prevent their separation during anaphase I in achiasmate cells . If so , depletion of Sgo1 , which eliminates centromere cohesion during anaphase I , should lead to frequent equational segregation of sister chromatids . Indeed , Sgo1 depletion led to a substantial increase in equational segregation in achiasmate cells . Equational segregation of sister centromeres was occasionally observed in sgo1 mutant cells but was more frequently observed in sgo1 rec12 and sgo1 rec14 double-mutant cells ( Figure 2B , sgo1Δ ) . When meiosis I was induced in haploid cells , Sgo1 depletion similarly increased equational segregation ( Figure 2D ) irrespective of Rec12 depletion ( data not shown ) . Therefore , the increased equational segregation is not specific to recombination-deficient cells but is common in achiasmate cells . These results confirm that the loss of chiasmata frequently leads to the bipolar attachment of sister chromatids during anaphase I . The SAC ensures faithful chromosome segregation by delaying anaphase initiation until all of the chromosomes become properly attached to the spindle [40] , [41] . We previously reported that the SAC becomes activated to delay anaphase initiation at meiosis I in rec12 mutant cells , which is likely associated with improper spindle attachment of chromosomes [42] . Similarly , analysis of spindle length showed that anaphase initiation was substantially delayed in sgo1 rec12 double-mutant cells in a Mad2-dependent manner , as previously observed in rec12 mutant cells ( Figure S2A , Text S1 ) . Therefore , we next examined whether the SAC contributes to the bipolar attachment of sister chromatids in achiasmate cells by depleting the SAC factor Mad2 . Mad2 depletion led to decreased equational segregation of sister chromatids in rec12 mutant cells , but equational segregation was still observed at substantial levels in rec12 sgo1 double-mutant cells ( Figure 3A ) . Likewise , Mad2 depletion decreased but did not abolish equational segregation during meiosis I in haploid cells ( Figure 3B ) . These results showed that the SAC promotes the bipolar attachment of sister chromatids but is not essential for this process in achiasmate cells . Furthermore , as seen in rec12 mutant cells , sister centromeres frequently dissociated and failed to move to the pole during anaphase I in mad2 rec12 double-mutant cells [40 . 9% ( 22 centromeres ) ; Figure 3C , mad2Δ rec12Δ] . However , lagging centromeres were rarely observed in mad2 mutant cells [0 . 1% ( 20 centromeres ) ; Figure 3C , mad2Δ] , although the timing of anaphase initiation was not much different between these mutants ( Figure S2A ) . These results indicated that the lagging centromeres seen in achiasmate cells were not caused by SAC activation or delayed anaphase initiation . Thus , we conclude that the bipolar attachment of sister chromatids depends only partially on the SAC in achiasmate cells . Spindle attachment of chromosomes is established before anaphase , and the chiasma may prevent the bipolar attachment of sister chromatids from occurring during the pre-anaphase stage . To test this possibility , we examined the dynamics of sister centromeres before anaphase by time-lapse analysis with 10-s intervals . The time-lapse analysis of cen2 loci on both homologous chromosomes in wild-type and rec12 mutant cells confirmed our previous observations from time-lapse analyses with 1-min intervals , although they exhibited slight differences in dynamic parameters ( Table S1 ) [42] . Homologous centromeres oscillated between the two spindle poles in a somewhat coordinated manner in wild-type cells; a pair of homologous centromeres often moved in the same direction ( Figure 4A , Table S2 ) . Accordingly , centromeres were mostly positioned around the middle point between the spindle pole and the spindle center with a tendency to be near the center ( Figure 4B ) . These centromere dynamics presumably reflect the frequent bipolar attachment of homologous chromosomes that are linked by the chiasmata ( Figure 4C ) . On the other hand , sister centromeres oscillated in an uncoordinated manner and tended to remain near the pole in rec12 mutant cells ( Figure 4A ) , and centromere positioning was shifted toward the pole ( Figure 4B ) . These centromere dynamics probably reflect the frequent attachment of each of the non-linked homologous chromosomes to one pole and the occasional switch in their attachment to the other pole ( Figure 4C ) . Notably , we found that sister centromeres occasionally underwent a transient dissociation in both wild-type and rec12 mutant cells ( Figure 4A and 4D , Table 1 ) . In both types of cells , centromere dissociation was observed in ∼20% of events on average ( Figure 4E ) . This dissociation was not the result of the integration into the chromosome of lacO repeats , which are used for visualization [33] , or of the dissociation of only the visualized pericentromeric region; when all three homologous sets of sister centromeres were visualized by GFP tagging of the centromere-specific histone H3 variant Cnp1 [43] , we observed more than six centromere signals together with a transient split of the signal into two ( Figure 4F ) . These observations showed that bipolar attachment of sister chromatids occasionally occurs during the pre-anaphase stage , irrespective of chiasma formation . Similar centromere dynamics were also observed in cells lacking Sgo1 . The occurrence of bipolar attachment in the presence of chiasmata is contradictory to the idea that chiasmata prevent the bipolar attachment of sister chromatids from occurring during the pre-anaphase stage . If chiasmata do not prevent the bipolar attachment of sister chromatids from occurring , they must contribute to the elimination of bipolar attachment of sister chromatids during the pre-anaphase stage . However , the overall frequency of centromere dissociation was not significantly different between wild-type and rec12 mutant cells ( Figure 4E , Table 1 ) , and chiasma-dependent elimination of the bipolar attachment was not evident . We hypothesized that if sister centromeres attach to both poles more frequently in the achiasmate background , the chiasma-dependent elimination of the bipolar attachment would be evident . Following this hypothesis , we examined mrc1 and moa1 mutant cells . The mrc1 gene encodes a conserved DNA replication checkpoint factor , which delays cell cycle progression upon DNA replication stress , promotes proper fork progression , and contributes to sister chromatid cohesion in mitosis [44]–[50] . On the other hand , the moa1 gene encodes a meiosis-specific centromere protein that contributes to the proper centromere localization of the meiotic cohesin component Rec8 [34] . In both mrc1 and moa1 mutant cells , chromosome segregation as well as spindle dynamics , recombination , and spore formation are largely normal ( Figures S2B and S3 , Text S1 ) [34] . However , sister chromatids are primarily segregated equationally in a manner partly dependent on Mad2 when chiasmata are not formed ( in the rec12Δ or the haploid background; Figure 5 ) [34] . Although these phenotypes are similar to the sgo1-mutant phenotypes , the equational segregation is primarily caused by defects in centromere features other than maintenance of centromere cohesion , because both mrc1 and moa1 mutant cells can maintain sister centromere cohesion until anaphase II if sister chromatids are not segregated equationally during meiosis I ( Figure S3D , Text S1 ) [34] . Therefore , the equational segregation seen in the mrc1 rec12 and moa1 rec12 mutant cells is likely to be caused by frequent bipolar attachment of sister chromatids , and we expected that the chiasma effects would be more evident in the mrc1 and moa1 mutants . To evaluate chiasma effects in the mrc1 and moa1 mutants , we first examined the pre-anaphase centromere dynamics in the achiasmate mrc1 rec12 and moa1 rec12 double-mutant cells . In the mrc1 rec12 mutant cells , the sister centromeres dissociated more frequently ( Figure 6A and 6B ) , with a significantly longer duration ( Table 1 ) , and were predominantly positioned around the spindle center , unlike those in the rec12 mutant cells ( Figure 6C ) . In the moa1 rec12 mutant cells , the centromeres were also frequently positioned around the spindle center ( Figure 6A and 6C ) , and in addition , the SAC was not activated as much as in rec12 mutant cells ( Figure S2A , Text S1 ) . These characteristics were expected to be associated with frequent bipolar attachment of sister chromatids ( Figure 6D ) . Indeed , the frequent dissociation of the centromeres and their positioning around the spindle center together with the low level of SAC activation were observed during meiosis I in achiasmate rec8 mutant cells ( Figure 6A–6C and Figure S2A , Table 1 ) , in which sister chromatids efficiently attach to both poles to fully undergo equational segregation [12] , [51] . They were also observed during mitotic division in wild-type diploid cells ( Figure S4 ) . These observations thus confirmed that sister centromeres attach to both poles more frequently in the mrc1 rec12 and moa1 rec12 double-mutant cells than in rec12 single-mutant cells . However , the centromere properties of the mrc1 and moa1 mutant cells differed from those of rec8 mutant or mitotic cells because the SAC substantially delayed anaphase initiation in mrc1 rec12 mutant cells ( Figure S2A , Text S1 ) , and centromere dissociation was not so frequent in moa1 rec12 mutant cells ( Figure 6B ) . We next examined the pre-anaphase centromere dynamics in the chiasmate mrc1 and moa1 single-mutant cells to evaluate chiasma effects . Remarkably , in mrc1 single-mutant cells , the level of centromere dissociation was almost identical to that in wild-type cells ( Figure 6A and 6B , Table 1 ) , indicating that bipolar attachment of sister chromatids was reduced to a wild-type level . Furthermore , centromere positioning and the distance between homologous centromeres were very similar to what was seen in wild-type cells ( Figure 6C and 6E ) , indicating that homologous chromosomes attach to both poles as frequently as in wild-type cells . These results show that chiasmata eliminate the bipolar attachment of sister chromatids and promote the bipolar attachment of homologous chromosomes during the pre-anaphase stage in mrc1 mutant cells . On the other hand , in moa1 mutant cells , centromere positioning and dissociation were not significantly different from those seen in achiasmate moa1 rec12 mutant cells ( Figure 6A–6C , Table 1 ) . Furthermore , homologous centromeres were not separated as widely as in wild-type cells ( Figure 6E ) . These results indicate that sister chromatids still attach to both poles at a level similar to that in moa1 rec12 mutant cells and pulling forces are not properly exerted on homologous chromosomes in moa1 mutant cells ( Figure 6E ) . Therefore , chiasmata fail to eliminate the bipolar attachment of sister chromatids during the pre-anaphase stage in moa1 mutant cells . Because the bipolar attachment of sister centromeres did not appear to be eliminated during the pre-anaphase stage in chiasmate moa1 mutant cells , we examined whether their bipolar attachment is retained during anaphase by analyzing anaphase centromere dynamics . In wild-type cells , sister centromeres moved swiftly toward the poles ( all 13 of the centromeres examined reached the poles within 130 s; Figure 7 ) and only occasionally dissociated during anaphase I [only three centromeres out of 13 ( 23 . 1% ) were dissociated; Figure 7 , Wt , lower panel] . The centromeres also moved swiftly to the pole and remained associated in mrc1 mutant cells ( all 11 centromeres examined reached the pole within 80 s without dissociation; Figure 7 ) . In contrast , in moa1 mutant cells , lagging and dissociation of centromeres were frequently observed during anaphase [10 out of 14 centromeres ( 71 . 4% ) failed to reach the poles within 130 s , unlike wild-type centromeres , and 5 of them ( 35 . 7% ) failed to reach the poles within 300 s; 6 centromeres ( 42 . 9% ) were dissociated; Figure 7] . Furthermore , elimination of anaphase centromere cohesion by Sgo1 deletion substantially increased the equational segregation of sister chromatids ( Figure 5B ) . These results showed that sister chromatids were frequently attached to both poles and pulled from opposite directions during anaphase I in moa1 mutant cells . Surprisingly , most of the lagging centromeres eventually moved to the proper pole ( Figure 5A and 5B , Figure 7 ) . This result indicates that although sister chromatids were pulled from opposite directions during anaphase , they were pulled toward the proper pole more strongly and/or continuously than they were pulled toward the improper pole in the chiasmate moa1 mutant cells . Therefore , the chiasma generates a bias toward the proper pole in poleward chromosome pulling from opposite directions that eventually results in proper chromosome segregation in moa1 mutant cells .
In the current study , we examined the role of chiasmata by analyzing the segregation and dynamics of chromosomes during meiosis I induced in recombination-deficient diploid cells and in haploid cells . The analysis of these two distinct types of achiasmate cells provided two lines of evidence to show that sister chromatids frequently attach to both poles and experience pulling forces from opposite directions during anaphase I in achiasmate cells . First , sister centromeres frequently became transiently dissociated and/or failed to move to the pole during anaphase I ( Figure 2C and Figure S1 ) . Second , when sister centromere cohesion was resolved during anaphase by Sgo1 depletion , sister chromatids frequently underwent equational segregation during anaphase I ( Figure 2B and 2D ) . Chiasmata therefore play a crucial role in preventing the bipolar attachment of sister chromatids during anaphase I . Because the bipolar attachment of sister chromatids has been observed during anaphase I in various achiasmate organisms [27]–[29] , it is probably common among eukaryotes . We further examined how chiasmata prevent the bipolar attachment of sister chromatids . Loss of chiasmata causes activation of the SAC [42] . However , we showed that the bipolar attachment of sister chromatids depends only partially on the SAC in achiasmate cells . The reduction of the bipolar attachment that normally generates tension in the achiasmate background is consistent with the idea that the SAC promotes attachments that generate tension [40] , [41] . We performed high time-resolution analysis of pre-anaphase centromere dynamics in several different types of chiasmate and achiasmate cells to understand how chiasmata contribute to the attachment . From this analysis , we have reached three conclusions . First , chiasmata cannot prevent occurrence of bipolar attachment of sister chromatids , based on the observation that the bipolar attachment occasionally occurred in chiasmate wild-type cells . Second , analysis of mrc1 mutant cells showed that chiasmata contribute to the elimination of the bipolar attachment of sister chromatids during the pre-anaphase stage ( Figure 8A ) . However , the elimination was not evident in wild-type cells in comparison with rec12 mutant cells . One possible explanation for this result is that the bipolar attachments occur more frequently in wild-type than in rec12 mutant cells because the centromere is positioned closer to the spindle center in wild-type cells ( Figure 4B ) . Alternatively , chiasmata may eliminate bipolar attachments in mrc1 mutant cells but not in wild-type cells because of distinct centromere structures or functions . Furthermore , we cannot completely exclude the possibility that the chiasmata-dependent elimination depends in part on unknown Rec12 functions . Third , analysis of moa1 mutant cells showed that chiasmata induced a bias toward the proper pole in poleward chromosome pulling from opposite directions that resulted in proper chromosome segregation ( Figure 8B ) . In moa1 mutant cells , sister centromeres were frequently pulled from opposite directions and dissociated during anaphase I , but they were pulled toward the proper pole more strongly and/or continuously than they were pulled toward the improper pole , and eventually moved to the appropriate pole . We also observed this chiasma effect , albeit occasionally , in wild-type cells ( Figure 7 , Wt , lower panel ) and thereby speculate that the chiasma-induced bias is a backup mechanism that ensures proper meiotic chromosome segregation even when improper attachments remain . How the chiasmata eliminate bipolar attachments and induce a bias in chromosome pulling remains elusive . Because chiasmata are essential for generating the tension that stabilizes kinetochore–microtubule interactions and increases kinetochore microtubules [9] , [52] , we speculate that chiasmata execute these different tasks via tension , as follows ( see also Text S1 ) . In wild-type cells , sister kinetochores occasionally attach to both poles ( Figure S5A ) . In the presence of chiasmata , microtubules that attach to the proper poles generate sufficient tension , but those that attach to improper poles probably do not . As a result , improper attachments are eliminated while proper attachments are increased . Even when improper attachments are not eliminated , the increase in proper attachments presumably promotes the exertion of segregation forces in the appropriate direction ( a similar scenario is shown in Figure S5A , rec12Δ ) . In contrast , improper attachments are not eliminated in rec12 mutant cells , possibly because the improper attachments also generate tension ( Figure S5A ) . In this model , chiasmata must prevent improper attachments from generating tension . During the pre-anaphase stage , chromosomes oscillate between the poles , and oscillation of the chiasma-linked chromosomes may reduce tension ( Figure S5B ) . When a pair of sister chromatids follows the other homologous pair that is moving toward the spindle pole , the leading sister chromatid pair presumably exerts pulling forces on the chromosome arms of the following pair via chiasmata . These pulling forces are likely to reduce the tension that improper attachments generate but not those generated by proper attachments . As a result , only proper attachments ( i . e . , bipolar attachment of the homologous chromosomes ) become stable and persist , whereas improper attachments ( i . e . , bipolar attachment of sister chromatids ) do not . Alternatively , the chiasmata-dependent pulling may make the kinetochores on the following chromosomes face the side opposite the direction of chromosome movement to physically eliminate improper attachments . Although the above model can account for the observed chiasmata-dependent effects , we cannot completely rule out the possibility that chiasmata directly contribute to centromere function or structure to affect spindle attachment and segregation of chromosomes . Chiasmata eliminated bipolar attachment of sister chromatids in the mrc1 mutant but did not eliminate it in the moa1 mutant . Distinct kinetochore arrangements may account for this difference ( Figure S5A , Text S1 ) . Given the frequent monopolar attachment of sister chromatids in the chiasmate mrc1 single-mutant cells together with the substantial SAC activation in achiasmate mrc1 rec12 double-mutant cells , sister kinetochores probably face the same side in mrc1 mutants . However , the frequent bipolar attachment of sister chromatids seen in mrc1 rec12 mutant cells conversely implies that the kinetochores face opposite sides . This contradiction may be explained by the flexibility of the kinetochore arrangement ( Figure S5A , Text S1 ) . It is possible that in the mrc1 mutant cells , although sister kinetochores are initially arranged side by side , the kinetochores end up facing opposite sides when they are pulled from opposite directions , leading to the subsequent efficient bipolar attachment of sister centromeres . On the other hand , in moa1 mutant cells , sister kinetochores perhaps face opposite sides to attach to both poles efficiently ( Figure S5A , Text S1 ) , as proposed previously [34] . Although kinetochore arrangement was previously proposed to be flexible in moa1 mutant cells [34] , we speculate that the arrangement is conversely inflexible because of strong centromere cohesion , considering increased centromere accumulation of cohesin [34] , infrequent sister centromere dissociation ( Figure 6B ) , and a narrower dissociation distance ( Figure S6 ) . Bipolar attachment was not eliminated in moa1 single-mutant cells , perhaps because bipolar attachment is easily re-established due to the back-to-back kinetochore arrangement . An alternative possibility is that moa1 mutant cells are defective in destabilizing the kinetochore–microtubule interaction and fail to eliminate improper attachments efficiently . Our findings have three important implications for understanding the mitotic chromosome segregation mechanism . First , the frequent bipolar attachment of sister chromatids seen in achiasmate cells indicates that kinetochore arrangement alone cannot prevent improper attachments and suggests that bipolar ( merotelic ) attachment of a single chromatid also occurs when sister chromatid cohesion is defective . Indeed , Courtheoux et al . recently reported that merotelic attachments occur during mitotic anaphase in rad21 fission yeast mutants defective in sister chromatid cohesion [53] . Furthermore , a lagging chromatid was frequently observed during anaphase II in sgo1 mutant of fission yeast , in which sister chromatids undergo precautious dissociation before anaphase II [17] . These observations may alter the interpretation of phenotypes associated with monopolin and heterochromatin mutants of fission yeast , which were proposed to be defective in the arrangement of microtubule-binding sites of kinetochores because these mutants frequently exhibited merotelic attachments during mitotic anaphase [54] , [55] . However , defective sister centromere cohesion in the monopolin and heterochromatin mutants may have caused the merotelic attachments [56]–[58] . Second , the fact that sister chromatids , despite their bipolar attachment , move to the same pole in chiasmate cells indicates that monopolar attachment of sister chromatids is not a prerequisite for their proper segregation . This feature is probably common during mitotic chromosome segregation because the proper segregation of a single chromatid that is attached to both poles has also been observed in higher eukaryotes during mitosis [59] . Therefore , generation of bias in the segregation forces is probably a general mechanism that ensures correct chromosome segregation . Finally , the chromosome oscillation-dependent model for the elimination of improper attachments may also account for the establishment of proper attachments during mitosis ( Figure S5B ) . During mitosis , chromosomes oscillate during the establishment of their spindle attachment ( Figure S4A ) [60] , [61] , and merotelic attachment occurs in higher eukaryotes [62] . Furthermore , in fission yeast , the physical linkage between two kinetochores induces their bipolar attachment during mitosis [63] . These facts suggest that the oscillation of cohesin-linked sister chromatids destabilizes improper attachments and contributes to the selection of proper attachments during mitosis . In summary , we have shown that chiasmata are essential for proper spindle attachment and segregation of sister chromatids during meiosis I . Based on our results , we propose that chiasmata play a pivotal role in the selection of proper attachments and establish a backup mechanism that promotes the appropriate segregation of chromosomes when improper attachments remain during anaphase I . Furthermore , we propose a model to explain how chromosome association contributes to correct spindle attachment of the chromosomes not only in meiosis but also in mitosis . Our findings increase understanding of the general mechanisms of chromosome segregation and contribute to knowledge about the mechanisms that underlie the chromosome mis-segregation associated with birth defects and/or tumorigenesis in humans .
Table S3 lists the yeast strains used in this study , and strains used in figures are described in Text S1 . Media used in this study have been described by Moreno et al . [64] . Yeast strains were grown on solid YES medium at 30°C . For the segregation analyses of homologous chromosomes , two types of cells , both of which contained GFP-labeled centromeres ( cen2 or lys1 ) , were crossed on solid ME medium . For sister chromatid segregation analyses , cells containing GFP-labeled centromeres were crossed with cells lacking GFP-labeled centromeres . The resulting diploid cells were then induced to enter meiosis by incubation at 25°C for 16–18 h . Nuclear DNA in meiotic zygotes was stained with the DNA-specific dye , Hoechst 33342 , as described [65] . GFP signal was examined in zygotes containing two round DNA masses that underwent meiosis I . Zygotes containing two DNA masses with a tear-drop shape and pointed ends facing each other were excluded because they were in the karyogamy stage . Haploid yeast cells were forced to enter meiosis by Pat1 inactivation following activation of the mating pheromone signaling pathway , as previously described [33] . Haploid pat1 temperature-sensitive mutant cells bearing the c-type mat gene of the opposite mating type , which is required for activation of the mating pheromone signaling pathway , were grown in YES-rich medium to a density of 3–5×106 cells/ml at 25°C . The cells were suspended in an equal volume of EMM2 medium lacking a source of nitrogen ( EMM2-N ) and incubated at 25°C for 14–16 h to synchronize the cells in G1 phase and activate the mating pheromone signaling pathway . The cells were resuspended in fresh EMM2-N medium and induced to enter meiosis by further incubation at 34°C . Meiotic progression was monitored by analysis of chromosomal DNA morphology at 1-h time intervals . Sister chromatid segregation was analyzed in cells containing two DNA masses that underwent meiosis I . The chromosome locus and spindle poles were visualized using the lacI/lacO recognition system and the GFP-tagged spindle pole component Sid4 , respectively , as described previously [42] . Cells were grown on solid YES medium at 30°C and induced to undergo meiosis by incubation on solid ME medium at 25°C for 16–18 h . The cells were observed to determine the dynamics of the GFP-labeled spindle pole or chromosome locus at 25°C using a DeltaVision microscope system ( Applied Precision Inc . ) equipped with a 60X/1 . 42 numerical aperture Plan Apo oil-immersion objective lens ( Olympus ) , as described previously [65] . The behavior of the GFP-labeled chromosome locus was observed every 1 min or 10 s . A set of images from six focal planes with 0 . 5-µm intervals or ten focal planes with 0 . 3-µm intervals was taken at each time point for 1-min or 10-s time-lapse analysis , respectively . Behavior of the GFP-tagged Cnp1 was observed in a manner similar to the 10-s time-lapse analysis of the GFP-labeled chromosome locus , except that a 100X/1 . 4 numerical aperture Plan Apo oil-immersion objective lens ( Olympus ) was used . All measurements were conducted in three dimensions . | Gametes form through a special type of cell division called meiosis . During meiosis , two nuclear divisions take place successively; the first division is specific only to meiosis , in that homologous chromosomes segregate from each other . Homologous chromosome segregation requires physical association of the homologous chromosomes by a structure called chiasma that forms at the site of recombination . This association is thought to contribute to proper attachment of homologous chromosomes to the spindle , leading to their proper segregation . In this study , we examined the functions of chiasmata during the first division in fission yeast by analyzing chromosome dynamics and segregation in several different mutants lacking chiasmata . We found that , in addition to proper spindle attachment of homologous chromosomes , chiasmata contribute to proper spindle attachment of replicated chromosomes more substantially than previously had been thought . In addition , even when chromosomes are improperly attached to the spindle , chiasmata eventually cause proper chromosome segregation . Our findings reinforce the significance of the physical association of homologous chromosomes in proper spindle attachment of chromosomes and have unveiled a previously unidentified , chiasma-dependent mechanism that ensures proper chromosome segregation . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"molecular",
"biology/centromeres",
"genetics",
"and",
"genomics/chromosome",
"biology",
"cell",
"biology/cytoskeleton"
] | 2011 | Chiasmata Promote Monopolar Attachment of Sister Chromatids and Their Co-Segregation toward the Proper Pole during Meiosis I |
Little is known about the presence/absence and prevalence of Rickettsia spp , Bartonella spp . and Yersinia pestis in domestic and urban flea populations in tropical and subtropical African countries . Fleas collected in Benin , the United Republic of Tanzania and the Democratic Republic of the Congo were investigated for the presence and identity of Rickettsia spp . , Bartonella spp . and Yersinia pestis using two qPCR systems or qPCR and standard PCR . In Xenopsylla cheopis fleas collected from Cotonou ( Benin ) , Rickettsia typhi was detected in 1% ( 2/199 ) , and an uncultured Bartonella sp . was detected in 34 . 7% ( 69/199 ) . In the Lushoto district ( United Republic of Tanzania ) , R . typhi DNA was detected in 10% ( 2/20 ) of Xenopsylla brasiliensis , and Rickettsia felis was detected in 65% ( 13/20 ) of Ctenocephalides felis strongylus , 71 . 4% ( 5/7 ) of Ctenocephalides canis and 25% ( 5/20 ) of Ctenophthalmus calceatus calceatus . In the Democratic Republic of the Congo , R . felis was detected in 56 . 5% ( 13/23 ) of Ct . f . felis from Kinshasa , in 26 . 3% ( 10/38 ) of Ct . f . felis and 9% ( 1/11 ) of Leptopsylla aethiopica aethiopica from Ituri district and in 19 . 2% ( 5/26 ) of Ct . f . strongylus and 4 . 7% ( 1/21 ) of Echidnophaga gallinacea . Bartonella sp . was also detected in 36 . 3% ( 4/11 ) of L . a . aethiopica . Finally , in Ituri , Y . pestis DNA was detected in 3 . 8% ( 1/26 ) of Ct . f . strongylus and 10% ( 3/30 ) of Pulex irritans from the villages of Wanyale and Zaa . Most flea-borne infections are neglected diseases which should be monitored systematically in domestic rural and urban human populations to assess their epidemiological and clinical relevance . Finally , the presence of Y . pestis DNA in fleas captured in households was unexpected and raises a series of questions regarding the role of free fleas in the transmission of plague in rural Africa , especially in remote areas where the flea density in houses is high .
The importance of fleas in human and animal health is largely related to their ability to transmit agents of infectious diseases [1] . The transmission of these zoonotic agents to human occurs mainly through bites or inoculation of feces into pruritic bite lesions [2] , [3] . Plague is the most notorious flea-borne disease known to man and is a reemerging public health issue mainly in Africa and South America [3] . The etiological agent of plague , Yersinia pestis , is a facultative gram-negative bacterium restricted nowadays to well defined endemic foci [4] , [5] . In the last decade , plague reemerged in old quiescent foci of Algeria [6] , the United Republic of Tanzania [7] and Libya [8] and caused remarkable bubonic and pneumonic outbreaks in known endemic foci in Madagascar [9] and in the Democratic Republic of the Congo [10] . Fleas are also associated with other bacterial diseases such as bartonelloses and rickettsioses . Rickettsia spp . , the etiological agents of rickettsioses , are intracellular gram-negative bacteria that represent an emergent global threat , particularly in the tropics [11] . R . felis , an emerging pathogen , and R . typhi , the agent of murine typhus ( MT ) , are the main rickettsial pathogens associated with fleas [1] , belonging to the spotted fever group ( SFG ) [12] and typhus group rickettsiae , respectively [13] . Although these two flea-borne rickettsiae are distributed worldwide , R . typhi appears to be more endemic in tropical regions , coastal areas and ports , where its transmission cycles between rats ( Rattus spp . ) and oriental rat fleas ( X . cheopis ) [14] . Also , several closely related rickettsiae , referred as Rickettsia felis–like organisms ( RFLO ) , identified in fleas and other arthropods around the world [15] . Likewise , bartonelloses are diseases caused by the fastidious , hemotropic bacteria of the genus Bartonella , especially in debilitated and immunocompromised individuals [16] . Importantly , the list of host species harboring Bartonella spp . includes a large number of mammals , mostly rodents , some of which are kept as pets [17] . An increasing number of papers have reported the occurrence of fleas and human flea-borne infections , especially in relation to wildlife and zoonotic risk . However , the identity and distribution of flea-borne agents in urban , domestic or peridomestic settings have been poorly documented in Sub-Saharan African countries such as the Democratic Republic of the Congo , the United Republic of Tanzania and Benin . Historical data about human infection with Rickettsia and Bartonella species are fragmentary , and virtually nothing is known about the current distribution of these flea-borne zoonotic agents in potential vectors and reservoir hosts in these countries . In the Democratic Republic of the Congo , recent small-scale surveys have reported serological evidence for Bartonella infection in human patients [18] and molecular data in rodents [19] and fleas [20] , suggesting a global underreporting at the country scale . Rickettsioses in humans are mentioned in historical reports; however , their notification remains anecdotal , and the species identification is likely erroneous . Recently though , among febrile military patients in Kisangani , Democratic Republic of the Congo , one patient tested positive in 1999 , for the R . typhi antigen using serological tools . In addition , R . felis has been found to circulate in arthropod vectors in Kinshasa [21] . As a general trend , flea-borne agents in fleas are underreported , whereas in the United Republic of Tanzania , a growing number of publications confirm their presence and wide distribution in humans [22] exposed to their bites and in infested rodents [19] . In recent years , our laboratory ( Unité de Recherche sur les Maladies Infectieuses et Tropicales , the WHO Collaborative Centre for Rickettsial Diseases and Other Arthropod-Borne Bacterial Diseases in Marseille , France ) initiated collaboration with correspondents and universities in the United Republic of Tanzania , the Democratic Republic of the Congo and Benin . The present survey pursued the objectives of detecting the presence and identity of Rickettsia spp . , Bartonella spp . and Y . pestis in flea specimens collected from domestic and peridomestic areas in the Democratic Republic of the Congo , the United Republic of Tanzania and Benin within the context of entomological studies .
Risk assessment was submitted to and approved by the ethical committee and decision board of each institution involved in small mammals trappings , and involved informed consent of the domestic animal owners; ethical approval are available from original publications on mammal hosts on which flea were collected [19] , [23] , [24] . The Ethical commitee of the University of Antwerp , Belgium and the Sokoine University of Agriculture Morogoro under the project RATZOOMAN granted by the European Commission Framework 5 Programme on International Cooperation , project contract number ICA4 CT 2002 10056 , approved the experiment in the South-eastern Africa . See here technical annex: http://projects . nri . org/ratzooman/docs/technical%20annex . pdf . The material analyzed consisted of fleas ( Siphonaptera ) collected in domestic and peridomestic areas in Benin , the United Republic of Tanzania and the Democratic Republic of the Congo ( Figure 1 ) . A portion of the collected fleas was used for the present study . A convenient sample was selected according to a good representation of species , host and localities . In 37 sites in the capital city of Benin , Cotonou ( 6°21′36″N; 2°26′24″E ) , rodent fleas were collected from rodents trapped monthly inside human residences and peridomestic areas between November 2009 to July 2010 , as described previously [24] . In the United Republic of Tanzania , 17 sites in the Lushoto district ( 04°40′00″S 38°19′00″E ) located in the Tanga Region were surveyed [23] , [25] . Lushoto district is a mountainous area where plague was reported from the first time in 1981; this endemic plague focus has however been quiescent since 2004 . Between May 2005 and November 2008 , fleas were collected – as in Benin – from small mammals in domestic and peridomestic habitats during the dry and rainy seasons . Further details on the rodent measurements and flea collection have been published elsewhere [23] , [25] . Finally , in March and April 2007 , rodent fleas and free domestic fleas were collected from 4 villages ( 15 capture sites ) in the Linga and Rethy health zones , Ituri district , Orientale Province , the Democratic Republic of the Congo; off-host fleas were collected in 4 villages during an investigation following a plague outbreak that occurred in the third trimester of 2006 [26] . Our investigation area was limited to Djalusene ( 2°12′10″5 N 30°88′02″7 E ) and Kpandruma ( 2°05′90″1 N 30°88′70″4 E ) , which had confirmed plague patients , and Wanyale ( 2°10′11″8 N 30°80′60″5 E ) and Zaa ( 2°14′03″2 N 30°85′65″9 E ) , which had several suspect cases but were considered control areas at the time of the study . We collected fleas in 40 houses ( bedroom ) in each village , for 3 nights in a row , using a kerosene lamp hung above a 45-cm diameter tray containing water as described in [27] . In April 2010 and July 2012 , additional flea samples were collected from the Ituri district in Rethy village ( 1°50′N 29°30′E ) and in Kinshasa ( 4°19′19″S 15°19′16″E ) by means of light traps in human residences ( bedroom ) and rodent burrows , and flat tweezers on dogs . All fleas collected in Benin , the United Republic of Tanzania and the Democratic Republic of the Congo were stored in 70% ethanol and identified morphologically using classical entomologic taxonomic keys . The samples were later processed in the WHO Collaborative Center for Rickettsial Diseases and Other Arthropod-Borne Bacterial Diseases , in Marseille , France . Fleas were rinsed twice in distilled water for 10 minutes and dried on sterile filter paper; the handling was performed in a laminar flow biosafety hood . The fleas were individually crushed in sterile Eppendorf tubes , as described [28] . A total of 50 µl of DNA was extracted from one half of each flea using the QIAamp Tissue Kit ( Qiagen , Hilden , Germany ) by QUIAGEN-BioRobot EZ1 , according to the manufacturer's instructions . The genomic DNA was stored at −20°C under sterile conditions until used as the template in PCR assays . The remaining portion of each flea was kept at −80°C for an additional control . All samples were screened by quantitative real-time PCR ( qPCR ) targeting the biotin synthase ( bioB ) gene , as previously described [29] . Positive results were confirmed by another qPCR targeting a membrane phosphatase gene with primers ( Rfel_phosp_MBF: 5′-GCAACCATCGGTGAAATTGA-3′ and Rfel_phosp_MBR: 5′-GCCACTGTGCTTCACAAACA-3′ ) and a probe ( Rfel_phosp_MBP: 6FAM-CCGCTTCGTTATCCGTGGGACC-TAMRA ) designed in our laboratory . The final mixture of the qPCR reaction was composed of 15 µL of mix that contained 10 µL of master mix QuantiTect Probe PCR Kit ( QIAGEN , Hilden , Germany ) , 0 . 5 µL ( 20 pmol ) of each primer , 0 . 5 µL ( 62 . 5 nmol ) of probe , 3 . 5 µL of RNase DNase-free water and 5 µL of DNA extracted from fleas . qPCR was performed as follows: 15 min at 95°C , followed by 40 cycles of 1 s at 95°C , 40 s at 60°C and 40 s at 45°C , as described [29] . The negative control consisted of DNA extracted from uninfected fleas from our laboratory colony and was used for all the PCR assays in this work . The positive control was DNA extracted from a diluted strain of R . felis from our laboratory in Marseille . Positive results were recorded if the cycle threshold ( Ct ) value obtained was lower than 36 using the 2 PCR systems . Samples were screened by qPCR targeting a fragment of the Rpr 274P gene coding for a hypothetical protein , as described previously [30] . Positive results were confirmed by qPCR targeting the glycosyltransferase gene using a previously described Rpr 331 system [31] . qPCR was conducted using the same method as described for R . felis detection . The positive control was DNA extracted from a diluted strain of R . typhi Wilmington ( ATCC VR-144 ) cultured in our laboratory in Marseille . DNA samples were screened by quantitative real-time PCR targeting the ITS region [32] . Positive samples with ITS primers were then confirmed by standard PCR performed with Bartonella-specific primers for the citrate synthase ( gltA ) gene , amplifying an approximately 334-bp fragment [33] . The positive control was B . alsatica strain IBS 382 ( CIP 105477 ) DNA extracted from a strain and previously diluted to 10−6 . The success of PCR amplification was verified by 2% agarose gel migration . The products were purified using NucleoFast 96 PCR plates ( Machery-Nagel EURL , France ) as recommended by the manufacturer . The purified PCR products were sequenced with gltA primers using the BigDye version 1 . 1 cycle sequencing ready reaction mix ( Applied Biosystems , Foster City , CA ) with the ABI 31000 automated sequencer ( Applied Biosystems ) . The sequences were assembled and analyzed with the ChromasPro program ( version 1 . 5 ) . DNA samples were screened by qPCR targeting the plasminogen activator gene ( Pla ) [6] using primers Yper_PLA_F ( 5′-ATG-GAG-CTT-ATA-CCG-GAA-AC-3′ ) and Yper_PLA_R ( 5′-GCG-ATA-CTG-GCC-TGC-AAG-3′ ) and probe Yper_PLA _P ( 6- FAM-TCC-CGA–AAG-GAG-TGC-GGG-TAA-TAGG-TAMRA ) . Positive results were confirmed with standard PCR targeting the glpD gene , as described [34] , and then sequenced using the same method used for Bartonella spp . sequencing . The positive control was Y . pestis DNA extracted from the CSUR P 100 strain , and diluted to 10−6 .
In Benin , 886 fleas were collected from 199 sexually mature small mammals of four species , namely , Crocidura olivieri ( 17/199 , 8 . 5% ) , Mastomys natalensis ( 36/199 , 18% ) , Rattus norvegicus ( 40/199 , 20 . 1% ) and Rattus rattus ( 109/199 , 54 . 7% ) . Three flea species were collected from rodents , with the oriental rat flea X . cheopis being the most abundant ( 861/886 , 97 . 1% ) , followed by X . brasiliensis ( 24/886 , 2 . 7% ) and Ct . felis strongylus ( 1/886 , 0 . 1% ) . In the present study , a convenient sample of 199 X . cheopis ( picked off Rattus rattus ) individuals – 55 . 78% females and 44 . 2% males – were selected for an initial molecular screening ( the remaining fleas were preserved for subsequent studies ) . All fleas tested negative for R . felis and Y . pestis . qPCR performed for the detection of R . typhi revealed 2 positive X . cheopis ( 2/199 , 1% ) , with a Ct of 32 . 6 and 34 . 5 , from 2 sites ( Bokossi Tokpa and Dédokpo ) . Bartonella spp . were detected in 69/199 ( 34 . 6% ) of the fleas ( Ct , 31 . 81 , +/−2 . 97 ) ( 24≤Ct≤35 ) collected from all studied sites ( Table 1 ) . DNA sequence analyses of the PCR products of the gltA gene of 8 representative samples ( with high Ct values ) showed 100% similarity with the Uncultured Bartonella sp . clone Pd5700t ( GenBank no . FJ851115 . 1 , 334/334 bp ) detected in Praomys delectorum rodents in Mbulu district , northern Tanzania [19] . More information about the Ct value and localization of each positive flea is reported in Supplementary data S1 . A total of 3821 fleas ( rodent fleas and free-roaming fleas present in the environment ) were collected from localities of the Lushoto district ( United Republic of Tanzania ) and were distributed into 23 species . A total of 94 fleas belonging to six common species were screened ( Supplementary data S2 ) ( 20 Ct . f . strongylus , 7 Ct . canis , 20 Ctenophthalmus calceatus calceatus , 20 X . brasiliensis , 20 Pulex irritans and 7 Nosopsyllus incisus . All tested fleas were negative for Y . pestis and Bartonella spp . DNA . However , R . typhi DNA was detected in 10% ( 2/20 ) of X . brasiliensis collected from 2 villages ( Magamba and Manolo ) . R . felis DNA was also detected in 20 . 2% ( 23/94 ) of analyzed fleas , including 65% ( 13/20 ) of Ct . f . strongylus , 71 . 4% ( 5/7 ) of Ct . canis and 25% ( 5/20 ) of Ct . ca . calceatus . In 2007 , in the Linga and Rethy health zones , Ituri district , 1190 fleas captured in households , belonging to 6 species ( 394 P . irritans , 153 Tunga penetrans , 280 Ct . f . strongylus , 89 Echidnophaga gallinacea , 88 L . a . aethiopica and 186 X . brasiliensis ) . A total of 123 fleas were conveniently selected for this work ( Supplementary data S3 ) . qPCR for R . typhi and Bartonella spp . was negative for all 123 fleas; however , 4 . 8% ( 6/123 ) , namely 19 . 2% ( 5/26 ) of Ct . f . strongylus and 4 . 7% ( 1/21 ) of E . gallinacea , contained R . felis DNA ( Table 1 ) . Y . pestis DNA was detected in 3 . 8% ( 1/26 ) of Ct . f . strongylus and 10% ( 3/30 ) of P . irritans from 2 villages ( Wanyale and Zaa ) . DNA sequence analyses of the PCR products targeting the glpD gene showed 100% similarity with Yersinia pestis Angola isolated from Angola ( GenBank accession no . CP000901 . 1 , 321/333 bp ) . In 2010 , 111 fleas , belonging to 3 species , were collected in the same district , namely , X . cheopis ( 62/111 , 55 . 8% ) , Ct . f . felis ( 38/111 , 34 . 2% ) and L . a . aethiopica ( 11/111 , 9 . 9% ) ( Supplementary data S4 ) . qPCR for R . typhi and Y . pestis detection was negative for all fleas ( Table 1 ) ; however , 9 . 9% ( 11/111 ) of two flea species ( Ct . f . felis and L . a . aethiopica ) were positive for R . felis . A total of 10 Ct . f . felis from 38 tested ( 26 . 3% ) and one of 11 L . a . aethiopica ( 9% ) contained R . felis . Bartonella spp DNA was detected in 3 . 6% ( 4/111 ) of fleas , with 36 . 36% ( 4/11 ) from only L . a . aethiopica . Sequencing of the gltA gene fragment from these four Bartonella-positive samples showed 100% similarity with Bartonella sp . MN-ga6 ( GenBank no . AJ583126 . 1 , 320/334 bp ) detected in fleas collected in South Africa . Finally , in 2012 , from the fleas collected in Kinshasa ( Table 1 ) , 56 . 5% ( 13/23 ) of Ct . f . felis collected from 3 dogs was positive for R . felis but negative for R . typhi , Bartonella spp . and Y . pestis by qPCR .
We report the first direct evidence of R . typhi and Bartonella sp . in X . cheopis fleas in Benin ( Cotonou ) . In Lushoto ( United Republic of Tanzania ) , we detected for the first time the presence of R . typhi DNA in X . brasiliensis and R . felis DNA in Ct . f . strongylus , Ct . canis and Ct . ca . calceatus . Finally , in the Democratic Republic of the Congo , we confirmed the presence of R . felis in Ct . felis in Kinshasa and for the first time report the presence of R . felis and Bartonella DNA in L . a . aethiopica and , most importantly Y . pestis DNA in P . irritans and Ct . felis from Wanyale and Zaa villages in the Rethy health zone . The robustness of our results and the detection of these pathogens in fleas on rodents are supported by the use of a validated method of real-time PCR and subsequent sequencing . The validity of the data that we report is based on strict laboratory procedures and controls that are commonly used in the WHO Center for Rickettsial Diseases , including rigorous positive and negative controls to validate the test . Each positive qPCR result was confirmed by another specific qPCR or confirmed with a successful DNA amplification and sequencing . R . typhi was detected in X . cheopis collected from Rattus rattus in Bokossi Tokpa and Dédokpo sites ( Cotonou , Benin ) and in X . brasiliensis from the United Republic of Tanzania . X . cheopis is the primary vector of R . typhi , the etiological agent of murine typhus ( MT ) , in most locations around the world , and X . brasiliensis appears to be an effective vector under experimental conditions [3] . MT is most often a relatively mild disease; yet R . typhi can cause acute febrile illness and death [35] . The diagnosis of MT may be missed or underreported due to its non-specific symptoms or the absence of epidemiological criteria [36] , [37] because laboratory tests and validated methods of diagnosis must be performed to confirm the diagnosis [30] . Before our study , R . typhi was never detected in Benin , and it is rarely directly reported in vectors and patients in Africa , specifically in sub-Saharan Africa . R . typhi in African fleas was only detected in X . cheopis fleas collected in Algeria [38] . Additionally , R . typhi has been reported in patients using serological methods in African countries [30] . Cases have been documented in international travelers returning from Tunisia , Morocco , Ivory Coast , Central African Republic , Madagascar , Reunion and Chad [30] . In the United Republic of Tanzania , a seroprevalence study among pregnant women from the port city of Dar es Salaam found a prevalence of 28% [39] and 0 . 5 to 9 . 3% in the town of Moshi and the Mbeya region , respectively [22] , [40] . R . felis is an emergent agent of infectious disease in humans , and this agent of spotted fever is known to be maintained in cat fleas ( Ct . felis ) [41] , [42] . To date , 12 species of fleas , 8 species of ticks and 3 species of mites have been found to be infected with R . felis [42] . This Rickettsiae has also recently been detected in several mosquito species in sub-Saharan Africa [29] , [43] , [44] . Interestingly , the R . felis genogroup seems large with recent organisms or genotypes related as R . felis like organisms ( RFLO ) . Our 2 qPCR were specifically designed to amplify R . felis type strain ( URRWXCal2 ) . However , the biotin synthase and membrane phosphatase gene sequences of many RFLO are not known . We however know that at least our qPCR system targeting the biotin synthase ( bioB ) gene do not amplify some RFLO such as Rickettsia sp . RF2125 and Rickettsia sp . SGL01 . Recently , a new qPCR assay has been proposed to address this issue by providing new qPCR primers and probe to specifically amplify R . felis OmpB gene fragments [15] . The clinical features of R . felis may include fever , fatigue , headache , generalized maculopapular rash and inoculation eschar ( s ) [42] . R . felis seems to be a frequent agent of unknown fever in Sub-Saharan Africa [44] . We detected R . felis in 5 species of fleas ( Ct . f . strongylus , Ct . canis , Ct . ca . calceatus , L . a . aethiopica and E . gallinacea ) ; some from the United Republic of Tanzania ( Lushoto district ) , and other from the Democratic Republic of the Congo ( Ituri District ) . R . felis had already been detected in the Ituri district [25] , but not in E . gallinacea , the fowl flea , and has been previously shown to circulate in arthropod vectors ( Ctenocephalides felis ) in Kinshasa , the capital city of the country [21] . E . gallinacea is usually found on poultry , and can occurs on rodents ( Rattus spp . ) foraging in fowl shelters around houses [45] . While chicken DNA has been found in blood meal of fleas collected on rodents in the same area [46] other Rickettsia spp . antibodies have been found in poultry in Brazil [47] , whether or not R . felis and R . typhi infects poultry or if poultry can act as a source of infection to human is unknown . Furthermore , no data on the potential vertical transmission of R . felis in E . gallinacea , or on the vectorial transmission of R . felis by E . gallinacea males ( females are semi-sessile ) between rodents and birds , are available . The questions raised by the findings of the present study in relation to Rickettsia in fleas are of real epidemiological significance and should be further investigated . Molecular evidence of Bartonella sp . in fleas from the Democratic Republic of the Congo is supported by a recent serological survey in human patients in the Ituri who tested seropositive for B . henselae , B . quintana or B . clarridgeiae [18] . Gundi and collaborators also found that local rodents harbor Bartonella spp . closely related to B . elizabethae or B . tribocorum which shows that a wide variety of Bartonella species is present in the country , and differ according to host [19] . Bitam and collaborators [48] report that B . elizabethae , which causes endocarditis , and B . tribocorum are usually known to be transmitted by X . cheopis fleas . However , while in our study , we detected an Uncultured Bartonella sp . , clone Pd5700t ( GenBank no . FJ851115 . 1 ) in X . cheopis of Benin , we also detected Bartonella sp . MN-ga6 ( GenBank no . AJ583126 . 1 ) in L . a . aethiopica , from Ituri . This Bartonella sp . had been previously found in the Democratic Republic of the Congo and the United Republic of Tanzania in rodents [19] . The detection of Y . pestis DNA in fleas collected in villages and houses where no current human plague cases had been reported for the last 6 months is puzzling . About 80 species and subspecies of Siphonaptera are known to be carriers and potentially vector of Y . pestis [49] , via various transmission mechanisms [50]; in particular in fleas from the genus Xenopsylla ( X . cheopis ) , which played a major role in historical plague pandemics [9] . In the present survey , DNA of Y . pestis was detected in the human flea , P . irritans , and the cat flea Ct . felis in a well known endemic focus of the Democratic Republic of the Congo [51] . In 2006 , in the Rethy and Linga health zone more than 600 human cases were reported [52] , which triggered the entomological investigation reported previously [25] and the collection of fleas analyzed herein . This survey occurred 6 months after the end of the epidemics , and at the time of the flea sampling , no confirmed human plague cases were reported to the Health centre of the villages ( Zaa and Wanyale ) or Rethy general Hospital . Several hypotheses can be proposed to explain this finding . A first hypothesis is that infected fleas from rodents , dogs or cats could have been imported in the infested houses , did not bite people and as such no human cases occurred , at the time of collection . A second hypothesis is that infected fleas containing Y . pestis DNA remained infected and alive without biting any potential host or that no human cases were reported to the health authorities which are unlikely due to the recent outbreak and constant surveillance . Other options are that Y . pestis DNA is reminiscent in the flea but the bacterium is either dead ( degraded DNA ) but the targeted sequences ( gene fragment and gene flanking regions are still complete ) or alive but in a quiescent form or VBNC state , possibly controlled by epigenetic mechanisms causing virulence gene repression . The human flea ( P . irritans ) may play an important role in spreading plague via human-to-human transmission as suggested in Lushoto district [27] and could possibly harbor Y . pestis without transmission for several months . Unfortunately no fleas were cultured in the field and the viability of the strain detected cannot be proven , but this finding calls for more research at times post outbreaks in order to answer this question . Similarly , cat fleas could play such a role both in northwest Uganda [53] and in Democratic Republic of the Congo ( Laudisoit and al 2014 , unpublished data ) , where C . felis spp . is the most common flea species collected in the domestic environment above a given altitude threshold . In conclusion , we widened knowledge of the repertoire of flea-borne bacteria present in Sub-Saharan Africa . In our study , we also illustrate the role of fleas in the entomological survey of vector -borne disease , which allow clinicians to confirm the etiological cause for some of the unknown cause of fever in African patients . Future studies on rickettsioses , bartonelloses and other vector-borne diseases should be performed to assess their epidemiological and clinical relevance in tropical and subtropical areas , to estimate the real prevalence and to allow the establishment of antivectorial control plans . | Fleas are associated with many bacterial diseases such as rickettsioses , bartonelloses and plague . These diseases may be severe , and little is known about their prevalence . Accordingly , we believe that our data shed light on the problem of unexplained fevers in tropical and subtropical African areas . Using molecular tools , we surveyed and studied selected flea-borne agents , namely Rickettsia spp . ( R . felis and R . typhi ) , Bartonella spp . and Y . pestis , in fleas collected in Ituri ( Linga and Rethy health zone ) and Kinshasa in the Democratic Republic of the Congo , the Lushoto district in the United Republic of Tanzania and in Cotonou in Benin . We found that these bacteria are present in the studied regions . R . typhi and an unidentified Bartonella sp . were detected in fleas from Cotonou ( Benin ) . R . felis and R . typhi were also detected in the Lushoto district ( United Republic of Tanzania ) . Finally , we detected R . felis , Bartonella sp . and Y . pestis in the Democratic Republic of the Congo . As these emerging zoonotic diseases have a global distribution and affect public health , the implementation of vector control strategies is urgent . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"infectious",
"disease",
"surveillance",
"plant",
"science",
"medicine",
"and",
"health",
"sciences",
"clinical",
"epidemiology",
"disease",
"dynamics",
"epidemiology",
"vector",
"biology",
"disease",
"surveillance",
"plant",
"pathology",
"biology",
"and",
"life",
"scie... | 2014 | Detection of Rickettsia felis, Rickettsia typhi, Bartonella Species and Yersinia pestis in Fleas (Siphonaptera) from Africa |
Despite major attempts to prevent cholera transmission , millions of people worldwide still must address this devastating disease . Cholera research has so far mainly focused on the causative agent , the bacterium Vibrio cholerae , or on disease treatment , but rarely were results from both fields interconnected . Indeed , the treatment of this severe diarrheal disease is mostly accomplished by oral rehydration therapy ( ORT ) , whereby water and electrolytes are replenished . Commonly distributed oral rehydration salts also contain glucose . Here , we analyzed the effects of glucose and alternative carbon sources on the production of virulence determinants in the causative agent of cholera , the bacterium Vibrio cholerae during in vitro experimentation . We demonstrate that virulence gene expression and the production of cholera toxin are enhanced in the presence of glucose or similarly transported sugars in a ToxR- , TcpP- and ToxT-dependent manner . The virulence genes were significantly less expressed if alternative non-PTS carbon sources , including rice-based starch , were utilized . Notably , even though glucose-based ORT is commonly used , field studies indicated that rice-based ORT performs better . We therefore used a spatially explicit epidemiological model to demonstrate that the better performing rice-based ORT could have a significant impact on epidemic progression based on the recent outbreak of cholera in Haiti . Our results strongly support a change of carbon source for the treatment of cholera , especially in epidemic settings .
The diarrheal disease cholera remains a major problem in developing countries . In 2012 , almost 250'000 cases were reported to the WHO; however estimated numbers , including non-reported cases , are argued to reach several million cases every year . The recent disease outbreak in Haiti demonstrated the devastating effects of cholera epidemics . Because of these dramatic consequences , the problem of how to stop an epidemic at an early stage , or at least to slow it down , was addressed and the employment of general intervention strategies was discussed . In this context , mathematical models were developed and used to predict the outcome of major interventions , such as vaccination or extended use of antibiotics and sanitation [1]–[6] . Remarkably , more general treatment strategies have not been considered . This fact appears surprising because , without treatment , the case fatality rate for severe cholera is approximately 50%; however , if handled properly , nearly all deaths can be avoided . The general treatment of cholera patients is based on a so-called oral rehydration therapy ( ORT ) , which is a cost-effective and easily applicable method to replace lost fluids and electrolytes . For the latter purpose , the administered solution contains a mixture of several compounds that were designated oral rehydration salts ( ORS ) , including sodium , chloride , and potassium ions as well as glucose . Indeed , glucose is the most commonly added carbohydrate because it stimulates sodium and therefore , water absorption in the small intestine [7] . However , in field studies , it was shown that ORT might be improved by the substitution of the carbon source ( or through the addition of amino acids and supplementation with trace elements ) [8] . The molecular mechanisms underlying these findings remain to be resolved . Moreover , a meta-analysis comparing the treatment with standard , glucose-based versus rice-based ORS illustrated the beneficial effects of the latter composition , such as reduced episodes of vomiting , a decrease of the stool volume , and a shortened recovery time [9]–[12] . The aim of this study was to unravel the molecular mechanisms underlying the better performance of rice-based treatments ( Fig . 1 ) . To do so , we analyzed the virulence gene expression of the causative agent of cholera , specifically the bacterium Vibrio cholerae , when cultured under conditions resembling ORT . Our results show that , in the presence of glucose , virulence-associated genes were induced and cholera toxin was produced and this in vitro induction was dependent on the main virulence regulatory proteins ToxR , TcpP and ToxT [13] . However , substituting glucose in ORT by an alternative non-PTS carbon source , such as rice-derived starch , led to the significantly reduced expression of virulence determinants in V . cholerae , which would have beneficial effects on disease progression . Based on this evidence , the impact of such optimized treatment on the outcome of cholera epidemics was simulated using a mathematical model , which recapitulates the recent Haitian cholera outbreak [4] . Our results indicate that rice-based treatment could have significantly reduced the burden imposed on the Haitian population .
The Vibrio cholerae strains that were used in this study are El Tor Inaba strain A1552 [14] , El Tor Inaba strain N16961 [15] , classical Ogawa strain O395 ( India , 1964 ) [16] , and derivatives of those strains ( see Supporting Table S1 ) . All strains belong to serogroup O1 . V . cholerae strains were grown as overnight cultures in Luria broth ( LB ) or M9 minimal medium at 30°C under shaking conditions . Antibiotics were added at the following concentrations when required: kanamycin at 75 µg/ml or ampicillin at 100 µg/ml . The genes toxR , ptsG , and nagE were deleted from the parental V . cholerae strain ( A1552 ) using the TransFLP gene disruption method [17]–[19] , which is based on natural transformation and FLP recombination . The same method was applied to introduce a functional hapR gene into strain N16961 to acquire strain N16961rep ( the natural transformation was performed in the presence of a hapR-complementing plasmid ) . To delete the genes tcpP and toxT from the parental strain A1552 , a gene-disruption method based on the counter-selectable plasmid pGP704-Sac28 [20] was used . The oligonucleotide sequences that were used to construct the transforming PCR fragments or plasmids pGP704-28-SacB-ΔVC0826 and pGP704Sac28ΔtoxT-II are indicated in Supporting Table S2 . Genomic DNA from V . cholerae strain A1552 served as a template . A detailed protocol for virulence gene expression in V . cholerae is provided in the Supporting Information ( Supporting Text S1 ) . Briefly , bacteria were grown in a M9 minimal salt medium ( Sigma-Aldrich , Buchs , Switzerland ) , which contained vitamins and casamino acids , as well as the carbon source of interest . Virulence gene expression was induced by the addition of sodium bicarbonate ( Fluka , St . Louis , USA ) as published for rich medium conditions [21] . Gene expression was evaluated using quantitative reverse transcription-based PCR ( qRT-PCR ) as described [22] . The gene-specific primers that were used for qRT-PCR are indicated in Supporting Table S2 . Cholera toxin ( CT ) concentrations were determined using a CT-ELISA , which was performed essentially as reported [23] , with the exception that phosphate-citrate buffer containing sodium perborate was used for detection . A mathematical model was applied to recapitulate the first year of the cholera outbreak that recently occurred in Haiti and to estimate the potential impact of rice-based ORT . In addition to the epidemiological processes that are relevant to cholera transmission , the model also takes hydrological pathogen transport , human mobility , and precipitation into account ( see Supporting Text S1 ) . Model parameters were estimated based on the available literature or calibrated using a Bayesian approach . The modeled progression of the cholera epidemic corresponded well with the observed cases ( details on the spatial match of reported cases and simulation results are provided elsewhere [4] , [6] , [24] , [25]; see also Supporting Text S1 ) . Next , the described effects of the rice-based treatment were implemented . This is compared with standard glucose-based ORS by assuming that ( a ) during the actual course of the Haiti epidemic all symptomatic infected people received a glucose-based ORT , and ( b ) all parameters ( namely patterns of human mobility , demography , bacterial ecology , and exposure rates ) are assumed unaltered by the ORT except for the shedding rate and disease duration ( according to [9]–[12] ) . Moreover , a 30-days lag period between the initial cholera case and the switch to the rice-based ORT was included because this timeframe seems reasonable to first confirm the causative agent of the diarrheal disease outbreak as being V . cholerae at the onset of the epidemic . Details regarding the model structure , the parameterization , and the validation can be found in the Supporting Information , which contains Supporting Text S1 , Supporting Figures ( Fig . S3 , Fig . S4 , Fig . S5 , Fig . S6 , and Fig . S7 ) and Supporting Tables ( Table S4 and Table S5 ) .
This study aimed to investigate how different carbon sources that are used for cholera treatment influence the production of the major virulence determinants in V . cholerae , cholera toxin ( CT ) and the toxin-coregulated pilus ( TCP ) ( Fig . 1 ) . CT is an AB5-toxin , which consists of two different subunits that are encoded by the genes ctxA and ctxB . Whereas the B subunit is important for the binding to host cells , the A subunit indirectly activates the enzyme adenylate cyclase within gut epithelial cells , which leads to an increased secretion of chloride and concomitantly increased secretion of water ( Fig . 1 ) . CT alone can cause diarrhea , and the amount of CT intake was shown to be proportional to the severity of symptoms [26] , [27] . The second major virulence factor of pathogenic V . cholerae strains , TCP , is important for intestinal colonization [28] . The pilus structure primarily consists of the major pilin TcpA , whereas the residual tcp gene products are important for the pilus structure and for its assembly ( Fig . 1 ) . We first established an assay to monitor the expression of these major virulence genes when the bacteria were grown under conditions that somewhat mimic those conditions that are encountered in patients undergoing ORT . To this extent , we used a bicarbonate-mediated in vitro virulence induction protocol [21] , [29] but changed the culture medium to minimal salts ( resembling ORS ) instead of the previously described rich-medium conditions . Using quantitative qRT-PCR , we demonstrated that , upon growth of the V . cholerae wild-type O1 El Tor strain ( A1552 ) in the presence of glucose , the genes ctxA , ctxB , tcpA and tcpB were significantly induced whereas significantly lower induction was observable when lactate was used as alternative carbon source ( Fig . 2A ) or under virulence-non-inducing conditions ( Fig . S1 and Table S3 ) . To confirm that the reduced virulence gene expression is also reflected in reduced protein levels of the virulence determinants , we analyzed the accumulation of CT in culture supernatants using an ELISA . Indeed , the amount of cholera toxin decreased to ∼25% for cultures that were grown in the presence of lactate compared with glucose ( Fig . 2B ) . Because it has become evident in recent years that the regulation of virulence genes differs between diverse pathogenic isolates [30] , [31] , we applied our method to two other epidemic V . cholerae strains . N16961 is an El Tor O1 strain ( as is strain A1552 ) , whereas O395 is a classical O1 biotype strain [15] , [16] . Indeed , these two strains also exhibited a glucose-dependent induction of virulence genes ( Fig . S1 ) , which indicates the general impact of carbon sources on the regulatory network of virulence-associated genes in V . cholerae . Most bacteria transport preferred sugars , such as glucose , by so-called phosphoenolpyruvate-phosphotransferase systems ( PTS ) ; therefore , the corresponding carbohydrates are often denoted as PTS sugars . During the sugar uptake via the PTS , a phosphoryl group is transferred onto the imported sugar , which prepares the sugar for central metabolism . In contrast , the enzyme adenylate cyclase , which is responsible for the production of the secondary messenger cyclic AMP ( cAMP ) , is activated when cells run short of PTS sugars . cAMP that is bound to the cAMP-receptor protein ( CRP ) is a crucial signaling compound and modulates several processes in bacteria [32] . Thus , we asked the question whether the observed virulence induction was specific to glucose , specific to PTS sugars , or a broader effect that was exerted by many carbon sources . Therefore , we tested other PTS sugars in addition to glucose ( e . g . , N-acetylglucosamine and sucrose ) and observed a comparable induction of the virulence genes , as demonstrated for glucose ( Tables 1 and 2 ) . In contrast , galactose and succinate , which are two non-PTS carbon sources , led to highly reduced virulence expression ( Table 1 ) . More importantly , we demonstrated that V . cholerae induces the virulence determinants at significantly lower levels when starch was provided as the sole carbon source , which mimics the conditions of rice-based ORT ( Table 1 ) . This reduced virulence gene expression was again reflected in the accumulation of CT , which decreased to 10–25% for cultures that were grown in the presence of non-PTS carbon sources ( e . g . , lactate , galactose , succinate or starch ) compared with glucose ( Fig . 2B ) . To elucidate whether the sugar-dependent virulence induction is linked to the transport of the PTS sugars we employed genetically modified V . cholerae strains , which lacked the PTS transporters for glucose , N-acetylglucosamine , or sucrose . In these strains , the signaling pathway that was described above is disrupted , thereby allowing cAMP to accumulate in the bacterial cell . As shown in Table 2 , the increase of virulence gene expression upon provision of the inducer bicarbonate was abolished in those strains , which were cultivated in the presence of the corresponding PTS sugars ( with the concomitant addition of lactate to maintain growth ) . Therefore , we conclude that the expression of the virulence genes is not dependent on any particular sugar per se but is dependent on the PTS-coupled regulatory circuit . The PTS-dependent signaling pathway ( including the secondary messenger cAMP ) is part of a process known as carbon catabolite repression in bacteria [32] . This mechanism allows bacteria to first utilize preferred sugars through the repression of uptake complexes and catabolic enzymes that are required for the utilization of less-preferred carbon sources . Earlier studies have indicated that carbon catabolite repression also acts as a modulator of quorum sensing ( QS ) in V . cholerae [33] . QS is commonly used by bacteria to assess population densities and to respond appropriately to these [34] . The master regulator of QS in V . cholerae is the protein HapR , which is produced at high cell density and which represses the virulence genes [35] , [36] . Consistent with these earlier studies , we observed increased virulence gene expression in a hapR mutant strain when compared with the wild-type strain ( Fig . S2 ) . Furthermore , the V . cholerae strain N16961 [15] , which has a frameshift mutation in hapR that renders the HapR protein non-functional [35] , also showed increased virulence gene expression compared with a genetically modified variant of N16961 in which the hapR gene was repaired ( Fig . S2 ) . However , the hapR mutant strain still displayed glucose-responsiveness with respect to virulence gene expression , which indicated that carbon catabolite repression also influences virulence in a QS-independent manner ( Fig . S2 ) . As the aforementioned observations were based on in vitro data the question arose as to how these data compare to the situation within human patients . It should be noted that field studies indicated that rice-based ( e . g . , starch-based ) ORT performs better compared to standard glucose-based ORT though the underlying mechanism remained elusive [9]–[12] . Apart from these important field study data it is difficult or even impossible to test the in vivo situation under laboratory conditions ( e . g . , outside the human intestine ) . The reason for that is that animal models do not recapitulate the disease well . Indeed , both adult rabbits and adult mice are naturally resistant to infection with the pathogen . Thus , the most prominent animal models of cholera are the rabbit ileal loop model and the infant mice model ( for recent review see [37] ) . The first model primarily reflects the enterotoxicity of the bacteria measured by the accumulation of fluid within ligated intestinal loops . However , this model does not take the natural oral infection route into consideration ( e . g . , external cues encountered throughout the gastrointestinal passage ) . Moreover , due to the feature of the model as being a closed system it does not allow any ORT treatment studies . Lately , infant mice ( suckling mice ) have been widely used to study virulence factors of V . cholerae . The infant mice model allows elucidating the potential of the pathogen to grow inside the animal and to colonize the small intestine . However , infant mice do not develop severe diarrhea and fluid replacement by ORT is therefore impossible . Notably , these models have been extremely valuable for deciphering the pathogenicity of the bacterial pathogen [37] . For example the involvement of TCP in intestinal colonization was established based on the suckling mice model [28] ( Fig . 1 ) , which was later confirmed to also be relevant in humans based on volunteer studies [38] . We therefore reasoned that the relevance of our in vitro data would be significantly strengthened if we could confirm that the observed in vitro virulence expression would still be subject to the well-established regulatory cascade of virulence induction in V . cholerae ( “the ToxR regulon” [39] ) , which includes the main regulatory proteins TcpP , ToxR [28] , [40] , and ToxT [41] ( for review see [13] , [42] ) ( Fig . 3A ) . To this extent , we first generated V . cholerae knockout strains lacking either toxR or tcpP ( Fig . 3A; Supporting Table 1 ) and subjected these strains and the parental wild-type strain to in vitro virulence induction in the presence of different carbon sources . Importantly , we saw a strong and statistically significant decrease in the expression of all tested virulence genes ( tcpA , tcpB , ctxA , ctxB ) in the presence of glucose when TcpP and ToxR were absent ( Fig . 3B ) . As both of these regulatory proteins are required for the production of ToxT ( Fig . 3A ) , the master regulator of virulence , which directly activates the tcp and ctx gene cluster by binding to so called toxboxes [13] , we next tested the contribution of ToxT to virulence induction under the here described in vitro virulence induction conditions . Indeed , the expression of toxT under virulence-inducing conditions was significantly increased in the presence of glucose compared to lactate ( Fig . 2A ) and this increase was dependent on the presence of the respective PTS transport system ( Table 2 ) . Importantly , when we deleted the toxT gene or grew the wild-type cells in the presence of the ToxT inhibitor virstatin [43] , [44] the glucose-dependent virulence induction was absent or significantly reduced ( Figs . 3C and 3D ) highlighting the importance of ToxT for in vitro virulence induction . Thus , as the in vitro induction of virulence determinants was dependent on the main regulatory proteins TcpP , ToxR , and ToxT we concluded that our approach recapitulates the in vivo situation to a certain extent and that the here described glucose-dependent expression of the main pathogenicity genes most likely also occurs in human patients . This finding would be consistent with the increased disease burden associated with glucose-based ORT compared to rice-based ORT [9]–[12] . Because we demonstrated above that starch does not support virulence gene expression to the same high level as glucose does , we aimed to test whether virulence could still be lowered by starch if previously induced . Indeed , such a scenario would occur at the onset of cholera symptoms and upon initial treatment with glucose-based ORS . Thus , we grew V . cholerae in glucose-containing virulence-inducing medium for a short period before washing the bacteria and shifting them to fresh medium , which contained either glucose or starch . Using this experimental approach , we observed a significantly reduced amount of CT ( down to ∼25% ) in starch-shifted cultures compared with the glucose-grown bacteria ( Fig . 4 ) . These data suggest that the administration of rice-based ORS , even after the initial virulence induction , would still significantly reduce CT accumulation along with reduced cholera symptoms . The data that were presented above provide a molecular explanation for why rice-based ORS performed better in field studies [9]–[12] . However , one wonders whether disease shortenings and reduced stool output , as described for rice-based ORT , would have any impact on large-scale patterns of cholera epidemics . We therefore used a spatially explicit epidemiological model [1] , [4] , [6] to recapitulate the recent Haitian cholera outbreak . The model consists of 365 nodes ( human communities ) that are spread over the Haitian territory ( Fig . 5A ) . Quantifying the reduction of shedding rate and disease duration is not easy . Our in vitro experiments suggest a strong reduction of the amount of CT , which matches the observed reduction of shedding rates from field studies [9]–[12] . This reduction can be as high as 50% ( especially compared to the glucose-based ORS , which was recommended by the WHO before 2002 ( WHO-ORS ) ; see discussion ) . Reduction of diarrhea duration is smaller and usually no larger than 30% . If one assumes a 10% reduction in both duration and shedding rate for rice-based compared with glucose-based ORT ( for HYPO-ORS; [12] ) , the model predicts a considerable decrease in disease incidence over the entire country ( Fig . 5B ) . Indeed the total number of cholera cases within the first 14 months of the epidemic would be reduced from 520 , 000 cases ( as reported by the Haitian Ministry of Health and available online at http://mspp . gouv . ht; our model predicts 535 , 000 cases ) to 375 , 000 cases ( i . e . 30% [22%–39%] less total cases until the end of 2011 ) according to the model . More importantly , if these parameters could be reduced by 15% , then the total number of cholera cases would drop by 59% [47%–67%] , and if the parameters were reduced by 20% , the number of cases would even decrease by 74% [71%–76%] ( Fig . 5C ) . The ranges of variation shown reflect the 2 . 5th–97 . 5th percentiles of the uncertainty related to parameter estimation . Such behavior ( i . e . the more than doubled reduction of total infections owing to a 10% to 20% reduction in bacterial shedding and disease duration ) is typical of nonlinear epidemiological models [6] . Interestingly , owing to a higher number of susceptibles ( Fig . S7 and Supporting Text S1 ) , a larger number of cholera cases would have been predicted for November 2011 , one year after the initial onset of the outbreak , most probably triggered by important rainfall events , which have been shown to play a major role in the dynamics of the epidemic [4] , [45] . However , such a one-year time span would have allowed other intervention strategies to be put into place , which could potentially avoid later cholera case peaks e . g . , by reducing exposure rates via improved water sanitation .
Cholera remains a major social emergency in developing countries and despite major research efforts this disease is far from being eradicated . Here , we aimed at understanding why an alternative cholera treatment , which relies on rice-based instead of glucose-based ORS , has beneficial effects on disease outcome as reported in field studies ( Fig . 1 ) . For this purpose , we tested the expression of virulence-associated genes in V . cholerae using defined virulence-inducing minimal medium conditions and a variety of carbon sources . Using this approach , we demonstrated that the virulence genes were upregulated in different clinical isolates of V . cholerae as long as those strains were grown in the presence of glucose or other PTS sugars ( Fig . 2A , Fig . S1 and Table 1 ) . The amount of cholera toxin produced under these conditions reflected the expression data ( Fig . 2B ) . Because the cholera toxin is primarily responsible for the severe symptoms that are associated with the disease , our study highlights the negative effects of glucose-based ORT . Interestingly , in 2002 , the WHO announced the recommendation to reduce the osmolarity in ORS [46] . Concomitantly , the final concentration of glucose was lowered from the initial 111 mM ( WHO-ORS ) to 75 mM ( HYPO- ORS ) . Nevertheless , we observed that lowering the concentration of glucose in this range ( from 111 mM glucose that was previously used , toward the 75 mM current recommendation , or even down to 50 mM , such as in our assays , because this concentration might better reflect the concentration within the intestine ) did not reduce virulence gene expression in V . cholerae ( Table 1 ) . To our knowledge , this study is the first to describe the PTS sugar-dependent expression of virulence genes in V . cholerae cells grown in minimal salt conditions , although a putative cAMP-CRP binding site in the tcpA promoter region has been proposed previously [47] and the repression of TCP and CT by cAMP-CRP has been described [48] . Furthermore , a link between carbon catabolite repression and QS-dependent virulence repression has been reported before [33] , [35] , [49]; however , our data indicate that PTS-sugars also influence virulence expression in a QS-independent manner ( Fig . S2 ) , which might be due to the binding of the cAMP-CRP complex to the putative cAMP-CRP binding site [47] mentioned above . In summary , our findings indicate that the virulence cascade of V . cholerae is inducible only in the presence of PTS sugars . Concomitantly , cholera toxin is produced on a larger scale when glucose is provided , which is a situation that most likely occurs when patients undergo standard glucose-based ORT . No such increase in CT production and virulence gene expression was observed when starch was used as alternative carbon source , which mimics rice-based ORT . Moreover , a switch from a glucose-based to a starch-based medium still resulted in decreased virulence gene expression in V . cholerae , which indicated that rice-based ORS would still be beneficial even if rice-based ORS were applied after the onset of symptoms . These results are supported by clinical studies , which demonstrated that rice-based ORS performed better in the field than standard glucose-based ORS [9]–[12] . The implementation of rice-based ORT has important effects on the epidemiological course of cholera . Indeed , the results of our model study of the Haitian cholera epidemic indicated a significant reduction in the total number of cholera cases using the range of parameters published for rice-based ORS . Notably , this significant reduction in cholera cases might even be an underestimation given that rice-based ORT might effectively down-regulate virulence gene expression in the pathogen thereby abolishing the hyperinfectivity state that has been described for V . cholerae [21] , [50] . This would ultimately lead to an even lower transmission rate within the population . However , current recommendations by the WHO do not fully support the use of rice-based ORS for reasons such as product stability , increased packaging size , mode of application , and a three-fold cost over glucose-based ORS ( $0 . 20 per liter of final ORS solution compared with $0 . 07 ) [51] . On the other hand , recent improvements in the food/nutrition industry might have optimized those handling- and cost-associated issues of starch-based products . Due to the described beneficial effects of rice-based ORS , the molecular explanations provided in this study , and the impact that those beneficial effects could have on epidemic progression , we recommend reconsidering starch-based ORS for cholera treatment . | Cholera research has so far mainly focused on the causative agent , the bacterium Vibrio cholerae , or on disease treatment , but rarely were results from both fields interconnected . Indeed , the treatment of this severe diarrheal disease is mostly accomplished by oral rehydration therapy ( ORT ) . ORT aims at rehydrating patients through the provision of water and oral rehydration salts; the latter being composed of electrolytes as well as glucose as a carbon source . Although glucose-based ORS is commonly used to treat diarrheal diseases and is recommended by the WHO , field studies on cholera indicated that rice-based ORT performs better than glucose-based ORT . Here , we investigated the impact that glucose , starch , or other carbon sources exert on V . cholerae . We demonstrated that glucose leads to an increased expression of the major virulence genes in the pathogen and , accordingly , to an enhanced production of cholera toxin during in vitro experimentation . Because the cholera toxin is primarily responsible for the severe symptoms that are associated with the disease , our study highlights the negative effects of glucose-based ORT . Next , we used a spatially explicit epidemiological model to demonstrate that the better performing rice-based ORS could have a significant impact on epidemic progression based on the recent outbreak of cholera in Haiti . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"bacterial",
"diseases",
"infectious",
"diseases",
"medicine",
"and",
"health",
"sciences",
"cholera",
"population",
"modeling",
"medical",
"microbiology",
"microbial",
"pathogens",
"biology",
"and",
"life",
"sciences",
"microbiology",
"infectious",
"disease",
"modeling",... | 2014 | Glucose- but Not Rice-Based Oral Rehydration Therapy Enhances the Production of Virulence Determinants in the Human Pathogen Vibrio cholerae |
The immortal strand hypothesis poses that stem cells could produce differentiated progeny while conserving the original template strand , thus avoiding accumulating somatic mutations . However , quantitating the extent of non-random DNA strand segregation in human stem cells remains difficult in vivo . Here we show that the change of the mean and variance of the mutational burden with age in healthy human tissues allows estimating strand segregation probabilities and somatic mutation rates . We analysed deep sequencing data from healthy human colon , small intestine , liver , skin and brain . We found highly effective non-random DNA strand segregation in all adult tissues ( mean strand segregation probability: 0 . 98 , standard error bounds ( 0 . 97 , 0 . 99 ) ) . In contrast , non-random strand segregation efficiency is reduced to 0 . 87 ( 0 . 78 , 0 . 88 ) in neural tissue during early development , suggesting stem cell pool expansions due to symmetric self-renewal . Healthy somatic mutation rates differed across tissue types , ranging from 3 . 5 × 10−9/bp/division in small intestine to 1 . 6 × 10−7/bp/division in skin .
The immortal DNA strand hypothesis , originally proposed by Cairns in 1975 , poses that adult mammalian stem cells do not segregate DNA strands randomly after proliferation [1] . Instead , stem cells might preferentially retain the parental ancestral strand , whereas the duplicated strand is passed onto differentiated cells with limited life span ( Fig 1 ) . In principle , such hierarchical tissues could produce differentiated progeny indefinitely without accumulating any proliferation-induced mutations in the stem cell compartment [2 , 3] . Experimental evidence supporting this hypothesis comes from BrdU stain tracing experiments both in vitro and in vivo [4–7] . Evidence from spindle orientation bias in mouse models of normal and precancerous intestinal tissue corroborated these findings , suggesting that strand segregation is then lost during tumourigenesis [8] . However , many of the experiments suffer from uncertainties in stem cell identity and a definite mechanism of strand recognition remains unknown [9] . Hence why Cairns hypothesis remains controversial [10] . Orthogonal studies based on the expected accumulation of somatic mutations in healthy human tissues have argued against the immortal strand hypothesis [11 , 12] . However , the mere accumulation of somatic mutations in healthy tissue neither supports nor negates the immortal strand hypothesis in vivo . Here , we show that measuring the change of the mutational burden and , most crucially , the change of the variance of the mutational burden with age allows determining the probability of DNA strand segregation and the per cell mutation rate in healthy human tissues . First , we outline the approach and then apply it to genomic data from healthy human colon , small intestine , liver , skin and brain tissue . The data comes from four recent independent studies on mutational burden in healthy tissues [13–16] , which contain information on in total 39 individuals at different ages and analysed genomes of 341 single cells . We find evidence for non-random strand segregation in all adult tissues and significant differences in somatic mutation rates between tissues , but less prominent strand-segregation in brain tissue during early development .
We describe the accumulation of mutations with time in hierarchically organised human tissues by a stochastic mathematical and computational model , Fig 1 . A detailed description and derivation of all equations is provided below ( Materials and Methods ) . Briefly , our model considers a constant number of N stem cells that contribute to tissue homeostasis . Stem cells divide with a certain constant rate λ , e . g . once every week or month . During each division , the parental DNA strand is copied and χ novel mutations might occur on the daughter strand . Here χ is a random number that follows a Poisson distribution with mutation rate μ per bp/division and genome size L . Cell fate is also probabilistic in our model . Cells with the parental strand will keep a stem-cell fate with probability p , e . g . for p = 1 they will always remain stem cell , or differentiate otherwise , e . g . for p = 1/2 cell fate decisions are purely random ( coin flip ) . We can understand the probability p as the probability of non-random strand segregation , e . g . p ≈ 1 suggest highly non-random strand segregation , whereas p = 1/2 corresponds to random strand segregation . With this model , we can describe the accumulation of mutations over time explicitly ( see Materials and Methods for more details ) . Assuming the mutation rate μ as well as the cell proliferation rate λ to be constant , we find that both the mutational burden μ˜ as well as the variance of the mutational burden σ2 are predicted to increase linearly with time t: 1λΔμ˜Δt= ( 1−p ) μL ( 1 ) 1λΔσ2Δt= ( 1−p ) μL+ ( μL ) 2p ( 1−p ) , ( 2 ) see Materials and Methods for a detailed derivation and Fig 2 for a verification by individual based computer simulations . However , the rates by which the mutational burden and the variance of the mutational burden increase over time depend differently on the mutation rate μ and the non-random strand segregation probability p . This allows us to independently measure the mutation rate μ and the non-random strand segregation probability p via: p=Δσ2Δμ˜−1Δσ2Δμ˜−1+1λΔμ˜Δt ( 3 ) μL=Δσ2Δμ˜−1+1λΔμ˜Δt . ( 4 ) Importantly , measuring the change in mutational burden Δμ˜Δt and variance Δσ2Δt over time in combination with Eqs ( 3 ) and ( 4 ) determines the mutation rate μ ( per cell divison ) and the non-random strand segregation probability p for healthy tissues . In a recent publication Blokzijl and colleagues [13] measured mutation accumulation in healthy colon , small intestine and liver tissue by whole genome sequencing multiple single stem cell derived organoids of healthy donors of different ages . In addition , Martincorena and colleagues [14] measured mutational burden in multiple skin samples of four individuals with ages between 58 and 73 years . Furthermore , two recent publications [15 , 16] performed large-scale single cell whole genome sequencing of neurons at different ages . In the experiments by Blokzijl and colleagues [1 , 13] , they isolated single cells and expanded those into organoids . These cells can therefore be thought of as tissue specific stem cells . In contrast , the other experiments [2 , 3 , 14–16] do not directly measure mutational burden in stem but more differentiated progenitor cells . However , compared to the total number of cell divisions in the tissue , the number of divisions separating stem and progenitor cells are neglectable . These datasets enable measurements for the change in mutational burden Δμ˜Δt and the variance Δσ2Δt of the mutational burden with age in those healthy human tissues , see Figs 3 & 4 . Eqs ( 3 ) and ( 4 ) have a single undetermined parameter , the stem cell proliferation rate λ . Strictly speaking , they therefore only provide possible ranges for the mutation rate and the strand segregation probability . However , the possible ranges are narrow for any biologically meaningful stem cell proliferation rate , see Fig 5 . For all tissues , the experimental observations confirm our expectation of a linearly increasing mean and variance of the mutational burden . Using linear regression on the data in [4–7 , 13–16] , we find for colon that the change in mutational burden over time was: Δμ˜Δt=37 . 2±3 . 1 , for small intestine: Δμ˜Δt=34 . 6±6 . 9 , for liver: Δμ˜Δt=30 . 5±2 . 1 , for prefrontal cortex: Δμ˜Δt=16 . 2±1 . 1 and for hippocampal dentate gyrus: Δμ˜Δt=21 . 8±7 . 9 mutations per whole genome per year . We found for skin: Δμ˜Δt=1 . 66±0 . 15 mutations per 0 . 69 Mb per year . We found for neurons during early development: Δμ˜Δt=4 . 2±1 . 3 mutations per whole genome per day . Uncertainties here are standard errors . Similarly , for the change of variance we found for colon: Δσ2Δt=985 . 5±103 , for small intestine: Δσ2Δt=747 . 3±304 , for liver: Δσ2Δt=1564±56 , for prefrontal cortex: Δσ2Δt=7500±965 , for hippocampal dentate gyrus: Δσ2Δt=15016±6234 mutations per whole genome per year , for skin: Δσ2Δt=5 . 23±0 . 37 mutations per 0 . 69 Mb per year and for neurons during early development: Δσ2Δt=252 . 2±191 mutations per whole genome per day ( Figs 3 & 4 ) . If stem cells divide once per week this implies ( Eq ( 3 ) ) for the probability of DNA strand segregation in colon: p = 0 . 973 ( 0 . 971; 0 . 974 ) , small intestine: p = 0 . 969 ( 0 . 966; 0 . 97 ) , liver: p = 0 . 988 ( 0 . 987; 0 . 989 ) , prefrontal cortex: p = 0 . 999 ( 0 . 998; 0 . 9993 ) , hippocampal dentale gyrus: p = 0 . 999 ( 0 . 998; 0 . 9998 ) , skin: p = 0 . 985 ( 0 . 983; 0 . 987 ) . In contrast for neurons during early development we find: p = 0 . 876 ( 0 . 78; 0 . 88 ) if cells divide every 48h . Numbers in brackets correspond to the range of the DNA strand segregation probabilities given the upper and lower bound of the error estimates of the linear regressions . Dependencies of the estimates on the proliferation rate can be found in the caption of Fig 5 . This suggests highly effective non-random DNA strand segregation in human adult stem cells and is in line with previous observations of predominantly asymmetric stem cell divisions [6–8 , 17 , 18] . It would require extreme stem cell proliferation rates of approximately one division per stem cell per year for the data to be consistent with solely random strand segregation ( p = 0 . 5 ) , Fig 5 . This is an unlikely scenario as all tissues analysed here are thought to have high stem cell proliferation rates [9 , 19 , 20] . Interestingly , during development non-random DNA strand segregation is less prominent . One explanation is an expanding stem cell population due to symmetric stem cell self-renewals during early development[1 , 10] , which also would explain the increased accumulation of mutations early , as well as typical increased telomere shortening early in life [11 , 12 , 21] . Based on Eq ( 4 ) we find for the in vivo mutation rate per base pair per cell division in colon: μ = 4 . 37 ( 4 . 26; 4 . 46 ) × 10−9 , small intestine: μ = 3 . 54 ( 2 . 61; 4 . 17 ) × 10−9 , liver: μ = 8 . 48 ( 8 . 22; 8 . 77 ) × 10−9 , prefrontal cortex: μ = 7 . 68 ( 7 . 18; 8 . 12 ) × 10−8 , hippocampal dentale gyrus: μ = 1 . 14 ( 1 . 04; 1 . 68 ) × 10−7 , neurons during early development: μ = 1 . 23 ( 0 . 43; 1 . 52 ) × 10−8 and skin: μ = 1 . 57 ( 1 . 54; 1 . 63 ) × 10−7 . The ranges of these values agree with a recent estimate of the somatic mutation rate in human fibroblasts [13–16 , 22] and are one to two orders of magnitude larger than germline mutation rates [13 , 23 , 24] . However , our method does not require precise estimates of the total number of cell divisions since conception ( Fig 5 ) . We find surprising differences in the somatic mutation rates across tissue types that cannot be explained by for example different stem cell proliferation rates alone . The mutation rate estimate in skin is particularly high . This might be due to the nature of the samples used by Martincorena and colleagues [14] , as the mutational burden was measured in eye lids of individuals that were exposed to high levels of UV radiation for decades . It is plausible that this contributed to the very high mutation rate estimate . It remains to be seen , if these differences across tissues prevail for denser sampling in more individuals . Our analysis suggests in general highly effective non-random DNA strand segregation in human colon , small intestine , liver , skin and brain . However , approximately 1% to 5% of divisions in adults do not seem to segregate strands properly and stem cells accumulate additional mutations over time . The reason for this improper segregation could be either wrongly segregated strands during an asymmetric stem cell division or the loss of a stem cell by either a symmetric stem cell differentiation or cell death followed by a symmetric stem cell self-renewal . Arguments are made for both symmetric and asymmetric stem cell divisions in human tissues [15 , 16 , 25–28] . We wondered if our approach provides a mean to distinguish both possibilities . We therefore implemented stochastic simulations of mutation accumulation in either asymmetric dividing stem cell populations with imperfect strand segregation or a stem cell population with a mix of symmetric and asymmetric divisions ( S1 Fig ) . Both scenarios lead to linearly increasing mean and variance of the mutational burden , with small differences in the actual rates . However , as predicted , the ratio of the variance and the mean σ2/μ˜ are in both scenarios independent of time and on average the same ( see also Eq ( 14 ) ) . Interestingly , the distribution of σ2/μ˜ differs . Whereas the variance of the distribution of σ2/μ˜ increases with time for symmetric stem cell divisions , it approximately remains constant for asymmetric stem cell divisions . However , measuring this effect reliably would require measuring the mean and variance of the mutational burden in many more independent samples of many more healthy humans of different ages than the currently available datasets . Hence , lack of resolution in currently available data precludes us to determine the cause of imperfect strand segregations . However , this effect might provide a future mean to quantitate the amount of symmetric self-renewal in human stem cell populations .
Stem cells in fast proliferating healthy adult tissues such as colon have been reported to accumulate approximately 40 new mutations per year [13] ( Fig 3 ) . However , if mutation rates are in the order of 10−9 per base pair per cell division , which seems to be the current consensus and agrees with our measurements here , and the human genome consists of 6 × 109 base pairs , this would on average only allow for 6 to 7 divisions per stem cell per year . This is in contradiction to current measures on stem cell turnover rates in for example healthy colonic crypts [19 , 29] . This discrepancy is resolved by non-random strand segregation , where many stem cell proliferations would not induce novel mutations on the stem cell level and the effective observed mutation accumulation on a population level can remain low despite high stem cell turnover rates . A clear molecular mechanism of strand recognition remains unknown . However , direct and indirect evidence to which our observations may contribute increasingly hint on the importance of strand segregation to maintain genomic integrity within healthy human tissues . Our joined inference of mutation rate and strand segregation probability also reveals that mutation rates per cell division are likely higher than was assumed in previous studies [11] . We therefore find stronger signals of strand segregation in human sequencing data than was thought previously [11] . Our inference neglects the effects of cell-division independent mutations that may contribute to mutational burden in tissues at a low rate . This can lead to an underestimation of the true strand-segregation probability as well as the per-cell mutation rate in human tissues , see S2 Fig . A loss of strand segregation in stem cells implies a 50 to 100 times increased effective mutation rate on the cell population level without any other changes to the intrinsic DNA repair machinery . In a non-homeostatic setting , such as a growing tumour , in which the number of self-renewing cells ( whether they are all or only a subset of cells ) increases , the rate of random strand segregation events is much higher . This effect may contribute to the usually high mutational burden in cancers [30–33] . However , we note that our model has been developed for normal tissue and does not account for chromosomal rearrangements in malignancies , which likely impact the estimation of mutation rates . It is an intriguing thought that early organ growth during development constitutes a very similar situation in which strand segregation is less effective within expanding stem cell populations and the increased rate of mutation accumulation early in life emerges as a natural consequence [34] .
We assume that homeostasis in a healthy adult human tissue is maintained by a constant pool of N stem cells . Each of these stem cells undergoes n cell divisions during a time interval Δt . With each division , a stem cell non-randomly segregates DNA strands with a probability p . If p = 1 the ancestral strand will remain in the stem cell and the duplicated strand will be passed onto a daughter cell that becomes a non-stem cell , whereas p = 0 . 5 implies random strand segregation ( i . e . no strand segregation ) , see Fig 1 . We assume the probability p to be the same for all stem cells and don’t account for possible variation by for example specific mutations that would change strand segregation probabilities for individual stem cells . The non-ancestral duplicated strand inherits on average μL novel mutations , where μ is the mutation rate per base pair per cell division and L the length of the copied genome ( e . g . L ≈ 6 × 109 base pairs in humans ) . Throughout the manuscript we assume a constant mutation rate μ . In principal the mutation rate could depend on time explicitly , e . g . μ → μ ( t ) . However , this would lead to non-linear dependencies , which is not supported by the currently available data , e . g . Figs 3 & 4 . Thus assuming a constant mutation rate is retrospectively justified by the actual change of the mutational burden in human tissues . It follows that for n cell divisions , the probability to segregate parental DNA strands k times is binomially distributed ( k successes in n draws given a success probability of p ) P ( k , n , p ) = ( nk ) pk ( 1−p ) n−k . ( 5 ) This implies that on average E[k , n , p] = np cell divisions do not induce additional mutations in stem cells . However , n ( 1 − p ) cell divisions will increase mutational burden within a single stem cell lineage , each division by a random number χ , given by a Poisson distribution: P ( χ ) = ( μL ) χχ ! e−μL . ( 6 ) The mutational burden χ˜ within a single stem cell lineage consequently increases by χ˜=∑i=1n−kχi . ( 7 ) Exact expressions for the mutational burden μ˜ and variance σ2 for such distributions are known[35] . The mutational burden μ˜ after n stem cell divisions is given by μ˜=E[χ˜]=E[n−k]E[χ]=n ( 1−p ) μL , ( 8 ) and the variance of the mutational burden σ2 is given by σ2=E[n−k]Var[χ]+ ( E[χ] ) 2Var[n−k]=n ( 1−p ) μL+n ( μL ) 2p ( 1−p ) . ( 9 ) These expressions allow quantifying the change of the mutational burden as well as the change of the variance of the mutational burden after a number of Δn divisions per stem cell Δμ˜Δn=μ2˜−μ1˜n2−n1= ( 1−p ) μL , ( 10 ) Δσ2Δn=σ22−σ12n2−n1= ( 1−p ) μL+ ( μL ) 2p ( 1−p ) . ( 11 ) However , in actual data the number of stem cell divisions is unknown and change would be measured in time t . Assuming a constant rate of stem cell proliferations λ we can write Δn = λΔt . This allows us to rewrite above equations for the change of the mean and the variance of the mutational burden over real time t via 1λΔμ˜Δt= ( 1−p ) μL ( 12 ) 1λΔσ2Δt= ( 1−p ) μL+ ( μL ) 2p ( 1−p ) . ( 13 ) Importantly , both the change of the mutational burden Δμ˜Δt as well as the change of the variance of the mutational burden Δσ2Δt can be measured from human somatic mutation data , see Figs 3 & 4 . Furthermore , Eqs ( 12 ) and ( 13 ) imply that the mutational burden as well as the variance of the mutational burden are expected to increase linearly with age in adult tissues . Even if strand segregation is highly effective , mutations still accumulate linearly with age . However , the rate of mutation accumulation is decreased by a factor of ( 1 − p ) . As we neither know the somatic mutation rate μ nor the stem cell proliferation rate λ with certainty , a linear increase in mutational burden with age at most suggests imperfect strand segregation ( e . g . 0 ≤ p < 1 ) . Importantly , the linear increase of both the mean and the variance in time is a result of the sum of Poisson distributed random variables and does by itself not imply the presence or absence of non-random strand segregation . However , the ratio of variance and mean is independent of time t and the stem cell proliferation rate λ Δσ2Δμ˜=1+μLp , ( 14 ) and therefore provides natural bounds for possible mutation rates per cell division μ and strand segregation probabilities p in human tissues , see S1 Fig . Furthermore , rearranging Eq ( 12 ) and substituting μLp=μL−1λΔμ˜Δt into Eq ( 14 ) , the strand segregation probability p and the mutation rate μ disentangle , allowing us independent estimates via p=Δσ2Δμ˜−1Δσ2Δμ˜−1+1λΔμ˜Δt ( 15 ) μL=Δσ2Δμ˜−1+1λΔμ˜Δt . ( 16 ) The relative change of the mutational burden and the variance variance of the mutational burden allow estimates of the mutation rate μ ( per cell division ) and the non-random strand segregation probability p . Estimating the mutation rate as well as the strand segregation probability , we need to measure the change of the mutational burden as well as the change of the variance of the mutational burden . This requires multiple measurements of the mutational burden within single cells of a single individual that ideally would be followed over time . This is unpractical and such data currently does not exist . We therefore measure the mutational burden and variance in multiple cells of multiple individuals of different ages . To calculate the variance and the mean of the mutational burden , we require at least 3 samples per individual , see Figs 3 & 4 . For completeness we also show expressions for the mutation rate μ and p in dependence of stem cell proliferations n . They are given by p=Δσ2Δμ˜−1Δσ2Δμ˜−1+Δμ˜Δn ( 17 ) and μL=Δσ2Δμ˜−1+Δμ˜Δn . ( 18 ) We recognize that our model is based on some assumptions and approximations . For example , telomeres , the protective ends of chromosomes , shorten with each cell division . Upon reaching a critically short telomere length , cells enter senescent . Senescence is not modelled in our model , however we argue that since this is likely to occur at very old ages [21 , 36] , this process is unlikely to influence our results significantly . | Cairn proposed in 1975 that upon proliferation , cells might not segregate DNA strands randomly into daughter cells , but preferentially keep the ancestral ( blue print ) template strand in stem cells . This mechanism would allow to drastically reduce the rate of mutation accumulation in human tissues . Testing the hypothesis in human stem cells within their natural tissue environment remains challenging . Here we show that the patterns of mutation accumulation in human tissues with age support highly effective non-random DNA strand segregation after adolescence . In contrast , during early development in infants , DNA strand segregation is less effective , likely because stem cell populations are continuing to grow . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"small",
"intestine",
"medicine",
"and",
"health",
"sciences",
"cell",
"cycle",
"and",
"cell",
"division",
"cell",
"processes",
"neuroscience",
"cell",
"differentiation",
"oncology",
"mutation",
"developmental",
"biology",
"stem",
"cells",
"digestive",
"system",
"anim... | 2018 | Variation of mutational burden in healthy human tissues suggests non-random strand segregation and allows measuring somatic mutation rates |
Protein modifications regulate both DNA repair levels and pathway choice . How each modification achieves regulatory effects and how different modifications collaborate with each other are important questions to be answered . Here , we show that sumoylation regulates double-strand break repair partly by modifying the end resection factor Sae2 . This modification is conserved from yeast to humans , and is induced by DNA damage . We mapped the sumoylation site of Sae2 to a single lysine in its self-association domain . Abolishing Sae2 sumoylation by mutating this lysine to arginine impaired Sae2 function in the processing and repair of multiple types of DNA breaks . We found that Sae2 sumoylation occurs independently of its phosphorylation , and the two modifications act in synergy to increase soluble forms of Sae2 . We also provide evidence that sumoylation of the Sae2-binding nuclease , the Mre11-Rad50-Xrs2 complex , further increases end resection . These findings reveal a novel role for sumoylation in DNA repair by regulating the solubility of an end resection factor . They also show that collaboration between different modifications and among multiple substrates leads to a stronger biological effect .
Efficient and accurate genome repair requires regulatory mechanisms that adjust DNA repair levels and pathway usage depending on the cellular context . For example , in response to increased lesion loads , DNA repair pathways are upregulated [1]–[3] . Additionally , DNA double-strand breaks ( DSBs ) are repaired by either homologous recombination ( HR ) or non-homologous end joining ( NHEJ ) depending on the cell cycle stage [4] , [5] . The regulatory changes in these situations occur rapidly , involve many targets , and are reversible [1]–[3] . They are often enabled by protein modifications that reversibly add modifier groups to multiple targets . The best-illustrated example of this is protein phosphorylation mediated by the DNA damage checkpoint and cyclin-dependent kinases , which occurs within minutes of changes in repair needs and affects hundreds of protein targets ( e . g . [6]–[9] ) . More recently , another protein modification , sumoylation , has emerged as a key regulator of genome repair ( reviewed in [10]–[13] ) . However , many important details of how sumoylation influences DNA repair have yet to be elucidated . For example , sumoylation is important for DSB repair in humans and yeast partly by promoting DNA end resection [14]–[18] . Yet , it has been unclear for which resection factor ( s ) sumoylation is relevant , how sumoylation influences specific attributes of such target ( s ) , and how this modification is coordinated with phosphorylation-based regulation . To address these questions , we used budding yeast as a model system to examine the sumoylation of a conserved DNA end resection factor , Sae2 . Sae2 collaborates with the Mre11-Rad50-Xrs2 ( MRX ) complex in processing multiple kinds of DSBs , including those with clean ends and ends capped with proteins or hairpin structures ( reviewed in [19] , [20] ) . Sae2 and MRX can remove the capping structure and 100–300 bp of single-stranded DNA from DSBs in a process called end clipping [21]–[29] . This first stage of DSB end resection is followed by long-range resection via parallel pathways mediated by the Exo1 exonuclease and the Sgs1/Dna2 helicase-nuclease pair [22] , [24] . End clipping commits DSB repair to HR , as resected DNA ends are poor substrates for NHEJ . This commitment point is tightly regulated in conjunction with cell cycle phase [30]–[32] , as NHEJ is more beneficial in G1 when sister chromatids are absent , whereas recombination constitutes more faithful repair during S and G2 phases when the synthesized sister chromatids provide accurate repair templates . Previous studies have shown that kinases confer cell cycle-dependent regulation of end clipping [31]–[37] . Both S phase cyclin-dependent kinase ( CDK ) and DNA damage checkpoint kinases phosphorylate Sae2 to promote end clipping [33] , [36] , [37] . This is achieved at least partly by dynamically increasing Sae2 protein solubility [37] . This form of regulation is critical as Sae2 is predominantly present as inactive aggregates in G1 , presumably to limit resection in this phase [36] , [37] . Upon treatment with DNA damaging agents in S phase , phosphorylation of Sae2 facilitates the rapid release of active monomeric and dimeric forms from the inactive aggregates to promote end clipping , and thus HR [37] . Here , we show that the conserved sumoylation of Sae2 also promotes its functions in the processing and repair of multiple kinds of DSBs . Interestingly , like phosphorylation , sumoylation also increases the levels of soluble Sae2 . We show that the two different modifications act in synergy to achieve a stronger effect on Sae2 function . Moreover , we present evidence that sumoylation of MRX also favors resection , suggesting that the coordinated sumoylation of multiple substrates leads to greater biological consequences .
We and others recently reported that five proteins involved in DNA end resection are sumoylated upon DNA damage in budding yeast [17] , [18] . Here we examined the sumoylation of the Sae2 protein . The Sae2 sumoylated form migrates ∼20 kDa higher than the unmodified form upon SDS-PAGE , as expected from the typical up-shift caused by mono-sumoylation ( Fig . 1A ) . The sumoylated form can be preferentially detected by an antibody against the SUMO moiety ( Fig . 1A and [17] ) . In addition , this modification was abolished by the simultaneous removal of both the homologous Siz1 and Siz2 SUMO ligases , but not of either single ligase ( Fig . 1A ) . These results validate Sae2 sumoylation and indicate that the Siz ligases redundantly sumoylate Sae2 . To evaluate the functional consequences of Sae2 sumoylation , we identified the lysine that is targeted for sumoylation . Sae2 possesses two sumoylation consensus motifs , ψKxE/D , where ψ is a large hydrophobic acid [38] , [39] . Mutating one of these sites , K97 , to arginine abolished its sumoylation in vivo ( Fig . 1B ) . This residue is conserved among Sae2 orthologs in closely related Saccharomyces species ( Fig . S1A ) . To confirm that K97 is the SUMO conjugation site , Sae2 was co-expressed with sumoylation enzymes in E . coli to enable its sumoylation ( see Methods ) . A higher-migrating form of Sae2 in the purified protein prep was specifically eliminated by treatment with the desumoylase Ulp1 , indicating that it is the sumoylated form of Sae2 ( Fig . S1B ) . Consistent with the in vivo finding , Sae2-K97R mutant protein was not sumoylated in vitro ( Fig . 1C ) . Together , our in vivo and in vitro data indicate that lysine 97 is the bona fide SUMO conjugation site on Sae2 . K97 is located within the N terminal domain that is important for self-association in several organisms ( Fig . 1D and [37] , [40]–[42] ) . As conserved modification of a protein in different organisms is indicative of functional importance , we examined whether Sae2 orthologs that share DNA resection functions are also targeted for sumoylation . To this end , we subjected recombinant Sae2 orthologs , namely fission yeast Ctp1 and human CtIP , to sumoylation in E . coli using reconstituted fission yeast and mammalian SUMO conjugating systems , respectively . Both SpCtp1 and hCtIP exhibited a slow migrating modified form only upon co-expression of SUMO and conjugating enzymes ( Fig . 1E–1F ) , suggesting that they can be sumoylated . The conserved sumoylation of Sae2 orthologs supports the notion that this modification can be functionally relevant . Next , we investigated whether abolition of Sae2 sumoylation affects its functions by studying the sae2-K97R allele . sae2-K97R did not affect Sae2 protein levels in normal growth conditions or after genotoxin treatment at either 30°C or 37°C , excluding an effect of sumoylation on general protein stability ( Fig . 2A and S1C Fig . ) . We then tested whether sae2-K97R affects the processing and repair of complex DNA ends , such as those capped by hairpin structures or covalently linked with proteins , using established assays . To query hairpin repair in vivo , we measured recombination that requires removal of hairpins formed through inverted Alu sequences at DSBs [27] , [43] . Consistent with a previous report , deleting SAE2 greatly reduced the recombination rate measured in this assay ( Fig . 2B and [27] , [43] ) . sae2-K97R showed a 2-fold reduction ( Fig . 2B ) , suggesting a moderate deficiency of Sae2 function in hairpin removal . To examine processing of DSB ends that are covalently linked with proteins , we first examined Sae2-mediated processing of DSBs capped with Top1 , which are induced upon camptothecin ( CPT ) treatment [29] , [44] . Consistent with previous reports , sae2Δ cells were sensitive to CPT ( Fig . 2C and [29] , [44] ) . sae2-K97R cells exhibited an increase in sensitivity to CPT at 37°C ( Fig . 2C ) , suggesting that sumoylation of Sae2 contributes to CPT repair . Consistent with this , Sae2 sumoylation is induced by CPT treatment and elevated temperature ( S1D Fig . ) . Second , we examined sporulation efficiency , as Sae2-mediated removal of Spo11 conjugated to DSB ends in meiosis is required for sporulation ( Fig . 2D and [28] , [45]–[47] ) . sae2-K97R homozygous diploid cells exhibited a reproducible 20% reduction in this assay , indicating a moderate defect ( Fig . 2D , p<0 . 05 ) . Taken together , these results show that sae2-K97R is partially defective in the processing and repair of complex DSBs . We proceeded to assess whether Sae2-mediated end clipping of clean DSB ends is affected by sae2-K97R . In yeast , end clipping can be directly assayed at DSBs induced by the endonuclease HO at the MAT locus [22] . As shown previously , because end clipping is an intermediate stage in end resection , it can be best monitored when the downstream extensive resection step is blocked by removing Sgs1 and Exo1 [22] . Both qPCR- and Southern blot-based assays can be used to assess Sae2 function in this setting . The two assays take advantage of the fact that single-stranded DNA generated by resection is resistant to restriction enzyme digestion . In the qPCR-based assay , PCR products amplified using primers flanking a StyI site located 700 bp distal to the DSB are compared between digested and undigested samples ( Fig . 2E , left panel , and [24] , [30] , [48] ) . PCR products from a control locus , ADH1 , are used for normalization ( see Methods ) . Using this assay , we found that sae2-K97R exhibited 60–80% of the wild-type level of resection in a time course of 120 minutes in the sgs1Δ exo1Δ background ( Fig . 2E , right ) . The lethality of sae2Δ sgs1Δ exo1Δ prevents comparison of sae2-K97R defects with sae2Δ in this setting [22] . In the Southern blot assay , end clipping products run as a smear of bands below the HO cut ( unprocessed ) fragment , and both types of bands are detected by a radio-labeled probe specific to a sequence flanking the DSB ( S2A Fig . and [22] ) . In sgs1Δ exo1Δ cells , the intensity of the smear moderately increased as the unprocessed fragment diminished with time ( S2A–S2C Fig . ) , signifying the progress of end clipping [22] . Introducing the sae2-K97R mutation reduced end clipping efficiency , as seen by the persistence of the unprocessed fragment and decreased intensity of the smear below ( S2A–S2C Fig . ) . Quantification of three independent strains indicated a reduction of up to 50% in the fraction of end clipping products among total cut fragments ( Fig . S2C ) . Taken together , both the qPCR- and Southern blot-based assays show that lack of Sae2 sumoylation impairs end clipping of clean DSBs . As end clipping disfavors NHEJ , its impairment would promote NHEJ [36] , [49] , [50] . Accordingly , a prediction of the observed end clipping defect in sae2-K97R sgs1Δ exo1Δ cells compared with sgs1Δ exo1Δ cells ( Fig . 2E and S2A–S2C Figs . ) is that the former should have higher NHEJ levels . Indeed , we detected an ∼30% increase in NHEJ in the triple mutant compared with the double , using a standard chromosomal NHEJ assay ( S2D Fig . ) . As this assay was performed side-by-side with the Southern blot-based resection assay , equal efficiency of HO cleavage between genotypes was confirmed ( S2A Fig . ) . We also used a well-established plasmid-based NHEJ assay in which cells are transformed with linearized or undigested plasmid DNA , and survival on selective media serves as a readout for successful repair by NHEJ [51] . sae2-K97R again exhibited a moderate increase in this assay in the sgs1Δ exo1Δ background , while its effect in the SGS1 EXO1 background was not statistically significant ( Fig . 2F ) . We noticed that sae2-K97R sgs1Δ exo1Δ showed more resistance to the DNA damaging agent methyl methanesulfonate ( MMS ) than sgs1Δ exo1Δ ( Fig . 2G ) . This is in contrast to the inviability of sae2Δ sgs1Δ exo1Δ [22] . One interpretation is that moderate reduction of end clipping in the absence of extensive resection allows more NHEJ , thus better survival , whereas complete loss of resection confers lethality even with endogenous levels of DNA damage . Supporting this idea , the higher MMS resistance of sae2-K97R sgs1Δ exo1Δ depends on the NHEJ factor Dnl4 ( Fig . 2G ) . These findings and the increased NHEJ seen for sae2-K97R are consistent with this mutant's impairment in end resection ( Fig . 2E and S2A–S2C Fig . ) . As phosphorylation of Sae2 is required for its resection function and DNA damage resistance [33] , [36] , [37] , we asked whether sae2-K97R interferes with this modification . The phosphorylated forms of Sae2 manifest as slower migrating bands , and mutating two main Mec1/Tel1 phosphorylation sites and an adjacent serine , namely S249 , S278 and T279 ( sae2-3A , [33] , [37] ) results in the loss of the top bands ( Figs . 1D and 3A ) . We found that Sae2-K97R exhibited a similar mobility shift as its wild-type counterpart ( Fig . 3A ) , suggesting that sumoylation of Sae2 does not interfere with its phosphorylation . To further elucidate the interplay between the two modifications , we assayed the sumoylation levels of Sae2 phosphorylation mutants . Neither sae2-3A nor sae2-S267A , which abrogates CDK-mediated phosphorylation , affected Sae2 sumoylation ( Fig . 3B–3C ) . Consistent with this , mec1Δ cells exhibited normal levels of Sae2 sumoylation ( Fig . 3D ) , despite being deficient for Sae2 phosphorylation [33] . Together , these results show that sumoylation of Sae2 does not require its phosphorylation . As phosphorylation and sumoylation of Sae2 occur independently , and both contribute to Sae2 function , we examined whether their biological effects are additive . We found that combining the K97R and 3A mutations , or the K97R and S267A mutations , resulted in greater sensitivity to MMS and CPT compared to mutants that were defective for only one modification ( Fig . 3E–3F ) . These results indicate that sumoylation and phosphorylation of Sae2 make separate contributions to DNA damage resistance . We proceeded to examine how sumoylation of Sae2 affects its function . As shown recently , an important means of regulating Sae2 by protein modification is through increasing its solubility [37] . Sae2 is primarily in inactive aggregate forms in G1 , whereas the soluble and active fraction of Sae2 increases upon entering S phase in the presence of DNA damage [37] . Phosphorylation of Sae2 by Mec1 and S-CDK promotes this increase [37] . Considering the genetic interaction between the two types of Sae2 modifications , we examined if sumoylation also alters the levels of soluble Sae2 . Using an established solubility assay , we examined G1-arrested cells after release into 0 . 03% MMS [37] . Consistent with our genetic data , levels of soluble Sae2-K97R-3A protein were significantly lower than that of Sae2-3A after cells were released from G1 ( Fig . 4A ) . We also observed a similar but smaller decrease in levels of soluble Sae2-K97R when compared with that of wild-type Sae2 ( Fig . 4B ) . In this case , the level of soluble Sae2-K97R protein is decreased by ∼25% compared to wild-type in a cell cycle-independent manner , suggesting that sumoylation by itself also affects Sae2 solubility . This reduction is less severe than that of Sae2-3A alone , which showed an S phase-specific decrease in Sae2 soluble levels by up to 50% , compared with wild-type Sae2 ( Fig . 4C ) . These results suggest that while both phosphorylation and sumoylation promote Sae2 solubility , the former has a stronger effect . As the solubility difference between Sae2-3A and Sae2-K97R mutants is only 25% , yet their MMS sensitivities differ greatly , it is possible that the Sae2-3A has additional defects besides solubility . Because sae2-K97R exhibited mild resection defects and genotoxin sensitivity ( Fig . 2 and S2 Fig . ) , and not to the level that has been reported for SUMO E2 ( e . g . [17] ) , we reasoned that sumoylation likely wields a strong influence on this process by additionally targeting other factor ( s ) , such as MRX . As mapping sumoylation sites on the three subunits of MRX proved difficult , we devised a genetic strategy to reduce MRX sumoylation . It has been shown that the catalytic domain of the de-sumoylating enzyme Ulp1 ( UD ) when fused with a protein can lead to the targeted removal of SUMO conjugated to the protein or its interactors [52] . A fusion with mutations of four residues required for catalytic activity and SUMO interaction ( UD* ) was used to control for the effect of tagging with this domain [52] . MRX physically interacts with the Ku complex , which arrives at DSBs concomitantly with MRX [53] , [54] . We tested if fusing the UD domain to the Ku70 subunit ( YKU70-UD ) can specifically decrease MRX sumoylation . To this end , we introduced the YKU70-UD or YKU70-UD* constructs at the endogenous YKU70 locus with its native promoter . As shown in Fig . 5A , sumoylation of all three subunits of MRX was either abolished or strongly reduced in cells expressing YKU70-UD compared with YKU70-UD* control cells . To assess the specificity of desumoylation , we examined proteins that arrive at DSBs around the same time as MRX [53] , [55] . YKU70-UD did not affect the sumoylation of Sae2 , or the Ku-interacting protein Lif1 ( Fig . 5B ) . In addition , sumoylation of the downstream recombination proteins Rad1 and Saw1 was not affected by YKU70-UD ( Fig . 5B ) . Together , these results suggest that YKU70-UD can limit MRX sumoylation with good specificity . We then examined whether YKU70-UD has a phenotype indicative of defective MRX-mediated resection . As MRX deficiency can exacerbate sae2Δ sensitivity to DNA damaging agents in certain contexts [44] , we tested whether reducing MRX sumoylation by YKU70-UD causes a similar phenotype . Indeed , we found that YKU70-UD worsened the MMS sensitivity of sae2Δ cells , while YKU70-UD* conferred suppression ( Fig . 5C ) . The latter effect is likely due to the tag's interference with Ku function , such as in inhibiting HR by Exo1 exclusion [44] , [56] , [57] . That YKU70-UD is additive with sae2Δ suggests that the defects caused by reduced MRX sumoylation overrides any suppression conferred by defective Ku function . This in turn suggests the possibility that reduced MRX sumoylation impairs its resection function . To test this idea , we examined resection dynamics in YKU70-UD vs . YKU70-UD* cells . In both qPCR- and Southern blot-based assays , DSB resection was decreased by 15–20% in YKU70-UD cells compared to YKU70-UD* , most obviously at early time points ( Fig . 5D–5E and S3A–S3B Fig . ) , suggesting that sumoylation of MRX facilitates resection . To assess if the sumoylation of MRX and Sae2 independently promotes resection , we measured end resection in YKU70-UD sae2-K97R cells . As shown in Fig . 5E , sae2-K97R further compromised resection in YKU70-UD , but not YKU70-UD* , cells . The moderate additivity in resection defects did not result in exacerbation of YKU70-UD's MMS sensitivity by sae2-K97R ( Fig . 5C ) , likely because it is insufficient to confer MMS sensitivity or YKU70-UD exerts compensatory effects due to impaired Ku function . Taken together , our results suggest that sumoylation of MRX , in addition to that of Sae2 , contributes to resection .
Regulation of DNA repair pathway levels and capacity in response to cell cycle changes and lesion loads is important for genome maintenance and damage resistance . Despite recent progress , many forms and targets of regulation are not yet identified or understood . One of the most highly regulated DNA repair proteins is the end resection factor , Sae2 . Its solubility is tightly controlled such that its activity and other functions are limited in G1 phase and increased in S phase under damage conditions [37] . We reveal here that regulation of Sae2 solubility is partly mediated by sumoylation . We also show that sumoylation collaborates with checkpoint-dependent phosphorylation in facilitating Sae2 function in end clipping . Furthermore , we provide evidence that sumoylation promotes DNA end resection by additionally targeting the MRX nuclease . This work reveals a novel mechanism of SUMO-mediated regulation of DNA repair and uncovers an example wherein sumoylation and phosphorylation , as well as multiple sumoylation events , collaborate to promote nuclease function ( see model in Fig . 5F ) . Proteins that cleave DNA are double-edged swords , and unscheduled DNA cleavage has to be minimized . In general , upstream constraints such as limiting recruitment to lesions can restrict the activity of downstream factors ( e . g . [58]–[61] ) . As Sae2 is one of the first resection proteins to arrive at DSBs without any known recruiters [55] , it is not surprising that it is subjected to other forms of regulation . Our findings and previous reports strongly suggest that Sae2 regulation can be achieved by different protein modifications that collaborate to ensure timely availability of the active forms of the protein [37] , [62] , [63] . Thus far , regulation of protein solubility by sumoylation has been reported only for proteins involved in neuronal diseases ( e . g . [64]–[66] ) . We now provide the first example wherein sumoylation regulates the solubility of a DNA metabolism protein . As protein aggregation is a widespread phenomenon caused by high intrinsic aggregation potential of the protein ( e . g . [67] , [68] ) , it is conceivable that SUMO-mediated protein solubilization is a general effect . Such an effect by SUMO may be similar to its promotion of solubility in recombinant protein applications [69] , [70] . Thus , in addition to the previously proposed glue effect of sumoylation in bridging interactions in complexes [18] , [71] , [72] , sumoylation can also have the opposite effect of “anti-glue” to disperse proteins from aggregates . As sumoylation occurs in the Sae2 self-association domain and at a region of high aggregation potential ( S4A Fig . and [41] , [73] ) , steric or charge changes conferred by sumoylation in these regions can disfavor aggregation . Several independent analyses show that lack of Sae2 sumoylation moderately reduces end resection and increases NHEJ ( Fig . 2B–2F and S2A–S2D Fig . ) . The correlation of these effects with changes in the levels of soluble Sae2 suggests that the decreased availability of active Sae2 can at least partly account for the end resection defects and NHEJ increase , though other possibilities such as defective DSB recruitment cannot be excluded . As the resection defect of sae2-K97R is less severe than that of the SUMO E2 mutant , sumoylation of additional resection factors also likely matters . Indeed , reduction of MRX sumoylation also impairs resection , and in a manner additive with sae2-K97R ( Fig . 5 ) . Although a thorough examination of MRX sumoylation awaits mapping of sumoylation sites on all three subunits , these results suggest that sumoylation achieves a large biological effect by simultaneously inducing small changes in multiple substrates ( “ensemble effect” ) . This suggestion is consistent with the observations that several dozen repair proteins are sumoylated upon DNA damage [17] , [18] , [74] , and individual non-sumoylatable mutants usually exhibit only mild phenotypes ( e . g . [75]–[77] ) . We propose that the ensemble effect model is common in DNA repair regulation and other processes to confer robustness to a system . We also note that the usefulness of this strategy is also seen for other protein modifications ( e . g . [32] , [62] ) , and that as in the case of sumoylation , the effects of a particular modification are unique to the substrate , rather than conforming to a general mechanism ( e . g . [8] , [78]–[80] ) . In summary , our work provides strong evidence for a new role for sumoylation in regulating DNA repair and its collaboration with phosphorylation-based regulation . Considering that only a few sumoylated substrates in DNA repair have been examined in detail thus far , future studies on additional substrates and the interplay between sumoylation and other forms of regulation will greatly expand our knowledge of how repair pathway levels and choice are determined in cells .
Strains used are listed in Table 1 . Only one strain per genotype is shown for simplicity , but at least two strains per genotype were tested in each assay . Standard yeast protocols were used for strain generation , growth and medium preparation . As siz1Δ siz2Δ results in amplification of the 2 micron plasmid [81] , strains with siz1Δ siz2Δ mutations were cured of the plasmid as described [82] . Detection of the sumoylated form of Sae2 was performed as described previously [17] . In brief , log phase cultures were treated with 0 . 3% methyl methanesulfonate ( MMS , Sigma-Aldrich ) or 50 ug/ml camptothecin ( CPT , Sigma-Aldrich ) or at 37°C for 2 h . Cells were lysed by bead beating under denaturing conditions , and TAP-or HA-tagged proteins were immunoprecipitated . These were then washed and eluted with loading dye , followed by SDS-PAGE and western blotting with antibodies against SUMO [83] , the protein A portion of the TAP tag ( Sigma-Aldrich ) or HA ( 12CA5 ) . We note that as the Fc portion of the SUMO antibody interacts with the Protein A part of TAP , it also detects the unmodified protein , but more strongly so for the sumoylated form because of additional high affinity for SUMO . Protein preparation for detecting Sae2 phosphorylation and protein levels was performed as described [33]; DNA damage treatment was performed as above . Assay was performed essentially as described [37] except that all Sae2 constructs were expressed from its own chromosomal locus . In brief , G1-arrested cells were released into 0 . 03% MMS and samples were harvested at the indicated time points for protein and FACS analyses . Upon complete cell lysis by bead beating and removal of DNA by DNaseI treatment , cell extract was centrifuged at high speed ( 14k rpm for 30 min ) to separate the soluble fraction from the insoluble . The soluble fractions were analyzed by SDS-PAGE and western blotting against the tag . The housekeeping protein Adh1 was used as loading control as its levels are invariant during the time course . FACS analyses show proper arrest and release for all the strains examined . We note that Sae2 is sumoylated in both G1 and S phases during this procedure ( S4B Fig . ) . To assess Sae2 soluble forms in different strains , two spore clones of each genotype were examined in at least two independent tests . For quantification , we compared solubility between the two strains for each time point . In brief , we first determined the soluble Sae2 protein level relative to loading control for each strain at each time point , and then calculated the ratio between the genotypes to represent it in Fig . 4 . The student's t test statistical analysis was performed for “Sae2 protein level relative to loading control” between the two genotypes from 6 repeats ( 2 trials with 3 spores ) . Both GST-tagged Sae2 and hCtIP were sumoylated in E . coli by co-expression with E1 ( Aos1-Uba2 ) , E2 ( Ubc9 ) and SUMO-1 [84] ( the pT-E1E2S1 plasmid was a gift from Dr . Hisato Saitoh ) . Plasmids for expression of GST-tagged Sae2 and hCtIP are derivatives of pGEX-4T1 and were a gift from Dr . Stephen Jackson [36] , [85] . The plasmid expressing GST-Sae2K97R ( pRS72 ) was made by site-directed mutagenesis of pGEX-4T1-Sae2 . Plasmids were transformed into BL21 ( DE3 ) cells , and single colonies were used to inoculate overnight cultures of LB ( plus 100 ug/ml ampicillin or 25 ug/ml ampicillin and 17 ug/ml chloramphenicol ) , which were incubated at 30°C with shaking at 250 rpm . These starter cultures were used to inoculate fresh LB ( with ampicillin plus chloramphenicol added as required ) cultures to an OD600 of 0 . 1 , which were then grown at 30°C with shaking at 250 rpm to an OD600 of 1 . 2 . The temperature was then lowered to 25°C and 250 uM IPTG added to induce expression of the proteins . The cultures were incubated for another 16 h at 25°C before harvesting by centrifugation . The cell pellet was resuspended in PBS ( pH 7 . 3 ) supplemented with 1 mM PMSF and 5 mM DTT . After sonication and centrifugation ( 47000×g at 4°C for 20 min ) , the soluble protein fraction was loaded onto an equilibrated Glutathione Sepharose Fast Flow column ( GE Healthcare Life Sciences ) . The column was washed with 10 column volumes of PBS and the GST-Sae2 and GST-Sae2-SUMO proteins eluted with 2 column volumes of 50 mM Tris-HCl ( pH 8 . 0 ) , 30 mM reduced glutathione . Proteins were dialyzed against 20 mM Tris , pH 8 . 0 containing 150 mM NaCl at 4°C . To perform SUMO cleavage reactions , recombinant purified Ulp1 ( 10 nM ) [86] was added to 5 uM partially purified GST-Sae2/GST-Sae2-SUMO and incubated at 23°C for 30 min in buffer containing 25 mM Tris-HCl ( pH 8 . 0 ) , 150 mM NaCl , 0 . 1% Tween-20 , and 2 mM DTT . The proteins were separated on a 10% SDS PAGE gel and analyzed on a western blot probed with an anti-GST antibody ( GE Healthcare Life Sciences ) . 3HA-tagged SpCtp1 was sumoylated in E . coli BL21 ( DE3 ) cells by co-expression with the S . pombe E1 ( Rad31+Fub2 ) , E2 ( Hus5 fused to a 6×His-tag ) and SUMO ( Pmt3GG ) . The 6×His-Hus5 is fused to 3HA-SpCtp1 and Pmt3GG is tagged with GST . Full details of the plasmid constructs will be provided elsewhere . Transformed BL21 ( DE3 ) cells were cultured and proteins were isolated as described above . These were performed as described [43] . Single colonies were picked from streakouts and allowed to grow for 3 days . Each single colony was resuspended in 0 . 25 ml water by vortexing ( 0th dilution ) and ten-fold serial dilutions were prepared . 100 ul of the 5th dilution was plated onto complete medium , while 100 ul of the 2nd dilution ( or 0th dilution for sae2Δ cells ) was plated onto medium lacking lysine . Successful recombination by processing Alu-generated hairpin DSBs generates LYS+ colonies . Fourteen colonies were analyzed in this manner for each genotype , and the recombination rate was calculated by fluctuation analysis . Both qPCR- and Southern blot-based assays were performed as described [17] , [48] . For both assays , a DSB at the MAT locus was introduced by galactose-induced expression of the HO endonuclease throughout the time course either in asynchronous ( Figs . 2E and 5D , S2A–S2C and S3A–S3B Figs . ) , or G2-arrested cultures ( Fig . 5E ) . Samples were collected at the indicated time points . Genomic DNA was isolated and an aliquot was subjected to digestion with XbaI and StyI . For Southern blot-based method , digested DNA was subjected to native agarose gel electrophoresis , transferred to Hybond XL ( GE Healthcare ) membranes , and hybridized with radiolabeled DNA probes . Quantification of intensities of bands on the Southern blots was done using ImageGauge . DSB end resection at each time point was calculated as the ratio of the signal intensity at that time point to that at the first time point after HO induction . Note that as Sae2 sumoylation was strongly increased at 37°C ( S1D Fig . ) , sae2-K97R phenotype in the above assays ( except Fig . 5E ) was examined at this temperature . qPCR-based resection assay was performed as described [48] . In brief , 150 ng of genomic DNA isolated as above was subjected to restriction enzyme digestion with StyI or mock-digested in a reaction volume of 15 ul . DNA was diluted by addition of 55 ul of ice-cold dH2O . 8 . 8 ul of the diluted DNA was used for each qPCR reaction in a total volume of 20 ul . Primer sequences are specified in [48] . PCRs were performed using SsoAdvanced Universal SYBR Green Supermix ( Bio-Rad ) with the Bio-Rad DNA Engine Chromo 4 system and corresponding software ( Opticon ) . All reactions were amplified using the following program: 95°C for 10 min , 40 cycles of ( 95°C for 15 s , followed by 58°C for 60 s ) , and melting curve 10 min . Reactions were set up in triplicates for all primer pairs and the resulting average threshold cycle ( Ct ) value was used for calculation . The percentage of DNA resected to 0 . 7 kb in HO-cut DNA was calculated by x = 200/{ ( 1+2ΔCt ) *f} , where ΔCt = Ct , digestion−Ct , mock , and f is the fraction cut by HO as quantified by Southern blot analysis . All Ct values were corrected for DNA concentrations by comparing with values for amplification at the ADH1 locus . For both resection assays , at least two spore clones of each genotype were examined in two or more trials . The analysis of chromosomal NHEJ levels was performed as previously described [87] . DSB induction was induced for 1 . 5 h , was performed side-by-side with the resection assay ( compare cleavage efficiency at 1 . 5 h ) . DSBs were induced in cells that cannot repair the break by HR and rely on NHEJ for repair; thus , NHEJ proficiency can be discerned by comparing the numbers of colonies that survive transient DSB induction . Plasmid-based NHEJ assay was performed by transforming either 1 ng of undigested or 20 ng of BamHI-digested pRS416 plasmid carrying URA3 into competent cells , and plating on medium lacking uracil . For yku70Δ control cells , 100 ng of digested plasmid was used for transformation . Successful NHEJ repair results in ligation of the linearized plasmid and thus growth on -URA medium . Transformation efficiency was calculated as the number of colonies on -URA medium divided by the amount of DNA transformed . NHEJ repair for each genotype was calculated as the ratio of transformation efficiencies of digested to undigested samples . For both NHEJ assays , at least two spore clones of each genotype were examined in two or more independent trials . Spot assays were performed as described previously [17] . Briefly , log phase cells were diluted 10-fold and spotted onto YPD media with or without CPT or MMS . Plates were incubated at 30°C ( Fig . 5C ) or 37°C ( Figs . 2D , 2G and 3E–3F ) , and photographed after 24–72 h . At least two spore clones of each genotype were examined in two or more independent trials . Sporulation assay was performed essentially as described [36] . Diploid SK1 cells were grown overnight in YPD medium , washed twice with warm sporulation medium , and left in sporulation medium for 36 h at 30°C . The percentage of sporulated cells was determined by light microscopy . | Proper repair of DNA lesions is crucial for cell growth and organism development . Both the choice and capacity of DNA repair pathways are tightly regulated in response to environmental cues and cell cycle phase . Recent work has uncovered the importance of protein modifications , such as phosphorylation and sumoylation , in this regulation . Sumoylation is known to be critical for the efficient repair of highly toxic DNA double-strand breaks in both yeast and humans , and this is partly mediated by influencing DNA end resection . However , it has been unclear for which resection factor sumoylation is important , how sumoylation influences specific attributes of the relevant targets , and how this modification is coordinated with phosphorylation-based regulation . Here , we provide exciting new insights into these issues by revealing that 1 ) a conserved end resection factor is a SUMO target relevant to this process , 2 ) this regulation favors a specific repair pathway , 3 ) sumoylation collaborates with phosphorylation to promote protein solubility , and 4 ) sumoylation influences DNA repair via an “ensemble effect” that entails simultaneous small alterations of multiple substrates . Our work reveals both a novel mechanism and a general principle for SUMO-mediated regulation of DNA repair . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biology",
"and",
"life",
"sciences"
] | 2015 | Sumoylation Influences DNA Break Repair Partly by Increasing the Solubility of a Conserved End Resection Protein |
Timing of cell division is coordinated by the Septation Initiation Network ( SIN ) in fission yeast . SIN activation is initiated at the two spindle pole bodies ( SPB ) of the cell in metaphase , but only one of these SPBs contains an active SIN in anaphase , while SIN is inactivated in the other by the Cdc16-Byr4 GAP complex . Most of the factors that are needed for such asymmetry establishment have been already characterized , but we lack the molecular details that drive such quick asymmetric distribution of molecules at the two SPBs . Here we investigate the problem by computational modeling and , after establishing a minimal system with two antagonists that can drive reliable asymmetry establishment , we incorporate the current knowledge on the basic SIN regulators into an extended model with molecular details of the key regulators . The model can capture several peculiar earlier experimental findings and also predicts the behavior of double and triple SIN mutants . We experimentally tested one prediction , that phosphorylation of the scaffold protein Cdc11 by a SIN kinase and the core cell cycle regulatory Cyclin dependent kinase ( Cdk ) can compensate for mutations in the SIN inhibitor Cdc16 with different efficiencies . One aspect of the prediction failed , highlighting a potential hole in our current knowledge . Further experimental tests revealed that SIN induced Cdc11 phosphorylation might have two separate effects . We conclude that SIN asymmetry is established by the antagonistic interactions between SIN and its inhibitor Cdc16-Byr4 , partially through the regulation of Cdc11 phosphorylation states .
Cell division is a fundamental and conserved process in all eukaryotes . The fission yeast Schizosaccharomyces pombe has already proved to be a very simple yet interesting model system to study and analyze eukaryotic cell division [1]–[3] . The onset of cytokinesis must be tightly coupled to the completion of mitosis for proper segregation of chromosomes into two daughter cells . In fission yeast , the initiation of cell division is controlled by a conserved signaling pathway known as the Septation Initiation Network or SIN [4]–[9] . Regulation of the SIN happens at the spindle pole bodies ( SPBs ) of fission yeast cells , where the scaffold proteins Cdc11 and Sid4 localize the rest of the molecules in the network [10] , [11] . At the top of the pathway sits the GTPase Spg1 , which controls a protein kinase pathway that triggers actomyosin ring contraction and positively regulates septum formation [12] . The Cdc16-Byr4 GAP complex negatively regulates SIN by inactivating Spg1 [13] . During interphase Cdc16-Byr4 keeps Spg1 inactive , but in metaphase the GAP complex is removed from SPBs , allowing the accumulation of the Cdc7 kinase to both SPBs [14] . As cells enter into anaphase Spg1-GTP gets hydrolyzed by the appearing Cdc16-Byr4 complex and Cdc7 disappears from the old SPB ( that was existing already in the mother cell [15] ) . At the same time Cdc7 level rises at the new SPB with Spg1 remaining in GTP bound form and without the presence of Cdc16-Byr4 [16]–[18] . Such asymmetric segregation of the active SIN ( Spg1-GTP and Cdc7 ) , and its inhibitory complex ( Cdc16-Byr4 ) is essential for proper activation and eventual inactivation of the SIN [19] . The role of this asymmetry was investigated recently and it was found that phosphorylation-dephosphorylation events on the scaffold protein Cdc11 by the downstream SIN kinase Sid2 and the SIN Inhibitory Phosphatase complex ( SIP ) play important roles in the establishment of SIN asymmetry between SPBs [20] , [21] . Still the detailed molecular mechanisms that ensure efficient and fast asymmetry establishment and turning off of SIN activity after cell division is not well understood [19] . Here we develop mathematical models of increasing complexity to understand what basic features such an asymmetry generating system might contain and what known interactions of SIN and its regulators might be important for such features . Mathematical modeling was already successfully used to capture dynamical features of the timing of SIN activation [4] and the orthologous pathway in budding yeast was also investigated this way [22] . Future experimental and modeling work will be needed to merge all knowledge on the spatio-temporal regulation of the SIN into a detailed model that could capture all molecular regulatory interactions in a quantitative way . Here we make the first steps on this line by focusing on the dynamics and regulation of SIN asymmetry establishment in a qualitative fashion .
The minimal mechanism whereby asymmetry could be established between the two SPBs needs to contain some type of positive feedback loop , which involves a non-linear step [23] , [24] . These are the minimal requirements to reach bistability , where one SPB ends up in a steady state with active SIN , while the other settles in an inactive SIN steady state . The two SPBs communicate through releasing and anchoring molecules from the cytoplasmic pool , thus these binding-unbinding steps could be the ideal ones to be controlled by the interacting molecules . Pure autocatalytic positive feedbacks could enforce collection of most of these autocatalytic molecules at one SPB , but that would not ensure that the other molecule type ends up at the other SPB ( not shown ) . Thus the simplest way of implementing a positive feedback loop that can bring the two molecule types to the opposite SPBs should be based on a double-negative type positive feedback loop [25] . In such a minimal model molecule X removes molecule Y from the SPBs , while molecule Y induces the unbinding of molecule X ( Fig . 1A ) . In this way both components remove their own inhibitor and with this they positively influence their own binding to the SPB . If X has a little bias at one of the SPBs it will remove all of Y from this place and help its own recruitment to this SPB . At the same time Y can pile up at the other SPB , since its inhibitor X was moved to the other SPB . Indeed Y speeds up the removal of X from this place and by this , speeds up the establishment of asymmetry . Computational simulation of such a minimal model shows that with a little noise in the initial amounts of X and Y at SPBs or a minimal ( 0 . 1% ) bias in the binding rate to the old SPB is enough to induce asymmetry from a symmetric initial condition ( Fig . 1B ) . The molecular interactions of Fig . 1A were translated into the computational model with a non-linear enzymatic reaction step for the action of X on Y unbinding ( see Materials and Methods for details ) . Thus a model with antagonistic interactions of two molecule types , with ( in biology often observed ) non-linear kinetics can serve as a minimal model of asymmetry establishment between two SPBs . Next we investigated if we have any evidence for the existence of such an antagonistic , double-negative feedback loop among regulators of cytokinesis timing in fission yeast cells . The SIN can be considered as a linear pathway from Spg1 through Cdc7 and Sid1 activation , leading eventually to the recruitment and activation of Sid2 [6] , [7] . The Cdc16-Byr4 complex inhibits Spg1 and as a result Cdc7 binding to the SPB , thus it is a negative regulator of SIN . It was also shown that Byr4 can bind to an SPB only if Cdc11 is fully dephosphorylated [26] and Sid2 is responsible for part of the phosphorylation on Cdc11 [20] . Cdc11 is known to be ( at least partially ) dephosphorylated by the SIN Inhibitory Phosphatase Complex SIP [21] , which we also consider as a regulator of the proposed minimal system . In summary Cdc16-Byr4 inhibits SIN and SIN inhibits Cdc16-Byr4 localization to SPB , giving an antagonistic double-negative feedback loop ( Fig . 1C ) . We can update the wiring diagram of Fig . 1A with the basics of the molecular details of this antagonistic interaction by joining the SIN members in a single variable and representing the Cdc16-Byr4 complex by its limiting component Byr4 . The wiring has to be further extended as SIN is not directly inhibiting Byr4 , but through phosphorylating Cdc11 , which form cannot support Byr4 recruitment to SPB . Thus , instead of direct activation of Byr4 removal ( as it is on Fig . 1A ) , SIN inhibits the facilitator of Byr4 binding ( Fig . 1D ) . This adds an extra step in the system , but does not change the signs of the interactions proposed above . This system can be also turned into a computational model and in this case we can move the non-linearity to the Cdc11 multistep phosphorylation-dephosphorylation reactions ( captured by an appropriate non-linear function [24] , [27] , [28] ) . Simulation of this model shows that asymmetry of SIN can be established from an initial metaphase state ( high SIN , low Byr4 at both SPBs ) . After the transition , the active SIN is localized together with phosphorylated Cdc11 to the new SPB , while Byr4 is at the old SPB with dephosphorylated Cdc11 ( Fig . 2A ) . Cdc11 is not moving between the two SPBs , it just changes its phosphorylation state depending on the presence of regulators at a given SPB . To reach this asymmetry all we had to assume is that Byr4 has a 0 . 1% higher affinity to bind to the old SPB than to the new SPB . This ( or a much higher ) initial bias could come from inherited phosphorylated proteins that are specifically present at the old SPB [15] . It is known that proper cytokinesis greatly depends on the total amount of SIN components and its regulators [29] , [30] . Overexpression of Spg1 , the uppermost member of SIN leads to hyperactivation of SIN and to a multiseptated phenotype when cells periodically lay down septa without cleaving them [12] . A similar phenotype is observed when Cdc16 , Byr4 or to some extent SIP function is lost [21] , [31] , [32] . On the other hand mutations in SIN components and Byr4 overexpression lead to SIN inactivation and to a multinucleate phenotype when septum formation and cell division is totally abolished [12] , [14] , [32] . We observe similar behavior in the simulations of the model if the total cellular levels of SIN and Byr4 are perturbed ( Fig . 2B–E ) . SIN level can be changed only in a very narrow window , even very small changes lead to delays in asymmetry establishment and doubling or halving of the original amount already shows the experimentally observed terminal phenotypes ( Fig . 2B ) . Byr4 cannot be increased either , small reductions do not lead to major delays in asymmetry but below a certain threshold the observed phenotype reveals ( Fig . 2C ) . The simulated high sensitivity to Cdc11 levels ( Fig . 2D ) is contradicting the literature data as overexpression should not lead to a phenotype [10] , while mutations in Cdc11 function should lead to multinucleate phenotype [33] . This latter problem comes from the fact that we initiate the model in late mitosis with high SIN levels , which cannot be reached in Cdc11 mutants as SIN binding to SPB requires Cdc11 function . Furthermore Cdc11 is also needed for the activity of downstream SIN components ( Sid1 , Sid2 ) [10] . A major extension of the model with the whole mitotic regulation of SIN could resolve this issue , here we keep our focus on asymmetry establishment after anaphase onset . Overexpression of Csc1 , a member of the SIP complex leads to multinucleate cells and some SIP mutant cells ( csc1Δ ) show multiple septa [21] . Although it is not clear if overexpression of one of the components of the SIP complex is enough to induce higher SIP phosphatase activity or if it has a dominant negative effect , the simulated high sensitivity to SIP levels ( Fig . 2E ) resembles experimental observations [21] . In summary the minimal molecular model of SIN asymmetry regulation properly simulates most experimental observations . The major failure of the model is on the high sensitivity to Cdc11 levels . The experimentally observed low sensitivity to Cdc11 overexpression [34] might be explained by a limiting effect of Sid4 , which helps Cdc11 to recruit SIN members to SPB [35] , but we can also investigate Cdc11 in more detail if we consider its different phosphorylation sites . Cdc11 is known to be phosphorylated on multiple sites by SIN ( specifically shown for Sid2 in [20] ) but Cdc11 also contains Cdk phosphorylation sites [20] , [35] . SIP was discovered as a SIN Inhibitory PP2A Phosphatase Complex as it can remove phosphate groups from Cdc11 [21] . PP2A complexes often counteract Cdk phosphorylations [36] , so it could be that SIP is working on the Cdk phosphorylation sites of Cdc11 and either SIP or another phosphatase removes the phosphates from SIN sites . Furthermore , it was observed that removal of SIN phosphorylation sites from Cdc11 ( mutating five serine to alanine ) leads to advanced asymmetry establishment [20] , which could not be captured by the minimal model . To overcome these issues we extended the model with Cdk phosphorylation of Cdc11 ( Fig . 3A ) . Cdc11 can exist in at least four different forms: Cdk phosphorylated ( Cdc11-CP ) , SIN phosphorylated ( Cdc11-SP ) , phosphorylated by both ( Cdc11-PP ) and non-phosphorylated ( Cdc11 ) and only this latest form can support Byr4 binding to SPBs . As we have no information on the target sites of SIP or other phosphatases acting on Cdc11 we investigate the effects of both dephosphorylation steps separately . We assume a hypothetical phosphatase ppC to remove phosphates from Cdk site , while another phosphatase ppS works on SIN sites ( Fig . 3A ) . Similarly to the simple model above , SIN and Byr4 dynamics at the two SPBs follows the experimentally observed trend ( Fig . 3B ) . The various forms of Cdc11 are converted into each other as cytokinesis proceeds , with ∼75% Cdc11 becoming dephosphorylated and 25% remaining Cdk phosphorylated at the old SPB ( solid black line of Fig . 3C ) and most of Cdc11 at the new SPB is phosphorylated mostly by SIN ( dashed green on Fig . 3C ) . This model is sensitive to changes in SIN and Byr4 levels ( Fig . S1A , B ) as the minimal model was ( Fig . 2B , C ) , but now the sensitivity of Cdc11 overexpression and the simulated multinucleate phenotype of the minimal model ( Fig . 2D ) is lost , since Cdk can phosphorylate even high levels of Cdc11 and by this inhibit Byr4 binding to the Cdc11 , which is present in excess ( Fig . S1C ) . With these we fixed the simulations of the major phenotypes . Literature data suggest that the timing of asymmetry establishment is highly sensitive to the Cdc11 phosphorylation state [20] . Fig . 4 shows how perturbations in the SIN and Cdk phosphorylation efficiencies and in the phosphatase efficiencies of ppC and ppS affect the timing of asymmetry establishment in the detailed model . Small decreases in SIN efficiency advance asymmetry , while severely reduced SIN phosphorylation on Cdc11 leads to a multinucleate phenotype . Advances were observed for the Sid2 phosphorylation site removed cdc11-S5A mutant [20] , which is matched with an approximate halving of SIN efficiency on Cdc11 ( arrow on Fig . 4A ) . Since the phosphorylation of SIN on Cdc11 in the model captures all negative effects of SIN on Byr4 activation and the experimentally observed effect of SIN sites removal from Cdc11 can be captured by a partial reduction of this effect , suggesting that SIN has to phosphorylate other targets which are regulating Byr4 activity/localization ( see details on this in the discussion ) . On the other hand , total reduction in Cdk phosphorylation efficiency has no effect on asymmetry timing , while an increase in the Cdk site phosphorylation , similar to high SIN efficiency led to serious delays and eventually to a multinucleate phenotype ( Fig . 4A ) . Thus , Cdk mostly serves as an initiator of the Cdc11 phosphorylation state and it is not directly involved in asymmetry timing , but if Cdk ( or SIN ) phosphorylation on Cdc11 is constantly high then Byr4 cannot bind to SPBs and this leads to multinucleate phenotype . Serious reduction in either hypothetic phosphatase activity leads to multinucleate phenotype , while milder reduction causes a delay . Interestingly increase in ppC efficiency ( overexpression of the hypothetical phosphatase ) does not cause any phenotype in the model , while ppS overexpression leads to multinucleate phenotype ( Fig . 4B ) . If we assume that the overexpression of the SIP component , Csc1 , induces higher SIP activity ( if this is the only limiting factor in the complex ) leading to the observed multinucleate phenotype [21] , then the model predicts that SIP should have roles in removing phosphates catalyzed by Sid2 to Cdc11 ( at least when it is overexpressed ) . Since other mitotic phosphatases , like the Cdc14 phosphatase , Clp1/Flp1 [37] , [38] or the PP2A phosphatases Par1 and Pab1 [39] , [40] have been associated with SIN function and recent results suggests a role for Clp1 in Cdc11 dephosphorylation [41] , we cannot conclude on the exact role of SIP only by simulating single perturbations on Cdc11 phosphorylation . In our first double perturbation test we investigated the interactions between perturbations in SIN and Cdk efficiency on Cdc11 phosphorylation versus mutations in the Byr4 effector Cdc16 efficiency on SIN inactivation ( Fig . 5A ) . Cdc16 mediates the GAP-activity that induces Spg1 inactivation and it is localized by Byr4 [13] , thus mutations in Cdc16 can be simulated in our model by changing the efficiency of Byr4 on SIN inactivation ( kSoff in Supplementary Text S1 ) . The temperature sensitive cdc16-116 mutant can proliferate at 25°C while at higher temperatures the activity of this mutant protein is gradually reduced and eventually the cells are unable to inactivate SIN leading to a multiseptated phenotype at 36°C [31] . Simulation of this mutant by setting Byr4 efficiency on SIN to 20% of the wild type value shows a strong delay in asymmetry establishment ( Fig . 5A ) . The model predicts that this delay can be compensated for mildly by removal of Cdk phosphorylation sites from Cdc11 but very efficiently by the cdc11-S5A mutants of SIN phosphorylation on Cdc11 ( Fig . 5A ) . To test this prediction first we used a Cdk site mutant version of Cdc11 [35] that substitutes the eight Cdk phosphorylation sites from Cdc11 [20] and tested its effects on cell viability . As reported previously [35] , removal of Cdk phosphorylation sites from Cdc11 has no major effect on cell viability , matching the simulation results ( Fig . 4A ) . The cdc11-S8A mutant could indeed mildly compensate for the defects of cdc16-116 ( Fig . 5B ) , while the SIN ( Sid2 ) sites removed cdc11-S5A mutation instead of rescuing the phenotype rather exacerbated it ( Fig . 5B ) . It was shown that SIP phosphatase complex removes phosphate groups from Cdc11 and that mutations in SIP components give an additive effect to cdc16 mutations [21] . To investigate the discrepancy between model and experiment further , we tested if cdc11-S5A and cdc11-S8A mutants can compensate this additive effect of SIP and cdc16 mutations . First we simulated the cdc16 mutation by reducing the effect of Byr4 on SIN to the half of the original value and the csc1Δ SIP mutation by setting both ppC and ppS to 75% of the wild type values . The simulations indeed match the additive effects of these mutations ( Fig . 5C ) . Greater decreases lead to even greater delays in asymmetry establishment and eventually to a multiseptate phenotype ( not shown ) . The simulations of cdc11 phosphosite mutants predict that major SIN sites removal ( cdc11-S5A ) can compensate the additive effect of SIP and Cdc16 quite well , while Cdk site removal has only minor compensatory effects ( Fig . 5C ) . Experimental tests show that the double mutants of cdc16-116 and csc1Δ is mildly compensated by Cdk phosphorylation sites removal from Cdc11 , matching the prediction ( Fig . 5D ) . At the same time the double mutant phenotype becomes more severe after Sid2 phosphorylation site removal ( Fig . 5D ) . Phenotypic analysis of these cells show that the number of multiseptated and cut cells increased in the cdc16-116 csc1Δ cdc11-S5A triple mutants ( Fig . 5E ) , suggesting that SIN might come too early and stays active longer in some of these cells . The discrepancies between simulations and experimental results show that blocking Sid2 phosphorylation of Cdc11 has consequences other than allowing enhanced Byr4 binding to SPBs [26] , furthermore , perturbation in the SIP phosphatase complex ( csc1Δ ) does not change the severe phenotype of cdc16-116 cdc11-S5A mutants . These , and other earlier findings [20] , [21] , [41] suggest that Sid2 phosphorylation might prime Cdc11 for dephosphorylation at other sites and Byr4 binding , making SIN an indirect activator of Byr4 . Recent results suggest that such dephosphorylation events might be catalyzed by the Cdc14-like Clp1/Flp1 phosphatase , even in the absence of SIP activity [41] . Removal of both SIN and Cdk phosphorylation sites from Cdc11 ( cdc11-S13A ) does not have a major effect on cell viability , furthermore SIP activity still has an effect on the phosphorylation state of Cdc11 in cdc11-S13A cells [41] , indicating that SIP dephosphorylates Cdc11 at sites modified by other kinases . Thus our findings , together with recent literature data , indicate that our understanding of Cdc11 regulation by phosphorylation-dephosphorylation events is incomplete . We have shown above that the model can capture the basic behavior of SIN mutants in asymmetry establishment and can accurately predict the behavior of some mutant combinations . There are a few , so far , unresolved experimental findings that ask for computational models to help understand them . Magidson et al . [42] found that if in anaphase , when SIN asymmetry is already established , the new SPB containing active SIN was ablated with a laser , then the SIN starts to get activated at the old SPB . To simulate this experiment we stopped the simulations when asymmetry was reached and uncoupled the new SPB from the rest of the cell . Fig . 6A shows that if some SIN from the ablated new SPB can fall back to the cytoplasm ( or constantly produced there – not shown ) then it can move to the old SPB and remove Byr4 activity there . This happens because the free cytoplasmic SIN now can start to bind to the only existing old SPB . Although this is slow at the beginning , as SIN starts to phosphorylate Cdc11 , Byr4 cannot be as efficiently recruited anymore . As this positive feedback of SIN activation ( through inhibiting the binding of its inhibitor ) speeds up , more and more SIN gets to the only existing SPB and at the same time Byr4 is getting removed . In another interesting experiment , by cleverly creating dikarions Garcia-Cortes and McCollum [43] investigated cells with four SPBs present at the time of mitosis . They found that when two SPBs with active SIN go to one daughter cell and two with inactive SPBs to the other , then cells separate properly and SIN gets inactivated right after division . In contrast , when both daughters inherit one active and one inactive SPB then the SIN could not turn off properly . We simulated these two scenarios by removing ( separated ) or maintaining ( non-separated ) the communication between the inactive , old SPB and the cytoplasm of the new SPB and followed the speed of SIN inactivation at the new SPB ( Fig . 6B ) . To mimic the unknown factors that induce SIN inactivation after cell separation we started to increase the cytoplasmic Byr4 level in the cells . We followed this approach as in our small model Byr4 acts as the only inhibitor of SIN , but any other abrupt change in the SIN/Byr4 ratio as a result of cytokinesis would have a similar effect in the model . Although the exact mode of SIN inactivation after completion of cytokinesis is not clear , the simulation results show that the same inactivation strength lead to a much faster SIN inactivation when the two SPBs were separated ( Fig . 6B ) . This happens , because in the separated case all inhibitors of SIN can start to work on the SPB with the active SIN , while in the non-separated case the newly produced inhibitors are still recruited to the already inactive SPB , thus they cannot reach the active SIN on the other SPB . A mechanical metaphor explains both situations on Fig . 6C . The antagonistic , double-negative feedback loop leads to situations when on one SPB SIN can always win against Byr4 . If two or more SPBs are in the same cytoplasm then this antagonism leads to asymmetry establishment and strong maintenance of this state . These results suggest that cells are sensitive to SIN/Byr4 ratio before establishing the asymmetry , but once they established SIN asymmetry the strong antagonism can compensate small changes in the SIN/Byr4 balance . After communication between the daughter nuclei is halted by the septum , the balance is important again and the SIN-Byr4 antagonism can help the fast inactivation of SIN .
Asymmetric activation of the SIN on one of the two SPBs is a necessary feature of proper cell division timing in fission yeast cells [18] , [19] . Similar asymmetry is established between the SPBs of the budding yeast Saccharomyces cerevisiae [44] , [45] . In the case of such asymmetrically dividing organisms , the asymmetry establishment is better characterized [46] and mathematical modeling has already facilitated discoveries of the detailed mechanism [22] . Here we establish a minimal model to understand the major driving forces of symmetry breaking in SIN activity at the two SPBs in fission yeast . This minimal model is based on the antagonistic interaction of two molecules that are inhibiting each other's localization to the SPB ( Fig . 1A ) . This system resembles the basic models of Notch-Delta antagonism that is used to model lateral inhibition [47] . Indeed the underlying dynamics in both cases leads to a pitchfork bifurcation ( [23] and Fig . S2 ) . The model behaves as an efficient switch [48] , which brings one molecule type to one SPB and its antagonist to the other , with some remaining in the cytoplasm . In the case of SIN asymmetry establishment the clear candidates for such antagonistic interactions are the members of the SIN and its inhibitory complex Byr4-Cdc16 . Byr4-Cdc16 inhibits SIN activity [13] , while there is also some evidence that SIN indirectly inhibits Byr4 localization [20] , [26] . Such antagonism is a special case of a positive feedback loop , where the two components cannot coexist , either one of them is winning and inhibiting the other [25] . In the case of SIN asymmetry establishment , the two antagonists are winning at different SPBs . Indeed when the new SPB is starting to get enriched in SIN , it means SIN has to drop a bit on the other SPB , which enables Byr4 to win on the old SPB . In this way SIN activation at one SPB helps Byr4 activation on the other SPB explaining some controversial observations which suggest that SIN components and mitotic phosphatases seem to activate both SIN and Byr4 [19] . Thus any signal that leads to the induction of asymmetry establishment basically activates SIN ( at the new SPB ) as well as Byr4 ( at the old SPB ) . The major initiating step is the drop in Cdk activity in anaphase in parallel with spindle elongation that moves the SPBs far apart . Our simulations are initiated exactly at this step . Possible spatial extensions of the model might reveal some role for SPB positioning , although the quick turnover of active Sid2 [20] might rule out any major effect of space in SIN asymmetry establishment . A crucial point here is that such a system with an antagonistic switch works properly only if the total amounts of the two antagonists are present in a given ratio ( 1 in our case , but this value is determined by the exact rate constants ) , any perturbation of this balance can lead to a situation where either SIN or Byr4 wins on both SPBs . Indeed fission yeast cells are very sensitive to the overexpression of either Byr4 or the SIN limiting factor Spg1 , but the joint overexpression of these two can be greatly tolerated by the cells [30] suggesting that indeed their ratio is important for proper asymmetry establishment . The model suggests that once the asymmetry is established this balance is not that crucial anymore , but later the same antagonism can help the fast inactivation of SIN after septation . At this stage only the new SPB inheriting daughter has active SIN signaling , but this is turned off for an unknown signal that most probably flips the SIN/Byr4 balance . The extended minimal model ( Fig . 3A ) is still a simplification of the whole system of SIN regulation as here we concentrated only on the interactions that are important for the asymmetry establishment in SIN activity ( see [4] for a model on SIN activation timing ) . Still this simple model was able to capture qualitatively multiple experimental results on single molecule perturbations ( Fig . 2B–E and Fig . S1 ) , explain results of experiments when the number of SPBs were perturbed in the cells ( Fig . 6 ) and predict the behavior of some double and triple mutants ( Fig . 5 ) . The prediction on the compensatory effects of Cdk sites removal from Cdc11 in a cdc16 and cdc16-116 csc1Δ mutants were verified experimentally ( Fig . 5A , B ) , the additive effects of SIP and Cdc16 mutants were also properly simulated , but the predictions on the double and triple mutants with cdc11-S5A failed ( Fig . 5C–E ) . The cdc11-S5A mutation amplified the phenotype of cdc16 and cdc16-116 csc1Δ mutants instead of compensating them . This does not mean that the model is totally wrong; it rather means that there is a hole in our knowledge about the backup mechanisms that regulate SIN activity when some of the major players are perturbed . Cdc11 is likely phosphorylated by other kinases ( perhaps Cdc7 [26] ) and proteomics screens found Clp1/Flp1 as a phosphatase acting on Cdk sites on Cdc11 [41] , adding extra layers to the interaction system . Another possibility is that the Cdc11 phosphomutants may not recapitulate the result of asymmetric loss of phosphorylation in which only one SPB is affected and/or the investigated mutant combinations show a phenotype that is a result of other functions of Cdc16 [49] . Furthermore , it was earlier proposed that Clp1 might form another positive feedback loop with the SIN [19] , [50] , which could also play a role in the robustness of SIN asymmetry establishment . The proposed core mechanism of antagonistic interactions between activators and inhibitors of SIN should hold in all cases , just the main players might change as kinases and phosphatases as well as their target molecules might be perturbed in various mutants . There could be several other layers , where SIN and Byr4 antagonistically interact , as many other SIN regulators are targets of Cdk , SIN and Polo kinase dependent phosphorylation events [19] . A related prediction of the model is that SIN components have to act on other Byr4 regulator targets than Cdc11 , as we could match the SIN phosphorylation sites removed cdc11-S5A phenotype only with a reduced efficiency of SIN , not with the total abolishment of this effect ( Fig . 4A ) . The simplest possible solution would be if one of the SIN components could directly phosphorylate and by this mechanism inactivate Byr4 . Since Byr4 has several candidate phosphorylation sites [29] , [51] we cannot rule out this possibility . The modeling results also predicted and the experiments verified that Cdk phosphorylation on Cdc11 is not a major factor in asymmetry establishment ( Fig . 5A ) , it might rather play a role in setting up the initial state in early mitosis , when the top components of the SIN pathway are bound to both SPBs and Byr4 is removed from there . Interestingly , all of our simulation results show that in the initial mitotic state Byr4 is not totally absent from SPBs . This assumption on the initial conditions we needed to take to be able to achieve a fast asymmetry establishment . If Byr4 is completely absent from both SPBs in mitosis then it would be difficult for Byr4 to appear at one SPB in sufficient amounts ( as it is sent away by active SIN ) to turn on the positive feedback loop and establish asymmetry . Since Byr4 is a low abundance protein , it is hard to visualize [29] , but the model suggests that even in mitosis some Byr4 might be localized at both SPBs . It is still unknown what signal ( s ) turns off SIN activity in the daughter inheriting the new SPB after the completion of cytokinesis . The model of SIN and Byr4 antagonistic interactions successfully simulated the experimental results , which have shown that SIN activity can take over Byr4 at the old SPB if the new SPB was laser ablated before cell division ( [42] and Fig . 6A ) and it could also explain why SIN has a harder time to turn off when the two spindle pole bodies remain in the same cell after cell division ( [43] and Fig . 6B ) . As we do not have information on the molecular details of the trigger that induces SIN inactivation in the daughter cell that inherited the SPB with active SIN , we needed to make a simple assumption that Byr4 production speeds up at this point , alternatively Byr4 degradation slows down when the daughters get separated [29] . Inactivation of SIN might happen even with a minor increase in Byr4 level , since once the old SPB is not in the same cytoplasm anymore it cannot serve as a sink for Byr4 , thus Byr4 can pile up at the daughter with the active SIN and eventually turn SIN off . The prerequisite for this mechanism to work is a very fast turnover of Byr4 , which has been suggested [29] . This and many other questions on the detailed regulation of SIN signaling still need to be addressed and as we have shown here , the system level view and computational modeling of the network can help our understanding and guide experimental discoveries . Here we could reach predictions on a semi-quantitative fashion ( e . g . : what happens earlier/later in various mutants ) , measurements on molecular levels of the regulators and kinetic contacts of the reactions will enable the development of quantitative models that contain all molecular details of SIN activity regulation .
The wiring diagrams of Fig . 1A , 1D , 3A were converted into systems of ordinary differential equations ( ODEs ) . Parameters of the models were identified by fitting their qualitative behavior to experimental observations . Molecular concentrations defined in arbitrary units . Future measurements of molecular levels could be used to convert the inferred parameter values to real biologically meaningful reaction rates . We assume fast diffusion between SPBs until cell separation cuts communication between SPBs . Parameter values , initial conditions and equations can be found in the Supplementary Text S1 . Equations were numerically solved and simulated by the freely available software WINPP ( http://www . math . pitt . edu/~bard/xpp/xpponw95 . html ) . S . pombe strains were grown in yeast extract ( YE ) medium . Strain construction was accomplished through standard methods . The relevant genotypes and strain numbers used in this study were cdc16-116 cdc11-S5A-GFP::kanR ( KGY1411 ) , cdc16-116 cdc11-GFP::kanR ( KGY3342 ) , cdc16-116 cdc11-S8A-GFP::kanR ( KGY8684 ) , cdc16-116 cdc11-GFP::kanR csc1::ura4+ ( KGY12982 ) , cdc16-116 cdc11-S5A-GFP::kanR csc1::ura4+ ( KGY12982 ) , and cdc16-116 cdc11-S8A-GFP::kanR csc1::ura4+ ( KGY12984 ) . | Rod shaped fission yeast cells , as the name suggests , divide by medial fission . The proper timing of this cytokinesis and septation event is controlled by a signaling pathway called the Septum Initiation Network , or SIN . The SIN is activated only after chromosomes start to separate in anaphase . At this stage , the two daughter spindle pole bodies ( SPBs - the yeast analog of centrosomes ) have separated and are on their way to the distant tips of the cell . SIN components are localized to SPBs , but the SIN is active only at one SPB , while the Cdc16-Byr4 complex keeps the SIN inactive at the other SPB . This asymmetric activation of the SIN is important for proper cell division as perturbation of this can lead to appearance of multiple septa or total lack of septation . The molecular mechanisms that are important for asymmetry establishment are emerging , but we lack a complete picture . Here we develop computational models to capture the dynamical features of asymmetry establishment and to determine the key components and interactions that are needed for proper asymmetric SIN activation . Our predictions and their experimental tests reveal some basic features of the system and highlight missing points in our knowledge . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"systems",
"biology",
"model",
"organisms",
"cell",
"division",
"cytokinesis",
"yeast",
"and",
"fungal",
"models",
"biology",
"computational",
"biology",
"molecular",
"cell",
"biology"
] | 2013 | Dynamics of SIN Asymmetry Establishment |
The cortex processes stimuli through a distributed network of specialized brain areas . This processing requires mechanisms that can route neuronal activity across weakly connected cortical regions . Routing models proposed thus far are either limited to propagation of spiking activity across strongly connected networks or require distinct mechanisms that create local oscillations and establish their coherence between distant cortical areas . Here , we propose a novel mechanism which explains how synchronous spiking activity propagates across weakly connected brain areas supported by oscillations . In our model , oscillatory activity unleashes network resonance that amplifies feeble synchronous signals and promotes their propagation along weak connections ( “communication through resonance” ) . The emergence of coherent oscillations is a natural consequence of synchronous activity propagation and therefore the assumption of different mechanisms that create oscillations and provide coherence is not necessary . Moreover , the phase-locking of oscillations is a side effect of communication rather than its requirement . Finally , we show how the state of ongoing activity could affect the communication through resonance and propose that modulations of the ongoing activity state could influence information processing in distributed cortical networks .
The brain processes sensory stimuli by an organized flow of neuronal activity across a distributed network of specialized cortical areas . This flow requires mechanisms that route neuronal signals from one cortical area to another . However , the exact nature of this routing process remains poorly understood . Experimental studies suggest that synchronization of spiking activity may play a pivotal role in the flow of neuronal activity , as synchronous neuronal firing can effectively drive downstream neurons [1]–[4] . To date , our understanding of synchrony-based neuronal routing has been dominated by two models which attribute the origin of synchrony to dissimilar mechanisms . According to the first model , synchronous spiking activity is both created and routed through dense and/or strong convergent-divergent connections between subsequent layers of feedforward networks ( FFNs ) . In this scenario , these connections are a source for shared and correlated input that provides sufficient synchronization for spiking activity to propagate across the FFN [5]–[10] . However , the requirements of either strong synapses or high connection probability pose serious constraints on the biological plausibility of these FFNs in the cortex , in which connectivity is in general sparse [11] and synapses are weak [3] , [9] , [12] . Even though , the sparse cortical connectivity could in theory host a large number of sparsely and weakly connected ( diluted ) FFNs , they would fail to generate enough synchronization to ensure propagation of spiking activity [7] , [13] , [14] . The second model suggests that population oscillations could soften the requirement of strong/dense connectivity by enhancing synchronization and neuronal excitability during the excitable phase of the oscillation [15] , [16] . A key requirement for this propagation mode is that oscillations , which are generated locally due to interactions between excitatory and inhibitory neurons , must maintain a consistent phase relationship ( coherence ) between the communicating networks ( “communication through coherence”; [15] , [17] , [18] ) . However , the mechanisms underlying the generation and maintenance of such coherent oscillations between distant brain areas have remained elusive despite a number of theoretical proposals [19] . Here , we propose a novel mechanism by which oscillatory activity exploits the presence of resonance frequencies in networks of excitatory and inhibitory neurons ( ) to promote the propagation of synchronous activity across diluted FFNs ( “communication through resonance” ) . The role of such network resonance is to amplify weak signals that would otherwise fail to propagate . According to our model , coherent oscillations emerge in the network during slow propagation of synchrony , while at the same time synchrony needs these oscillations to be propagated . Thus , spreading synchrony both generates oscillations and renders them coherent across different processing stages . This abolishes the requirement for separate mechanisms providing the local generation of oscillations and establishing their long-range coherence . Moreover , coherence between oscillations may be viewed as a consequence of propagation instead of being instrumental to establish communication through synchrony . Our results also suggest that the emergence of coherent oscillations is influenced by the dynamical state of the ongoing activity . We propose that changes in the ongoing activity state can have an influence on cortical processing by altering the communication between different brain areas .
The network models were multi-layered FFNs . Each layer consisted of two recurrently connected homogeneous neuronal populations . In Figure 1 , Figure 2 and Figure 3 we used 2 , 000 excitatory ( ) and 500 inhibitory ( ) neurons . For the rest of the figures , we reduced the number of neurons to 1 , 000 while keeping the number of interlayer projecting neurons fixed to 300 . This reduction , which was done in order to improve simulation efficiency , did not affect the results in any qualitative manner . The connectivity within each layer was random with the following connection probabilities: and , where denotes the probability of connection from a neuron in population to a neuron in the population . Connections between layers were strictly feedforward and excitatory , and restricted to a sub-population of 300 randomly-chosen neurons ( in the rest of the paper referred to as ) in every layer . The interlayer connectivity was sparse with probability ( cf . Table 1 ) . Neurons were modeled as leaky integrate-and-fire neurons , with the following membrane potential sub-threshold dynamics:where is the neuron's membrane potential , is the total synaptic input current , and are the membrane capacitance and leak conductance respectively . When the reached a fixed threshold a spike was emitted and the membrane potential was reset to After the reset , the neuron's membrane potential remained clamped to during a time period mimicking the period of absolute refractoriness . All other parameters are detailed in Table 2 . Synaptic inputs consisted of transient conductance changes:where is the synapse reversal potential . Conductance changes were modeled using exponential functions with and . Synaptic delays were set to and in Figure 4a , Figure 5b–c and Figure S4a . In the rest of the figures , delays were set to , and . Longer delays produced a stronger and more reliable propagation and therefore were chosen to illustrate the propagation across layers in Figure 4a . The choice of delays influenced the resonance properties of the network [20] . However , the general principle remained unaffected . Other parameters are detailed in Table 3 . Each neuron was driven by 1 , 000 independent Poisson excitatory spike trains with an average rate of 1 Hz each ( i . e . , a total average input rate of 1 kHz ) , which mimicked uncorrelated background inputs coming from other brain areas . In Figure 6 , neurons received this external drive ( referred to as drive ) with larger rates than 1 kHz as indicated in the figure . The synchronous stimuli consisted of periodic trains of synchronous spikes ( pulse packets ) with different frequencies . Only neurons received these additional spikes . The individual pulse packets consisted of a fixed number ( ) of spikes per neuron , distributed randomly around an arrival time . The time of each individual spike was drawn independently from a Gaussian probability distribution centered around and with s . d . ( ) . In Figure 2e , Figure 3b–c , Figure 4a , spikes and ( i . e , perfectly synchronous ) . In the remaining cases spikes and . When the stimulus was a periodic train of pulse packets , we set the frequency of stimulation by adjusting the period ( ) between arrival times ( i . e . , the center of the Gaussian p . d . f . ) . When , the spikes were spread around , as indicated above , and therefore the time distance between the last spike from a given pulse packet and the first spike from the next was always variable for the same input frequency . The smallest interval that was used between arrival times was 10 ms ( 100 Hz ) and the largest 100 ms ( 10 Hz ) . Additionally , in Figure 3b we used 1 Hz stimulation . In simulations where the arrival times were jittered , the size of the jitter was drawn from a uniform distribution centered on the arrival time . The extent of the jitter window was chosen to be a function of the interval , where , 4 or 2 in order to make the effect comparable across different frequencies . To compute the auto-covariance functions ( inset in Figure 2c and Figure 3c bottom right; only positive time lags are shown ) , time was divided into bins of and the population spike trains were transformed into spike count vectors , where denotes the population . The auto-covariance functions were then computed as follows:where , , indicates the population mean firing rate and the superscript denotes ongoing ( computed from a single simulation in absence of pulse packet stimulation ) and activated ( computed from of activity during stimulation starting 5 s after the stimulus onset and averaged across 20 trials ) , respectively . We used the population Fano factor ( pFF ) to classify the population spiking activity states as synchronous or asynchronous ( dashed line in Figure 6a ) . We used the central value of ( variance ) normalized by the mean population firing rate: The signal-to-noise ratio ( SNR ) in Figure 6d was computed as follows:where indicates the variance of the spiking activity of neurons as indicated above . Pairwise correlations were computed using the Pearson correlation coefficient between the spike count vectors of pairs of neurons ( and ) . where: and indicates time average and vectors and were computed using a time window of . We used 10 , 000 pairs to compute the distributions shown in Figure 2c and Figure 3d . The correlation coefficients were computed from simulations with a length of . The power spectrum of the population spike train ( PS ) was calculated as follows ( from [21] ) :where , in Figure 4b , Figure S4a and Figure S5a and in Figure 6b indicating the corresponding value of ( cf . description of the auto-covariance function above ) . Network simulations were performed using the simulator NEST , interfaced with PyNest [22] , [23] . The differential equations were integrated using forth order Runga-Kutta with a time step of 0 . 1 ms . Simulation data was analyzed using the Python scientific libraries: SciPy and NumPy . The visualization of the results was done using the library Matplotlib [24] . The code to reproduce several results presented in this work ( Figure 1b , Figure 3a , Figure 4a , Figure 5b and Figure S6a ) is available at https://github . com/AlexBujan/ctr . Other results can be reproduced by modifying that code .
We studied the propagation of synchronous spiking activity across diluted FFNs with sparse interlayer connectivity . In this model , each layer represented a small neocortical network with 2 , 000 excitatory ( ) and 500 inhibitory ( ) neurons . The connectivity within each layer was sparse and random . The connections between layers , which modeled long-range projections between different cortical networks , were strictly feedforward and excitatory . These interlayer projections were restricted to a sub-population of 300 neurons which we refer to as projecting neurons or neurons throughout the manuscript ( refers to all the projecting neurons in a layer with the subscript indicating the position of the layer in the FFN; cf . Figure 1a ) . Interlayer connection probability and hence , each neuron received , on average , connections from the previous layer ( ; cf . Table 1 ) . All layers were driven by external Poisson input spike trains and the synaptic weights were adjusted ( cf . Table 3 ) to bring the network into an asynchronous-irregular ( AI ) activity regime [25] , [26] , consistent with the statistics of cortical activity in awake behaving animals [27]–[30] . The mean firing rate of individual excitatory neurons showed a heavy tailed distribution with a mean of ( ; s . d . across the population ) . The mean coefficient of variation of the inter-spike interval distribution ( ) was ( ) and the distribution of pairwise correlations was centered around zero with a mean of ( ) ( cf . Figure 1b and Figure 2a–c ) . The activity of the population was also irregular and asynchronous although with slightly higher mean firing rates ( ) . These results were computed from a single simulation of duration . To study the propagation of synchrony , we stimulated all neurons in the first layer ( ) with synchronous events or pulse packets ( cf . Methods ) . The synaptic strength of these input synapses was equivalent to the other synapses in the FFN ( cf . Table 3 ) . First , we checked that the connectivity between layers was indeed too weak to support the propagation of single synchronous events . To this end , we generated an amplitude transfer map which we used to estimate the change in amplitude undergone by pulse packets as they travel across the FFN . This map , shown in Figure 2d , was generated using the ongoing membrane potential distribution ( black trace ) and depolarization transfer function ( dark gray solid trace ) of the population . The measured membrane potential distribution ( computed from 100 s of ongoing activity ) is shown as the cumulative density function ( c . d . f . ) of the distance to threshold ( ) . When represented as such , the probability of being at a certain distance from threshold can be interpreted as the fraction of cells ( here named , where indicates layer index ) that will spike if a depolarization ( “jump” ) equivalent to such distance is applied to all cells . The membrane potential transfer function was calculated by measuring the averaged maximum depolarization across neurons induced by a pulse of perfectly synchronous spikes ( ) with different amplitudes . The mapping between the two curves can be done by knowing the relationship between the activation level of the th layer and the amplitude of the pulse packet received by the subsequent layer which in this case is as follows: . Knowing this relationship , it is then possible to project a point from one curve to the other , thereby drawing an estimated trajectory of the pulse packet's amplitude across the chain . In the figure , an example of such a trajectory is illustrated with red dots and dotted lines . To make a convincing case , we started the trajectory with a fully activated first layer ( ; upper red dot ) and followed the pulse packet until it reached a stable point ( intersection between the two curves ) . Such trajectories will always end at an intersection between the curves which in this case ( ) is found only at zero . This shows that any single pulse traveling across this FFN will eventually vanish , regardless of the initial value of . Similarly , it can be shown that if the connectivity is raised to ( dashed light gray line ) a single pulse can undergo a stable propagation for some initial values . After a perturbation caused by a synchronous pulse , the network's activity relaxed back to ongoing levels while displaying a stereotypical damped oscillation ( Figure 2e ) . This dynamics , which was observed both at the spiking level ( shown as conductances in neurons in Figure 2e top ) and the level of the membrane potential ( Figure 2e bottom ) , indicated that the network had resonance frequencies . The presence of such resonance frequencies suggested that stimulating the network with a periodic train of pulse packets , within a specific frequency range , could induce a large response even for weak stimuli ( e . g . , pulse packets consisting of a few weakly synchronized spikes ) . The existence of resonance behavior in networks has already been shown elsewhere [20] . Here , we analyzed the network response to a pulse packet stimulation in order to understand in more detail how resonant dynamics can emerge in these networks . During the transient damped oscillatory response , there was a brief time period of a few milliseconds ( indicated approximately as a gray region in Figure 2e ) during which neurons were slightly more depolarized ( higher mean ) , more synchronous ( decreased s . d . ) and their inhibitory conductance was reduced . This suggested that the arrival of a second pulse packet inside this brief time window ( e . g . , around after the arrival of the first pulse packet; shown as a green dot in Figure 2e ) should result in a larger activation as compared to the first pulse . Conversely , the arrival of a second pulse outside of this window ( magenta dot in Figure 2e ) would only lead to a similar or even weaker activation . To confirm this , we stimulated neurons in an isolated layer with a sequence of 100 periodic pulse packets ( identical to the ones described in the previous section; and ) and computed the mean firing rate within 20 ms after the arrival of each synchronous event ( which was found to be an appropriate time window to capture the pulse packet induced modulation of the firing rate ) . We repeated the experiment using three different time intervals : 35 , 45 and 1 , 000 ms ( Figure 3b ) and in each case the results were averaged across 100 trials . Pulse packets separated by , which matched the optimal window described above , resulted in an average spiking activity of 48 Hz ( , s . d . across trials; green bar in Figure 3b ) . By contrast , stimulation with pulse packets separated by could only induce a mean network response of ( ) . This response was comparable to a stimulation in which pulse packets arrived at an interval of one second , long after the transient response to each individual event had died out ( compare magenta and blue bars in Figure 3b ) . This result confirmed that a train of periodic pulses , with a period adjusted to match the optimal time window , was able to elicit a stronger response as opposed to a single pulse packet . Additionally , the fact that a higher input frequency resulted in a lesser activation suggested that this effect was not merely due to the temporal integration of the individual pulse packets . To further understand the emergence of resonance in these networks , we analyzed the temporal evolution of the membrane potential distribution ( mean and s . d . sampled 1 ms prior to the arrival of each pulse packet ) during stimulation with a train of 100 pulse packets separated by 45 ms ( Figure 3c ) . The results were averaged across 100 trials . A brief initial depolarization , caused by the first two pulse packets , was followed by a sustained hyper-polarization in both and neurons as more pulse packets were presented . The hyper-polarization reflected that a larger fraction of neurons was refractory ( or close to the spike reset potential ) due to the increase in firing rate ( light gray bars in Figure 3c ) and recurrent inhibition . The fact that most neurons were more hyper-polarized seemed to be at odds with the observation that the pulse packets were more effective in driving neurons . Furthermore , a decrease of the s . d . ( Figure 3c inset ) indicated that neurons were overall more synchronized , namely , that the hyper-polarization was shared across the entire population . Essentially , the increased responsiveness was a consequence of the fact that neurons were effectively refractory at the time of the arrival of pulse packets , as indicated by the progressive reduction in their firing rates ( Figure 3c dark gray bars ) . That is , although neurons moved farther away from the spiking threshold , they received less inhibition at the time of the arrival of the pulse packets which resulted in stronger activation . This observation hinted to an important role of connections in the emergence of resonance in these networks . We investigated the contribution of the loop to the generation of resonance by conducting simulations in which we progressively reduced the strength of the recurrent inhibitory connections ( Figure S1 ) . We compensated the reduction in input by adding an additional source of external inhibitory conductance in order to keep the firing rate of the neurons ( measured during the ongoing state ) constant across conditions . Our results showed that although the loop had a substantial effect on the resonance peak's amplitude and frequency , the network still had resonant properties in the absence of an loop . This indicates that while connections are sufficient to create resonance , dynamics play a facilitating role . In addition to the hyperpolarizing inhibition used in our model , other biologically plausible mechanisms , such as shunting inhibition or gap junctions , could also enhance resonance [31]–[33] . Although the overall activity of an isolated layer became more synchronized , with network oscillations that were locked to the stimulus , the overall activity of neurons remained fairly irregular ( ) , and mean pairwise correlations were still relatively low ( ; compare ( Figure 3d and Figure 2c ) . Hence , the activity of neurons during stimulation was still consistent with biological data , which shows that cortical firing is highly irregular despite the presence of oscillations at the population level as measured by local field potentials [34] , [35] . Note however that the activity of the neurons was more regular ( they skipped fewer cycles ) than the other neurons . Such a level of regularity in the population was needed in order to induce oscillations in the post-synaptic layer and was a consequence of the small number of neurons together with the sparse inter-layer connectivity . Thus , the choice of a larger population size and/or a higher connection probability could make propagation compatible with a more irregular firing in the projecting population ( cf . below ) . Additionally , we explored whether our network model operated in a linear regime in which case the tools of linear systems analysis could be applied to further understand the resonance [20] . To this end , we calculated the amplitude of the network's response when stimulated with synchronous pulses for different values of . Our results indicated that the behavior of the simulated network was generally non-linear showing a saturation of the response amplitude with high and a progressive shift in the resonance frequency ( Figure S2 ) . However , we also found that within a restricted range of input amplitudes the network's response approached linearity ( cf . straight lines in Figure S2e ) . Next , we addressed the question whether the network resonance-induced amplification of stimulus responses , observed in isolated layers , could be sufficient to enable the transmission of synchrony in diluted FFNs , which did not support the propagation of individual pulse packets . To this end , we stimulated a 5-layer FFN with three different frequencies , that were analogous to the ones introduced in the previous section ( cf . Methods ) . The amplification , observed when the input frequency matched the resonance frequency of , proved to be sufficient to induce a successful transmission across the entire FFN ( Figure 4a bottom ) . As expected , when the stimulus had a different frequency from the resonance frequency , or it consisted of a single pulse packet , the synchronous activity did not reach the last layer ( Figure 4a top and middle ) . Since the transmission relies on the network resonance , we refer to this mode of synchronous activity propagation as “communication through resonance” ( CTR ) . After receiving a few input cycles at the resonance frequency , nearly all neurons started to fire near synchronously every time a new pulse was presented . At this point , even though a large number of synchronous spikes were produced in the first layer , the sparse interlayer connectivity ( ) reduced this increased activation to a train of weak pulse packets with an average of spikes ( spikes ) and ( ) , which prevented the propagation of synchronous volleys immediately after amplification had taken place in . Therefore , amplification through resonance was needed at every layer to propagate the activity across the FFN due to the diluted interlayer connectivity . Next we investigated how the frequency of stimulation affected the propagation of synchrony across a 10-layer diluted FFN . Expectedly , we found a correlation between resonance-induced increase in synchrony in and the successful communication of synchronous events across the entire FFN ( Figure 4b ) . To quantify the synchrony we calculated the variance of the population spike train ( where indicates the stimulus frequency; cf . Methods ) . We then used to construct resonance curves as shown in Figure 4b bottom . A propagation was labeled as successful when was significantly increased ( ; white dots in Figure 4b bottom ) with respect to the baseline value . The spectral analysis of the spiking activity revealed that the increase in power in the last layer was always more pronounced at , which was approximately the resonance frequency of the network ( cf . Figure 4b lower-middle subpanel ) . Furthermore , CTR was not restricted to the FFN architecture discussed thus far . Our results showed that at least two alternative interlayer connectivity patterns also supported CTR: when receiving neurons were restricted to a specific sub-population of neurons but any neuron could project to the next layer ( Figure S3a ) ; when any neuron could receive and send projections ( Figure S3b ) . However , even when neuronal activity propagated to the last layer ( white dots in Figure 4b ) , was significantly lower than in ( compare red and blue curves in Figure 4b bottom ) . This result indicated that propagation was occasionally characterized by failures of synchronization of the last layers . Thus , the ratio could be used as a proxy for the propagation reliability when activity was observed during long time periods ( 10 s ) . Generally , networks that produce a moderate amplification of the signal at the resonance frequencies would be more sensitive to noise fluctuations , which can transiently reduce the degree of synchrony and lead to frequent propagation failures . A larger amplification , which in our model was achieved by introducing longer delays within each layer , lead to a perfectly reliable propagation at the resonance frequencies ( Figure S4a ) . For the parameters used here , the range of frequencies that led to a successful propagation approximately spanned from 22 to 26 Hz . The extent of this frequency range can be varied by an appropriate choice of network parameters ( cf . Figure S4a; see [20] for a more detailed study on the effect of different parameters on resonance ) . The effect of different parameters on the resonance profile of the network can be estimated using the network's average response to a single pulse packet stimulation ( cf . Figure 2e ) . When the input frequency is expressed as the time interval between pulse packets ( dashed black trace in Figure 2e top ) , the resonance profile can be related to the average network response . Note that is again represented as a function of the input frequency in Figure 4b ( blue trace ) . As can be seen in Figure 2e , the dominant peak in closely matches the trough of the average inhibitory conductance response ( red curve ) . This suggests that the network's response to a single pulse packet stimulation can predict its resonance curve and thus can be used to understand how different changes in the network parameters may affect the resonance properties of the network . While different network parameters can alter its resonance curve , the activity propagation based on network resonance would remain essentially the same . For this specific choice of parameters , ( Figure 4b top; cf . Methods ) revealed that the resonance occurred mainly around two main stimulus frequencies: 23 Hz and 58 Hz ( see also Figure S5b bottom row ) . Note that similar resonance frequencies were found when neurons were stimulated with a sinusoidally modulated Poisson input , which indicates that the faster resonance frequency can not be explained by the existence of harmonics of the base frequency present in the periodic input pulse train ( Figure S6 ) . Naturally , the smaller resonance frequency precisely matched the time window described in the previous section . The frequency of the second resonance peak can be explained using the network's average response as indicated earlier . To understand this effect , we can consider a simpler stimulus consisting of three pulses the frequency of which is systematically increased with respect to the main resonance frequency ( 23 Hz ) . Initially , the rise in frequency will cause the second and third pulses to arrive outside the optimal time window . However , as the frequency is further increased , a frequency will be reached for which the third pulse will fall inside the optimal window giving rise to an increase of the spiking response . Intuitively , this latter frequency should be approximately twice as large as the main resonance frequency , which is inconsistent with our results . This discrepancy can be understood when we notice that the second pulse , although not strong enough to activate neurons , does accelerate their re-polarization , thereby advancing the optimal time window within which the third pulse should arrive . That is , the subthreshold effect of these incommensurate pulses will speed up the network response resulting in the second resonance peak being faster than twice the main resonance frequency . Experimental evidence suggests that brain oscillations in the gamma range are not perfect periodic oscillators with a consistent phase [36]–[39] . Consequently , to be a biologically plausible mode of communication , CTR should be robust enough to facilitate the transmission of oscillatory spiking activity when the constraint of a constant phase has been relaxed . To quantify the extent to which CTR could afford unstable phases within an oscillation , we probed 10-layer diluted FFNs with periodic trains of pulse packets whose arrival times were jittered . The jitter was drawn from a uniform distribution centered on the arrival time ( ) of the pulse packet . The extent of the jittering window was chosen to be a function of the interval where , 4 or 2 . The results showed that CTR could still enable the transmission in the presence of moderate amounts of jitter ( Figure 5a ) . For this particular selection of network parameters , a jitter of did not alter the main characteristics of the amplification process and the activity propagated to the last layer ( Figure 5a top right ) . However , if the jitter was further increased the activity propagated to fewer layers and the propagation was more unreliable . Interestingly , for a jitter of , which corresponds to completely aperiodic pulse packet train , we observed that activity propagation increased with increasing the stimulus frequency . However , in this case also the pulse packets propagated with a frequency of , close to that of the network resonance frequency ( Figure S4b ) . That is , each FFN layer acted like a bandpass filter , which suggested that a broad-band noise stimulus could also trigger the transmission since it can generate oscillations close to the resonance frequency . Indeed , it is well known that the dynamics of networks can display oscillations at the population level when they are stimulated with strong unstructured external drive [26] , [40] . We hypothesized that in the FNN a constant rate Poisson input could bring the activity of the first layer into an oscillatory regime , thereby generating a train of weak pulse packets that provide rhythmic input to the subsequent layers . We tested this hypothesis by replacing the oscillatory input to by an additional source of constant Poisson input to all neurons in the first layer . When in the network shown in Figure 5b-c the drive was increased from 1 to 1 . 8 kHz the activity became oscillatory with enough power to ignite the resonance in the second layer . Interestingly , the frequency of the oscillations in was comparable to the resonance frequency of the network . This is not surprising as both resonance and oscillations at higher input regimes are shaped by the same network time constants , e . g . , synaptic delays and membrane time constants [20] . Thus , we show that both slightly phase-jittered oscillatory inputs at the resonance frequency and broad-band stimulation are compatible with CTR in diluted FFNs . Thus far , we have assumed that ongoing activity in each individual layer of the FFN was AI with low firing rates . However , there is ample experimental evidence suggesting that cortical networks in vivo can display more synchronized ongoing activity regimes depending on the behavioral state of the animal [30] , [41] . We therefore explored how the propagation of pulse packets via CTR is influenced by the dynamical state of the spontaneous network activity . The level of synchrony in recurrent networks can be modulated by adjusting the firing rate of the external excitatory input [9] , [26] , [42] . Here , we changed the dynamical state by increasing the drive from 1 to 1 . 6 Hz . Lower rate drive gave rise to very sparse and asynchronous firing patterns , which progressively became more synchronous as the drive was increased ( synchrony measured as population Fano factor; red line in Figure 6a ) . The spiking activity of individual neurons remained irregular ( ) for the parameter space explored here ( cf . blue line in Figure 6a ) . increased in the range between 10 and for larger values of drive . This increase was more pronounced around the peaks , which progressively shifted towards faster frequencies as the external input became stronger ( Figure 6b ) . To study the effect of network synchrony on CTR , we stimulated in 10-layer FFNs with periodic trains of pulse packets for the different levels of drive and computed the signal-to-noise ratio ( SNR ) in ( cf . Methods ) . Generally , more synchronized activity states enabled CTR within a broader range of input frequencies , however the largest SNR values in were found at the low input regimes ( Figure 6d ) . Independent of the synchrony level , resonance frequency and subsequently CTR were always confined within a range of input frequencies that closely matched the frequency around the peaks of ( compare Figure 6d and Figure 6b ) . Hence , the resonance frequencies also became faster at higher levels of drive . This shift reflected the reduction of the time that neurons needed to recover from the effective refractory state ( absolute refractory period and hyper-polarization time ) due to the presence of larger amounts of excitation as drive was increased . The main peak in when activity reached was invariably found at 20 Hz . This value was slower than the mean peak measured in which was 28 Hz ( Figure 6c ) . The values of were larger than those of for all the frequencies analyzed here . Notably , this difference was more pronounced in the gamma range ( ) as compared to lower frequencies ( ; cf . Figure 6c ) . Interestingly , network synchrony improved the propagation for faster input oscillatory regimes ( , Figure 6d ) . In summary , our results showed that the ongoing state had opposite effects on CTR depending on the input frequency range . For lower input frequencies , AI activity increased SNR , while for larger input frequencies SI could enable the propagation which was absent during AI . A direct validation of the model will involve the induction of coherent oscillations between distant brain areas by stimulating excitatory neurons in the presynaptic area at the resonance frequency . The resonance profile of a neuronal population can be obtained by recording its activity during periodic stimulation of the neurons with different frequencies . Similar experiments , which made use of optogenetic tools , have already been performed to study the role of specific cell types in the generation of gamma oscillations [43] . According to our model , even weakly connected distant networks ( verified , e . g . , by anatomical or electrophysiological studies ) with a similar resonance profile can engage in a coherent oscillation by stimulating the presynaptic population at the resonance frequency . In contrast , a stimulation protocol , which does not induce a strong oscillation in the stimulated area , will fail to form such a coherent activity with the distant population . Our model also predicts a progressive entrainment characterized by a gradual increase in the measured power over multiple stimulation cycles in the stimulated presynaptic network . A similar entrainment should be found in the postsynaptic network with a certain delay which should be a function of the connectivity strength ( see discussion ) . Moreover , in CTR mode of propagation the oscillations emerge only after a delay and not directly at the onset of the stimulus . This feature of the model is consistent with the observation that - band oscillations appear after 100 ms of the stimulus onset ( e . g . [44] ) . This would confirm that CTR is by definition a slow mode of communication and therefore it is not suited for the communication of signals which have to propagate across multiple areas within a short period of time . Note that , e . g . , in the FFN shown in Figure 4a , synchronous activity reached the fifth layer only after approximately 10 stimulation cycles ( at 40 Hz ) . We further quantified this result by testing the number of cycles required in a given layer until a significant synchronization level was found in the subsequent layer . A significant degree of synchrony was reached when the instantaneous rate of neurons , computed using 5 ms time bins , hit a threshold value equal to plus five times its s . d . . The results , computed using 100 trials , are shown in Figure 7 as a function of stimulus frequency ( represented as the inter-pulse interval ) . Our results showed that when stimulated within the main resonance frequency range ( 39–42 ms intervals ) the average speed of propagation was approximately two cycles/layer with small variability . Small deviations from that resonance frequency range resulted in higher trial-to-trial variability of the propagation speed and increased mean while larger deviations resulted in propagation failure . The results obtained with our example FFN are indicative of how much time it will take to encode a stimulus using CTR at each stage of a processing chain . Naturally , the amount of time will be proportional to the number stages that the activity has to traverse . However , synchrony-based coding using FFNs seems to be suited only for communicating binary signals , i . e . , the asynchronous/synchronous activity of a given layer indicates the absence/presence of a particular stimulus ( e . g . , a specific orientation of a bar of light ) . By contrast , the encoding of graded signals would require a monotonic relationship between the input and the output of the FFN . We tested the capacity of a diluted FFN to communicate continuous signals using CTR . To this end , we applied periodic stimuli with different amplitude and computed the amplitude response of the network . Our results showed that for these network parameters it was possible to find an input range within which the system's response changed monotonically . Moreover the response remained linear for a restricted range of inputs strength ( cf . gray lines in Figure S2e ) . Such a linear operating regime including even a modest degree of saturation , could allow for the communication of graded signals . We note that our model supports communication of activity between areas that have similar resonance profiles . This automatically ensures selective communication and gives possibility of gating the propagation by small change in the resonance frequency of a network . The experiments proposed above could demonstrate , whether CTR is in principle compatible with the neuronal hardware and physiology , even though they will not necessarily rule out other proposed mechanisms like CTC [17] .
Here we propose a novel mechanism for propagation of synchronous spiking activity within weakly coupled FFNs based on the presence of resonance in networks . In our model , resonance is a network property that emerges due to the interactions between excitatory and inhibitory neurons in each FFN layer . Using numerical simulations of spiking neuronal networks , we show that a weak and sustained stimulus can be gradually amplified in every layer , thereby overcoming the limitations of synchrony transmission imposed by the diluted interlayer connectivity . We refer to this mode of synchronous activity propagation as “communication through resonance” ( CTR ) . Until recently , resonance was considered mostly at the level of single cells in both experimental [45]–[47] and theoretical studies [48] , [49] . Now , there is increasing experimental evidence showing that resonance also exists at the network level in inhibitory [43] as well as excitatory neuronal populations [50] , and may play a crucial role in the generation of cortical rhythms . Theoretical studies have shown that resonance is a fundamental property of networks [20] and could be used to gate neuronal signals [51] . In our model , such network resonance is used to enable the propagation of synchronous spiking activity in diluted FFNs . In previous theoretical studies , propagation of neuronal activity was restricted to either densely and weakly connected FFNs , which promote the propagation of synchronous activity [7] , [9] , [42] , [52] , [53] , or sparsely and strongly connected FFNs , which are capable of propagating asynchronous firing ( [54] , [55]; see [14] for a review ) . However , biological neuronal networks are typically neither densely connected nor have strong synapses [11] and therefore the mechanisms that govern the propagation of neuronal activity in dense/strong FFNs are not always applicable . Our results indicate that propagation is possible in diluted FFNs , when aided by network resonance , but is restricted to synchronous activity . Oscillations in the gamma range ( ) , which are a key feature of task-related population activity in several brain areas [56] , [57] , have emerged as a prominent mechanism that may facilitate propagation of synchronous spiking activity in weakly connected networks [15] . These oscillations can synchronize neuronal activity and provide appropriate temporal windows of excitability , which enable communication between different brain areas . Within these temporal windows , effective functional connections are generated where otherwise only weak structural links may exist [17] , [58] . This mode of propagation , however , requires communicating brain areas to oscillate with matched phase and frequency ( i . e . , their oscillations are coherent ) such that synchronous activity from the sender can reach the receiver during its excitable phase and maximize its spiking response . It is commonly believed that coherent oscillations are generated by two independent mechanisms , one responsible for the local generation of oscillations [59] and another mechanism that can flexibly modulate the coherence between spatially distant oscillators [19] . However , the precise nature of the process responsible for achieving such long-range coherence still remains elusive . Here , we argue that coherent oscillations arise due to the propagation of periodic synchronous spiking activity . In our model , weak rhythmic synchronization provided by the input initially fails to propagate further down the FFN due to the diluted connectivity . The crucial role of the oscillations is to amplify this weak synchronous stimulus by promoting resonance dynamics of the receiving network and enable its propagation across the FFN . This is in contrast to the idea that oscillations are generated independently at every layer and locally synchronize unstructured background input . Our results show that oscillations arise in the network as a consequence of the stimulus propagation , and at the same time the stimulus exploits these oscillations to propagate . Due to this propagation , oscillations in each layer are driven by the previous layer and are hence naturally coherent with a phase that is determined by the conduction delay between the layers [17] . From this perspective coherence becomes a side effect of the propagation dynamics . Thus , a separation of distinct mechanisms that create oscillations and provide coherence is not necessary , as both arise naturally as consequence of CTR . Indeed , recent experimental studies suggest that there is an unidirectional entrainment of coherent oscillations between areas [60]–[62] , making the feedforward spread of coherent oscillatory activity , as explained by our model , biologically plausible . We show that while CTR still works for moderate deviations from periodicity , it is most efficient for propagating periodic stimuli . Notably , the same FFN architecture can transform a sustained firing rate signal into a weak rhythmic stimulus that can then be propagated . Even though it can be argued that environmental stimuli are often not periodic , it has been recently suggested that sensory information could be actively converted into periodic signals by sensing organisms [63] , [64] . CTR requires amplification of activity in each layer and , as a consequence , the propagation is slow requiring several cycles to reach the target network . The numbers of cycles needed to transmit synchronous activity across the entire FFN is a function of the connectivity strength between the layers . As the synaptic weights become stronger , the number of cycles required to spread synchrony to the final layer of the FFN decreases and transmission becomes more reliable . Once the weights are sufficiently strong , synchrony flows through the network in one oscillation cycle , which is equivalent to the propagation of synchronous activity in dense/strong connected FFNs investigated by previous studies ( cf . [14] for a review ) . Thus , CTR could generate FFNs with strong connections capable of propagating isolated synchronous events , when certain types of synaptic plasticity are recruited to strengthen the synapses between the different FFN layers . Indeed , coherent oscillations , like those generated by CTR , can provide an ideal dynamical environment to promote synaptic potentiation [65] . In this way , CTR could be regarded as an initial means to propagate activity before strong connections have been formed , while providing the ideal substrate for the generation of fast and reliable communication channels . In the present study , we describe activity propagation in single FFNs . However , other more complicated network architectures in which multiple FFNs interact may also be possible . In such a scenario , the input could create a stronger response in one such FFN , while partially and weakly activating other FFNs with unmatched resonance frequencies , thereby generating a broadband increase in power around the resonance frequency of the activated FFN . Thus , such a scheme could indeed explain the increase in broadband gamma power of the LFP signal observed during behavioral tasks [66] . Signal gating is an intrinsic property of CTR , since in a given FFN only the stimuli that match its resonance frequency are able to propagate . Selective gating of signals through network resonance has been suggested by previous theoretical studies [51] , [67] . Interestingly , the resonance frequency of the network can be dynamically modulated offering the possibility to gate signals differently in time . In our study , we show that modifying the level of external excitation shifts the resonance frequency of the FFN . Additionally , other mechanisms such as neuromodulator mediated changes of the effective connectivity within each layer can have similar effects on the resonance properties of the network . Another alternative gating mechanism is the use of gating signals [14] . Gating activity in dense/strong FFNs requires highly precise and strong gating signals [53] . However , the fact that in CTR the initial phase of the propagation in a given layer is characterized by low amplitude synchrony , which is still insufficient to elicit responses in the next layer , makes CTR suited for a gating mechanism that utilizes relatively imprecise and weak gating signals . Thus , overall CTR constitutes a flexible process that could implement complex spatio-temporal routing of neuronal signals . As we show here , the dynamical properties of the background activity affect the quality ( SNR ) of the neuronal signals that are communicated using CTR . More specifically , SNR at low frequency stimulation ( ) was maximized when background activity state was asynchronous-irregular . This result is in line with experimental evidence which found oscillations in the gamma range to be associated with cortical desynchronization [68] , [69] . In contrast , the propagation of stimuli was successful only when ongoing activity was in a synchronous-irregular state . These findings hint at a hypothetical scenario in which slow periodic modulations of the background dynamics could rhythmically improve or even gate signals that propagate using fast oscillations . The fact that the nesting of slow and fast cortical oscillations ( e . g . , beta-gamma ) is commonly found in experiments ( see [70] for a review ) could be indicative of such a collaborative effort between different cortical rhythms . These findings open up the possibility that top-down signals may provide the change of background activity state required for coherent feedforward oscillations to be generated . Importantly , CTR is not restricted to the specific neuron and network model used in this work . The resonance mechanism , which is the essence of the model , is a general property of recurrently connected populations of excitatory and inhibitory neurons [20] and therefore it is widely applicable . Notably , a specific range of propagating frequencies can be achieved by a proper selection of network parameters . In summary , we have shown that communication of neuronal signals across weakly connected networks can be achieved by combining oscillatory activity with resonance dynamics . | The cortex is a highly modular structure with a large number of functionally specialized areas that communicate with each other through long-range cortical connections . It is has been suggested that communication between spiking neuronal networks ( SNNs ) requires synchronization of spiking activity which is either provided by the flow of neuronal activity across divergent/convergent connections , as suggested by computational models of SNNs , or by local oscillations in the gamma frequency band ( 30–100 Hz ) . However , such communication requires unphysiologically dense/strong connectivity , and the mechanisms required to synchronize separated local oscillators remain poorly understood . Here , we present a novel mechanism that alleviates these shortcomings and enables the propagation synchrony across weakly connected SNNs by locally amplifying feeble synchronization through resonance that naturally occurs in oscillating networks of excitatory and inhibitory neurons . We show that oscillatory stimuli at the network resonance frequencies generate a slowly propagating oscillation that is synchronized across the distributed networks . Moreover , communication with such oscillations depends on the dynamical state of the background activity in the SNN . Our results suggest that the emergence of synchronized oscillations can be viewed as a consequence of spiking activity propagation in weakly connected networks that is supported by resonance and modulated by the dynamics of the ongoing activity . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"computational",
"neuroscience",
"biology",
"and",
"life",
"sciences",
"computational",
"biology"
] | 2014 | Communication through Resonance in Spiking Neuronal Networks |
Interleukin 18 ( IL18 ) is a cytokine that plays an important role in inflammation as well as host defense against microbes . Mammals encode a soluble inhibitor of IL18 termed IL18 binding protein ( IL18BP ) that modulates IL18 activity through a negative feedback mechanism . Many poxviruses encode homologous IL18BPs , which contribute to virulence . Previous structural and functional studies on IL18 and IL18BPs revealed an essential binding hot spot involving a lysine on IL18 and two aromatic residues on IL18BPs . The aromatic residues are conserved among the very diverse mammalian and poxviruses IL18BPs with the notable exception of yatapoxvirus IL18BPs , which lack a critical phenylalanine residue . To understand the mechanism by which yatapoxvirus IL18BPs neutralize IL18 , we solved the crystal structure of the Yaba-Like Disease Virus ( YLDV ) IL18BP and IL18 complex at 1 . 75 Å resolution . YLDV-IL18BP forms a disulfide bonded homo-dimer engaging IL18 in a 2∶2 stoichiometry , in contrast to the 1∶1 complex of ectromelia virus ( ECTV ) IL18BP and IL18 . Disruption of the dimer interface resulted in a functional monomer , however with a 3-fold decrease in binding affinity . The overall architecture of the YLDV-IL18BP:IL18 complex is similar to that observed in the ECTV-IL18BP:IL18 complex , despite lacking the critical lysine-phenylalanine interaction . Through structural and mutagenesis studies , contact residues that are unique to the YLDV-IL18BP:IL18 binding interface were identified , including Q67 , P116 of YLDV-IL18BP and Y1 , S105 and D110 of IL18 . Overall , our studies show that YLDV-IL18BP is unique among the diverse family of mammalian and poxvirus IL-18BPs in that it uses a bivalent binding mode and a unique set of interacting residues for binding IL18 . However , despite this extensive divergence , YLDV-IL18BP binds to the same surface of IL18 used by other IL18BPs , suggesting that all IL18BPs use a conserved inhibitory mechanism by blocking a putative receptor-binding site on IL18 .
Poxviruses are a family of large , complex DNA viruses , infecting a variety of organisms including insects , reptiles , birds and mammals [1] . The poxvirus family is further subdivided into genera based on shared characteristics such as host range , morphology , antigenicity , and sequence similarity [2] . Four genera of poxviruses are known to be pathogenic to humans , including molluscipoxvirus , orthopoxvirus , parapoxvirus , and yatapoxvirus . As an immune evasion strategy , poxviruses encode an assortment of decoy receptors for chemokines and cytokines [3] . One such strategy for evasion of the host immune response is through modulation of the interleukin 18 ( IL18 ) signaling pathway . IL18 is a pro-inflammatory cytokine belonging to the interleukin 1 superfamily and plays an important role in both innate and acquired immune responses by inducing interferon-γ ( IFN-γ ) production from T lymphocytes and macrophages while also enhancing the cytotoxicity of natural killer cells [4] . IL18 activity is modulated in vivo by a negative feedback mechanism involving a naturally occurring IL18 inhibitor , the IL18 binding protein ( IL18BP ) [5] . Homologues of IL18BPs are also encoded by many poxviruses including molluscum contagiosum virus and orthopoxviruses [6] , [7] such as variola virus , the causative agent of smallpox . Yaba-Like Disease Virus ( YLDV ) along with Yaba Monkey Tumor Virus ( YMTV ) are members of the yatapoxvirus genus of poxviruses . These viruses produce a very distinct disease in primates that is characterized by epidermal histiocytomas and vesicular lesions of the head and limbs [8]–[10] . Although their exact host reservoir is not well established , it is presumed that the immunomodulatory proteins expressed by these viruses can at least partially cope with the primate/human immune system . Analysis of YLDV and YMTV genome revealed yatapoxviruses encode a predicted IL18BP family member , designated as 14L [11] , [12] . YLDV and YMTV 14L proteins share approximately 54% sequence identity between each other but less than 14% with IL18BPs of the orthopoxviruses such as the variola virus and the mousepox ectromelia virus ( ECTV ) . Despite this low sequence similarity , YMTV 14L was previously shown to be a functional IL18 inhibitory protein with comparable affinity as orthopoxvirus IL18BPs [13] . The high-resolution crystal structure of the ECTV-IL18BP in complex with human IL18 revealed the structural basis by which orthopoxvirus IL18BPs antagonize IL18 signaling through direct competition with IL18 cognate receptor for binding [14] . The crystal structure along with mutagenesis studies identified a set of conserved residues from IL18 and IL18BPs as key to complex formation . In particular , a phenylalanine ( F67 in ECTV-IL18BP ) residue that is highly conserved in IL18BPs was found indispensible for IL18 binding [14]–[17] . Mutations of this site in all IL18BPs examined to date significantly decreased or even completely abolished the binding to IL18 . In addition , a residue on IL18 , K53 , was shown as a ‘hot spot’ for binding IL18BPs , since mutations at this site drastically decreased binding affinity [18] . A strong π-cation interaction between IL18 K53 and the conserved phenylalanine residue ( F67 ) of ECTV-IL18BP was revealed in the structure of ECTV-IL18BP:IL18 complex , explaining its important role in binding . Surprisingly , phylogenetic analysis and sequence alignment revealed the presence of a threonine ( T64 ) in yatapoxvirus IL18BPs at the position equivalent to the conserved phenylalanine ( Figure 1 ) [19] . Furthermore , mutation of K53 on IL18 only modestly affected the binding with YMTV 14L [13] . To understand how yatapoxvirus IL18BPs bind IL18 , we determined the high-resolution crystal structure of YLDV-IL18BP:IL18 complex . This structure along with the functional analysis through mutagenesis and Surface Plasmon Resonance ( SPR ) provide new insights into the mechanism by which IL18BPs inhibit IL18 . The result provided here could be helpful for developing inhibitors for IL18 or IL18BP .
Recombinant human IL18 and mature YLDV-IL18BP ( residue 20–136 ) proteins were individually purified from E . coli , and were subsequently used to reconstitute a complex of IL18:YLDV-IL18BP . Initially , wild-type ( WT ) IL18 and YLDV-IL18BP were used , but no quality crystals were obtained . In efforts to improve crystallization , we mutated some non-essential cysteines and additional surface residues in IL18 . Substitution of four cysteines with serines in IL18 was previously shown to increase the stability of IL18 without affecting IL18 activities [20] . We found that the use of an IL18 mutant [IL18 ( 8S ) , see Materials and Methods] , containing substitutions of the four-cysteines with serines and substitutions of four surface residues opposite to the IL18BP binding interface with alanines , greatly improved crystal quality and reproducibility . The crystal structure of IL18 ( 8S ) in complex with YLDV-IL18BP was determined to 2 . 7 Å . Furthermore , a crystal from the complex of IL18 ( 8S ) and a three-cysteine mutant of YLDV-IL18BP ( ΔC21 , C87S , C132S ) diffracted to 1 . 75 Å ( see Materials and Methods , Table 1 ) . The two structures are essentially identical to each other with a root mean square deviation ( r . m . s . d ) of less than 0 . 4 Å , and the three mutated cysteines in YLDV-IL18BP are distant from the IL18 binding interface . Therefore , for discussion of protein:protein interactions between YLDV-IL18BP and IL18 , we will mainly focus on the higher resolution structure of the mutant YLDV-IL18BP . The structure of the complex shows that the YLDV-IL18BP forms a homo-dimer in a back-to-back fashion , with each protomer binding to a molecule of IL18 , forming a hetero-tetramer complex with a stoichiometry of 2∶2 ( Figure 2A ) . The complex displays an elongated v-shaped architecture with the IL18BP homo-dimer at the center and IL18 at the ends . As seen in the ECTV-IL18BP:IL18 complex [14] , IL18 adopts the same β-trefoil fold , which is comprised of 12 β-strands ( β1–β12 ) with one short α-helix and one 310-helix . The IL18 molecules in the two structures show very little conformational changes , with an r . m . s . d . of only 0 . 5 Å from 134 aligned IL18 Cα backbones ( Figure 2B ) . The two IL18 molecules in the current complex structure are also nearly identical , having only a 0 . 1 Å r . m . s . d from 150 aligned cα backbone residues . Each protomer of YLDV-IL18BP adopts a canonical h-type immunoglobulin ( Ig ) fold [21] comprised of mainly β-sheets as observed previously for ECTV-IL18BP ( Figure 3A ) . However , there is an r . m . s . d . of 2 . 9 Å over the 106 aligned cα backbone residues between the two viral IL18BPs . ECTV-IL18BP has an extended β-sheet architecture with predominantly shorter loops connecting the β-sheets and β-strands , while YLDV-IL18BP has comparatively shorter β-sheets with extended connecting loops . The two IL18BP structures differ mostly at one sheet of the β-sandwich fold ( βA , βB , βE and βF ) that is not involved in binding IL18 ( Figure 2B ) . There is a major rigid-body movement on this β-sheet , especially at strands βA , βE and βF where twists are estimated at about 30 to 45 degrees . Compared to ECTV-IL18BP , there are two additional α-helices located between β-strands F and G ( H2 ) , and at the C-terminus ( H3 ) . YLDV-IL18BP contains five cysteine residues , forming two intra-molecular disulfide bonds ( SS ) ( C21–C87 , C43–C111 ) and one inter-molecular SS bond ( C132–C132 ) , in contrast to only two intra-molecular SS bonds in ECTV-IL18BP . C43–C111 is the conserved SS bond among many Ig-fold proteins , which connects the two β-sheets and plays a key role in maintaining overall integrity of the Ig-fold structure . C21–C87 SS bond connects the very N-terminus prior to βA with the loop between βE–βF . In contrast , the very N-terminus prior to βA of ECTV-IL18BP is SS bonded to the neighboring βB . In the triple-cysteine mutant YLDV-IL18BP:IL18 structure , the βE–βF loop is not visible in the electron density map , suggesting that C21–C87 SS bond stabilizes the local structure , particularly the βE–βF loop ( Figures 2 , 3A ) . However , the C21–C87 SS bond is not critical for the overall structure of YLDV-IL18BP or its binding with IL18 ( shown later ) . YLDV-IL18BP dimerizes back-to-back , abutting on one edge of the β sandwich , exposing the opposite edge for binding IL18 ( Figures 2 , 3 ) . The homo-dimer interface involves mainly βA , βB , βH and the C-terminal helix H3 . It involves extensive hydrophobic interactions in the center ( I29 , H30 , V31 , P32 and V33 from βA , the aliphatic side chain of E122 , V124 from βH and I129 from H3 ) flanked by hydrogen bonding and charge-charge interactions , burying approximately 1 , 700 Å2 and 1 , 465 Å2 solvent accessible surface area ( ASA ) for the wild-type complex and the triple-cysteine mutant complex , respectively ( Figure 3 ) . The lack of the C132-C132 SS bond in the triple-cysteine mutant complex caused the disorder of four residues at the C-terminus beyond S132 , resulting in a slightly smaller ASA . At the dimer interface , the imidazole ring of H30 from one protomer stacks on the P32 from the other ( Figure 3C ) , while V33 is forming favorable van der Waals interactions with a hydrophobic platform comprised of I29 , V31 , V124 and I129 from the other protomer ( Figure 3D ) . E42 from one protomer appears to be protonated , forming favorable hydrogen bonds with E42 from the other protomer . E42 also forms both intra-chain and inter-chain hydrogen bonds with R44 ( Figure 3B ) . The inter-chain C132-C132 SS bond covalently links the two promoters . Indeed , a non-reducing SDS-PAGE showed that YLDV-IL18BP proteins expressed in E . coli or in mammalian cells formed disulfide-bonded dimers ( Figure 4 ) . However , the triple-cysteine mutant of YLDV-IL18BP displays nearly identical structure as the WT protein , and it exists as a dimer in solution judging by size exclusion chromatography ( Figure 5 ) and dynamic light scattering analysis ( data not shown ) . Therefore , the dimerization is not solely dependent on the C132-C132 inter-chain SS bond . We performed additional mutagenesis studies to verify the importance of the residues at the homo-dimerization interface . E42R/C132S ( EC ) mutant remained a dimer in solution as the WT ( data not shown ) , while mutants bearing H30A/V33R/C132S ( HVC ) substitutions or H30A/V33R/E42R/C132S substitutions ( HVEC ) appeared as monomers in solution based on size exclusion chromatography ( Figure 5 ) and dynamic light scattering analysis ( data not shown ) . Therefore , hydrophobic interactions as well as the disulfide bonding together contribute to the dimerization . When measuring the binding affinity with IL18 by Surface Plasmon Resonance ( SPR ) , we found that the monomeric YLDV-IL18BP ( HVEC ) had a 3-fold decrease ( student t-test P-value<0 . 05 ) in binding affinity than the dimeric WT YLDV-IL18BP ( Table 2 ) . Sequence analysis shows that the residues key to YLDV-IL18BP homo-dimerization ( H30 , P32 , V33 , C132 ) are conserved only in yatapoxvirus IL18BPs but not in other IL18BPs ( Figure 1 ) . Therefore , dimer formation and bivalent binding of IL18 seems to be unique to yatapoxvirus IL18BPs . In fact , ECTV-IL18BP and human IL18BP were reported to be monomers in solution [14] , [22] . YLDV-IL18BP binds IL18 by using the same edge of the β sandwich as observed in ECTV-IL18BP:IL18 complex structure . Specifically , the following regions on YLDV-IL18BP are observed at the interface: loop connecting βB–βC , βC , the short βD , helix H1 , βG , and loop connecting βG–βH ( Figures 2 , 3 ) . Similar to what was observed in the ECTV-IL18BP:IL18 complex structure , YLDV-IL18BP molecule sits atop the opening of the IL18 β-barrel and binds the cytokine through extensive hydrophobic and hydrogen bonding interactions ( Figure 6 ) , covering about 1 , 957 Å2 of ASA , which is comparable to the ECTV-IL18BP:IL18 complex at 1 , 930 Å2 ASA as identified by the program AreaIMol of the CCP4 suite [23] . The numbers of residues involved at binding interfaces are also comparable between the two inhibitory complexes . YLDV-IL18BP contributes mainly 19 residues to the complex interface while IL18 contributes 23 residues , in comparison to 17 residues from ECTV-IL18BP and 25 residues from IL18 using NCont of the CCP4 suite [23] . To assess the energetic contributions to binding by residues at the binding interface , we performed site-directed mutagenesis on both IL18 and the monomeric YLDV-IL18BP ( HVEC ) and assessed the effects of the mutations on the binding affinity by SPR ( Figures 7 , 8 , 9 , 10 ) . In addition , to probe the difference in IL18 binding by IL18BPs of YLDV and ECTV , we performed binding studies of various IL18 mutants with the two IL18BPs simultaneously ( Figures 9 , 10 ) . We will describe the results from these functional studies in context of our depiction of the YLDV-IL18BP:IL18 complex interface . As we described in the previous ECTV-IL18BP:IL18 complex structure , we will continue to use the three identified binding sites , labeled as A , B and C on IL18 here ( Figure 6 ) . Site A of IL18 contains key residues Y1 , L5 , K53 , D54 , S55 and P57 , making extensive interactions with YLDV-IL18BP ( Figure 6A , B ) . The detailed interactions are similar to those observed in the previous ECTV-IL18BP:IL18 complex structure except for the loss of one of the ‘hot spot’ interactions involving a phenylalanine ( F67 in ECTV-IL18BP , described below ) . Substitution of S55 of IL18 with alanine was previously shown to decrease its binding affinity to orthopoxvirus IL18BP by 7-fold [18] . As observed in the ECTV-IL18BP:IL18 complex , the side chain hydroxyl of S55 is tethered to the main chain amino group of Y1 via a hydrogen bond in the YLDV-IL18BP:IL18 complex . Y56 on βC of YLDV-IL18BP occupies nearly identical position as observed in the ECTV-IL18BP complex without any conformational changes . Its phenolic group is tethered to the D54 main chain of IL18 , while its aromatic side chain together with the methyl group of T64 from YLDV-IL18BP , and L5 of IL18 constructs a hydrophobic wall , entrenching the aliphatic side chain of K53 from IL18 ( Figure 6B ) . We found mutation of Y56A on YLDV-IL18BP completely abolished its binding to IL18 ( Figures 7 , 8 , Table 2 ) , which is consistent with previous functional analysis of other IL18BPs [16] . Therefore , a tyrosine residue at this location in IL18BPs seems to be a conserved ‘hot spot’ , as the most important point to anchor the inhibitory protein to IL18 . As predicted , T64 of YLDV-IL18BP , located on the tip of βD , indeed occupies the same location of F67 from ECTV-IL18BP . This phenylalanine residue is highly conserved in IL18BPs of various species including human , all orthopoxviruses and MCV . Mutations at this location of various IL18BPs were shown to dramatically reduce the binding with IL18 [15]–[17] . In the ECTV-IL18BP:IL18 complex , F67 is inserted into an induced hydrophobic pocket , forming strong interactions with IL18 residues located on the surface while forming a strong π-cation interaction with the charged head group from the side chain of K53 of IL18 [14] . These interactions are absent in YLDV-IL18BP complex due to the presence of a threonine instead of phenylalanine . Interestingly , T64F substitution in YLDV-IL18BP did not increase the binding of IL18 ( Table 2 , Figures 7 , 8 ) , while T64A substitution only caused a 2 . 5-fold decrease ( student t-test P-value<0 . 05 ) in binding affinity . Consistent with the lack of π-cation interaction , IL18 K53 contributes less to the binding with YLDV-IL18BP than with the orthopoxvirus IL18BPs . While K53A mutation drastically reduced the binding affinity to orthopoxvirus IL18BPs [i . e . , more than 100-fold decrease for variola IL18BP [18] , Figures 9 and 10] , this mutation had less impact on binding affinity with YLDV-IL18BP ( about 30-fold decrease , Table 3 , Figures 9 , 10 ) . K53 of IL18 nevertheless remains important for binding to YLDV-IL18BP , because polar interactions involving K53 are preserved in the current structure . Specifically , the positively charged amino head group on the side chain of K53 forms salt bridges with D66 and E76 of YLDV-IL18BP , similar to the interactions of K53 with two glutamate residues of ECTV-IL18BP in the ECTV-IL18BP:IL18 complex structure . Site A differences also include the side chain rotation and repositioning of Q67 on YLDV-IL18BP ( equivalent to H70 of ECTV-IL18BP ) and Y1 on IL18 , creating a novel interaction that was absent in the ETCV-IL18BP:IL18 structure . Q67 is located in close vicinity to T64 and rotates about 90 degree ( vs . H70 of ECTV-IL18BP ) forming bifurcated hydrogen bonds with the hydroxyl group and the main chain amide nitrogen of T64 . Y1 of IL18 rotates about 80 degrees and stacks on the aliphatic portion of the Q67 side chain , forming favorable van der Waals interactions ( Figure 6B ) . Y1A substitution of IL18 and Q67A substitution of YLDV-IL18BP decreased the binding affinity by 20- and 4-fold , respectively ( student t-test P-value<0 . 05 , Table 2 , 3 , Figures 7 , 8 , 9 , 10 ) . In contrast , Y1A mutation did not affect the affinity with ECTV-IL18BP ( Table 3 , Figures 9 , 10 ) . Since T64A substitution of YLDV-IL18BP showed very little effect on binding to IL18 ( Table 2 , Figures 7 , 8 ) , the hydrogen bond between the side chain of Q67 and the main chain of T64 seems to be more significant than its interaction with the side chain . It is likely Q67 further stabilizes the local structure , including helix H1 where D66 locates , allowing correct positioning of this acidic residue for interacting with K53 of IL18 . Site B is a large , predominantly hydrophobic cavity spatially adjacent to Site A on the surface of IL18 ( Figure 6A ) . As observed in the ECTV-IL18BP complex , three non-contiguous residues , Y54 , I114 and P119 ( Y51 , T113 and V118 in ECTV-IL18BP ) from YLDV-IL18BP βC , βG and G-H loop reside but do not fully occupy the pocket . Y54A substitution in YLDV-IL18BP had negligible effects on IL18 binding , while I114A mutation caused only a 1 . 8-fold decrease in binding affinity ( statistically not significant , student t-test P-value>0 . 05 ) ( Figures 7 , 8 , Table 2 ) , similar to substitutions of the equivalent positions of other IL18BPs [16] , [17] . Therefore , site B interactions appear not essential for binding of YLDV-IL18BP , similar to observations for other IL18BPs [14] . Site C of IL18 is next to Site B and mainly comprised of 10 IL18 surface residues involving a mixture of charged and hydrophobic interactions . Similar to what was observed in the ECTV-IL18BP complex , the loops connecting βB–βC and βG–βH of YLDV-IL18BP interact with Site C on IL18 predominantly through hydrophobic interactions . YLDV-IL18BP F52 adopts nearly identical conformation as ECTV-IL18BP F49 , which is inserted into the large hydrophobic pocket of Site C [14] . Surprisingly , F52A mutation of YLDV-IL18BP had negligible effect on binding with IL18 ( Figures 7 , 8 ) , in contrast to the mutation at this location in other IL18BPs significantly decreasing binding affinity to IL18 ( 83–fold decrease for human IL-18BP:human IL18 , 8-fold decrease for ECTV-IL18BP:human IL18 and 138-fold decrease for ECTV-IL18BP:murine IL18 ) [16] , [17] . This difference can be explained by several unique interactions of YLDV-IL18BP at Site C . Residue P116 of YLDV-IL18BP is situated in a hydrophobic groove formed by aliphatic side chains from M60 , Q103 and M113 of IL18 and is stabilized by a stair-wise hydrophobic stacking by side chains from F49 , Y48 and K23 of YLDV-IL18BP ( Figure 6A , C ) . In addition , P116 main chain is hydrogen bonded with the hydroxyl group from the side chain of IL18 S105 , further stabilizing the complex interface . Indeed , P116A substitution reduced binding of YLDV-IL18BP to IL18 by approximately 6-fold ( student t-test P-value<0 . 005 , Figures 7 , 8 , Table 2 ) . Mutation of the equivalent residue in human IL18BP ( P153 ) showed negligible effect on its binding affinity to IL18 [15] , so P116 appears to be specific for YLDV-IL18BP in binding to IL18 . Y48 and K23 of YLDV-IL18BP are hydrogen bonded with IL18 D110 through side chain to side chain interactions . K117 of YLDV-IL18BP forms a bifurcated hydrogen bond with side chains from IL18 S105 and D110 ( Figure 6C ) . D110A of IL18 decreased affinity with YLDV-IL18BP by 6-fold ( student t-test P-value<0 . 05 ) but had no effect on binding with ECTV-IL18BP ( Figures 9 , 10 ) . Similarly , S105R of IL18 caused more than 30-fold decrease ( student t-test P-value<0 . 005 ) in binding affinity with YLDV-IL18BP but had no impact on binding with ECTV-IL18BP ( Figures 9 , 10 ) . Therefore , D110 and S105 of IL18 are specifically required for the binding of IL18 with YLDV-IL18BP but not for ECTV-IL18BP . The specificity of S105 towards binding of YLDV-IL18BP is further signified by a double mutant , S105R/P57R of IL18 . This double mutant completely abolished the binding with YLDV-IL18BP , while it had a much smaller impact on binding with ECTV-IL18BP ( Table 3 , Figures 9 , 10 ) .
We previously determined the crystal structure of the ectromelia virus IL18BP in complex with IL18 , revealing the structural basis for the binding and inhibition of IL18 by the IL18BPs [14] . Despite an overall low sequence homology between the diverse viral and host IL18BPs , the key residues of ECTV IL18BPs at the IL18 binding interface are highly conserved . Mutagenesis studies on human and viral IL18BPs also showed that these key residues are almost universally critical for the binding of IL18BPs to IL18 . Thus it was enigmatic that functional yatapoxvirus IL18BPs lack a key phenylalanine residue ( Figure 1 ) that has been identified to be essential for many other IL18BPs to bind IL18 [14]–[17] . It is similarly puzzling that a residue of IL18 ( K53 ) that is critical for binding orthopoxvirus IL18BPs only played a modest role in binding with the YMTV-IL18BP [13] . In this report , we resolved these questions by determining the crystal structure of YLDV-IL18BP:IL18 complex and by performing extensive mutagenesis and SPR studies . We revealed two unique signature features of YLDV-IL18BP that distinguish yatapoxvirus IL18BPs from the rest of IL18BP family members . First , YLDV-IL18BP forms a homo-dimer and interacts with IL18 in a 2∶2 binding mode . Second , the binding of YLDV-IL18BP and IL18 does not rely on two of the ‘hot spot’ interactions that were shown to be essential for the binding of all previously studied IL18BPs , including a phenylalanine ( F67 in ECTV-IL18BP ) at site A and another phenylalanine at site C ( F49 in ECTV-IL18BP ) . Instead , yatapoxvirus IL18BPs evolved interactions with some IL18 residues ( Y1 , D110 , S105 ) that are specifically important for binding with YLDV-IL18BP . It appears that YLDV-IL18BP shifts and disperses the binding energy across the IL18-binding interface rather than concentrating the binding energy on a few hot spots as is the case for all other IL18BPs examined to date . It was previously reported that ECTV-IL18BP and human IL18BP are monomeric in solution [14] , [22] . In contrast , YLDV-IL18BP forms a disulfide-bonded dimer , which was demonstrated not only in the crystal structure but also in solution by non-reducing SDS-PAGE and gel filtration analysis . The dimer interface is quite large ( about 1 , 700 Å2 ) and involves extensive hydrophobic interactions in addition to the intermolecular disulfide bond , indicating that the YLDV-IL18BP dimer is intrinsically stable in solution . The dimer could only be separated into monomers by mutations that disrupt both the hydrophobic interactions as well as the inter-chain SS bond . Analysis of the monomeric YLDV-IL18BP ( HVEC ) showed that the dimerization was not essential for binding IL18 but enhanced the binding affinity by 3-fold in our in vitro assay . Although this enhancement in binding affinity as measured by SPR is modest , it is possible that the dimerization may be more important for the function of YLDV-IL18BP during infection of the host , perhaps by increasing the half-life of the protein in the infected tissue or by increasing the avidity of binding to IL18 at low protein concentration . In fact , divalent or multivalent binding is an important , inherent feature of many biological systems to enhance the effectiveness of binding of ligands to receptors and of antibodies to antigens [24]–[28] . More specifically , this has been a feature for quite a few poxvirus cytokine binding proteins . For example , ectromelia virus IFN-γ binding protein forms a tetramer , which is required for efficient IFN-γ antagonism [29] . Myxoma virus T2 protein , a Tumor Necrosis Factor ( TNF ) Receptor homolog , is secreted as both monomer and dimer , and the dimeric T2 is a more potent TNF inhibitor [30] . Because residues of YLDV-IL18BP involved in dimer formation are only conserved in yatapoxviruses , yatapoxviruses IL18BPs may be unique among IL18BPs in that they use bivalent binding to increase the affinity and avidity for IL18 . Another difference between YLDV-IL18BP and all other IL18BPs is the lack of two of the ‘hot spot’ interactions at the binding sites A and C on the surface of IL18 . The structure of ECTV-IL18BP:IL18 complex showed that a conserved phenylalanine ( F67 ) is engaged in hydrophobic and strong π-cation interactions with K53 of IL18 at binding site A [14] . Alanine substitutions of K53 of IL18 significant decreased binding with orthopoxvirus IL18BPs [18] , while alanine substitutions of the conserved phenylalanine ( equivalent to ECTV-IL18BP F67 ) in human , MCV and orthopoxviruses IL18BPs significantly decreased or completely abolished binding of IL18 [15]–[17] . The current structure of YLDV-IL18BP:IL18 complex showed that a threonine residue ( T64 ) is present at the position equivalent to the phenylalanine , indicating the π-cation interaction with K53 is not important for YLDV-IL18BP: IL18 complex . Indeed , T64F or T64A substitution of YLDV-IL18BP had negligible or minor ( 2 . 5-fold decrease in affinity for T64A , student t-test P-value<0 . 05 ) effect on the binding with IL18 , while K53A of IL18 had a more modest effect on binding with YLDV-IL18BP than with orthopoxvirus IL18BPs . A similar loss of ‘hot spot’ interaction was also observed at binding site C . A phenylalanine residue on IL18BPs that binds to site C of IL18 was previously shown to be important for binding IL18 in orthopoxvirus , MCV and human IL18BPs [15]–[17] . Although a phenylalanine ( F52 ) is present at the equivalent position in YLDV-IL18BP , it is not important for binding to IL18 ( Table 2 , Figures 7 , 8 ) . Through structural and mutagenesis studies , we have identified contact residues that are unique to the YLDV-IL18BP:IL18 binding interface . This includes Q67 of YLDV-IL18BP and Y1 of IL18 at site A , P116 of YLDV-IL18BP and S105 and D110 of IL18 at site C . Our data are in agreement with the conclusion of a more delocalized energy distribution for binding of IL18 to YMTV-IL18BP [13] . The structural and functional studies of two different IL18BP complexes suggest that there is a degree of plasticity in the IL18BP:IL18 interface that could accommodate certain mutations in IL18BPs without compromising their binding affinity to IL18 . Despite the differences in several key residues for binding IL18 , the current YLDV-IL18BP:IL18 complex structure showed that YLDV-IL18BP targets the same surface of IL18 as ECTV-IL18BP does in the previous complex structure . This suggests that all IL18BPs inhibit IL18 function by blocking a putative receptor-binding site on the surface of IL18 . Similar to previous findings on human and poxvirus IL18BPs , Y56 of YLDV-IL18BP ( interacting with site A of IL18 ) was found to be absolutely essential for binding to IL18 , indicating that this conserved tyrosine residue is an obligatory ‘anchor’ for binding of all IL18BPs to IL18 . The conservation and variation in functional residues and their specific interactions with IL18 suggest that IL18BPs share a common ancestor but may have undergone significant evolution through different selection pressures , resulting in a conserved inhibitory mechanism albeit with mutations of interface residues . The biological activity of IL18 is determined in part by its relative affinities for IL18 receptors and IL18BP . The binding of IL18 to its receptors triggers multiple cellular responses vital to immunity , but excessive IL18 activities are associated with many autoimmune and inflammatory diseases [4] , [31]–[37] . Functional IL18BPs are present in many poxviruses including variola virus and vaccinia virus , providing a key strategy of poxvirus immune evasion by inhibition of IL18 cytokine activity . Therefore the studies on IL18BP:IL18 inhibitory complexes could serve dual purposes by providing important clues on how to develop functional inhibitors targeting either IL18 or poxvirus IL18BP . These inhibitors could potentially modulate IL18 and poxvirus IL18BP activities , which may benefit efforts in developing treatments against some autoimmune and inflammatory diseases and in developing treatments for potential pathogenic outbreaks associated with poxvirus infections .
Mature IL18 and YLDV-IL18BP ( residues 20–136 ) were individually cloned into a modified pET vector as SUMO fusion proteins with N-terminal 6×His tags and expressed in E . coli BL21 ( DE3 ) gold ( Stratagene ) or Rosetta-Gami 2 ( Invitrogen ) strains , respectively . An IL18 mutant ( C38S , C68S , C76S , C127S , K67A , E69A , K70A , I71A ) with substitutions of four nonessential cysteines [20] and four additional surface residues opposite to the IL18BP binding interface , IL18 ( 8S ) , and the triple-cysteine mutant of YLDV-IL18BP ( residues 22–136 , C87S , C132S ) were cloned and expressed in the same way as WT proteins . The individual proteins were purified using the similar double Ni-nitrilotriacetic acid ( Ni-NTA ) procedure as described [14] . Briefly , the his-tagged fusion proteins were first purified from cell lysate by Ni-NTA affinity column ( Qiagen ) and then co-dialyzed with ULP1 protease to remove the SUMO moiety , exposing the authentic N-terminus for both proteins . The cleaved protein mixtures were subsequently passed through a second subtracting Ni-NTA column and further purified by size exclusion chromatography on a Superdex s200 column . The YLDV-IL18BP and IL18 ( 8S ) proteins were mixed together and the complexes were subsequently purified from size exclusion chromatography and each concentrated to 9 mg/ml . The complex of wild-type YLDV-IL18BP:IL18 ( 8S ) crystallized in a condition containing 18% PEG3350 , 0 . 1 M Bis-Tris Propane/Citric Acid , pH 6 . 5 , while the crystallization condition for the complex of triple-cysteine mutant YLDV-IL18BP:IL18 ( 8S ) is 12% PEG3350 , 0 . 1 M Tris , pH 8 . 0 . 25% ethylene glycol was added step-wise to the mother liquid as cryoprotectant . To make biotinylated IL18 , mature human IL18 ( residues 37–193 ) was cloned into a modified pET vector containing a C-terminal 6×His tag along with the coding sequence ( GLNDIFEAQKIEWHE ) for biotinylation . This plasmid and a plasmid encoding biotin ligase ( Avidity ) were co-transformed into E . coli BL21 ( DE3 ) gold ( Stratagene ) for expression . One step Ni-NTA affinity purification was used to purify biotinylated and his-tagged IL18 . For mammalian expression of YLDV-IL18BP , a mammalian expression plasmid for YLDV 14L was constructed as described previously for the construction of expression vector for human IL18BP [16] . Briefly , YLDV 14L was amplified by PCR from genomic DNA of YLDV and cloned into pYX45 with NheI and BamHI sites , so that 14L ORF was appended with a C-terminal biotinylation tag and 6-His tag . 293T cells were transfected with the expression plasmid and then infected with vTF7 . 3 , a vaccinia virus expressing T7 polymerase . 3 days later , the medium was harvested and incubated with Ni-NTA resin ( Qiagen ) . The resin was then washed and added with E . coli biotin holoenzyme synthetase ( Avidity ) . After the biotinylation reaction , the protein was eluted with phosphate-buffered saline containing 300 mM imidazole . The ECTV-IL18BP was expressed in HEK293T cells , purified from the culture medium and biotinylated essentially as described previously [18] . An initial set of data for the tripe-cysteine mutant YLDV-IL18BP:IL18 ( 8S ) was collected at beamline X29 of National Synchrotron Light Source , Brookhaven National Laboratory . Initial phases were determined by molecular replacement using Phaser [38] of the CCP4 suite [23] and a search model containing IL18 along with a trimmed poly-alanine model of the ECTV-IL18BP ( PDB ID 3F62 ) . A subsequent data set from a crystal of WT YLDV-IL18BP:IL18 ( 8S ) complex was collect at beamline 19-ID of the Advanced Photon Source , Argonne National Laboratory . The structure of the complex containing WT YLDV-IL18BP was solved similarly as the structure of the complex containing the triple-cysteine mutant YLDV-IL18BP . All datasets were processed with HKL3000 [39] . PHENIX [40] was used for refinement and Coot [41] was used for iterative manual model building . Translation , libration and screw-rotation displacement ( TLS ) groups used in the refinement were defined by the TLMSD server [42] . The structure of the triple-cysteine mutant YLDV-IL18BP:IL18 ( 8S ) complex was refined to 1 . 75 Å resolution with Rwork and Rfree of 19 . 0% and 23 . 1% respectively . The structure of the WT YLDV-IL18BP:IL18 ( 8S ) complex was refined to 2 . 7 Å resolution with Rwork and Rfree of 21 . 9% and 27 . 0% respectively . The final models are of good refinement statistics for both complexes as shown in Table 1 . All molecular graphic figures were generated with PYMOL [43] . The SPR analysis was done essential as described previously [15] , [16] . Briefly , biotinylated IL18BP [∼300 resonance units ( RU ) ] or IL18 ( ∼250 RU ) was captured onto a BIAcore CM5 chip coated with streptavidin . Various concentrations of IL18 ( from ∼1 nM to ∼40 nM ) or IL18BP ( from ∼2 nM to ∼60 nM ) were injected at a flow rate of 20 µl/min . The chip coated with IL18BP was regenerated with a 10-µl injection of 1 M NaCl , 50 mM NaOH , while the chip coated with IL18 was regenerated with a 10-µl injection of 10 mM glycine ( pH 2 . 5 ) . The sensorgrams were analyzed with BIAEVALUATION software ( BIACORE ) . The binding data from the injection of at least five different concentrations of analyte were globally fitted to a 1∶1 binding model . Analyses with the same concentration series were done twice . Protein Data Bank: structure factors and atomic coordinates for the WT YLDV-IL18BP:IL18 ( 8S ) and the triple-cysteine mutant YLDV-IL18BP:IL18 ( 8S ) complexes have been deposited with accession codes 4EEE and 4EKX , respectively . | Interleukin 18 ( IL18 ) is an important cytokine in inflammation and immunity . Mammals and poxviruses encode homologous inhibitory proteins of IL18 , named IL18BPs , which regulate IL18 activity and , in the case of the viral proteins , contribute to virulence . Previous structural and functional studies revealed residues at IL18:IL18BP interface that are critical for the high-affinity binding , including a phenylalanine on IL18BPs , which is conserved among nearly all IL18BPs with the notable exception of yatapoxvirus IL18BPs . To understand the mechanism by which yatapoxvirus IL18BPs neutralize IL18 , we solved the high-resolution crystal structure of the Yaba-Like Disease Virus ( YLDV ) IL18BP:IL18 complex . The structure revealed a 2∶2 bivalent binding complex , which has not been observed in any other IL18BPs . Through mutagenesis and functional studies , we found a set of interacting residues that are unique for the association of YLDV-IL18BP and IL18 , likely compensating for the lack of the interactions involving the conserved phenylalanine . Despite this extensive divergence , however , YLDV-IL18BP binds to the same surface of IL18 used by other IL18BPs . Our study suggests that all IL18BPs use a conserved inhibitory mechanism by blocking a putative receptor-binding site on IL18 but the interface on IL18 is malleable by a broad and diverse family of mammalian and poxvirus IL18BPs . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biochemistry",
"protein",
"chemistry",
"macromolecular",
"assemblies",
"biology",
"computational",
"biology",
"biophysics",
"macromolecular",
"structure",
"analysis"
] | 2012 | A Unique Bivalent Binding and Inhibition Mechanism by the Yatapoxvirus Interleukin 18 Binding Protein |
Molecular interactions between killer immunoglobulin-like receptors ( KIRs ) and their MHC class I ligands play a central role in the regulation of natural killer ( NK ) cell responses to viral pathogens and tumors . Here we identify Mamu-A1*00201 ( Mamu-A*02 ) , a common MHC class I molecule in the rhesus macaque with a canonical Bw6 motif , as a ligand for Mamu-KIR3DL05 . Mamu-A1*00201 tetramers folded with certain SIV peptides , but not others , directly stained primary NK cells and Jurkat cells expressing multiple allotypes of Mamu-KIR3DL05 . Differences in binding avidity were associated with polymorphisms in the D0 and D1 domains of Mamu-KIR3DL05 , whereas differences in peptide-selectivity mapped to the D1 domain . The reciprocal exchange of the third predicted MHC class I-contact loop of the D1 domain switched the specificity of two Mamu-KIR3DL05 allotypes for different Mamu-A1*00201-peptide complexes . Consistent with the function of an inhibitory KIR , incubation of lymphocytes from Mamu-KIR3DL05+ macaques with target cells expressing Mamu-A1*00201 suppressed the degranulation of tetramer-positive NK cells . These observations reveal a previously unappreciated role for D1 polymorphisms in determining the selectivity of KIRs for MHC class I-bound peptides , and identify the first functional KIR-MHC class I interaction in the rhesus macaque . The modulation of KIR-MHC class I interactions by viral peptides has important implications to pathogenesis , since it suggests that the immunodeficiency viruses , and potentially other types of viruses and tumors , may acquire changes in epitopes that increase the affinity of certain MHC class I ligands for inhibitory KIRs to prevent the activation of specific NK cell subsets .
Natural killer ( NK ) cells are able to lyse infected or malignant cells without prior antigenic stimulation , and thus provide an important innate defense against infectious agents and tumors [1] , [2] . NK cell activation in primates is regulated in part through interactions between the highly polymorphic killer immunoglobulin-like receptors ( KIRs ) expressed on NK cells and their MHC class I ligands on target cells [1] , [2] . KIRs are type I integral membrane proteins with either two or three immunoglobulin ( Ig ) -like extracellular domains ( 2D or 3D ) that transduce either inhibitory or activating signals via long ( L ) or short ( S ) cytoplasmic domains , respectively . Engagement of inhibitory KIRs by MHC class I molecules on healthy cells normally suppresses NK cell activation [1] , [3] , [4] . However , if these interactions are perturbed , for instance as a result of MHC class I downregulation by HIV-1 Nef [5] , [6] , or presentation of a peptide antagonist [7] , this inhibition is lost resulting in NK cell activation and target cell lysis . In contrast to the T cell receptor , which is highly specific for a given peptide-MHC complex , KIRs typically recognize subsets of MHC class I molecules with common amino acid motifs in their α1 domains . Based on serological epitopes that correspond to defined sequences at positions 77-83 , all HLA-B molecules , and some HLA-A molecules , can be classified as either Bw4 or Bw6 allotypes [8] . Allotypes of KIR3DL1 have broad specificity for HLA-Bw4 ligands [9] , whereas KIRs specific for HLA-Bw6 have not been identified . All inhibitory KIRs that have been examined thus far also exhibit selectivity for peptides bound by their MHC class I ligands [10] , [11] , [12] , [13] , [14] , [15] , [16] . These observations are consistent with crystal structures of KIR2DL1 and KIR2DL2 in complex with their HLA-C ligands showing that KIR residues contact surfaces of the HLA class I α1 and α2 domains in an orthogonal orientation across C-terminal residues of the bound peptide [17] , [18] . However , the molecular basis for the selectivity of KIRs for different peptides bound by a particular MHC class I ligand has not been defined . Genetic evidence suggests that polymorphic differences in the KIR and HLA class I genes play an important role in determining the course of infection for a number of human viral pathogens , including HIV-1 [19] , [20] , hepatitis C virus [21] , human papillomavirus [22] and cytomegalovirus [23] . In the case of HIV-1 , combinations of both activating and inhibitory KIR3DL1/S1 and HLA-Bw4 alleles have been associated with delayed progression to AIDS [19] , [20] . NK cells expressing KIR3DS1 were also shown to suppress the in vitro replication of HIV-1 in target cells expressing HLA-Bw4 [24] . While these observations point to a role for KIR-MHC class I interactions in determining the outcome of HIV-1 infection , studies to address the functional significance of these interactions have been limited , in part , by the lack of a suitable animal model . Simian immunodeficiency virus ( SIV ) infection of the rhesus macaque is an important animal model for lentiviral pathogenesis and for AIDS vaccine development [25] . Rhesus macaques express MHC class I molecules that correspond to products of the classical HLA-A and -B genes ( Macaca mulatta; Mamu-A and -B ) , but not the HLA-C gene [26] , [27] . Consistent with the co-evolution of KIR and MHC class I molecules , genes for the two-domain KIRs specific for HLA-C have not been identified in macaques [28] , [29] . Instead , macaques have an expanded repertoire of KIR3DL genes characterized by extensive polymorphism and gene duplication [28] , [29] , [30] , [31] , [32] . Here we identify Mamu-A1*00201 , a common rhesus macaque MHC class I molecule with a Bw6 motif , as a ligand for multiple allotypes of Mamu-KIR3DL05 . We show that the binding of Mamu-KIR3DL05 to Mamu-A1*00201 is peptide-dependent , and that the relative avidity and peptide-selectivity of binding is determined by polymorphisms in the D0 and D1 domains . We also demonstrate that target cells expressing Mamu-A1*00201 suppress the degranulation of primary Mamu-KIR3DL05+ NK cells . These observations reveal a previously unappreciated role for D1 polymorphisms in determining the selective recognition of MHC class I-bound peptides by KIRs , and define the first functional KIR-MHC class I interaction in the rhesus macaque .
Samples of peripheral blood from Mamu-A1*00201+ rhesus macaques were stained with Mamu-A1*00201 tetramers folded with SIV peptides to establish baseline CD8+ T cell responses prior to beginning a vaccine study . To our surprise , Mamu-A1*00201 in complex with the Gag71-79 GY9 peptide stained a subset of CD8+CD3– lymphocytes from one animal ( Mm 337-07 ) . Plasma from this animal tested negative for SIV RNA and for antibodies to viral antigens , indicating that this animal had not been previously exposed to SIV . The majority of tetramer-positive cells expressed CD8α and CD16 , characteristic of NK cells that are capable of mediating cytolytic activity [33] , as well as additional NK cell markers including NKp46 , NKG2A , and NKG2D ( Fig . 1A ) . A subset of these cells also cross-reacted with an antibody to human KIR2D ( Fig . 1A ) . Although most of the tetramer-positive cells were CD16+CD3– NK cells , staining was also observed for CD8+CD3+ T cells ( Fig . 1B ) . A longitudinal comparison of the frequency of tetramer-positive CD8+CD3+ versus CD16+CD3– lymphocytes revealed that these two populations were relatively stable in this animal over more than a year , ranging from 0 . 16% to 0 . 69% for CD8+ T cells and from 5 . 1% to 9 . 8% for CD16+ NK cells . To investigate the contribution of the peptide bound by Mamu-A1*00201 to this unusual pattern of tetramer staining , whole blood was stained with Mamu-A1*00201 tetramers folded with peptides corresponding to eight different CD8+ T cell epitopes of SIV [34] . In addition to Gag71-79 GY9 , staining was also observed for Env788-795 RY8 , but not for any of the other tetramers ( Fig . 1C ) . Thus , the tetramer staining observed for primary NK cells and CD8+ T cells from Mm 337-07 was dependent on the peptide bound by Mamu-A1*00201 . Since KIRs are known to be expressed on subsets of human NK cells and CD8+ T cells [4] , [35] , [36] , we hypothesized that this pattern of tetramer staining might reflect Mamu-A1*00201 binding to a rhesus macaque KIR . Full-length KIR cDNA sequences were therefore cloned from the PBMCs of Mm 337-07 and sequenced . Six KIR3DL alleles , three KIR3DS alleles and two KIR2DL04 alleles were identified in this animal , and their predicted amino acid sequences are shown in Fig . 2 . To identify the receptor for Mamu-A1*00201 , Jurkat cells were transfected with constructs expressing each of the KIR alleles cloned from Mm 337-07 and stained with Mamu-A1*00201 tetramers . To differentiate transfected from untransfected cells , the KIR alleles were expressed from a bicistronic vector that co-expresses enhanced green fluorescent protein ( eGFP ) . Since not all KIRs are well expressed on the cell surface , and antibodies are not available to macaque KIRs , an HA tag was introduced at the N-terminus of the D0 domain of each KIR . Our rationale for introducing the HA tag at this position is based on a recent three-dimensional model showing that the N-terminus of KIR3DL1 is free and oriented away from surfaces that are predicted to contact the peptide-MHC class I complex [37] , and experiments demonstrating that the introduction of an epitope tag at the N-terminus of the D0 domain does not interfere with ligand recognition [38] . Following the electroporation of Jurkat cells with these KIR expression constructs , the cells were stained with Mamu-A1*00201 tetramers and with a monoclonal antibody to the HA tag . Transfected cells were identified by gating on the eGFP+ population , the surface expression of each KIR was verified by HA staining , and binding to Mamu-A1*00201 in complex with Gag71-79 GY9 versus Nef159-167 YY9 was assessed by tetramer staining . All of the KIRs were expressed on the cell surface under the conditions of this assay , as indicated by HA staining ( Fig . 3 ) . However , only Mamu-KIR3DL05*008 resulted in a detectable level of staining with the Gag71-79 GY9 tetramer ( Fig . 3A ) . At higher levels of surface expression , staining was also observed for Nef159-167 YY9 , indicating that this tetramer can bind to Mamu-KIR3DL05*008 under conditions of protein over expression ( Fig . 3A ) . These results identify Mamu-KIR3DL05*008 as a receptor for Mamu-A1*00201 , and indicate that the peptide bound by Mamu-A1*00201 can modulate this interaction . Phylogenetic comparisons of macaque KIR3DL sequences revealed that Mamu-KIR3DL05*008 belongs to a group of similar alleles found in both rhesus and cynomolgus macaques [29] . To determine if other allotypes of Mamu-KIR3DL05 could also bind to Mamu-A1*00201 , Jurkat cells were transfected with constructs expressing six additional Mamu-KIR3DL05 alleles , as well as six Mamu-KIR3DL07 alleles . The transfected cells were then stained with Mamu-A1*00201 tetramers folded with four different SIV peptides to assess binding; Gag71-79 GY9 , Env788-795 RY8 , Nef159-167 YY9 and Vif89-97 IW9 . One , or more , of the Mamu-A1*00201 tetramers bound to cells expressing each of the Mamu-KIR3DL05 alleles . Cells expressing Mamu-KIR3DL05*004 , -KIR3DL05*003 , -KIR3DL05*010 , -KIR3DL05*008 and -KIR3DL05*005 stained with Gag71-79 GY9 , Env788-795 RY8 and Nef159-167 YY9 , whereas cells expressing Mamu-KIR3DL05*001 and mmKIR3DL05x stained only with Gag71-79 GY9 or Nef159-167 YY9 ( Fig . 4A ) . In contrast , none of these KIRs bound to the Vif89-97 IW9 tetramer ( Fig . 4A ) . Furthermore , none of the Mamu-KIR3DL07 alleles resulted in a detectable level of staining for any of the Mamu-A1*00201 tetramers ( Fig . S1 ) . Hence , this interaction is dependent on the peptide bound by Mamu-A1*00201 and is specific for Mamu-KIR3DL05 . Mamu-KIR3DL05*003 , -KIR3DL05*008 and -KIR3DL05*010 , were indistinguishable in their pattern of tetramer staining ( Fig . 4A ) . This is reflected by the similarity in their values for the mean fluorescence intensity ( MFI ) of tetramer staining divided by the MFI of HA staining , which are provided in Table 1 as a quantitative comparison of tetramer binding corrected for differences in surface expression for each KIR . In accordance with the rank order of tetramer staining observed for primary NK cells ( Fig . 1C ) , staining was highest for Gag71-79 GY9 , followed by Env788-795 RY8 , and then Nef159-167 YY9 ( Fig . 4A and Table 1 ) . With the exception of a single amino acid difference in the first position of the D0 domain of Mamu-KIR3DL05*010 , each of these KIRs have identical Ig-like domains ( Fig . 4B ) . Relative to Mamu-KIR3DL05*008 , Mamu-KIR3DL05*004 exhibited an increase in the intensity of tetramer staining for Gag71-79 GY9 ( 1 . 3 fold ) , Env788-795 RY8 ( 2 . 0 fold ) and Nef159-167 YY9 ( 4 . 3 fold ) ( Fig . 4A and Table 1 ) . Since the Ig-like domains of Mamu-KIR3DL05*004 and Mamu-KIR3DL05*008 only differ by a single amino acid at position 138 ( Fig . 4B ) , the histidine residue at this position accounts for the increase in Mamu-KIR3DL05*004 binding to Mamu-A1*00201 . Based on a recently proposed three-dimensional model of KIR3DL1*015 bound to HLA-A*2402 [37] , this residue is predicted to lie at the base of the second MHC class I-contact loop of the D1 domain , and may alter the conformation of this loop in a way that enhances binding to Mamu-A1*00201 . Compared to Mamu-KIR3DL05*008 , decreases in the intensity of tetramer staining were observed for both Mamu-KIR3DL05*005 and -KIR3DL05*001 ( Fig . 4A ) . The intensity of staining for Mamu-KIR3DL05*005 , which differs from Mamu-KIR3DL05*008 by 14 amino acids ( Fig . 4B ) , was 2 . 7-fold lower for Gag71-79 GY9 , 2 . 3-fold lower for Env788-795 RY8 and 3 . 0-fold lower for Nef159-167 YY9 ( Table 1 ) . A much greater reduction in the intensity of tetramer staining was observed for Mamu-KIR3DL05*001 . Tetramer staining for Mamu-KIR3DL05*001 was only detectable with Gag71-79 GY9 at an intensity that was 75-fold lower than for Mamu-KIR3DL05*008 ( Table 1 ) . Since Mamu-KIR3DL05*001 and -KIR3DL05*008 differ by ten amino acids in D0 , but are otherwise identical in D1 and D2 ( Fig . 4B ) , this reduction in the avidity of binding to Mamu-A1*00201 is due to polymorphic differences in the D0 domain . Thus , similar to KIR3DL1-HLA-Bw4 interactions in humans [37] , [39] , polymorphisms in the D0 domain of Mamu-KIR3DL05 can dramatically affect binding to MHC class I ligands . In the case of mmKIR3DL05x , tetramer staining was observed for Nef159-167 YY9 , but not for Gag71-79 GY9 or Env788-795 RY8 ( Fig . 4A ) . This shift in the pattern of Mamu-A1*00201 tetramer staining almost certainly reflects differences in D1 , since mmKIR3DL05x has a unique D1 domain , but nearly identical D0 and D2 domains to other allotypes of Mamu-KIR3DL05 ( Fig . 4B ) . Using cryopreserved PBMCs from the original source of mmKIR3DL05x , we verified that mmKIR3DL05x represents a bona fide allele , and not a PCR artifact , by independently cloning and confirming the cDNA sequence for this allele , and by PCR amplification of a 2 . 0 kb region spanning intron 4 from genomic DNA with primers to unique sequences in exons 4 and 5 . Additional sequence comparisons revealed that the D1 domain of mmKIR3DL05x , as well as the leader peptide and the D0 domain , are identical to Mamu-KIR3DS02*00402 and mmKIR3DHa ( Fig . S2 ) . Thus , mmKIR3DL05x appears to be the product of a recombination event in which exon 4 ( encoding D1 ) was acquired , either by the introduction of exons 1-4 of a Mamu-KIR3DS gene into -KIR3DL05 or by the introduction of exons 5-9 of Mamu-KIR3DL05 into a -KIR3DS gene . A closer examination of mmKIR3DL05x revealed that seven of the thirteen differences in the D1 domain coincide with , or are immediately adjacent to , loops predicted to contact surfaces of the peptide-MHC class I complex [37] . These include a charge difference at position 144 in the second loop ( L2 ) and a cluster of six residues at positions 164–170 in the third loop ( L3 ) ( Fig . 4B and Fig . 5A ) . To determine if these differences account for the unique binding pattern exhibited by mmKIR3DL05x , we constructed recombinants in which these sequences were exchanged with the corresponding sequences of Mamu-KIR3DL05*008 , and tested them for binding to Gag71-79 GY9 versus Nef159-167 YY9 . Reciprocal L2 substitutions affected the avidity , but not the specificity , of tetramer binding ( Fig . 5B ) . In contrast , exchanging L3 residues switched the specificity , and altered the avidity , of binding to the Mamu-A1*00201 tetramers . The 3DL05*008/xL3 recombinant bound Nef159-167 YY9 , but not Gag71-79 GY9 , and the 3DL05x/*008L3 recombinant bound both Gag71-79 GY9 and Nef159-167 YY9 ( Fig . 5B ) . Hence , these results reveal a role for polymorphisms in the third predicted contact loop of the D1 domain in determining the selective recognition of different peptides bound by the same MHC class I molecule . Additional Mamu-KIR3DL05+ rhesus macaques were identified by sequence-specific PCR and screened for tetramer-positive NK cells and CD8+ T cells . The Gag71-79 GY9 tetramer stained subsets of CD8+CD3– and CD8+CD3+ lymphocytes in in peripheral blood from each of the Mamu-KIR3DL05+ animals , but not from Mamu-KIR3DL05- animals ( Fig . 6A and 6B ) . In accordance with the complex regulation of KIR expression , which is influenced by a number of factors including differences in gene content on different KIR haplotypes , differences in the repertoire of MHC class I genes and polymorphic differences in KIR genes [40] , [41] , [42] , [43] , there was considerable animal-to-animal variation in the frequency and intensity of tetramer staining ( Fig . 6B ) . Variation in the frequency of tetramer-positive CD8+CD3+ lymphocytes , particularly in Mamu-A1*00201− animals that do not have Gag71-79-specific CD8+ T cells , may also reflect changes in KIR expression on memory CD8+ T cells related to age and/or prior exposure to infectious agents , since similar changes have been associated with age and encounters with viral pathogens in humans [36] , [44] , [45] . Overall , these results demonstrate that the presence of the Mamu-KIR3DL05 gene is predictive of Gag71-79 GY9 staining in peripheral blood , that this pattern of tetramer staining is independent of Mamu-A1*00201 and SIV infection , and that the variability of staining is typical of the heterogeneity of KIR expression on human NK cells and CD8+ T cells [36] , [46] . The role of NK cells and CD8+ T cells that express Mamu-KIR3DL05 in SIV-infected animals remains to be determined . However , among the eight animals represented in Fig . 6B , there were no obvious differences in the percentage of tetramer-positive cells for either population that could be attributed to the status of SIV infection . Indeed , of the two uninfected animals ( Mm 177-05 and Mm RHAX18 ) , the two animals infected with attenuated SIVmac239 Δnef ( Mm 350-04 and Mm 376-04 ) , and the two animals infected with pathogenic SIVmac239 ( Mm R03035 and Mm 20-05 ) , each pair had among the lowest and the highest frequencies of tetramer-positive lymphocytes ( Fig . 6B ) . Some of the tetramer-positive CD8+CD3+ lymphocytes in the Mamu-A1*00201+ animals probably represent virus-specific CD8+ T cells , since we cannot differentiate binding of the Gag71-79 GY9 tetramer to Mamu-KIR3DL05 versus the T cell receptor . Nevertheless , in two of the three SIV-infected animals ( Mm 350-04 and Mm R02020 ) , the percentage of tetramer-positive cells was actually higher for the CD8+CD3− population than for the CD8+CD3+ population ( Fig . 6B ) . Although the explanation for this is presently unclear , it is possible that Mamu-KIR3DL05 interactions with Mamu-A1*00201 may suppress CD8+ T cell responses to the Gag71-79 GY9 epitope in Mamu-KIR3DL05+ animals , which could explain the inconsistent , and often weak , CD8+ T cell responses to Gag71-79 GY9 that we and others have observed in SIV-infected , Mamu-A1*00201+ macaques . To investigate the functional consequences of NK cell recognition of Mamu-A1*00201 , PBMC from four Mamu-KIR3DL05+ animals were stimulated with the MHC class I-deficient 721 . 221 cell line [47] , or with 721 . 221 cells expressing either Mamu-A1*00201 , -A1*01101 ( Mamu-A*11 ) or -B*010101 ( Mamu-B*01 ) , and stained for CD107a as a degranulation marker . The cells were also stained with Gag71-79 GY9 to differentiate Mamu-KIR3DL05+ NK cells from Mamu-KIR3DL05− NK cells . CD107a was upregulated on the surface of both tetramer-positive and tetramer-negative NK cells in response to parental 721 . 221 cells and 721 . 221 cells expressing Mamu-A1*01101 or -B*010101 ( Fig . 6C ) . In contrast , CD107a was suppressed on tetramer-positive NK cells , but not on tetramer-negative NK cells , in the presence of target cells expressing Mamu-A1*00201 ( Fig . 6C ) . The same pattern of NK cell activation/inhibition was also observed by intracellular cytokine staining for IFNγ ( Fig . S3 ) . Moreover , CD107a was suppressed on tetramer-positive NK cells from both Mamu-A1*00201+ and -A1*00201− animals ( Fig . 6C ) , indicating that these cells were responsive to Mamu-A1*00201 whether or not they were educated in the presence of this MHC class I molecule . These results are therefore consistent with the functional inhibition of Mamu-KIR3DL05+ NK cells by Mamu-A1*00201 .
Polymorphic differences in the KIR and HLA class I genes play an important role in determining the course of infection for HIV-1 and for a number of other viral pathogens [19] , [20] , [21] , [22] , [23] , [24] . However , studies to address the functional significance of KIR-MHC class I interactions have been hampered by the lack of a suitable animal model . In the present study , we identify Mamu-A1*00201 , an MHC class I molecule present in approximately 20% of Indian origin rhesus macaques [48] , as a ligand for multiple allotypes of Mamu-KIR3DL05 . Although the frequency of specific alleles of Mamu-KIR3DL05 remains to be determined , the Mamu-KIR3DL05 gene was present in 42% of the rhesus macaques ( 43 of 103 animals ) recently screened at the New England Primate Research Center by sequence-specific PCR . This suggests that animals expressing both Mamu-KIR3DL05 and -A1*00201 are sufficiently common for use in future studies to investigate the functional implications of this interaction with respect to the pathogenesis of SIV infection . Genotyping for Mamu-KIR3DL05 was predictive of Mamu-A1*00201 tetramer staining for primary NK cells and CD8+ T cells in peripheral blood . The pattern of staining observed for subsets of CD8+CD3− and CD8+CD3+ lymphocytes from Mamu-KIR3DL05+ animals , but not from Mamu-KIR3DL05− animals , is consistent with the variegated expression of KIRs on human NK cells and CD8+ T cells [4] , [35] , [36] , [49] , [50] , [51] . Tetramer staining was independent of Mamu-A1*00201 , reflecting the segregation of KIR and MHC class I genes on different chromosomes , and was detectable regardless of the status of SIV infection . Moreover , the variability in the frequency and intensity of tetramer staining among Mamu-KIR3DL05+ animals was typical of the heterogeneity of KIR expression on human NK cells and CD8+ T cells [36] , [46] . Although tetramer staining has been reported for NK cell clones and for transfected cells expressing human KIRs [11] , [12] , to our knowledge this is the first report of direct ex vivo tetramer staining of primary NK cells . Incubation of peripheral blood lymphocytes from Mamu-KIR3DL05+ macaques with target cells expressing Mamu-A1*00201 specifically suppressed the degranulation of tetramer-positive NK cells . These results are consistent with the functional inhibition of primary NK cells expressing Mamu-KIR3DL05 by Mamu-A1*00201 . Furthermore , this inhibition was observed for tetramer-positive NK cells from Mamu-A1*00201− as well as from Mamu-A1*00201+animals , indicating that these cells were responsive to Mamu-A1*00201 , whether or not they were educated in animals that express this ligand . Although the mechanisms of NK cell education are not fully understood [52] , there is evidence that the maturation of NK cells expressing inhibitory KIRs is dependent on interactions with self-MHC class I molecules , and that NK cells expressing a particular inhibitory KIR in the absence of an appropriate MHC class I ligand are rendered hyporesponsive [3] , [53] , [54] . Thus , the in vitro suppression of tetramer-positive NK cells from Mamu-A1*00201− animals by target cells expressing Mamu-A1*00201 implies that these cells were educated for recognition of another MHC class I ligand . This is perhaps not surprising given the complexity of the rhesus macaque MHC class I genes [55] , [56] , and the ability of KIRs to recognize multiple MHC class I ligands with common amino acid motifs in their α1 domains [9] , [57] . Based on haplotype modeling and phylogenetic comparisons , Mamu-KIR3DL05 is predicted to represent a single genetic locus [29] . Although KIR3DL05 is not orthologous to any of the human KIR genes , interactions between Mamu-KIR3DL05 and Mamu-A1*00201 resemble features of KIR3DL1 binding to HLA-Bw4 . A three-dimensional model of KIR3DL1*015 bound to HLA-A*2402 was recently constructed based on a crystal structure of KIR2DL1 in complex with HLA-C*04 [17] , [37] . This model predicts that surface-exposed loops in each of the three Ig-like domains of KIR3DL1 contact the HLA class I molecule over the C-terminus of the bound peptide , and that the specificity of KIR3DL1 for HLA-Bw4 is dependent on a salt bridge between glutamate 282 in the D2 domain of KIR3DL1 and arginine 83 in the α1 domain of HLA-Bw4 [37] . Consistent with this model , polymorphisms in the Ig-like domains of Mamu-KIR3DL05 were associated with differences in binding to Mamu-A1*00201 . Amino acid differences in D0 affected the relative avidity of Mamu-KIR3DL05 binding to Mamu-A1*00201 . Compared to Mamu-KIR3DL05*003/*008 , tetramer staining was diminished for both Mamu-KIR3DL05*001 and -KIR3DL05*005 , which differ by eight and ten residues in D0 respectively . In the case of Mamu-KIR3DL05*001 , which is otherwise identical to Mamu-KIR3DL05*003/*008 in D1 and D2 , binding to Mamu-A1*00201 was all but eliminated . These results are analogous to previous observations showing that polymorphisms in the D0 domain of KIR3DL1 modulate the avidity of binding to HLA-Bw4 ligands [37] , [39] . Polymorphisms in D1 altered the selective binding of Mamu-KIR3DL05 to Mamu-A1*00201 in complex with different SIV peptides . In contrast to other allotypes of Mamu-KIR3DL05 , mmKIR3DL05x preferentially bound to Mamu-A1*00201 folded with Nef159-167 YY9 rather than Gag71-79 GY9 . This difference in peptide preference mapped to six amino acids in the third D1 loop predicted to contact surfaces of the peptide-MHC class I complex . These results support a recent three-dimensional model of KIR3DL1*015 bound to HLA-A*2402 [37] , and reveal a role for polymorphisms in the D1 domain in determining the selectivity of KIRs for MHC class I-bound peptides . Interestingly , mmKIR3DL05x appears to be the product of a recombination event in which exon 4 sequences coding for the D1 domain were derived from a KIR3DS gene; an observation that is consistent with domain shuffling as a mechanism of KIR evolution in primates [58] . Unlike previously identified ligands for human KIRs , the α1 domain of Mamu-A1*00201 contains a Bw6 motif . In contrast to Bw4 , the Bw6 motif has a glycine rather than an arginine at position 83 ( N77LRNLRG83 ) . Yet , Mamu-KIR3DL05 retains a glutamate at position 285 , which corresponds to glutamate 282 of KIR3DL1 . Since the peptides recognized by Mamu-KIR3DL05 each contain a positively charged residue at position 6 or 8 ( Gag71-79 GSENLKSLY , Env788-795 RTLLSRVY and Nef159-167 YTSGPGIRY ) , it is conceivable that glutamate 285 may form an alternative salt bridge with the peptide that accounts for the peptide-dependence of Mamu-KIR3DL05 . However , a charge at this position does not appear to be sufficient for binding , since the Vif89-97 IW9 peptide , which also contains a lysine at position 6 ( ITWYSKNFW ) , did not result in detectable Mamu-A1*00201 tetramer staining . While the molecular interactions underlying the binding of Mamu-KIR3DL05 to Mamu-A1*00201 remain to be fully defined , these observations offer a potential explanation for the contribution of the peptide to this interaction , and perhaps suggest a more prominent role for certain peptides in KIR recognition of other Bw6 ligands . The extent to which KIR recognition of Bw6 ligands has been elaborated in the rhesus macaque is presently unclear . However , since this motif is retained in the MHC class I molecules of humans and macaques , the absence of human KIRs that recognize HLA-Bw6 appears to reflect the loss of receptors of this specificity during the course of human evolutionary history . While the reason for this is not understood , it may be related to the expansion of the lineage III KIR genes coding for KIR2DL/S receptors with a D1-D2 configuration , and a greater dependence on the regulation of NK cell activation through interactions with their HLA-C ligands . The identification of inhibitory KIRs that bind with high avidity to a common MHC class I molecule in the rhesus macaque in complex with SIV-derived peptides suggests a potential mechanism of immune evasion . The Nef proteins of HIV-1 and SIV selectively downregulate MHC class I molecules from the surface of infected cells to evade destruction by virus-specific CD8+ T cells [6] , [59] . However , the removal of these molecules from the cell surface increases the susceptibility of infected cells to elimination by NK cells [6] . By acquiring changes in CD8+ T cell epitopes that increase the binding of MHC class I ligands to inhibitory KIRs , the virus may prevent the activation of NK cells under conditions of incomplete downregulation by Nef . This possibility is supported by recent evidence that peptides can modulate NK cell activation by varying the affinity of HLA ligands for inhibitory KIRs [7] . Whereas Fadda et al . show that antagonistic peptides that disrupt MHC class I interactions with inhibitory KIRs leads to NK cell activation [7] , our data suggests that viruses may acquire changes in epitopes that stabilize these interactions to suppress NK cell activation in a way that favors virus replication . KIRs are also expressed on subsets of memory CD8+ T cells in HIV-1 infected individuals , and have been associated with a decrease in the responsiveness to TCR-dependent stimulation [44] , [60] . Thus , peptides that stabilize interactions with inhibitory KIRs may also suppress CD8+ T cell activation . Deleterious combinations of KIR and MHC class I alleles may therefore select for changes in epitopes of HIV-1 and SIV that inhibit certain NK cell and CD8+ T cell responses; a scenario that may further undermine the host's ability to contain virus replication . Consistent with this hypothesis , a single nucleotide polymorphism was recently identified as a marker for two Mamu-KIR3DL05 alleles that were more prevalent among SIV-infected rhesus macaques with high viral loads in animals [61] . The identification of Mamu-A1*00201 as a ligand for Mamu-KIR3DL05 now affords an opportunity to investigate the functional implications of KIR-MHC class I interactions . Using KIR- and MHC class I-defined animals , experiments can now be designed to examine the phenotypic changes that occur in a specific population of NK cells during the course of virus infection in a way the was previously only possible for CD8+ T cells . Characterization of the molecular interactions underlying the binding of Mamu-KIR3DL05 to Mamu-A1*00201 also promises to yield fundamental insights regarding the role of viral peptides in modulating KIR recognition of MHC class I ligands . The binding of Mamu-KIR3DL05 to Mamu-A1*00201 in complex with SIV peptides suggests that these interactions may be particularly important in determining the course of SIV infection .
All of the animals used for these studies were Indian origin rhesus macaques ( Macaca mulatta ) . These animals were housed at the New England Primate Research Center ( NEPRC ) and were maintained in accordance with standards of the Association for Assessment and Accreditation of Laboratory Animal Care and the Harvard Medical School Animal Care and Use Committee . Animal experiments were approved by the Harvard Medical Area Standing Committee on Animals and conducted according to the principles described in the Guide for the Care and Use of Laboratory Animals [62] . Rhesus macaque KIR sequences were submitted to Genbank and to the Immuno-Polymorphism Database ( www . ebi . ac . uk/ipd/kir/ ) [63] . Sequences that have been assigned official names are indicated with the prefix Mamu-KIR . In cases where official names have not yet been assigned , sequences are referred to using a provisional nomenclature indicated by the prefix mmKIR . The names and Genbank accession numbers for each of the KIR alleles in this study are listed in Table S1 . Whole blood was stained with Mamu-A1*00201 tetramers folded with the SIV peptides Gag71-79 GY9 , Env788-795 RY8 , Env317-325 KM9 , Nef248-256 LM9 , Nef159-167 YY9 , Env296-304 RY9 , Vif97-104 WY8 , or Vif89-97 IW9 ( 30 min , 37°C ) followed by antibodies to cell type-specific markers ( 30 min , 20°C ) . Mamu-A1*00201 tetramers were obtained from David Watkins' laboratory ( Wisconsin National Primate Research Center ) , and the quality of each tetramer lot was verified by staining CD8+ T lymphocytes from SIV-infected rhesus macaques . For polychromatic assays , samples were stained with anti-CD3-Pacific blue ( SP34-2 , BD Pharmingen ) , anti-CD4-AmCyan ( L200 , BD Pharmingen ) , anti-CD16-FITC ( 3G8 , BD Pharmingen ) , anti-HLA-DR-PE Texas Red ( Immu-257 , Immunotech ) , anti-CD20 PE-Cy5 . 5 ( L27 , BD Pharmingen ) , anti-CD56 PE-Cy7 ( NCAM16 . 2 , BD Pharmingen ) , anti-CD8α-Alexa 700 ( RPA-T8 , BD Pharmingen ) , anti-CD14-APC-Cy7 ( MphiP9 , BD Pharmingen ) , and either anti-NKG2A-PE ( Z1999 , Beckman Coulter ) , anti-NKp46-PE ( BAB21 , Immunotech ) , anti-KIR2D-PE ( NKVFS1 , Miltenyi Biotec Inc . ) , or anti-NKG2D-PE ( BAT221 , Miltenyi Biotec Inc . ) For four-color assays , samples were stained with anti-CD3-FITC ( SP34-2 , BD Pharmingen ) , anti-CD16-PE ( 3G8 , BD Pharmingen ) , and anti-CD8α-PerCP ( SK1 , BD Pharmingen ) . Samples were treated with FACS Lysing solution ( BD Biosciences ) to eliminate red blood cells , washed and fixed in 2% paraformaldehyde PBS . Data was acquired using a LSRII flow cytometer ( BD Biosciences ) and analyzed using FlowJo 8 . 8 . 6 ( Tree Star Inc . ) . Peripheral blood lymphocytes were isolated over Ficoll ( Sigma ) and aliquots of 2–10 million PBMC were frozen in Trizol ( Invitrogen ) . Total RNA was extracted using the RNeasy kit ( Qiagen ) according to the manufacturer's instructions . KIR cDNAs were amplified by reverse transcription-polymerase chain reaction ( RT-PCR ) using the Superscript III One-Step RT-PCR kit ( Invitrogen ) with modified versions of the Ig3Up and Ig3Down primers [64] . Cycling conditions included an RT step at 55°C for 30 min , a denaturation step at 94°C for 2 min , followed by 40 cycles of denaturation ( 94°C for 15 sec ) , annealing ( 55°C for 30 sec ) and extension ( 68°C for 90 sec ) , and a final extension step at 68°C for 5 min . PCR products were cloned into the pGEM-T Easy vector ( Promega ) and sequenced with T7 and SP6 sequencing primers . Sequences were analyzed using Sequencher 4 . 8 ( Gene Codes Inc . ) and MacVector 9 . 5 . 2 ( MacVector Inc . ) software packages . At least three identical cDNA clones were identified for each KIR allele . Rhesus macaque KIRs were PCR amplified from cDNA clones using primers to introduce an HA tag at the N-terminus of the D0 domain . The KIR cDNAs were then cloned into pCGCG , a bicistronic vector that co-expresses eGFP , in frame with an upstream sequence for the leader peptide of Mamu-KIR3DL05*008 . Jurkat cells ( 1×107 cells ) were electroporated ( 250V , 975µF ) with plasmid DNA ( 40 µg ) in serum-free RPMI ( 400 µl ) in a 0 . 4 cm cuvette ( BioRad ) . After resting ( 10 min , 20°C ) , the cells were re-suspended in RPMI medium ( 9 ml ) with 10% FBS and incubated overnight at 37°C , 5% CO2 . After 22 hours , the cells were stained with APC-conjugated tetramers ( 30 min , 37°C ) , followed by PE-conjugated anti-HA PE ( GG8-IF3 . 3 , Miltenyi Biotec Inc . ) ( 20 min , 20°C ) . The cells were washed and fixed in 2% paraformaldehyde PBS . At least 200 , 000 events were acquired using a FACSCalibur flow cytometer ( BD Biosciences ) and the data was analyzed using FlowJo 8 . 8 . 6 . Genomic DNA was extracted from 1–2 million PBMC using the DNAeasy kit ( Qiagen , Valencia , CA ) , and 10 ng was used as template in a 25 µl PCR reaction with forward and reverse primers ( GAGACCCATGAACTTAGGCTTC & GCAGTGGGTCACTGGGGA ) for amplification of a 156 bp sequence in exon 5 specific to Mamu-KIR3DL05 . Primers specific for a conserved 300 bp region of Mamu-DRB were included as an internal control [48] . Cycling conditions included a denaturation step at 96°C for 2 min followed by 30 cycles of denaturation ( 94°C for 30 sec ) , annealing ( 63°C for 45 sec ) and extension ( 72°C for 45 sec ) , and a final extension step at 72°C for 10 min . PCR products were separated on a 1% agarose gel containing ethidium bromide and visualized by UV transillumination . PBMC ( 1×106 cells ) were stimulated for 18 hours with 721 . 221 cells , or with 721 . 221 cells expressing rhesus macaque MHC class I molecules , at a 5:1 ratio in the presence of anti-CD107a PE-Cy5 ( clone H4A3 , BD Pharmingen ) , Golgi-Stop and Golgi-plug ( BD Pharmingen ) . The cells were then stained with APC-conjugated tetramers ( 30 min , 37°C ) , followed by anti-CD16-FITC , anti-NKG2A-PE , anti-CD8α-Alexa 700 and CD3 APC-Cy7 ( 20 min , 20°C ) . The cells were then permeabilized and stained for 30 min with anti-IFN-γ-PE-CY7 ( Clone 4S . B3 , BD Pharmingen ) . Samples were washed and fixed in 2% paraformaldehyde PBS . At least 200 , 000 lymphocyte events were collected using an LSRII flow cytometer , and the data was analyzed using FlowJo 8 . 8 . 6 . | NK cells provide an important first line of defense against infectious diseases and tumors by virtue of their ability to kill infected or malignant cells without prior sensitization . NK cell activation is regulated in part through interactions between KIRs expressed on the surface of NK cells and their MHC class I ligands on target cells . Here we identify Mamu-A1*00201 ( Mamu-A*02 ) , a common MHC class I molecule in the rhesus macaque , as a ligand for Mamu-KIR3DL05 . We show that this interaction is peptide-dependent , since soluble Mamu-A1*00201 tetramers folded with certain SIV peptides , but not others , stained cells expressing Mamu-KIR3DL05 . Differences in binding avidity were associated with polymorphisms in the D0 and D1 domains of Mamu-KIR3DL05 , whereas differences in peptide-specificity mapped to the D1 domain . These observations reveal a previously unappreciated role for D1 polymorphisms in determining the selectivity of KIRs for MHC class I-bound peptides , and identify the first functional KIR-MHC class I interaction in the rhesus macaque . These observations suggest that SIV , and potentially also HIV-1 , may acquire changes in epitopes that increase the avidity of MHC class I ligands for inhibitory KIRs as a mechanism of immune evasion to prevent the activation of certain NK cell subsets . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"virology/immunodeficiency",
"viruses",
"immunology/immune",
"response",
"microbiology/innate",
"immunity",
"immunology/innate",
"immunity",
"virology/animal",
"models",
"of",
"infection",
"infectious",
"diseases/hiv",
"infection",
"and",
"aids",
"infectious",
"diseases/viral",
... | 2011 | KIR Polymorphisms Modulate Peptide-Dependent Binding to an MHC Class I Ligand with a Bw6 Motif |
Chromosomal organization in 3D plays a central role in regulating cell-type specific transcriptional and DNA replication timing programs . Yet it remains unclear to what extent the resulting long-range contacts depend on specific molecular drivers . Here we propose a model that comprehensively assesses the influence on contacts of DNA-binding proteins , cis-regulatory elements and DNA consensus motifs . Using real data , we validate a large number of predictions for long-range contacts involving known architectural proteins and DNA motifs . Our model outperforms existing approaches including enrichment test , random forests and correlation , and it uncovers numerous novel long-range contacts in Drosophila and human . The model uncovers the orientation-dependent specificity for long-range contacts between CTCF motifs in Drosophila , highlighting its conserved property in 3D organization of metazoan genomes . Our model further unravels long-range contacts depending on co-factors recruited to DNA indirectly , as illustrated by the influence of cohesin in stabilizing long-range contacts between CTCF sites . It also reveals asymmetric contacts such as enhancer-promoter contacts that highlight opposite influences of the transcription factors EBF1 , EGR1 or MEF2C depending on RNA Polymerase II pausing .
We propose to use a generalized linear model with interactions ( GLMI ) to analyze the effects of genomic features such as architectural protein co-occupancies on chromatin contacts at genome-wide level: log ( E [ y | X ] ) = β 0 + β X = β 0 + β d d + β B B + β C C + β g g ( 1 ) Variable y denotes the number of Hi-C contacts for any pair of bins on the same chromosome . Variable set X = {d , B , C , g} comprises several variable subsets: the log-distance variable d , the bias variables B , the confounding variable set C and the genomic variable of interest g . The log-distance variable d accounts for the background polymer effect ( log-log relation between distance and Hi-C count ) [14] . Bias variables B = {len , GC , map} are known Hi-C biases including fragment length ( len ) , GC-content ( GC ) and mappability ( map ) that are computed as in [15] ( S1 Appendix , Bias variable computation ) . Including those bias variables into the model allows to correct for biases in Hi-C data . Bias normalization by matrix balancing methods [16] is avoided , because these methods might remove effect of genomic variable of interest . Variable g represents the genomic feature of interest , whose associated βg parameter value reflects its effects on chromatin contacts . Variable set C comprises confounding variables included to properly estimate βg . Model ( 1 ) is very general and can be developed in multiple versions depending on the variable g of interest . In the following paragraphs , we will see the different kinds of variables g . The corresponding models are detailed in Subsection Materials and Methods , The different models . We illustrate the different model variables in Fig 1 . For simplicity , we illustrate our model with protein binding sites , yet the same model is applicable to many other genomic features such as motifs or promoters . Let consider a pair of bins that we call left bin ( L ) and right bin ( R ) . The attribution for left and right bins is arbitrary . Let also consider 3 genomic features Fi ( whose binding is colored in blue in Fig 1 ) , Fj ( in red ) and Fk ( in green ) that represent binding sites of 3 different proteins . For the genomic feature Fi , occupancy variables ziL and ziR denote the occupancies of Fi on left and right bins , respectively . For an occupancy variable , a value of 0/1 means absence/presence of the corresponding feature on the bin , e . g . absence/presence of the protein on the bin ( a value between 0 and 1 means partial overlap of the feature ) . Occupancy variables are used to build 4 main kinds of model variables as follows . A “homologous interaction” variable nii is the product of ziL and ziR ( nii = ziL × ziR ) . The associated β n i i parameter reflects the extent by which the genomic feature Fi interacts with itself through chromatin contacts ( Fig 1a ) . For instance , distant CTCF binding sites were shown to form loops in human [10 , 17] . A “heterologuous interaction” variable nij is the average of the product ziL × zjR and the product zjL × ziR ( n i j = 1 2 ( z i L × z j R + z j L × z i R ) ) , because both products are identically associated to y . The associated β n i j parameter reflects the extent by which the genomic feature Fi interacts with another genomic feature Fj through chromatin contacts ( Fig 1b ) . For instance , enhancers are in long-range contacts with promoters to regulate target gene expression [14 , 18] . A “homologous interaction cofactor” variable ciik is the product of an interaction variable nii and an interaction variable nkk ( ciik = nii × nkk = ziL × ziR × zkL × zkR ) . Here we consider the cofactor Fk as a protein that does not directly bind to DNA , but which is instead bound by an insulator binding protein Fi ( IBP ) to DNA , such as cohesin is recruited by CTCF to DNA . Hence we expect that a cofactor will be found at both bins L and R in contact , e . g . cohesin ring entraps both chromatin fibers and is thus observed at both bins [10 , 17] . That explains why ciik is the product of nii and nkk . The associated β c i i k parameter reflects the extent by which chromatin contacts between genomic feature Fi and itself are mediated by a genomic feature Fk , the cofactor ( Fig 1c ) . A “heterologous interaction cofactor” variable cijk is the product of an interaction variable nij and an interaction variable nkk ( c i j k = n i j × n k k = 1 2 ( z i L × z j R × z k L × z k R + z j L × z i R × z k L × z k R ) ) . Here we consider the cofactor Fk as a protein that does not directly bind to DNA , but which is instead bound to two IBPs Fi and Fj . For instance , a loop can be mediated by CP190 that binds to BEAF-32 and GAF sites that are distant [19] . The associated β c i j k parameter reflects the extent by which chromatin contacts between genomic features Fi and Fj are mediated by a third genomic feature Fk , the cofactor ( Fig 1d ) . In the previous paragraphs , we introduced numerous variables that were the products of simpler variables , namely the occupancy variables . In ( generalized ) linear regression , those product variables are called “interaction” terms . To detect such interaction effects , one usually needs a large number of observations . We will see in the next subsections that the tremendous amount of data provided by Hi-C experiments allows to detect such interaction effects with accuracy . The model and the different variables will be illustrated with real world scenarios in the next subsections . We first sought to validate our model using experimental data . For this purpose , we focused on the Drosophila model because several insulator binding proteins ( IBPs ) that mediate long-range interactions have been well characterized in this organism . Drosophila IBPs comprise suppressor of hairy wing ( Su ( Hw ) ) , Drosophila CTCF ( dCTCF ) , boundary-element-associated factor of 32 kDa ( BEAF-32 ) , GAGA binding factor ( GAF ) , Zeste-White 5 ( ZW5 ) [20] , the general transcription factor dTFIIIC [9] and DNA replication-related element factor ( DREF ) [7] . We analyzed Kc167 Hi-C data at 10 kb resolution and focused on 20kb-1Mb distances for which contact frequencies were accurately measured experimentally [21] . At this distance range , the log-log relation between Hi-C count and distance was linear ( R2 = 0 . 99 , S1 Fig ) , supporting the use of the log-distance term in the model . The data comprised approximately 1 million of observations , which allowed to detect higher-order interactions with enough precision ( tight parameter confidence intervals reflected by low p-values , see below ) . Because of Hi-C count overdispersion , we used negative binomial regression as the most appropriate specification of the generalized linear model . It has been shown that BEAF-32 motifs can form long-range interactions with each other using both fluorescence cross-correlation spectroscopy [22] and high-resolution microscopy [23] . Following this observation , we first validated our model by successfully estimating long-range contacts between the BEAF-32 CGATA motifs using model ( 2 ) ( β ^ n i i = 6 . 7 × 10 3 , p < 10−20; Fig 2a; model ( 2 ) and all other models used in the following are described in Subsection Materials and Methods , The different models ) . This result was confirmed as we observed that the Hi-C count increased with co-occupancy of BEAF-32 motifs ( variable nii ) ( Fig 2b ) . We also observed long-range contacts between dCTCF motifs ( β ^ n i i = 2 . 4 × 10 4 , p = 3 × 10−14 ) , highlighting their important roles in loop formation in Drosophila as observed in human [10 , 17] . Over the 7 known IBPs , the model correctly identified all IBP motifs as involved in long-range contacts among themselves ( Fig 2c ) . Next the same approach was used to evaluate the model’s ability to discriminate between the 7 IBP motifs ( true positives ) and 83 other DNA-binding protein motifs ( false positives ) . This approach obtained good predictions ( area under the curve ( AUC ) = 0 . 855; Fig 2d ) . Among the motifs that we considered as false positives , M1BP and Ttk69K motifs presented high and significant interaction effects ( M1BP: β ^ n i i = 1 . 7 × 10 5; Ttk69K: β ^ n i i = 2 . 3 × 10 4 , p < 10−12 , resp . ) . These results suggested that M1BP and Ttk69K might represent new insulator-binding protein candidates . Accordingly , M1BP protein binds to the promoters of paused genes that were shown to be involved in long-range contacts [18 , 24] . Ttk69K protein has a homomeric dimerization BTB/POZ domain that could help bridging two distant proteins through long-range contacts [22] . We then used GLMI to study the role of cofactors that cannot directly bind to DNA , but are instead recruited by IBPs , and are required to mediate or stabilize long-range contacts between two IBP binding sites . In Drosophila , well-known cofactors include condensin I , condensin II , Chromator , centrosomal protein of 190 kDa ( CP190 ) , cohesin [19–22] , Fs ( 1 ) h-L [25] and lethal ( 3 ) malignant brain tumor ( L ( 3 ) Mbt ) [7] . Most notably , fluorescence cross-correlation spectroscopy ( FCCS ) experiments have shown that CP190 is required to bridge long-range contacts between two BEAF-32 binding sites [22] . Using ChIP-seq peak data with model ( 4 ) , we estimated a significant and positive effect of CP190 in mediating long-range contacts between BEAF-32 sites ( β ^ c i i k = 878 , p < 10−20; Fig 2e ) , in complete agreement with recent work [22] . Similar result was obtained for Chromator in mediating long-range contacts between BEAF-32 sites ( β ^ c i i k = 3 . 4 × 10 3 , p < 10−20 ) [22] . In addition , previous BEAF-32 mutation by our group has revealed that cofactor CP190 is also required to bridge long-range contacts between BEAF-32 and GAF binding sites [19] . Using ChIP-seq peak data with model ( 5 ) , we estimated a significant and positive effect of CP190 in bridging distant BEAF-32 and GAF sites ( β ^ c i j k = 1 . 3 × 10 3 , p < 10−20; Fig 2e ) [19] . We applied the same modeling approach to the 6 other known cofactors and found that all were associated with significant positive effects in mediating contacts between BEAF-32 and GAF binding sites ( all betas β ^ c i j k > 326 , all p-values p < 10−20; Fig 2f ) . Because CP190 was also shown to mediate long-range contacts between BEAF-32 and dCTCF , and between BEAF-32 and Su ( Hw ) [19] , we estimated the corresponding cofactor effects . We again found significant positive effect of CP190 between BEAF-32 and dCTCF ( β ^ c i j k = 892 , p < 10−20 ) , but our method only detected a slightly significant mediating effect of CP190 between BEAF-32 and Su ( Hw ) ( β ^ c i j k = 175 , p = 0 . 02 ) . In human , the most studied cofactor is cohesin that is able to entrap two chromatin fibers thereby stabilizing long-range contacts between CTCF sites [10 , 17] . Hence we assessed the impact of cohesin in mediating long-range contacts between two dCTCF binding sites in Drosophila . We found a significant and positive effect of cohesin ( β ^ c i i k = 105 . 8 , p < 10−20; Fig 2g ) , thus supporting a conserved function of cohesin in stabilizing long-range contacts between CTCF sites in metazoans . We further tested our model for cofactor effects using perturbed conditions such as the removal of these cofactors , as obtained through knocking-down ( KD ) followed by Hi-C experiment . Of note , Hi-C experiments are expensive and complex to carry out , and the possibility to predict long-range contacts upon such KD is of major importance . We compared the impact of cohesin in the context of long-range contacts bridging CTCF sites in WT and Rad21 ( cohesin subunit ) KD Hi-C data . Our model estimated a significant but lower cofactor effect of cohesin in Rad21 KD ( β ^ c i i k = 75 . 7 , p = 9 × 10−12 ) , compared to WT ( β ^ c i i k = 105 . 8 , p < 10−20 ) . The difference between WT and Rad21 KD associated coefficients was negative and significant ( beta difference = −30 . 1 , p = 0 . 027 ) , corresponding to a beta decrease of 28% ( Fig 2h ) . This result therefore validated the estimated effect of cohesin in mediating distant dCTCF binding sites , which decreased upon cohesin depletion as expected . Using real data , we concluded that our model successfully predicted the roles of IBP motifs in long-range contacts between distant loci , as well as the roles of known cofactors in bridging distant IBP binding sites . The GLMI predictions were validated in the literature and using protein KD followed by Hi-C experiment . We then compared GLMI with existing methods for their ability to identify genomic features known to be involved in long-range contacts . For this purpose , we compared GLMI with ( 1 ) enrichment test ( ET ) on highly confident chromatin interaction pairs as previously [26] , ( 2 ) correlation ( Cor ) on highly confident chromatin interaction pairs [27] and ( 3 ) random forests ( RF ) discriminating highly confident chromatin interaction pairs from non-interacting pairs [28] . As a first and simple benchmark , we assessed the different methods to identify long-range contacts between protein binding sites of the same proteins ( model ( 2 ) ) . We evaluated the ability to discriminate between architectural proteins known to be involved in long-range contacts ( 13 true positives including IBPs and cofactors ) and random protein peaks ( 100 false positives ) using receiver operating characteristic ( ROC ) curves . We observed that all four methods were very efficient to detect long-range contacts between known architectural protein binding sites ( Fig 3a ) . In particular , GLMI and Cor showed perfect predictions ( AUC = 1 ) . RF and ET were also very accurate ( AUC > 0 . 94 ) . Previous benchmark was an easy task because it relied on random protein peaks whose binding was very different from real protein binding . For a more realistic benchmark , we then evaluated the ability to discriminate between motifs whose proteins are known to be involved in long-range contacts ( 7 true positives ) and other DNA-binding protein motifs ( 83 false positives ) using ROC curves . Using this benchmark , all the four methods performed less well ( Fig 3b ) . However we found that GLMI clearly outperformed the three other methods to detect long-range contacts between DNA motifs known to be involved in chromatin interactions ( AUCGLMI = 0 . 855 ) . Another benchmark consisted in identifying long-range contacts between binding sites of a protein and active promoters . Here , as previously , we evaluated the ability to discriminate between architectural proteins known to be involved in enhancer-promoter contacts ( 13 true positives including IBPs and cofactors ) and random protein peaks ( 100 false positives ) using ROC curves . We observed that all four methods were very efficient to detect long-range contacts between known architectural protein binding sites and active promoters ( Fig 3c ) . In particular , GLMI and Cor showed excellent predictions ( AUCGLMI = 0 . 985 and AUCCor = 1 ) . We then evaluated the ability to discriminate between motifs whose proteins are known to be involved in enhancer-promoter contacts ( 7 true positives ) and other DNA-binding protein motifs ( 83 false positives ) using ROC curves . Both GLMI and Cor performed well ( AUCGLMI = 0 . 797 and AUCCor = 0 . 807; Fig 3d ) . Conversely , ET and RF showed lower perfomance ( AUCET = 0 . 728 and AUCRF = 0 . 601 ) . We next analyzed the impacts of mutations in the consensus dCTCF motif . Single nucleotide polymorphisms ( SNPs ) play an important role in common genetic diseases and recent works have uncovered differential long-range contacts due to variations in the CTCF motif in human [17 , 29 , 30] . Hence we evaluated the methods to detect the impacts of single nucleotide mutations in the dCTCF motif . For this purpose , we considered the dCTCF consensus motif AGGTGGCG ( wild-type motif ) [31] and generated dCTCF motifs with single nucleotide mutations for each position ( mutated motifs ) . For instance , for the first position , the mutated motifs were TGGTGGCG , GGGTGGCG and CGGTGGCG . Over the 24 possible mutated motifs ( 8 positions × 3 alternative nucleotides ) , GLMI detected 17 motifs ( 71%; Fig 3e ) with homologous interaction variable betas that were lower than the one of the wild-type motif , indicating that the corresponding mutations diminished the ability of dCTCF to bridge long-range contact . Compared to GLMI , other approaches showed lower performance ( Cor: 14/24; RF = 10/24; ET = 8/24 ) . In addition to its better prediction performances , our model presents several theoretical advantages over the three other methods as summarized in Fig 3f . All the methods can assess long-range contacts between protein binding sites . However , GLMI is the only model that , at the same time , ( 1 ) accounts for the contact frequency which can vary among highly confident loops , ( 2 ) can deal with the presence of colocalization among proteins using conditional independence , ( 3 ) allows variable selection using lasso or stepwise , and ( 4 ) can assess the effect of cofactors by including higher-order interaction terms . Given the biological validation of our model , we next sought to address the roles of IBP motifs in establishing or maintaining long-range interactions in Drosophila . We first assessed how IBP motifs were coupled to form loops ( i . e . for all combinations of distant IBP motifs ) . For this purpose , we estimated homologous and heterologous interaction variable effects for any couple of IBP motifs using models ( 2 ) and ( 3 ) , and using the same Hi-C data , distance range and resolution as above ( Fig 4a ) . The strongest long-range contacts were between dCTCF and DREF motifs ( β ^ n i j = 2 . 8 × 10 4 , p < 10−20 ) , between dCTCF motifs ( β ^ n i i = 2 . 4 × 10 4 , p < 10−20 ) and between DREF motifs ( β ^ n i i = 2 × 10 4 , p < 10−20 ) . High levels of long-range contacts were also found between BEAF-32 and DREF motifs ( β ^ n i j = 1 . 9 × 10 4 , p < 10−20 ) and between BEAF32 and dCTCF motifs ( β ^ n i j = 1 . 9 × 10 4 , p < 10−20 ) . Thus in Drosophila , chromatin loops not only involve dCTCF motifs but also DREF and BEAF-32 motifs that all work together . We then explored if these long-range contacts depended on the distance between motifs . At short distance ( <100kb ) , long-range contacts were mainly detected between DREF motifs ( β ^ n i i = 1 . 8 × 10 4 , p < 10−20 ) , whereas at long distance ( > 750kb ) , they were more frequent between dCTCF and DREF motifs ( β ^ n i j = 3 . 5 × 10 4 , p = 7 × 10−9 ) ( Fig 4b ) . In addition , long-range contacts between dCTCF motifs peaked at 500 kb . Our results therefore raise the possibility that long-range contacts between IBP motifs could be distant-dependent . This observation might provide a molecular explanation for the observed hierarchical nature of 3D chromatin structure [32 , 33] , for which loops could be formed at different scales by the interplay of specific proteins . Next we sought to comprehensively test whether motif orientation could influence long-range contacts , as originally shown for CTCF motifs in human [10] and more generally in mammals [34] . We distinguished the motifs that were on the positive DNA strand ( denoted + ) , from those that were on the negative DNA strand ( denoted - ) . Then it was possible to compute four types of homologous interaction variables: nii+− = ziL+ × ziR− ( orientation →← ) , nii−+ = ziL− × ziR+ ( orientation ←→ ) , nii−− = ziL− × ziR− ( orientation ←← ) , nii++ = ziL+ × ziR+ ( orientation →→ ) . The corresponding models are detailed in Subsection Materials and Methods , The different models . Here we processed data at 1 kb resolution for better accuracy in distinguishing the different orientations . Similarly to in human and mammals , we found significant long-range contacts for motifs in convergent orientation ( β ^ n i i = 570 , p = 2 × 10−3 ) , and no significant contacts for the 3 other possible orientations ( ←→ , →→ and ←←; Fig 4c ) , revealing conservation of convergent CTCF mediated loops in agreement with 4C analyses [35] . We then assessed motif orientation for all other IBP motifs . Of note , the orientation of DREF TATCGATA motifs could not be assessed because of its palindromic property . For BEAF-32 , dTFIIIC and Su ( Hw ) motifs , we could not detect any strong orientation effect ( Fig 4c ) . Conversely , for GAF and ZW5 motifs , we found stronger contacts for motifs in divergent orientation ( ←→ ) compared to convergent orientation ( →← ) , suggesting a different mode of binding of the corresponding protein to DNA or a different constraint depending of its interaction with cofactors . Thus motif orientation in loops depends on the protein involved , and the dependence on convergent orientation of motifs does not apply to all insulator binding proteins . IBP binding sites might significantly vary depending on the cell type and stage . Hence we reanalyzed the roles of IBP binding in Kc167 Drosophila cells using available ChIP-seq data ( same cell type with Hi-C data; ZW5 data were not available ) . As in the previous subsection , we estimated interaction effects for any couple of IBP motifs using models ( 2 ) and ( 3 ) . Similarly to the analysis of IBP motifs , we observed high levels of long-range contacts involving DREF and dCTCF ( Fig 5a ) . In particular , we found strong long-range contacts between distant DREF binding sites ( β ^ n i i = 147 , p < 10−20 ) and between dCTCF and DREF binding sites ( β ^ n i j = 133 , p < 10−20 ) . However , we also observed strong long-range contacts between DREF and dTFIIIC ( β ^ n i j = 119 , p < 10−20 ) , and between DREF and GAF ( β ^ n i j = 112 , p < 10−20 ) , which could not be detected by previous analysis of IBP motifs . We then built a graph using estimated betas by adding an edge between two proteins Fi and Fj with a weight β ^ n i j , and by adding an edge between a protein Fi and itself with a weight β ^ n i i ( Fig 5b ) . Analysis of the graph clearly revealed the role of DREF as a hub , i . e . DREF was involved in many long-range contacts with other IBPs , such as BEAF-32 , DREF , dTFIIIC and GAF . Such DREF-mediated loops might be in apparent contradiction with recent experiments showing that DREF motifs tag proximal activation of housekeeping genes , in contrast to long-range activation of developmental genes [36] . However such DREF-mediated loops can be explained by long-range contacts between promoters ( β ^ n i i = 203 , p < 10−20 ) . Previous results should be carrefully interpreted since IBPs often linearly colocalize ( i . e . correlate ) with each other on the chromosome [31] . Such correlations can lead to “indirect” long-range contacts between IBPs . For instance , if a loop is maintained by two distant dCTCF binding sites , and that BEAF-32 colocalizes to dCTCF , then it is likely that we will also observe loops between distant BEAF-32 and dCTCF sites , and even between BEAF-32 sites . The impact of such correlations between proteins in the study of 3D chromatin has been discussed in details [12] . Models ( 2 ) and ( 3 ) could not account for such correlations between IBPs because only one interaction variable term was included . Instead one should use another model that includes all possible interaction variable terms between IBPs ( model ( 10 ) , see Subsection Materials and methods , The different models ) . To better discard indirect long-range contacts between the 6 IBPs , we thus re-estimated interaction variable beta parameters using model ( 10 ) that included all marginal variables ( 6 variables , one for each IBP ) and all interaction variables ( 21 variables , one for each combination of IBPs ) . Using model ( 10 ) , we obtained rather different results ( Fig 5c ) . We still observed strong long-range contacts between DREF binding sites ( β ^ n i i = 25 , p < 10−11 ) . However other long-range contacts were observed such as between BEAF-32 sites ( β ^ n i i = 30 , p < 10−20 ) . In turn , such analysis showed that an IBP tended to interact more with itself ( homologous interactions ) than with another IBP ( heterologous interactions ) ( p = 0 . 018; Fig 5d ) , in agreement with insulator bodies observed by microscopy [37] . In addition , the model ( 10 ) allowed to infer negative and significant interaction effects , such as between distant DREF and BEAF-32 ( β ^ n i j = - 25 , p < 10−11 ) , which could not be detected before . This negative effect means that BEAF-32 and DREF tend to avoid each other in long-range contacts , i . e . they tend to have a repulsive effect . This might reflect the known antagonistic relationship between BEAF-32 and DREF in competing for binding to overlapping binding sites [38 , 39] . As previously , we built a graph of betas and could detect groups of IBPs that may cluster together through long-range contacts as found for the two connected components BEAF-32/dTFIIIC/GAF and DREF/Su ( Hw ) /dCTCF , respectively ( Fig 5e ) . Interestingly , these two classes of IBPs that worked together in 3D were different from the two classes that were previously identified by 1D analysis: dCTCF/BEAF-32 and Su ( Hw ) , respectively [40] . Such observations strenghtened the importance of analyzing protein complexes in 3D in complement to 1D analysis ( see Discussion ) . In human and mammals , the main model of loop formation involves CTCF and cohesin [10 , 17] . According to this model , a loop may form by the homodimerization of two CTCF proteins bound to two distant CTCF motifs that are in convergent orientation [10] . The loop also involves cohesin that is recruited by CTCF and that has the ability to entrap the two DNA fibers inside a ring . In addition to CTCF and cohesin , other architectural proteins have been recently uncovered such as ZNF143 [41] and PcG proteins [42] . In order to systematically analyze proteins mediating loops , we considered integrating available protein binding data ( 73 proteins ) together with high-resolution Hi-C data in human GM12878 cells using our GLMI model . As previously done for Drosophila , we analyzed Hi-C data at 10 kb resolution and focused on 20kb-1Mb distances [10] . At this distance range , the Hi-C data comprised a very large number of bin pairs ( around 22 millions ) , and hence , its analysis often required subsampling to few million pairs to achieve tractable regression parameter estimation . As for Drosophila , the log-log relation between Hi-C count and distance was linear at this distance range ( R2 = 0 . 992 , S2 Fig ) , supporting the use of the log-distance term in the model . We first investigated contacts between distant CTCF binding sites using model ( 2 ) . As expected , we observed strong long-range contacts ( β ^ n i i = 37 , p = 6 × 10−12 ) [10] . Moreover high levels of long-range contacts were detected between cohesin subunit Rad21 binding sites as expected ( β ^ n i i = 89 , p < 10−20; Fig 6a ) [10] , as well as between cohesin subunit SMC3 ( β ^ n i i = 75 , p < 10−20 ) . We then used the same approach to estimate long-range contacts for all 73 proteins available ( S1 Table ) . Among the proteins that significantly interacted among themselves , we found several proteins known to colocalize to CTCF binding sites including YY1 ( β ^ n i i = 31 , p < 10−20 ) , MAZ ( β ^ n i i = 16 , p < 10−20 ) and JUND ( β ^ n i i = 258 , p = 10−9 ) [7] . We also found P300 , an important transcriptional coactivator [43] ( β ^ n i i = 264 , p < 10−20 ) . In addition , histone marks including H3K27me3 , H3K36me3 , H3K4me2 , H3K4me3 , H3K9ac and H3K9me3 showed homologous long-range contacts , as previously shown by polymer simulations [44] ( all β ^ n i i > 0 . 05 , p < 10−20 ) . Curiously , H4K20me1 sites presented repulsive effects with each other ( β ^ n i i = - 0 . 07 , p < 10−20 ) , indicating that distant H4K20me1 marked sites may avoid each other . We further estimated the well-known influence of cohesin in mediating long-range contacts between distant CTCF binding sites in human using model ( 4 ) [8 , 10] . Interestingly , we found that the effect of cohesin depended on the distance between CTCF binding sites , with no significant contacts for short distances ( 20-300kb: β ^ c i i k = - 3 × 10 3 , p = 0 . 63; 300-700kb: β ^ c i i k = - 1 × 10 4 , p = 0 . 15 ) and significant contacts for long distances ( 700-1000kb: β ^ c i i k = 4 × 10 4 , p = 3 × 10−6 ) ( Fig 6b ) . This suggested that cohesin is required for stabilizing CTCF-mediated loops for long distances , but is not necessary for short distances for which homodimerization of CTCF might be sufficient . We also sought for other proteins whose loops could be mediated by cohesin for long distances ( S2 Table ) . Most notably , we found that cohesin positively influences long-range contacts between architectural protein ZNF143 binding sites ( β ^ c i i k = 4 . 8 × 10 4 , p = 2 × 10−9 ) , between PolII binding sites ( β ^ c i i k = 446 , p = 6 × 10−16 ) , and between transcriptional factor binding sites ( EGR1 , ELF1 , FOXM1 , MAZ , MXI1 , NRF1 , YY1 ) , which suggests a wider role for cohesin in mediating long-range contacts . Further analyses of long-range contacts for every couple of proteins were performed using model ( 10 ) that included together all possible interaction variables . We considered 73 proteins , 7 histone modifications , active enhancers and active promoters . The model thus comprised ( 82 × 83 ) /2 = 3403 interaction variables . To deal with such a large number of interaction variables , we used a Poisson lasso estimation [45] . An interaction variable beta of zero was expected to reflect the absence of direct long-range contact between two proteins . From the estimated betas , we built a first graph that we called “attraction graph” by adding an edge between two proteins Fi and Fj if β ^ n i j > 0 , and by adding an edge between a protein Fi and itself if β ^ n i i > 0 ( Fig 6c ) . To identify hubs in the graph , we used eigenvector centrality that reflected how central is a node ( Fig 6d ) . Both active and repressed chromatin marks as well as enhancers were the most central nodes ( H3K9ac: score = 1; H3K9me3: score = 0 . 98; H3K4me3: score = 0 . 948; Enhancer: score = 0 . 84 ) . Among DNA-binding proteins , CTCF and Rad21 showed high values ( CTCF: score = 0 . 619; Rad21: score = 0 . 555 ) . Surprisingly , however , other proteins MEF2C and FOXM1 presented the highest values ( MEF2C: score = 0 . 725; FOXM1: score = 0 . 692 ) . Previous studies showed that MEF2C is necessary for bone marrow B-lymphopoiesis ( GM12878 is a lymphoblastoid cell line ) [46] , and that FOXM1 has an important role in maintenance of chromosomal segregation [47] . We then looked for cliques in the graph , i . e . a group of nodes that were all connected to each other ( complete list in S3 Table ) . As expected , we found a clique composed of CTCF and the cohesin subunits Rad21 and SMC3 , that are known to mediate together loops [10] . But we also found novel protein complexes that were specific to lymphocyte B such as the clique IKZF1/RFX5/PolII . IKZF1 plays a role in the development of lymphocytes [48] , RFX5 is involved in bare lymphocyte syndrome [49] and polymerase II catalyzes gene transcription . In addition , we found many cliques involving Polymerase III ( PolIII ) such as the cliques MEF2C/RUNX3/PolIII and MEF2C/WHIP/PolIII , which might reflect the influence of architectural protein RNA polymerase III-associated factor ( TFIIIC ) at tRNA genes [2 , 50] . Very little is known about repulsion effects between distant binding sites . Such repulsive effects could result from allosteric effects of loops [51] , or factors that disassociate protein complexes involved in loops [52] . To investigate repulsive effects , we built a second graph that we called “repulsion graph” by adding an edge between two proteins Fi and Fj if β ^ n i j < 0 , and by adding an edge between a protein Fi and itself if β ^ n i i < 0 ( Fig 6e ) . The repulsion graph was very different from the attraction graph . Different histone marks were central in the repulsion graph , including H3K36me3 ( score: 1 ) and H4K20me1 ( score: 0 . 974 ) , except histone mark H3K9me3 ( score: 0 . 798 ) that was central in both the attraction and repulsion graphs ( Fig 6f ) . Interestingly , we found that enhancers presented a high centrality score in the repulsion graph ( score: 0 . 766 ) , as found in the attraction graph . This result highlights the ability of enhancers to specifically interact with distant protein partner binding sites while avoiding others . Supporting this interpretation , we found enhancers to be in attraction with CFOS , NRF1 or POU2F2 , and in repulsion with RXRA , NFE2 or P300 . We then looked at pairs of proteins that were in repulsion . Most notably , we found CTCF to be in repulsion with EZH2 , which might result from steric effects of CTCF-mediated loops [10] with Polycomb-mediated loops [42] . Enhancer-promoter ( EP ) interactions play an essential role in the regulation of gene expression [14 , 18] . Therefore , we explored the roles of DNA-binding proteins in establishing or maintaining EP interactions . Before assessing the role of proteins , we first measured long-range contacts between active enhancers and promoters depending on gene expression using model ( 3 ) ( Fig 7a ) . We observed an attraction effect between active enhancers and highly expressed gene promoters ( β ^ n i j = 2 , p = 3 × 10−5 ) , and conversely , a repulsion effect between active enhancers and low expressed gene promoters ( β ^ n i j = - 1 . 7 , p < 1 × 10−20 ) , in complete agreement with the established positive influence of long-range contacts on gene expression [53] . To identify the influence of DNA-binding proteins , we then assessed the presence of long-range contacts between lymphocyte B transcriptional activator binding sites ( ChIP-seq data ) and promoters using the same model ( 3 ) . All lymphocyte B transcriptional activators including BCL11A , EBF1 , EGR1 , MEF2C , PAX5 and TCF12 showed long-range contacts with highly expressed gene promoters , compared to weakly transcribed gene promoters ( Fig 7b ) . This clearly showed that lymphocyte B transcriptional activators regulate expression of target genes through long-range contacts . Among the proteins available , we could not identify any that acted as silencers , i . e . proteins whose long-range contacts are high with low expressed gene promoters and low with highly expressed gene promoters . However when we focused on histone modifications , we found that long-range contacts of H3K27me3 mark were stronger to weakly transcribed gene promoters ( β ^ n i j = 0 . 06 , p < 10−20 ) , compared to highly expressed gene promoters ( β ^ n i j = - 0 . 2 , p < 10−20 ) ( Fig 7c ) . This suggested that H3K27me3 mark not only acts as a transcriptional silencer in linear proximity [54] , but could also repress target genes at distance through loops . Conversely , active marks such as H3K4me3 and H3K9ac interacted more with highly expressed genes . Because enhancer-promoter contacts were previously shown to be associated with Polymerase II pausing [18] , we then assessed enhancer-promoter interactions depending on gene transcription pausing . As expected , we found higher EP contacts at paused genes ( β ^ n i j = 62 . 2 , p = 10−3 ) , compared to genes in elongation ( β ^ n i j = 49 . 3 , p = 2 × 10−3 ) . We then looked at the influence of DNA-binding proteins ( Fig 7d ) . For instance , EBF1 sites showed higher long-range contacts with promoters of genes in pause ( β ^ n i j = 39 . 7 , p = 1 × 10−13 ) , compared to those in elongation ( β ^ n i j = 17 . 8 , p = 3 × 10−5 ) , in agreement with [18] . But , surprisingly , we also found that BCL11A sites showed higher long-range contacts with promoters of genes in elongation ( β ^ n i j = 72 . 8 , p < 10−20 ) than with genes in pause ( β ^ n i j = 60 . 9 , p = 2 × 10−11 ) . These observations suggest that , depending on the protein involved , long-range contacts with promoters are not always associated with pausing , but could also be linked to elongation . Here , we propose to use a generalized linear regression with interactions ( GLMI ) to study the roles of genomic features such as DNA-binding proteins , motifs or promoters to bridge long-range contacts in the genome , depending on transcriptional status or motif orientation . GLMI has multiple assets over existing approaches such as enrichment test , correlation and random forests . Compared to enrichment test [2 , 55] or correlation [27] that respectively assesses the protein enrichment or correlation at highly confident loops , GLMI quantitatively links the frequency of all long-range contacts to complex co-occupancies of proteins while accounting for known Hi-C biases and polymer background . Moreover , GLMI accounts for colocalizations among protein binding , a strong issue when analyzing protein binding sites known to largely overlap over the genome . In contrast to random forests [28] which are efficient predictive models but sometimes poor explanatory ones , GLMI allows to identify key chromatin loop driver proteins and motifs . GLMI can also uncover numerous mechanisms behind loop formation using higher-order interaction terms and proper confounding variables . For instance , GLMI can determine if a cofactor is necessary to mediate long-range contacts between distant protein binding sites . Using real Drosophila Hi-C and ChIP-seq data , we validate numerous GLMI predictions of long-range contacts that involve insulator binding proteins , cofactors and motifs , and which were confirmed by previous microscopy and mutational studies . For instance , our model estimates long-range contacts between distant BEAF-32 motifs , which were previously observed with both fluorescence cross-correlation spectroscopy [22] and high-resolution microscopy [23] . In addition , our model finds a mediating role of CP190 in bridging long-range contacts between distant BEAF-32 and GAF binding sites , in agreement with mutational experiments [19] . Of interest , GLMI analyses highlight a role of cohesin in stabilizing long-range contacts between CTCF sites in Drosophila , similarly to its role in human [7] . Supporting this role , we show that such influence is reduced upon cohesin subunit Rad21 depletion . It has to be noted that the absence of complete loss of contacts between CTCF sites after Rad21 depletion can be explained by the fast turnover of chromosome-bound cohesin in interphase [56] . Moreover , GLMI outperforms enrichment test , correlation and random forests in the identification of known architectural proteins and motifs , and in the detection of the effects of mutations in the dCTCF motif . The proposed model also uncovers several novel results . In Drosophila , GAF and ZW5 motifs are shown to act in divergent orientation to form loops , in contrast to CTCF motifs that are found in convergent orientation in Drosophila and human [10 , 17] , suggesting a different mode of action of corresponding proteins . In addition , we identify two groups of proteins that act in 3D to form loops . The first group comprises BEAF-32 , dTFIIIC and GAF , and the other group includes DREF , Su ( Hw ) and dCTCF . Those groups are different from the ones observed with 1D analysis only ( i . e . linear colocalization on the genome ) [40] , highlighting the importance of 3D analysis using GLMI . In human , we identify numerous long-range contacts between protein binding sites . In addition to the well-known protein complex CTCF/RAD21/SMC3 , we uncover new protein complexes that are specific to lymphocyte B such as IKZF1/RFX5 . We also found that enhancers could be either in long-range contact or repulsion with certain protein binding sites , highlighting potential specificity in selecting protein partners for long-range contacts . Our observations therefore support the idea that enhancer-promoter contacts are not solely driven by insulators or TAD borders that physically constrain such long-range interactions [29 , 36 , 57] . Rather , enhancer-promoter contacts may also be encoded by the specificity of protein-protein interactions . In addition , our results suggest that repressive mark H3K27me3 does not only repress genes that are contigous [54] , but it could also repress from a distance through the juxtaposition of H3K27me3 with genes in 3D . We also find that , depending on the protein involved , long-range enhancer-promoter contacts are not always favored by PolII pausing [18] , which may highlight distinct mechanisms by which proteins can influence transcription-associated long-range contacts . There are several limitations of the proposed approach . First , the present analysis is restricted to a 10-kb resolution because of the quadratic complexity of Hi-C data . Second , our analysis is limited by the amount of higher-order interaction variable parameters that can be learned within the same model ( full model ) using current parameter learning programs . Most notably , all possible interaction cofactor variables cannot be included in the same model because of the cubic complexity of such model , and hence they are learned separately instead ( using models ( 4 ) and ( 5 ) ) . In addition , although generalized linear models can include interactions of any order involving large protein complexes ( for instance , complexes of more than 4 proteins ) , parameter learning is limited by the availability of data and computational resources . Increasing depth of Hi-C data will allow inference of more complex models in the near future . Moreover the development of new big data learning algorithms could be used to process the data at a higher resolution that would allow in-depth analysis of 3D chromatin drivers [58] . An alternative to the exploration of all possible higher-order interactions together might be to guide the search using prior information , such as protein-protein interaction network [55] . Lastly , in order to explore all possible higher-order interaction variables within the same model ( full model ) , one should use a lasso regression model with hierarchically constrained interactions [59] .
We used publicly available high-throughput chromatin conformation capture ( Hi-C ) data from Gene Expression Omnibus ( GEO ) accession GSE62904 [21] . Hi-C experiments have been done for Drosophila melanogaster wild-type and Rad21 knock-down Kc167 cells with DpnII restriction enzyme . Hi-C data were binned at 1 and 10 kb resolutions . For human data analysis , we used publicly available Hi-C data of lymphoblastoid cells GM12878 cells from Gene Expression Omnibus ( GEO ) accession GSE63525 [10] . We used Hi-C data binned at 10 kb resolution . For Drosophila analysis , we used publicly available binding profiles of chromatin proteins of Drosophila melanogaster wild-type embryonic Kc167 cells . ChIP-seq data for CP190 , Su ( Hw ) , dCTCF and BEAF-32 were obtained from GEO accession GSE30740 [60] . ChIP-seq data for Barren ( condensin I ) , Cap-H2 ( condensin II ) , Chromator , Rad21 ( cohesin ) , GAF and dTFIIIC were obtained from GEO accession GSE54529 [9] . ChIP-seq data for DREF and L ( 3 ) Mbt were obtainted from GEO accession GSE62904 [21] . ChIP-seq data for Fs ( 1 ) h-L and Fs ( 1 ) h-LS were obtained from GEO accession GSE42086 [25] . Peak calling was done using MACS 2 . 1 . 0 ( https://github . com/taoliu/MACS ) . For human analysis , we used publicly available binding peaks of 73 chromatin proteins ( RAD21 , CTCF , YY1 , ZBTB33 , MAZ , JUND , ZNF143 , EZH2 , ATF2 , ATF3 , BATF , BCL11A , BCL3 , BCLAF1 , BHLHE40 , BRCA1 , CEBPB , CFOS , CHD1 , CHD2 , CMYC , COREST , E2F4 , EBF1 , EGR1 , ELF1 , ELK1 , FOXM1 , GABP , IKZF1 , IRF4 , MAX , MEF2C , MTA3 , MXI1 , NFATC1 , NFE2 , NFIC , NFKB , NFYA , NFYB , NRF1 , NRSF , P300 , PAX5 , PBX3 , PML , POL2 , POL3 , POU2F2 , RFX5 , RUNX3 , RXRA , SIN3A , SIX5 , SMC3 , SP1 , SPI1 , SRF , STAT1 , STAT3 , STAT5 , TBLR1 , TBP , TCF12 , TCF3 , TR4 , USF1 , USF2 , WHIP , ZEB1 , ZNF274 , ZZZ3 ) and histone marks ( H3K27me3 , H3K36me3 , H3K4me2 , H3K4me3 , H3K9ac , H3K9me3 , H4K20me1 ) of GM12878 cells from ENCODE [61] . We downloaded peaks that were uniformly processed ( Uniform Peaks ) . For human analysis , we divided promoters into quartiles of gene expression using RNA-seq data [61] . We also divided promoters into quartiles of gene pausing and into quartiles of gene elongation using PolII ChIP-seq data [61] . For enhancer mapping , we used lymphocyte of B lineage differentially expressed enhancers identified from the Fantom5 project [62] . For both Drosophila and human analyses , we used transcription factor binding site ( TFBS ) motifs from the MotifMap database ( http://motifmap . ics . uci . edu/ ) . The proposed GLMI assumed a linear relation between logarithm of Hi-C counts and the logarithm of distance between bins as previously shown in [5] . This assumption only holds locally , i . e . for a specific distance scale . Hence we restricted GLM modeling to a certain range of distances , e . g . for 20kb to 1Mb . In addition , we tested this assumption on data before using GLMI . We considered that this assumption holds when the R2 > 0 . 95 . Before computing variables for the GLMI presented above , intermediate variables from the genomic features such as DNA-binding proteins needed to be calculated . Intermediate “occupancy” variable zi denoted the presence ( zi = 1 ) or absence ( zi = 0 ) of the protein Fi within the genomic bin . If the protein only overlapped 60% of the genomic bin , then zi = 0 . 6 . Here are described the different models derived from model ( 1 ) that we used . In order to assess a homologous interaction variable nii = ziL × ziR ( here g = nii ) , model ( 1 ) becomes: log E y | X = β 0 + β d d + β B B + β C C + β g g = β 0 + β d d + β B B + β m i m i + β n i i n i i ( 2 ) Following the hierarchy principle in ( generalized ) linear models , the assessment of a statistical interaction variable , such as nii = ziL × ziR , must include both ziL and ziR as confounding variables . Because ziL and ziR are identically associated to y ( the attribution for left and right bins is arbitrary ) , their values are averaged to give m i = 1 2 ( z i L + z i R ) . Hence C = mi is used as a confounder of nii . In order to assess a heterologous interaction variable n i j = 1 2 ( z i L × z j R + z j L × z i R ) ( here g = nij ) , model ( 1 ) becomes: log E y | X = β 0 + β d d + β B B + β C C + β g g = β 0 + β d d + β B B + β m i m i + β m j m j + β n i j n i j ( 3 ) Following the hierarchy principle , ziL , ziR , zjL and zjR have to be included as confounding variables . As previously , ziL and ziR are averaged to give m i = 1 2 ( z i L + z i R ) . Similarly , zjL and zjR are averaged to give m j = 1 2 ( z j L + z j R ) . Hence C = {mi , mj} is used as confounder of nij . In order to assess a homologous interaction cofactor variable ciik = nii × nkk ( here g = ciik ) , model ( 1 ) becomes: log ( E [ y | X ] ) = β 0 + β d d + β B B + β C C + β g g = β 0 + β d d + β B B + β m i m i + β m k m k + β m i k m i k + β n i i n i i + β n k k n k k + β n i k n i k + β n i i × m k ( n i i × m k ) + β n k k × m i ( n k k × m i ) + β c i i k c i i k , ( 4 ) Here variable ciik is a four-way interaction term and hence there are a large number of confounding variables included in variable set C = {mi , mk , mik , nii , nkk , nik , nii × mk , nkk × mi} . We need to introduce a new type of variable , noted mij , the average of product ziL × zjL and product ziR × zjR ( m i j = 1 2 ( z i L × z j L + z i R × z j R ) ) . For a detailed explanation of the confounder set C , see S1 Appendix , Confounder sets . In order to assess a heterologous interaction cofactor variable cijk = nij × nkk ( here g = cijk ) , model ( 1 ) becomes: log ( E [ y | X ] ) = β 0 + β d d + β B B + β C C + β g g = β 0 + β d d + β B B + β m i m i + β m j m j + β m k m k + β m i k m i k + β m j k m j k + β n i j n i j + β n j k n j k + β n i k n i k + β n k k n k k + β n i j × m k n i j × m k + β n k k × m i n k k × m i + β n k k × m j n k k × m j + β c i j k c i j k . ( 5 ) Here variable cijk is a four-way interaction term and hence there are a large number of confounding variables included in variable set C = {mi , mj , mk , mik , mjk , nij , njk , nik , nkk , nij × mk , nkk × mi , nkk × mj} . For a detailed explanation of the confounder set C , see S1 Appendix , Confounder sets . In addition , we formulated models for homologous interaction variables , depending on motif pair orientation . For a pair of motifs in convergent orientation ( →← ) , model ( 1 ) becomes: log ( E [ y | X ] ) = β 0 + β d d + β B B + β C C + β g g = β 0 + β d d + β B B + β z i L + z i L + + β z i R − z i R − + β n i i + − n i i + − ( 6 ) with nii+− = ziL+ × ziR− . Symbol “+” denoted motifs that were on the forward DNA strand , while symbol “-” denoted motifs that were on the reverse DNA strand . For instance , variable ziL+ was the occupancy of a motif on the forward DNA strand within genomic bins . For a pair of motifs in divergent orientation ( ←→ ) , model ( 1 ) becomes: log ( E [ y | X ] ) = β 0 + β d d + β B B + β C C + β g g = β 0 + β d d + β B B + β z i L − z i L − + β z i R + z i R + + β n i i − + n i i − + , ( 7 ) with nii−+ = ziL− × ziR+ . For a pair of motifs in same orientation ( →→ ) , model ( 1 ) becomes: log ( E [ y | X ] ) = β 0 + β d d + β B B + β C C + β g g = β 0 + β d d + β B B + β z i L + z i L + + β z i R + z i R + + β n i i + + n i i + + , ( 8 ) with nii++ = ziL+ × ziR+ . For a pair of motifs in same orientation ( ←← ) , model ( 1 ) becomes: log ( E [ y | X ] ) = β 0 + β d d + β B B + β C C + β g g = β 0 + β d d + β B B + β z i L − z i L − + β z i R − z i R − + β n i i − − n i i − − , ( 9 ) with nii−− = ziL− × ziR− . Moreover , we formulated an additional “full” model where all possible homologous and heterologous interaction variables were included . For instance , if we study two proteins Fi and Fj that tend to linearly colocalize , then the following “full” model would be: log ( E [ y | X ] ) = β 0 + β d d + β B B + β C C + β G G , = β 0 + β d d + β B B + β m i m i + β m j m j + β n i i n i i + β n j j n j j + β n i j n i j , ( 10 ) where G is the set of all possible homologous and heterologous interaction variables . Here G = {nii , njj , nij} for two proteins Fi and Fj . The confounder set C = {mi , mj} includes all marginal variables . The general linear regression with interactions is implemented in R language . The model is available in the R package “HiCglmi” which can be downloaded from the Comprehensive R Archive Network . | Chromosomal DNA is tightly packed in three dimensions ( 3D ) such that a 2-meter long human genome can fit into a microscopic nucleus . Recent studies have revealed that such packing of DNA is not random but instead structured into functional DNA loops . Those loops are essential to numerous key processes in the cell , such as genome expression and DNA replication . In addition , disruption of DNA loops can lead to genetic diseases and cancers . Understanding how DNA loops are formed and what are their molecular determinants is thus a fundamental issue . In this work , we propose a computational model to identify the molecular determinants of loops , including protein and DNA sequence . Most notably , the model offers insights in the different mechanistic scenarios behind loop formation . Using this model , we uncover numerous novel DNA loops and underlying mechanisms in Drosophila and human . We find that the orientation-dependent specificity between CTCF motifs is conserved in metazoans . We show how loops between DNA-binding proteins can be mediated by additional cofactors . Our analyses further reveal opposite influences of transcription factors depending on RNA Polymerase II pausing . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods"
] | [
"invertebrates",
"protein",
"interactions",
"dna-binding",
"proteins",
"enzymology",
"animals",
"invertebrate",
"genomics",
"animal",
"models",
"insulators",
"drosophila",
"melanogaster",
"model",
"organisms",
"materials",
"science",
"experimental",
"organism",
"systems",
"... | 2017 | Uncovering direct and indirect molecular determinants of chromatin loops using a computational integrative approach |
The Kaposi's sarcoma-associated herpesvirus ( KSHV ) genome encodes a G protein-coupled receptor ( vGPCR ) . vGPCR is a ligand-independent , constitutively active signaling molecule that promotes cell growth and proliferation; however , it is not clear how vGPCR is negatively regulated . We report here that the KSHV K7 small membrane protein interacts with vGPCR and induces its degradation , thereby dampening vGPCR signaling . K7 interaction with vGPCR is readily detected in transiently transfected human cells . Mutational analyses reveal that the K7 transmembrane domain is necessary and sufficient for this interaction . Biochemical and confocal microscopy studies indicate that K7 retains vGPCR in the endoplasmic reticulum ( ER ) and induces vGPCR proteasomeal degradation . Indeed , the knockdown of K7 by shRNA-mediated silencing increases vGPCR protein expression in BCBL-1 cells that are induced for KSHV lytic replication . Interestingly , K7 expression significantly reduces vGPCR tumorigenicity in nude mice . These findings define a viral factor that negatively regulates vGPCR protein expression and reveal a post-translational event that modulates GPCR-dependent transformation and tumorigenicity .
Kaposi's sarcoma-associated herpesvirus ( KSHV , also known as human herpesvirus 8 ) is believed to be the etiologic agent for Kaposi's sarcoma ( KS ) [1] . KSHV infection is also linked to primary effusion lymphoma [2] and multicentric Castleman's disease , rare lymphoproliferative malignancies of B-cell origin 3 , 4 . The KSHV genome encodes over 80 viral polypeptides , many of which are capable of promoting cell proliferation and/or modulating host responses , when expressed in gene transfer experiments ( for review see reference [5] ) . One such gene product consistently detected in KS lesions is the viral G protein-coupled receptor ( vGPCR , or open reading frame 74 ) [6] , [7] . vGPCR is a homolog of the human interleukin-8 receptor and possesses promiscuous chemokine-binding activity [8] . In tissue culture , vGPCR expression activates various signaling pathways and up-regulates the transcription of numerous cellular and viral genes that encode cytokines , signaling molecules , and transcription factors that culminate in promoting cell proliferation and endothelial tube formation [9]–[15] . Additionally , vGPCR transgenic mice developed tumors that resemble human KS lesions [9] , [16] , [17] . Although ligand binding is not required for vGPCR-mediated signaling , cognate chemokines appear to modulate vGPCR activity in tissue culture and in mice as well [8] , [18] . Despite the fact that proliferative and prosurvival activities of vGPCR have attracted extensive attention in the past , accumulating evidence suggests that tightly regulated expression and signaling are important for vGPCR function in KSHV infection . Indeed , over-expression of vGPCR induced cell death in COS-1 cells and constitutive expression of vGPCR was toxic to PEL cells [6] , [19] . Furthermore , vGPCR is predominantly translated from a bicistronic mRNA transcript downstream of K14 ( vOX2 ) , presumably reducing vGPCR protein expression [6] , [20] . These observations suggest that KSHV likely has evolved mechanisms to achieve a temporary expression of the constitutively active vGPCR during lytic infection . A post-translational degradation is one of these mechanisms . Regulated protein degradation is important for a variety of cellular events including cell cycle , apoptosis , signal transduction , immune response , and development . Cellular GPCR can be degraded either by the ubiquitin-proteasome system ( UPS ) or by the lysosome . Within the lysomsome , proteins are cleaved by diverse acidic proteases upon fusion with endosomes or autophagosomes . For UPS substrates , proteins destined for destruction are tagged with ubiquitin through sequential actions of the E1 activating enzyme , E2 conjugating enzyme , and E3 ligase [21] . Relying on the UPS , the endoplasmic reticulum ( ER ) -associated degradation ( ERAD ) pathway is a major route to remove mis-folded proteins post-translationally , and plays an essential role for ER quality control . Indeed , alteration of ERAD pathways has been implicated in diverse clinical presentations such as neurodegeneration and cystic fibrosis . Furthermore , viruses usurp components of this pathway to evade host recognition and possibly modulate other host responses [22]–[24] . We previously identified a small membrane protein , K7 , which induces protein degradation of IκB and p53 . K7 specifically interacts with the ubiquitin-associated domain of cellular protein linking integrin-associated protein and cytoskeleton ( PLIC1 ) and antagonizes PLIC1 , thereby promoting protein degradation [25] . Additionally , K7 was shown to deregulate cellular apoptosis by targeting Bcl-2 and an ER resident calcium modulating cyclophilin ligand [26] , [27] . Although these data imply that K7 inhibits apoptosis to facilitate viral replication , its biological roles in KSHV infection remain obscure . We report here that K7 interacts with vGPCR and induces its proteasomeal degradation . The knockdown of K7 by shRNA-mediated silencing increased vGPCR protein expression in BCBL-1 cells that are induced for KSHV lytic replication . Biochemical and confocal microscopy analyses support that K7 retains vGPCR in the ER , thereby facilitating the proteasome to degrade vGPCR . Consequently , K7 significantly reduces vGPCR transformation in vitro and tumorigenicity in nude mice . These data establish a negative regulation of vGPCR protein expression and tumorigenicity by KSHV K7 .
To understand K7's functions , we searched for cellular interacting proteins with K7 as bait using the yeast two-hybrid screen . One clone contained a partial sequence of a putative G protein-coupled receptor that encodes its last four transmembrane ( TM ) domains . Since the KSHV genome encodes a vGPCR , we speculated that K7 interacts with vGPCR . To test this possibility , whole cell lysates of 293T cells transiently transfected with plasmids expressing vGPCR-Flag and/or K7-V5 were precipitated with the M2 anti-Flag antibody and precipitates were analyzed by immunoblot with anti-V5 ( K7 ) antibody . Indeed , K7 was readily detected in immune complexes containing vGPCR ( Figure 1A , left panels ) . Notably , vGPCR expression greatly increases K7 protein expression and the glycosylated form ( the slower migration band ) is only detected in the presence of vGPCR . Reciprocally , vGPCR was also precipitated by anti-V5 ( K7 ) antibody ( Figure 1A , right panels ) . Of note , the interaction between K7 and vGPCR was also identified by the yeast two-hybrid screen with a high throughput approach [28] . To further characterize the vGPCR-K7 interaction , K7 mutants that contain various deletions as described in our previous publications [25] , [26] were used for a co-immunoprecipitation ( co-IP ) assay . The internal hydrophobic region ( amino acid 22–74 ) containing the putative TM domain was sufficient to interact with vGPCR ( Figure 1B ) . Unfortunately , K7 mutants lacking the TM domain were expressed at an undetectable level compared to the wild type ( wt ) K7 . Thus , we failed to obtain any deletion mutant that no long interacts with vGPCR . Nevertheless , these data indicate that K7 interacts with vGPCR and suggest that its predicted TM domain is important for this interaction . K7 contains a putative TM domain and vGPCR is a seven-membrane-spanning protein , therefore we examine whether the predicted K7 TM domain is necessary for this interaction . The K7 mutant whose putative TM region was replaced by a heterologous TM from the Saimiri transforming protein C ( Stp C ) , designated K7TMStp C , was constructed and expressed in 293T cells . We found that K7TMStp C failed to interact with vGPCR under the same co-IP conditions ( Figure 1C ) . Of note , K7TMStp C was expressed and localized to intracellular organelles similarly to the wt K7 ( unpublished data ) . Furthermore , appending the putative TM region ( amino acids 23–45 ) of K7 to GFP renders it capable of binding vGPCR ( Figure 1D ) . Thus , these data collectively support that K7 interacts with vGPCR and that the putative K7 TM region is necessary and sufficient for this interaction . K7 and vGPCR proteins are confined to distinct intracellular organelles . Particularly , vGPCR was reported to reside primarily in the trans-Golgi network ( TGN ) [7] , whereas K7 localizes to both the ER and mitochondrial compartments [26] , [27] . To examine the intracellular distribution of vGPCR and K7 , indirect immuno-fluorescence microscopy was performed . To this end , human lymphoid BJAB and HeLa cells were transfected with plasmids expressing vGPCR-Flag and K7-V5 , and analyzed by confocal microscopy . In both HeLa and BJAB cells , vGPCR predominantly localizes to a subcellular structure reminiscent of the TGN , while K7 distributes throughout the cytoplasm mainly as punctate vesicles ( Figure 2A and 2B ) . In support of the interaction between K7 and vGPCR , K7 had an intracellular staining pattern similar to that of vGPCR in both HeLa and BJAB cells ( Figure 2C ) . Despite the overall colocalization between K7 and vGPCR , there are some regions that either K7 or vGPCR is predominant , likely reflecting their distinct intracellular compartments that vGPCR and K7 reside in when they are separately expressed ( Figure 2C , insets of BJAB cells ) . The fact that K7 interacts with vGPCR prompted us to investigate the temporal expression kinetics of K7 and vGPCR in KSHV lytic replication . Both K7 and vGPCR were reported to be expressed early during KSHV lytic reactivation and/or de novo infection [6] , [7] , [27] , [29]; however , the relative temporal expression of vGPCR and K7 remains unclear . Our interaction study suggests that vGPCR and K7 are possibly expressed at the same time . Thus , we examined mRNA levels of vGPCR and K7 by reverse-transcriptase ( RT ) -polymerase chain reaction ( PCR ) . The KSHV latently infected PEL cell lines BCBL-1 ( KSHV only ) and JSC-1 ( KSHV and EBV co-infected ) were treated with TPA to induce KSHV lytic replication . Alternatively , lytic replication was reactivated by Rta expression that was induced by doxycycline using the BCBL-1/T-Rex_Rta cell line [30] . RT-PCR analyses were performed using primers specific for vGPCR , K7 , the polyadenylated nuclear RNA ( PAN ) , and cellular β-actin . When treated with TPA ( 20 ng/ml ) , lytic replication was initiated in both BCBL-1 and JSC-1 cells which was indicated by potent induction of PAN transcripts ( Figure 3 ) . The residual PAN RNA in untreated BCBL-1 cells and BCBL-1/T-Rex_Rta cells ( lane 3 of left two sets in Figure 3 ) are likely due to spontaneous lytic replication of KSHV or leaky Rta expression in these PEL cells , respectively . Upon TPA induction , vGPCR transcripts peaked at 72 h , which coincided with the highest mRNA level of K7 in BCBL-1 cells . Upon Rta expression induced by doxycycline addition , vGPCR was highly expressed as early as 12 h post induction and was sustained for more than 24 h ( Figure 3 , middle panels ) , while K7 transcripts gradually increased and peaked at 36 h after TPA induction when vGPCR mRNA started to decline . This indicates that K7 expression predominantly overlaps with that of vGPCR in response to the KSHV lytic switch protein , Rta . Similar results were obtained for TPA-induced JSC-1 cells in which vGPCR was highly expressed at 12 and 24 h after treatment . Meanwhile , K7 was highly expressed at 24 h after induction ( Figure 3 , right panels ) . The most abundant lytic transcript of KSHV , PAN , was significantly induced by TPA and sustained high transcript levels in BCBL-1 and JSC-1 cells throughout the entire induction period . This was more pronounced by Rta induction ( Figure 3 third panel from top ) , while cellular β-actin transcript remained the same . Overall , these data indicated that K7 and vGPCR are expressed at the same time and suggest that the interaction between these two molecules is biologically relevant . We have consistently observed that K7 co-expression significantly reduces the protein level of vGPCR ( Figure 1A and 1B ) , suggesting that K7 modulates vGPCR biosynthesis . Because our previous data implicate K7 in regulating protein degradation [25] , we speculated that K7 induces the degradation of vGPCR . To examine K7's effect on vGPCR protein expression , human endothelial ECV cells were transiently transfected with a plasmid expressing vGPCR-Flag and increasing amounts of a plasmid expressing K7-V5 . Whole cell lysates were analyzed by immunoblot for vGPCR protein expression . The result shows that K7 reduces vGPCR protein in a dose-dependent manner ( Figure 4A ) . The specificity of K7 is further supported by the observation that the K7TMStp C chimera , a mutation that abolished its interaction with vGPCR , failed to suppress vGPCR protein expression ( Figure 4B ) . Previous publications have convincingly shown that vGPCR activates a number of signaling pathways , leading to the activation of NF-AT , NF-κB , and AP-1 transcription factors [19] , [31] , [32] . To further correlate K7's effect on vGPCR protein expression , the transcription activation of NF-AT , NF-κB , and AP-1 response elements by vGPCR were measured by luciferase assays in transiently transfected 293T cells . Consistent with published data , vGPCR activated NF-κB , NF-AT , and AP-1 transcription factors by approximately 4 , 25 , and 4 . 5 fold , respectively . In contrast , K7 exhibited no effect on the transcription of NF-κB , NF-AT , and AP-1 ( Figure 4C ) . In agreement with our observation that K7 reduces vGPCR protein , K7 suppressed the transcription activation by vGPCR to approximately two-fold for NF-κB and AP-1 , and eight-fold for NF-AT , respectively ( Figure 4C ) . These data indicate that K7 reduces vGPCR protein expression and mitigates vGPCR-activated downstream signaling . Although our studies clearly indicate that K7 reduces vGPCR protein expression , these experiments relied on exogenous protein expression . To corroborate K7-reduced vGPCR protein expression during KSHV infection , the shRNA-mediated silencing experiments were designed to knock down K7 expression and vGPCR protein level was examined by confocal microscopy . Both K7 and vGPCR are expressed in the lytic phase during KSHV infection . Given the fact that K7 open reading frame overlaps with the transcribed region of PAN ( or T1 . 1 ) , four pairs of short hairpin RNA ( shRNA ) molecules targeting the 5′ untranslated region of K7 transcripts were cloned ( Figure 4D ) and lentiviral particles were produced in 293T cells . Lentivirus was then used to infect KSHV-positive BCBL-1 cells that were subsequently treated with TPA to induce KSHV lytic replication . A scrambled shRNA was used as a control for all silencing experiments . Among the shRNAs , K7 shRNA#1 and #3 significantly reduced the level of K7 transcripts , while these two shRNA molecules had no discernable alteration on mRNA levels of PAN and vGPCR , when compared to BCBL-1 cells expressing the scrambled shRNA ( Figure 4D , right panels ) . Densitometry of RT-PCR products showed that K7 shRNA#3 and shRNA#1 had a silencing efficiency of 60% and 50% ( Figure 4D ) . Semi-quantitative PCR analyses using serial dilution of cDNA templates further support that K7 transcripts were reduced by 60%–70% ( Figure S1 ) . Notably , the knockdown of K7 did not significantly affect cell viability after lytic induction , suggesting that additional viral proteins such as vBcl-2 and vFLIP play a redundant antiapoptotic role . BCBL-1 cells infected with lentiviruses expressing K7 shRNA#1 , shRNA#3 , or the scrambled shRNA were induced with TPA for KSHV lytic replication . At 48 h after induction , cells were fixed and subjected to confocal microscopy analysis to examine vGPCR protein level . As shown in Figure 5 , the knockdown of K7 significantly increased vGPCR protein expression ( second and third rows from the top ) , while the ER resident protein calreticulin was not affected . The vGPCR-positive cells increased from 20% in BCBL-1 cells expressing the scrambled shRNA to 65% in BCBL-1 cells expressing K7 shRNA#3 and 45% in BCBL-1 cells expressing K7 shRNA#1 ( Figure 5 , middle panels ) . Furthermore , merged images clearly indicate the increased vGPCR protein expression upon K7 knockdown , because image color shifted from red ( calreticulin ) in BCBL-1 cells expressing the scrambled shRNA to green ( vGPCR ) in BCBL-1 cells expressing K7 shRNA ( Figure 5 , right panels ) . Taken together , these findings support the conclusion that K7 suppresses vGPCR protein expression in tissue culture and in KSHV lytic infection . Our previous publication indicated that K7 induces protein degradation dependent on the UPS [25] . To investigate the mechanism by which K7 downregulates vGPCR protein expression , the half-life of vGPCR was measured by a pulse chase experiment . Transient transfection of ECV cells expressing vGPCR or vGPCR and K7 were pulse labeled with [35S]-methionine/cysteine ( Met/Cys ) . After extensive washing , ECV cells were chased with cold medium . Precipitated vGPCR was quantified by autoradiography and its half-life was calculated . As shown in Figure 6A , vGPCR has a half-life of about 6 . 5 h and K7 expression reduced its half-life to approximately 3 . 4 h , indicating that K7 promotes vGPCR degradation . Cellular GPCRs are 7-membrane-spanning proteins that can be degraded through the lysosome or the UPS [33] . To examine whether K7-induced vGPCR degradation is dependent on the proteasome or the lysosome , vGPCR protein stability was examined by a pulse chase experiment with either a lysosome inhibitor ( chloroquine ) or proteosome inhibitors ( lactacystin and MG132 ) . It was found that lactacystin and MG132 , but not chloroquine , completely blocked K7-induced vGPCR degradation , indicating that this process relies on the proteolytic activity of the proteasome ( Figure 6B ) . Proteasome substrates are often marked with polyubiquitin chains that facilitate delivery to and subsequent degradation by the proteasome . To further corroborate the proteasome-dependence of K7-induced vGPCR degradation , vGPCR ubiquitination was examined by immunoprecipitation and immunoblot . vGPCR was precipitated with anti-Flag sepharose and analyzed by immunoblot with anti-HA ( ubiquitin ) antibody . Consistent with the increased degradation of vGPCR , K7 promoted vGPCR polyubiquitination in the presence of a proteasome inhibitor , lactacystin ( Figure 6C , first panel from left ) . Recent findings have shown that K48-linkage ubiquitin chains mediate protein degradation and K63-linkage ubiquitin chains are involved in signal transduction . Thus , these ubiquitin mutants were included in the vGPCR ubiquitination assay . Indeed , the K48R mutant , but not the K63R mutant , completely abolished vGPCR ubiquitination induced by K7 ( Figure 6C ) . Of note , the protein level of precipitated vGPCR and vGPCR in whole cell lysate in the presence of K7 is significantly lower than vGPCR alone ( Figure 6C , second panel , lanes 2–5 , and Figure S2 ) . These data collectively support the conclusion that K7 increases vGPCR ubiquitination and promotes its proteasomeal degradation . To further define the molecular action of K7 in inducing vGPCR degradation , vGPCR intracellular localization was analyzed by confocal microscopy using human HeLa cells . Consistent with a previous report [7] , vGPCR primarily localized to the TGN stained by anti-TGN46 antibody ( Figure 7A ) . Upon K7 expression , vGPCR localized to intracellular structures that resemble the ER and nuclear membrane ( Figure 7B ) , suggesting that K7 retains vGPCR in the ER compartment . Indeed , HeLa cells expressing both K7 and vGPCR revealed that these two proteins colocalized significantly with protein disulfide isomerase ( PDI ) , an ER resident protein ( Figure 7C ) , supporting the notion that K7 retains vGPCR in the ER . Furthermore , K7 expression reduced vGPCR localization in the TGN when intracellular distribution of vGPCR and K7 was examined in relation to TGN46 ( Figure 7D ) . These results clearly indicate that K7 retains vGPCR in the ER and suggest that K7 induces vGPCR degradation via the ER-associated degradation pathway . To examine K7's effect on vGPCR biological functions , NIH3T3 cell lines stably expressing K7 , vGPCR , and vGPCR+K7 were established with lentivirus infection . As shown in Figure 8A , K7 detectably reduced vGPCR protein expression without affecting its mRNA levels ( Figure 8B ) . Of note , vGPCR did not further increase K7 protein after treatment by the proteasome inhibitor MG132 ( Figure 8A ) . During the course to establish these stable cell lines , we noticed that NIH3T3 cells expressing vGPCR grow more slowly than the control NIH3T3 cells . In contrast to what was reported [34] , NIH3T3/vGPCR cells had a doubling time of approximately 31 h that is significantly longer than 22 . 6 h of NIH3T3/vector cells . RT-PCR analysis indicated that vGPCR is expressed at similar levels in NIH3T3 , and reactivated BCBL-1 and JSC-1 cells ( Figure S3 ) . This observation rules out the possibility that the inhibitory effect on cell growth is due to over-expression . Interestingly , K7 expression also increased NIH3T3 doubling time to roughly 28 . 2 h . Consistent with K7-reduced vGPCR protein expression , K7 co-expression slightly decreases the doubling time of NIH3T3 cells to 30 h ( Figure 8C ) . Due to K7's inhibitory effect on cell growth and vGPCR-increased K7 expression ( unpublished data ) , the subtle difference in cell growth may be significant . Given the inhibitory effect of vGPCR on cell growth , we suspect that NIH3T3 cells expressing higher vGPCR will gradually decrease when continuously cultured without selection . To test this , NIH3T3/vGPCR and NIH3T3/vGPCR+K7 cells were passaged for a week and RT-PCR analyses were performed to assess the mRNA levels of vGPCR . Indeed , the vGPCR mRNA level significantly decreased after 1 wk of passage and K7 reduced the vGPCR loss ( Figure 8D ) . Semi-quantitative RT-PCR and real-time PCR analyses revealed that the vGPCR mRNA in NIH3T3/vGPCR+K7 was approximately 5-fold of that in NIH3T3/vGPCR cells at day 7 ( Figure 8E and S4 ) . The rapid loss of vGPCR transcripts suggests that NIH3T3 cells that lost vGPCR have a growth advantage . We and others have shown that K7 inhibits apoptosis induced by various stress stimulations [25]–[27] . To examine whether vGPCR affects K7's antiapoptotic function , NIH3T3 stable cells were stimulated with TNF-α and cyclohexamide and cell viability was measured by trypan blue staining as described previously [25] . It was found that vGPCR expression had no significant effect on cell survival upon TNF-α stimulation , while K7 expression increased cell survival rate by 20% compared to NIH3T3/vector cells ( Figure 8F ) . Interestingly , vGPCR co-expression with K7 further promotes cell survival rate by approximately 30% , indicating that vGPCR potentiates K7's antiapoptotic effect . This is consistent with our observation that vGPCR increases K7 protein expression ( unpublished data ) . These results indicate that K7 reduces vGPCR-induced stress and suggest that K7 likely co-operates with vGPCR to promote cell survival during KSHV lytic replication . In a mouse pathogenesis model , vGPCR is sufficient to induce tumor formation in nude mice and vGPCR transgenic mice developed lesions that resemble human KS , suggesting its potential contribution to KSHV-associated malignancies [17] , [18] , [34] . To assess K7's effect on vGPCR tumorigenicity , NIH3T3 stable cells expressing K7 , vGPCR , or vGPCR+K7 were mixed with NIH3T3 cells and colony formation on soft agar was examined . Similar to the human cytomegalovirus US28 [35] , vGPCR-expressing cells stimulated anchorage-independent growth of NIH3T3 cells , whereas neither NIH3T3/vector , nor NIH3T3/K7 cells supported colony formation ( Figure 9A ) . In support of the observation that K7 suppressed vGPCR protein expression , NIH3T3/vGPCR+K7 cells formed smaller colonies than NIH3T3/vGPCR cells ( Figure 9A , left panels ) . Furthermore , K7 expression also reduced the number of colonies from 258 of NIH3T3/vGPCR to 131 of NIH3T3/vGPCR+K7 ( Figure 9A , right diagram ) . To further investigate K7's effect on vGPCR tumorigenicity in vivo , these stably transfected cells were injected into nude mice and tumor growth was assessed . Mice injected with NIH3T3/vGPCR developed visible tumors within two weeks and all mice harbored tumors after 6 wk . Neither NIH3T3/vector cells nor NIH3T3/K7 cells induced apparent tumor in nude mice . In agreement with results from the soft agar assay , K7 significantly reduced vGPCR capacity to promote tumor growth in nude mice as shown by the number of mice harboring tumor and tumor weight ( Figure 9B ) . All four nude mice injected with NIH3T3/vGPCR developed tumors after 6 wk , whereas only two mice injected with NIH3T3/vGPCR+K7 developed tumors , which were substantially smaller ( Figure 9B ) . The mean weight of tumors derived from NIH3T3/vGPCR cells is approximately 8-fold higher than that of tumors derived from NIH3T3/vGPCR+K7 cells ( Figure 9B and unpublished data ) . Interestingly , we found that K7 transcripts were expressed at a higher level in the smaller tumor than the bigger tumor , suggesting that K7 inhibits the vGPCR-dependent tumor growth in vivo ( Figure 9C ) . This result is consistent with the observation that K7 expression reduces vGPCR tumorigenicity ( Figure 9B ) . In contrast , the vGPCR transcript was expressed more abundantly in tumors derived from NIH3T3/vGPCR+K7 cells than those derived from NIH3T3/vGPCR cells ( Figure 9C ) . This likely represents the relative expression of vGPCR in stable NIH3T3 cells before mice injection . Overall , K7 negatively regulates vGPCR tumorigenicity in vitro by a soft agar assay and in vivo in nude mice .
We report here that KSHV K7 interacts specifically with vGPCR and induces the rapid degradation of vGPCR , thereby reducing vGPCR protein expression . The putative K7 TM domain is necessary and sufficient for its interaction with vGPCR , indicating a specific interaction between vGPCR and K7 . However , the K7/vGPCR interaction may involve multiple residues within the putative TM domain of K7 because further mutational analyses within this domain failed to identify critical residues that are essential for this interaction ( unpublished data ) . Alternatively , additional cellular components such as membrane proteins or lipids could be involved , as our co-IP procedure does not exclude this possibility . Nevertheless , these data support the conclusion that K7 interacts specifically with vGPCR . We have previously shown that K7 antagonizes cellular PLIC1 , a factor that inhibits proteasome-mediated protein degradation , and induces rapid degradation of p53 and IκB [25] . Our current study enlists vGPCR as an additional proteasome substrate whose degradation is accelerated by K7 . The specificity of K7-induced degradation appears to be derived from an interaction with either a proteasome substrate such as vGPCR or a key component of the UPS pathway such as PLIC1 . It is possible that binding of K7 to cellular PLIC1 also contributes to K7-dependent reduced expression of vGPCR , given that PLIC1 has been shown to promote protein expression of multiple transmembrane proteins [36] , [37] . Indeed , we have observed that PLIC1 overexpression increases vGPCR protein , while the knockdown of PLIC1 by shRNA-mediated silencing greatly reduces vGPCR protein expression . These data indicate that PLIC1 is a positive regulator for vGPCR expression ( unpublished data ) . Future experiments will determine whether K7 binding to PLIC1 is sufficient for suppressing vGPCR protein expression . Confocal microscopy analyses and biochemical assays examining vGPCR protein degradation support the conclusion that K7 retains vGPCR in the ER and allows vGPCR to be removed by the proteasome . The rapid degradation of vGPCR induced by K7 also correlates with increased ubiquitination upon treatment with a proteasome inhibitor . vGPCR appears to carry polyubiquitin chains and K7-induced polyubiquitination of vGPCR is specifically inhibited by the K48R ubiquitin mutant , but not by the K63R ubiquitin mutant ( Figure 6C ) . Interestingly , the K63R mutant significantly increased unmodified- as well as ubiquitinated-vGPCR protein . This is likely due to the inhibitory effect of K63R ubiquitin on vGPCR signaling that is presumably coupled to vGPCR degradation . For example , the K63R mutant may inhibit signaling downstream vGPCR such as NF-κB activation , therefore stabilizing vGPCR . Alternatively , vGPCR polyubiquitin chains may contain a mixture of K63- and K48-linkages . The fact that the K48R mutant abolished , while the K63R mutant increased vGPCR ubiquitination suggests that K48-linkage is necessary to initiate ubiquitination , whereas K63-linkage is important for degradation . These intriguing possibilities are not mutually exclusive and require further experimental investigation . Our data , however , do not exclude the possibility that vGPCR undergoes ubquitination-independent proteasomeal degradation . In transfected cells , K7 consistently altered vGPCR intracellular distribution , showing a more diffused ER/nuclear membrane pattern that was confirmed by staining with anti-PDI antibody . This observation suggests that K7 retains vGPCR in the ER in order to induce vGPCR degradation . This also implies that K7 likely engages the ERAD pathway to facilitate vGPCR degradation in similar ways employed by human cytomegalovirus US11 and murine γ-herpesvirus 68 mK3 [23] , [24] , [38] , [39] . Future experiments will be directed to test whether K7-induced protein degradation is dependent on any critical components of the ERAD pathway . Interaction with K7 was found to reduce vGPCR protein , thereby dampening vGPCR-mediated signaling . Both vGPCR and K7 are expressed during KSHV lytic replication and it appears that K7 and vGPCR share an identical or overlapped expression profile . The observation that the K7 transcript peaks at a later time point than the vGPCR transcript raises the possibility that K7 serves as a negative regulatory factor to shut off vGPCR protein during KSHV lytic infection . Indeed , the knockdown of K7 by shRNA-mediated silencing increased vGPCR protein without altering vGPCR transcription level in BCBL-1 cells that are induced for KSHV lytic replication ( Figure 5 ) . Interestingly , K7 protein expression was substantially increased when co-expressed with vGPCR ( unpublished data ) , revealing a negative feedback loop that culminates in dampening vGPCR protein expression . These observations are consistent with the notion that diverse regulatory mechanisms operate to achieve a temporary expression of vGPCR in KSHV infection . In addition to the K7-reduced vGPCR expression , known mechanisms also include the bicistronic translation and the vMIP-mediated regulation [18] , [40] . Interestingly , modulation by its cognate chemokines is important for vGPCR tumorigenicity in transgenic mice [18] . Our findings that K7 interacts with vGPCR and directs it for proteasome-mediated degradation further support the notion that KSHV has evolved intricate mechanisms to regulate vGPCR activity . Additionally , K7 expression provides antiapoptotic activity under various conditions [25]–[27] and vGPCR co-expression potentiates K7's antiapoptotic activity ( Figure 8F ) . This implies that K7 can cooperate with vGPCR in the tumorigenesis of KSHV infection , analogous to the paradigm in which Bcl-2 cooperates with c-myc [41] . However , our transformation assay in vitro and tumor growth in nude mice ruled out this possibility . Together with the biscitronic translation and modulation by vMIP chemokines , vGPCR downregulation by K7 raises an intriguing speculation that KSHV has evolved these mechanisms to monitor vGPCR pathogenicity , permitting a persistent infection within its host . K7 expression suppressed vGPCR transformation on soft agar assay and more pronouncedly reduced vGPCR tumorigenicity in nude mice . Although K7 reduced vGPCR protein expression by approximately two-fold ( Figures 6A and 8A ) , it was found that K7 inhibited vGPCR tumorigenicity by more than 8-fold ( Figure 9B ) . This suggests that additional mechanisms , other than reduced protein expression , may contribute to K7's effect on vGPCR tumorigenicity . One likely mechanism is a K7-dependent retention of vGPCR in the ER , given that vGPCR predominantly localizes to the TGN and cell surface under normal circumstances . Conceivably , vGPCR functions in the TGN and on the cell surface are abolished by K7 expression . Interestingly , we have found that vGPCR is tyrosine sulfated in the TGN and tyrosine sulfation is important for vGPCR tumorigenicity ( unpublished data ) . In addition to tyrosine sulfation , post-translational modifications in the ER ( such as ubiquitination and glycosylation ) altered by K7 may cause impaired vGPCR signaling and tumorigenicity . These mechanisms are not mutually exclusive and warrant further investigations of post-translational events underlying vGPCR tumorigenicity . Mounting evidence points to vGPCR expression inducing a stress in mammalian cells including KSHV infected PEL cells [6] , [13] . Indeed , our vGPCR-expressing NIH3T3 cells have a longer doubling time than control NIH3T3/vector cells ( Figure 8C ) . Furthermore , NIH3T3/vGPCR cells gradually lost vGPCR expression when continuously passaged in vitro , suggesting that NIH3T3 cells gain a growth advantage by reducing vGPCR expression . Indeed , K7 alleviated vGPCR-mediated inhibition of NIH3T3 growth and the rate of vGPCR transcript loss ( Figure 8C–8E ) . In contrast , vGPCR expression was necessary for tumorigenicity in nude mice , and K7-reduced vGPCR expression correlated with less transformation in vitro and tumorigenicity in vivo ( Figure 9A and 9B ) . Interestingly , the endothelial progenitor cell line containing Bac36 ( a KSHV Bacmid ) behaves similarly to NIH3T3/vGPCR cells , demonstrating reduced cell growth in vitro and increased tumor formation in vivo [42] . The seemingly paradox between in vitro stress and in vivo tumorigenicity may be explained by a paracrine mechanism supported by accumulating studies [9] , [43] , [44] . In fact , vGPCR-induced tumor formation is highly dependent on growth factors and chemokines that stimulate the angio-proliferation of neighboring cells [43] , [44] . In KS lesions , vGPCR expressing cells presumably stimulate the proliferation of spindle cells that are latently infected by KSHV . The fact that slower growth of NIH3T3 stable cell lines in vitro correlates with higher tumorigenicity in vivo suggests that the nude mice model primarily assesses the paracrine function of vGPCR . This is also supported by our in vitro transformation assay where the proliferation of regular NIH3T3 cells was examined in the presence of NIH3T3/vGPCR cells ( Figure 9A ) . Additionally , it is not unprecedented that oncogenic proteins exploit cellular stress responses to induce tumor formation . Perhaps , these stress responses represent various barriers that oncogenesis has to overcome . For example , H-RAS triggers the ER-associated unfolded protein response , cellular senescence and sensitizes cells to apoptosis [45] , [46] . Similarly , the myc-mediated stress is overcome by Bcl-2 expression [41] . Taken together , the fact that the stress in tissue culture accompanies the tumorigenicity in vivo for many oncogenic proteins suggests that the stress response may serve as an indicator for tumorigenicity in vivo . Similar to vGPCR , K7 also reduces NIH3T3 growth and it will be interesting to examine K7's tumorigenicity in nude mice . All members of the beta- and gamma-herpesvirus family encode up to four GPCRs in their genomes . Some of them have been shown to constitutively activate signaling events downstream of various G proteins ( for review see [47] ) . Although it was demonstrated that KSHV vGPCR can be uncoupled from downstream signal activation by overexpressed G protein-coupled receptor kinase 5 and arrestins [48] , it is largely unknown how these unconventional viral GPCRs are differentially regulated as opposed to cellular GPCRs under normal physiological conditions . This study established an example of post-translational regulation of vGPCR pathogenicity by which a viral factor-induced degradation greatly influences its tumorigenicity . Similar regulatory mechanisms may exist for other viral GPCRs of herpesviruses . Therefore , viral factors that modulate these viral GPCRs likely have a profound effect on various biological activities during herpesvirus infection .
Unless specified , all constructs were derived from pcDNA5/FRT/TO ( Invitrogen ) . A DNA fragment corresponding to the KSHV vGPCR was amplified from BCBL-1 genomic DNA by polymerase chain reaction ( PCR ) and cloned into pcDNA5/FRT/TO between BamHI and XhoI . For protein expression , either the HA epitope or the Flag epitope was inserted upstream or downstream of vGPCR coding sequence , respectively . Plasmids expressing wild-type and mutant K7 polypeptides were described in previous publications [25] , [26] . For lentiviral expression , K7-Flag was cloned into pCDH-EF-puro ( System Bioscience ) between EcoRI and BamHI . HA-vGPCR was digested with EcoRI and BglII , and ligated to pCDH-EF-puro or pCDH-EF-CopGFP that was digested with EcoRI and BamHI . To generate the K7TMStpC , the K7 transmembrane ( TM ) domain was replaced with a heterologous TM segment of Stp C by PCR-based mutagenesis using overlapping PCR primers . All constructs were sequenced for verification . For the shRNA-mediated knockdown of K7 , four pairs of synthetic DNA oligos were annealed and cloned into pLKO . 1 ( Sigma ) that was digested with AgeI and EcoRI . The pLKO . 1 expressing the scrambled shRNA was purchased from Sigma . Plasmids expressing HA-tagged wt and mutant ubiquitin were a kindly gift from Dr . James Z . J . Chen ( UT Southwestern ) . HEK293T ( 293T ) , HeLa , and NIH3T3 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum , 5 mM L-glutamine , 100 U/ml penicillin , and 100 µg/ml streptomycin . BJAB , JSC-1 , BCBL-1 , and BCBL-1/T-Rex_Rta cells were grown in RPMI 1640 supplemented with 10% fetal calf serum , 5 mM L-glutamine , 100 U/ml penicillin , and 100 µg/ml streptomycin . BCBL-1 cells were treated with phorbol-12-teradecanoate-13-acetate ( TPA , 20 ng/ml ) to induce lytic replication . HeLa cells were transfected with Fugene 6 ( Roche ) , 293T cells were transfected with calcium phosphate ( Clontech ) , ECV cells were transfected with lipofectamine ( Invitrogen ) , and BJAB cells were transfected with electroporation at 220 V/975 µF . The stable BCBL-1/T-Rex_Rta inducible cell line was maintained and induced as previously described [30] . Immunoprecipitation and immuno-blot analyses were performed as previously described [26] . Immunoblot detection was performed with anti-V5 antibody ( 1∶5000 , Invitrogen ) , anti-Flag M2 antibody ( 1∶5000 , Sigma ) , anti-HA ( 1∶2000 , Covance ) , anti-tubulin ( 1∶250 , Santa Cruz ) , or anti-actin ( 1∶30 , 000 , Abcam ) . Proteins were visualized with chemical luminescent detection reagent ( Pierce ) and a Fuji LAS-3000 camera . One million KSHV latently infected BCBL-1 or JSC-1 cells were treated with either TPA ( 20 ng/ml ) to induce viral lytic replication and harvested at various time points . Alternatively , KSHV lytic replication was induced in BCBL-1/T-Rex_Rta stable cells with doxycline ( 1 µg/ml ) . Total RNA was extracted with RNAeasy column ( Qiagen , CA ) and digested with DNase I at 37°C for 1 h . After phenol/chloroform extraction , 1 µg of total RNA was used for first-strand cDNA synthesis using an oligo ( dT ) primer . Then , 1 µl of cDNA was added to 19 µl of PCR mixture and gene of interest was amplified using specific primers . PCR products were resolved on agarose gel and photographed . For each gene of interest , dilution of original cDNA and cycle number were determined to warrant that PCR products were generated within the linear range of PCR reaction . Total RNA from tumor tissues was extracted with triazol ( Invitrogen , CA ) and ethanol precipitation as previously described . Transiently transfected ECV cells were pulse labeled with 35S-methionine/cysteine ( Met/Cys ) for 30 min . After extensive washing with phosphate buffered saline ( 137 mM NaCl , 2 . 7 mM KCl , 10 mM Na2HPO4 , 2 mM KH2PO4 , pH 7 . 4 ) , cells were chased with cold medium up to 16 h . At various time points , cells were harvested , washed with cold PBS , resuspended in RIPA buffer ( 50 mM Tris-HCl [pH 7 . 4] , 150 mM NaCl , 0 . 5% sodium deoxycholate , 0 . 1% SDS , 1% NP40 , 5 mM EDTA/EGTA ) , and lysed by passing through 26-G syringe for 15 times . Centrifuged supernatant was pre-cleared with protein A/G agarose and mixed with 2 µg of anti-Flag M2 antibody . Incubation was carried out at 4°C for 4–6 h . Protein A/G agarose was added and incubation was further extended for 90 min . After extensive washing with RIPA buffer , precipitated proteins were resolved by SDS-PAGE and analyzed by autoradiography . The relative intensity of a selected protein band was quantified and its half-life was calculated . When vGPCR degradation route was investigated , 20 µM of lactacystin and MG132 ( proteasome inhibitors ) or 50 µM of chloriquine ( a lysosome inhibitor ) was added during the chase period . IP and autoradiography were performed similarly . The reporter cocktail consists of plasmids expressing fire fly luciferase ( 50 ng/µl ) and β-galactosidase ( 100 ng/µl ) . While β-galactosidase expression is driven by a housekeeping glucophosphokinase promoter , the expression of fire fly luciferase is under control of response elements of NF-κB , NF-AT , and AP-1 transcription factor . 293T cells were transiently transfected with 2 . 5 µl of reporter cocktail , and 200 ng of plasmids expressing vGPCR and K7 . For each transfection , the total amount of plasmid was balanced with an empty vector ( pcDNA5/FRT/TO ) . At 36 h after transfection , cells were harvested and lysed on ice . Centrifuged supernatant was used to measure luciferase and β-galactosidase activity according to manufacturer's protocol ( Promega ) . NIH3T3 stable cells were treated with vehicle ( DMSO ) , cyclohexamide ( CHX , 1 µg/ml ) , or TNF-α ( 5 ng/ml ) plus CHX ( 1 µg/ml ) for 24 hours . Cells were harvested and live cells were scored by trypan blue staining as previously described [25] . Viable cells treated with drugs divided by viable cells treated by DMSO was used to obtain cell viability in percentage . BJAB , HeLa , or BCBL-1 cells were fixed with paraformaldehyde and permeabilized with Triton X-100 ( 0 . 2% in PBS ) . After stained with primary and secondary antibodies , cells were analyzed by immunofluorescence microscopy as previously described [26] , [49] . vGPCR in BCBL-1 cells was detected with a gift rabbit polyclonal antibody provided by Dr . Gary Hayward [7] . For commercial antibodies , mouse monoclonal anti-Flag antibody ( 1∶1500 ) , rabbit polyclonal anti-Flag antibody ( 1∶400 , Sigma ) , mouse monoclonal anti-V5 antibody ( 1∶500 , Invitrogen ) , sheep anti-TGN46 ( 1∶200 , Serotec ) , rabbit anti-PDI ( 1∶200 , Calbiochem ) were used . All conjugated secondary antibodies were obtained from Molecular Probes and diluted at 1∶1000 ( Alexa 488-conjugated ) or 1∶500 ( Alexa 568 or Alexa 647-conjugated ) . Four shRNA seuquences were designed using Dharmacon software and cloned into pLKO . 1 . These sequences are: 5′ TCATCCGTATTGTGTATAT 3′; 5′ CATCGTGAGTTGGTTAATA 3′; 5′ TGGCTACTCTGCTCGATTA 3′; 5′ TGAAGGATGATGTTAATGA 3′ . Together with packaging plasmids DR8 . 9 and VSV-G , pLKO . 1 plasmids expressing various K7 shRNA molecules were transfected into 293T cells with Fugene 6 ( Roche ) . Lentivirus expressing the scrambled shRNA was produced similarly . Filtered lentivirus was used to infect BCBL-1 cells at 20 MOI in medium containing 10 µg/ml polybrene . To increase infection efficiency , cells were centrifuged at 1 , 800 rpm , 30°C for 1 h and incubation was further extended for up to 12 h . The infection was repeated once and cells were selected with puromycin at 1 µg/ml . At 48 h later , BCBL-1 cells were treated with TPA ( 20 ng/ml ) to induce KSHV lytic replication . NIH3T3 cells were infected with lentiviruses to establish stable cell lines expressing K7 with puromycin selection . Then , NIH3T3/puro and NIH3T3/K7 cells were further infected with lentivirus expressing GFP or vGPCR . This lentiviral infection was repeated once to obtain stable cells expressing K7 , vGPCR , or vGPCR and K7 . Cells were cultured in complete DMEM medium containing puromycin ( 1 µg/ml ) . To measure the doubling time , 2×105 cells were plated and cells were counted at 24 h and 48 h later . The soft agar assay was performed as described by Liang et al [50] . Stable NIH3T3 cells ( 5×104 ) were mixed with 1×105 normal NIH3T3 cells and cultured for two weeks in regular culture medium without puromycin . All animal experiments were performed according to the National Institutes of Health principles of laboratory animal care and approved by the University of Texas Southwestern Medical Center . Stable NIH3T3 cells ( 3×106/site ) expressing GFP , K7 , vGPCR , or vGPCR and K7 were injected subcutaneously into the flanks of 6- to 8-wk-old mice ( athymic , nu/nu , Jackson Laboratory ) . | Kaposi's sarcoma-associated herpesvirus ( KSHV ) is the etiological agent of Kaposi's sarcoma . KSHV is also found in primary effusion lymphoma and multicentric Castleman's disease , rare lymphoproliferative diorders associated with immuno-suppression . The KSHV genome encodes a G protein-coupled receptor ( vGPCR ) that is believed to contribute to the KSHV-associated malignancies . vGPCR is a ligand-independent , constitutively active signaling molecule . It is not clear how vGPCR is negatively regulated . Here , we report that the KSHV small membrane K7 protein interacts with vGPCR through its putative transmembrane domain . Interaction with K7 retains vGPCR in the ER and facilitates its degradation by the proteasome , thereby reducing vGPCR protein expression . Consequently , K7 significantly reduces vGPCR-mediated transformation in vitro and tumor formation in nude mice . Our findings reveal that K7 functions as a viral factor to dampen vGPCR protein expression and negatively modulate the tumor-inducing capacity of vGPCR , implying that KSHV has evolved mechanisms to avoid deleterious effects and to permit persistent infection within its host . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"microbiology/cellular",
"microbiology",
"and",
"pathogenesis"
] | 2008 | Kaposi's Sarcoma-Associated Herpesvirus K7 Induces Viral G Protein-Coupled Receptor Degradation and Reduces Its Tumorigenicity |
St . Louis encephalitis virus ( SLEV ) is a causative agent of encephalitis in humans in the Western hemisphere . SLEV is a positive-sense RNA virus that belongs to the Flavivirus genus , which includes West Nile encephalitis virus , Japanese encephalitis virus , Dengue virus and other medically important viruses . Recently , we isolated a SLEV strain from the brain of a horse with neurological signs in the countryside of Minas Gerais , Brazil . The SLEV isolation was confirmed by reverse-transcription RT-PCR and sequencing of the E protein gene . Virus identity was also confirmed by indirect immunofluorescence using commercial antibodies against SLEV . To characterize this newly isolated strain in vivo , serial passages in newborn mice were performed and led to hemorrhagic manifestations associated with recruitment of inflammatory cells into the central nervous system of newborns . In summary this is the first isolation of SLEV from a horse with neurological signs in Brazil .
St . Louis encephalitis virus ( SLEV ) is a mosquito-borne virus that causes human and animal encephalitis in the Western hemisphere . SLEV is considered endemic in the Americas , with encephalitis cases being diagnosed from Canada to Argentina [1]–[3] . There is no vaccine or treatment available for St . Louis encephalitis . SLEV is a single-stranded positive sense RNA virus , with approximately 50 nm in diameter and a genome of 11 kb . SLEV is a member of the Flavivirus genus in the Flaviviridae family , together with several important pathogens such as West Nile virus ( WNV ) , Japanese encephalitis virus ( JEV ) , Dengue virus ( DENV ) , Yellow fever virus ( YFV ) and others [4] , [5] . Viral life cycle is enzootic and birds are the natural amplifying host [6] . Other vertebrates ( e . g . wild animals , horses , and humans ) are considered accidental/final hosts [7]–[9] . Human infections with SLEV are mostly asymptomatic . Infected individuals can present mild malaise or flu-like symptoms , especially young or middle-aged patients [6] , [10] . Severe cases are clinically characterized by high fever , neurological dysfunction , altered consciousness , and headache; which are accompanied by encephalitis or meningoencephalitis that affects more often the elderly [11]–[13] . Lethality rates in severe cases can reach 30% , and are associated to direct damage to the central nervous system ( CNS ) [3] . Acute illness can be followed by prolonged convalescence with cognitive and psychosocial deficits for over a year [6] , [14] . Disease in wild or domestic animals has not been described , although many species are infected or are serologically positive for SLEV in endemic areas [6] , [15]–[19] . SLEV has been detected in Brazil for over 40 years , isolated from arthropods [19] or by serological surveys in birds [20] and mammals [18] , [21] . SLEV was isolated from two patients in the Amazon region in 1970's [22] , [23] and isolated again from a dengue-suspected patient in Southeastern Brazil , in the early 2000's [2] . Interestingly , SLEV infections in humans were identified in southeast Brazil in the following years , under an outbreak of DENV-3 , together with the first a human case of DENV-3 and SLEV co-infection [24] , [25] . Here we describe the first isolation of SLEV from a horse with neurological signs in Brazil . SLEV identity was confirmed by molecular and serological techniques , and by inoculation of newborn mice . Our findings highlight the importance of effective arboviral surveillance .
Our animal study followed national guidelines ( Law number 11 . 794 , 8/10/2008 ) , which governs the use of animals for experimental procedures . All experimental procedures were approved and complied with the University of Minas Gerais ( UFMG ) Committee for Ethics in Animal Experimentation ( CETEA ) regulations , under protocol number 163/2011 . Pregnant female mice were acquired from Centro de Bioterismo ( CEBIO ) of UFMG ( Belo Horizonte , Brazil ) . Newborn Swiss mice ( 24 hours old ) were used in animal model development experiments . All mice were kept under controlled temperature ( 23°C ) with a strict 12 h light/dark cycle , food and water available ad libitum , under specific pathogen-free conditions , in the animal warren at the Departamento de Clínica e Cirurgia Veterinárias of UFMG . Brain tissue from horses that presented neurological symptoms before death were sent to Laboratório de Saúde Animal at Instituto Mineiro de Agropecuária ( LSA/IMA ) in Belo Horizonte , Brazil . Samples that were PCR negative for Rabies virus were sent to Laboratório de Patologia Molecular at UFMG and stored at −80°C . Tissue samples were processed for RNA extraction and screened by nested RT-PCR . There were no tissue samples available for histopathology . Reaction parameters and primers were used as described by Ré and colleagues [26] . Primers used for the initial amplification were SLE 1497 ( + ) RRYATGGGYGAGTATGGRACAG , SLE 2517 ( − ) CTCCTCCACAYTTYARTTCACG , and primers for the final amplification were SLE ( + ) 2002 TGGAYTGGACRCCGGTTGGAAG and SLE ( − ) 2257 CCAATRGATCCRAARTCCCACG . SLEV RT-PCR amplicon band was purified from an agarose gel and sequenced in Megabace 1000 sequencer . Edited sequences were aligned by CLUSTAL/W , using the BioEdit program , version 5 . 09 . The resulting sequence was deposited in GenBank ( accession number KF718857 ) . A phylogenetic tree was generated using the Molecular Evolutionary Genetics Analysis software ( MEGA - www . megasoftware . net ) , version 4 ( MEGA 4 ) [27] . The neighbor-joining method was used to generate bootstrap of 1 , 000 replications using p-distance . Nucleotide sequences were used to perform a similarity search in sequence databases , using BLAST algorithm ( http://blast . ncbi . nlm . nih . gov/ ) . Construction of the phylogenetic was based on 39 other SLEV sequences available at GenBank , which were distributed within the following genotypes: IA , IB , IIA , IIB , IIC , IID , IIE , IIF , IIG , III , IV , VA , VB , VI , VII , VIIIa; VIIIB and 3 out-groups: WNV , JEV , and DENV , as detailed in Table 1 . The isolated viral strain was categorized within genotypes as previously described [19] , [28]–[31] . A tissue fragment of the SLEV-positive brain was homogenized and clarified by centrifugation . The brain homogenate was inoculated on monolayers of the mosquito cell lineage C6/36 and cultures were monitored for cytopathic effect ( CPE ) daily . Viral stocks were collected for up to five days post-infection and re-inoculated ( 500 µL of supernatant on a fresh culture ) two times , for a total of three passages . The viral titer from supernatant of the third passage was determined by focus immunodetection assay ( FIA ) as previously described [32] . We obtained SLEV stocks at 1×103 focus forming units , or FFU , per mL of culture supernatant . To characterize the isolated SLEV strain , 40 µL of SLEV stocks were inoculated in newborn mice by intracranial route ( frontal left region of the brain ) . Homogenates of a pool of brain samples ( 10% in PBS ) from the littermates were used for preparing the inoculum for the next passage . Infected newborn brains were collected at the onset of neurological disease or at day 7 post-infection , to produce new virus stocks or to process for histological analysis . Brain suspensions were passed seven times in newborn mouse brains and relevant experimental controls were maintained . Clinical alterations were assessed daily in newborns after each SLEV passage and representative pictures were taken . Tissue samples , including brain , kidney , liver , lung , heart , and fragments of the thoracic and pelvic limbs , were obtained from two newborn mice at each SLEV passage , immediately fixed in 4% buffered formaldehyde , processed and embedded in paraffin . Tissue sections ( 4 µm thick ) were stained with hematoxylin and eosin ( HE ) , and examined under light microscopy . Micrographs were taken using a Spot Insight color camera coupled to Olympus BX41 microscope . C6/36 cells were harvested and seeded in 24-well plates with gelatin-coated coverslips , and incubated at 28°C for at least 3 hours . Cells were infected with a virus stock derived from the 7th SLEV passage in newborn mice , at a multiplicity of infection ( MOI ) of 1 , for 1 h . At 24 hours post-infection , cells were fixed , permeabilized and stained with a mouse anti-SLEV monoclonal antibody ( MSI-7 , clone 6B6C-1 , MAB8744; Merck Millipore , USA ) followed by Alexafluor 488-labelled secondary anti-mouse ( Molecular Probes , Invitrogen , USA ) . Experiments were performed with control group with or without primary antibodies . Stained coverslips were mounted in Mowiol 4–88 ( Polysciences , Inc . , USA ) and analyzed using an Olympus BX61WI microscope equipped with a FV300 confocal scanning unit . Images were analyzed with imageJ software .
From a total of 170 brain samples from horses with neurological disease received and analyzed by PCR at the Laboratório de Patologia Molecular at UFMG , one sample was identified as positive for SLEV . This brain sample was obtained in March 2009 ( late summer ) from a 12 years-old male horse of undefined breed , which died 72 hours after presenting neurological signs . Those neurological signs were described as incoordination , depression , and flaccid paralysis of the hind limbs . The horse came from a farm in Abaeté , countryside of Minas Gerais State , 207 kilometers from the capital , Belo Horizonte . In the same farm there were two additional horses that remained clinically healthy . Importantly , this SLEV-positive brain sample was negative for other pathogens commonly associated with encephalitis in horses , including Rabies virus , Equine Herpesvirus-1 , Equine Herpesvirus-4 , West Nile virus , Eastern equine encephalitis virus , Western equine encephalitis virus , Venezuelan encephalitis virus , and Sarcocystis neurona . The SLEV amplicon originated by the RT-PCR reaction was sequenced and deposited in GenBank ( Submission ID #1663732 ) , referring to SLEV strain MG150 . Phylogenetic analysis of a 903 bp amplified sequence from partial Envelope ( E ) gene region [26] indicated that the isolate from the horse was within the cluster of the VB genotype ( Figure 1 ) . A higher degree of nucleotide identity ( 97–98% ) was observed among ten SLEV strains from the Brazilian Amazon region , of the VB genotype ( BRA-71 , BRA-78 , BRA-72 , BRA-60 , BRA-84B , BRA-74B , BRA68B , BRA-84A , BRA-74D , BRA-73E ) by comparison with nucleotide sequences previously deposited in GenBank using BLAST algorithm ( http://blast . ncbi . nlm . nih . gov/ ) . Among the foreign isolates with a similar level of nucleotide identity ( i . e . 98% ) , one ( F72M022 ) , also genotype VB originated in Florida from opossum in 2006 [33] . After confirmation of the virus identity by sequencing , we focused our efforts on isolating SLEV from the horse tissue . A brain fragment from the SLEV-positive horse was homogenized and inoculated in C6/36 mosquito cells , which is a cell lineage suitable for arbovirus propagation . All passages were tested for SLEV by RT-PCR , and were all positive beginning at the second passage . The virus was isolated after three passages , obtaining a SLEV stock of 1×103 FFU/mL of supernatant . To characterize this newly isolated strain in vivo , we performed serial passages of SLEV by intracranial inoculation of newborn mice , which allowed us to gather some data on MG150 strain pathogenicity . Increase of clinical signs and circulatory changes were associated with increased mortality that reached 100% at the 7th passage ( Figure 2 ) . At third passage , edema and necrosis at the distal extremity of the hind limbs and at the tip of the tail were observed in SLEV infected newborn mice ( Figure 3 ) . After the fourth passage SLEV infection was associated with behavioral changes ranging from excitability to apathy and neurological changes including tremors , loss of proprioception , and walking in circles , which were accompanied by the same circulatory changes as observed at the third passage ( Figure 3 ) . Importantly , neurological changes in the final SLEV passages were accompanied by hemorrhage in the CNS and peritoneum ( Table 2 ) . Neither mortality nor clinical signs were observed in uninfected control mice . Histological analysis indicated hyperemia and discrete multifocal hemorrhage in CNS from all infected newborn mice at the 4th passage . Multifocal mild lympho-hystiocytic inflammatory infiltrate in the leptomeninges and around blood vessels in the brain was observed in one newborn mouse at the 4th passage ( Figure 4A ) . Focal mild neuronal degeneration , and multifocal hemorrhage were also observed ( Figure 4B ) . Circulatory changes such as hyperemia and multifocal hemorrhage were noticed in several organs , including brain , kidney , liver , lung , and heart at the 5th passage ( Figure 4C ) . Newborn mice at the 6th passage also developed hyperemia in the kidney and lung , and multifocal hemorrhage in liver and brain ( Figure 4D ) . Newborn mice at the 7th passage had hyperemia in the kidney , liver , brain , and lung ( Figure 4E ) . No histological changes were detected in uninfected control mice . Together , these data indicates that SLEV can cause disease in newborn mice . Virus adaptation to the murine host increased after each passage , resulting in neurological and circulatory changes consistent with Flavivirus infection , providing further evidence that our isolated virus strain was indeed SLEV . Inocula ( i . e . CNS tissue homogenate pools ) resulting from each passage in mice were submitted for RNA extraction for confirmation of viral detection by RT-PCR ( data not shown ) . The same procedure was performed with organs that had gross changes . Pool of organs , including the liver , heart , kidney , and lung from the 4th to the 7th passage were analyzed . CNS and other organs were positive in all RT-PCR assays , confirming the presence of viral RNA in tissues . To further confirm SLEV identity , we performed an immunofluorescence assay , using a commercial monoclonal antibody to detect SLEV proteins in cell culture . Monolayers of C6/36 mosquito cells were infected with the 7th SLEV passage , fixed and stained 24 hours post-infection . Infected cells were positively stained with the anti-SLEV antibody , which was not observed in mock-infected cells , confirming the identity of this SLEV strain in an antibody-based test ( data not shown ) .
The first isolation of SLEV from a horse that died due to a neurological disease in Brazil is a significant event . Virus isolation from horse brain tissue , together with molecular and immunofluorescence data , confirms that SLEV was the agent that caused disease and , ultimately , the horse death in this case . To our knowledge , this is the first observation that SLEV can cause disease in wild or domestic animals , which indicate that some aspects of SLEV viral cycle and its ability to cause disease need further studies . Furthermore , a model of newborn mice infection for characterization of SLEV was thoroughly described . In terms of public health and epidemiology , the first identification and isolation of SLEV in the State of Minas Gerais adds to previous reports regarding SLEV detection in Brazil [2] , [19] , [21] , [23] , [24] , and neighboring South American countries [34] , [35] , which strongly indicates that SLEV circulates in Brazil . Importantly , for SLEV epidemiological surveillance purposes , dengue is endemic and is an important health problem in Brazil . Antigenic similarity between SLEV , DENV and other flaviviruses , especially in terms of their envelope protein , generates cross-reactive antibodies that make serological detection of SLEV infections problematic , especially during the frequent dengue outbreaks [36] . Furthermore , SLEV infection can cause febrile illness and even hemorrhagic manifestations that are indistinguishable from mild and severe dengue fever cases , respectively [37] . In spite of these difficulties , molecular screening methods are available and could be employed to monitor SLEV circulation , allowing for preparedness in case of virus re-emergence or SLEV encephalitis outbreak , which has taken place in Argentina [38] and several times in the United States [39] . Considering that the SLEV strain isolated in this study came from a horse with neurological signs , and that the virus was able to induce systemic and neurological sings in mice , the virulence of the circulating strains should be evaluated . A serological survey involving five Brazilian states , including Minas Gerais , resulted in a prevalence of 36% with a total of 753 horses sampled [40] . Despite the existence of eight lineages and fifteen subtypes of SLEV , namely IA , IB , IIA , IIB , IIC , IID , IIG , III , IV , VA , VB , VI , VII , VIIIA , and VIIIB , phylogenetic studies based on the E gene indicate that genotypes I and II are found predominantly in North America , whereas genotypes III to VIII have been isolated in South and Central Americas [19] . Brazilian SLEV isolates have been classified within the genotypes II , III , V ( A and B ) , and VIII ( A and B ) . Genotypes V and VIII are predominately Brazilian Amazon region , whereas genotypes II and III have been isolated in the State of São Paulo ( Southeast Region ) . In this study , the isolate from a horse had a higher degree of identity ( 97–98% ) with the VB genotype , suggesting that this sample was likely originated from the Brazilian Amazon Region . The circulation of SLEV from the Amazon Region in the Southeast Region of Brazil suggests a possible involvement of migratory birds in disseminating the virus , since SLEV has been detected in 49 species of wild birds in Brazil , many of which are migratory [19] , [21] . SLEV strains genotype VB were isolated since 1960s from wild birds and mosquitoes or sentinel animals at a surveillance site for arboviroses in a forested area of Pará state [19] . Migratory birds may have also been related to the periodic introduction of South American SLEV genotype V in Florida ( USA ) in 2006 , originated possibly from Brazil , Mexico , or Panama [33] . In our efforts to characterize SLEV strain MG150 in vivo , to study its pathogenicity , we noticed the virus progressively adapted to serial passagens in newborn mice . The mouse has been previously used as a model for assessing SLEV virulence [41] . Disease presented by newborn mice infected with the last SLEV passages had some similarities to human disease , such as the development of hemorrhagic manifestations [19] , mortality , and neurological changes [12] , [14] . Importantly , lesions in organs such as the liver and heart are likely to reflect a systemic circulatory change rather than any specific viral tropism for these organs . | St . Louis encephalitis virus ( SLEV ) , a member of the Flavivirus genus , which includes West Nile encephalitis virus , Japanese encephalitis virus , Dengue virus , and other medically important viruses , is a cause of encephalitis in humans and animals . SLEV is considered endemic in the Americas , and currently there is no vaccine or specific treatment available for controlling of preventing SLEV-induced encephalitis . In this study we describe the first isolation of SLEV from an adult male horse with neurologic disease , which was further characterized by molecular and serological methods . Phylogenetic analysis of a 903 base pairs amplified sequence from partial Envelope ( E ) gene region indicated that the isolate from the horse was within the cluster of the VB genotype . In addition , inoculation of the SLEV isolate intracranially in newborn mice resulted in circulatory and neurological changes . This is the first report of isolation of SLEV from a horse with neurological disease in Brazil . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [] | 2013 | Isolation of Saint Louis Encephalitis Virus from a Horse with Neurological Disease in Brazil |
Typhoid fever is endemic in many developing countries . In the early 20th century , newly industrializing countries including the United States successfully controlled typhoid as water treatment ( chlorination/sand filtration ) and improved sanitation became widespread . Enigmatically , typhoid remained endemic through the 1980s in Santiago , Chile , despite potable municipal water and widespread household sanitation . Data were collected across multiple stages of endemicity and control in Santiago , offering a unique resource for gaining insight into drivers of transmission in modern settings . We developed an individual-based mathematical model of typhoid transmission , with model components including distinctions between long-cycle and short-cycle transmission routes . Data used to fit the model included the prevalence of chronic carriers , seasonality , longitudinal incidence , and age-specific distributions of typhoid infection and disease . Our model captured the dynamics seen in Santiago across endemicity , vaccination , and environmental control . Both vaccination and diminished exposure to seasonal amplified long-cycle transmission contributed to the observed declines in typhoid incidence , with the vaccine estimated to elicit herd effects . Vaccines are important tools for controlling endemic typhoid , with even limited coverage eliciting herd effects in this setting . Removing the vehicles responsible for amplified long-cycle transmission and assessing the role of chronic carriers in endemic settings are additional key elements in designing programs to achieve accelerated control of endemic typhoid .
Typhoid fever caused by Salmonella Typhi was controlled in developed countries after widespread water and sanitation improvements were introduced , but remains a pressing public health problem in many developing countries [1–3] , with estimates of global burden ranging from approximately 10 to 20 million cases per year [4–7] . Effective control of typhoid is often impeded by lack of knowledge of the local dominant transmission routes , age-specific incidence of disease and role played by chronic carriers ( individuals with S . Typhi-colonized gallbladders who can transmit the pathogen for decades ) . Modeling endemic typhoid in specific epidemiologic niches can guide investments and prioritization of control strategies such as identifying cost-effective targets for immunization with new and existing typhoid vaccines and improving water/sanitation/hygiene ( WASH ) infrastructure [8–11] . The usefulness of mathematical models is enhanced when comprehensive and precise input data are available . Regrettably , such data are often unavailable where typhoid is currently endemic . Thus , modeling data from sites where typhoid has already been successfully controlled offers an opportunity to dissect the mechanisms of transmission in those data-rich settings and also allows those models to be adapted to study transmission and the impact of potential interventions in modern endemic settings . Enigmatically , typhoid was hyper-endemic in Santiago , Chile from the mid-1970s through early 1990s [12] , even though 96% of Santiago households had access to treated , bacteriologically-monitored water and ~80% had toilets connected to the municipal sewerage system [12–14] . Typhoid fever incidence in Santiago was stable through the early 1970s but doubled in 1977 and 1978 without an obvious explanation [12 , 13] . This prompted the Head of the Epidemiology Unit of the Ministry of Health of Chile ( MoHC ) , Dr . José-Manuel Borgoño , and the Pan American Health Organization in 1978 to invite two external advisers to Chile , Drs . Branco Cvjetanovic and Myron M . Levine , to provide independent unbiased assessments of the endemicity of typhoid and offer recommendations . One recommendation was to establish a Typhoid Fever Control Program ( TFCP ) , which was instituted in 1979 by Dr . Borgoño with Dr . Levine as external adviser . During the ensuing 13 years , the Chilean TFCP: strengthened clinical , epidemiologic and bacteriological surveillance within the Metropolitan Region ( Santiago and environs ) [13]; quantified the reservoir of chronic gallbladder typhoid carriers ( prevalence , 694/105 adults ) [15 , 16]; identified risk and protective factors for transmission [14]; hypothesized that an unusual mechanism of amplified transmission was maintaining hyper-endemic typhoid disease across all socioeconomic levels and neighborhoods in an urban population with widespread access to potable water and flush toilet sanitation [17 , 18]; and undertook environmental bacteriology investigations to confirm the hypothesis [17 , 18] . The environmental bacteriology studies identified irrigation of crops with untreated raw sewage wastewater during the rainless summer months as the predominant mechanism that was sustaining amplified transmission [17 , 18]; 90% of the crops were vegetables ( lettuce , cabbage , celery ) eaten uncooked . A computer-based model of endemic typhoid in Santiago was developed based on pre-intervention incidence data ( 1968–1976 ) to explore the impact of future vaccination and sanitation interventions to control typhoid [8] . Two major preventive interventions were instituted in Santiago during the period of the TFCP , with each followed by marked drops in typhoid incidence . First , from 1982 through 1991 , were four large-scale field trials of Ty21a live oral vaccine among 514 , 150 schoolchildren , the age group that accounted for >60% of cases of typhoid [19–24] . Second was a sanitation intervention . In April 1991 , following an outbreak of 41 confirmed cholera cases that occurred in Santiago [25] , the practice of irrigating crops with raw sewage-containing wastewater was prohibited [13 , 25] . Thenceforth , this strictly-enforced intervention abruptly interrupted the long-standing amplified transmission of typhoid in Santiago [13 , 18] . The availability of data from an extended period of endemic transmission and well-documented non-coincident vaccine and sanitation interventions with surveillance across both time periods offered unique data to examine the drivers of endemicity in Santiago , and to estimate the impacts of multiple interventions across a single population . By examining assumptions and constraints through mathematical modeling of data from a historical site , one can better understand typhoid transmission in current hyper-endemic loci .
The typhoid model was built upon the EMOD 2 . 11 framework [26] . The structure was created by modifying a previous individual-based typhoid model [27] . Modifications include adding multiple transmission routes and simplifying immunity; sterilizing and clinical immunities were combined into a single immunity structure , absent data to inform individual durations . Santiago’s population was simulated using age-specific fertility and mortality rates estimated with Instituto Nacional de Estadisticas census data [28] . The model was initialized with the earliest reported age distribution , with individuals entering and exiting the model through age-specific fertility and mortality . Typhoid transmission occurs through either the “short-cycle” or “long-cycle” . Short-cycle denotes infections transmitted from person-to-person through proximate contaminated food vehicles [29] . Long-cycle signifies infections transmitted through environmental mediators such as contaminated water [30] , crops irrigated with untreated sewage [18] or widely-distributed contaminated commercial food products [31] . Our model captures both transmission routes; individuals can both contaminate and become exposed to the short-cycle and long-cycle “composite of contaminated vehicles of transmission , ” or CCVT . Infectious individuals shed into both short-cycle and long-cycle CCVTs at a daily rate per their infectiousness , in colony-forming units ( CFU ) . In the absence of quantitative shedding data , this value is an estimated free parameter for acute infectiousness ( AI , Table 1 ) , with multipliers for non-acute disease states , explained in detail below . The die-off of S . Typhi over time outside the human body varies in different environmental niches [32 , 33] . Our estimate of the long-cycle CCVT daily decay at rate LD ( Table 2 ) was based on the S . Typhi die-off results reported by Cho et al [32] . Assuming short-cycle transmission ensues primarily via proximate food vehicles that are prepared and consumed daily , short-cycle CCVT decays at 100% each day . The number of times a susceptible individual is exposed to each CCVT is determined by a daily Poisson process , with rates for both short-cycle and long-cycle ( EL , ES , Table 1 ) left free for model fitting . In the model , the probability of clinical response post-exposure differs by transmission route . Short-cycle transmission is assumed to always involve a food vehicle that protects S . Typhi against gastric acid or contains a large inoculum ( or both ) . Thus , we assume that small inocula that may not transmit through the long-cycle are successfully transmitted by short-cycle vehicles . Thus , short-cycle is modeled in a direct transmission framework , where probability of infection is the population-scaled short-cycle CCVT divided by the total potential short-cycle CCVT . We undertook to address the heterogeneity inherent in long-cycle transmission , which leads to potential variation in inoculum size ( i . e . , water-borne vs . food-borne long-cycle exposure ) . Therefore , the probability of infection through the long-cycle is determined by a dose-response function where population-scaled CCVT for the long-cycle is the infecting inoculum size . The dose-response function is a beta-Poisson curve fitted to Maryland experimental challenge dose-response data where S . Typhi inocula were administered without buffer ( Table 2 ) [34 , 35] . We assume Maryland challenge data represent Santiago as a whole , absent additional data . Infected individuals in the model transition through disease states beginning with the incubation period ( Fig 1 ) , an asymptomatic infectious period monitored via stool cultures in typhoid challenges [36 , 37] . The incubation period is informed by the Maryland challenge model [38] , which demonstrated shorter incubations for those challenged with high ( 108−109 CFU ) versus lower ( 105 CFU ) inocula . Incubation period in the model can be drawn from one of two estimated lognormal distributions [38] , with the cutoff for high vs . low dose being the mid-point between 105 and 108 CFU . We are unaware of direct comparisons of S . Typhi counts in stools of incubating , sub-clinical and clinical infections . To approximate , we used isolation rates from stool cultures to inform the relative infectiousness of the incubation period versus acute clinical infections . In the Oxford human challenge model , 26% of stools cultured within 72 hours post-challenge were positive for S . Typhi among those eventually diagnosed with typhoid disease [37] . Similarly , the maximum percentage of positive stools in the first three days post-challenge was ~30% in the Maryland model , assuming equal stool isolation rates each day [36] . In the following week , the maximum daily percentage of positive stool cultures was approximately 60% . The pre-antibiotic era report of Ames and Robins described stool isolation rates to be in the range of 60–70% , during the three weeks following diagnosis of clinical illness [39] . Using these isolation rates as a proxy for relative infectiousness , we assume a baseline relative infectiousness , rI , of 0 . 5 for incubating individuals compared to acute cases ( Table 2 ) . Individuals who excrete virulent S . Typhi following ingestion of an inoculum may or may not develop clinical typhoid [34] , and clinical severity does not correlate with infectious dose [34 , 37 , 38 , 41] . Individuals progress to either clinical or subclinical infection , according to probability of clinical infection pA . During the primary bacteremia of acute typhoid infection , be it clinical or subclinical , S . Typhi always reaches the gallbladder [42] . Further , older adults have abnormal gallbladder mucosa more often than younger adults or children . Therefore , even though all adults with gallbladder disease who have acute typhoid fever or acute sub-clinical infection do not become chronic carriers , they may nevertheless have delayed clearance of S . Typhi from their gallbladder . This feature was observed by Ames and Robins in the pre-antibiotic era [39] . Duration of clinical and subclinical shedding is sampled from a lognormal distribution , stratified by <30 and ≥30 years of age , derived from non-chronic carrier shedding durations of acute infections [39] . Following the infectious period , individuals can revert to susceptible class or become chronic carriers . Both clinical and subclinical infections may lead to chronic carriers in this model , and the probability of becoming a carrier after each infection is age- and gender-specific . The propensity for S . Typhi ( or S . Paratyphi A or B ) to reside long-term in the gallbladder is related to whether a patient has chronic gallbladder disease due to gallstones [43 , 44] , though long-term carriage may occasionally occur in persons without gallstones . Regardless , the prevalence of chronic typhoid carriers parallels the prevalence of cholelithiasis and chronic gallbladder disease . Both are much greater in females than males and the prevalence increases with age [15] . Our model utilizes the sex- and age-specific prevalence of gallstones in Santiago [15] , and multiplies this value by an additional free parameter , pC , to inform the probability of chronic carriage per infection when an individual has gallstones . When examining the age-specific incidence of typhoid fever in Area Norte , Santiago , from 1971–1981 , there appear to be abrupt increases in incidence at 3 and 6 years of age ( Fig 2A ) . As these are the common entry ages of preschool and elementary school , this finding suggests that differential age-specific exposures may influence the occurrence of pediatric typhoid . Our model assumes all individuals are born into an unexposed class and move to the susceptible class at probabilities for each age . Specifically , at each month of age a fitted curve determines the probability of an individual entering the susceptible class . The curve is anchored at 0% exposure at birth , and 100% exposure at age 20 years , with a free slope parameter ( S ) determining the concavity/shape of the function ( Fig 2B ) . We include a mechanism for reinfection with immunity in our model . All individuals enter the model with no prior infections ( Ni = 0 ) . When an individual returns to the susceptible class after a subclinical or clinical infection , the number of previous infections , Ni , increases by 1 . Upon exposure to either short-cycle or long-cycle CCVT , the probability of becoming infected is multiplied by the value resulting from the equation: ( 1-P ) Ni . Currently , there is no decay of Ni over an individual’s life because we assume that in the hyper-endemic Santiago setting , repetitive exposures to S . Typhi achieve immunologic boosting . This assumption is reasonable for endemic settings with frequent exposures to S . Typhi but may not hold where typhoid endemicity is unstable or when individuals leave the hyper-endemic setting . P , the reduction in susceptibility after clinical or sub-clinical infection , is left as a free parameter to be fitted to Santiago dynamics . An individual’s dose through long-cycle transmission in the model is attenuated by a mechanism for seasonality , chosen to represent the likely constant shedding of individuals into the long-cycle vehicle but with differential exposure to it influenced by seasonal crop selection and need for irrigation . In the high ( warm , rainless ) season , we assume individuals are exposed to 100% of long-cycle CCVT . At low season , we assume no exposure to long-cycle CCVT . Degrees of exposure in the intermediate stages are mediated by a trapezoidal function , where ramp-up and ramp-down durations ( RUD and RDD ) , in combination with high-season duration and timing ( PD , EPS ) , determine a linear function connecting high and low seasons outlined in Fig 1 . These parameters are left free for model fitting . Four large-scale field trials of Ty21a vaccine encompassing 514 , 150 schoolchildren were performed in Santiago , beginning in 1982 [19–24]; three trials included placebo control groups . The timing and number of vaccinees for each of the trial years are summarized in Table 3 , with age ranges restricted between 6 and 17 years of age in the absence of exact age distributions . Formulation , number of doses and inter-dose intervals varied in the trials , affecting vaccine efficacy and duration of protection ( Table 3 ) . Vaccination is implemented as a multiplier on an individual’s probability of infection , equal to one minus the vaccine efficacy corresponding to the year and dose listed in Table 3 . The multiplier exists for the duration listed for partially protective formulations ( Table 3 ) , or the estimated parameter , D , for trials equivalent to full efficacy doses or better ( Table 1 ) . After the defined duration of protection , there is no residual immunity or protection assumed in the model , and individuals return to a fully susceptible state . This choice was made according to the observations of cohorts that received sub-optimal regimens ( too few doses ) of otherwise protective vaccines during the field trials . Whereas significant efficacy was observed during years 1 and 2 of a Ty21a vaccine efficacy trial in Area Norte , Santiago , little or no residual immunity was observed during the 3–5 year follow-up [20] . Shifts in age-specific incidence during the vaccine trials offer insight into the true duration of protection from Ty21a vaccine , as maximum follow-up times were 7 years , 5 years and 3 years , with most trials still indicating measurable protection at the last follow-up point . Thus , duration ( D ) of efficacy for full-dose vaccinees was left as a free parameter and included in model fitting . Longitudinal trends in the pre-vaccine era indicate a relatively stable incidence from 1970 to 1976 , with notable increases in 1977–1978 and 1982–1983 ( Fig 3 ) . Apparent shifts in GDP growth and copper prices during these years may be loosely correlated with changes in irrigation practices or food purchase habits ( Fig 3 ) [45] , and may therefore drive exposure to the long-cycle CCVT . Our model included multipliers on the long-cycle CCVT exposure frequency ( mEL ) , for these pre-defined time periods plotted in Fig 3 . We also explored whether the increase in cases from 1982 to 1983 , beginning not long after initiation of the TFCP , could be attributed to increased diagnostics , by fitting a higher pA value for the years after 1982 . The multiplier mEL was assumed to be zero from 1992 to 2000 , when exposure through the long-cycle was interrupted due to a prohibition of irrigation with sewage . Thus , the model was constrained exclusively to short-cycle transmission during this time ( Table 2 ) . Longitudinal changes in exposure to long-cycle CCVT may also have occurred during the vaccine period . After fitting vaccine duration parameters , a linear multiplier on long-cycle CCVT exposure frequency ( mEL ) was fitted between 1984 and 1992 to capture additional incidence reduction seen during the vaccine period that wasn’t captured by the vaccine . Parameters that were not identified from the literature were estimated through model fitting . The estimated parameter values were identified from pre-defined ranges ( Table 1 ) using a gradient ascent algorithm , which iteratively maximizes a combined log-likelihood to approach a local optimum , calculated from the fit of the model to identified data components ( S1 Appendix ) . The TFCP and CMoH collected and reported data from many independent sources , leading to heterogeneity in years of reporting , age tranches and timeframes of each data component . Data were collated from the authors’ personal files ( S1 dataset ) , with additional components and age tranches shared from recent exploratory initiatives [46] . Data from simulated years were extracted to match to report years for calculation of each component of the likelihood . Data related to typhoid incidence was informed by passive surveillance , meaning the patient or the parent/caretaker of a sick child had to make the decision to actively seek health care . The surveillance activities and vaccine field trials were carried out in parts of Santiago and in an era when the vast majority of the population ( except the very wealthy ) sought acute care at health centers ( consultorios ) run by the government . Typhoid fever was a notifiable disease , meaning all healthcare providers were required to report diagnosed cases . Infections reported to the TFCP and CMoH were assumed to be represented by the acute infection disease state in the model ( Fig 1 ) . Four distinct components of the data were used in the first stage of model fitting ( S1 Table ) , which estimated all free parameters in the model with the exception of mEL_C ( Table 1 ) . Age distribution before , during and after the vaccination period ( 1971–1992 ) and pre-vaccine period seasonality ( 1970–1979 ) were derived from Ministry of Health reports . Age distribution of incidence for the year 1984 was excluded due to missing age-specific demographic data from that year . Chronic carrier prevalence ( 1980 ) was obtained from literature estimates , which multiply population cholelithiasis prevalence by gallbladder carriage of S . Typhi . Annual incidence in the pre-vaccine period ( 1970–1983 ) and post environmental intervention ( 1993–1996 ) were derived from Ministry of Health reports , and used for the first stage of model fitting . Annual incidence during the vaccination period was withheld in the first stage of model fitting , due to the potential for the estimate of the free parameter for vaccine duration ( D ) to be influenced by the changing incidence rates during the vaccination period . Due to our assumptions of long-lasting immunity after repetitive exposures and infections in endemic locations , we similarly assume that the age distribution of typhoid in the model is robust to incidence rate changes in the short term , and therefore attribute changes to age distribution during the vaccine period to be a result of the vaccine . The second stage of model fitting involved fixing the parameters estimated in the first stage , and utilizing annual incidence data from the vaccine period ( 1983–1992 ) to fit mEL_C . This allowed us to independently estimate exposure-related changes during the vaccine period , with vaccine duration and other individual-level parameters fixed .
Fit of the model to Santiago dynamics of age distribution , typhoid seasonality , and prevalence of chronic carriers is shown in Fig 4 . Fits of the model to longitudinal trends in reported typhoid fever are plotted in Fig 5A , with estimated parameter values summarized in Table 1 . Increases in exposure to the long cycle CCVT in simulated years 1978 and 1983 resulted in an increased estimated population immunity ( measured by percentage of the simulated population exposed at age 25 ) prior to the vaccination period ( Fig 5B ) , which is a likely contributor to the decline in incidence in the following years . We also modeled a scenario where the increase in cases in 1983 was due to improved diagnostics , which led to a poor fit of the model to data for years after 1983 ( S1 Fig ) . The model estimates that the majority of typhoid infections result from long-cycle transmission in the endemic period but the ratio to short-cycle infections varies seasonally ( Fig 4B ) . Parameters driving seasonality estimate an asymmetrical exposure , with a short ramp-up beginning on day 279 ( October 6th ) , a duration of peak long-cycle exposure of 108 days ending on January 22nd , and a gradual ramp down of 227 days . Predictions from Chile in 1979 based on temperature estimate the growing season to begin August 30th [47] , somewhat preceding our estimated exposure start . Population immunity drives the adult age distribution of simulated typhoid fever in Santiago , which is created by both immunity after clinical typhoid and a high incidence of immunity-boosting repetitive sub-clinical infections . The best-fit model estimates a protection per-infection parameter ( P ) of 99 . 8% , indicating that each initial acute clinical or subclinical S . Typhi infection causes a substantial reduction in susceptibility to subsequent clinical typhoid in the model . Repeat episodes of typhoid fever have been observed in individuals who participated in experimental human challenge/re-challenge studies [48] , or who were members of circumscribed populations that experienced successive typhoid epidemics [49]: these data indicate only modest protection against subsequent typhoid conferred by the initial clinical infection . It is presumed that recurring subsequent exposures in hyper-endemic areas repetitively boost immunity and maintain long-lived protection [50] . No recent studies have addressed clinical reinfections in endemic settings . The best-fit model estimates a large proportion of sub-clinical infections , with the symptomatic fraction of typhoid infection ( pA ) at 5 . 3% . When fitting the model to the estimated chronic carrier prevalence in Santiago , the best-fit probability of carriage following acute or subclinical infection of persons with gallstones ( pC ) was 10 . 8% . When multiplied by the gallstone carriage prevalence rates [15] , we independently estimate the probability of carriage due to infection at age-specific rates that are lower than pre-antibiotic era estimates from New York State , with an age-adjusted rate of 1 . 5 vs . 2 . 9% ( Table 4 , S1 Table ) [39] . During the period of hyper-endemic transmission in the Santiago model , one could offset low levels of short-cycle transmission ( ES ) with high levels of chronic carrier infectiousness ( rC ) to capture longitudinal trends , based on likelihood values ( Fig 5C ) . The availability of incidence data after the interruption of long-cycle transmission in 1991 allowed us to constrain these two parameters that would otherwise be unidentifiable . This is additionally aided by our prevalence estimate of chronic carriers , which is unknown in most endemic locations . The model estimated the relative infectiousness of chronic carriers to be 24% of the infectiousness of acute cases and a short-cycle transmission rate of 0 . 0093 . Other parameters were less identifiable . Specifically , a trade-off exists between acute infectiousness ( AI ) and exposure to the long-cycle ( EL ) , the dominant route of transmission in this context . As acute infectiousness is the primary driver of infectious dose in the model , we can simulate high or low-dose scenarios by adjusting values for acute infectiousness . A higher value of long-cycle exposure frequency ( EL ) can offset a lower value of dose , and vice-versa , resulting in a stable probability of infection despite fluctuating values . Fitted values of 13 , 436 CFU and 0 . 54 were estimated for AI and EL , respectively ( Table 1 ) , but further studies to help identify one or both of these parameters would be valuable . The best-fit model estimates duration of the efficacy of Ty21a vaccine to be 8 . 4 years , as determined by the model’s fit to age distributions during and after the vaccination period ( Fig 4C ) . The model estimates that in addition to the vaccination-related decline beginning in 1983 , there was an estimated 23–53% reduction in exposure to the long-cycle over this period , increasing linearly until after 1991 ( Fig 5A ) . We estimated vaccine impact by comparing simulations of WASH-only and WASH+vaccine scenarios , using best-fit parameters . We see a maximum estimate of 11 . 7% reduction of cases across all age groups in the year 1985 . With 5 . 3% of the overall population receiving full dose vaccines by the end of the trial , and an additional 4 . 6% receiving partially protective formulations , this indicates indirect protection of non-vaccinated age groups is likely occurring . The high coverage and direct protection within vaccinated age groups is reflected in the shifts in age distribution of incidence between 1982 and 1991 , with peak age-specific incidence shifting from 15–19 years of age in 1982 , to 5–9 years of age by 1991 ( Fig 4C ) .
Between 1979 and 1993 , through a multi-faceted applied public health research agenda , the Chilean TFCP generated data on the magnitude of the human chronic carrier reservoir , modes of transmission and impact of vaccine and sanitation interventions in Santiago [13] . We utilized these data in a mathematical model to understand the mechanisms of transmission in this setting , and to estimate the impact of both vaccine and environmental control measures . Similar to observations in many current typhoid-endemic locations [1] , the age distribution of typhoid in Santiago has a paucity of adult cases relative to children . Two primary mechanisms can create this pattern in the mathematical model: i ) the degree of immunity after infection; ii ) and the incidence rate of clinical and sub-clinical infections . Both mechanisms appear to play a role in Santiago . Parameter estimates when fitting the mathematical model to the age distribution of typhoid in Santiago suggest robust immunity after clinical and sub-clinical infection , with a very low probability of repeated infection after an initial infection . Our model-estimated immunity after infection is much higher than what has been demonstrated in challenge studies [48] , but studies of repeat infections in endemic settings are lacking . Additionally , the model estimates the occurrence of approximately 19 sub-clinical infections for each clinical case , leading to a large amount of circulating S . Typhi infection that remains ‘unreported’ in the model . Population-based seroprevalence surveys that detect long-lived anti-flagella H:d responses from both prior sub-clinical and clinical typhoid infections cumulatively over time offer insight into the levels of circulating disease that may go undetected by clinical surveillance [13 , 50] . A cross-sectional prevalence survey of S . Typhi H antibody in Santiago in 1978 found that approximately 50% of 25 year-olds had a reciprocal titer ≥ 40 . Estimates from our model ( Fig 5B ) corroborate the estimate derived by seroepidemiology and predict that 60–70% of the population has been infected by the age of 25 during the pre-vaccination period . Absent data informing decay rates of the H antibody over time , the higher percentage of estimated individuals ever having been exposed compared to antibody prevalence may be explained by the decay of H antibody over time , and at a minimum supports our finding that many subclinical or mild infections occur for each reported clinical case [13] . The proportion of incident typhoid cases reported to public health authorities is notoriously variable among modern typhoid-endemic healthcare locations and can be attributed to differences in treatment-seeking , volumes of blood drawn for culture , and microbiological methods . If we assume the estimated parameter specifying immunity after initial infection is consistent across diverse locations , differences in the adult age distribution of typhoid should only be driven by the rate of disease transmission . The shape of the adult age-specific case distribution may better indicate the force of infection than incidence rate based on an unknown case reporting fraction [13 , 50] . Serological surveillance is needed to confirm these observations in modern typhoid-endemic locations . We estimated a probability of carriage after infection that is lower than the age-specific rates estimated in the pre-antibiotic era ( Table 4 ) [39] . This is expected due to the ability of certain antibiotics ( fluoroquinolones , e . g . , ciprofloxacin ) to diminish chronic carriage after treatment of acute infection . Indeed , a longer ( 4-week ) course of these antibiotics can even eliminate established chronic gallbladder carriage without cholecystectomy [51 , 52] . Our estimate was dependent on the accepted dogma that sub-clinical cases can lead to chronic carriage , while Ames and Robins only followed-up clinically detected cases [39] . We utilized two pieces of data , the prevalence of chronic carriers and the incidence after the environmental intervention , to estimate parameters that typically are unidentifiable: the infectiousness of chronic carriers and the short-cycle transmission rate . When investigating persistence after extreme WASH interventions in modern endemic locations , one should consider that the Santiago results are likely a lower-bound for short-cycle transmission rates , due to the widespread availability of potable water in Santiago households and other improved WASH indicators . Because we see sustained but progressively diminishing transmission in Santiago after interruption of long-cycle transmission in 1991 , we posit that chronic carriers transmitting through the short-cycle are largely responsible . The contribution of carriers should be studied intensively in future projects aimed to achieve accelerated control ( and eventually local elimination ) of typhoid , once amplified long-cycle endemic transmission has been curtailed , including after widespread vaccination with effective vaccines that alter the susceptibility of the population . The impact of large-scale use of Ty21a vaccine was assessed within the context of other potential changes occurring during the 1980s in Santiago . Pre-vaccination incidence increases during 1977 and 1982 led to a subsequent decrease of naïve individuals in the model , leading to an estimated decline in incidence over time independent of vaccination ( Fig 5A ) . Shifts in the age distribution of typhoid fever incidence during the vaccination period were valuable for understanding the duration of efficacy of Ty21a , which the model estimates to be ~1 . 5 years longer than the maximum duration of protection documented in field trials ( 8 . 4 versus 7 years ) . Data additionally support longer durations of efficacy for some less protective formulations and immunization regimens of Ty21a vaccine , past the follow-up times published in literature [19] . For example , over six years of follow-up , three doses of enteric coated capsule and gelatin capsule formulations given in long intervals ( 21 days ) between doses exhibited 55 . 1% efficacy ( 95% CI , 38 . 2–67 . 4 ) and 35 . 0% efficacy ( 95% CI , 13 . 7–51 . 0 ) , respectively , while the enteric-coated capsule formulation administered at short interval ( 2 days ) conferred 62 . 7% ( 95% CI , 47 . 7–73 . 5 ) efficacy ( S2 Table ) . When evaluating vaccination impacts on a population level , it will be important to consider potential shifts in population immunity and age distribution of infection , in the case of an age-targeted vaccine . The model shows that with Ty21a use typhoid incidence at the population level falls more than expected based on direct protection only , indicating indirect effects of the vaccine , which have been described using other methods of analysis [24] . Our finding depends on modeling reduced shedding in protected vaccinees , an assumption documented in Ty21a challenge studies [53] . Additionally , this finding is influenced by the model structure , which assumes a well-mixed environmental reservoir , consistent with the known transmission route . In locations without known transmission routes , the assumption of a well-mixed pool of infection may not be valid and may over-estimate indirect effects . Investigations into spatial scales of transmission are needed when modeling modern endemic locations . Utilizing the model’s trade-offs between dose-response and exposure frequency , driven by parameters acute infectiousness ( AI ) and long-cycle exposure frequency ( EL ) , typhoid dynamics can be simulated at both high-dose and low-dose scenarios . The unidentifiability of these parameters is a limitation of model structure , at present , and studies quantifying estimated exposure levels and frequency would greatly improve our understanding of transmission dynamics and vaccine efficacy in relation to infectious dose . The vaccine efficacy estimates we used in the model were derived from field trial data , which do not account for potential differences in vaccine efficacy in relation to variations in the size of the inoculum ingested . Thus , our impact estimates did not account for potential variation in infectious dose . Experimental challenge studies that assessed the efficacy of parenteral killed whole-cell typhoid vaccines in volunteers showed that high inocula could overwhelm the protective effect of vaccines efficacious against lower doses [34] . Since a similar concern was raised by investigators who assessed the efficacy of a Vi conjugate vaccine in a challenge model [54] , this should be considered when projecting the impact of new conjugate vaccines . In summary , this study utilized unique datasets collected during multiple stages of endemicity and control in Santiago , Chile . Paired with mathematical modeling , we aimed to better understand both the complex dynamics contributing to sustained transmission in this setting . Modeling also allowed us to estimate the contributions of mass use of vaccine , in a time when other water and sanitation measures were underway . Our findings support the use of typhoid vaccines to reduce transmission , but also highlight the importance of identifying and intervening upon the critical long-cycle transmission pathways , allowing for targeted and sustained control . | Typhoid fever was successfully controlled in Santiago , Chile , after a series of interventions including vaccination with a live oral vaccine ( Ty21a ) , and an environmental sanitation improvement , when a ban was put on the irrigation of salad vegetable crops with untreated sewage . Data collected during this period inform seasonality , age distribution and longitudinal trends of disease . We developed an individual-based , mathematical model to both simulate the dynamics of typhoid seen in Santiago , as well as to investigate relative impacts of the vaccine and sanitation interventions . We found that herd immunity resulted from field trials of the Ty21a vaccine and that chronic carriers were a likely driver of sustained transmission at low incidence levels . Modeling typhoid fever in areas that have demonstrated successful control provides insight for control strategies in modern settings . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"age",
"distribution",
"immunology",
"microbiology",
"salmonella",
"typhi",
"health",
"care",
"vaccines",
"preventive",
"medicine",
"bacterial",
"diseases",
"signs",
"and... | 2018 | Typhoid fever in Santiago, Chile: Insights from a mathematical model utilizing venerable archived data from a successful disease control program |
The extent of epigenetic variation is currently well documented , but the number of natural epialleles described so far remains very limited . Determining the relevance of epigenetic changes for natural variation is an important question of research that we investigate by isolating natural epialleles segregating in Arabidopsis recombinant populations . We previously described a genetic incompatibility among Arabidopsis strains based on the silencing of a gene involved in fitness . Here , we isolated a new epiallele resulting from the silencing of a transfer-RNA editing gene in an Arabidopsis accession from the Netherlands ( Nok-1 ) . Crosses with the reference accession Col-0 show a complete incompatibility between this epiallele and another locus localized on a different chromosome . We demonstrate that conversion of an unmethylated version of this allele occurs in hybrids , associated with modifications of small RNA populations . These epialleles can also spontaneously revert within the population . Furthermore , we bring evidence that neither METHYLTRANSFERASE 1 , maintaining methylation at CGs , nor components of RNA-directed DNA methylation , are key factors for the transmission of the epiallele over generations . This depends only on the self-reinforcing loop between CHROMOMETHYLASE 3 and KRYPTONITE , involving DNA methylated in the CHG context and histone H3 lysine 9 methylation . Our findings reveal a predominant role of this loop in maintaining a natural epiallele .
Epialleles , identified in different organisms and predominantly in plants ( reviewed by [1] ) , are gene variants based on epigenetic marks stably transmitted between generations . Most of the plant epialleles described so far depend on DNA methylation of cytosines , an epigenetic mark influencing the way genes are transcribed ( reviewed by [2] ) . Few reports depict plant natural epialleles associated with phenotypes or agronomical traits . For instance , Colourless non-ripening is a tomato natural epivariant corresponding to a hypermethylated version of an SBP-box gene and resulting in the silencing of the gene . Ripening is inhibited in plants carrying the Cnr epimutation , and consequently fruits are colorless [3] . Similarly , the peloric phenotype in Linaria vulgaris is due to methylation variations [4] . In melon , spreading of DNA methylation from a transposon influences the transcription of the CmWIP1 gene that controls sex determination [5] . Several epialleles were also identified in rice , all of them presenting severe morphological phenotypes: Epi-d1 [6] , Epi-df [7] and Epi-rav6 [8] . In natural populations of Arabidopsis , epivariants of QQS [9] or PAI [10] were similarly described , although no specific phenotypes associated with the epiallelic version of the gene were discovered , beyond the change in transcription . Other epivariants arose only in mutants with drastically modified epigenomes such as deficient in dna methylation1 ( ddm1 ) encoding a chromatin-remodelling factor [11] . In our previous study [12] , we demonstrated that an incompatibility between two Arabidopsis accessions is mediated through natural DNA methylation variation . In the progeny of a cross between the reference accession Col-0 and an accession from Tajikistan called Shahdara ( Sha ) , a particular allelic combination at two loci , localised on chromosome 4 and 5 , respectively , is counter-selected and very rare . In Sha , both loci carry duplicated FOLT genes encoding transporters of folate essential for fertility . FOLT1 , localised on chromosome 5 , is silenced by DNA methylation , while the chromosome 4 locus comprises a complex genomic rearrangement including functional and truncated copies of a paralogue , FOLT2 . Col-0 contains only an unmethylated version of FOLT1 at chromosome 5 . Consequently , after a cross between Col-0 and Sha , plants inheriting only the methylated FOLT1 epiallele from Sha are mostly sterile because the function of FOLT is missing . Since we discovered small RNAs ( sRNAs ) targeting the FOLT genes in Sha , we reasoned that FOLT1 is silenced by RNA-directed DNA methylation ( RdDM ) . We also found that unmethylated FOLT1 alleles can be converted in trans when FOLT2 copies are present at the other locus [12] . Different key steps , involving two plant-specific polymerases , define a feed-forward loop that controls the RdDM pathway in Arabidopsis ( recently reviewed by [13] ) : the RNA polymerase IV ( PolIV ) is first recruited at methylated DNA and transcribes them to single-stranded RNA . The complementary strand of this transcript is then rapidly synthesized by the RNA-DEPENDENT RNA POLYMERASE 2 ( RDR2 ) , cleaved by DICER-LIKE 3 ( DCL3 ) into sRNAs of 24-nt that are guiding ARGONAUTE 4/6 ( AGO4/6 ) to regions transcribed by another RNA polymerase ( PolV ) , finally attracting DNA methyltransferases . In addition to the RdDM , other pathways are involved in maintaining the methylation . In plants , methylation occurs at cytosines in CG , CHG and CHH ( where H = A , T , C ) contexts , and the different proteins controlling these pathways are now well characterized , particularly in Arabidopsis . DNA METHYLTRANSFERASE 1 ( MET1 ) maintains CG methylation , while CHROMOMETHYLASE 2 ( CMT2 ) and CMT3 are DNA methyltransferases responsible for CHH and CHG maintenance , respectively . Methylation on lysine 9 of histone H3 ( H3K9me ) and non-CG DNA methylation are tightly correlated . CMT2 and CMT3 are recruited to regions enriched in H3K9me [14 , 15] and in a reciprocal way , H3K9 histone di-methyltransferases , predominantly KRYPTONITE ( also known as SUVH4 and hereafter called KYP ) , bind CHG-methylated cytosines through their SRA domains [16] to methylate nearby histones . Thus , CMTs and KYP participate in a self-reinforcing loop between DNA and histone methylation , essential to silence transposons and repeated sequences , but deleterious to genes [17 , 18] . Recent data point toward the contribution of CMT2 in epigenetic variation of natural Arabidopsis populations [19 , 20] . Here , we describe the molecular mechanism that underlies a new allelic incompatibly identified in natural populations of Arabidopsis , which is based on the silencing of a gene homologous to the yeast transfer RNA ADENOSINE DEAMINASE 3 ( TAD3 ) , that is crucial for transfer RNA ( tRNA ) editing [21] . We identified spontaneous revertants and moreover determined that unmethylated alleles could be converted in hybrids , associated with a complete change in the population of sRNAs targeting this region . Finally , we provide evidence that only CMT3 and KYP are essential to maintain the epiallele over generations .
A recombinant inbred lines ( RIL ) population was generated from a cross between two accessions of Arabidopsis thaliana , Nok-1 and Col-0 ( http://publiclines . versailles . inra . fr/ ) . In this population comprising 222 genotyped individual lines , linkage disequilibrium ( LD ) analyses revealed two physically unlinked loci segregating dependently from each other: a locus at ~14–15 Mb near the centromere of chromosome 1 appears as in significant LD with a locus at ~ 8 . 5 Mb on chromosome 5 . Indeed , one of the four homozygous allelic combinations expected from the segregation of two independent loci—the combination of genotypes Col-0 at the chromosome 1 locus and Nok-1 at the chromosome 5 locus , named “incompatible”—is missing in the RIL population . A similar allelic incompatibility involving colocalizing loci was also observed in a different RIL set obtained by crossing Col-0 and Est-1 , as documented previously [22] . We fine-mapped the two loci involved in the allelic incompatibility using recombinant plants in the progenies of Nok-1 x Col-0 RILs that were still heterozygous for either of the two loci and fixed for the other one ( S1 Fig ) . The two incompatible loci were restricted , respectively , to 1 . 7 Mb near the centromere of chromosome 1 ( S1A Fig ) and to 13 kb on chromosome 5 ( S1B Fig ) . In the reference accession Col-0 , the chromosome 5 interval contains four genes ( Fig 1A ) and shares no sequence homologies with the second interval localized on chromosome 1 . We isolated T-DNA insertion lines in a Col-0 background for each of these four genes: we recovered homozygous plants for all genes , except when the T-DNA was inserted in the coding region of AT5G24670 in the mutant line GABI_141G12 ( Fig 1A ) . Moreover , in this line , plants heterozygous for the T-DNA insert exhibited partial seed abortion , with embryos arresting before the globular stage ( Fig 1B and [23] ) . The same phenotype was observed in siliques of three different RILs homozygous Col-0 at chromosome 1 and heterozygous Nok-1/Col-0 at chromosome 5 ( Fig 1B and 1C ) . Interestingly , we recovered homozygous plants when T-DNAs were inserted in the 5’-UTR region of AT5G24670 ( Fig 1A ) , implying that the disruption of TER2 , producing a non-coding RNA associated with the telomerase [24] , is not embryo-lethal , contrarily to AT5G24670 . These results make AT5G24670 , which encodes a protein homologous to the tRNA ADENOSINE DEAMINASE 3 ( TAD3 ) essential in yeast [21] , an excellent candidate gene to be responsible for the incompatibility between Nok-1/Est-1 and Col-0 . To understand whether Nok-1 , Col-0 and Est-1 were carrying different genetic variants of TAD3 , we amplified the whole gene plus 1 . 4 kb before the ATG in Nok-1 , Est-1 and Col-0 ( S2A Fig ) . Sequencing this unique amplicon of ≈3 . 7 kb revealed several differences between Nok-1 or Est-1 and the Col-0 genomic sequence publicly available . In particular , Nok-1 has ( 1 ) an insertion of three Ts and a deletion of TCTTCT within the promoter sequence , ( 2 ) one SNP ( A>T ) in the 5’-UTR , ( 3 ) two SNPs changing two amino acids in the gene body ( Fig 2A ) . We also found two other SNPs localized in an intron ( S2B Fig ) . The SNP localized in the 5’-UTR was the only polymorphism common to both Est-1 and Nok-1 . The TCTTCT deletion was mapped at chromosome 5 during the fine-mapping process , therefore , this copy , hereafter called TAD3-1 , is located on chromosome 5 in Nok-1 and Est-1 as in Col-0 ( S2B Fig ) . When we amplified TAD3 on genomic DNA with primers anchored within the coding sequence , we detected different copies in both Nok-1 and Est-1 , in contrast to Col-0 . By cloning and sequencing the corresponding PCR products , we found other copies in Nok-1 and Est-1 that present a number of polymorphisms with respect to TAD3-1 , including a 17 bp deletion within an intron ( Fig 2A , red star , and Fig 2B ) . Using primers surrounding this deletion , we maped these extra copies in the Nok-1 x Col-0 RILs and showed that they are located at the incompatible locus on chromosome 1 ( S1 Table ) . We designed new primers ( named TAD3K1_F/R; Fig 2B ) overlapping this deletion to specifically amplify the TAD3 genes localized on chromosome 1 ( Fig 2C ) , revealing two other copies carrying SNPs specific for chromosome 1 ( S1 Text ) . Therefore , Nok-1 and Est-1 are carrying at least two copies of TAD3 at chromosome 1 ( named TAD3-2 and TAD3-3 ) and one at chromosome 5 . All of them are polymorphic between them and compared to the unique TAD3-1 gene of Col-0 . We first analyzed the expression of the genes using primers anchored in the coding region of all copies: two mRNAs were detected in both Nok-1 and Est-1 in contrast to Col-0 ( S3A Fig ) . The two TAD3 cDNAs amplified in Nok-1 and Est-1 were isolated and sequenced . Compared to Nok-1 and Est-1 TAD3-1 sequences of chromosome 5 , both are carrying seven SNPs ( S3B Fig ) , identical to those identified on the TAD3-2 genomic sequence mapped on chromosome 1 ( S1 Text ) , indicating that the corresponding mRNAs are transcribed from this paralog . We designed specific primers ( S4 Fig ) to study the expression of TAD3-1 in Nok-1 , Est-1 and Col-0 . In Col-0 , TAD3-1 is transcribed in two isoforms ( AT5G24670 . 1 and . 2 ) , differing only by their UTRs ( Fig 2A ) . After RT , we amplified a region common to the two transcripts in both Col-0 and Nok-1 ( PCR#3 ) , and we found that the Nok-1 TAD3-1 transcripts were expressed at very low levels compared to Col-0 ( S5 Fig ) . qRT-PCR analyses using the same primers confirmed that TAD3-1 is about 30 times more expressed in Col-0 than in Nok-1 and Est-1 ( see Col-0 , Nok-1 and Est-1 controls in Fig 3C ) . Thus , in Nok-1 and Est-1 , TAD3-1 is expressed at very low levels , suggesting that their functional copy is localized on chromosome 1 . Our data point toward a duplication carrying at least two extra copies of TAD3 on chromosome 1 in both Nok-1 and Est-1 . In these two accessions , one chromosome 1 copy is transcribed whereas TAD3-1 is not . Consequently , seeds that are Col-0 at the chromosome 1 locus and Nok-1 at the chromosome 5 locus have no functional copy of this essential gene and then abort . Altogether , these results indicate that TAD3 is the gene involved in this allelic incompatibility between the Arabidopsis Nok-1/Est-1 and Col-0 accessions . To determine why TAD3-1 levels of expression are different from Col-0 in Nok-1 and Est-1 , we monitored its levels of methylation in these accessions . The epigenome public data [25] indicate that , in contrast to Col-0 , the region of TAD3-1 is heavily methylated in both Nok-1 and Est-1 . PCR amplification of the TAD3-1 5’-UTR after digestion with the methylation sensitive endonuclease McrBC confirmed that this region is methylated in both Nok-1 and Est-1 but not in Col-0 ( Fig 3A , PCR#2 and #3 ) . To confirm that expression of TAD3 and level of methylation are related , we grew the parental accessions on medium containing 5-aza-2’deoxycytidine , which inhibits cytosine methylation . The treatment activated the expression of TAD3-1 in both Nok-1 and Est-1 ( Fig 3B ) to levels similar to those of Col-0 ( Fig 3C ) . TAD3-1 is thus methylated and silenced in both Nok-1 and Est-1 , contrarily to Col-0 and this epiallele was hereafter referred as tad3-1 . Using the short indel of 7 bp that differentiates TAD3 copies from chromosomes 1 and 5 ( S1 Text ) , we designed primers specific for the 5’ region of the chromosome 1 TAD3 copies ( S6A Fig ) . No PCR products were obtained when the DNAs were digested with McrBC prior to the amplification , indicating that the chromosome 1 copies are methylated ( S6B Fig ) . In the course of this study , we unexpectedly identified , in the progeny ( n = 60 ) of one F8 plant fixed Col-0 at chromosome 1 and heterozygous Nok-1/Col-0 at chromosome 5 , 21% of plants carrying the incompatible allelic combination . We hypothesized that this F8 plant or one of its ancestors was already carrying a reverting tad3-1 epiallele since revertants were not detected within the 7 , 872 F7 plants genotyped to map the TAD3 loci . To determine the molecular basis of this spontaneous reversion , we examined the methylation pattern of TAD3-1 in the F9 progeny of this plant . Different parts of the gene were amplified from three plants fixed Col-0 at chromosome 1 and Nok-1 at chromosome 5 and three sibling plants fixed Col-0 at both loci after digestion with McrBC ( Fig 4A ) . In all plants carrying the incompatible combination ( i . e . K1ColK5Nok ) , the promoter and the 5’-UTR were not methylated anymore , contrarily to Nok-1 ( Fig 4A , PCR#1–3 ) , while the second part of the gene and the 3’-UTRs remained methylated ( Fig 4A , PCR#4–6 ) . By sequencing the promoter region and the last part of the gene after bisulfite conversion , we confirmed that the promoter was not methylated and we determined that cytosines were methylated in all contexts in the gene body ( Fig 4B and S7 Fig ) . In these plants , qRT-PCR analyses revealed that TAD3-1 is expressed at levels intermediate between the Col-0 and Nok-1 alleles ( Fig 4C ) . We concluded that the silencing of the methylated Nok-1 tad3-1 epiallele is reversible , although this seems to be a rare event . In revertants , the promoter and the first part of the gene , including the 5’-UTR , are demethylated contrarily to the rest of the gene , representing very few nucleosomes . Thus , it is likely that different TAD3-1 epiallelic versions are present in Arabidopsis and that the status of this epiallele can change within few generations . We genotyped the progeny of two F9 plants fixed Col-0 at chromosome 1 and heterozygous at chromosome 5 ( Fig 4D ) . 19% of the F10 plants had the incompatible allelic combination , indicating that the reverting tad3-1 epiallele is stably transmitted to the next generations . To further understand how the TAD3-1 methylation is established , we crossed Col-0 and Nok-1 and we determined the methylation states of TAD3-1 in hybrids by sequencing their genomic DNA after bisulfite conversion . We focused our analysis on the promoter , a region appearing to be essential for the silencing of TAD3-1 , as shown above . To distinguish the Col-0 and Nok-1 TAD3-1 alleles , we sequenced one polymorphic region ( S8 Fig ) in the promoter containing three additional Ts in Nok-1 compared to Col-0 ( Fig 2A ) . In F1 hybrid plants , the Nok-1 allele of the TAD3-1 promoter is methylated in the three cytosine contexts at levels comparable to those of Nok-1 ( Fig 5A ) . However , we showed that the Col-0 allele gained methylation in both CG ( 5 to 6 times more compared to the Col-0 parent ) and CHG ( 2 to 4 times more compared to the Col-0 parent ) contexts ( Fig 5A and S9 Fig ) . This result indicates that the Col-0 allele has been de novo methylated in hybrid plants . F1s obtained from both Nok-1 x Col-0 and Col-0 x Nok-1 reciprocal crosses gave similar results ( S9 Fig ) . We then examined whether the newly acquired methylation of the Col-0 allele in F1s could interfere with the expression of TAD3-1 . Pyrosequencing analyses using the SNP A>T in the 5’-UTR region ( Fig 2A ) confirmed that the TAD3-1 Nok-1 allele is not expressed in F1s ( Table 1 and S10 Fig ) . Additionally , expression analyses by qRT-PCR revealed that the Col-0 allele is expressed in leaves of F1s at levels intermediate between the parents ( Fig 5B ) , which is the expected expression level when only one allele is transcribed . These results imply that the methylation gained in one generation by the Col-0 allele has a minor effect on the expression of TAD3 . To determine whether an RdDM process mediated by sRNAs targets TAD3 , we profiled , by deep sequencing , the sRNA populations of Col-0 , Nok-1 and their F1 hybrid , mapping them to the Col-0 reference genome ( Fig 6 and S11 Fig ) . In Col-0 , we only detected 23/24-nt sRNAs corresponding to transposons localized upstream of the TAD3-1 gene . In both Nok-1 and the F1 , the number of 24-nt sRNAs matching these transposons increased by 2 . 5 times compared to Col-0 . In addition , sRNAs , mostly 23/24-nt , mapping to the region localized between the transposons and the 5’-UTR of TAD3-1 were detected in Nok-1 and the F1 . In comparison , we found a limited number of sRNAs dispatched along the TAD3-1 sequence in both the F1 and Nok-1 . Thus , in both the F1 and Nok-1 , 23/24-nt sRNAs matching the regions upstream of TAD3-1 are more abundant than in Col-0 . We hypothesize that the sRNAs , which cover this entire region in Nok-1 and the F1 , potentially initiate the methylation through the RdDM pathway . It would be intriguing to examine in future studies whether the precursors of these sRNAs originate in trans or in cis . To decipher the molecular mechanisms involved in maintaining the tad3-1 epiallele , we determined whether the incompatibility depends on particular epigenetic pathways . To this end , we crossed Nok-1 with several mutants ( in a Col-0 background ) , which are impacted in the maintenance or the establishment of different epigenetic marks . The resulting hybrids were then back-crossed with the same mutants to obtain plants that were fixed for the mutation and heterozygous Nok-1/Col-0 at both TAD3 loci ( S12 Fig ) . We then followed the segregation of both loci in the selfed progenies of these plants by genotyping . As a control , we genotyped 337 F2 plants from the Nok-1 x Col-0 F1 progeny and we confirmed that Col-0 and Nok-1 alleles segregated as expected , that is missing the incompatible allelic combination ( Col-0 at chromosome 1 and Nok-1 at chromosome 5; Table 2 ) . We obtained similar results for all mutants involved in RdDM pathways , namely mutants of the DNA-dependent RNA polymerases PolIV and PolV , AGO4 and RDR2 ( Table 2 ) . Comparable results were obtained in a met1-1 background ( Table 2 ) , implying that the tad3-1 epiallele is maintained independently of the RdDM or MET1 pathways . Nevertheless , we found plants with an 'incompatible' allelic combination in both cmt3 and kyp backgrounds ( Table 2 ) . To further determine the methylation patterns of TAD3-1 in this context , we extracted genomic DNAs from 'incompatible' plants obtained in the cmt3 background and we amplified different parts of the gene , after digestion with McrBC ( Fig 7A ) . The profiles of methylation obtained in a cmt3 mutant were similar to the ones obtained in Col-0 , confirming that CMT3 is necessary to maintain Nok-1 tad3-1 epiallele methylation . Sequencing the promoter regions after bisulfite conversion revealed that the non-CG methylation from the Nok-1 allele is missing in these cmt3 backgrounds ( Fig 7B ) , while the level of CG methylation was intermediate between Col-0 and Nok-1 ( Fig 7B and S9 Fig ) . In the progeny ( n = 128 ) of one cmt3 plant fixed Col-0 at chromosome 1 and heterozygous Nok-1/Col-0 at chromosome 5 , 23% of the plants were carrying the incompatible allelic combination ( Fig 7C ) , showing that CMT3 maintains tad3-1 over generations .
Transposable Elements ( TEs ) or repetitive sequences , like those localized upstream and downstream of the TAD3 gene in Col-0 ( Fig 6 ) , are prone to silencing and can influence the transcription of genes . First , the spreading of epigenetic marks from the TE toward neighbouring genes localized in cis can be deleterious for their transcription . Not surprisingly , there are many examples of plant epialleles whose silencing depends on an adjacent TE or repeat: Epi-1d [6] and Epi-rav6 [8] in rice , CmWIP1 in melon [5] , FLOWERINGWAGENINGEN ( FWA ) [29 , 30] and BONSAI ( BNS ) [31] in Arabidopsis . Second , sRNAs produced from a TE or repetitive sequences can also trigger the silencing of homologous regions in trans , including genes , through the RdDM pathway . In a previous study [12] , we identified two incompatible loci in a RIL population obtained by crossing Col-0 and Shahdara . We demonstrated that this incompatibility is based on a duplication and rearrangement of the FOLT gene . The Shahdara copy on chromosome 4 produces sRNAs ( detected by northern blot ) that cause a FOLT copy on chromosome 5 to be methylated and silenced in trans . In this case , sRNAs are abundant and homologous to the FOLT gene promoter and the first part of the coding region [12] . Therefore , it is likely that FOLT sRNAs are directly targeting the FOLT gene , triggering the methylation by RdDM . In the case of the TAD3 gene , we were unsuccessful in detecting , by northern blot analyses , sRNAs homologous to the coding region in Nok-1 , F1 and RILs . This was confirmed by the low level of sRNAs targeting TAD3 identified by sequencing the whole population of sRNAs in both Nok-1 and the F1 ( Fig 6 and S11 Fig ) . Furthermore , the 24-nt sRNAs homologous to TAD3 are dispatched along the 3 kb of the gene and neither the 5’-UTR nor the promoter region immediately adjacent to the gene are massively targeted ( S11 Fig ) . Therefore , the TAD3-1 coding region is probably not directly targeted by sRNAs in the F1 . On the opposite , a large amount of sRNAs found in Nok-1 and the F1 are matching the region corresponding to transposons localized upstream of TAD3-1 , going over the edges of the transposons ( S11 Fig ) . Therefore , we cannot exclude their importance in establishing a new epigenetic state near TAD3-1 in F1s , leading to the progressive methylation of the Col-0 allele ( Fig 5A ) . Indeed , several reports point toward drastic modifications of sRNA populations between hybrids and their parents , associated with changes in DNA methylation patterns [32–35] . Interestingly , regions localized in the vicinity of TEs seem to be particularity prone to changes in sRNA contents [36] and DNA methylation [35] in hybrids . Nevertheless , while sRNAs could initiate the spreading of methylation toward TAD3-1 in an initial step , none of the genes involved in 24-nt sRNA synthesis tested in our study ( Table 2 ) are important to maintain the tad3-1 epiallele . We conclude that sRNAs are not essential to maintain the silencing of TAD3-1 over generations but are potentially required in an early initiation step . Indeed , RdDM plays a critical role in controlling most of the DNA methylation interactions occurring in an Arabidopsis Col-0 x C24 hybrid [35] . In Arabidopsis , FWA and BNS are two examples of epialleles revealed and stabilized in hypomethylated ddm1 mutants [31 , 37] . FWA is an imprinted gene involved in flowering that is silenced in vegetative tissues , due to the methylation of a SINE-related sequence [29] . The lack of methylation results in expression of FWA and late flowering [30] and its maintenance depends on MET1 [38] . MET1 is , however , not involved in the maintenance of TAD3-1 methylation: first , the incompatible allelic combination is absent from the progeny of two individual met1 plants segregating the TAD3 loci ( Table 2 ) and second , plants are fertile when the promoter of TAD3-1 is partially methylated in the CG-context ( Fig 7B ) , maintained by MET1 . Thus FWA and TAD3 are maintained silenced via different pathways . Another possibility is that more generations are needed for tad3-1 to revert in the met1-1 background or that this requires a stronger met1 allele . In addition to fwa epialleles , the bns epiallele arose in hypomethylated mutant backgrounds due to changes of epigenetic patterns at nearby transposons . However , after several generations this resulted in an increasing DNA methylation at BNS , leading to its silencing [31] . Very similarly to the tad3-1 epiallele , bns maintenance depends on both CMT3 and KYP , but not on the RdDM machinery [39] . Indeed , we demonstrate that CMT3 is involved in maintaining the tad3-1 epiallele found in Nok-1 since we recovered incompatible plants in the cmt3 mutant background ( Table 2 ) . We also obtained the same results in a kyp background ( Table 2 ) , providing further evidence that both CMT3 and KYP , involved in the same self-reinforcing loop [15 , 16] , are key factors to maintain the tad3-1 epivariant between generations . Additionally , BNS is also hypermethylated when the function of INCREASE IN BONSAI METHYLATION 1 ( IBM1 ) , a histone demethylase removing H3K9 methylation , is compromised [17] and TAD3 is likewise targeted by IBM1 [40] . Recently , the molecular bases involved in maintaining Cnr through generations were identified in tomato , with a major role played by CMT3 [27] . In this natural epivariant of the LeSPL-CNR gene , the promoter is methylated and the gene is transcriptionally silenced , leading to the non-ripening phenotype observed . Silencing CMT3 in Cnr , but not MET1 , resulted in the almost complete rescue of fruit ripening , associated with a reduction in CHG methylation at eight positions in the promoter of LeSPL-CNR . Therefore , together with the tad3-1 epiallele , both Cnr and bns are epialleles maintained by the CMT3/KYP loop . An intriguing question that remains to be answered is to determine the molecular events that break this loop , allowing the promoter of TAD3-1 to be demethylated and transcribed again in revertants . Further studies are needed to clarify the role and the extent of this feed-forward loop between histone and DNA in maintaining natural epigenetic variation in plants .
A . thaliana RILs , HIFs and accessions were obtained from the Versailles Arabidopsis stock center ( http://publiclines . versailles . inra . fr/ ) . The following mutants were used: the GABI_141G12 T-DNA line [41] , met1-1 [42] , cmt3-11 ( SALK_148381 , [43] ) , ago4-2 [44] , suvh4/kyp ( SALK_069326 , [45] ) , rdr2-2 ( SALK_059661 , [46] ) , nrpd1a-4 ( pol IV; SALK_083051 , [47] ) and nrpe1-11 ( pol V; SALK_029919 , [48] ) . Col-0 sequences are corresponding to the TAIR10 version of the reference genome . To genotype plants , we used two molecular markers , one located within the 5'-UTR of TAD3-1 ( MSAT5 . 08448 ) , amplifying 142 bp in Col-0 and 138 bp in Nok-1 , and another one that co-segregates with the interval of 14–15 Mb at chromosome 1 ( MSAT1 . 15597; S1 Fig ) , and amplifying 128 bp in Col-0 and less in Nok1 ( ≈120 bp ) . The GeneRuler DNA Ladder Mix ( Ref SM0331 , Thermo ) is the DNA ladder used for all figures . Genomic DNA ( 300 ng ) was digested for 8 h at 37°C with the McrBC enzyme ( New England Biolabs ) , the same amount of undigested genomic DNA was used as control . The methylation of a region was assessed by PCR amplification ( 35 cycles ) using 20 ng of digested or undigested genomic DNA . The primers are listed in S3 Table . For each sample , 1 to 2 μg of genomic DNAs were extracted from leaves , using the NucleoSpin Plant II kit ( Macherey-Nagel ) . DNAs were treated with bisulfite using the EpiTect Bisulfite Kit ( Qiagen ) . Treated DNAs were amplified using primers listed in S3 Table . PCR fragments were then cloned in pTOPO ( Life Technologies ) and sequenced individually . Results were analyzed with the Kismeth tool [49] . Total RNAs were isolated from the aerial parts of 21 day-old seedlings grown in vitro using the RNeasy Plant Mini kit ( Qiagen ) followed by a DNAse treatment ( Fermentas ) . RT-PCR was performed on 500 ng of total RNAs with the M-MLV reverse transcriptase ( Fermentas ) and cDNAs were diluted 10 times . 5 μl were used for qRT-PCRs using a CFX96 real-time PCR machine ( BioRad ) with a SYBR solution ( Eurogentec ) using primers listed in S3 Table . Expression levels were normalized against the Arabidopsis UBC21 gene ( AT5G25760 ) . Analyses were done as described before [50] . Briefly , pyrosequencing ( PyroMark Q24; Qiagen ) was used to estimate relative allele-specific expression in the F1 . RNAs from leaves or flowers of Col-0 , Nok-1 and seven different F1 plants were extracted to prepare cDNAs as described above . The region amplified corresponds to PCR#3 ( Fig 2A ) and is specific for TAD3-1 ( S4 Fig ) . Pyrosequencing reactions were performed using the SNP ( A>T ) localised in the 5’-UTR of TAD3-1 . F1 genomic DNA was used for technical control to normalize the ratios against possible pyrosequencing or PCR biases . The sequencing primer is listed in S3 Table . sRNAs were extracted and sequenced as previously described [51] . Reads were first trimmed to discard reads shorter than 15 nt and to remove the adapter ( AGATCGGAAGAGCACACGTCT ) . Clean reads were then aligned to the Col-0 genome ( TAIR10 . 30 ) with bowtie allowing one mismatch and a maximum of 50 multi-mappings per read . Data were plotted with the viRome R package [52] . The reads were normalized to reads per million ( RPM ) of mapped reads . Statistics of the bioinformatics analyses are presented in S2 Table . | Epialleles are gene variants based on epigenetic marks stably transmitted between generations . Most of the known epialleles existing in the wild were described in plant populations but very few are associated with phenotypes or agronomical traits . In this study , we isolated a new natural epiallele resulting from the silencing of a RNA editing gene essential for plants . We demonstrated that an incompatibility between two Arabidopsis strains depending on this epiallele , is based on DNA methylation of cytosines , an epigenetic mark influencing gene function . In F1 hybrids , obtained by crossing the incompatible parental lines , unmethylated versions of the allele can be converted to methylated ones . The epiallele can also spontaneously revert in very rare cases , within the population . The methylation status of this epiallele can therefore potentially change within the population and is maintained in a metastable state . Indeed , two enzymes promoting histone or DNA methylation , respectively , and acting in loop , are involved in maintaining the epiallele in natural populations , over generations . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"chromosome",
"5",
"biotechnology",
"brassica",
"chromosome",
"1",
"plant",
"science",
"model",
"organisms",
"experimental",
"organism",
"systems",
"plant",
"genomics",
"epigenetics",
"dna",
"molecular",
"biology",
"techniques",
"plants",
"dna",
"methylation",
"chromati... | 2017 | An Arabidopsis Natural Epiallele Maintained by a Feed-Forward Silencing Loop between Histone and DNA |
In Tunisia , almost 77% of clinically and bacteriologically diagnosed cases of extrapulmonary tuberculosis ( EPTB ) are zoonotic TB , caused by M . bovis . Although several studies have analyzed bovine TB in cattle in Tunisia , no study has evaluated the risk of transmission to humans in such an endemic country . We aimed to study the genetic diversity of M . bovis human isolates , to ascertain the causes of human EPTB infection by M . bovis and to investigate the distribution and population structure of this species in Tunisia . A total of 110 M . bovis isolates taken from patients with confirmed EPTB were characterized by spoligotyping and MIRU-VNTR typing methods . Among the 15 spoligotypes detected in our study , 6 ( SB0120 , SB0121 , SB2025 , SB1200 , SB1003 and SB0134 ) were the most prevalent ( 83 . 5% ) of which SB0120 , SB0121 and SB2025 were the most prevailing . MIRU-VNTR typing method showed a high genotypic and genetic diversity . The genetic differentiation based on MIRU-VNTR was significant between populations from South East ( Tataouine , Medenine ) and Central West ( Gafsa , Sidi Bouzid , Kasserine ) regions . Of note , 13/15 ( 86 . 7% ) spoligotypes detected in our study were previously identified in cattle in Tunisia with different frequencies suggesting a peculiar ability of some genotypes to infect humans . Using combined spoligotyping and MIRU-VNTR method , a high clustering rate of 43 . 9% was obtained . Our results underlined that human EPTB due to M . bovis was more commonly found in female gender and in young patients . Most of our patients , 66 . 4% ( 73/110 ) were raw milk or derivatives consumers , whereas 30 . 9% ( 34/110 ) patients would have contracted EPTB through contact with livestock . The findings suggest that the transmission of Zoonotic TB caused by M . bovis to humans mainly occurred by oral route through raw milk or derivatives . Our study showed the urgent need of a better veterinary control with the implementation of effective and comprehensive strategies in order to reach a good protection of animals as well as human health .
In developing countries , Mycobacterium bovis , the agent of bovine tuberculosis ( bTB ) represents a threat for livestock and human health [1 , 2] . Humans can acquire M . bovis either by aerogenous route due to a close contact with infected animals or by consuming unpasteurized dairy products [1] . According to the World Health Organization ( WHO ) , among the estimated 10 million new cases of tuberculosis ( TB ) in 2017 , almost 20% were extrapulmonary tuberculosis ( EPTB ) [3] . Furthermore , in the region with high bTB prevalence , M . bovis would frequently be associated with extrapulmonary disease in humans [1 , 3 , 4] . In developing countries , the exact percentage of M . bovis in human TB cases is generally underestimated , since the species diagnosis is rarely performed , especially in EPTB cases [1 , 3 , 4] . Human TB ( 30/100 . 000 ) is endemic in Tunisia with high frequency of EPTB cases ( 56 . 9% ) despite the low rate of HIV infection [5] . Recently , Ghariani et al have reported high prevalence of M . bovis ( 76% ) in lymphadenitis TB ( LTB ) cases in North Tunisia by conventional methods [6] . In South Tunisia , Siala et al have detected M . bovis by qPCR in almost 77% of extrapulmonary samples taken from patients with EPTB [7] . In addition , LTB was estimated to be 50% of all EPTB cases of which cervical localization was the most frequent ( 70% to 90% ) [5 , 8] . Djemal et al , have demonstrated that M . bovis is still spreading in Tunisia causing bTB with a frequency of 64 . 4% in 2014–2015 , whereas Lamine-khmiri et al , reported a frequency of 35% in 2013 [9 , 10] . Consequently , the risk of contracting EPTB is still high since the control measures for herd , livestock and unpasteurized dairy products as well as milk are consistently declining [1 , 7] . Nevertheless , no study evaluated the risk of transmission of zoonotic TB due to M . bovis to humans in such an endemic country . Furthermore , the genetic background of M . bovis causing human EPTB is still not known in Tunisia . The availability of such information is critically important in order to identify the source , the route of infection and thus to control the disease . The main goal of this work was to genetically characterize the M . bovis population involved in human EPTB in order to improve our understanding of the bTB epidemiology and M . bovis spread in Tunisia . Spoligotyping and mycobacterial interspersed repetitive units-variable number of tandem repeats ( MIRU-VNTR ) analysis were used , since the combination of these two genetic markers is known to be a powerful tool for the molecular epidemiology study of M . bovis [11 , 12] . We finally compared M . bovis data obtained from humans with data from Tunisian cattle [9 , 10] .
Approval for usage of M . bovis isolates , demographic , epidemiological and clinical data for our study was obtained by the Ethics Committee of Hedi Chaker Hospital-Sfax-Tunisia The study was prospectively conducted over a 32-month period ( January 2013 to September 2015 ) by the regional hygiene care mycobacteriology laboratory of the Department of Microbiology of the Hedi-Chaker University Hospital in Sfax ( South Tunisia ) . During this period , all patients with a positive diagnosis of extrapulmonary tuberculosis ( EPTB ) , microbiologically confirmed and attributed to M . bovis , were included in the study . All data were anonymized . The laboratory provided routine mycobacterial diagnostic tests for TB specimens obtained during consultation or hospitalization of suspected TB patients in Hedi-Chaker University Hospital . In total , 110 patients were included in this study . One hundred and six patients were from 7 Tunisian governorates localized in East Central Tunisia ( Sfax ( n = 23 ) , Gabes ( n = 21 ) ) , West Central Tunisia ( Gafsa ( n = 13 ) , Sidi bouzid ( n = 7 ) , Kasserine ( n = 4 ) ) and South East ( Tataouine ( n = 23 ) , Medenine ( n = 15 ) ) of Southern Tunisia regions . Four patients were from Libya . The other demographic and clinical characteristics that were collected include sex , age , clinical site of TB , patient origin , past history of TB , raw milk consumption , contact with livestock and BCG vaccination status ( Table 1 and S1 Table ) . Specimens from various extrapulmonary sites were collected: lymph nodes ( n = 100 ) , osteoarticular , intestinal , meningeal , peritoneal and pleural samples ( n = 10 ) . Solid specimens were homogenized with pestle and mortar . All specimens were directly inoculated on Lowenstein–Jensen ( LJ ) and Coletsos slants . After decontamination using the standard N-acetyl- L-Cysteine sodium hydroxide method , the samples were centrifuged at 3000 rpm for 15 min . The pellets were then inoculated on Lowenstein–Jensen ( LJ ) , Coletsos media , and in liquid media Mycobacterial Growth Indicator Tube 960 ( MGIT 960 , Becton Dickinson Biosciences , Sparks , MD , USA ) . The tubes were incubated at 37°C for 42 days and the cultures were incubated on solid media at 37°C for up to 12 weeks . The isolates were identified as mycobacteria and M . bovis species by Ziehl-Neelsen ( ZN ) staining for Acid Fast Bacilli ( AFB ) , morphological and biochemical criteria including niacin test , nitrate reductase test , and susceptibility for certain inhibitors such as thiophene-2-carboxylic acid hydrazide ( TCH ) , pyrazinamid and p-nitrobenzoic acid ( PNB ) [13] . Strain identification was confirmed by the commercial multiplex line probe assay , GenoTypeMTBC ( Hain Lifescience GmbH , Nehren , Germany ) [14] . DNA extraction from mycobacteria isolates and high-throughput spoligotyping on Luminex 200 ( Luminex Corp . , TX ) were performed as previously described [15 , 16 , 17 , 18] . The obtained data were compared with those of the international database ( www . mbovis . org ) . The isolates were also genotyped by PCR amplification of 8 Mycobacterial Interspersed Repetitive Units-Variable Number Tandem Repeats ( MIRU-VNTR ) loci: ETR-A ( VNTR 2165 ) , ETR-B ( VNTR 2461 ) , QUB 11a ( VNTR 2163a ) , QUB 11b ( VNTR 2163b ) , QUB 26 ( VNTR 4052 ) , QUB 3232 ( VNTR 3232 ) , ETR-C ( VNTR 0577 ) and MIRU 4 , ETR-D ( VNTR 580 ) as described previously [19 , 20 , 21] . These eight loci were selected according to the recommendation of the European Union Reference Laboratory ( EURL ) [21] . The results were combined into 8-digit allelic profiles for each isolate [19] . The spoligotyping and MIRU-VNTR data obtained from M . bovis isolated from Tunisian cattle were used for comparison [9 , 10] . Several diversity indices , including the genotypic diversity ( Gd = the number of different genotypes divided by the total number of samples using the combination of MIRU-VNTR and Spoligotyping data ) , the allelic diversity per locus and the mean genetic diversity ( Hs ) were calculated . The population structure was explored by estimating the Fst ( index of genetic differentiation between samples ) value ( 0 = no differentiation and 1 = fixation of alternative alleles ) . The allelic diversity , the Hs and the Fst were calculated using F-STAT , version 2 . 9 . 3 using the 8 MIRU-VNTR loci data [22] . The discriminatory power of each locus and of the typing method was estimated using the Hunter-Gaston discriminatory Index ( HGDI ) according to the previously described formula [23] . Neighbor-joining tree was constructed based on Cavalli-Sforza and Edwards distance methods , using Populations 1 . 2 . 30 ( Olivier Langella , CNRS UPR9034 , France ) [24] and MEGA6 [25] software . The topology robustness was estimated by performing bootstrap analysis with 1000 replicates . Treedyn and Inkscape were used for tree visualization and annotation [26] . The clustering rate was calculated using the N-1 method formula as described previously [27] to estimate the proportion of TB potentially attributable to recent transmission . Univariate analysis was performed to test the association of risk factors with clustering ( proportion of isolates in clusters versus non-clustered ones ) . Statistical analyses ( chi-square , Fisher’s exact 2-tailed and student test ) were performed using SPSS 16 . 0 statistical software ( SPSS Inc . , Chicago , IL ) . Differences were considered significant at values of P ≤ 0 . 05 .
The sociodemographic and clinical data linked to the EPTB human isolates are shown in Table 1 and S1 Table . For some patients , the epidemiological data could not be collected as mentioned in the tables . LTB was the most common form ( n = 100/110 , ( 90 . 9% ) ) among the patients with M . bovis culture confirmed-EPTB , followed by osteoarticular ( n = 4 ) , pleural ( n = 3 ) , meningeal ( n = 1 ) , peritoneal ( n = 1 ) and intestinal ( n = 1 ) forms . Two out of the 110 M . bovis isolates under study were resistant to streptomycin and one to rifampicin and ethambutol . The age of the patients ranged between 1 and 72 years , with a mean age of 30 . 36 ± 17 . 07 years ( Table 1 ) . The majority of the patients ranged between 15–59 years ( n = 84 , ( 76 . 4% ) ) with 49% of cases between 20 and 40 years old . The male/female ratio in the study population was 0 . 43 . Age distribution differed significantly between men and women ( mean age 24 . 21 ±20 . 03 years versus 33 years ±15 . 02; P = 0 . 013 , respectively ) . The majority of isolates were collected from Tunisian patients who originated from the Central Eastern ( 40% ) and South Eastern Tunisia ( 34 . 5% ) , with the highest percentages in Tataouine ( n = 23 , 21% ) , Sfax ( n = 23 , 21% ) and Gabes ( n = 21 , 19 . 1% ) governorates . Four patients ( 3 . 6% ) were from Libya ( see Table 1 , and S1 Table ) . A high proportion of patients ( n = 73/110 , ( 66 . 4% ) ) used to consume raw milk or derivatives with a significant increase observed in patients between 15–59 years ( p = 0 . 04 ) . TB history was recorded only for 6 patients ( 5 . 4% ) . None of them was older than 15 years ( p = 0 . 02 ) . About half of the patients received BCG vaccination and 34 ( 30 . 9% ) had contact with livestock or other animals ( Table 1 ) . Patients in contact with livestock ( 14/31 ( 45 . 2% ) ) and consuming raw milk ( 21/73 ( 28 . 8% ) ) were significantly more represented in the Central West region including Gafsa , Sidi Bouzid and Kasserine , than in the other regions ( p = 0 . 004 and p = 0 . 039 , respectively ) ( Table 1 ) . Of note , 100% of patients from the Central West region had LTB . Isolates from Libyan patients were excluded in this latter statistical analysis . Using the combined analysis of Spoligotyping and MIRU-VNTR methods , the 106 isolates revealed 61 different genotypes with a high level of discrimination ( HGDI = 0 . 980 ) . A total of 65 isolates ( 61 . 3% ) were grouped in 20 clusters and the other 41 isolates ( 38 . 7% ) had a unique pattern giving a CR of 42 . 45% ( Fig 2 , S1 Table ) . The cluster size ranged from 2 to 10 isolates per cluster . Out of these 20 clusters , four ( C5 , C6 , C7 , C9 with n = 4 , 3 , 2 , 2 , respectively ) belonged to M . bovis SB0120 ( BCG like ) isolates . SB0121 was found in 5 clusters ( C13 , C14 , C15 , C16 , C17 with n = 3 , 3 , 3 , 2 , 4 , respectively ) . Two clusters ( C18 , C19 ) including SB2025 spoligotype were made up of 3 and 10 isolates , respectively . Two clusters ( C10 , C11 ) of SB1200 spoligotype included 2 and 6 isolates , respectively . Two other clusters ( C8 , C12 ) were characterized by the spoligotypes SB0871 and SB1003 including three and five isolates , respectively . For the isolates from Libyan patients , three were not clustered and one belonged to SB1003 which was clustered ( C12 ) with four Tunisian isolates ( two from Sfax , one from Sidi Bouzid and one from Kasserine ) . SB0134 was represented only in one cluster ( C4 ) including two isolates . Three clusters ( C1 , C2 , C3 ) included two isolates with the spoligotypes SB1062 , SB1148 , and SB0133 , respectively . Only one cluster ( C20 ) including two isolates belonged to M . caprae ( SB0866 ) . The two other M . caprae ( SB0866 , SB2024 ) were not clustered . In Tunisia , 13/15 ( 86 . 7% ) spoligotypes detected in our study were reported in cattle samples collected from November 2014 to April 2015 by Djemal et al [9] . From these 13 spoligotypes , seven ( SB0120 , SB0121 , SB2025 , SB1200 , SB1003 , SB0134 and SB2024 ) were previously identified by Lamine-khmiri et al in animal samples collected between 1991–2012 [10] and two ( SB1003 and SB1200 ) in milk samples [28] . The six most detected M . bovis spoligotypes in our human isolates were also reported in cattle in Tunisia [9 , 10] with different frequencies ( S2 Table ) . The SB0120 spoligotype found in 29 out of 109 ( 26 . 6% ) Tunisian human isolates , was detected with a higher frequency in the cattle in the south of Tunisia as demonstrated by Djemal et al ( 36 . 4% ) and Lamine-khmiri et al ( 37 . 1% ) [9 , 10] ( S2 Table ) . SB0121 , SB2025 , SB1200 were more prevalent in our study ( 21/110 ( 19 . 3% ) , 20/110 ( 19% ) and 8/110 ( 7 . 3% ) , respectively ) compared to the studies of Djemal et al [9] and Lamine-khmiri et al [10] ( S2 Table ) . SB1200 was detected in cattle but also in one milk sample as previously reported in Tunisia [9 , 10 , 28] . SB1003 was detected in milk samples and in the cattle in Tunisia [9 , 10 , 28] . SB0134 identified as the second predominant spoligotype in cattle in Tunisia ( 11 . 4% [9] and 20% [10] ) , was detected only in 5 . 5% of our human isolates ( S2 Table ) . Regarding M . caprae isolates , SB2024 was reported in human and cattle Tunisian isolates [9 , 10] but SB0866 was not found in cattle in Tunisia ( S2 Table ) . It is worth noting that the spoligotypes observed in our isolates and shared with cattle ( S2 Table ) , were mainly associated with raw milk-consuming patients and in contact with livestock . Indeed , 23/29 ( 79 . 3% ) and 14/21 ( 66 . 7% ) of the SB0120 and SB0121 , respectively , were taken from patients who consumed raw milk and/or had contact with livestock . SB2025 was detected in 13/20 ( 65% ) of raw milk-consuming patients of which one was in close contact with livestock . SB1200 and SB0134 , were also detected in raw milk-consumers ( 5/8 ( 62 . 5% ) , 3/6 ( 50% ) , respectively ) or in close contact to livestock ( 3/8 ( 37 . 5% ) , 3/6 ( 50% ) , respectively ) . For M . caprae isolates , SB0866 were detected in raw milk-consuming patients ( 3/3 ( 100% ) ) of whom 2 were in close contact with goats . SB2024 was found in patient in close contact with livestock ( 1/1 ( 100% ) ) . The comparison of the MIRU-VNTR genotypes based on the 6 common loci ( ETR-A , ETR-B , QUB 11a , QUB 11b , QUB 3232 and MIR U4 ( ETRD ) between human and cattle isolates from Tunisia showed a total of 14 common VNTR profiles ( CVP ) ( Table 5 , S2 Table ) . Among them , 13 CVP were recently identified in cattle from Tunisia by Djemal et al [9] . CVP 1 , 2 , 10 , 11 , 12 and 14 were also previously detected in Tunisian cattle isolates [10] ( Table 5 and S2 Table ) . It is worth noting that 13/14 ( 92 . 85% ) and 9/14 ( 64 . 3% ) CVP were detected in our isolates from patients who consumed raw milk ( 35/73 ( 47 . 9% ) ) or who were in close contact with livestock ( 12/34 ( 35 . 3% ) ) , respectively . The recent transmission rate among the 98 M . bovis isolates from Tunisia ( excluding M . caprae isolates ( n = 4 ) , isolates from Libyan patients ( n = 4 ) and isolates with missing genotyping data ( n = 4 ) ) was 43 . 9% based on the MIRU-VNTR patterns plus spoligotyping . The unique genotypes were detected in children ( 0–4 years ) , in old patients ( ≥ 60 years ) , in patients from Sidi bouzid , Kasserine and in patients with personal TB history and with no BCG vaccination ( S3 Table ) . The clustering was then analyzed in function of the demographic and clinical data ( S3 Table ) . The number of isolates in clusters was high and statistically significant in female patients ( 75 . 8% ) , in patients between 15 to 59 years ( 79% ) , in patients with urban lifestyle ( 72 . 6% ) , in patients living in Gabes ( 24 . 2% ) , Sidi Bouzid ( 8 . 1% ) and in Central Eastern Tunisia ( 43 . 5% ) ( S3/A Table ) . The univariate statistical analysis showed that the clustering was not affected by age , sex , area of residence , raw and/or unpasteurized milk consumption , contact with livestock , personal TB history and BCG vaccination status ( P >0 . 05 ) ( S3 Table ) . Considering the three main M . bovis , SB0120 , SB0121 and SB2025 ( corresponding to 66 patients ) detected in our samples , the Gd ( 0 . 58 ) , the HS ( 0 . 12 ) and the CR ( 42 . 4% ) values were equivalent to the data obtained in the whole Tunisian M . bovis samples based on the MIRU-VNTR patterns plus spoligotyping ( S4 Table ) . Among the 66 patients infected by the three main M . bovis SBs ( SB0120 , SB0121 and SB2025 ) , the univariate analysis showed that the clustering was statistically associated with female patients ( 87 . 2% vs 63% , p = 0 . 021 ) but not affected by age , area of residence , raw and/or unpasteurized milk consumption , contact with livestock , personal TB history and BCG vaccination status ( p >0 . 05 ) ( S4 Table ) .
In South Tunisia , almost 77% of clinically and bacteriologically diagnosed cases of EPTB are attributed to M . bovis [7] . To our knowledge , this is the first study that investigates molecular characterization of M . bovis causing human EPTB in the South of Tunisia . The geographical distribution of human EPTB cases due to M . bovis in our study showed that the disease was widespread in the seven governorates in Southern Tunisia . SB0120 , SB0121 and SB0134 have been described in humans and livestock across the globe [29; 30] indicating the zoonotic importance of these M . bovis spoligotypes . The most detected spoligotype in our study was SB0120 ( 26 . 9% ) which is also the most common in humans in Italy ( 63 . 8% ) [31] , France ( 69% ) [32] , Germany ( 20% ) [33] compared to 1% in the United Kingdom [34] . SB0120 is also detected in cattle in some European countries like France [35] , Italy [30] , Portugal [30 , 36] , Spain [37] , and Germany [33] as well as in neighboring countries ( Algeria and Morocco ) [38 , 39] . The frequency of SB0121 ( 19 . 3% ) among our human isolates was in agreement with those mentioned in humans in previous studies ( e . g . France ( 15% ) [32] and Spain ( 14 . 1% ) [40] ) . However , it was scarce in human M . bovis isolates from individuals in London and the southeast of England ( 6% ) [29] . SB0134 , detected in our study at 5 . 5% , was found in human M . bovis cases in France [32] , in Spain ( 10 . 5% to 13 . 1% ) [40; 41] , in London , in the southeast of England ( 6% ) [29] , in UK ( 2% ) [34] and in Italy ( 2% ) [31] . SB0121 and SB0134 were previously detected in animal isolates all over the world [30] including neighboring countries such as Algeria [38] and Morocco [39] . Interestingly , this is the first study to identify SB2025 and SB1200 in human M . bovis isolates causing EPTB . It is worth noting that these two spoligotypes were isolated for the first time in milk and cattle samples in Tunisia , respectively [10; 28] and they were not detected in cattle in France [35] . Over the previous years , more emphasis was put on human infection due to M . caprae because of the increasing implication of this species in TB infection in cattle or other animals [42; 43] . Among our isolates , M . bovis subsp . caprae SB2024 was represented by one isolate and was previously found in Tunisian isolates from cattle [9 , 10] . This spoligotype was never reported in humans . In our study , M . bovis subsp . caprae SB0866 was localized in the same region ( Gafsa in Central West Tunisia ) but was never found in Tunisian cattle samples [9 , 10] . Only one EPTB case infected by this spoligotype was reported in humans in Germany [33] . The proportion of spoligotypes detected in our human isolates and shared with those previously found in cattle in Tunisia was of 86 . 7% . This was in agreement with the high values ( 61–90 . 9% ) previously published [44 , 45] . Nevertheless , there are differences in frequency of certain spoligotypes among human and cattle isolates reported in Tunisia . For example , SB0134 was the second predominant spoligotype ( 11 . 4% to 20% ) [9 , 10] in cattle while it was the fifth most common ( 5 . 5% ) in our human strains . This finding suggests differential abilities of spoligotypes to infect humans and animals [30 , 44 , 45] . Regarding MIRU-VNTR data , human isolates shared some CVP genotypes based on the 6 common loci ( ETR-A , ETR-B , QUB 11a , QUB 11b , QUB 3232 and MIR U4 ( ETR-D ) with cattle in Tunisia . However , humans and cattle showed different M . bovis populations suggesting a host-specific pool of genotypes as described in the Table 5 . All these findings showed a specificity of M . bovis populations in terms of geography and hosts as observed for M . tuberculosis . It would be pertinent to go further in the evolutionary and biological mechanisms linked to host specificity in order to better understand the transmission mechanism between humans and cattle . As previously reported in Tunisia [6 , 7] , our results underlined that human EPTB cases due to M . bovis were more common in women [6 , 7 , 46 , 47] . Female gender represents a major risk factor for EPTB according to several published reports in Turkey , USA , Asia , Egypt and North Africa [48 , 49; 50; 51] . The sociodemographic data in our current study showed that EPTB was also more frequent in “younger patients” between 15–59 years of whom 49% ranged between 20 and 40 years and 81% were female . This finding is in agreement with data reported in Tunisia as well as in several areas with moderate or high TB burden [6 , 7 , 46 , 47 , 49 , 51 , 52 , 53 , 54] . This can be explained by a higher consumption of raw milk or derivatives among “young people” than the elderly . Furthermore , due to the current lifestyle , females are generally more exposed to livestock or products from cattle . Indeed , our data showed that female patients are more than twice as likely as the male patients to have contact with livestock or other animals and 68% of “young patients” ( ranging between 15–59 years ) declared to consume raw milk or derivatives of whom almost 80% were females . Furthermore , the consumption of unpasteurized milk and dairy products has been indicated as the most likely source of transmission in clinical cases of LTB , which is the main form of zoonotic TB . Of note , most of our patients ( 73/110 , ( 66 . 4% ) ) are consumers of raw milk or derivatives , whereas only 34/110 ( 30 . 9% ) patients had contact with livestock , suggesting that , in our study , dairy products are one of the main sources of M . bovis human infection . This could explain the increased proportion of cases with positive LTB diagnosis ( 100/110 ( 91% ) ) caused by M . bovis among EPTB patients in Southern Tunisia . In these areas , people working in dairy farms commonly take raw milk home and sell it clandestinely to markets , increasing the risk of zoonotic TB transmission to humans . It has been shown that M . bovis was able to survive and to remain virulent for extended periods in a variety of cheeses made with raw milk [55; 56; 57] . MIRU-VNTR typing method showed a high genotypic and genetic diversity with 6/8 loci showing high allelic diversity and HGDI values . In our study , the global genetic differentiations based on MIRU-VNTR were very low between governorates and regions . Nevertheless , it was significant between populations from the South East ( Tataouine , Medenine ) and the Central West ( Gafsa , Sidi Bouzid , Kasserine ) regions of Tunisia . These findings suggest a slight structuring in the M . bovis population in humans . These findings are in concordance with those of Smith et al who mentioned that some M . bovis genotypes could be more localized compared to others [58] . The diversity observed in the South East ( Tataouine , Medenine ) and the Central West regions could be explained by the fact that these two regions are transit areas for livestock traffic through the Sahara coming from different neighboring areas with high bTB burden ( e . g . Arab Maghreb ( Libya and Algeria ) and African countries ) . Using combined spoligotyping and MIRU-VNTR method , a high clustering rate of 43 . 9% was obtained . These data do not corroborate previous reports showing much lower rates for LTB , 29 . 5% [59] and 35% [60] in Southwest Ethiopia and 33% for M . bovis TB infection in United Kingdom [61] . These results suggest a high level of recent transmission of zoonotic bTB in humans in Tunisia . SB0120 , SB0121 and SB2025 represent the spoligotypes harboring the highest frequencies of isolates among clustered strains ( 42 . 4% ) . This indicates that the predominant spoligotypes , SB0120 , SB0121 and SB2025 , are mainly responsible for recent transmissions in Southern Tunisia . According to the clustering analysis , female patients , patients aged between 15–59 years , from Central East Tunisia ( Sfax , Gabes ) , South East regions , patients who not have a personal TB history and who have a positive BCG vaccination status were significantly clustered cases . Up to 60% and 58% of isolates from EPTB patients who consumed raw milk/milk products and who were in close contact to livestock , respectively , were also highly regrouped in clusters . Among human isolates that shared CVP genotypes with cattle isolates , 75 . 5% were in clusters . To explain these recent human transmission cases , oral transmission route seems evident since 72/73 ( 98 . 6% ) and 33/34 ( 97% ) patients who consumed raw milk/derivatives or who had animal contact , respectively , had extrapulmonary manifestations in Lymph nodes . It is worth noting that , from the analysis , the elderly , children , young population , patients with personal TB history and with no BCG vaccination were significantly more associated with non-clustered isolates , which can be interpreted as ancient infection . Nevertheless , the recent transmission rate can be underestimated since only an exhaustive study including human and animal isolates could rigorously determine the bTB epidemiology . Furthermore , our study was limited to the South of Tunisia , and the majority of patients were from urban areas . Thus , a larger study with sampling combining rural and urban areas from both North and South of the country , humans and animals would provide a more accurate view of M . bovis molecular epidemiology . This kind of study would give a better evaluation of bTB transmission risk to humans and a better identification of the sources and routes of transmission of M . bovis in humans in Tunisia . Besides , the complementary detailed genomic analysis would also allow study the genes/proteins involved in the biological processes responsible for host specificity and/or pathogenesis of the isolates . In conclusion , our study described for the first time , in southern Tunisia , the transmission of zoonotic TB to humans due to M . bovis . Our findings underlined that recent transmission was the possible explanation of most M . bovis EPTB infections . The high genetic diversity reflects a large number of contamination sources . Nevertheless , the high clustering rates of certain genotypes such as the SB0120 and the strong association with milk or milk products consumption suggest that some genotypes have a higher ability to infect humans and that the source is mainly linked to oral ingestion . Our study highlighted the urgent need for a better veterinary control with the implementation of effective and comprehensive strategies in order to reach an effective protection not only of humans but also of animals . | In South Tunisia , the prevalence of bovine TB is high with Mycobacterium bovis as causative agent and cattle as reservoir of the bacteria . However as previously mentioned in several studies , M . bovis is also responsible for human extrapulmonary tuberculosis ( EPTB ) cases in South Tunisia . Despite the veterinary and medical problems , M . bovis is still little studied . In this context , this work aimed to study the molecular epidemiology of M . bovis in EPTB patients in south Tunisia in order to determine the main risk factors of transmission . Our results underlined that SB0120 , SB0121 and SB2025 , previously described in cattle in Tunisia , represent the predominant genotypes . The findings highlighted that human EPTB caused by M . bovis mainly occurred through the consumption of raw milk or derivatives . These data demonstrate the urgent need to implement strategies for preventing and controlling zoonotic TB . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"livestock",
"medicine",
"and",
"health",
"sciences",
"milk",
"body",
"fluids",
"population",
"genetics",
"geographical",
"locations",
"tropical",
"diseases",
"diet",
"bacterial",
"diseases",
"nutrition",
"population",
"biology",
"bacteria",
"africa",
"infectious",
"dis... | 2019 | A first insight into genetic diversity of Mycobacterium bovis isolated from extrapulmonary tuberculosis patients in South Tunisia assessed by spoligotyping and MIRU VNTR |
The molecular complexity within a cell may be seen as an evolutionary response to the external complexity of the cell’s environment . This suggests that the external environment may be harnessed to interrogate the cell’s internal molecular architecture . Cells , however , are not only nonlinear and non-stationary , but also exhibit heterogeneous responses within a clonal , isogenic population . In effect , each cell undertakes its own experiment . Here , we develop a method of cellular interrogation using programmable microfluidic devices which exploits the additional information present in cell-to-cell variation , without requiring model parameters to be fitted to data . We focussed on Ca2+ signalling in response to hormone stimulation , which exhibits oscillatory spiking in many cell types and chose eight models of Ca2+ signalling networks which exhibit similar behaviour in simulation . We developed a nonlinear frequency analysis for non-stationary responses , which could classify models into groups under parameter variation , but found that this question alone was unable to distinguish critical feedback loops . We further developed a nonlinear amplitude analysis and found that the combination of both questions ruled out six of the models as inconsistent with the experimentally-observed dynamics and heterogeneity . The two models that survived the double interrogation were mathematically different but schematically identical and yielded the same unexpected predictions that we confirmed experimentally . Further analysis showed that subtle mathematical details can markedly influence non-stationary responses under parameter variation , emphasising the difficulty of finding a “correct” model . By developing questions for the pathway being studied , and designing more versatile microfluidics , cellular interrogation holds promise as a systematic strategy that can complement direct intervention by genetics or pharmacology .
The divalent calcium cation , Ca2+ , occupies an unusual position in respect of cellular behaviour . It is highly toxic , being especially prone to precipitate phosphate , and most cells go to considerable lengths to exclude it , typically maintaining a 20 , 000-fold differential between the concentration of Ca2+ in the extracellular medium , in the low millimolar range , and the typical concentration in the cytoplasm , of around 100nM [1] . At the same time , Ca2+ is widely used as an intracellular second messenger and the substantial transmembrane potential difference makes it particularly useful for the fastest cellular events , such as synaptic release of neurotransmitters . In view of the tension between these features , it is not surprising that the dynamics of intracellular Ca2+ show distinctive patterns in time and space . In response to stimulation by a variety of hormones and other agonists , mammalian cells exhibit repetitive spikes of cytoplasmic Ca2+ [2 , 3] , thereby allowing Ca2+ to be deployed without accumulating to toxic levels . The frequency of oscillations often increases with agonist concentration and downstream cellular processes which are sensitive to Ca2+ can show striking frequency dependence . This has suggested that Ca2+ implements a frequency-modulated form of information processing , allowing exogenous signals to selectively tune a broad range of cellular responses [4] . To orchestrate the information processing required for a given cell type and physiological context , cells exploit a conserved calcium signalling “molecular toolkit” [5] , from which appropriate Ca2+ -handling proteins are picked and mixed ( Fig 1A ) . Several molecular networks have been put forward to account for the Ca2+ oscillations , typically based on interlinked positive and negative feedback loops , a motif that is believed to implement robust frequency control [6] . The corresponding mathematical models show stable limit-cycle oscillations which resemble those shown by Ca2+ , as reviewed in [7 , 8] . However , it has proved difficult to conclusively determine which network is responsible for Ca2+ oscillations in any cell type . We approached this problem by asking whether cells could be probed with more complex forms of stimulation in such a way that their responses told us more about the underlying network . It is well known , for instance , that a linear system can be reconstructed from its response to different frequencies of stimulation ( such engineering analogies are reviewed further in the Discussion ) and this strategy has been explored previously in other signalling networks [9–12] . There are several difficulties in applying this strategy to Ca2+ signalling . In addition to being nonlinear , Ca2+ responses are also non-stationary , with neither the amplitude nor the period of spikes reaching a steady state , and different cells , even in a clonal population , can exhibit markedly different oscillatory responses [13] , making it difficult to glean information from population-averaged measurements . However , the problem of cell-to-cell variation may potentially be turned to our advantage . It seems reasonable to assume that each cell in a clonal , isogenic population has the same network and what makes the cells respond differently is that the effective parameters of the network differ between cells . Accordingly , by examining single-cell responses and comparing the experimental responses to that of a mathematical model exposed to parameter variation , it might , in principle , be possible to gain more information with which to constrain the underlying network . Cell-to-cell variation has also been exploited in other ways [14 , 15] and its potential as a methodology has been noted [16] . To develop our strategy , we designed and built two-layer polydimethylsiloxane ( PDMS ) microfluidic devices which can reproducibly generate a train of pulses of an appropriate hormone , such as histamine ( Fig 1B and Materials and Methods ) . Pressure-regulated on-chip valves [17] allow pulse width and inter-pulse period to be controlled by computer . To avoid cellular stress responses due to growth on PDMS and exposure to high shear flow , the device was bonded to a dish in which the cells were grown as normal ( “Chip-In-A-Dish” ) . Total cellular Ca2+ was measured at single-cell resolution using the cell-permeant Ca2+ -sensitive dye Fluo4-AM and fluorescence microscopy . Fig 2A and 2B show features of the Ca2+ response of individual cells to steps and pulses , respectively , of histamine . We kept to a fixed amplitude of 10 μM histamine throughout . A step increase elicits a large initial Ca2+ spike followed by repeated spikes with increasing inter-spike periods . The time-averaged period has a roughly Gaussian distribution over the cell population , with a mean ± SD of 130 ± 40 seconds ( inset ) . This “natural period” gives a timescale for the free-running oscillator . Repetitive histamine pulsing at a smaller-than-natural period entrains the oscillator . Initially , each histamine pulse elicits a Ca2+ spike but this leads eventually to skipping , with some pulses generating only tiny spikes , followed again by larger ones ( Fig 1C ) . Such “phase locking” is well known in forced oscillators [18] . It is visible in early calcium-spiking data [19] and has been studied previously [20 , 21] . In our experiments , the relative timescale of pulse to step stimulation , given by the ratio of pulse width in seconds to the mean natural period of 130 seconds , varied between 5/130 ( 0 . 04 ) and 38/130 ( 0 . 29 ) . Note that both step and pulse stimulations elicit non-stationary responses . To relate this data to the circuitry within cells , we examined eight networks , for which mathematical models exhibit oscillatory spiking ( Fig 3 ) . We chose these models to include both small distinctions ( SB2 vs SB3 ) and large ( MST vs GDB ) , as well as the important distinction [22] between class 1 , or receptor-controlled , ( AT1 , LR1 ) , and class 2 , or second-messenger controlled , ( AT2 , LR2 , MST ) , models , which is discussed further below . In separate work , we developed a computational infrastructure called Proteus for building a spectrum of models in a modular fashion from a basic set of components [23] , in keeping with the idea of a calcium signalling toolkit . All the models used in this paper are publicly accessible through the Proteus website . The chosen models exhibit oscillatory spiking in response to step stimulation , eventually reaching a stable limit cycle from the reference initial conditions and parameter values given in the original papers ( S1 Text ) . They also exhibit phase-locking in response to pulse stimulation , in a similar way to the cells in Fig 2B . We wanted to develop suitable forms of interrogation that could discriminate between these models , but , in view of the cell-to-cell variation in response , we wanted to avoid fitting the models to data . Fitting to the population average data would not be meaningful , given such oscillatory behaviour , as no cell would behave like the average , and fitting to the data from any “representative” cell , however that cell might be chosen , runs the risk of overfitting , or capturing what is unique to that cell instead of what is general to all cells . We therefore developed a form of nonlinear frequency analysis that allows for nonstationarity , which , when coupled to parameter variation , enabled us to rule out three of the models . We then developed a nonlinear amplitude analysis that similarly ruled out three further models , leaving only the models AT1 and LR1 , which passed both interrogations . These two models are schematically identical ( Fig 3 ) but differ in their mathematical details . The models yielded unexpected predictions that we confirmed experimentally and further analysis showed how subtle mathematical differences can markedly change the distribution of model responses ( Discussion ) .
To develop a nonlinear frequency analysis that allows for non-stationarity , we translated a phase-locked pattern of Ca2+ spikes into a bitstring by determining which peak in the data gave rise to a “spike” ( binary 1 ) and which to a “skip” ( binary 0 ) ( Fig 4A ) . We then counted the skipping patterns between consecutive appearances of the bitstring “10” , which we took as the onset of a bout of skipping . A skipping pattern , represented as a fraction i/n , signifies i Ca2+ spikes out of n histamine pulses . The set of skipping patterns contains more information than a single time-averaged phase-locking ratio , as used previously [21] , and better captures the heterogeneity of the response . Identical algorithms were applied , after spike identification , to experimental data from an individual cell and to the simulation output from each model , using the entire transient response from the onset of stimulation to incorporate the non-stationarity ( Fig 4A ) . Cell-to-cell variation in signalling responses are thought to arise primarily from extrinsic cell-to-cell variation in the concentrations of molecular components [24 , 25] . As explained in the Introduction , we sought to exploit this additional information rather than average over it . We therefore aggregated the skipping pattern counts over all cells and over multiple experiments at different periods ( Materials and Methods ) , to yield the skipping-pattern histogram in Fig 4B . The patterns 1/2 and 1/3 dominate , with more than half of all patterns being 1/2 . For a model , extrinsic variation between cells corresponds directly to variation in initial conditions ( ICs ) and also indirectly , through the influence of component concentrations on reaction rates , to variation in effective parameter values ( PVs ) . Accordingly , for each model , we randomly selected sets of ICs and PVs at which the model exhibited oscillatory spiking in response to step stimulation . The empirical distributions of ICs and PVs are not well understood and measurements of them are difficult and largely lacking . In the absence of such data , we independently selected ICs and PVs using either uniform sampling or lognormal sampling around the reference values in the original papers ( Materials and Methods ) . For each sampled set of ICs and PVs we determined the “natural period” of the corresponding model , as above for the experimental data . We then subjected each set of ICs and PVs to pulse stimulation at 20 different inter-pulse periods . Because inter-spike periods vary widely , we set the relative timescale of pulse to step stimulation in the middle of the experimental range , at 0 . 125 . For each model , we aggregated the skipping-pattern counts over the sampled sets of ICs and PVs and all 20 stimulations ( Fig 5A ) . We also calculated as a control the pattern histogram for a population of randomly chosen bitstrings . We found this to be markedly different from the simulated histograms in exhibiting symmetry around the 1/2 skipping pattern ( Fig 5C ) . The simulation histograms stabilise after several thousand samples and we chose 20 , 000 samples to provide a balance between coverage and efficiency . However , this number is still small in respect of the dimension of the space being sampled and we checked further for sampling errors . We undertook sub-sampling and super-sampling with both the uniform and the lognormal distributions . We found no evidence for heterogeneity ( Figs J and K in S1 Text ) , except in the case of AT2 . Excluding AT2 for the moment , we found by inspection of Fig 5A three groups of histograms with slightly different membership of the groups depending on whether the sampling method was uniform or lognormal . Under uniform sampling , GDB , SB2 and SB3 have high values at 1/3; MST has moderate values at 1/2 and 1/3 , with 1/3 being the higher; and AT1 , LR1 and LR2 also have moderate values at 1/2 and 1/3 , with 1/2 being the higher . These qualitative relationships were all supported by sub- and super-sampling ( Fig J in S1 Text ) . The last group best matched the experimental distribution ( Fig 5B ) , as was further confirmed by two metrics , a distance measure , Δ , that we defined and the Kolmogorov–Smirnov ( KS ) statistic ( Materials and Methods ) . When comparing histograms , it is important to use the metrics in conjunction with visual inspection , as histograms which are metrically close may still look quite different , as in the case of RAN . Under lognormal sampling , MST still has moderate values at 1/2 and 1/3 but now 1/2 is higher , placing it in the same group as AT1 , LR1 and LR2 which best matches EXP . The other group of GDB , SB2 and SB3 with high values at 1/3 remains the same . These qualitative relationships were also supported by sub- and super-sampling ( Fig K in S1 Text ) . LR1 is now the closest to EXP under both metrics with AT1 , LR2 and MST next closest . The position of AT2 within these groups was ambiguous , with some samples showing 1/3 higher than 1/2 and some samples showing 1/2 higher than 1/3 ( Figs J and K in S1 Text ) . It seemed unlikely that further sampling would decisively resolve this heterogeneity . In summary , nonlinear frequency analysis rules out GDB , SB2 and SB3 and allows AT1 , LR1 and LR2 , under both uniform and lognormal sampling , while leaving AT2 and MST as possibilities under particular circumstances . Although nonlinear frequency analysis classified models into meaningful groups , it was unable to separate class 1 from class 2 models . In class 1 , IP3 may initiate but does not drive Ca2+ oscillations , so that if IP3 oscillates , it does so only passively; in class 2 , Ca2+ feeds back upon IP3 , whose oscillation is thereby required for that of Ca2+ [22] . When HeLa cells respond to a step of histamine , IP3 has been found not to oscillate [26] , indicating that whatever model is appropriate should be of class 1 . The analysis correctly placed AT1 in the group closest to EXP but did not separate LR1 from LR2 , while AT2 and MST are both class 2 and remain possibilities . In contrast to previous work [21] , we find that phase-locking alone does not resolve key features of network architecture . We therefore sought further questions and turned to amplitude in place of frequency . The experimental response to step stimulation shows steadily decreasing spike amplitude after a high initial spike ( Fig 2A ) . When initial Ca ER is increased from its reference value , we found similar behaviour in the AT1 model while LR2 settled immediately to a lower constant spike height and LR1 showed intermediate behaviour ( Fig 6A ) . Accordingly , we defined a measure of amplitude decay rate based on counting the number of spikes before a specified cut-off ( Fig 6B ) and determined the histogram of this measure from all cells for all step stimulation experiments ( Fig 6C ) . HeLa cells show a steady decrease in frequency of the number of spikes before cut-off . We then calculated the distribution of this measure by simulation over all models under both uniform and lognormal sampling ( Fig 7A ) . 20 , 000 samples gave stable distributions , with no exceptions found by sub- and super-sampling ( Figs L and M in S1 Text ) . AT2 , LR2 and MST were quite different by inspection from EXP under both uniform and lognormal sampling and this was supported by both metrics ( Fig 7B ) . We were therefore able to rule out all the class 2 models which had survived nonlinear frequency analysis . As for the class 1 models , under uniform sampling , AT1 is a good match to EXP by inspection and is closest to EXP under both metrics , while LR1 is conspicuously worse . However , under lognormal sampling , AT1 and LR1 are the closest to EXP by visual inspection and their distance to EXP is almost the same under both metrics . Nonlinear amplitude analysis therefore leaves us with both AT1 and LR1 as possibilities . AT1 and LR1 have identical schematics in Fig 3 but differ in the mathematical details of how components and feedbacks are represented . Their similarity makes it difficult to identify a third interrogation that would distinguish them . However , the analysis above does reveal a difference between the models: LR1 , unlike AT1 , shows a marked sensitivity to the sampling method ( Fig 7A ) . We sought to understand how this sensitivity arises . We constructed 11 hybrid AT1-based models , as listed by number below , by starting with AT1 and replacing individual mathematical assumptions with those used in LR1 . We similarly constructed 11 LR1-based models by starting with LR1 and doing the opposite change to that listed above for models 1 , 2 , 4 , 6 , 7 , 8 , 9 , 11 and the corresponding combination of changes as listed above for models 5 , 10 . Full details of these hybrid models are given in S1 Text . For the purposes of comparison , we also considered the unmodified AT1 ( model 12 ) , unmodified LR1 ( model 13 ) and , as a further control , the original LR2 model ( model 14 ) . We subjected the AT1-based and LR1-based models to nonlinear amplitude analysis under both uniform and lognormal sampling and found a wide range of histograms ( Figs 8A and 9A ) . Each individual model assumption has its own distinctive effect on the amplitude decay of the spiking behaviour . In particular , for the AT1-based models under uniform sampling , model 7 shows an excellent match , and the best match among all models , to the experimental data under visual inspection and both metrics , exceeding in quality that of AT1 itself . As for the sensitivity to the sampling method , all the LR1-based models , with one exception , exhibit similar sensitivity as LR1 itself ( Fig 9A ) . The exception is also model 7 . Model 7 concerns the Ca2+ -dependence of IP3 receptor inactivation , which changes from having a Hill coefficient of 2 in LR1 to a Hill coefficient of 1 in AT1 . However , the loss of sampling sensitivity cannot be attributed to this feature in isolation , as the AT1-based model 7 does not acquire sampling sensitivity when the opposite change is made . The AT1-based models are generally as insensitive to the sampling method as AT1 itself ( Fig 8A ) . We see that subtle mathematical details , as well as the choice of sampling method , can make a substantial difference to achieving a good match to the experimental data , confirming how difficult it can be to find the “correct” model when population variation is taken into account . We sought predictions from the AT1 and LR1 models about HeLa cell responses . In these class 1 models , IP3 acts as a passive link between the input and the core oscillator ( Fig 3 ) and decays exponentially when no histamine is present . If the gap between histamine pulses is large compared to the timescale of IP3 decay , IP3 should fall below the oscillator’s threshold and there should be no spike—indicating that the gap has been detected . Conversely , gaps that are small compared to the IP3 decay timescale should not be detected . Experiments with increasing gaps between histamine pulses showed a steadily increasing proportion of cells in which such gaps produce a detectable response ( Fig 10A ) . The complexity of the spiking pattern makes it difficult to specify in computational terms whether or not a gap is detected but we found that manual scoring of the trend was perfectly consistent , with 5 out of 5 observers independently reporting an increase in cell proportions ( Materials and Methods ) . Interestingly , we noticed that cells could sometimes detect a second inter-pulse gap without detecting the first . Both the AT1 model ( Fig 10B ) and the LR1 model reproduce this unexpected behaviour and further predict that the level of ER Ca2+ sets the detection threshold: for an IP3 decay timescale at which only the second gap is detected , decreasing just the initial ER Ca2+ allows both gaps to be detected ( Fig 10C ) . The interplay between IP3 decay timescale and ER Ca2+ level in detecting inter-pulse gaps has also been reported for the response of HEK293 cells to carbachol [27 , Fig 7] .
There is a natural temptation to analyse cellular mechanisms “inside-out” , by intervening within cells and pulling the mechanisms apart , for which powerful tools exist in genetics , biochemistry and pharmacology . This approach has been less successful for understanding the networks underlying Ca2+ oscillation , perhaps because the Ca2+ signalling toolkit offers many alternative components to implement these networks in different cell types and because many different networks are capable of yielding such oscillations ( Fig 3 ) . A complementary strategy is to recognise that the molecular complexity inside cells is , in some sense , a response over evolutionary time to the complexity of the environments in which those cells have existed , which has selected the information processing tasks that the cells have evolved to carry out . This raises the possibility of an “outside-in” strategy , in which the external environment is used to probe the internal molecular network [28] . Engineering offers several analogies for such an “outside-in” strategy . For instance , in communications engineering , it is well known that a linear system can be reconstructed from its frequency response . In a nonlinear biological system , the language of “bandwidth” , “filters” and “resonance” , based on measuring responses over a range of stimulation frequencies , can still be informative [9 , 12] , especially for homeostatic mechanisms close to their set points [10] , where it may be reasonable to assume that the nonlinear system is well approximated by a linear one . Another analogy is provided by the Internal Models Principle from control theory , which , in informal terms , states that if a system is to be controlled in such a manner that it is robust to some class of perturbations , then its controller must include a suitable model of those perturbations [29] . A particular example of this principle is the mechanism of integral feedback control , which , in the linear approximation , must be present if the system exhibits “perfect adaptation” to perturbation [30 , 31] . Here , inferences are made about the internal mechanism ( ie: the existence of integral feedback control ) without intervening inside the cell . The idea of “internal models” has also been influential in neurobiology of motor control [32 , 33] A third engineering analogy lies in the Principle of Requisite Variety from cybernetics , which informally states that the information capacity of a system must match , in a suitable sense , that of its environment [34] . While resembling the justification given above for an “outside-in” strategy , this principle lacks an evolutionary motivation and has , so far , proved to be of limited biological value . These analogies suggest that an outside-in strategy may be worth exploring further in systems biology but this has been hampered by several challenges . We know little about the actual spatio-temporal profile of the environments to which cells are exposed during development and physiology and it is difficult to manipulate such environments in vivo . It is also hard to reliably and reproducibly construct artificial environments and signals . Microfluidics has provided an important step forward in this respect , although , at the level of chemical signals , it remains difficult to reliably construct environmental signals that are more complicated than the pulse trains used here . Nevertheless , further progress in this direction is likely and raises an interesting conceptual problem . What kinds of signals are most useful for discriminating models ? For instance , if models are chosen from some class , perhaps defined by picking and mixing components in the manner of the calcium signalling toolkit [23] , is it possible to define from among those signals that can be practically generated , a hierarchy of questions , which can efficiently discriminate one model from the others ? Very little seems to be known about this problem but identifying a highly discriminatory class of signals could , in turn , encourage the development of microfluidic devices that can implement them . The greatest challenge , of course , lies in the nature of living cells , which , among many other features , exhibit high degrees of nonlinearity , nonstationarity and heterogeneity . It has , accordingly , rarely been possible to adapt methods of engineering analysis directly . Instead , quantitative measures have to be defined that are appropriate to the particular biological context being studied , as was done here . For Ca2+ oscillation in response to hormone stimulation , the skipping phenomenon during phase-locking ( Fig 2B ) leads naturally to skipping pattern analysis ( Fig 4A ) while the amplitude decay in spiking ( Fig 6A ) leads naturally to the “spikes-before-cut-off” measure ( Fig 6B ) . These measures appear reasonable for the spiking data that we found when we exposed HeLa cells to steps and pulses of histamine but other biological contexts may require different measures . Crucially , for such quantitative measures to gain discriminatory power , it is essential to exploit cell-to-cell variation , as opposed to averaging over it . Each cell in the population is doing its own experiment in response to environmental stimulation and by accumulating this information , we can place greater constraints on the underlying model . We can do so , moreover , not by fitting the model to the data but , rather , by exploring its parametric sensitivity and matching that to what is found in the cell population . Exploiting cell-to-cell variation in this way distinguishes the methodology introduced here from previous efforts at developing “outside-in” strategies [10 , 21] . The difficulty in exploiting population variation is that we lack empirical information about initial conditions and parameters . The latter , in particular , are usually “effective” parameters which may summarise complex mechanistic details . Their values are difficult to measure in individual cells in the first place , let alone over a population of cells , and very little is known about their distributions . We have adopted the tactic here of using two contrasting sampling methods , uniform and lognormal , and allowing models to succeed interrogation with either method . Although interrogation leads to two possible models , AT1 and LR1 , these are so similar ( Fig 3 ) that they give identical predictions which we were able to verify experimentally ( Fig 10 ) . We were also able to find differences between the models , in their sensitivity to uniform versus lognormal sampling ( Figs 8 and 9 ) , and to implicate the Ca2+ -dependence of IP3 receptor inactivation in this feature . However , this difference in behaviour cannot be readily exploited experimentally to distinguish the models . The impact of cellular heterogeneity is particularly evident in the hybrid AT1-based and LR1-based models , which show distinctive changes in the distribution of amplitude decay rates to seemingly small changes in the mathematical details ( Figs 8A and 9A ) . We see that finding the “correct” model for the cell type and hormone in question can be very difficult . We must emphasise that cellular interrogation may be able to discriminate between models but it cannot confirm that a model is the “correct” one . It is here that an “inside-out” strategy becomes invaluable; the two strategies , “outside-in” and “inside-out” , are not exclusive but complementary . We note further that a model is only “correct” in terms of the data against which it has been tested; there is no guarantee that it will remain correct if new data are acquired . As we have argued elsewhere [35] , models are better seen as being “useful” , rather than as being “correct” , and it is in this spirit that the AT1 and LR1 models which emerges from interrogation should be viewed . Oscillatory Ca2+ signalling has many advantages for exploring an “outside-in” strategy , not the least of which is that Ca2+ can be measured in single cells in real-time with considerable accuracy . However , oscillatory signalling is emerging as a much broader theme in biology [36–39] . Many other cellular components , such as key transcription factors , are now known to exhibit oscillations , to whose frequency downstream responses show differential sensitivity . It appears that evolution has exploited frequency modulation as a form of information processing , even when it is not a protection against toxicity . We hope that the methods introduced here for exploiting nonstationarity and heterogeneity may be usefully extended to these other biological systems . By asking cells the right questions , they may tell us more about themselves .
The microfluidic device was a simplified version of one developed in previous work [40] . Device construction uses soft lithography [17 , 41] , enabling automated fluid handling on the device through computer-controlled valves . The valves require two layers , each made from the elastomer polydimethylsiloxane ( PDMS; Dow Corning , Midland , MI , USA ) , a control layer containing the control lines and a flow layer containing the flow lines . The PDMS layers are formed by replica molding from silicon masters on which features are constructed by standard photolithographic techniques . The two layers are bonded together to form the microfluidic device . Valves are created where a control line crosses a flow line , at which juncture a thin , flexible membrane separates the two lines . By increasing the fluid pressure in the control line , the membrane is deformed , thereby closing the valve and blocking the flow line . For proper closing , the flow line needs a semi-circular cross section . In our device , valves could be reliably opened and closed at a frequency of approximately 1Hz . The masters for each layer were created by bonding photosensitive epoxy ( a photoresist ) to 3” silicon wafers . Patterns were created on the master by using photolithographic masks ( Fig A in S1 Text ) printed at 20 , 000dpi ( CAD/Art Services Inc . ) yielding 10 μm minimum feature resolution . The control layer master consisted of SU-8 photoresist ( MicroChem , Newton , MA , USA ) patterned into control lines 40 μm thick using MicroChem’s SU-8 protocol ( http://microchem . com/pdf/SU-82000DataSheet2025thru2075Ver4 . pdf ) . The flow layer master required a two step process , with a layer of 22 μm AZ 50XT photoresist ( AZ Technology , Huntsville , AL , USA ) for the flow lines , followed by a layer of 42 μm SU-8 photoresist for the cell chamber and output port . ( The cell chamber was cut off in the CIAD device , as explained below . ) The three masks are shown in Fig A in S1 Text . The protocol for the AZ resist was adapted by us as follows . The silicon wafer was etched in the plasma etcher at 115W for 5 minutes to loosen organic matter . The wafer was then spun at 1000 rpm , rinsed and swabbed twice with acetone , twice with methanol and twice with isopropanol . The wafer was then heated to 200°C for 5 minutes to dehydrate . The wafer was then covered with MCC primer ( MicroChem , Newton , MA , USA ) for 10 seconds , which was then spun off at 1000 rpm for 30 seconds and then baked at 100°C for 5 minutes . The AZ 50XT photoresist was carefully applied so as not to create air bubbles . The wafer was then spun at 2800 rpm for 30 seconds and then baked at 90°C at high humidity for 30 minutes . Humidity was achieved by placing beakers of water on the hotplate with the wafer and covering the entire assembly with aluminum foil . After baking , the wafer was placed in a 39% to 42% relative humidity chamber for 45 minutes , to equilibrate the water content of the photoresist to an optimal value . The wafer was exposed to UV light of 15 mW/cm2 constant intensity at 365 nm wavelength for 60 seconds , rested for 60 seconds , and then exposed again for 60 seconds . The wafer was developed in AZ400 developer ( AZ Technology , Huntsville , AL , USA ) diluted three parts water to one part developer and rinsed with deionised water . The temperature was ramped to 250°C over one hour to reflow and harden the resist . Reflow changes the cross-section of the resist from rectangular to semi-circular , which ensures proper closing of the flow lines during valve operation . The wafer was then cooled slowly to prevent cracking . The SU-8 photoresist was then applied following MicroChem’s SU-8 protocol , as above for the control layer master . The flow layer itself was made by placing mixed and degassed PDMS , in the ratio 20:1 of pre-polymer to curing agent , on the flow layer master and spinning at 2100 rpm for 45 sec to achieve the desired thickness of 50 μm . The control layer was made by placing mixed and degassed 5:1 PDMS on the control layer master inside a foil container for a resulting PDMS thickness of about 5mm . The flow and control assemblies were baked at 65°C for at least one hour . The control layer was then peeled from the master and cut into 6 individual pieces as marked by fiduciary lines . The pieces were then blown dry with nitrogen at 70 psig and bonded to the flow layer by plasma etching both pieces at 115 W , 115 mTorr for 15 sec and then baking the assembly at 65°C for one hour . The flow layer was carefully cut around the perimeter of the six control layer pieces using a curved-blade knife and peeled from the flow-layer master . Biopsy punches at 0 . 50 mm for control lines and 0 . 75 mm for flow lines were used to punch the holes in the devices at fiduciary marks in the PDMS . Razor blades were used to cut the devices to the desired size , indicated by line-shaped indentations in the PDMS . The devices were then rinsed with 100% isopropanol and blown dry with nitrogen . Other standard protocols , such as those for cleaning silicon wafers , may be found at ( https://microfluidics . hms . harvard . edu/protocols . html ) . The device described above included a chamber for cells but we found that baseline Ca2+ measurements in the absence of hormone differed between cells grown on the device to those grown on a dish . We opted , therefore , to cut off the chamber from the device and to grow cells in a dish ( Fig 1B ) , leaving the microfluidic device to undertake the fluid handling and pulse generation . A disadvantage of this design is attenuation and spreading of the pulse as it leaves the device output port . To counteract this , we measured Ca2+ responses only in those cells in the vicinity of the output port and used fluorescein dye to check that the cells in this region were receiving a properly-shaped pulse . We also determined skipping histograms in different areas of the dish and found no qualitative difference between them ( Fig D in S1 Text ) . To construct the CIAD , a microfluidic device fabricated as described above was plasma etched ( 115 W , 115 mTorr for 15 s ) together with a 35 mm glass-bottom dish ( In Vitro Scientific , D35-20-1-N ) and the device was placed near the edge of the glass portion of the dish to bond . The whole assembly was then baked for at least 90 min at 65°C . To prepare the CIAD for experiments , it was filled with Milli-Q purified water ( EMD Millipore , Billerica , MA , USA ) and placed under vacuum for at least 90 minutes to remove air from the control lines . Tygon tubing ( Cole-Parmer , 06418-02 ) was cut to 25 cm lengths and connected to stainless steel tubes ( New England Small tube , NE-1301-03 ) bent in half at a 90 degree angle . These were filled with water and the steel-tube end inserted into the control lines of the water-filled CIAD . The CIAD was then air dried and filled with collagen solution ( Sigma , C8919 , diluted with Milli-Q water 10:1 ) , so as to also cover the glass portion of the dish . This was incubated for 24 hours , the collagen removed , the CIAD flushed with sterile water to remove excess collagen and vacuumed dry by inserting a syringe into the fluid lines and applying negative pressure for at least 10 seconds . Finally , the CIAD was sterilised by being placed under UV radiation inside a laminar flow cabinet for 24 hours . To perform an experiment , water-filled control lines were connected to pneumatic lines at a pressure of 25 psig . These lines are controlled by a PCI card which is run by a purpose-built software program developed previously [40] , which allows the valves to to be actuated to within millisecond time resolution . The two solutions used in the flow lines were 2 . 5 mM probenecid ( Invitrogen , P36400 ) in HBSS ( Sigma , H8264 ) , with and without 10 μM histamine ( Sigma , 53290 ) , both pressurised to 5 psig ( air pressure measured in fluid reservoir ) . The histamine valve and buffer valve were closed and the purge valves opened , allowing the two solutions to flow through their respective parts of the device . This process purges any air bubbles and remaining fluid from within the device and is done for 10 seconds or until all air bubbles are removed from the flow lines . At this point , the purge valves are closed . HeLa Cells ( ATCC , CCL-2 , lot number 4965442 ) were incubated in T-25 flasks at 37°C with 5% CO2 . They were passaged twice per week by trypsinizing the cells by rinsing with 5 mL HBSS and 1 mL trypsin EDTA ( Cellgro , 25-052-Cl ) and incubating for 5 minutes at 37°C in 5% CO2 and then splitting the suspension into three T-25 flasks , each with 3 mL to 5 mL media consisting of DMEM ( Cellgro , 10-013-CV ) with 10% FBS ( Gibco , 26140 ) and 1% penicillin-streptomycin solution ( 100X , 10 , 000 I . U . penicillin , 10 , 000 μg/mL streptomycin ( Cellgro , 30-002-CI ) ) . Confluent cells were trypsinized as described above and each CIAD was given 170 μL cell suspension ( approximately 4 . 7 × 105 cells ) with 1 . 83 mL media and 83 μM trolox ( Sigma , 238813-1G ) as an antioxidant . This was incubated at 37°C in 5% CO2 for 24 hours . To perform an experiment , a CIAD that had been incubated with cells , as above , was rinsed twice with 1 mL HBSS and then filled with 1 mL HBSS with 2 . 5 mM probenecid and 1 μM Fluo-4 AM ( Invitrogen , F14217 ) and left at room temperature ( 18°C to 20°C ) for 45 minutes . Fluo-4 AM is a Ca2+ -sensitive , cell-permeant dye whose acetoxymethyl group is cleaved upon entering the cell , trapping the dye in the cytoplasm and excluding it from the mitrochondrion and other membrane-bound organelles . Probenecid reduces dye efflux from cells by inhibiting organic-anion transporters in the plasma membrane . The solution was then aspirated and the cells were soaked in HBSS with 2 . 5mM probenecid for another 45 minutes . During this period , the tubes leading to the flow lines were cleaned with ethanol and the appropriate histamine or buffer solution was run through the tubes to remove any remaining ethanol and air bubbles . The tubes were inserted into the CIAD and , after moving fluid through the device with the purge valves open for at least 10 seconds to remove air bubbles , the purge valves were closed . At the end of the 45 minute period , the device was imaged as described next . For imaging , we used a Zeiss Axiovert 135 TV inverted epi-fluorescence microscope equipped with a Chroma 49002 filter set . When bound to Ca2+ , Fluo-4 has an absorption maximum at 494 nm and an emission maximum at 516 nm . We used an ET470/40x excitation filter , which covers a large part of Fluo-4’s absorption spectrum ( although it does not overlap the absorption maximum ) along with a T495lpxr dichroic mirror and ET525/50m emission filter , which cover Fluo-4’s emission spectrum very well . An HBO W/2 lamp was used as a broad-band light source . A Zeiss fluar 10x/0 . 5 M27 objective was chosen for its high transmittance in the range of Fluo-4’s absorbtion and emission wavelengths . A Hamamatsu C10600-10B camera provided a wide field of view and high spatial resolution , although , due to memory constraints , 4X binning was used during sampling . Micro-Manager 1 . 3 , an open-source program developed by Ron Vale’s lab ( http://valelab . ucsf . edu/~MM/MMwiki/index . php/Micro-Manager_Reader_as_ImageJ_plugin ) , which runs as a plugin to ImageJ [42] , provided fully-automated control of the microscope for image acquisition . A 40 ms excitation UV pulse and a 10 ms exposure were used for each image , with images taken at a frame rate between 1 second/frame and 3 seconds/frame , so as to provide a minimum of 4 frames per inter-pulse gap . A low frame rate was used , relative to exposure time , to minimise phototoxicity and increase imaging time . For each experiment , cells were segmented using the ImageJ macro in Fig B in S1 Text: all image frames were superimposed , thereby avoiding cells that were not adhered and , after background subtraction , the image was converted to a binary mask and segmented by the watershed algorithm . This created a template that defined the subsequent regions of interest ( ROIs ) in which fluorescence will be measured . Fluorescence integrated density—the product of the event area and the mean gray-scale value across the area—was measured in each frame for each ROI as a proxy for Ca2+ concentration in the cytoplasm , where the dye is localised . Full details of the eight models which are schematically described in Fig 3 are provided in S1 Text . Because of the stiffness of the equations and the importance of numerical accuracy , we implemented our own 4th-order Runge-Kutta algorithm in Fortran for numerical simulation ( available on request ) . The resulting trajectories , either for step or pulsed stimulation , are analysed in-situ and the results saved for later analysis using Matlab , as described further below . In response to step stimulation , HeLa cells exhibit repetitive spiking with generally decreasing spike amplitude ( Fig 2A ) . This decrease in amplitude is not due to photobleaching . We subjected cells to two step functions of histamine with a gap of 10 minutes between the steps and found that individual cells responded to the second step much as they responded to the first step ( Fig H in S1 Text ) , suggesting that the decrease in amplitude is due to the dynamics of the molecular circuitry . A measure of spike amplitude decay rate was defined ( Fig 6B ) and was applied to both experimental data and simulation output . The experimental data was taken from the three step stimulation experiments in Table 1 . For each selected cell in each experiment , spikes were identified as described above and the amplitude decay rate estimated . The distribution of decay rates is shown in Fig 6C . For each of the models , 20 , 000 sets of ICs and PVs were randomly chosen , following the same procedure as described above for nonlinear frequency analysis . However , in addition to requiring oscillatory behaviour , the spike amplitudes were also required to decrease monotonically once the highest spike in the trajectory was reached . A flow chart and pseudo-code of this algorithm are shown in Fig I in S1 Text . The distribution of decay rates for all models over all sets of ICs and PVs is shown in Fig 7A . | We have developed a cellular interrogation methodology that combines programmable microfluidics , fluorescence microscopy and mathematical analysis and have used it to discriminate between models of repetitive Ca2+ spiking in HeLa cells . Our approach exploits the natural variability in response of individual cells in a clonal population and the non-steady state behavior of the response in each cell , thereby providing more powerful discrimination . Interrogation consists of steps or pulses of histamine of fixed concentration and width but varying frequency . Eight mathematical models of repetitive Ca2+ spiking were chosen from the literature and methods of nonlinear frequency and nonlinear amplitude analysis were developed which ruled out all but two of the models , without having to fit the models to the data . Further analysis of the remaining models yielded predictions that were experimentally confirmed . Cellular interrogation offers a general approach to ruling out competing hypotheses about molecular mechanisms , which is complementary to traditional methods of genetics and biochemistry . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"neurochemistry",
"classical",
"mechanics",
"chemical",
"compounds",
"fluid",
"mechanics",
"engineering",
"and",
"technology",
"hela",
"cells",
"signal",
"processing",
"biological",
"cultures",
"mathematical",
"models",
"neuroscience",
"organic",
"compounds",
"simulation",
... | 2016 | Cellular Interrogation: Exploiting Cell-to-Cell Variability to Discriminate Regulatory Mechanisms in Oscillatory Signalling |
Total internal reflection fluorescence microscopy ( TIR-FM ) has become a powerful tool for studying clathrin-mediated endocytosis . However , due to difficulties in tracking and quantifying their heterogeneous dynamic behavior , detailed analyses have been restricted to a limited number of selected clathrin-coated pits ( CCPs ) . To identify intermediates in the formation of clathrin-coated vesicles and factors that regulate progression through these stages , we used particle-tracking software and statistical methods to establish an unbiased and complete inventory of all visible CCP trajectories . We identified three dynamically distinct CCP subpopulations: two short-lived subpopulations corresponding to aborted intermediates , and one longer-lived productive subpopulation . In a manner dependent on AP2 adaptor complexes , increasing cargo concentration significantly enhances the maturation efficiency of productive CCPs , but has only minor effects on their lifetimes . In contrast , small interfering RNA ( siRNA ) depletion of dynamin-2 GTPase and reintroduction of wild-type or mutant dynamin-1 revealed dynamin's role in controlling the turnover of abortive intermediates and the rate of CCP maturation . From these data , we infer the existence of an endocytic restriction or checkpoint , responsive to cargo and regulated by dynamin .
Clathrin-mediated endocytosis ( CME ) is the major endocytic pathway in eukaryotic cells . It occurs via clathrin-coated pits ( CCPs ) that are assembled from cytosolic coat proteins . CCPs capture transmembrane cargo molecules , invaginate , and then pinch off to form clathrin-coated vesicles ( CCVs ) . CME is a constitutive , yet highly regulated process . Biochemical assays of endocytosis score ligand uptake and measure only the ensemble average of successful internalization events , thereby obscuring critical , rate-limiting early stages of maturation and alternative outcomes that might cause variability in individual CCP dynamics . Indeed , live cell imaging has revealed striking heterogeneity in the dynamic behavior of plasma membrane–associated CCPs [1–5] . An important parameter for analyzing CCP heterogeneity is their lifetimes . The lifetime of an individual CCP at the plasma membrane , i . e . , the time required for ( 1 ) coat initiation , ( 2 ) coat propagation , ( 3 ) neck constriction , and ( 4 ) vesicle budding , is critical for understanding CME . Changes in lifetimes caused by specific molecular perturbations can reveal mechanisms that regulate each of these steps . However , selective probing of all stages of CCP maturation is only possible by mild perturbation of the underlying molecular processes . Detection and interpretation of these necessarily milder phenotypes requires sensitive and comprehensive analysis of individual CCP lifetimes and behavior . To this end , we have employed total internal reflection fluorescence microscopy ( TIR-FM ) , the premier assay to detect early intermediates in CCV formation and visualize the dynamics of CCPs in living cells [1 , 3–9] . By selectively exciting fluorophores associated with molecular components of CCPs at the ventral plasma membrane , TIR-FM provides excellent signal-to-background ratio and high time resolution . In spite of these strengths , it has remained a challenge to extract reliable measurements of CCP lifetimes from TIR-FM videos . Lifetime measurements are notoriously susceptible to tracking errors , which typically break CCP trajectories into two or more subtrajectories , leading to systematic bias of lifetimes towards shorter values . As a result , tracking has previously been accomplished either manually for a low number of well-discernable , high-intensity CCPs [1 , 6] , or using semiautomated tracking restricted to isolated CCPs , for which no close neighbors are likely to confuse the tracking algorithm [2 , 4] . Both approaches sample the behavior of arbitrary and typically small subpopulations with relatively uniform properties . To solve these problems and to better exploit the heterogeneity of CCP dynamics as a source of mechanistic information , we have employed particle-tracking software [10] capable of detecting and tracking all CCPs visualized by TIR-FM in an unbiased fashion . Automated detection and tracking enabled analysis of several tens of thousands of trajectories per condition , 100 times more than previous studies , thus providing a comprehensive and accurate measurement of CCP lifetime distributions .
We used TIR-FM and our automated tracking assay [10] ( see Materials and Methods , Figure S1 , and Videos S1 , S2 , and S4 ) to obtain large and unbiased datasets of CCP dynamics in well-characterized BSC1 cells expressing a fully functional enhanced green fluorescent protein ( EGFP ) -tagged clathrin light chain a ( LCa-EGFP; Figure 1A and 1B ) [2] . To capture both fast events at the timescale of seconds and slower events at the timescale of minutes , we combined data from time-lapse sequences taken at a frame rate of 0 . 4 s for ≥3 min with data from sequences taken at a frame rate of 2 s for 10–15 min ( see Materials and Methods ) . Objects not detected for at least five consecutive frames were not counted , to exclude transient , highly motile structures . Nonetheless , CCPs displayed a nearly exponential decay of lifetimes ( Figure 1C ) , indicating that a large number appear and disappear on the timescale of a few seconds . To further analyze CCP lifetimes , the raw data were fitted to a series of models that differed in the number and types of subdistributions . Our goal was to identify the minimal number of kinetically distinct subpopulations that could account for the overall lifetime distribution observed ( see Figure S2A ) . Model selection was achieved by three strategies . The first two involved minimization of the Bayesian Information Criterion ( BIC ) [11 , 12] , which defines the optimal tradeoff between the goodness of fit of the model and the number of free model parameters . The application of the BIC requires a priori knowledge of the distribution of experimental errors , which is unknown for lifetime data . Thus , we performed BIC minimization , first , in a fit of the cumulative lifetime histogram , i . e . , with the lifetime and the cumulative frequency as the independent and dependent , error-perturbed variables , respectively; and second , in a fit of the inverse of the cumulative histogram , i . e . , with the percentage rank as the independent and the lifetime as the dependent error-perturbed variables . In both cases , we approximated the distribution of the fitting errors as normal . The third strategy for model selection involved a nonparametric test of the distribution of the fitting residuals , which did not require a priori assumptions ( see Figure S2 ) . All three strategies identified three statistically significant subpopulations ( Figure 1D ) with distinct time constants , but broad and overlapping lifetime distributions ( Figure 1F ) . Importantly , at the noise level of our TIR-FM images , accurate assignment of these subpopulations required analysis of >5 , 000 trajectories ( see Figure S2C ) , and hence , our results strongly relied on the accurate and automatic tracking of all CCPs in multiple videos per experimental condition ( see Materials and Methods ) . Our data contained two short-lived subpopulations with time constants of 5 . 2 ± 0 . 1 s and 15 . 9 ± 1 s , respectively ( ±jackknifed cell-to-cell error , see Table S1 ) that were best fit with Rayleigh distributions , i . e . , the shortest and longest lifetimes within the population occur less frequently than the intermediate ones . This suggests that our time sampling of 0 . 4 s per frame was sufficient to capture all events of significant clathrin coat accumulation . A single longer-lived subpopulation ( time constant 86 . 9 ± 5 . 8 s ) was best approximated by an exponentially decaying distribution . Given its long time constant , accurate measurements of the mean lifetime of this population required imaging for ≥10 min . The longer-lived subpopulation was designated the “productive population , ” because ( 1 ) its kinetics match those of surface-bound transferrin internalization measured biochemically ( t1/2 ≈ 104 s; Figure S3A ) , and ( 2 ) manual tracking of 450 CCPs ( for which internalization was confirmed by sequential disappearance from the TIR-FM and the epifluorescence microscopy ( EPI-FM ) field [5 , 6] ) also yielded a mean lifetime of approximately 100 s ( Figure S3B ) . Accordingly , we hypothesized that the two shorter-lived species correspond to transient , nonproductive events and therefore termed them “early abortive” and “late abortive” CCPs , respectively . In BSC1 cells , the productive population constituted only 38 . 6 ± 3 . 4% of total CCPs at the plasma membrane , with early and late abortive CCPs representing 38 . 1 ± 3 . 1% and 21 . 9 ± 1 . 4% , respectively ( ±jackknifed cell-to-cell error; Figure 1E , Table S2 ) . Taking into account the different mean lifetimes and relative contributions of the three individual subpopulations , the mean lifetime of all CCPs is 39 s , much shorter than the half-time of transferrin ( Tfn ) uptake . This further supports the hypothesis that the two short-lived subpopulations are abortive and do not contribute to Tfn uptake . To determine whether the existence of these three subpopulations was affected by the nature of the fluorescent tag or the cell type used , we performed lifetime analyses on BSC1 and HeLa cells transiently expressing LCa-tomato and NIH3T3 cells stably expressing LCa-DsRed . We consistently observed one long-lived and two short-lived populations ( Table S1B ) , suggesting that this categorization into kinetically distinct CCP populations is a universal phenomenon of CME . As described by others [4 , 13] , however , we also observed in both HeLa cells and NIH3T3 cells , a higher proportion of larger clathrin-coated structures ( CCSs ) from which multiple CCSs emerge and disappear . These so-called “nonterminal” events are rarely detected in BSC1 cells , consistent with previous findings [2] . We also analyzed CCP lifetimes by tracking the adaptor protein , AP2 , in BSC1 cells expressing the EGFP-tagged σ2 subunit ( σ2-EGFP ) , shown not to interfere with AP2 function [2] ( Videos S3 and S5 ) . Our model selection again identified three kinetically distinct populations of σ2-containing CCPs ( Figure 2C and 2D , Tables S1 and S2 ) ; the preference for three versus two subpopulations was weaker but still highly significant ( p < 10−4 as compared to p < 10−10 in LCa-EGFP–expressing cells ) . This further supports our conclusion that each of these subpopulations represents bona fide plasma membrane–associated CCPs rather than clathrin-bearing endosomal structures transiently approaching the cell surface . The percentage of productive CCPs with σ2-EGFP labeling was higher than those labeled with LCa-EGFP ( 56 . 3 ± 10 . 1% as compared to 38 . 6 ± 3 . 4%; compare Figure 2A and 2C , Table S2 ) , confirming previous suggestions [2] that adaptors enhance the maturation efficiency of CCPs . The characteristic lifetimes of early abortive CCPs labeled with σ2-EGFP ( 4 . 8 ± 0 . 4 s ) were similar to those observed for LCa-EGFP ( 5 . 2 ± 0 . 1 s; compare Figure 2B and 2D , Table S2 ) , suggesting that this very short-lived subpopulation represents stochastic coated pit nucleation events , perhaps triggered by low-affinity interactions between AP2 and phosphatidylinositol-4 , 5-bisphosphate at the plasma membrane . In contrast , the characteristic lifetimes of both late abortive and productive subpopulations of σ2-EGFP–labeled CCPs ( 8 . 4 ± 1 . 8 s and 71 . 5 ± 6 s , respectively ) were significantly shorter than their LCa-EGFP–labeled counterparts ( 15 . 9 ± 1 s and 86 . 9 ± 5 . 8 s , respectively ) . This observation is consistent with findings of others , reporting a general shift of AP2-containing structures toward shorter lifetimes [2 , 8] , although interpretations have varied . It may reflect dissociation of AP2 complexes from CCPs prior to clathrin [8] and/or a nonuniform distribution of AP2 complexes in the clathrin lattice resulting in their differential illumination by the TIRF field [9] . The former interpretation would be consistent with recent findings that Sar1p and the Sec23/24 adaptors can dissociate from budding vesicles prior to the Sec13/31p-containing outer shell of the COPII coat [14 , 15] and the notion that multivalent clathrin interactions dominate over AP2 interactions at later stages of CCV formation [16] . Using semiautomated analysis , Ehrlich et al . [2] previously detected a single , short-lived subpopulation of CCPs in BSC1 cells , which displayed a mean lifetime and relative contribution similar to our late abortive population . In addition , they reported that cargo-associated CCPs rarely failed to proceed to completion , leading to the suggestion that cargo might stabilize abortive CCPs . However , a direct link between CCP maturation and cargo load has not been established . To test the hypothesis that CCP maturation is responsive to the concentration of cargo , we infected BSC1 cells with an adenovirus coding for the human transferrin receptor ( TfnR ) in a tetracycline ( tet ) -repressible system . The TfnR is constitutively internalized even in the absence of ligand [17] and thus serves as a model transmembrane cargo molecule . Removal of tet induced TfnR overexpression by >30-fold in nearly all cells ( Figure 3A–3C ) . There have been conflicting observations regarding the effect of TfnR overexpression on surface density of CCPs [18–20] . We confirmed previous biochemical measurements [20] showing that at high levels of TfnR overexpression , the endocytic machinery becomes saturated , i . e . , the bulk endocytic efficiency of TfnR declines ( Figure S3A ) , thus obscuring any effect that cargo might have on CCP dynamics . In addition , we did not observe an increase in membrane recruitment of either clathrin or the TfnR adaptor protein AP2 , as measured by subcellular fractionation and western blot analysis ( Figure S3C and S3D ) or by TIR-FM ( Figure 3D and 3E ) . In cells overexpressing TfnRs , we could no longer detect a significant ( p > 0 . 05 ) late abortive subpopulation labeled with LCa-EGFP , and this population was completely undetectable in σ2-EGFP labeled cells . The contribution of the productive population increased to 67 . 9 ± 7 . 9% of LCa-EGFP–labeled CCPs ( Figure 2A ) and 76 . 8 ± 5 . 6% of σ2-EGFP–containing CCPs ( Figure 2C ) . From this , we conclude that CCPs mature with their highest efficiency when a threshold amount of AP2 and cargo are incorporated into the nascent clathrin lattice . In contrast , the relative contribution of early abortive CCPs is unchanged by cargo overexpression further supporting the notion that these are transient structures assembled in a cargo-independent manner . The mean lifetime of the productive population was only slightly decreased upon cargo overexpression when compared to control conditions ( Kolmogorov-Smirnov test [KS-test] p = 0 . 03 , Table S1 ) in LCa-EGFP–expressing cells ( Figure 2B ) and showed no decrease in σ2-EGFP–expressing cells ( Figure 2D ) . Thus , we conclude that elevated TfnR concentrations result in more efficient CCP maturation , but that TfnR concentration is not rate-limiting for CCV formation . The approximately 2-fold increase in efficiency of CCP maturation in the presence of an approximately 40-fold increase in cargo concentration points to the existence of other limiting factors and is consistent with the saturation of endocytic efficiency observed biochemically ( Figure S3A ) . To further probe the role of AP2 adaptors in pit maturation , we decreased cellular AP2 levels by approximately 50% through short-term small interfering RNA ( siRNA ) -mediated knockdown of the μ2-subunit ( unpublished data ) [21] . AP2 depletion reduced the densities of all classes of pits ( from 0 . 477 ± 0 . 012 μm2 in control to 0 . 276 ± 0 . 069 μm2 in the knockdown ) , although the relative contributions of the three populations and their lifetimes remained unchanged ( Figure 2A and 2B ) . This suggests that AP2-dependent nucleation events lead proportionally to both short-lived abortive and productive pits , implying a precursor/product relationship . In contrast to control cells , TfnR overexpression in AP2-depleted cells did not lead to an increase in the relative contribution of the productive population ( Figure 2A ) . We conclude that the increase in CCP maturation efficiency with increased cargo concentration requires AP2 and/or is limited by AP2 concentration . Given that TfnR/cargo concentration only marginally affects lifetimes of CCPs , we next investigated which factor ( s ) might regulate this aspect of CCP dynamics . The self-assembling GTPase dynamin has been suggested to play a dual role in CME , both as a regulator and/or fidelity monitor during early , rate-limiting steps in endocytosis , and as a well-documented component of the fission apparatus late in CCV formation [22–24] . Dynamin is recruited along with clathrin and AP2 [2] , and the early activities of dynamin presumably occur while it is associated with coated pits in its unassembled state , utilizing its basal GTP binding and hydrolysis activities [22 , 23] . In contrast , dynamin function in membrane fission requires its self-assembly and assembly-stimulated GTPase activities [22 , 23 , 25] and occurs subsequent to a burst of recruitment at late stages of CCP formation [5–7] . We sought direct evidence for dynamin's dual role in CME by examining the effects of well-characterized dynamin mutants on CCP dynamics by siRNA-mediated knockdown of dynamin-2 and reintroduction of siRNA-resistant wild-type ( WT ) or mutant dynamin-1 . Because dominant-negative dynamin mutants block endocytosis and lead to clustering of nonproductive CCPs ( unpublished data ) , we focused our analysis on three well-characterized hypoactive dynamin mutants: ( 1 ) Dyn1K694A is impaired in self-assembly and hence specifically in assembly-stimulated GTPase activity [22] . Overexpression of dyn1K694A was shown to increase rates of CME [22] . ( 2 ) Dyn1S61D exhibits WT GTP binding , but reduced basal and assembly-stimulated GTP hydrolysis rates [26] . Overexpression of dyn1S61D was shown to reduce rates of CME [26] . ( 3 ) Dyn1T141A exhibits reduced GTP binding in the unassembled , basal state , but increased basal GTP hydrolysis rates . It also exhibits a slight increase in assembly-stimulated GTPase activity and overexpression of dyn1T141A slightly increases the rate of CME [26] . The effects of dynamin-2 knockdown and expression of these mutants on the relative contributions of the different subpopulations were minor compared to the effect of cargo overexpression ( Figure 4A ) ; however , after dynamin-2 knockdown we now detect a substantial increase in the fraction of long-lived ( >10 min ) , “persistent” CCPs that were negligible in control BSC1 cells ( Figures 4B , S1E , and S1F ) . Reintroduction of WT dynamin-1 ( dyn1WT ) or dyn1K694A reduced the number of persistent CCPs , whereas reintroduction of dyn1S61D or dyn1T141A mutants increased their numbers . In contrast to cargo overexpression , perturbations of dynamin function had dramatic effects on the lifetimes of CCP subpopulations . Knockdown of dynamin-2 to approximately 17% of endogenous levels ( see Figure S3E and S3F ) significantly ( KS-test , p < 10−10 ) increased the characteristic lifetime of productive CCPs ( Figure 4C and Table S1 ) , confirming a role for dynamin as a rate-limiting factor in CCV formation . Dynamin-2 knockdown also increased the characteristic lifetime of late abortive CCPs ( Figure 4C ) . This observation is consistent with its proposed role during early stages in CCV formation . After knockdown of dynamin-2 , overexpression of dyn1WT significantly decreased the characteristic lifetime of the productive population ( KS-test , p < 10−7 , Figure 4C , Table S1 ) , again consistent with dynamin controlling rate-limiting steps in CME . In addition , overexpression of dynWT decreased the lifetime of late abortive coated pits , suggesting a role for dynamin as a fidelity monitor that initiates rejection and disassembly of nonviable CCPs . Overexpression of the self-assembly–impaired dyn1K694A mutant also significantly decreased the characteristic lifetime of both the productive and late abortive subpopulations ( KS-test , p < 10−16 , p < 10−2 , respectively; Figure 4C ) . These data suggest that unassembled dynamin functions early , that this assembly-impaired mutant stimulates CME by enhancing the rate of CCP maturation , and that dynamin self-assembly and subsequent assembly-stimulated GTPase activities per se are not rate-limiting for CCV formation [22] . Overexpression of the GTPase-defective dyn1S61D was unable to fully restore the rate of productive CCV formation in dynamin-2 siRNA-treated cells ( Figure 4C ) , and indeed increased the lifetimes of late abortive and productive CCPs even when overexpressed in the presence of endogenous dynamin-2 ( unpublished data ) . Lastly , overexpression of dyn1T141A was unique in that it differentially affected the abortive and productive lifetimes ( Figure 4C ) : the lifetimes of abortive CCPs remain slightly increased relative to control cells ( KS-test , p = 0 . 08 ) , whereas the lifetime of productive CCPs became shorter ( KS-test , p = 0 . 008 ) . Together with the dyn1S61D findings , these results demonstrate that GTP binding and hydrolysis in the basal state are required for an early surveillance function of dynamin and that the basal rate of GTP hydrolysis might be rate-limiting for maturation towards the productive CCP subpopulation . These data provide direct evidence that dynamin plays a dual role in CCP maturation and vesicle budding . To provide further evidence for dynamin's role in CCP maturation , we extracted the intensity time courses of the LCa-EGFP signal during CCP maturation . For this purpose , we focused our analysis on a subset of long-lived , isolated CCPs , which are highly likely ( >99% ) to represent productive events . The typical intensity time course is skewed ( see example in Figure 5A ) and can be divided into three distinct phases: ( 1 ) an “assembly phase” corresponding to the initial fast increase of signal intensity , which occurred during the first approximately 20 s of detectable clathrin lattice assembly; followed by ( 2 ) a “maturation phase , ” during which LCa-EGFP signal intensity plateaus or increases only moderately; followed by ( 3 ) a “departure phase , ” characterized by a sudden final drop of signal intensity . To measure the duration of these phases , the intensity time courses of individual CCP trajectories were fitted with a smoothing spline ( Figure 5A ) to identify the points where the approximated slope drops below a set threshold . siRNA-mediated knockdown of dynamin-2 markedly increased the duration of the maturation phase ( t-test , p < 10−10 ) , without significantly altering assembly or departure phases ( see Table S3 ) . These data are consistent with a role for dynamin in regulating rate-limiting steps in CCP maturation and with the fact that dynamin-mediated membrane fission is not rate-limiting for CME [22] . The average length of the assembly and departure phases ( Figure 5B ) were largely unaffected by reintroduction of the various dynamin mutants . In contrast , the length of the maturation phase increased and decreased depending on the GTP binding and hydrolysis activities of the reintroduced dynamin . Specifically , re-expression of dyn1WT or dyn1K694A mutants decreased , whereas re-expression of GTPase-defective dyn1S61D or dyn1T141A increased the duration of the maturation phase . These findings support a role for the basal GTP binding and hydrolysis activities of dynamin in CCP maturation .
We report a comprehensive and unbiased analysis of CCP dynamics in living cells . This was accomplished using a new particle tracking algorithm that defines the correspondence between CCP images in consecutive frames based on spatial and temporal global optimization [10] , which allowed us to reliably follow the fate of CCPs in areas of both low and high pit density . The algorithm incorporates a gap closing scheme that permitted tracking of faint and temporarily unstable CCPs . The performance of this algorithm was extensively validated for its application to CME analysis [10] . To capture the sub-second scale events of CCP formation and the much slower events of CCP maturation , we merged lifetime data from high frequency , shorter time-lapse videos with lower frequency , longer time-lapse videos . Thus , we tracked tens of thousands of both short-lived and long-lived species for each experimental condition without biasing the selection of CCPs to a subpopulation with a specific characteristic , e . g . , only isolated or bright pits . The large sample number enables application of statistical model-selection methods to determine the minimum number of subpopulations necessary to accurately fit the measured lifetime distribution . Indeed , application of these statistical methods requires n > 5 , 000 , a criterion met in each of our analyses , but which greatly exceeds the 100–600 selected events previously assessed in other studies [1–5] . Importantly , subsequent molecular perturbations identified certain conditions in which the contribution of subpopulations to the overall lifetime distribution changed while their time constants were unaffected , whereas other conditions left contributions unchanged while significant shifts occurred in the time constants . This indicates the orthogonality of the two parameters we extracted and establishes that they can be independently determined to probe distinct mechanistic aspects of CCP maturation . In control BSC1 cells expressing LCa-EGFP , we detected three CCP subpopulations ( early abortive , late abortive , and productive ) , with distinct time constants ( ∼5 s , ∼15 s , and ∼90 s , respectively ) . A previous analysis of CCP dynamics in the same cells suggested an average lifetime of approximately 46 s , assuming a single population of CCPs [2] . Taking into consideration the different contributions of these subpopulations to the entire ensemble of CCPs , we obtain a value of 39 s , consistent with this previous data . We suspect that the slightly lower ensemble lifetime in our measurements may be associated with a more systematic exclusion of the very faint , short-lived early abortive CCPs in the previous study [2] . The longer CCP lifetimes ( 60–90 s ) consistently reported by others [1 , 6] reflect selection and analysis of a single subset of typically productive CCPs . The functional assignment of productive CCPs rested on the agreement of the lifetimes of the long-lived subpopulation with biochemical rates of TfnR uptake , as well as with the lifetimes of LCa-EGFP structures tracked manually in quasi-simultaneous TIRF- and epifluorescent images showing CCV internalization . Therefore , we interpret the short-lived CCPs to represent abortive events , but it is also conceivable that they could represent clathrin-coated intracellular structures , e . g . , endosomes , that transiently move as visitors through the evanescent field of the TIRF microscope . However , for the following reasons , we think this possibility is unlikely: ( 1 ) Although we occasionally observe fast-moving visitors in LCa-EGFP–labeled cells , their displacement between consecutive frames is so much above average that particle correspondences are generally outside the tracking algorithm's self-adaptive search range [10] . Thus , the trajectories of visitors are typically fragmented into short sub-trajectories ( less than three frames ) , similar to trajectories associated with false-positive detections . To exclude both types of false structures from the lifetime analysis , trajectories shorter than five frames were removed from the dataset ( see Materials and Methods ) . ( 2 ) Early and late abortive events detected with LCa-EGFP are also detected by our statistical model selection following σ2-EGFP , a selective marker of plasma-membrane–associated CCPs . In addition , the early abortive σ2-labeled CCPs have virtually the same lifetime as early abortive LCa-labeled CCPs , giving us confidence that these are bona fide plasma membrane–associated structures . ( 3 ) The relative contributions of both abortive and productive CCPs are affected by transferrin receptor overexpression , and their lifetimes are affected by dynamin . There is no reason why these parameters should be affected for visitors unrelated to CCPs . ( 4 ) The strongest indication that the vast majority of the structures we have studied are plasma membrane–associated CCPs comes from the AP2 depletion experiments in which the numbers of all three subpopulations are proportionally decreased . This would not be expected if the shorter-lived species were derived from internal membranes . Furthermore , AP2 depletion prevents the shift to productive CCPs induced by TfnR overexpression . Thus , although it is possible that there is a minor contamination of CCPs by clathrin-coated internal membrane vesicles or clathrin-coated nonendocytic structures , their contribution appears not to be significant enough to affect our findings . Having identified three kinetically distinct subpopulations of CCPs , we next showed that their relative distributions and lifetimes could be affected by systematically manipulating cargo concentration , adaptor protein levels and the level and activity of the GTPase dynamin . Based on the shifts in contribution and lifetime of the three subpopulations we propose that CME might be governed by an endocytosis checkpoint or restriction point , which is regulated , in part , by dynamin . The following observations support the existence of this checkpoint: ( 1 ) the identification of abortive and productive CCPs ( see also [2] ) , ( 2 ) the finding that AP2-containing ( and presumably cargo-enriched ) CCPs are more likely to be productive , ( 3 ) the finding that cargo load enhances the efficiency of CCV formation leading to more productive CCPs at the expense of abortive ones , and ( 4 ) the finding that this effect of cargo concentration is dependent on or limited by AP2 adaptor concentrations . Progression through a restriction or checkpoint requires the tight interaction of a monitor and an activator system . As the major GTPase involved in CME , dynamin was a prime candidate to regulate the endocytosis checkpoint . Two models have been proposed for dynamin function in endocytosis: as a regulatory molecule [27] and as a component of the fission machinery [28–30] . However , these are not mutually exclusive , and recent data support aspects of both [23–26] . Our data on the effects of dynamin depletion and dynamin mutants on CCP dynamics provide several lines of evidence that unassembled Dyn•GTP acts early and controls progression through the endocytosis checkpoint: ( 1 ) siRNA reduction of dynamin decreases the rate of both productive CCV formation and turnover of abortive CCPs , whereas overexpression of WT dynamin accelerates each of these rates; ( 2 ) a mutant defective in self-assembly ( K694A ) that is predicted to increase cellular levels of unassembled Dyn•GTP further increases the rates of abortive CCP turnover; and ( 3 ) dynamin GTPase domain mutants predicted to be defective in basal GTP binding ( T141A ) or hydrolysis ( S61D ) selectively reduce the rates of abortive CCP turnover . Importantly , these conclusions rely on the use of well-characterized , hypomorphic dynamin mutants that mildly alter the kinetics of CME , yet have robust and readily detectable effects when assessed by large-scale image analysis . Strong dominant-negative dynamin mutants that stop CME lead to the accumulation of aggregated CCPs , thus limiting their usefulness for mechanistic interpretation . Dynamin is also positioned to act as a monitor of factors that satisfy restriction point requirements through its many SH3 domain–containing binding partners . These have additional domains that recognize coat proteins ( e . g . , amphiphysin , SNX9 , ) membrane curvature ( e . g . , amphiphysin , endophilin , SNX9 ) , and/or cargo molecules ( e . g . , SNX9 , grb2 , TTP ) . It is known that these proteins can differentially affect dynamin's basal GTPase activity and assembly properties [31–33] . Hence , they provide a potential mechanism for regulating dynamin function in response to these parameters of CCP maturation . Dynamin has also been shown to interact with auxilin and hsc70 [34] , thus providing a potential mechanism for the dynamin-dependent turnover of abortive CCPs that we have observed . A model describing these results is illustrated in Figure 6 . In this model , productive CCP formation is a stochastic event initiated by the cargo-independent association of AP2 at the plasma membrane , which nucleates clathrin assembly . If a critical mass of the additional factors required for CCP stabilization is not reached during this assembly phase , these structures , which correspond to early and late abortive CCPs , fail to pass through the restriction point and disassemble . We propose that dynamin regulates the checkpoint and controls entry into and progression through the CCP maturation phase . The basal GTP binding and hydrolysis activities may enable unassembled dynamin to function either as a sensor or timer of CCP assembly , through its SH3 domain–containing partners , and thus directly or indirectly control the termination or progression of early endocytic intermediates . Further work will be needed to test this hypothesis and to determine the mechanistic details of how dynamin may monitor CCP assembly and maturation . In sum , we propose that the presence of sufficient cargo , a threshold concentration of AP2 adaptors and perhaps other parameters such as the recruitment of endocytic accessory factors , acquisition of membrane curvature , etc . , are detected by dynamin to signal progression beyond the endocytosis checkpoint . Whereas three kinetically distinct subpopulations are detected with statistical significance in our analyses , the lifetime distributions—particularly of the productive population—remain very broad . Thus , we expect that there are other aspects of functional heterogeneity and other factors regulating the endocytosis checkpoint masked within this subpopulation . Future studies involving mild perturbation of other endocytic accessory factors together with comprehensive quantitative analysis of CCP dynamics should provide further mechanistic insight into this functional heterogeneity .
TIR-FM was performed on BSC1 monkey kidney epithelial cells stably expressing rat brain clathrin LCa-EGFP or the AP2 rat brain σ2-adaptin fused to EGFP ( provided by Dr . T . Kirchhausen , Harvard Medical School , and described in [2] ) using a 100 × 1 . 45 NA objective ( Nikon ) mounted on a Nikon TE2000U inverted microscope ( Nikon ) . CCP lifetimes range from a few seconds to several minutes . To fully capture this range of dynamics would require image sampling over minutes at a high frame rate ( <1 s ) . Such exposure leads to significant photobleaching and also substantially impairs cell viability , both affecting the accuracy of lifetime measurements . To avoid these problems , we applied a multi-timescale imaging approach . For each experimental condition , three to nine videos with a frame rate of 0 . 4 s/frame were acquired over >200 s , and five to 21 videos with a frame rate 2 s/frame , each from different cells , were acquired over approximately 10 min , using a 14-bit mode operated Hamamatsu Orca II-ERG camera . CCPs were detected using à-trous wavelet transform decomposition of the image [35] . Tracking of CCP was accomplished using spatially and temporally global particle assignment described in detail elsewhere [10] . The histograms of CCP lifetimes extracted from the two TIR-FM video categories were merged for the final lifetime analysis by normalizing the relative contribution of the CCP population with a lifetime in the range 30 to 50 s . Thus , the integrals of the measured lifetime probability density functions gi , [0 . 4] from all N[0 . 4] videos sampled at 0 . 4 s/frame and the integrals of the measured lifetime functions fj , [2] from all M[2] videos sampled at 2 s/frame were set to equal values: From the merged lifetime histograms , the underlying distributions of multiple CCP populations with different lifetime dynamics were extracted using both parametric and nonparametric model selection as described in detail in Text S1 . For the intensity analysis , we extracted the trajectories of CCPs that were long-lived ( >60 s ) and isolated ( no nearest neighbors within six pixels ) ; this criterion ensured that the chosen CCPs were in fact “productive , ” i . e . , that they underwent full maturation and internalization , and that there was no crossover to neighbors , both in terms of the physical material of the CCP and in terms of the tracked CCP trajectories , both of which can lead to artifacts in the intensity time course . Text S1 contains a more detailed description of methods including particle tracking , lifetime analysis , cell culture , adenoviral infection , cell fractionation , western blotting , immunofluorescence , siRNA transfection , and biochemical measurement of endocytic uptake . Tables S1 and S2 summarize the mean lifetimes and relative contributions of CCP subpopulations in each experimental condition . Table S3 summarizes the intensity phase data . Figure S1 shows data relating to the validation of tracking and lifetime analysis . Figure S2 shows three statistical methods used to identify CCP subpopulations . Figure S3 shows the effect of TfnR overexpression on ( 1 ) endocytosis efficiency and ( 2 ) subcellular distribution of AP2 and clathrin , and a western blot of dynamin knock-down and corresponding quantification . Videos S1–S6 show examples of particle detection and tracking in simulated videos ( Video S6 ) and those obtained imaging live cells ( Videos S1–S5 ) . | Clathrin-mediated endocytosis is the major pathway for the uptake of molecules into eukaryotic cells and is regulated by the GTPase dynamin . Adaptor proteins recruit clathrin to the plasma membrane , where clathrin-coated pits capture transmembrane cargo molecules , again via adaptors . The pits invaginate and pinch off to form clathrin-coated vesicles that carry the cargo into the cell . Live cell imaging has revealed striking heterogeneity in the dynamic behavior of clathrin-coated pits associated with the plasma membrane , yet the nature of this heterogeneity and its functional implications are unknown . We used particle-tracking software to establish an unbiased and complete inventory of the trajectories of clathrin-coated pits visible by total internal reflection fluorescence microscopy . Through statistical analyses , we identified three dynamically distinct subpopulations of coated pits: two short-lived subpopulations corresponding to aborted intermediates , and one longer-lived productive subpopulation . The proportion of each subpopulation and their lifetimes respond independently to molecular perturbations . As a result of systematic modulation of cargo concentration , adaptor levels , and analysis of dynamin mutants , we postulate the existence of an endocytic restriction or checkpoint that governs the rate of clathrin-mediated endocytosis by gating the maturation of clathrin-coated pits . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"cell",
"biology"
] | 2009 | Cargo and Dynamin Regulate Clathrin-Coated Pit Maturation |
During various inflammatory processes circulating cytokines including IL-6 , IL-1β , and TNFα elicit a broad and clinically relevant impairment of hepatic detoxification that is based on the simultaneous downregulation of many drug metabolizing enzymes and transporter genes . To address the question whether a common mechanism is involved we treated human primary hepatocytes with IL-6 , the major mediator of the acute phase response in liver , and characterized acute phase and detoxification responses in quantitative gene expression and ( phospho- ) proteomics data sets . Selective inhibitors were used to disentangle the roles of JAK/STAT , MAPK , and PI3K signaling pathways . A prior knowledge-based fuzzy logic model comprising signal transduction and gene regulation was established and trained with perturbation-derived gene expression data from five hepatocyte donors . Our model suggests a greater role of MAPK/PI3K compared to JAK/STAT with the orphan nuclear receptor RXRα playing a central role in mediating transcriptional downregulation . Validation experiments revealed a striking similarity of RXRα gene silencing versus IL-6 induced negative gene regulation ( rs = 0 . 79; P<0 . 0001 ) . These results concur with RXRα functioning as obligatory heterodimerization partner for several nuclear receptors that regulate drug and lipid metabolism .
In a variety of acute and chronic illnesses , including bacterial or viral infection , tissue injury , many chronic diseases and most cancers , proinflammatory cytokines such as interleukin ( IL ) 6 , IL-1β , and TNFα evoke a major reorganization of hepatic gene expression resulting in the massive synthesis of acute phase proteins such as C-reactive protein ( CRP ) [1] . It has long been known that under such conditions the drug metabolism capacity and other hepatic functions can be impaired , largely due to strong and broad downregulation of most drug metabolizing enzymes and transporters ( DMET ) at the transcriptional level [2–4] . As 60 to 80% of all used drugs are extensively metabolized in the liver [5] , these changes may lead to unrecognized drug overdosing and adverse events especially for drugs with narrow therapeutic index , including many cardiovascular , anti-cancer and central nervous system drugs [6–9] . DMET genes are regulated at the constitutive level by hepatic nuclear factors ( HNF ) such as HNF-1α , HNF-4α , and CCAAT-enhancer binding proteins ( C/EBPs ) [10 , 11] , while inducible expression involves several ligand-activated receptors including the aryl hydrocarbon receptor ( AhR ) , the constitutive androstane receptor ( CAR ) , the pregnane X receptor ( PXR ) , the peroxisome proliferator-activated receptor-α ( PPARα ) and others , which function as pleiotropic sensors for a large variety of endogenous and xenobiotic compounds [12 , 13] . The potential involvement of several of these transcription factors in the downregulation of DMETs by proinflammatory cytokines has been suggested in numerous reports on both mouse and human model systems [14–18] . Taken together , current evidence indicates that the downregulation of hepatic DMET genes by proinflammatory cytokines involves intense crosstalk between signaling components and the transcriptional machinery , potentially involving several and overlapping receptor-dependent mechanisms . Some authors also suggested coordinated mechanisms , e . g . involving the major hepatic retinoid X receptor , RXRα , which is required as a heterodimerization partner for several nuclear receptors including CAR , FXR , LXR , PPAR , PXR , and VDR [19 , 20] . Further upstream , the signaling pathways involved in DMET regulation also remained largely unclear . IL-6 is known to activate janus kinase/ signal transductors and activators of transcription ( JAK/STAT ) , mitogen activated protein kinase/ extracellular regulated kinase ( MAPK/ERK ) , and phosphoinositide 3 kinase ( PI3K ) /AKT pathways [21 , 22] . Earlier work has shown that downregulation of the major human drug metabolizing cytochrome P450 , CYP3A4 , in response to IL-6 occurs independently of the JAK/STAT pathway [14] , although it remained unknown whether this also applies to other DMET genes . On the other hand there is evidence that MAPKs are able to phosphorylate nuclear receptors , which may lead to their subcellular relocalization [19 , 23] , and PI3K/AKT may induce nuclear translocation of NF-κB , which has been shown to antagonize nuclear receptor function by mutual repression as well as by direct binding of NF-κB to DMET promoter regions [24] . To enhance understanding of the complex interactions within signaling pathways and transcriptional networks , different kinds of systems biology modeling techniques have been increasingly employed [25–30] . The most prominent types of logical models are Boolean models , which permit individual components to be only in active or inactive state , thus allowing only a qualitative description of the input-output behavior of signaling pathways . While very large Boolean models can be constructed , they are often not adequate for describing biological reality . By contrast , logic-based ordinary differential equation ( ODE ) modeling enables a more quantitative simulation of signaling dynamics over time [27] . However , the requirement for extensive time-resolved experimental data as well as prior knowledge about the involved signaling mechanisms limits application of ODE modeling to small networks . An intermediate alternative is provided by “fuzzy logic” , a highly flexible methodology that enables system component states to be in a continuous interval . Recent studies have shown that fuzzy logic can be applied to complex biological problems . Some studies established the use of fuzzy logic to convert prior ( e . g . , literature-based ) knowledge networks to computable models that can be trained to multi-factorial experimental data in order to understand complex signaling pathways [28–30] . Here we used primary human hepatocytes ( PHH ) stimulated by IL-6 , the most potent mediator of the acute phase response in liver , to characterize cellular responses in high-throughput quantitative gene expression and ( phospho- ) proteomics data sets . Using a previously developed large-scale Boolean model of IL-1 and IL-6 signaling [22] and extensive literature survey we constructed a fuzzy logic model comprising IL-6 signal transduction and DMET gene regulation . Selective inhibitors used in perturbation experiments to disentangle JAK/STAT , MAPK , and PI3K signaling pathways generated the necessary data for model training . Our approach suggests a major role of MAPK and PI3K pathways with the orphan nuclear receptor RXRα at a central position as link between inflammatory signaling and downregulation of drug detoxification genes . We finally validated these findings by RXRα knock-down experiments . This study emphasizes fuzzy logic modeling as a useful alternative to elucidate complex signaling interactions .
We first constructed a prior knowledge network ( PKN ) comprising IL-6 signal transduction and downstream gene regulation ( S1 Fig ) . The core of the signal transduction part of the network was taken from the Boolean model by Ryll and colleagues [22] , which comprises several signaling pathways . As we were primarily interested in identifying logical nodes and not in dynamic features , we did not attempt to include a time scale . This allowed us to simplify the model by removing feedback loops , including those involving SOCS1 and SOCS3 , without negatively impacting our model , because feedback loops by definition require a time scale . Furthermore , we deleted several input and output nodes not relevant for our study . The AND , OR , and NOT gates of the remaining network were transformed into activating or inhibiting transitions from the respective input species of the gate to the output species . All changes introduced to the model by Ryll et al . are summarized in S1 Text . The IL-6 signal transduction module of the PKN was supplemented with a gene regulation module by compiling biological knowledge from various cell types as provided by databases and scientific literature , including e . g . , BIOBASE TRANSFAC and Pubmed ( S2 Text ) . The resulting network contained all DMET genes including their transcriptional regulators STAT3 , NF-κB , AhR , HNF-1α , HNF-4α , ELK1 , glucocorticoid receptor , and cFOS , as well as a species RXR/NR , which represents the complexes between RXRα and any of the nuclear receptors known to be a potential partner of RXRα . We used PHHs because these cells are considered the “gold standard” model for the investigation of hepatic metabolism of drugs and its regulation at the cellular level [31 , 32] . Despite considerable inter-individual variability and limitations in availability , PHHs are superior to immortalized cell lines , whose de-regulated cell cycle control is a result of massive changes in mitogenic and apoptotic signaling , and to primary mouse hepatocytes , whose genomic response poorly resembles that of humans , especially during an inflammatory response [33] . It should be pointed out that the limited availability of PHH , their rather short live-span in the fully differentiated state , and the lack of appropriate cryopreservation protocols preclude the possibility to perform complete experiments or replications in the same donor . We determined signaling pathway activation upon IL-6 stimulation in PHH of three independent donors ( donors D1-D3 , Table 1 ) by quantification of a large panel of phosphoproteins using reverse-phase protein array ( RPA ) technology ( Fig 1A ) . Among the 32 detected phosphoproteins , consistently induced phosphorylations ( i . e . at 10 and 30 min after stimulation ) of AKT , c-JUN , ERK1/2 , STAT1 , STAT3 , and STAT6 were observed , although not all were statistically significant ( Fig 1B–1E , left panel ) . Western blot analyses confirmed the RPA findings ( Fig 1B–1E , right panels ) . Thus , increased phosphorylation of AKT , ERK1/2 , STAT1 , and STAT3 at their respective phosphorylation sites was demonstrated , indicating activation of the associated signaling pathways . Gene expression upon IL-6 stimulation was measured after 24 h for major DMET genes as well as for genes indicating inflammation or activation of a specific pathway ( see S1 Table for a list of all measured genes ) . We used specific chemical inhibitors to selectively interfere with STAT3 , PI3K , and MAPK signaling as confirmed by RPA measurements ( Fig 2 ) . Gene expression analysis following chemical inhibitions of pathways were conducted in PHH from five liver donors ( donors D4-D8 , Table 1 ) . The resulting five gene expression data sets contained single inhibitions of STAT3 , PI3K , and MAPK as well as combinatorial inhibitions of STAT3 + MAPK and PI3K + MAPK , while the combined inhibition of STAT3 + PI3K rapidly induced cell death and was therefore not included in the analysis . In agreement with previous observations , IL-6 elicited a profound transcriptional downregulation of many genes of the drug detoxification system . Hierarchical clustering based on the log 2 linear fold change ( log2FC ) values shows major clusters of genes and treatments ( Fig 3 ) . One major gene cluster comprises CRP and SOCS3 , which were strongly upregulated , indicating IL-6-dependent activation of the acute phase response . Except for CYP2E1 , all CYPs as well as several important ATP-binding cassette ( ABC ) and solute carrier ( SLC ) transporters were downregulated upon IL-6 stimulation . The upregulation of CYP2E1 was not seen in all donors , causing a separation from the two other upregulated genes in the hierarchical clustering . The cluster of treatments shown on the upper part of Fig 3 consisted of treatments including a MAPK and/or a PI3K inhibitor . Most of the IL-6-induced effects were attenuated in this cluster . Single treatments with the STAT3 inhibitor clustered together with the five IL-6 treatments in the absence of inhibitors , although almost complete loss of STAT3 pY705 was demonstrated by RPA analysis ( see Materials and Methods ) . In conclusion , both MAPK and PI3K signaling appeared to play a more important role in IL-6 induced DMET gene regulation as compared to STAT3 . We used adapted routines of the R library CNORfuzzy [34] in order to create an optimized fuzzy logic model from the gene expression data and the prior knowledge network ( see Materials and Methods ) . The model was trained with the five perturbed gene expression data sets that contained single inhibitions of STAT3 , PI3K , and MAPK as well as combinatorial inhibitions of two pathways ( donors D4-D8 , Table 1 ) . The resulting model family ( Fig 4 ) illustrates the connections between signaling molecules and regulated genes quantitatively , with the line width of each transition representing the percentage of the optimized models ( N = 100 ) containing this particular transition . By far most of the transitions are connected to the RXR/NR complex species , which represents heterodimeric complexes between RXRα and a number of nuclear receptors including PXR , FXR , and others [20] . Additional striking nodes that connect to several regulated genes are identified as NFkB , HNF4A , and HNF1A . For a few genes the model further suggests the inhibition of the glucocorticoid receptor by MAPK , and inhibition of AHR by NF-κB . The involved transitions for these events are present in nearly all of the optimized models ( Fig 4 ) . Comparison of the predictions of the model family with the respective data showed marked agreement for most genes , as indicated by bright-colored fields ( Fig 5 ) . Some higher deviations between prediction and data were seen only for few genes in certain conditions , in particular SOCS3 and to a lesser extent some DMET genes , e . g . CYP2A6 and CYP2C8 , in the presence of STAT3 inhibitor ( darker colored fields ) . This may indicate unknown regulatory events not contained in the prior-knowledge network . In the case of SOCS3 , the deviations reflect induced measured levels while the model predicts baseline levels due to the assumed inhibition of STAT3 . Despite confirmed effective STAT3 inhibition ( vide supra ) , we cannot exclude the possibility that residual activity leads to “leaky” upregulation of SOCS3 , since this is one of the most strongly regulated STAT3 target genes . It should be pointed out that this discrepancy cannot be due to the missing feedback loop in our model , which precludes secondary effects , as mentioned above . The optimized fuzzy logic model ( Fig 4 ) suggested an important role of the RXRα/NR complexes . We used siRNA-mediated selective RXRα gene silencing to analyze the impact on gene expression of major DMET genes via high-throughput real-time qPCR analysis . RXRα protein expression was almost completely suppressed as demonstrated by Western blot analysis ( Fig 6A ) . Fig 6B illustrates the IL-6- and RXRα knock-down ( KD ) -induced gene expression changes in PHH from three independent donors ( D9-D11 , Table 1 ) . RXRα KD was further confirmed by more than 90% downregulation of RXRA mRNA . Upon IL-6 stimulation , APR genes were highly upregulated and a coordinated downregulation of major DMET genes was observed in all donors , similar to those described for donors D4-D8 ( Fig 3 ) . The impact of RXRα KD on expression of DMET and modifier genes was very pronounced with similar patterns compared to the effects of IL-6 . This visual impression was supported by Spearman correlation analysis , showing a highly significant correlation between the mean fold changes of IL-6 and RXRα KD treatments ( rs = 0 . 79; N = 86; P<0 . 0001; Fig 6C ) . Among the phase I metabolism genes , most of which were strongly and significantly downregulated , only CYP2E1 reacted differently , being downregulated by the RXRα KD while it showed ( nonsignificantly ) higher levels after IL-6 treatment . The phase II metabolism genes NAT1 , NAT2 , and SULT1A1 were also downregulated to similar extent . Of note , the transporters ABCB1 and SLC10A1 showed opposite regulation , being upregulated by RXRα KD and downregulated by IL-6 . Among the DMET modifiers , only AHR , ARNT , and PPARA expression was significantly impaired by IL-6 but not after the KD of RXRα . As expected , the RXRα KD experiment did not indicate significant induction of most acute phase genes .
In this study we investigated the response of primary human hepatocytes to stimulation with IL-6 , the most potent pro-inflammatory cytokine for the hepatic APR . Using quantitative gene expression and time-resolved ( phospho- ) proteomics data sets of unprecedented comprehensiveness , we optimized a fuzzy logic model comprising all known major IL-6 signal transduction pathways as well as a broad spectrum of DMET gene regulation pathways . Our model suggested a major role of MAPK and PI3K pathways with the orphan nuclear receptor RXRα playing a central role as link between inflammatory signaling and downregulation of drug detoxification genes . Experimental RXRα knock-down by RNA interference further substantiated a coordinating role of RXRα to downregulate a wide variety of drug detoxification genes during inflammation . Based on our high-throughput ( phospho- ) proteomic analysis only a few major signaling molecules demonstrated increased phosphorylation status following IL-6 treatment: AKT ( S473 ) , ERK1/2 , ( T202/Y204 ) , STAT1 , ( Y701 ) and STAT3 ( Y705 ) . Whereas activation of STATs as well as established APR factors and ERK1/2 by IL-6 was shown previously [21 , 35 , 36] , increased phosphorylation of AKT at S473 following IL-6 treatment has not been shown before to our knowledge . Of note , PHH cultures of several donors showed increased phosphorylation at AKTS473 already at the steady state prior to treatment . This may indicated drug- or disease-induced basal deregulation of PI3K signaling pathway in the hepatocytes of these patients [37] . Our chemical inhibitions , which had been confirmed to be effective at the used concentrations by RPA phosphorylation analysis , indicated that blocking STAT3 signaling pathway compromised the IL-6 effect on DMET mRNA expression only marginally and only for few DMET genes , in agreement with a previous study showing that STAT3 was not required for CYP3A4 downregulation [14] . Inhibition of the MAPK and PI3K pathways however markedly interfered with IL-6-induced DMET expression changes , particularly in combination ( Fig 3 ) . Thus , co-inhibition analyses of PI3K and MAPK as well as of STAT3 and MAPK signaling pathways abolished almost all IL-6-mediated effects on DMET gene expression , suggesting a higher relevance of MAPK/PI3K compared to the JAK/STAT pathway in mediating the IL-6 triggered effects . Some limitations of this approach should be noted . Chemical inhibitors may have unspecific effects [38] and may also activate NRs by themselves [39] . However , we believe that the concentrations used here were low enough to show primarily true effects on the intended pathways . Furthermore , extensive pathway crosstalk [21] poses general difficulties in the interpretation of such data . In order to elucidate the regulatory events responsible for IL-6 regulation of DMET gene expression , we developed a fuzzy logic model . This modeling technique avoids the requirement for estimating numerous kinetic parameters as in ODE modeling and allows more realistic approximations to biological systems compared to simple Boolean logic . Fuzzy logic modeling has been previously used in studies involving rather tedious manual calibration of model parameters [28] as well as parameter estimations with heuristic optimization routines [29 , 30] . In our study we applied the latter approach due to the lack of prior knowledge about model parameters . This required training of the model by experimental data . Morris et al . [29] introduced a method to train signal transduction pathways to protein data that we adapted to the use of gene expression data sets . As we combined several data sets for model training , we constructed a “mean” model over these data sets . A principal problem in this respect could be variability in the gene expression data throughout different donors . However , as shown by our datasets representing 8 individual donors ( Figs 3 and 6 ) , IL-6 effects on both APR and DMET genes were remarkably similar for all liver donors . In principle , donor-specific models could have been created by model calibration with respect to only the data for a specific donor . In order to obtain a reliable model , this would , however , require that all experimental perturbations are conducted in this donor , which is practically very difficult with PHHs . The model suggested the inhibition of the complexes of RXRα and nuclear receptors by MAPK and Nf-κB as the major event for the downregulation of most DMET genes by IL-6 . RXRα is required as heterodimerization partner for several important nuclear receptors including CAR , FXR , LXR , PPAR , PXR , and VDR [20] . A coordinating role of RXRα based on its biological function has been previously proposed [19] but only few mouse genes had been observed in that study and to our knowledge the hypothesis has not been tested for humans . Here we used siRNA-mediated gene KD in PHH to confirm the model-proposed role of RXRα in DMET gene downregulation . In three independent donors we observed highly similar patterns of regulation with comparably few interindividual differences ( Fig 6 ) , resulting in a highly significant correlation ( rs = 0 . 79 , N = 86; P<0 . 0001 ) between mean fold changes elicited by the two treatments . Regarding the underlying mechanisms , it had been shown previously that endotoxin leads to rapid loss of nuclearly localized RXRα , while RXRα mRNA levels were not affected [19] , which is in agreement with our findings ( Figs 6 and S3 ) . The detailed molecular events leading to RXRα inhibition remain to be investigated . Modulation of the phosphorylation status of nuclear receptors including RXRα has been proposed as a possible event in this process [19] . In conclusion , this study provides new insights into the coordinated negative regulation of DMET genes by the proinflammatory cytokine IL-6 . Using extensive datasets that characterize the activation of signaling pathways and the regulation of a broad range of APR and DMET genes in primary human hepatocytes we found that MAPK and PI3K/AKT signaling pathways appear to be more important than STAT3 signaling in mediating the response of DMET genes to IL-6 . A fuzzy logic model based on gene expression data sets from five different hepatocyte donors identified RXRα as a key player in downregulation of DMET gene expression , which was confirmed by gene silencing experiments . While previous fuzzy logic modeling approaches mainly focused on describing signaling events , our model also involves gene regulation . Hence , our study is a novel example for the elucidation of key gene-regulatory events from biological data and prior knowledge using fuzzy logic .
The use of PHH was approved by the local ethics committee and written informed consent was obtained from all donors ( number 025–12 , Ethics Committee of the Medical Faculty of the Ludwig-Maximilians-Universität München ) . William’s E Medium was purchased from Invitrogen Life Technologies ( Darmstadt , Germany ) . Fetal bovine serum ( FBS ) was from PAA Laboratories GmbH ( Pasching , Austria ) , human insulin from Sanofi ( Frankfurt , Germany ) , and hydrocortisone from Pfizer Pharma GmbH ( Karlsruhe , Germany ) . Hepes , L-glutamine , MEM non-essential amino acids ( NEAA ) , penicillin/streptomycin ( Pen/Strep ) , phosphate-buffered saline ( PBS ) , and sodium pyruvate were purchased from GIBCO ( Carlsbad , CA , USA ) . Bovine serum albumin ( BSA ) , dexamethasone , and dimethyl sulfoxide ( DMSO ) were from Sigma-Aldrich ( Steinheim , Germany ) , hydrocortisone from Pfizer ( Karlsruhe , Germany ) . Human recombinant intereukin-6 ( IL-6 ) was purchased from Promo Cell GmbH ( Heidelberg , Germany ) . Human recombinant interleukin 1β ( IL-1β ) and tumor necrosis factor α ( TNF α ) were purchased from Sigma-Aldrich ( Steinheim , Germany ) . All cytokines were reconstituted and stored as high concentration stocks according to manufacturer specifications . Chemical inhibitors were purchased from the following suppliers: LY294002 ( Merck , Darmstadt , Germany ) , U0126 ( Promega , Madison , WI , USA ) , and Stattic ( Sigma-Aldrich , Steinheim , Germany ) . Inhibitor stock solutions ( 20 mM each ) were prepared in DMSO . All TaqMan assays were purchased from Applied Biosystems ( Foster City , CA , USA ) . Silencer Select Pre-designed siRNA was purchased from Applied Biosystems ( Foster City , CA , USA ) . PHH were isolated from partial liver resections by collagenase digestion as described previously [32 , 40] . Donor data are shown in Table 1 . Isolated cells with a viability of more than 70% as determined via trypan exclusion test were seeded at a density of 4 × 105 viable cells/well onto BioCoat Collagen I Cellware 12-well culture plates ( Becton Dickinson , Bedford , USA ) in William’s E Medium , supplemented with 10% FBS , 100 U/ml Pen/Strep , 2 mM L-glutamine , 32 mU/ml human insulin , 1 mM sodium pyruvate , 1X NEAA , 15 mM hepes , and 0 . 8 μg/ml hydrocortisone . After 24 h , cells were equilibrated for another 24 h in cultivation medium , containing William’s E Medium , supplemented with 10% FBS , 100 U/ml Pen/Strep , 2 mM L-glutamine , 32 mU/ml human insulin , 0 . 1% DMSO , and 0 . 1 μM dexamethasone . Cells were maintained at 37°C in 5% CO2 throughout the experiment with the exception of the shipping period . All cells were cultured for a minimum of 48 h between isolation and treatment . Media were changed every 24 h . PHH were treated for up to 24 h with 10 ng/ml human recombinant IL-6 in PBS , supplemented with 0 . 1% BSA , or vehicle only ( PBS + 0 . 1% BSA ) . This concentration had been previously shown in various cell models to activate STAT3 and to induce CRP expression without being toxic [41 , 42] . Furthermore , maximal induction of acute phase protein mRNA expression including CRP and SAA1/2 was recently demonstrated by dose-response experiments in PHH [4] . For inhibition of signaling pathways , three specific chemical inhibitors were applied , targeting three major signaling proteins: LY294002 for PI3K ( upstream of AKT ) , U0126 for MEK1/2 ( upstream of ERK1/2 ) , and Stattic for STAT3 . LY294002 was shown to be a potent inhibitor of PI3K in hepatocytes , where concentrations of > 20 μM inhibited the enzyme’s activity by more than 90% [43] . U0126 is a selective inhibitor for MEK-1 and -2 [44] . It was shown to effectively inhibit wild-type MEK1 phosphorylation of ERK2 in concentrations between 20 and 100 μM in in vitro experiments [45] . Stattic is a selective inhibitor of the activation , dimerization , and nuclear translocation of STAT3 shown to inhibit STAT3 in vitro with an IC50 value after one hour of incubation of 5 . 1 ± 0 . 8 μM [46] . For inhibition , medium was aspirated and replaced by fresh medium containing one or a combination of chemical inhibitors in final concentrations of 10 μM ( Stattic ) and 50 μM ( LY294002 and U0126 ) . DMSO-treated cells served as control . After incubation for 1 h , cells were treated with IL-6 or vehicle as described above . Successful inhibition of signal propagation was assessed in PHH using phosphoproteomics RPA technology . IL-6-dependent AKT S473 , ERK1/2 , and STAT3 Y705 phosphorylation was confirmed to be nearly abolished by LY294002 , U0126 , and Stattic , respectively . KD of RXRα via Silencer Select Pre-designed siRNA ( P/N4392420 , #s12384; sense: UCGUCCUCUUUAACCCUGAtt , antisense: UCAGGGUUAAAGAGGACGAtg ) was carried out in PHH according to the manufacturer’s instructions . In short , transfection mix was prepared and after 20 min incubation at RT added , giving a total volume of 1 . 2 ml per well ( 12-well plate ) . Total RNA was isolated from PHH and HepaRG cells using the RNeasy Mini Kit , including on-column genomic DNA digestion with RNase free DNase Set ( Qiagen , Hilden , Germany ) . The RNA integrity ( RIN ) and quantity were analyzed with the Agilent 2100 Bioanalyzer using the RNA 6000 Nano Kit ( Agilent Technologies , Waldbronn , Germany ) . Only samples with a RIN value larger than 7 were used . Synthesis of cDNA was performed with 500 ng RNA using TaqMan Reverse Transcription Reagents ( Applera GmbH , Darmstadt , Germany ) . Quantification of expression of 95 genes was performed using Fluidigm’s BioMark HD high-throughput quantitative 96x96 chip platform ( Fluidigm Corporation , San Francisco , CA , USA ) , following the manufacturer’s instructions [47] . All used predesigned TaqMan assays are listed in S1 Table . The mRNA expression levels were normalized to the most stably expressed gene among a selection of housekeeping genes ( ACTB , GAPDH , GUSB , HMBS , POLR2A , RPLP0 , TBP ) by using the Normfinder Excel Add-in as described by Andersen and colleagues [48] . Relative gene expression changes were calculated using the delta delta Ct ( ΔΔCt ) method [49] . ΔΔCt values were calculated by subtracting the ΔCt value of the calibrator sample ( e . g . , PBS , 0 . 1% BSA-treated ) from the ΔCt of the experimental sample ( e . g . , IL-6-treated ) . As the Ct is on a log 2 scale , linear fold changes ( FCs ) were calculated as 2 ( -ΔΔCt ) . RPA technology and Western blot analysis were used for relative quantification of protein phosphorylations . In the RPA , pL amounts of protein mixtures are immobilized in a microarray format and the presence of specific target proteins is screened by using highly selective antibodies [50] . This technology allows for the simultaneous quantification of more than 100 proteins and phosphoproteins by direct two-step immunoassay using specific primary antibodies [51] . Proteins were isolated using the CLB1 lysis buffer . Sample preparation and measurements were carried out as described elsewhere [51] . Western blots of selected phosphoproteins and of RXRα were performed with total cell lysate ( 20 μg of protein ) . β-Actin staining served as loading control . Detection was performed with an Odyssey infrared imaging system . Details on the antibodies used can be found in S2 Table . Each DMET gene is represented by a vector of fold changes for all treatments . The R function heatmap . 2 [52] was used to create a heat map of the genes and treatments based on the logarithmized fold changes . The genes as well as the treatments are thereby clustered hierarchically with average-linkage clustering using Euclidean metrics [53] . CNORfuzzy is an add-on to the CellNOptR , which constructs a fuzzy logic model that enables the model species to be in a continuous state in the interval [0 , 1] . A state of 0 for a model species then represents inactivity of the species and a state of 1 the highest possible activity . States in between stand for intermediate activity levels of the species . This routine has been used with proteomic data [29] and was here applied to gene expression data . The two discrete states for model species in CellNetOptimizer ( CNO ) corresponding to an active and inactive species have proven to be suitable for modeling the activity of signaling molecules [54] . However , for most genes instead of this on-off-pattern , we rather expect a gene to have several activation states . Therefore , we use CNORfuzzy for creating our model of IL-6 induced DMET gene regulation . CNORfuzzy first tries to remove from the network all species that are neither measured nor perturbed in the experimental data , i . e . , the only species that are additionally maintained in the network are those that are necessary for logical consistency . The program then expands the PKN with possible AND-gates to supplement the already implemented OR-gates . For model inference a genetic algorithm that optimizes the mean squared error between model prediction and normalized experimental data was used . This algorithm fits transfer functions for each gate to the data . In the following model reduction step , gates that do not significantly affect the mean squared error ( MSE ) between model prediction and data are removed based on a chosen selection threshold that determines the maximum tolerated increase in the MSE , when a model is reduced by removing logic gates . CNORfuzzy thus reduces the network to a topology that is sufficient to explain the experimental data . The genetic algorithm for optimization and the following reduction procedure of CNORfuzzy were run 100 times , resulting in a family of optimized and reduced models . The mean number of parameters in the optimized model family depends on the chosen selection threshold ( S2 Fig ) . At a selected threshold of 0 . 01 , the average MSE for the 100 models was 0 . 013 and the mean number of parameters in the optimized model family was approximately 110 . Details on network compression and optimization are presented in the supplemental S3 Text . The methods of Cell Net Optimizer and CNORfuzzy are based on normalized data in the interval [0 , 1] . However , our gene expression data have a different structure and the normalization method is neither suitable for the Ct values , nor for the calculated fold changes . Therefore , we adapt the given method in order to enable a transformation of the fold change values into the interval [0 , 1] . We transformed all fold change values fci of a gene with the following Hill function , which depends on the Hill coefficient h and the value m standing for the midpoint of the normalization function: vi=fcihmh+fcih This Hill function is similar to the Hill function in the CNO routine [54] . The main difference to the normalization method provided by CNO is the lack of fold change computation in our approach , because our data already contained fold changes . The fold change values for genes downregulated by IL-6 are usually smaller than 1 in the experimental data , whereas for genes upregulated by IL-6 they are greater than 1 . In order to reasonably transform the values into [0 , 1] for both classes of genes , the midpoint of the Hill function m had to be chosen differently . An important aspect to consider for this transformation is that model simulation with CNORfuzzy only produces species states of 0 or 1 in the case of inactivity of IL-6 ( see S3 Text ) . Therefore , normalization of gene states for control treatments should also lead to values near 0 or 1 . For all genes downregulated by IL-6 , their gene activity after control treatment must be high compared to after IL-6 treatment and thus their normalized values should be close to 1 . To this end , m was set to 0 . 5 , which proved effective . For CYP2E1 , which is upregulated , we correspondingly set m = 2 , because the control treatments represent a comparably low activity in this case . The other genes upregulated by IL-6 ( SAA , CRP , SOCS3 ) showed large fold changes upon IL-6 stimulation and m = 2 was not suitable for their midpoint of the normalization function . Therefore , we set m to half of the mean fold change value of the IL-6 treatments over the data sets . For all genes h was set to 4 , which led to a transformation of the control fold change values ( 1 ) to a value near 0 ( for the genes upregulated by IL-6 ) or 1 ( for the genes downregulated by IL-6 ) . In this way , we ensured that the transformed data points were spread throughout the entire interval [0 , 1] . We also conducted model calibrations with modified values of m for the genes downregulated by IL-6 . Increasing m led to worse fitting results ( MSE of approximately 0 . 025 for m = 0 . 7 ) , while decreasing m produced fitting results with similar MSE but unsatisfying spread of the data over the interval [0 , 1] . Statistical significance of ( phospho ) proteomic and gene expression changes was analyzed by grouped t-test ( two-tailed ) . Spearman correlation coefficients ( rs ) were calculated for averaged fold changes from three independent experiments . Statistical significance was defined as P<0 . 05 . All statistical calculations were performed using GraphPad Prism software ( version 5 . 04; GraphPad Software Inc . , San Diego , CA ) . | During inflammation , circulating proinflammatory cytokines such as TNFα , IL-1ß , and IL-6 , which are produced by , e . g . , Kupffer cells , macrophages , or tumor cells , play important roles in hepatocellular signaling pathways and in the regulation of cellular homeostasis . In particular , these cytokines are responsible for the acute phase response ( APR ) but also for a dramatic reduction of drug detoxification capacity due to impaired expression of numerous genes coding for drug metabolic enzymes and transporters . Here we used high-throughput ( phospho- ) proteomic and gene expression data to investigate the impact of canonical signaling pathways in mediating IL-6-induced downregulation of drug metabolism related genes . We performed chemical inhibition perturbations to show that most of the IL-6 effects on gene expression are mediated through the MAPK and PI3K/AKT pathways . We constructed a prior knowledge network as basis for a fuzzy logic model that was trained with extensive gene expression data to identify critical regulatory nodes . Our results suggest that the nuclear receptor RXRα plays a central role , which was convincingly validated by RXRα gene silencing experiments . This work shows how computational modeling can support identifying decisive regulatory events from large-scale experimental data . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2016 | Coordinating Role of RXRα in Downregulating Hepatic Detoxification during Inflammation Revealed by Fuzzy-Logic Modeling |
The parasite Giardia lamblia must remain attached to the host small intestine in order to proliferate and subsequently cause disease . However , little is known about the factors that may cause detachment in vivo , such as changes in the aqueous environment . Osmolality within the proximal small intestine can vary by nearly an order of magnitude between host fed and fasted states , while pH can vary by several orders of magnitude . Giardia cells are known to regulate their volume when exposed to changes in osmolality , but the short-timescale effects of osmolality and pH on parasite attachment are not known . We used a closed flow chamber assay to test the effects of rapid changes in media osmolality , tonicity , and pH on Giardia attachment to both glass and C2Bbe-1 intestinal cell monolayer surfaces . We found that Giardia detach from both surfaces in a tonicity-dependent manner , where tonicity is the effective osmolality experienced by the cell . Detachment occurs with a characteristic time constant of 25 seconds ( SD = 10 sec , n = 17 ) in both hypo- and hypertonic media but is otherwise insensitive to physiologically relevant changes in media composition and pH . Interestingly , cells that remain attached are able to adapt to moderate changes in tonicity . By exposing cells to a timed pattern of tonicity variations and adjustment periods , we found that it is possible to maximize the tonicity change experienced by the cells , overcoming the adaptive response and resulting in extensive detachment . These results , conducted with human-infecting Giardia on human intestinal epithelial monolayers , highlight the ability of Giardia to adapt to the changing intestinal environment and suggest new possibilities for treatment of giardiasis by manipulation of tonicity in the intestinal lumen .
The parasitic protozoan Giardia lamblia is a major cause of diarrheal illness and infects millions of people per year , primarily via waterborne outbreaks in developed countries and long-term water contamination in developing countries [1] , [2] . Giardia trophozoites infect many hosts , including humans , preferentially colonizing the proximal small intestine , which is sparsely populated by other microbes in comparison to the rest of the intestinal tract [3] , [4] . Symptoms of giardiasis vary among patients , ranging from asymptomatic infections to malabsorption and severe chronic diarrhea . The health impacts of giardiasis can be severe , as chronic infection or reinfection may cause malnutrition and growth retardation [1] , [5] . Giardiasis is generally treated with metronidazole or other nitroimidazoles , which target the parasite's anaerobic metabolism [6] , as well as with furazolidone or quinacrine . However , these chemotherapeutic agents are not always available , fully effective , or tolerated [6] , [7] . While the exact mechanism of virulence is not well understood [5] , attachment of Giardia to the small intestine wall is a requirement , since unattached cells would simply pass through the intestine during peristalsis . Thus attachment , also a requirement for cell proliferation in the host , can be considered a virulence factor in giardiasis . Trophozoites in culture have been observed to attach and detach in less than one second [8] , [9] , and they are able to attach to a variety of surfaces , both biological and inert . Attachment itself has been studied extensively ( reviewed in [8] , [10] ) , but the contributing environmental factors remain unclear , as does an understanding of how Giardia are able to remain attached ( or reattach when necessary ) in the highly variable environment of the small intestinal lumen . In addition to resisting shear stresses resulting from peristaltic flow , Giardia must withstand variations in the osmolality and pH of the contents of the small intestine resulting from normal feeding behavior of the host . Osmolality is defined as the total concentration of solutes per kilogram of solvent . Strictly speaking , for a semi-permeable membrane that passes some solutes , such as a cell membrane , the most relevant measure is tonicity , or the concentration of membrane-impermeant solutes per kilogram of solvent . Osmolality in most organisms is tightly regulated , and in humans , serum osmolality is held at 290 mOsm/kg . In the upper digestive tract , however , osmolality and pH can vary considerably from fed to fasted states . The largest variability is in the antrum and the duodenum , where osmolality has been found to range from approximately 30% to more than 200% of serum levels in fasted and fed states [11] , and pH fluctuations below pH 5 and above pH 7 . 5 have been observed [11] , [12] . Both pH and osmolality stabilize somewhat in more distal portions of the small intestine [13]–[15] . To study the effect of osmolality and pH changes on Giardia lamblia attachment , we developed a flow chamber assay to monitor cell attachment during changes in media concentration and composition ( Figure 1 ) . The closed chamber prevented exposure to excessive oxygen , as Giardia are microaerophilic , and it allowed us to monitor attachment to both glass surfaces and intestinal epithelial monolayers under controlled flow rates , while quickly changing media . We tracked changes in the number of attached cells during the experiments to quantify effects on attachment . Our results show that Giardia rapidly detached from both glass and enterocyte monolayer surfaces when osmolality was altered from standard growth medium conditions . We found that the fraction of detached cells increased with the magnitude of osmolality change . Experiments with iso-osmotic solutions containing solutes that can and cannot pass through the cell membrane ( non-isotonic and isotonic , respectively ) indicate that tonicity is the critical factor causing detachment . Interestingly , we observed that Giardia are able to adapt to moderate changes in tonicity , indicated by the dependence of detachment on the change in tonicity rather than the absolute tonicity . Based on Giardia's adaptation to environments of different tonicities , we were able to vary solution tonicity periodically in a way that forced extensive detachment of Giardia from both glass and epithelial monolayers .
Giardia lamblia WBC6-strain trophozoites were cultured in 15-mL polystyrene tubes at 37C in growth medium [16] supplemented with bovine bile ( Sigma Aldrich , St . Louis , MO ) , adult bovine serum ( Biosource International , Camarillo , CA ) and penicillin/streptomycin/Fungizone ( Cambrex Bio Science , Baltimore , MD ) . Cultures were passaged three times per week . For experiments , cells were grown to confluency and detached with a 20-minute cold shock at 4C . Two experiments were performed to test the effect of hypo- and hyperosmotic media conditions on growth . First , to test cell viability after prolonged exposure to test media , Giardia were passaged as usual into growth media containing approximately 20–200% normal solute concentration , or osmolality ranging from 62 to 625 mOsm/kg . Second , to test viability of cells after brief exposure to these media , cells grown in normal growth medium were detached by cold shock , centrifuged at 1500 g , and resuspended in test media for three minutes . Cells were then centrifuged again and resuspended in normal growth medium . For both viability and growth controls , cells were counted using a hemacytometer after 24 hours of growth . Cells were also regularly monitored in tubes for confluency levels . C2BBe-1 cells , which are brush border expressing clones of Caco-2 intestinal epithelial cells [17] , were obtained from the American Type Culture Collection ( ATCC ) and cultured at 37C in Dulbecco's modified Eagle's medium ( DMEM , ATCC 30-2002 ) with 10% fetal bovine serum ( Mediatech , Herndon , VA ) , penicillin/streptomycin , and 0 . 01 mg/mL human apo-transferrin ( Sigma , St . Louis , MO ) . Culture medium was changed three times per week and cells were passaged at 80−90% confluency , approximately every six days . For experiments , C2BBe-1 cell monolayers were grown on glass coverslips for at least three weeks beyond confluency to allow complete differentiation and microvillus brush border growth [17] , [18] . C2BBe-1 cells were not used beyond passage 68 due to possible brush border instability [17] . To confirm brush border expression , fully differentiated C2BBe-1 cell monolayers grown on pieces of glass coverslips were gently washed with serum-free DMEM , fixed in 2% glutaraldehyde in 0 . 1 M Na cacodylate for 1 . 5 hours , rinsed with Na cacodylate buffer , post-fixed with 1% osmium tetroxide in 0 . 1 M Na cacodylate buffer for one hour , rinsed again , and dehydrated in an ethanol series . For coincubated samples , Giardia cells were centrifuged at 1500 g for five minutes and the pellet was resuspended in cysteine-buffered DMEM and added to tissue culture dishes with C2BBe-1 monolayers on glass coverslip pieces . Giardia were allowed to attach to the C2BBe-1 monolayer for 45 minutes , then both were fixed together using the procedure previously described . All samples were critical point dried in an AutoSamdri 815 drier ( Tousimis Research , Rockville , MD ) , mounted to stubs using carbon tape , sputter coated with 2 . 4 nm of platinum using a MED020 sputter coater ( Bal-Tec AG , Liechtenstein ) and viewed with an S-5000 scanning electron microscope ( Hitachi High Technologies America , Pleasanton , CA ) .
To determine the effect of solution properties on attachment , we exposed attached Giardia trophozoites to changes in osmolality , tonicity , and pH using a flow chamber assay . Media variations of approximately 60 to 620 mOsm/kg and pH 5 to pH 8 were selected to span the range measured in the human proximal small intestine [11]–[15] . We limited bulk flow rate to 0 . 3 mL/min , a rate that flushed unattached cells from the chamber but did not perturb attached cells . At significantly higher flow rates ( approximately 3 mL/min and higher ) , cells first oriented upstream into the flow and then were forced to detach by the shear stress of the flow ( data not shown ) . We found that when detachment of cells occurred , the response was very rapid , and detachment was complete within less than two minutes ( including media exchange time ) for a given population , so we limited test media presentation to three minutes unless otherwise noted . Longer experiments ( up to 60 min ) did not result in significant additional detachment unless the medium composition was not suitable for cell growth ( e . g . sucrose solutions or media lacking cysteine; data not shown ) . As a first test of osmotic stress response , we exposed cells to a range of dilutions and concentrations of standard complete growth medium . We monitored cell attachment numbers during the three-minute exposure period and found no significant effect on attachment for osmolality shifts smaller than ∼±30% from the baseline medium osmolality of 300−330 mOsm/kg . Above ∼30% shifts in osmolality ( osmolality below ∼230 mOsm/kg or above ∼430 mOsm/kg ) both hypo- and hyperosmotic solutions resulted in rapid detachment of a statistically significant fraction of attached cells ( Figure 2A , B ) . Detachment began within 10 seconds of exposure to test media and depended on the magnitude of osmotic change . In extremely hypo-osmotic media ( 62 . 5 mOsm/kg or lower ) , 97 . 7% ( SD 3 . 03% , n = 3 ) of cells detached from the glass surface , while in highly hyperosmotic media ( 620 mOsm/kg or higher ) , 80 . 2% ( SD = 18 . 5% , n = 3 ) of cells detached . Chi-square analysis found these to be significantly different ( p<10−6 for both ) . For those experiments in which cells detached , they did so with a time constant τ = 25 seconds ( SD = 10 sec , n = 17 ) when fitted to an exponential decay function . Control experiments showed that Giardia did not detach from glass on the three-minute timescale of our experiments in the presence of unmodified Giardia growth medium or when exposed to the iso-osmotic Dulbecco's Modified Eagle's medium ( DMEM ) used for culturing C2BBe-1 intestinal epithelial cells ( Figure 2C ) . In previous studies , the actin depolymerizing agent cytochalasin D was associated with detachment after 24 hours of exposure [19]–[21] . Here , iso-osmotic addition of cytochalasin D ( 0 . 1 mM ) to Giardia growth medium had no significant effect on attachment after three minutes of exposure ( Figure 2C ) . Finally , metronidazole ( 30 µg/mL ) , the current drug of choice for treatment of giardiasis [6] , also did not cause Giardia to detach from glass on the three-minute timescale ( Figure 2C ) . To test whether exposure to growth media of different osmolality caused detachment by killing cells , we conducted additional control experiments . We found that Giardia populations briefly exposed ( three minutes ) to hypo- or hyperosmotic growth media and then returned to normal growth medium had normal morphology and behavior under the microscope and grew at normal rates ( Figure S1A ) . This suggests that short-timescale osmolality shifts did not induce detachment by killing cells . However , cells incubated for 24 hours in hypo- or hypertonic media , though unaffected by small changes in osmolality , grew more slowly under moderate osmotic stress , and were not viable in the most extreme hypo- and hypertonic media tested ( Figure S1B ) . We also tested the effect of pH on detachment because it is known to vary considerably in the duodenum [11] , [15] and somewhat in the jejunum [13] , [14] . After three minutes , low pH ( pH = 4 . 94 ) medium had attachment levels of 98 . 7% ( SD = 1 . 1% , n = 3 ) , while high pH ( pH = 8 . 03 ) medium had attachment levels of 96 . 3% ( SD = 1 . 4% , n = 3 , Figure 2C ) . Even though these physiologically relevant changes in pH did not result in significant detachment , media pH during the time course of other experiments was held relatively constant . We next explored whether Giardia detachment behavior was dependent on the absolute osmolality of the solution or on the magnitude of osmotic change from one solution to the next . That is , would Giardia still react similarly to changes in solution osmolality like those documented above if they started out in a medium of different osmolality than normal growth medium ? To this end , Giardia were initially incubated for 10 minutes in growth media of slightly elevated or lowered osmolality ( by +79 mOsm/kg and −107 mOsm/kg , respectively ) . Cells were then exposed to a moderate osmotic shift in the opposite direction . For the cells incubated in +79 mOsm/kg medium , the test medium osmolality was 182 mOsm/kg , or 124 mOsm/kg below the incubation medium and 197 mOsm/kg below normal medium tonicity . For cells incubated in the –107 mOsm/kg medium , test medium osmolality was a normal 308 mOsm/kg . In both cases , when cells were subsequently exposed to these osmolality shifts , which were of an absolute osmolality that would not normally result in much detachment , a statistically significant percentage of cells detached . The amount of detachment depended on the osmotic difference between the incubation medium and test solution instead of on the absolute value of the incubation medium and test solution osmolality . That is , cells appeared to adapt to the osmolality of their incubation medium . As a result , detachment for the given osmolality change is similar to that for cells initially incubated in normal medium ( 300−330 mOsm/kg ) and exposed to the same magnitude of osmotic change ( Figure 3A ) . To examine whether Giardia detachment behavior depended on the specific composition of the medium , we tested several different solutions with a range of osmolalities . Surprisingly , the detachment trends observed in hypo- and hyperosmotic growth media ( Figure 3B , dashed line ) were also seen for solutions that do not support cell growth or survival ( Figure 3B ) . Pure NaCl and pure sucrose solutions induced no detachment of Giardia cells when the solution osmolarity matched that of growth medium . Detachment levels in both pure NaCl and pure sucrose solutions of varying osmolality were found to match results for those in growth medium , indicating that cell detachment is not an immediate response to nutrient deprivation . However , cells left in sucrose for a longer time period gradually began to detach after approximately 10 minutes of exposure , and long-timescale exposure ( >60 minutes ) resulted in cell death . The detachment-inducing effect of reduced-osmolality growth medium can be prevented by restoring diluted medium to its original osmolality with the addition of solutes , e . g . sucrose , before exposing cells to it ( Figure 3C ) . In this experiment , we diluted medium by 150 mOsm/kg , causing 36 . 9% of cells to detach after three minutes ( Figure 3C , column VI , p = 0 . 00001 ) in comparison to undiluted medium ( column I ) . However , supplementing diluted medium with a sufficient amount of sucrose to restore its original osmolality resulted in 99 . 6% ( SD = 3 . 1 , n = 3 , column II ) of cells remaining attached . Direct substitution of an iso-osmotic solution of either pure sucrose ( column III ) or pure NaCl ( column IV ) for growth medium also prevented detachment , with attachment at 100 . 6% ( SD = 2 . 8% , n = 3 ) and 103 . 1% ( SD = 2 . 4% , n = 3 ) of control levels , respectively . Thus , the observed detachment is not due to the absence of specific osmolytes in the test medium . However , restoration of medium osmolality from a dilute 162 mOsm/kg to 324 mOsm/kg with cell-permeant mono ( ethylene glycol ) , or MEG , resulted in detachment levels comparable to those in the diluted growth medium , with 28 . 8% of cells detaching ( column V , p = 0 . 009 ) . Thus , changes in media tonicity , not osmolality , cause detachment , where tonicity is the effective osmolality due to cell-impermeant osmolytes . This effect appears to be independent of medium composition for all media tested . We refer to the sudden detachment of Giardia in response to tonicity changes as “tonic shock” . To test whether the observed tonicity-dependent detachment was specific to glass substrates , we repeated key experiments with Giardia attached to a monolayer of C2BBe-1 cells , which are a brush border-expressing sub-clone of Caco-2 cells and are well established as a model for the intestinal epithelium [17] , [18] , [22] . Fluorescently-labeled Giardia attached readily to these monolayers ( Figure 4A ) and scanning electron microscopy confirmed that our method produced cell monolayers with dense microvilli ( Figure 4B ) to which Giardia can attach ( Figure 4C ) . Giardia attached to intestinal monolayers exhibited tonicity-dependent detachment behavior remarkably similar to that of Giardia attached to glass . Cells detached from the monolayers in both hypo- and hypertonic media , and the number of detached cells depended on the magnitude of tonic change ( Figure 5A , triangles ) . This was observed for both experiments conducted in Giardia growth medium and C2BBe-1 DMEM growth medium ( Figure 5A ) . As in the osmolality restoration experiments for Giardia on glass ( Figure 3C ) , the detachment response on C2BBe-1 monolayers was prevented only by isotonic media . In undiluted medium at 335 mOsm/kg , 95 . 8% ( SD = 4 . 12% , n = 3 ) of cells remained attached after experiments ( Figure 5B , column I ) , while medium diluted to 206 mOsm/kg ( column V ) dropped attachment levels to 44 . 9% ( p = 0 . 0007 ) . When this diluted medium was restored to its original osmolality with sucrose ( column II ) , 99 . 2% of cells remained attached . Similarly , an iso-osmotic pure sucrose solution ( column III ) resulted in 102 . 6% of initial attachment ( 100 . 9% and 104 . 4% , n = 2 ) . Finally , medium diluted by 140 mOsm/kg and then supplemented to iso-osmotic levels with cell-permeant MEG ( column IV ) failed to prevent detachment , with only 31 . 5% of cells remaining attached after exposure ( p = 0 . 00005 ) . As previously described , Giardia cells that remain attached after exposure to changes in medium tonicity appear to adapt to the new tonicity ( Figure 3A ) . We hypothesized that if these cells were truly adjusting to a different medium tonicity , this adaptation would render them sensitive to a small deviation from the original medium tonicity that was insufficient to cause detachment before adaptation . For example , cells that had previously adapted to a slightly hypertonic medium would be more sensitive to a small hypotonic challenge than if they did not adapt . The small hypotonic challenge , in turn , would make the cells more sensitive to a subsequent exposure to hypertonic medium if the cells adapted to the hypotonic medium . To test this idea , we designed an experiment in which Giardia attached to C2BBe-1 monolayers were exposed to hypertonic medium flow for three minutes , then left to adjust to the new tonicity for five minutes , then challenged with hypotonic medium flow for three minutes , then allowed to adjust again for five minutes . The tonicity fluctuation pattern was then repeated , alternating between hypo- and hypertonic values that were either ±100 mOsm/kg or ±150 mOsm/kg from the original growth medium tonicity . These tonicity values were selected to correspond with values producing only mild detachment in previous experiments ( Figure 2 ) , which are also consistent with osmolality fluctuations measured in the human small intestine [11] . For these experiments we used a PEG/electrolyte mixture as the medium base . Figure 6 shows that cells remaining attached do indeed adapt to the tonicity of the surrounding solution and that tonicity manipulations can force a larger fraction of cells to detach . In hyper- and hypotonic media with swings ±150 mOsm/kg from standard medium tonicity , cells detached in a stair-step pattern , and average attachment was reduced to 13 . 6% ( SD = 4 . 8% , n = 5 ) after 2 periods each of hypotonic and hypertonic medium exposure ( Figure 6 , Video S1 ) . In media ranging ±100 mOsm/kg from standard tonicity , detachment was more gradual , but over half of initially attached cells had detached by the third exposure to hypertonic medium ( 44 . 7% and 47 . 4% attachment , n = 2 ) . Cells incubated for an additional 24 hours prior to experiments appeared to detach more readily ( Figure 6 , dotted line ) , with average attachment dropping to 8 . 4% after one period each of hyper- and hypotonic medium exposure ( SD = 7 . 6% , n = 3 ) . In comparison , the constant tonicity control showed 81 . 6% attachment at this time point . Some detachment is expected in long timescale experiments since cells detaching for any reason are swept away by the flow . However , this gradual decline in cell numbers is clearly different from the stair-step pattern seen during the periodic tonicity swings .
Giardia trophozoites rely on attachment to stay within the host and must accommodate a large range of perturbations in flow and chyme composition . In these experiments , we used live cell microscopy to monitor immediate detachment events upon exposure to test media in a flow chamber . Short-timescale detachment of Giardia has not been examined in the numerous studies of forced detachment , as most assays measure population attachment levels after a period of 2−24 hours incubation with a given drug or medium composition ( e . g . [19]–[21] , [23] ) . Such studies have shown that Giardia attachment and survival rates are inhibited by changes in pH and osmolality [23] , but the long incubation prior to quantification of attachment makes it very difficult to separate direct effects on attachment from an overall effect on cell viability . Our results demonstrate that not only are Giardia susceptible to tonic shock , but that the response is rapid , with a detachment time constant of 25 seconds . This is consistent with the extremely rapid attachment and detachment behavior of trophozoites noted in the literature [8] , [9] . Notably , remaining cells appear to adapt to the new tonicity and are not susceptible to further detachment unless subjected to a second tonic shock . It is unclear whether the tonicity-dependent detachment we observe is because tonic changes directly affect Giardia's mechanism of attachment or whether detachment is a secondary result , perhaps due to observed osmotic shrinking and swelling of the cell . Attachment is associated with a unique microtubule-based structure called the ventral disk , which is located on the underside of the cell . Other studies have hypothesized several methods of attachment , including specific binding by lectins to the surface of intestinal cells ( reviewed in [10] ) . High resolution ( 100X ) phase contrast video microscopy of attached Giardia under hypotonic shock ( −196 mOsm/kg ) showed that detachment events are non-catastrophic , with affected cells simply “falling off” the glass surface ( Video S2 ) . Beating of the ventral flagella was sinusoidal as previously observed [24] , even during detachment , and cell morphology appeared unchanged . Given Giardia's ability to adhere to glass and other inert surfaces and the observation that detachment forces are unaffected by surface chemistry [25] , a pressure-based mechanism of attachment [26]–[28] appears most likely . Yet the major question of how pressure is generated remains open . We propose that the detachment behavior observed in this study is consistent with an osmotic , pressure-based mechanism of attachment in which cells create an osmotic pressure differential to generate attachment force . Osmolyte leak rate or a compromised seal around the ventral disk would be the primary modes of attachment failure under tonic shock . Although this model is consistent with our data , we know of no example in the literature of a single-celled organism that generates an osmotic gradient for attachment . Previous studies of the regulatory volume decrease of unattached cells showed that Giardia cells release alanine and potassium to decrease intracellular osmolality upon exposure to hypotonic media [29]–[31] . Regulatory volume changes are present in most eukaryotes and aid the cell in adjusting to a range of medium tonicities; if Giardia do indeed use an osmotic method of attaching to surfaces , the correspondence between volume changes and attachment in Giardia raises the question of whether Giardia evolved their robust attachment mechanism from an ancestrally conserved volume regulation mechanism or whether these mechanisms are otherwise coupled . The response of Giardia trophozoites to changes in media tonicity has several potentially important implications . Giardiasis is a disease with a global distribution but a disproportionate impact on developing countries , contributing to malnutrition problems and to diarrheal illnesses that are particularly dangerous to young children and the immunocompromised . Current drug treatment methods exist , namely the administration of metronidazole , but they are not always effective or tolerated . Additionally , Giardia are capable of multiple drug resistance in vitro [7]; drug-resistant isolates have been found in patients although epidemics of resistant strains have not yet been reported [6] . As such , a simple , cost-effective , safe treatment that is easily administered and effective against all strains of the parasite would be particularly useful . Our results demonstrate that sufficient changes in tonicity that are timed to take advantage of Giardia's adaptation mechanisms can strongly and rapidly promote Giardia detachment . Since the tonicity changes leading to detachment in our experiments are within the range measured in the human small intestine [11] , this method could be explored as a possible way to treat giardiasis or to enhance the effectiveness of present treatment methods . Ingestion of a low tonicity solution , such as water , followed by an interval that allows parasites to adjust to the lower lumenal tonicity , then followed by ingestion of a high tonicity solution , containing perhaps sucrose or PEG , may induce parasites to detach from the intestinal epithelium , as they did from human intestinal epithelial monolayers in our experiments ( Figure 6 , Video S1 ) . Our experiments used a PEG/electrolyte mixture approved for use in humans as the medium base to support the feasibility of this treatment possibility . One central question facing application of this approach is whether sufficiently rapid swings in osmolarity can be achieved in the human intestine . Our results provide clear evidence of Giardia's sensitivity to and ability to adapt to the changing chemical environment of the host intestine . If tonicity-dependent detachment of Giardia is confirmed in human or animal model studies , our proposed “tonic shock therapy” may provide a useful strategy for enhancing the effectiveness of current treatments , helping to relieve symptoms in patients unable to take metronidazole or other antigiardial medications , or combating drug-resistant parasite strains . | The single-celled organism Giardia lamblia colonizes the small intestine of a wide variety of hosts , including humans . Giardiasis infections can cause severe gastrointestinal symptoms and pose a major health concern in the developing world . Giardia are known to attach robustly to a variety of surfaces , but the conditions that influence this attachment are not known . In this study , we examined the behavior of attached Giardia parasites exposed to rapid changes in solution properties , like those Giardia might encounter in the intestine . After systematically varying media concentration and composition , we found that only one solution property caused rapid detachment of Giardia cells: tonicity , which is a measure of the total concentration of solutes in the solution that are unable to pass through a semi-permeable membrane ( here , the cell membrane of Giardia ) . We found similar results for Giardia initially attached to monolayers of intestinal cells . Giardia cells remaining attached after a change in tonicity are able to adapt to the change , highlighting the general ability of this organism to weather normal changes in the intestinal environment . We propose that Giardia's susceptibility to large changes in tonicity could be explored as a possible new route for treatment of giardiasis . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"infectious",
"diseases/protozoal",
"infections",
"gastroenterology",
"and",
"hepatology/gastrointestinal",
"infections",
"microbiology/parasitology"
] | 2008 | Tonic Shock Induces Detachment of Giardia lamblia |
The development of primordial germ cells ( PGCs ) involves several waves of epigenetic reprogramming . A major step is following specification and involves the transition from the stably suppressive histone modification H3K9me2 to the more flexible , still repressive H3K27me3 , while PGCs are arrested in G2 phase of their cycle . The significance and underlying molecular mechanism of this transition were so far unknown . Here , we generated mutant mice for the Mad2l2 ( Mad2B , Rev7 ) gene product , and found that they are infertile in both males and females . We demonstrated that Mad2l2 is essential for PGC , but not somatic development . PGCs were specified normally in Mad2l2−/− embryos , but became eliminated by apoptosis during the subsequent phase of epigenetic reprogramming . A majority of knockout PGCs failed to arrest in the G2 phase , and did not switch from a H3K9me2 to a H3K27me3 configuration . By the analysis of transfected fibroblasts we found that the interaction of Mad2l2 with the histone methyltransferases G9a and GLP lead to a downregulation of H3K9me2 . The inhibitory binding of Mad2l2 to Cyclin dependent kinase 1 ( Cdk1 ) could arrest the cell cycle in the G2 phase , and also allowed another histone methyltransferase , Ezh2 , to upregulate H3K27me3 . Together , these results demonstrate the potential of Mad2l2 in the regulation of both cell cycle and the epigenetic status . The function of Mad2l2 is essential in PGCs , and thus of high relevance for fertility .
In mice , PGCs are induced by BMP signaling at the onset of gastrulation at day 7 . 25 of embryonic development ( E7 . 25 ) in the posterior epiblast . They enter the extraembryonic mesoderm and the hindgut endoderm , and then migrate through the dorsal mesentery , until they accumulate in the genital ridges to participate in the generation of the future gonads [1] . Once specified , PGCs undergo various changes of their transcriptional profile and epigenetic status , which together establish the unique germ cell fate separate from surrounding somatic cells [2] , [3] . Two PR-domain containing proteins , Prdm1 ( Blimp1 ) and Prdm14 , initiate the PGC-specific program [4] , [5] . The reactivation of the pluripotency-associated gene Sox2 that had been silenced in the epiblast of the egg cylinder is an immediate early change upon PGC specification [6] , [7] . It leads to the acquisition of a potential to become pluripotent under specific culture conditions [8]–[10] . Around E7 . 5 the transcription of somatic genes like Hox , Snail or Brachyury become repressed as a result of Prdm1 function , and the characteristic PGC gene Dppa3 becomes upregulated . Together , the typical transcriptional signature of PGCs has developed by E9 . 0 [11] . The chromatin of PGCs undergoes extensive remodeling , affecting both DNA and histone configurations [3] , [12] . De novo DNA methylation is suppressed as the result of the downregulation of the DNA methyltransferases Dnmt3b and Uhrf1 [7] . Consequently , a passive DNA demethylation is initiated at around E8 . 0 , and by E9 . 5 , PGCs become hypomethylated [3] . At E7 . 75 , PGCs harbor a high , genome-wide level of the repressive histone modification H3K9me2 , similar to the surrounding somatic cells . This modification is gradually lost , and by E9 . 25 suppressed in most PGCs . The corresponding histone methyltransferases GLP and G9a , which methylate lysine residue 9 of histone 3 , are downregulated by E7 . 5 or E9 . 0 , respectively [11] , [13] . In parallel to H3K9me2 downregulation , H3K27me3 , a repressive histone modification providing more plasticity , accumulates in PGCs and finally replaces the H3K9me2 completely at E9 . 25 [2] , [3] , [11] . H3K27 trimethylation is catalyzed by Ezh2 , a subunit of the polycomb repressive complex 2 ( PRC2 ) , and downregulates the expression of typical somatic or differentiation related genes [14] , [15] . Ezh2 is subject to phosphorylation at different motifs by the cyclin dependent kinases Cdk1 or Cdk2 , which modulate the activity or stability of Ezh2 , and thus affect the level of H3K27me3 [16]–[18] . Cdk1/Cyclin B1-mediated phosphorylation of Ezh2 at threonin 487 ( pEzh2-T487 ) disrupts its binding to the other components of PRC2 complex , leading to its inactivation , and therefore to H3K27me3 attenuation [18] . It was previously shown that murine and porcine PGCs , and also PGCs derived in vitro from mouse embryonic stem cells arrest their cell cycle in a G2 phase briefly after their specification [11] , [19]–[21] . This phase , which is accompanied by transcriptional silence , may provide time for epigenetic reprogramming . So far , the molecular mechanism coordinating the epigenetic reprogramming and cell cycle prolongation in early PGCs is not clear . Mad2l2 is a chromatin binding protein involved in both cell cycle control and DNA repair [22]–[24] . Mad2l2 was previously described as an accessory , non-catalytic subunit of the translesion DNA polymerase zeta , and its knockdown led to hypersensitivity towards DNA damage [25] , [26] . Mad2l2 appears to function by binding to a diverse spectrum of proteins via its conserved HORMA domain . Several , but not all of these partners bind via the conserved sequence motif PXXXPP [27] . Reported binding partners include Cdh1 and Cdc20 , the substrate binding proteins of the APC/C complex , the two translesion polymerases Rev1 and Rev3 , the transcription factors Elk-1 and TCF4 , the clathrin light chain A , and others [23] , [24] , [28]–[32] . Accordingly , functions for Mad2l2 were previously claimed in such diverse processes as DNA repair , cell cycle control , and the regulation of gene expression . However , the biological significance of the reported interactions and activities remained unclear due to the lack of appropriate mouse mutants . In this work we describe a mouse mutant lacking the Mad2l2 gene . Embryos lose PGCs briefly after their specification , and do not proceed in epigenetic reprogramming . We investigated the function of Mad2l2 also by gain- and loss-of-function analysis in fibroblasts , and in biochemical assays . We suggest new functions of Mad2l2 as a regulator of epigenetic reprogramming , which is particularly relevant for primordial germ cells , and therefore required for fertility of males and females .
Low levels of Mad2l2 mRNA are widely expressed in adult and E14 . 5 embryonic cells , with a particularly high level in testis ( Figure 1A ) . High levels of Mad2l2 protein were detected in pachytene spermatocytes by immunohistochemistry ( Figure 1E ) , while the antibody did not lead to specific signals above background in other tissues , including PGCs . Significant amounts of Mad2l2 RNA were previously detected in E9 . 5 PGCs by microarray analysis ( NCBI database Gene Expression Omnibus GEO; Hayashi et al . , 2011 ) . A conditional knockout of the Mad2l2 gene in embryonic stem cells was generated and ubiquitously active Cre recombinase was introduced through breeding ( Figures S1A , B ) . Heterozygous Mad2l2 mutants were viable , healthy and fertile . Homozygous embryos and postnatal mice were significantly smaller than their littermates , but no morphological abnormalities were observed ( Figures S1C–F ) . Offspring before and after birth appeared in sub-Mendelian ratios , indicating a loss of embryos in midgestation ( Table S1 ) . Homozygous males and females were infertile , and gonads were significantly underdeveloped . Ovaries were not formed at all or were small organ rudiments that did not contain ovarian follicles or germ cells ( Table S2 and Figure 1B ) . Such structures may be indicative that some germ cells were present in the gonad during granulosa cell differentiation ( Figure 1B ) . Mutant testes were drastically smaller than control organs of the same age , and seminiferous tubules were devoid of spermatogonial cells ( detected by Plzf ) , pre-meiotic ( identified by Stra8 ) and meiotic cells ( detected by γH2AX; Figure 1C , D , F–H ) [33]–[36] . Leydig cells appeared hyperplastic , and Sertoli cells , identified by Wt1 , were mislocalized and highly vacuolated ( Figure 1I ) [37] , [38] . In summary , finding these deficiencies in both males and females suggested that developmental problems arose earlier during embryogenesis . For the determination of PGC numbers , embryos were collected at different time points during their early development , were staged as outlined under experimental procedures , and PGCs were identified by the presence of alkaline phosphatase ( AP ) or Oct4 ( Figure 2A ) [39] . At the early head fold ( EHF ) stage , the numbers of PGCs at the base of the allantois were similar in wild type , heterozygous and homozygous embryos . However , while the number of normal PGCs increased at the late head fold ( LHF ) stage , the number of Mad2l2−/− PGCs fell behind ( Figure 2B ) . It decreased drastically from E8 . 5 onward , and at E9 . 0 only few instead of normally ca . 120 PGCs were found in the hindgut endoderm . At E9 . 5 and E10 . 5 Oct4-positive PGCs were no longer detected ( Figure 2B ) . At E8 . 25 , both wild type and remaining mutant PGCs co-expressed Oct4 together with Prdm1 , Tcfap2c , and Dppa3 , indicating a normal specification of mutant PGCs ( Figure S2A , B , D ) . Oct4 and Sox2 were co-expressed in all wild type PGCs with no exception . In contrast , above 40% of Oct4-positive Mad2l2−/− PGCs did not express Sox2 at E9 . 0 , and thus had either failed to reactivate , or at least to maintain its expression ( Figure S2C ) . Emigration to the dorsal mesentery did not occur , and as a result , gonad primordia at E13 . 5 were devoid of germ cells ( Figure 2A ) . All E9 . 0 Mad2l2−/− PGCs had accumulated active , acetylated p53 protein , reflecting an activated stress response and impending apoptosis ( Figure S3A ) [40] . As judged by the TUNEL assay ( See Text S1 ) , some SSEA1-positive PGCs undergoing cell death were detected in E9 . 0 hindgut endoderm ( Figure 2C ) . In addition , the same territory contained accumulations of SSEA1-negative , apoptotic cells . Based on their size we suspected them to be germ cells having lost already expression of their typical marker , although we could not exclude that they represented mutant somatic cells . In summary , Mad2l2−/− PGCs were specified normally , but their numbers decreased progressively , and no PGCs could be detected in Mad2l2−/− embryos beyond E9 . 5 . This time window correlates with an epigenetic transition of PGCs and cell cycle arrest between E7 . 5-E9 . 5 [3] , [11] . Proper development of PGCs relies on their endogenous program as well as on exogenous signals emanating from surrounding somatic cells that support their induction , migration or survival in various organisms [41]–[44] . To address the cause of early PGC loss in Mad2l2 deficient embryos , we employed a Prdm1-Cre mouse line , which would be expected to delete the Mad2l2 gene specifically in nascent PGCs [4] . The TUNEL assay demonstrated apoptosis in SSEA1-positive PGCs of Prdm1-Cre+ , Mad2l2fl/fl embryos at E8 . 75 ( Figure 3 ) . In addition , TUNEL-positive , SSEA1-negative cells with a high nuclear to cytoplasmic ratio were observed in the hindgut . Also some TUNEL-negative , SSEA1-positive PGCs were found , which is explainable by the incomplete efficiency of Prdm1-Cre mediated deletion , although the actual recombination could not be confirmed here for the few available cells [4] . In contrast , no apoptosis was observed in Prdm1-Cre+ , Mad2l2fl/+ PGCs of the same age , excluding toxic effect of Cre recombinase on PGCs [45] . Together , these findings demonstrate that Mad2l2 deficient PGCs did not survive even in a wild type somatic environment . Since Mad2l2 is the subunit of a repair DNA polymerase , we asked if Mad2l2 deficient PGCs are affected by DNA damage . We applied an antibody detecting phosphorylated ATM/ATR substrates ( pATM/ATR-S ) including Chk1 , Chk2 , and MDM2 , as well as specific antibodies against pChk1 and pChk2 , respectively . No double-positive PGCs were detected in either wild type or knockout embryos in such staining ( Figure S3B–D ) . Together , these observations indicate that Mad2l2 deficient PGCs are not lost due to DNA damage . Immediately after their induction in the epiblast , PGCs begin to undergo massive epigenetic reprogramming with regard to both DNA and histone modifications . The genome-wide demethylation of the DNA in PGCs is partially due to a downregulation DNA methyltransferases , which is accompanied by loss of cytidine methylation . To address the epigenetic reprogramming in Mad2l2−/− PGCs , first we performed whole mount staining ( See Text S1 ) against Dnmt3b DNA methyltransferase . Both wild type and Mad2l2 deficient PGCs suppressed Dnmt3b expression ( Figure 4A ) . Immunohistochemistry analysis of DNA methylation showed loss of the 5-methylcytosine ( 5 mC ) at E9 . 0 in both wild type and knockout sections ( Figure 4B ) . These observations seem to indicate that DNA hypomethylation had been properly initiated and progressed in the absence of Mad2l2 . In PGCs , the repressive histone H3K9me2 should become downregulated during the cell cycle arrest between E7 . 5 and E9 . 5 . A comparison of stage-matched E9 . 0 embryos revealed that the majority of mutant , Oct4-positive PGCs had not downregulated H3K9me2 , while wild type PGCs mostly had lost this histone modification ( Figure 5A ) . Correspondingly , also G9a and GLP , two H3K9 methyltransferases , were still found in mutant , but not in wild type PGCs ( Figure 5B , C; S4A , B ) . Addressing the cell cycle profile of PGCs , we confirmed a cytoplasmic localization of Cyclin B1 in the majority of wild type PGCs on E9 . 0 , indicating that they were in the G2 phase of the cell cycle ( Figure 6 ) [11] . In Oct4-positive Mad2l2−/− PGCs , on the other hand , the Cyclin B1 protein was either localized in the nucleus , in the cytoplasm or not present at all ( Figure 6 ) . Thus , it appeared that mutant PGCs did not arrest in G2 phase of their cell cycle . A highly elevated , global H3K27me3 modification could be confirmed for the majority of wild type PGCs , while levels in Mad2l2−/− PGCs were mostly indistinguishable from surrounding somatic cells ( Figure 7A ) . Ezh2 , the relevant methyltransferase for residue K27 of histone 3 , is expressed in PGCs at a similar level to that of neighboring somatic cells , at least during their specification period [46] . However , we observed that the inactivation of Ezh2 was completely suppressed in the majority of wild type PGCs at E8 . 5 , while above 60% of knockout PGCs contained high or low levels of such inactive Ezh2 protein ( Figure 7B ) . Thus , a significant portion of the Mad2l2−/− PGCs failed to acquire an epigenetic status dominated by H3K27me3 , probably due to presence of inactive phosphorylated Ezh2 . The number of early PGCs is too small for biochemical and transfection approaches . Therefore , we performed a set of experiments in fibroblasts with the intention to provide evidence for a function of Mad2l2 in epigenetic and cell cycle regulation . Since the Mad2l2 protein contains a protein-binding HORMA domain Co-immunoprecipitation was applied to identify Mad2l2 interacting partners related to histone modifications ( See Text S1 ) . First , to explore a physical interaction between Mad2l2 and G9a or GLP , NIH3T3 fibroblasts were transfected with a plasmid encoding HA-Mad2l2 ( See Text S1 ) . Co-immunoprecipitation of NIH3T3 protein extract with anti-G9a , anti-GLP or anti-HA antibodies demonstrated that Mad2l2 interacts with both methyltransferases ( Figure 8A , B ) . Transfection of NIH3T3 cells with a vector encoding a GFP-fused Mad2l2 protein showed that G9a mRNA levels were specifically downregulated in the presence of GFP-Mad2l2 ( Figures S5A ) . G9a protein levels were always low in Mad2l2-GFP transfected cells , while untransfected cells had either high or low levels ( Figures 8C ) . Correspondingly , the level of H3K9me2 became completely suppressed in transfected cells ( Figure 8C ) , while levels of H3K4me2 , an unrelated histone modification , remained unaffected ( Figure S5B ) . For the analysis of loss-of-function conditions Mad2l2 deficient MEFs were prepared , and elevated levels of G9a and H3K9me2 were observed ( Figure 8D ) . Together , these findings indicate a negative correlation between the presence of Mad2l2 and the expression and activity of the methyltransferase G9a . To test whether ectopic expression of Mad2l2 can arrest the cell cycle , NIH3T3 cells were transfected with a HA-Mad2l2 encoding vector . Expressing cells did not enter mitosis , as evident by the complete absence of pH 3 or Cyclin B1 from nuclei , as well as the presence of unseparated centrosomes ( Figure 8E ) [47] , [48] . Several pathways regulating the entry into mitosis converge at the cyclin dependent kinase 1 ( Cdk1 ) , which needs to be dephosphorylated and associated with phosporylated Cyclin B1 to be active [49] , [50] . We hypothesized that Mad2l2 might interact physically with Cdk1 or Cyclin B1 to regulate the G2/M transition . Protein lysate from HA-Mad2l2 transfected NIH3T3 cells was precipitated with antibodies against Cdk1 , pCdk1 ( phosphorylated Cdk1 ) , Cyclin B1 , and the HA-tag . Co-precipitate analysis revealed a physical interaction of Mad2l2 with Cdk1 , but not pCdk1 or Cyclin B1 ( Figure 8F–H ) . We then looked for a regulatory effect of Mad2l2 on the kinase activity of Cdk1/Cyclin B1 in an in vitro assay ( See Text S1 ) , containing recombinant GST-Mad2l2 , Cyclin B1 and Cdk1 , as well as the specific substrate Cdc7 [51] . GST-Mad2l2 , but not GST alone could specifically attenuate the kinase activity of Cdk1-Cyclin B1 in a concentration-dependent manner ( Figure 8I ) . Together , our experiments suggest that the ectopic presence of Mad2l2 prolongs the cell cycle . To address whether Mad2l2 can principally be involved in H3K27me3 upregulation , gain-of-function experiments with a GFP-Mad2l2 fusion protein were performed in NIH3T3 cells . Immunocytochemistry showed a very high level of H3K27me3 in all GFP-positive cells , while surrounding untransfected cells had mostly low levels , with some exceptions possibly dependent on the state of their cell cycle ( Figure 8J ) . Given the inhibitory function of Mad2l2 on the kinase activity of Cdk1 , we asked if it might attenuate the inhibitory phosphorylation of Ezh2 ( Figure 8K , L ) . The highest level of pEzh2 was observed in mitotic cells correlating with the highest activity of Cdk1/Cyclin B1 ( Figure 8K ) [18] . In contrast , Mad2l2 over-expressing cells showed the lowest level of pEzh2 , even less than that in untransfected interphase cells ( Figure 8K ) . Consistently , western blot analysis confirmed the drastic suppression of pEzh2 in Mad2l2 over-expressing FACS-sorted fibroblasts , while the overall level of Ezh2 itself remained unchanged ( Figure 8K ) . The loss-of-function situation was analyzed in Mad2l2 deficient MEFs , which showed an increased level of pEzh2 , while the amount of H3K27me3 was decreased ( Figure 8L ) . Apparently , here the Cdk1/Cyclin B1 was active , and could phosphorylate and thereby inactivate Ezh2 . Our analysis of fibroblasts and of a cell free system demonstrate the capacity of Mad2l2 to suppress the kinase activity of Cdk1/Cyclin B1 , and thus to support the activity of Ezh2 and by that promote the tri-methylation of histone 3 on K27 .
Several mutations are known to affect or terminate the development of PGCs ( for review see [44] ) . In principal , every step proved to be sensible , particularly the primary induction by BMP signaling , the early specification , the migration to the developing gonad , and the pre- or postnatal oogenesis or spermatogenesis . The early BMP response genes , Prdm1 and Prdm14 , are crucial for PGC specification directly after induction , where numbers of mutant PGCs are drastically reduced already on E8 . 0 , and only few mutant PGCs survive to E9 . 5 [4] , [5] . Similar kinetics for PGC loss were observed in mice lacking the transcription factor Tcfap2c , which mostly phenocopy the Prdm1−/− mice [52] . A slightly later timing , shifted by about one day , was found for the Mad2l2 mutants in our study . Although embryos at EHF stage were relatively small , they harbored stage-adequate numbers of PGCs expressing Prdm1 and the commitment markers Dppa3 and Tcfap2c arguing for a normal specification in the epiblast . A reduction of PGC numbers was observed in the LHF stage , and there was no survival beyond E9 . 5 . At this point of development , PGCs would normally have undergone a major epigenetic reprogramming , would recover from their cell cycle arrest , and resume transcription . This timing suggests a failure of epigenetic reprogramming and cell cycle arrest in Mad2l2−/− PGCs . In principle , it is conceivable that wrongly developed PGCs might either revert to a somatic fate , or undergo apoptosis . PGCs are lost without evidence for apoptosis in mutants of the Prdm1 , the Prdm14 , and the Tcfap2c gene , whereas mutations in the Oct4 , the Kit and the Mad2l2 genes remove wrongly programmed PGCs by apoptosis [4] , [5] , [52]–[54] . Somatic Mad2l2−/− cells apparently do not rely on a specific epigenetic reprogramming and cell cycle arrest , and at least some Mad2l2-deficient mice develop normally and live until adulthood . Still , mutants are born in sub-Mendelian ratio and adults are usually smaller , as is the case in many mutant mice . Together , this points to a highly specialized function of Mad2l2 in the unique development of germ cells , but does not exclude lower penetrance effects in somatic cells . H3K9 methylation is critical for formation of heterochromatin and transcriptional silencing . At the onset of PGC development , H3K9me2 is the dominant epigenetic mark in the genome of embryonic cells [3] , [11] . This modification requires the activity of the two methyltransferases G9a and GLP [55] . G9a , the major mammalian H3K9 methyltransferase , plays a critical role in germ cell development , particularly in gametogenesis . The specific deletion of G9a in PGCs after E9 . 5 leads to germ cell loss during the meiotic prophase , and thus to sterility of both males and females [56] . During the S phase of the cell cycle , G9a binds to DNA methyltransferase DNMT1 and loads on to the DNA at replication foci , ensuring a coordination of DNA methylation and H3K9 methylation in heterochromatin regions [57] . Nascent PGCs leave asynchronously the S phase of their cycle and enter G2 at around E8 . 0 . At this time , the de novo methylation of the daughter chromatin is suppressed , and both Prdm1 and Prdm14 were suggested to be involved [58] , [59] . In parallel , the maintained activity of histone demethylases like Jmjd1a erases further the remaining H3K9me2 [60] . Our results indicate that similar to Prdm14 deficient PGCs , the majority of Mad2l2−/− PGCs fail to suppress H3K9me2 . The maintenance of a high H3K9me2 level in Prdm14 mutant PGCs was attributed to a failure in downregulation of GLP . Released from repression by genome-wide H3K9me2 , PGCs repress RNA Pol-II dependent de novo transcription until they acquire the alternative repressive histone mark , H3K27me3 . This probably ensures the maintenance of separate PGC and somatic programs , established previously via combinational functions of Prdm1 , Prdm14 , and Tcfap2c [61] . A significant portion , but not all , of the Mad2l2−/− PGCs failed to proceed with their epigenetic reprogramming , as it is the case in Prdm14 mutant PGCs . Obviously , shortly before their elimination around E9 . 0 , the Mad2l2−/− PGCs represent a heterogeneous population with respect to their transcriptional and epigenetic status . Thus , Mad2l2 is absolutely essential for the development of PGCs . We observed that Mad2l2 suppresses G9a on the level of gene expression , which could be related to its ability to interact with transcription factors [29] , [32] . The binding of Mad2l2 to the two histone methyltransferases G9a and GLP was previously identified in a systematic analysis of human protein complexes , and represented a first hint for an involvement of Mad2l2 in the generation of epigenetic modifications [62] . We confirmed this evidence by co-immunoprecipitation of both G9a and GLP with HA-Mad2l2 from transfected fibroblasts , where the level of H3K9me2 was significantly downregulated . Noteworthy , both G9a ( PXXXPP ) and GLP ( PXXXyP ) have the sequence motif suggested to be responsible for Mad2l2 binding [27] . G9a and GLP form homo- and heteromeric complexes in vitro , which are necessary for histone methyltransferase activity [13] , [55] . Indeed , several proteins , bind to G9a or GLP , and alter their activities [63] , [64] . Among those is Prdm1 , which binds to G9a and recruits it to assemble silent chromatin [65] . Similarly , the direct interaction between Mad2l2 and G9a or GLP may disrupt formation of the G9a-GLP active heterodimer complex , and thus suppress the methylation of histone 3 . Supportive evidence for such an inhibitory binding comes from the negative correlation between Mad2l2 and H3K9me2 levels in PGCs ( Fig . 5A ) and fibroblasts ( Fig . 8D ) . However , the actual significance of the observed protein-protein interactions needs further investigation . Cdk1 is a regulatory kinase of central importance for several processes , in particular also in cell cycle control and in epigenetic reprogramming [66] , [67] . Our study in transfected fibroblasts and in a cell-free system suggests that Mad2l2 can bind directly to dephosporylated Cdk1 , and thus inhibit its kinase activity . Possibly this interaction involves the Cdk1 sequence PXXXPy , which is related to the previously identified Mad2l2 binding motif PXXXPP [27] . The entry into mitosis is mediated by a complex network of proteins that finally activate the Cdk1-Cyclin B1 complex [50] . One of the first functions of Cdk1-Cyclin B1 is the phosphorylation and therefore disruption of Eg5 , a protein involved in centrosome adhesion [68] . Overexpression of Mad2l2 abrogated centrosome separation , and caused a cell cycle arrest at the G2 phase . Dephosphorylated Cdk1 in association with phosphorylated Cyclin B1 translocate to the nucleus and initiates prophase by the phosphorylation of a variety of substrates [50] . Thus , via direct binding to Cdk1 , Mad2l2 would have the capacity to inhibit Cdk1-Cyclin B1 complex formation , and thus to block the entry into mitosis . Inhibition and/or disruption of the Cdk1-Cyclin B1 complex through direct interaction were previously also observed for Gadd45 proteins , stress factors implicated in the activation of the G2/M DNA damage checkpoint [51] , [69] , [70] . Previous analyses of Mad2l2 had indicated inhibitory interactions with Cdh1 , and possibly also with Cdc20 [23] , [24] . These proteins would normally exert their function only after the onset of mitosis , either as part of the spindle assembly checkpoint , or as the substrate recognizing protein of the APC/C protein ubiquitination complex , respectively . However , early knockout PGCs divide relatively normal and only fail to arrest in the G2 phase . Therefore , it is less likely that Mad2l2 functions in mitosis of PGCs via binding to Cdh1 , or Cdc20 . Overexpression in fibroblasts indicated the possibility that Mad2l2 can be involved in a G2 arrest . This might correlate with the G2 arrest , which coincides with the epigenetic transition of PGCs from a H3K9me2 to a H3K27me3 configuration , and with the timing of PGC loss in Mad2l2 mutants . Among the many functions of the widely distributed kinase Cdk1 is the inhibition of the histone 3 methyltransferase Ezh2 by phosphorylation [66] , [67] . Our analysis in fibroblasts indicates that Mad2l2 can interfere with this inactivation , and thus in effect , promote the activation of Ezh2 . Consequently , we observed an increase of H3K27me3 levels upon overexpression of Mad2l2 . Our data do not allow at present to decide if the primary defect in knockout PGCs lies in the regulation of the cell cycle , if the epigenetic failure precedes misregulation of the cycle , or if the two tightly coupled processes are not separable . Nevertheless , the outcome is that Mad2l2 mutated PGCs are not able to make the developmental transition from E7 . 5 to E9 . 5 , and are quickly eliminated from the embryo ( Figure 9 ) . Thus , Mad2l2 is absolutely required for the development of PGCs , and thus for fertility . While this manuscript was under revision , a related set of data was published demonstrating the necessity of Mad2l2 for PGC maintenance [71] . However , detailed characterization of knockout PGCs and the mechanism by which Mad2l2 may function were not studied .
All animal works have been conducted according to relevant national and international guidelines . Genomic sequences were amplified from a 129 strain mouse PAC clone . The vector was assembled using the recombineering protocol and materials as described ( see Figure S1; [72] . The loxP sites were introduced 113 bp upstream of the first coding exon , and 20 bp dowstream of the last exon , deleting finally a region of 5330 bp . The vector was introduced into MPI-II ES cells , which were selected with G418 and Ganciclovir . Cells with homologous recombination were aggregated with morula-stage embryos . The Mad2l2 gene was inactivated by crossing of heterozygote mice with CMV-Cre mice [73] , and then breeding to homozygocity . Genotyping was performed using the primers #1 ( GCTCTTATTGCCTTGACATGTGGCTGC ) , #2 ( GGACACTCAGTTCTGGAAAGGCTGG ) , and #3 ( CTGCAGCCCAATTCCGATCATATTCAATAAC ) . The day of the vaginal plug was taken as E0 . 5 , and embryos were dissected accordingly . Embryos were staged [11] by corresponding time and morphology as follows: before E8 . 0 ( EHF ) , E8 . 0 ( LHF ) , E8 . 25 ( less than 5 somites ) , E8 . 5 ( before turning , 6 to 8 somites ) , E8 . 75 ( turning embryos , 10 to 12 somites ) , E9 . 0 , ( after turning , 14 to 18 somites , with only the first branchial arch obvious , and with open otic vesicles , E9 . 5 ( two branchial arches , closed otic vesicles , 20–24 somites ) . The following antibodies were used . Rabbit anti-Cyclin B1 ( Sigma-Aldrich ) , 1∶100; mouse anti-phospho-Histone H3 ( ser10; Cell Signaling ) , 1∶200; rat anti-HA ( Roche ) , 1∶100; mouse anti-γTubulin ( Abcam ) , 1∶200; mouse anti-Cdk1 ( Santa Cruz ) , 1∶50; rabbit anti-pCdk1 ( Cell Signaling ) , 1∶50; mouse anti-Oct4 ( BD ) , 1∶100; rabbit anti-Oct4 ( Abcam ) , 1∶100; mouse anti-SSEA1 ( Santa Cruz ) , 1∶100; rabbit anti-Nanog ( abcam ) , 1∶100; rabbit anti-Sox2 ( Millipore ) , 1∶200; rabbit anti-H3K9me2 ( Upstate ) 1∶100; and ( Millipore ) , 1∶100; rabbit anti-G9a ( Cell Signaling ) , 1∶25; mouse anti-GLP ( Abcam ) , 1∶50; rabbit anti-Mad2l2 ( Abcam ) , 1∶100; mouse anti-γH2AX ( Millipore ) , 1∶200; rabbit anti-pChk2 ( Cell Signaling ) , 1∶200; mouse anti-Vimentin ( gift of M . Osborn ) , 1∶100; rabbit anti-WT1 ( Abcam ) , 1∶1000; rabbit anti-Ezh2 ( Cell Signaling ) , 1∶2000; rabbit anti-pEzh2 T487 ( Epitomics ) , 1∶1000; rabbit anti-H3K4me2 ( Active Motif ) , 1∶100; rabbit anti-H3K27me3 ( Active Motif ) , 1∶100; rabbit anti-Dppa3 ( abcam ) , 1∶500; rabbit anti-Stra8 ( abcam ) , 1∶2000; rabbit anti-Plzf ( abcam ) , 1∶100; rabbit anti-Dnmt3b ( abcam ) , 1∶100; rabbit anti-Tcfap2c ( Santa Cruz ) , 1∶100; mouse anti-5mC ( abcam ) , 1∶200 . GST-fused Mad2l2 protein was expressed in and purified from E . coli . Full length Mad2l2 cDNA was cloned in frame with the N-terminal GST-tag into the pGEX-KT vector . Expression was induced by the addition of 1 mM IPTG ( isopropyl-β-D-thiogalactopyranoside , Sigma ) . Bacterial cells were harvested; proteins were lysed on ice in 50 mM Tris , pH 7 . 5 , 500 mM NaCl , 2 mM EDTA , 5 mM DTT , 10% glycerol , freshly added 1 mM PMSF and Complete EDTA-free protease inhibitor cocktail tablet ( Roche ) . Glutathione Sepharose 4B ( Amersham Biosciences ) was used to purify the GST-fused protein . The elution was done twice , each time with 2 ml elution buffer ( 500 mM Tris , pH 8 . 0 , 100 mM Glutathione supplemented with protease inhibitor ) . The protein was dialyzed in dialysis buffer ( 20 mM Tris-Cl pH 7 . 5 ) using a dialysis cassettes ( Pierce ) at 4°C overnight . The protein concentrations were measured and determined according to the standard curve . Kinase activity of Cdk1-cyclin B1 was analyzed using purified , recombinant proteins ( CycLex ) , and a human Cdc7 peptide as substrate , applying an assay system from CycLex [51] . To test effect of Mad2l2 on kinase activity of Cdk1-Cyclin B1 , dilutions of GST-Mad2l2 or GST alone protein were incubated for 15 min at 37°C with 12 . 5 mUnits of recombinant kinase . These protein mixes were individually given into substrate-coated wells , and incubated for 45 min at 37°C . For detection of phospho-Cdc7 a specific monoclonal antibody ( TK-3H7 ) and HRP-conjugated anti-mouse IgG was applied , and the absorbance at 450 nm was measured . | Primordial germ cells ( PGCs ) are the origin of sperm and oocytes , and are responsible for transferring genetic information to the next generation faithfully . PGCs are first specified from pluripotent epiblast cells early in embryonic development . Second , they reprogram their epigenetic signature by changing histone modifications . This developmental event is specific to germ cells but not somatic cells . Although many players in the specification of PGCs are identified , only little is known about the genes essential for the regulation of the second phase . Here , we report that the Mad2l2 gene product plays an important role in the epigenetic reprogramming of PGCs . In wild type PGCs the cell cycle is arrested , and the methylation of histone 3 on residue K9 is replaced by methylation on K27 . Our findings indicate that Mad2l2 is involved in this coordination of cell cycle and epigenetic reprogramming . The elucidation of this mechanism would help to identify the genetic basis of infertility . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2013 | A Critical Function of Mad2l2 in Primordial Germ Cell Development of Mice |
Human genetic factors such as blood group antigens may affect the severity of infectious diseases . Presence of specific ABO and Lewis blood group antigens has been shown previously to be associated with the risk of different enteric infections . The aim of this study was to determine the relationship of the Lewis blood group antigens with susceptibility to cholera , as well as severity of disease and immune responses to infection . We determined Lewis and ABO blood groups of a cohort of patients infected by Vibrio cholerae O1 , their household contacts , and healthy controls , and analyzed the risk of symptomatic infection , severity of disease if infected and immune response following infection . We found that more individuals with cholera expressed the Le ( a+b− ) phenotype than the asymptomatic household contacts ( OR 1 . 91 , 95% CI 1 . 03–3 . 56 ) or healthy controls ( OR 1 . 90 , 95% CI 1 . 13–3 . 21 ) , as has been seen previously for the risk of symptomatic ETEC infection . Le ( a–b+ ) individuals were less susceptible to cholera and if infected , required less intravenous fluid replacement in hospital , suggesting that this blood group may be associated with protection against V . cholerae O1 . Individuals with Le ( a–b− ) blood group phenotype who had symptomatic cholera had a longer duration of diarrhea and required higher volumes of intravenous fluid replacement . In addition , individuals with Le ( a–b− ) phenotype also had lessened plasma IgA responses to V . cholerae O1 lipopolysaccharide on day 7 after infection compared to individuals in the other two Lewis blood group phenotypes . Individuals with Lewis blood type Le ( a+b− ) are more susceptible and Le ( a–b+ ) are less susceptible to V . cholerae O1 associated symptomatic disease . Presence of this histo-blood group antigen may be included in evaluating the risk for cholera in a population , as well as in vaccine efficacy studies , as is currently being done for the ABO blood group antigens .
Cholera continues to cause severe diarrheal illness in people with inadequate public health who live in resource-limited settings . Cholera is endemic in countries in Asia and Africa , with new outbreaks reported each year in several countries including , most recently , in Zimbabwe and Haiti [1] , [2] . Vibrio cholerae O1 is the predominant cause of endemic and epidemic cholera , and this infection is the most common bacterial cause of acute watery diarrhea in adults and children in Bangladesh [3] . There is a close interplay between the organism and the human host in the disease process , and understanding the nature of this interaction is important for understanding pathophysiology , as well as for designing the most appropriate preventive and therapeutic strategies to reduce the morbidity and mortality associated with this infection . In previous studies , we have analyzed the genes expressed by V . cholerae O1 during human infection [4] , [5] , [6] , as well as the human genes expressed in the gut mucosa in response to the infection [7] . These studies have suggested that human innate immune responses are up-regulated in response to V . cholerae O1 infection , and these innate immune responses may be important in controlling the disease . Studies of protection from cholera in exposed household contacts indicated that there is a genetic basis for at least some portion of protection from infection [8] , [9] , and a candidate gene analysis in these contacts identified a polymorphism in the human gene for LPLUNC1 , an innate immune response gene , as linked to protection [10] , [11] . The other set of human genetic factors that have been studied in relationship to susceptibility to enteric infections are the blood group antigens . For cholera , blood group O has been associated with a lower risk of colonization in exposed household contacts [12] but if colonized , a higher risk of more severe disease [12] , [13] , [14] . In contrast , blood groups AB or A have been shown to be associated with more severe illness in individuals infected with a related pathogen , enterotoxigenic Escherichia coli ( ETEC ) , in children in Bangladesh [15] . Another set of blood group antigens , the Lewis blood group antigens Lewis a ( Lea ) and Lewis b ( Leb ) , are carbohydrate antigens related to the ABO blood group antigen that are synthesized in epithelial tissues and adsorbed to the surface of red blood cells [16]; these antigens can also be detected in saliva and other secretions , as well as on cells of mucosal epithelia [17] , [18] . The Lewis antigen system has three different phenotypes; Le ( a+b− ) ( these individuals have the nonsecretor phenotype ) ; Le ( a–b+ ) , in which a fucosyltransferase converts Lea to Leb ( these individuals have the secretor phenotype ) ; or Le ( a–b− ) , in which there is a failure to express either antigen ( these individuals can be either secretors or non-secretors ) [19] . In a previous study of ETEC diarrhea in Bangladesh , we showed that the approximate proportions of these three phenotypes in the population were: Le ( a+b− ) 26%; Le ( a–b+ ) 58%; and Le ( a–b− ) 16% [20] . We also showed that patients with the Le ( a+b− ) phenotype had an increased risk of having symptomatic ETEC diarrhea compared to the other two phenotypes , particularly if infected with an ETEC strain expressing a CFA/I group colonization factor; this increased risk of symptomatic disease was not seen in patients infected with ETEC expressing other colonization factors , or with rotavirus . Previous studies have suggested that the CFA/I group colonization factors of ETEC bind the Lea antigen on epithelial cells of the small intestine [21] . Conversely , susceptibility to Helicobacter pylori infection was higher in Le ( a–b+ ) individuals [22] . In the present study , we analyzed the relationship of Lewis blood group antigen to the risk of symptomatic V . cholerae O1 infection in a cohort of patients and their household contacts in Bangladesh , as well as the relationship of Lewis antigen phenotype to severity of and immune responses following disease .
The study was carried out on patients with cholera presenting to the icddr , b diarrheal disease hospital in Dhaka , Bangladesh . Hospitalized patients with acute watery diarrhea were confirmed by stool culture to be infected with V . cholerae O1 as previously described and enrolled on the 2nd day of hospitalization after informed consent [8] , [23] . On the same day as patients were enrolled in the study ( defined as day 2 ) , field workers enrolled all consenting household contacts of each index patients , defined as individuals who shared the same cooking pot as the index patient for three or more days [23] . Index patients were assessed for other clinical parameter . The type of dehydration status and recovery of patients was assessed by experienced physicans in the icddr , b diarrheal hospital [24] . Household contacts were followed prospectively on study days 2–10 , providing daily rectal swabs for cultures for V . cholerae O1 , as well as giving clinical histories for diarrheal illness . Blood specimens were obtained from index patients and household contacts on study days 2 , 7 and 30 . Saliva specimens were collected from all participants on study day 2 . Saliva specimens were also obtained at one time point from 283 healthy individuals who were from an urban setting and in a similar socio-economic status as the index patients , to determine the distribution of the Lewis blood group antigens in the general population . Blood and saliva samples obtained at day 2 were used for the determination of the ABO and Lewis blood group phenotypes , respectively . Blood samples at each time point were assessed for vibriocidal antibody , and IgG and IgA antibodies against cholera toxin B subunit ( CTB ) and lipopolysaccharide ( LPS ) antigens . This study was conducted according to the principles expressed in the declaration of Helsinki . We obtained written consent from each individual prior to participation . Written informed consent was obtained from adults participating in the study . This study was approved by the Ethical and Research Review Committees of the International Centre for Diarrhoeal Disease Research , Dhaka , Bangladesh ( icddr , b ) and the Institutional Review Board of Massachusetts General Hospital , Boston , MA . For all index cases , stool specimens were cultured on taurocholate-tellurite gelatin agar ( TTGA ) plates for isolation of V . cholerae . After overnight incubation of plates , specific monoclonal antibodies were used to detect V . cholerae O1 , and the Ogawa and Inaba serotypes by slide agglutination test [25] , [26] . Rectal swabs from household contacts were collected in Cary-Blair transport media , taken to the icddr , b , and cultured on TTGA followed by colony identification as above . Some specimens were also enriched in alkaline peptone water for 4 hours prior to culturing [3] . For ABO blood group typing , a slide agglutination test was carried out according to the manufacturer's instruction ( Biotec laboratories , UK ) . Lewis blood group phenotype was determined using saliva samples and a dot blot immunoassay procedure [20] , [27] . For this purpose , 2 µl of saliva were applied to nitrocellulose membrane strips and allowed to dry . After blocking with 1% bovine serum albumin , mouse monoclonal anti-Lea and anti-Leb antibodies ( Abcam , Cambridge , UK ) were added and the strips were incubated for 30 min at room temperature with gentle shaking . The strips were then washed and incubated with secondary , horseradish peroxidase-conjugated antibody for another 30 min . After washing , the strips were developed with 4-chloro-1-naphthol and 3% hydrogen peroxide . A specimen was considered positive when a dark black spot appeared on the membrane . Vibriocidal antibody assays were performed using guinea pig complement and the homologous serotype of V . cholerae O1 isolated from the patient , either El Tor Ogawa ( strain 25049 ) or El Tor Inaba ( strain T-19479 ) as previously described [28] . The vibriocidal titer was defined as the reciprocal of the highest plasma dilution resulting in >50% reduction of the optical density compared to that of control wells without plasma . Seroconversion was defined as a 4-fold or higher increase in vibriocidal titer after infection . Plasma IgG and IgA antibodies specific to CTB and LPS were measured by kinetic ELISA procedure as previously described [29] , [30] . In brief , 96-well microtiter plates were coated with either purified V . cholerae O1 LPS ( 250 ng/well ) , or GM1 ganglioside ( 100 ng/well ) followed by recombinant CTB ( 50 ng/well ) . Plates were incubated with diluted patient sera ( 1∶50 for LPS ELISA and 1∶200 for CTB ELISA ) , washed , and horseradish peroxidase-conjugated secondary antibodies to human IgG or IgA ( Jackson Laboratories , Bar Harbor , Maine ) were applied in separate wells . Plates were developed using 0 . 1% orthophenylene diamine ( Sigma , St . Louis , Missouri ) in 0 . 1 M sodium citrate buffer with 0 . 1% hydrogen peroxide , and optical densities ( OD ) were read kinetically at 450 nm for 5 minutes at 19-s intervals and results expressed as milliabsorbance/min ( mAb/min ) . ELISA values were calculated by taking the ratio of the value obtained for the test specimen to that obtained for the positive control specimen and multiplying by a factor of 100 . Pooled plasma was prepared using specimens from convalescent stage cholera patients from an earlier study [29] . Statistical analyses were performed on SPSS 17 . 0 and SigmaStat 3 . 1 programs . Graphs were prepared using the Prism 5 . 0 software ( GraphPad Software Inc . ) . The association between Lewis blood groups and symptomatic cholera was assessed by the chi-square test . Associations were also carried out by calculating the odds ratio ( OR ) with 95% confidence intervals ( CI ) using EpiInfo 3 . 3 . 2 . The Wilcoxon signed rank test was used to compare immune responses of patients on different follow-up days and the Mann-Whitney U test was used for comparison among different groups . All reported P values are two tailed and significance was defined as P<0 . 05 .
Ninety five cholera patients , 144 household contacts , and 283 healthy controls were enrolled in the study overall ( Table 1 ) . The median age of the patients enrolled in the study was 28 years while that for the household contacts was 23 years and of healthy controls was 18 years . The controls were younger than the patients and healthy contacts ( P<0 . 001 ) . The proportion of males and females in each group was not significantly different . Thirty five household contacts had positive rectal swabs for V . cholerae O1 during follow up and of these , 20 had diarrhea and were considered to have symptomatic cholera; these 20 were excluded from the analysis of Lewis blood group types in contacts . Among the index patients , 80% ( 76/95 ) were infected with the Ogawa serotype of Vibrio cholerae O1 and 20% ( 19/95 ) were infected with Vibrio cholerae O1 Inaba . At the time of hospitalization , 92% ( 87/95 ) of the index patients were severely dehydrated . The average duration of diarrhea for all index patients was 57 hours and patients received on average 7 . 5 liters of intravenous rehydration . Among the 95 index patients , 43% were blood group O positive , 34% were blood group B , 19% were blood group A , and 4% were blood group AB . The asymptomatic contacts had a similar distribution of ABO blood groups ( 47% , 27% , 18% and 8% respectively ) . We did not determine the ABO blood group of individuals enrolled as healthy controls but this has been done in earlier studies [12] , [46] . The distribution of ABO blood group in a similar setting in Bangladesh has been shown to be for the O∶A∶B∶AB groups to be 28%∶23%∶38%∶11% respectively [46] . In the 522 study participants overall , 28% were Le ( a+b− ) , 55% were Le ( a–b+ ) , and 17% were of the Le ( a–b− ) blood group phenotype , very similar to the proportions shown in this population previously [15] . In comparing the Lewis blood group phenotype distributions between patients symptomatic with cholera compared to asymptomatic household contacts and healthy controls , patients were enriched for the Le ( a+b− ) phenotype ( 39% ) and had fewer individuals in the Le ( a–b+ ) phenotype ( 40% ) ; both these were significantly different than the frequencies of these phenotypes in asymptomatic contacts and healthy controls ( Figure 1 ) . In contrast , the distribution of the three phenotypes in asymptomatic contacts and healthy controls were virtually identical to each other and to the overall population . The Le ( a+b− ) blood group phenotype was significantly associated with having symptomatic cholera , as compared to the household contacts who were asymptomatic ( OR 1 . 91 , P = 0 . 039 , 95% CI 1 . 03–3 . 56 ) or to the healthy controls ( OR 1 . 90 , P = 0 . 014 , 95% CI 1 . 13–3 . 21 ) . Similarly the frequency of Le ( a–b+ ) was lower in patients than the contacts ( OR 0 . 45 , P = 0 . 006 , 95% CI 0 . 25–0 . 81 ) or healthy controls ( OR 0 . 48 , P = 0 . 003 , 95% CI 0 . 29–0 . 79 ) . No relationship was found between the presence of Le ( a–b− ) blood group phenotype and susceptibility to cholera comparing the three groups of study participants ( Figure 1 ) . We also analyzed the presence of different combinations of Lewis blood group antigens and ABO blood groups in relation to susceptibility to cholera ( Figure 2 ) . In individuals with the A blood group , the Le ( a–b+ ) phenotype was less common in patients than household contacts ( P = 0 . 001 ) , while the percentage of the Le ( a+b− ) phenotype in patients trended toward being higher than in contacts ( P = 0 . 071 ) , as seen for the group overall . Similarly , in individuals with blood group B , we found a lower frequency of the Le ( a–b+ ) phenotype in patients compared to contacts ( P = 0 . 048 ) , as for the analysis in the study population overall . However , we did not find any significant associations of Lewis blood group antigens and symptomatic cholera in patients with blood group O . The small number of individuals with blood group AB ( n = 14 ) prevented any firm conclusions for this blood group . We next assessed whether there were any differences in immune responses to V . cholerae O1 infection in individuals with the various Lewis blood group phenotypes . We found no differences in plasma vibriocidal titers on days 2 , 7 , or 30 between index patients with the different Lewis blood group phenotypes ( data not shown ) . In analyzing IgG and IgA responses to CTB and LPS , we also found no differences in either IgG or IgA responses to CTB or the IgG responses to LPS ( Figure 3 ) . However , patients with the Le ( a–b− ) phenotype developed significantly lower LPS IgA responses on day 7 compared to the individuals with the Le ( a–b+ ) phenotype ( P = 0 . 034 ) ; there was a trend of lower responses when compared to those of the Le ( a+b− ) phenotype ( P = 0 . 064 ) . The responses in patients in the different Lewis groups were comparable by day 30 ( Figure 3A ) . Because of the differences in IgA responses to LPS between individuals with the different Lewis blood group antigens , we also compared the severity of cholera in these three groups . There were no differences between the groups in the time between onset of symptoms and presentation to the icddr , b , in the duration of diarrhea pre-hospitalization , in the use of antibiotics prior to presentation , or in the average ORS consumed before presentation ( data not shown ) . However , once hospitalized , patients with the Le ( a–b+ ) phenotype required significantly less intravenous fluids compared to individuals with either the Le ( a+b− ) ( P = 0 . 017 ) or Le ( a–b− ) ( P<0 . 001 ) phenotypes ( Table 2 ) . In addition , patients with the Le ( a–b− ) phenotype had a significantly longer duration of diarrhea than did the patients of Le ( a–b+ ) or Le ( a+b− ) groups ( P = 0 . 012 and 0 . 017 , respectively ) , correlating with their increased need for intravenous hydration ( Table 2 ) .
In this study , we investigated the relationship of the Lewis blood group antigens with susceptibility to cholera and to the clinical course of the illness . We determined the Lewis blood group using saliva samples , which have been previously shown to be concordant with typing carried out using blood specimens [20] . The overall ABO and Lewis blood group antigen distribution was similar to that seen in other studies carried out recently in Bangladesh [15] , [20] . Our first finding in this study was that individuals with the Le ( a+b− ) phenotype were more likely to get symptomatic cholera compared to the other two groups , suggesting that this Lewis blood group may be associated with an increased risk of being colonized with V . cholerae O1 or if colonized , of becoming symptomatic . This same Lewis blood group has previously been shown to increase susceptibility to symptomatic ETEC infection if the organism is expressing a CFA/I group colonization factor [15] , [20] . Interestingly , in looking at the inter-relationship between risk of symptomatic cholera and both ABO and Lewis blood groups , the increased risk of symptomatic infection in Le ( a+b− ) individuals was only seen in individuals who had the A or B blood groups , and not blood group O . Since blood group O is itself a risk factor for more severe cholera , it is possible that an effect of the Lewis blood group types was not evident because of the higher risk of symptoms already conferred by the O blood group . Our second finding , that individuals with the Le ( a–b+ ) phenotype required less intravenous fluids following hospitalization than individuals of the Le ( a+b− ) phenotype , is also consistent with a difference in severity of cholera , once it occurs , between these two Lewis blood groups . Our third finding was that individuals in the Le ( a–b− ) phenotype admitted to the icddr , b with cholera required the most intravenous fluids and had the longest duration of diarrhea , suggesting that this phenotype , while not over-represented in patients with cholera , was associated with an increased severity of disease once it occurs . We observed that there was a trend of susceptibility to cholera for those in the Le ( a–b− ) group also , but possibly because of the small sample size , the analysis did not reach statistical significance . The fourth finding in our study was that individuals with the Le ( a–b− ) phenotype had reduced IgA responses to LPS compared to individuals in the other two phenotypes although comparison with Le ( a+b− ) did not reach significance . The plasma level of IgA reactive to LPS on exposure is correlated with protection from subsequent infection with V . cholerae O1 [8] . Index patients in the different Lewis blood group types did not have any significant differences in baseline IgA reactive to LPS , just a difference in magnitude of response at day 7 . It is not known if this reduced magnitude of LPS-specific IgA on day 7 is associated with the longer duration of diarrhea and therefore higher requirement for intravenous fluid in this subgroup of individuals; the differences in LPS-specific IgA between groups was not evident by day 30 post infection . Histo-blood group antigens can predispose individuals to genetic , metabolic , and infectious diseases , including enteric illnesses . Blood group antigens are fucosyloligosaccharides that are expressed in the gut epithelium and hence can act as potential receptors for enteric pathogens [22] , [31] . This can make individuals of one blood group type more susceptible to a particular pathogen compared to individuals expressing other blood group antigens . Another mechanism of association with disease is that soluble forms of these antigens can be secreted into the gut lumen and might prevent colonization of pathogens by competitive neutralization [32] . Earlier studies have shown that individuals with Lewis blood group Le ( a–b+ ) are at higher risk for colonization by H . pylori [22] , [33] . Campylobacter jejuni binds to intestinal H ( blood group O ) antigen and it has been shown that fucosyloligosaccharides in human milk can inhibit binding and infection by this organism [34] . Norovirus has been particularly well studied for the association with blood group antigens . This pathogen binds specifically to A , H and difucosylated Lewis antigens but not to the B antigen [35] , which is supportive of earlier studies in which it was shown that individuals with O blood group were more prone to Norovirus infection , while it was less likely in individuals of the B blood group [36] . Susceptibility to V . cholerae infection is believed to result from a combination of factors including exposure , lack of immunity on encountering the organism [8] , [12] , nutritional deficiencies [37] , [38] , and human genetic polymorphisms [11] . Individuals with blood group O are at a higher risk of developing severe cholera than those with other blood groups [12] , [13] , [14] , [39] . It is hypothesized that this may have resulted in a selective pressure for human genetic evolution that may explain the lower prevalence of the O blood group in cholera endemic regions such as Bangladesh and other areas near the Ganges delta [40] . In the present study , we did not find any association of cholera with the presence of specific ABO blood groups , perhaps related to our smaller sample size . However , we did find a strong association with the Lewis blood group phenotypes , although the reasons behind this observation are not yet defined . Two possibilities are that , as for ETEC , the Lea antigen may act as a receptor on mucosal epithelia for a cholera ligand . However , Le ( a+b− ) individuals are also non-secretors , so it is possible that the association of this phenotype with symptomatic cholera is intertwined with the non-secretor status rather than the Lea and Leb antigens themselves . In contrast to cholera , individuals with blood groups A and AB are at higher risk for ETEC infection [15] , [41] , [42] . On the other hand , for the Lewis blood group antigens , individuals with the Le ( a+b− ) phenotype are more susceptible to both symptomatic cholera as well as ETEC infection . The distribution of Lewis blood group phenotypes in this study was different from that reported in a Caucasian population in a cholera non-endemic area , but similar to populations studied in India and Africa , where cholera is endemic [19] , [43] , [44] . However , unlike the relationship between ABO blood group and cholera , the Lewis blood group phenotypes have apparently not been selected for by cholera , as the more susceptible type , Le ( a+b− ) , is more frequent in the endemic areas than in those areas without endemic cholera [19] , [43] , [44] . Perhaps selective pressure for the Lewis blood group antigens is weaker than for the ABO blood group system , and the endemicity of cholera and ETEC diarrhea in settings such as Bangladesh may be partially explained by the increased presence of the Le ( a+b− ) phenotype . Factors influencing susceptibility to cholera may also play a role in responses to cholera vaccines . For example , in a large scale field trial conducted in Matlab , Bangladesh of the role of ABO blood group and efficacy of an oral , killed cholera vaccine , there was substantially lower protection in recipients who were blood group O [45] . We have also shown previously that in Bangladeshi children receiving a live , oral attenuated cholera vaccine , Peru 15 , the frequency of serological responders was higher in children of the A blood group compared to the O blood group [46] . Thus , the ABO blood group system is a potential factor that may affect vaccination efficacy in different settings , and this factor has now been included in the assessment of ongoing cholera vaccine trials . In the present study , we found that individuals who were negative for both Lewis antigens ‘a’ and ‘b’ had impaired LPS-specific IgA responses on day 7 compared to individuals with Le ( a–b+ ) Lewis antigen phenotypes . This suggests that inclusion of the Lewis blood group should be considered in cholera vaccine efficacy trials in the future as well as the ABO blood group types . | Cholera remains a severe diarrheal disease , capable of causing extensive outbreaks and high mortality . Blood group is one of the genetic factors determining predisposition to disease , including infectious diseases . Expression of different Lewis or ABO blood group types has been shown to be associated with risk of different enteric infections . For example , individuals of blood group O have a higher risk of severe illness due to V . cholerae compared to those with non-blood group O antigens . In this study , we have determined the relationship of the Lewis blood group antigen phenotypes with the risk of symptomatic cholera as well as the severity of disease and immune responses following infection . We show that individuals expressing the Le ( a+b− ) phenotype were more susceptible to symptomatic cholera , while Le ( a–b+ ) expressing individuals were less susceptible . Individuals with the Le ( a–b− ) blood group had a longer duration of diarrhea when infected , required more intravenous fluid replacement , and had lower plasma IgA antibody responses to V . cholerae LPS on day 7 following infection . We conclude that there is an association between the Lewis blood group and the risk of cholera , and that this risk may affect the outcome of infection as well as possibly the efficacy of vaccination . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"immunology",
"biology"
] | 2011 | Individuals with Le(a+b−) Blood Group Have Increased Susceptibility to Symptomatic Vibrio cholerae O1 Infection |
Recognition of single-stranded DNA ( ssDNA ) or single-stranded RNA ( ssRNA ) is important for many fundamental cellular functions . A variety of single-stranded DNA-binding proteins ( ssDBPs ) and single-stranded RNA-binding proteins ( ssRBPs ) have evolved that bind ssDNA and ssRNA , respectively , with varying degree of affinities and specificities to form complexes . Structural studies of these complexes provide key insights into their recognition mechanism . However , computational modeling of the specific recognition process and to predict the structure of the complex is challenging , primarily due to the heterogeneity of their binding energy landscape and the greater flexibility of ssDNA or ssRNA compared with double-stranded nucleic acids . Consequently , considerably fewer computational studies have explored interactions between proteins and single-stranded nucleic acids compared with protein interactions with double-stranded nucleic acids . Here , we report a newly developed energy-based coarse-grained model to predict the structure of ssDNA–ssDBP and ssRNA–ssRBP complexes and to assess their sequence-specific interactions and stabilities . We tuned two factors that can modulate specific recognition: base–aromatic stacking strength and the flexibility of the single-stranded nucleic acid . The model was successfully applied to predict the binding conformations of 12 distinct ssDBP and ssRBP structures with their cognate ssDNA and ssRNA partners having various sequences . Estimated binding energies agreed well with the corresponding experimental binding affinities . Bound conformations from the simulation showed a funnel-shaped binding energy distribution where the native-like conformations corresponded to the energy minima . The various ssDNA–protein and ssRNA–protein complexes differed in the balance of electrostatic and aromatic energies . The lower affinity of the ssRNA–ssRBP complexes compared with the ssDNA–ssDBP complexes stems from lower flexibility of ssRNA compared to ssDNA , which results in higher rate constants for the dissociation of the complex ( koff ) for complexes involving the former .
Interactions between nucleic acids and proteins are essential and central to many biochemical processes . Protein–nucleic acid complexes have very diverse structures and the interface may depend on both the shape of the protein and the structure of the nucleic acid . The diversity of DNA and RNA sequences dictates their structures , which in turn control their binding specificity to proteins . The structure of protein–DNA complexes may vary and sometimes even small nuances in the geometrical parameters of the major or minor grooves are fundamental to achieving specificity [1 , 2] and therefore function . An RNA strand can fold into diverse three-dimensional ( 3D ) structures , including double-stranded A-form helices and higher-order tertiary structures [3] that interact specifically with proteins . Stable complexes between proteins and nucleic acids are essential and their disruption can lead to a range of diseases [4] , including several neurodegenerative disorders [5] and cancers [6] . Structures can be formed transiently between proteins and double-stranded DNA ( dsDNA ) during transcription , replication , recombination , and dsDNA repair . Structures between proteins and single-stranded ( ss ) DNA and RNA are also essential for function , for example , in telomeric overhangs at the end of chromosomes , at double stranded breaks , and at replication forks [7 , 8] . Compared with dsDNA , ssDNA structures are highly flexible [9–12] and their functional form is thermodynamically less stable , such that they are vulnerable either to forming secondary structures in which the nucleotide groups are non-accessible or to re-annealing with complementary DNA strands . They are also susceptible to detrimental chemical or enzymatic attacks . Various proteins function to specifically bind to and thereby protect ssDNA molecules so that they can take part in necessary cellular processes . Some ssDNA binding proteins ( ssDBPs; often called SSBs ) have the functional ability to recruit partner proteins and present the ssDNA substrate to them [13] . The structures of ssDBPs can vary in size and shape , and many of them consist of one or more copies of unique binding domains . Four such domains having distinct structural topologies have been characterized so far and their available structures reveal their mode of interaction with ssDNAs . These ssDBP domains are oligonucleotide/oligosaccharide/oligopeptide-binding ( OB ) folds , K homology ( KH ) domains , RNA recognition motifs ( RRMs ) , and whirly domains . In a multi-domain ssDBP , domains either repeat in the same subunit or monomeric domains fold into a homo-oligomeric tertiary structure and all the domains conjointly bind ssDNA [14] . The situation is somewhat similar with respect to ssRNAs , which are an important component of RNA biology [15 , 16] . RNA binding proteins ( RBPs ) bind single-stranded RNA ( ssRNA ) and act either as essential cofactors for their functional activity or to protect them from degradation . The structures of ssRNA binding proteins ( ssRBPs ) vary in shape and size , and some of them consist of more than one copy of the binding domain . The complex structures that some of the abundant ssRBP domains form with ssRNA , such as RRMs , Pumilio repeat domains ( PUFs ) , KH domains , OB fold domains , and tristetraprolin and CCCH-type zinc fingers ( e . g . , Tis11d ) , have been solved . However , the structural basis of their sequence specificity is often not clear . Studying the conformational heterogeneity of ssDNA and ssRNA is challenging using common approaches because they can provide only limited information either on the global conformation or on the detailed molecular characteristics . Nevertheless , ssDNA and ssRNA were studied by atomic force microscopy ( AFM; [17 , 18] ) , fluorescence resonance energy transfer ( FRET;[19] ) , nuclear magnetic resonance ( NMR;[20–22] ) and small angle x-ray scattering ( SAXS;[23 , 24] ) . The interactions between proteins and ssDNA or ssRNA were studied , however , the number of studied crystal structures is much smaller for ssDNA and ssRNA compared with dsDNA or dsRNA and it is unclear how they interact in solution . Interactions between ssDNA and ssDBP or between ssRNA and ssRBP are fundamentally different from the interactions of dsDNA or dsRNA with proteins . Predicting their structures is complicated by the much greater flexibility of ssDNA/ssRNA compared with their double-stranded analogs . In many cases , ssDNA/ssRNA molecules of variable sequences but of similar length are able to adopt different conformations to engage with the same protein binding site . It was reported that the binding mode adopted is affected by salt concentration . For example , an ssDBP interacts differently with ssDNA at low and high salt concentrations [25] . In the case of ssRNA binding , although the same RRM surface is used to contact various ssRNAs , substantial variation exists in their interaction modes , in the number of interacting bases , and in their degree of specificity [26] . Additionally , the complexity of these interactions is reflected in the high thermodynamic stability of the formed complexes even when they interact with homopolymeric single stranded nucleic acids . Some of these complexes can even have an experimentally resolved structure in which the ssDNA can participate in extensive diffusion along the protein [27] . Although electrostatics ( in which the negatively charged backbone of the nucleic acid is attracted to the positively charged residues on the binding surface of the protein ) plays a crucial role in the interactions of both single and double-stranded nucleic acids with proteins , ssDNAs and ssRNAs are highly flexible in solution and thus they do not possess a definite shape [28] . By contrast , dsDNA and dsRNA are much more rigid and therefore their complexes with proteins often possess shape complementarity . Unlike dsDNA , the bases of ssDNA can be unstacked in the unbound form and thus are capable of engaging in π–π stacking interactions with the aromatic side chains ( tryptophan ( W ) , tyrosine ( Y ) , phenylalanine ( F ) , and histidine ( H ) ) of ssDBPs . This scenario is also valid for the interaction of ssRNA with ssRBPs . Since there is little experimental information on the conformations of ssDNA or ssRNA in solution , most reported studies have focused on the conformations of the protein . The interactions between single stranded nucleic acids and proteins have different biological functions , some of which demand sequence specificity . Complexes that are formed to protect the ssDNA from hybridization with another ssDNA are expected to be less specific and some of them were also shown to involve diffusion of the DBP along the ssDNA , so indicating the formation of various interfaces between the ssDNA and the DBP [29 , 30] . Indeed , several ssDBPs were crystallized with a non-specific ssDNA sequence , such as poly-T [14 , 31–33] . Some ssDBP molecules , however , such as the telomere-binding proteins , bind ssDNA in a sequence-specific manner . For example , Pot1p from S . pombe binds a hexanucleotide ssDNA sequence strongly with an equilibrium dissociation constant , KD , in the nano-molar range but does not bind when a single nucleotide at the center of the sequence is altered [34] . For homo oligonucleotide single strand sequences ( poly-A , poly-T etc . ) , the KD equilibrium binding constant of a particular DBP can vary depending on the nucleotide type; poly-A ssDNA binds RPA with a KD that is orders of magnitude higher than that of poly-T [35 , 36] . Even for the same binding mode , the OB-fold from cold shock protein ( CSP ) binds T rich sequences tighter than C rich ones [37] . Also , ssDNA sequences bind much tighter than ssRNA sequences [34 , 38 , 39] . In cases where the cognate ssDNA sequence binds tightly , other non-cognate sequences can also bind to the same binding site [14] . This accommodation of different sequences is possible where the protein adjusts its backbone , relocates its flexible side chains , and alters its hydrogen bonding networks and where the DNA strand undergoes structural rearrangements , mainly by rotating its bases [14 , 40] . Sometimes , specificity is biased toward one end of the ssDNA . For example , both S . pombe Pot1 and Cdc13 recognize a particular telomeric sequence in the 5’ region but their binding at the 3’ region is less specific [41 , 42] . Computational approaches can provide a powerful means of studying the complexes between proteins and ssDNA or ssRNA , particularly with respect to their dynamics and functional motions . However , only a few such studies have been reported . Atomistic molecular dynamics simulations were applied to study complexes of ssDNA with the SSB protein [43] , with the RPA protein [44] , and with a KH domain [45 , 46] . Coarse-grained molecular dynamics simulations were used to study the self-assembly of several protein-ssDNA complexes [47] . A few studies have reported the development of a computational algorithm to study the interactions of ssRNA with proteins . The major examples are an atomic distance- and orientation-based scoring function [48] , a machine learning-based docking-score in RosettaDock [49] , an energy-based coarse-grained force field [50 , 51] , and a fragment-based flexible docking score [52 , 53] . Most of these knowledge-based algorithms were evaluated on small data sets because of the limited number of experimental structures available , which limits their coverage . Moreover , considering the RNA structure as a rigid body makes them inapplicable for the modeling of ssRNA binding , in which flexibility plays a crucial role . Applying similar approaches to ssDNA-protein complexes is challenging mostly because of the small number of available structures and low sequence similarities , which hampers efforts to construct knowledge-based potentials . Here , we applied a physical interaction-based coarse-grained approach to construct a transferable model to study the recognition of ssDNAs and ssRNAs by ssDBPs and by ssRBPs , respectively . The method does not require any structural information on ssDNA/ssRNA , nor does it utilize any prior knowledge of the binding site . Earlier , we reported a similar model that successfully predicted the crystal complex structures of homopolymeric ssDNAs with ssDBPs coming from different domains of life [47] . New parameters have been incorporated into the current model to account for sequence-specific interactions with ssDNA/ssRNA . The two major components of the coarse-grained model that govern the interactions and stability of the complexes formed between ssDNA/ssRNA and their corresponding proteins are the flexibility of the nucleic acids and the strength of interactions between each nucleotide and the aromatic sidechains . The interface between ssDNA/ssRNA and proteins is defined by electrostatic interactions between the phosphate backbone and positively charged residues and by aromatic interactions between nucleic acid bases and aromatic residues . The sequence specificity is mostly introduced by different strengths of interactions between the four types of nucleotides ( A , T/U , G , C ) and the four types of aromatic side chains ( W , F , Y and H ) . The new coarse-grained model was applied to six ssDNA–ssDBP and six ssRNA–ssRBP complexes involving binding proteins having different protein folds and ssDNA/ssRNA molecules having different lengths and sequences . The model predicted their structures successfully , was sensitive to the sequence variation of the ssDNA or the ssRNA , and qualitatively predicted their experimental binding affinities .
For a comprehensive analysis of a variety of interactions between proteins and single-stranded nucleic acids , we studied 12 complexes: six ssDNA–ssDBP complexes and six ssRNA–ssRBP complexes whose three-dimensional structures are known ( summarized in Table 1 ) . The sets of protein–DNA and protein–RNA complexes include proteins having different functions , with folds of different sizes , and with heterogeneous ssDNA/ssRNA having different lengths and sequences . The proteins in these ssDNA–ssDBP complexes belong to different structural domains: the oligonucleotide/oligosaccharide-binding ( OB ) fold , the RNA recognition motif ( RRM ) domain , and the K homology ( KH ) domain . We note that the four complexes with OB-folds differ in their structures ( i . e . , protein length ) and sequences . Likewise , the six ssRNA–ssRBP complexes were also selected to cover different structural domains , namely , the OB-fold , RRM , PUF domain , zinc-finger domain , RAMP protein , and a Fab . Overall , we covered different folds in which the electrostatic and aromatic stacking energy contributions vary from a very high stacking energy fraction ( the OB-fold ) to a high electrostatic energy fraction ( KH-domain and RAMP ) . Judging from the available structures of the 12 complexes studied here and based on the available unbound structures , it appears unlikely that the proteins undergo a considerable conformational change in order to bind their ssDNA/ssRNA ligands . The ssDNA and ssRNA molecules are much more flexible in solution than folded proteins . Accordingly , one may conclude that the binding surfaces in ssDBPs and ssRBPs are predefined , and large conformational change occurs for ssDNA/ssRNA only . In many cases , proteins bind to cognate ssDNA or ssRNA partners in a sequence specific manner , where the binding specificity depends on the interactions between nucleotide bases and aromatic side chains . To model the sequence specific interaction of ssDNA and ssRNA with proteins , we adopted the coarse-grained model that was originally developed to study nonspecific ssDNA–ssDBP interactions [47 , 54] . Starting from the experimentally determined structures , the ssDBPs and ssRBPs were modeled by their native topology , where each amino acid residue was represented by two beads at the Cα- and Cβ-positions except Gly , which has only Cα . Charged amino acids were modeled by placing a point charge of +1 ( Lys and Arg ) or -1 ( Asp and Glu ) on the Cβ-bead . In some cases , His was also considered as positively charged depending on its estimated pKa values on the Propka server [55] . The ssDNA and ssRNA molecules were modeled using a coarse-grained approach as ‘beads-on-a-string’ polymers in which each nucleotide was represented by three beads representing the phosphate ( P ) , sugar ( S ) , and nucleo-base ( B ) moieties , which were positioned at the geometric center of each represented group . The phosphate bead in the model bears a -1 charge . In order to maintain chain connectivity and local geometry , the neighboring beads were constrained using bonds , bond angles , and dihedral angles . Non-bonded interactions are crucial to model the dynamics of ssDNA and ssRNA molecules . In our model , we included base-stacking interactions and hydrophobic interactions , as described below . Given the short length of ssDNA and ssRNA for all the systems studied here , the present model did not consider the possibility of intra-DNA base-pairing interactions . In the simulation , the native contact interactions of the protein were maintained by the Lennard-Jones ( L-J ) potential , whereas nonspecific electrostatic interactions were allowed among all charged residue beads . Overall , we followed a coarse-grained protein modeling approach that was used in previous studies [47 , 56–58] . The internal energy of the protein Eprot comprises the following three bonded and three non-bonded terms: Eprot=EprotBond+EprotAngle+EprotDihedral+EprotNativecontacts+EprotElectrostatics+EprotRepulsion The potential of a particular conformation Γ ( Γ0 is the native conformation ) in the molecular dynamics ( MD ) trajectory is then described as: Eprot ( Γ , Γ0 ) =∑bondsKbonds ( bij‐bij0 ) 2+∑anglesKangles ( θijk‐θijk0 ) 2+∑dihedralsKdihedrals[1‐cos ( ϕijkl‐ϕijkl0 ) −cos ( 3 ( ϕijkl‐ϕijkl0 ) ) ]+∑ i≠jKcontacts[5 ( Aijrij ) 12‐6 ( Aijrij ) 10]+∑ i , jKelectrostaticsB ( κ ) qiqjexp−krεrrij+∑ i≠jKrepulsion ( cijrijr ) 12 The value of the constant Kbonds was set to 100 kcal mol-1 Å-2 , the value of Kangles was set to 20 kcal mol-1 Å-2 and the values of constants Kdihedrals , Kcontacts and Krepulsion were set to 1 kcal mol-1 . For a given conformation along the trajectory , bij is the distance between bonded beads i and j and bij0 is the optimum inter-bead distance in Å . Similarly , θijk is the angle between sequentially bonded beads i-k and θijk0 is their optimum angle in radians; ϕijkl is the dihedral angle between sequentially bonded backbone beads i-l and ϕijkl0 is their optimal dihedral angle in radians . Finally , rij is the distance between non-bonded beads i and j that are in contact and Aij is their optimal distance in Å . Optimal values were calculated from the atomic coordinates of the corresponding PDB structure . For the repulsion term , Cij is the sum of the radii for any two non-bonded beads not forming a native contact and rijr is the distance between them in Å; the repulsion radii for the backbone and side chain ( Cβ ) beads were set to 1 . 9 Å and 1 . 5 Å , respectively . The electrostatic interactions were modeled by the Debye-Hückel potential , and we followed the parameters used in previous studies in our group [57 , 58] . In the coarse-grained model , the inherent flexibility of protein segments varies as a function of the density of the native contacts in the local surroundings . In addition , we incorporated enhanced flexibility for segments either with high B factors ( i . e . , higher than the mean B factor ) or with missing coordinates . For complexes that were resolved by NMR ( 1s40 . pdb ) , the flexible regions were predicted using FlexServ [59] . In order to retain the unimpaired native fold of the protein including its binding site , all simulations were run at relatively low temperatures to allow the protein to fluctuate around its native state but not to unfold . All simulations were started from the extended conformation of the ssDNA or ssRNA . In contrast to the modeled proteins , there were no native contact interactions for ssDNA/ssRNA . There are several models for ssDNA that aim to capture sequence-dependent polymeric properties ( e . g . , persistence length and force-extension profiles ) [60–65] . The current model was based on one we developed for homopolymeric ssDNA that successfully predicted binding with ssDBPs[47] . In this model , intra-molecular electrostatic repulsions were not allowed between negatively charged phosphate beads , and the ssDNA/ssRNA flexibility was modulated by the two dihedral potentials described below . Consistently with previously reported studies [50 , 60 , 61 , 66] , the following are the potential energy terms of the ssDNA and ssRNA used in our model: EssDNA/ssRNA=EssD/RNABond+EssD/RNAAngle+EssD/RNADihedral+EssD/RNABasepairing+EssD/RNAStacking+EssD/RNARepulsion Here , the first three terms are responsible for retaining the ssDNA/ssRNA backbone and their forms are identical to the corresponding terms in Eprot . The term EssD/RNABond represents the contribution from the covalently linked beads and comes from the following bead-pairs: ( Pi-Si ) , ( Si-Bi ) , and ( Si-Pi+1 ) with Kbonds = 100 kcal mol-1 Å-2 . The term EssD/RNAAngle is the bond angel potential and comes from the following bead-trios: ( Pi-Si-Bi ) , ( Bi-Si-Pi+1 ) , ( Pi-Si-Pi+1 ) , and ( Si-Pi+1-Si+1 ) , with Kbonds = 20 kcal mol-1 Å-2 . The term EssD/RNADihedral is the potential for the dihedral angles included to mimic the flexibility of ssDNA or ssRNA in the model . We introduced two types of dihedral potentials: i ) those formed between the following four consecutive base and sugar beads to modulate the flexibility of the base–sugar moiety: Bi , Si , Si+1 , and Bi+1 with Kdihedrals = 0 . 5 kcal mol-1 and 1 . 5 kcal mol-1 for ssDNA and ssRNA , respectively; and ii ) those formed between four consecutive phosphate beads to modulate the flexibility of the phosphate backbone: with Pi , Pi+1 , Pi+2 , and Pi+3 Kdihedrals = 0 . 3 kcal mol-1 and 0 . 9 kcal mol-1 for ssDNA and ssRNA , respectively . The values were calibrated so that the persistence length of ssDNA/ssRNA calculated from the simulations qualitatively resembled that observed in experiments ( see below ) . The values of the native bond lengths and angles were obtained from the PDB atomic coordinates of the helical structure that ssDNA adopts in the duplex form . The first two terms in the potential energy equation dictate the connectivity of the ssDNA/ssRNA and the other four terms dictate the global conformation . Base-pairing and base stacking may contribute to the structural stability of the ssDNA/ssRNA . All ssDNA and ssRNA systems studied here were of short length , moreover , the homopolymeric nature of some of the sequences restricted the possibility of base-pair formation . We thus set EssDNABasepairing = 0 and kept this potential for future studies with longer ssDNA segments . The attractive nature of the π-stacking between consecutive bases was incorporated by using a short range L-J potential between consecutive ssDNA/ssRNA bases: EssDNAStacking=‐εB−B[5 ( rij0rij ) 12‐6 ( rij0rij ) 10] , with rij0 being the typical distance between consecutive bases and set to 3 . 6 Å[67] . Different stacking interaction strengths ( εB−B ) represent different depths of the potential well between the neighboring stacked bases and can take different values depending on the nature of the two bases . Previous efforts have endeavored to estimate the interaction energies of stacked nucleobase dimers experimentally [68] and from quantum chemical calculations [69] . Though obtained differently , their trends are similar , as expected . For example , in both cases , Guanine was found to have lower interaction energies ( engage in stronger interactions ) compared with Thymine . This set of interaction energies was used to assess base–base stacking interaction strengths in ssDNA coarse-grained models to elucidate ssDNA dynamics [70] and DNA hybridization[71] . We adopted the energetic values for stacking εB-B for different base pairs from an earlier study [71] and rescaled the values to fit the experimental persistence length of poly-T ssDNA ( Table 2 ) . In the model , the base stacking is strongest for purines and weakest for pyrimidines . Adopting an approach similar to that used with the proteins , we applied a repulsion term ( i . e . , excluded volume ) to all non-bonded beads in ssDNA/ssRNA . This repulsion energy was applied to any beads of non-adjacent nucleotides; the radii of the base , phosphate and sugar beads were 1 . 5 Å , 3 . 7 Å , and 3 . 7 Å , respectively . A major challenge in predicting the complexes formed between proteins and ssDNA/ssRNA stems from the considerable flexibility of the latter . Their flexibility is linked to electrostatic repulsions between negatively charged phosphate groups and can , therefore , be modulated by salt concentration . Indeed , the persistence length of ssDNA decreases whereas its contour length increases with increasing salt concentration [12 , 19 , 72] . In our coarse-grained simulation , the effect of salt concentration was incorporated by using the Debye-Hückel potential , which modulates the ssDNA persistence length as well as electrostatic interactions at the protein–ssDNA/ssRNA interface . To modulate the flexibility of the ssDNA and ssRNA , we omitted ion condensation effects and simplified the representation of the effect of electrostatics on the ssDNA/ssRNA persistence length by adding the two dihedral potentials described above . We calculated the persistence length ( Lps ) of the modeled ssDNA and ssRNA using the expression for a flexible polymer ( L/Lps >> 1 ) : Lps = <Ll0>/<l0> , where l0 is the vector between the first two monomers ( the bond vector between the two phosphate beads at the 5’ end ) , and L is the end to end vector ( the bond vector between the phosphate beads at the 5’ and 3’ end ) of the polymer . In the model , Lps for a T40 polymer initially increased with the backbone dihedral potential ( from 0 to 1 . 2 kcal mol-1 ) and saturated thereafter . The persistence length obtained using this approach is in agreement with experimental values and is consistent with the values from other computational approaches[47] . Here , the values of Kdihedral were chosen such that the relative persistence lengths of ssDNA and ssRNA agreed with the experimentally determined range . The experimentally reported values of the persistence lengths of ssDNA and ssRNA span a wide range of 1 . 0–6 . 0 nm that is sensitive to the solution condition ( e . g . , ionic strength and ion types ) and experimental technique ( e . g . , FRET , SAXS , and AFM ) . Several studies reported higher persistence length for ssRNA than ssDNA [12 , 19] . In a recent comparative study , by fitting SAXS data with a worm-like chain model , the persistence length of dT40 ( 16–19 Å ) was found to be less than that of dU40 ( 19–22 Å ) at a particular salt concentration [12] . We mimicked the lower persistence length of ssDNA by modeling it with a lower backbone dihedral constant ( Kdihedral = 0 . 3 ) and the higher persistence length of ssRNA with a higher dihedral constant ( Kdihedral = 0 . 9 ) . The Lps values for ssDNA and ssRNA were 30 Å and 42 Å , respectively , which is in the range of the experimentally measured flexibility of ssDNA and ssRNA [11 , 67] . These values yielded a persistence length for ssRNA that was 30% larger than that of ssDNA , consistently with the ratio estimated by the SAXS measurements [12] . This difference between the flexibility of ssDNA and ssRNA was needed to reproduce their different binding affinities to ssDBPs and ssRBPs , respectively . In our model , the interaction potential between a protein and ssDNA/ssRNA comprises the following three components: ( i ) the electrostatic interaction between the Cβ-beads representing the side chain of charged residues ( K , R , H , D , and E ) and the negatively charged phosphate beads of ssDNA/ssRNA; ( ii ) the aromatic stacking interaction between the Cβ-beads representing aromatic side chains ( W , F , Y , and H ) and the ssDNA/ssRNA base bead; and ( iii ) the repulsive interactions between other beads of the protein and ssDNA/ssRNA . Thus , Eprot−ssD/RNA=Eprot−ssD/RNAElectrostatics+Eprot−ssD/RNAAromatic+Eprot−ssD/RNARepulsion The electrostatic interactions between all of the charged beads in the system are modeled by the Debye–Hückel potential . These interactions are nonspecific , and the phosphate groups of the ssDNA/ssRNA can interact with any charged residue of the ssDBP or ssRBP , respectively . The repulsion is applied to all beads of the protein and all beads of the ssDNA/ssRNA . Unlike dsDNA , the nucleobases of extended ssDNA/ssRNA are free to engage with aromatic residues in π–π stacking , which plays a crucial role in protein–ssDNA/ssRNA interactions . These stacking interactions were characterized and compared using detailed quantum chemical calculations [73 , 74] . The energies of these interactions were estimated to range between -9 . 4 and -28 . 5 kJ∙mol-1 in water; suggesting that the π–π stacking interactions play an important role in stabilizing the interface between proteins and ssDNA/ssRNA . Stacking energy increases with the amino acid according to Phe < His ≈ Tyr < Trp , while the stacking energy is generally larger for purines compared with pyrimidines[73] . Similarly to base stacking , the aromatic interactions between the Cβ-beads of aromatic side chains ( W , F , Y , and H ) and the nucleotide base bead is also modeled by the L-J potential and weighted by the base–aromatic interaction strength εB−AA ( Fig 1 ) . Thus , EssDNA/ssRNAAromatic=‐εB−AA[5 ( rij0rij ) 12‐6 ( rij0rij ) 10] where rij0 is the average distance between an aromatic side chain and a base and was set as 3 . 6 Å . The value of εB−AA varies depending on the B-AA pair . We adopted these pairwise base-aromatic energy values from the studies of Rutledge et al . , [73] . To scale these values to fit appropriately into our model , we reweighted the sets of εB−AA values by a factor of 0 . 15 to maximize the populations of the native state binding mode and to minimize its binding energy . The composition of ssRNA differs from that of ssDNA only by a single nucleotide ( uracil in place of thymine ) . Based on their chemical similarity , uracil and thymine are expected to have similar stacking energies . Indeed , it was estimated that the stacking energy for uracil is only 8–10% lower than that of thymine[73 , 74] . In the model , the base–aromatic energy for ssDNA and ssRNA only differs for uracil and thymine . We compared the εB−AA values used in the model with the similar values reported recently , which were calculated by the potential mean force [75] and free energy estimation [76] methods from all-atom molecular dynamics simulations with explicit water . The εB−AA parameters in the model correlated well with both of these sets , the corresponding correlation coefficients are 0 . 57 and 0 . 71 with the potential mean force and free energy methods , respectively ( Fig 1B ) . We note that the correlation coefficients improved when the values corresponding to interactions with Tyr were excluded ( r = 0 . 80 and 0 . 87 , respectively ) . The details of the model are schematically summarized in Fig 2 . The dynamics of the protein and ssDNA were simulated using Langevin dynamics and deploying the total potential energy Eprot + EssDNA/ssRNA + Eprot−ssD/RNA of the system . All simulations were performed with an implicit solvent model of dielectric constant 70 ( water ) and at a 10 mM salt concentration . We point out that , because of the coarse-grained representation of the systems , the effective salt concentration may correspond to a higher value ( by a factor of ~3 ) than for an atomistic representation . We chose a temperature of 0 . 3 ( arbitrary units ) , at which the protein was shown to fluctuate around its native fold , and the ssDNA/ssRNA was able to perform an extensive search that included diffusion over the protein surface . At this temperature , the bound state was thermodynamically more favorable and thus more populated than the dissociated state . Importantly , at this temperature , the persistence length of the modeled ssDNA/ssRNA fit the related experimental values . The model was initially tested for its ability to maintain the native bound structure of all of the six ssDNA–DBP complexes and the six ssRNA–RBP complexes when the simulations were started from the bound conformations . We started the predictive simulations by placing an unbound ssDNA/ssRNA molecule ( in its linear form ) at one of six different positions around the DBP or RBP at a distance of 35–40Å . For each ssDNA/RNA position , 100 replications were performed and , in each case , a unique random seed was used to generate different velocity distributions . Thus , a total of 600 simulations were run for each system in order to perform extensive sampling of association mechanisms having multiple binding routes . Each trajectory was simulated for 107 molecular dynamics steps with a time step of 0 . 005 . Conformations were saved every 1000 molecular dynamics steps , thus 10000 conformations were saved in each trajectory . Finally , to consider the part of the trajectory that was equilibrated well , the last 2000 conformations were collected from each trajectory for analysis , such that we analyzed a total of 12 × 105 ( 6 × 100 × 2000 ) conformations per system . To evaluate the sampled conformational ensemble and especially to examine the deviation of the simulated bound conformations from the experimental structure , we utilized two similarity parameters: DSite and DConf . Both these parameters quantify the similarity between the crystal and simulated conformations of the binding interface in the ssDNA–protein or ssRNA–protein complex . The DSite term achieves this by probing the protein patch used for interaction with either ssDNA or ssRNA , whereas the Dconf term probes the conformation of the ssDNA/ssRNA in this site . Consequently , DConf is sensitive to whether the ssDNA/ssRNA is located at the binding site from 3' to 5' or vice versa , however , DSite quantifies the location and conformation irrespective of ssDNA/ssRNA directionality . To calculate DSite and DConf , the Cβ-beads of all the positively charged ( K and R ) as well as the aromatic ( W , Y , F and H ) residues in the experimentally resolved structures of the ssDNA–DBP or ssRNA–RBP interfaces were identified . Any positive residue bead lying within a cutoff distance ( 9 Å ) from any phosphate bead , and any aromatic residue bead lying within the same cutoff distance from any base bead were defined as the native interfacial residues . The size of the interface varied between the different systems . The following equation was then applied to the interfacial residue and the ssDNA/ssRNA phosphate or base to calculate the crystal structure similarity parameter , DSite . DSite=1NprotNssDNA∑iNprot ( ∑jssDNArij−∑jssDNArij0 ) Here , i and j are the ith bead of the selected Cβ of the protein and the jth bead of the ssDNA/ssRNA , respectively , and thus , rij and rij0 are the pairwise distances between those beads in the simulated structure and the crystal structure , respectively . The pairwise distances were calculated either between each selected aromatic amino acid and all of the base beads of the ssDNA/ssRNA or between each of the selected charged amino acids and all of the phosphate beads of the ssDNA/ssRNA . Nprot is the total number of selected interfacial amino acid residues ( positively charged or aromatic ) , and NssDNA is the number of nucleotides in the length of ssDNA/ssRNA examined . Thus , the term DSite quantifies the overall conformational similarity between the predicted binding interface in the ssDNA–protein complex and the crystal structure , with a DSite = 0 indicating 100% conformational similarity . To obtain a finer structural quantification of the interface , we divided the selected interfacial residues into two groups that covered two different regions of the interface . We then calculated two order parameters , D1Site and D2Site , each characterizing the accuracy of the prediction for the corresponding region of the interface . The advantage of the DSite measure ( compared with other structural measures , such as root mean square deviation ( RMSD ) ) is that it quantifies the location and conformation of the ssDNA/ssRNA relative to each potential interfacial residue of the ssDBP or the ssRBP and ignores the directionality of the ssDNA/ssRNA . For example , a situation in which the ssDNA ( which is poly T and lacks polarity in the model ) is perfectly located at the protein interface but is flipped from 3′ to 5′ ( instead 5′ to 3′ ) will result in a large RMSD value but a very low DSite value . The other structural similarity parameter , Dconf , was calculated where the 5’ to 3’ direction of the bound ssDNA/ssRNA was taken into consideration . The same set of interface residues was identified first using the same criteria as used for DSite . Next , a list was made of native pairwise interactions between a base bead and the Cβ-bead of the nearest aromatic residue or between a phosphate bead and the Cβ-bead of the nearest positively charged residue . The following equation was then applied to this pairwise interfacial interaction to calculate Dconf: Dconf=1NssDNA∑iNssDNA ( ri−ri0 ) Where , ri0 is the distance of the ith pair of base–aromatic or phosphate–positive beads from the above list , ri is the corresponding value for the simulated structure . NssDNA is the total number of ssDNA/ssRNA base and phosphate beads . Thus , the term Dconf quantifies the conformational similarity between the predicted binding interface in the ssDNA/ssRNA–protein complex and the crystal structure considering ssDNA/ssRNA direction . As with DSite , Dconf = 0 Å corresponds to 100% conformational similarity . The values of DSite and DConf can be calculated also for a specific region of the interface formed between the ssDNA or the ssRNA and the protein . In this case , for DSite only the relevant interfacial residues will be used and for DConf the relevant pairwise distances will be taken into account .
Although different ssDBPs and ssRBPs perform different cellular functions , the actual number of distinct domains found in both cases is limited . The ssDBPs arrange these domains in a modular way to achieve different structures for distinct activities , including ligand specificities . In this study , we considered all three types of ssDBP domain whose complete structures are available , namely , OB folds , KH domains , and RRMs . We studied four different kinds of single OB-fold structures of variable sizes ( 67 to 187 residues ) and different lengths and sequences of ssDNA ( Table 1 ) . These proteins are capable of binding specific ssDNA sequences with different affinities ( particularly for the telomere proteins ) . The six studied ssDNA–ssDBP complexes differ in the relative contributions made by electrostatic and aromatic energies . For example , ssDNA typically binds OB-folds such that the bases facing the protein participate in both intra- and inter-molecular aromatic stacking interactions and the backbone remains exposed to the solvent , but for the KH domain , the electrostatic energy is much larger . For simplicity , we did not include multi-domain ssDBPs , such as PARP1 [77] , which bind folded ssDNA with definite secondary structures , or proteins such as RPA , which demand high flexibility ( i . e . , undergo conformational changes ) in order to bind ssDNA [33] . The selected ssRNA–ssRBP systems ( Table 1 ) represent the four most abundant ssRNA-binding domains in proteins: RRMs , PUF , CCCH-type zinc fingers , and OB fold domains . Their abundance suggests that these folds have the versatility to function as diverse recognition modules . Indeed , they possess modular structures of multiple repeats that arrange to create versatile RNA-binding surfaces . Additionally , two more unique structures , an engineered synthetic antibody fragment and a RAMP that binds single-stranded CRISPR Repeat RNA , were also included . The RRM is among the most abundant structural motifs and approximately 500 human proteins contain RRMs , often in multiple copies in the same polypeptide chain . The performance of the developed coarse-grained model in studying protein–ssDNA/ssRNA interactions was tested by quantifying the binding mode of the sampled conformations of the twelve simulated systems and by comparing them with the corresponding X-ray or NMR structures . Considering both the flexible nature of ssDNAs/ssRNAs and their linear shape , a more detailed structural comparison can be achieved by dividing the ssDNA–ssDBP and the ssRNA–ssRBP interfaces of the experimental structures into two moieties . Splitting the interface into two moieties is useful to estimate the similarity of each of them to the corresponding region in the experimentally resolved structure . Thus , in the context of calculations to determine the similarity between the crystal and simulated conformations of the binding interface in the ssDNA–protein or ssRNA–protein complex , D1Site and D2Site indicate whether the ssDNA or ssRNA interacts with the native patches on the protein linked with the two moieties that comprise the experimental interface . Similarly , D1Conf and D2Conf describe the conformation and directionality of the ssDNA or ssRNA ( details in Methods ) . We note that the two moieties have similar contribution to the interface stability ( each contributes 40–60% to the interface energy ) . DConf thus reflects the molecular identity of the interactions at the interface and is a more appropriate measure than DSite when examining binding specificity . Small values for these structural measures correspond to conformations having a greater degree of similarity with the experimental structure , in which ( DSite1 , DSite2 ) or ( DConf1 , DConf2 ) equals ( 0 , 0 ) . Fig 3 shows the sampled conformational ensembles for three simulated ssDNA–ssDBP complexes and three ssRNA–ssRBP complexes projected along ( DSite1 , DSite2 ) or ( DConf1 , DConf2 ) ( the other six simulated ssDNA–DBP and ssRNA–RBP conformations are shown in the Supporting Information ) . The free energy surface of the binding process for each of the different studied folds ( in which the interaction between the ssDBP and ssDNA or between the ssRBP and ssRNA was modeled by combining electrostatic and aromatic interactions ) reflects that , in all six cases , near-native conformations with low values of DSite1 and DSite2 are highly populated ( blue in Figs 3 and S1 ) . We note that the sequence independent model captures the complexes of telomeric proteins with polyT ssDNA [47] similarly to that using the sequence-dependent model , yet with lower probabilities ( S2 Fig ) . Three representative conformations of each of the studied complexes that correspond to densely populated ( i . e . , low total energy ) regions are shown in Figs 4 and S3 for the six systems . We note that in all cases , the ssDNA and ssRNA conformations possessing minimum binding energies bind at or very close to the actual binding site . A more heterogeneous conformational space is illustrated when projecting the sampled structures along ( DConf1 , DConf2 ) , which measures not only the conformation of the DNA at the binding site but also its directionality ( i . e . , 5’ to 3’ , see Methods ) . These maps show that near-native conformations with low values of DConf1 and DConf2 are reasonably populated . Decomposition the contribution of the ssDNA/ssRNA backbone and bases to the accuracy of the predicted near native conformations ( region 1 ) , reveal that the accuracy of the backbone conformation is slightly higher by about 2Å than the predicted conformations of the bases ( S4 Fig ) . A few additional regions , however , corresponding to non-native conformations of the DNA , are also found to be populated , and some of them possess low binding energy . Their representative conformations in Fig 4 suggest that , although they bind to the actual binding site with a similar alignment but a different orientation to that of the experimental structure ( the 5’ and 3’ ends are flipped ) , their binding energy approaches the minimum . Overall , similar trends were found for the six remaining systems ( see Supporting Information ) . To examine the shape of the binding energy landscape for the interaction of proteins with single stranded nucleic acids , we plotted the potential energy of binding , Ebind ( i . e . , EssDNA/ssRNA-Prot ) , for the simulated systems along DSite or DConf ( Fig 5 ) . For all 12 systems , the distribution of Dsite follows a funneled energy landscape in which near-native structures correspond to a lower binding energy . When the direction of the DNA is not considered in the structural measure , the distribution shows a more funneled shape , with the near-native structures at the minimum energy positions for all proteins except for the two sequence-specific telomere proteins Pot1pc ( 4HIO , Fig 5 ) and Cdc13 ( 1S40 , S5 Fig ) , which have a rugged bottom in their binding free energy surface . Indeed , Pot1pc can accommodate ssDNAs with variable sequences by adjusting the side chains of its interface residues[78] , whereas Cdc13 shows variable specificity at the two terminals of the ssDNA[42] . Similarly , for the ssRNA–ssRBP complexes , plotting the binding energy along DSite reveals global funneled energy landscapes . However , when the order parameter is described by DConf , a more rugged landscape is observed for some systems ( e . g . , Pot1pc and nuclear polyadenylation ) . This suggests that the detailed conformations of ssDNA and of ssRNA at more rugged binding sites can vary , and their energies can compete with that of the native conformation of the single-stranded nucleic acids . Various factors may affect the specificity of the recognition between proteins and nucleic acids . Major determinants for specificity are the conformational ensembles of the two molecules in solution and the network of interactions ( e . g . , aromatic , charged–charged interactions , and hydrogen bonds ) at the interfaces between the proteins and the nucleic acids [14 , 40 , 79] . In our model , sequence specificity is expected to be governed by aromatic–base interactions rather than by electrostatic interactions . We note that , in eleven of the twelve systems studied here , >30% of the total aromatic side chains are located at the binding interface . The only exception is the KH domain , which uses solely electrostatic interaction . We postulate that stacking interactions between specific ssDNA or ssRNA bases and aromatic side chains play a major role in sequence-specific binding . We examined the degree of specificity by investigating the effect that shuffling of the nucleic acid sequences had on the binding energy landscape with the corresponding protein . For this , we chose two telomeric proteins that are expected to interact specifically with ssDNA . We note that some ssDBPs interact similarly with homopolymeric ssDNA ( e . g . , polyT ) and thus are not sensitive to ssDNA sequence . For each telomeric ssDNA sequence , a few other sequences were designed by shuffling the nucleotides while keeping their content fixed and then examining whether the binding pattern changed in the shuffled sequences . The energy landscape for the shuffled sequences ( depicted by plotting Ebind ( i . e . , EssDNA/ssRNA-Prot ) as a function of DConf; Fig 6 ) shows that the systems are sensitive to the ssDNA sequences . The overall shape of the energy landscape , as well as its high-density regions , change with altered sequences . Fig 6 shows the binding pattern for three sequences: the wild-type ( left ) , a shuffled sequence showing inferior binding compared with the wild-type ( middle ) , and another shuffled sequence with better binding ( right ) ( binding energies of additional ssDNA sequences are shown in S6 Fig ) . We note that , for the two telomeric ssDNA binding systems , Pot1pc ( 4HIO ) and Cdc13 ( 1S40 ) , the wild-type sequence tends to show better binding behavior compared with most of the shuffled sequences; the minimum energy structure of the wild-type sequence corresponds to the near-native structure . However , in both the cases , there are ssDNA sequences that show better binding behavior in terms of their similarity with the native structure as well as binding energy . The calculated binding energy for shuffled sequences demonstrate that the specific positions of ssDNA bases with respect to the aromatic residues ( e . g . , interactions between Trp and T or between Phe/Tyr with C , see Fig 1A ) dictates the binding specificity for heterogeneous sequences for Pot1pc . The effect of sequence shuffling is weaker for Cdc13 that lacks any Trp at the interface . These observations suggest that base-mediated stacking interactions are critical for DNA specificity and that modeling enables a reliable prediction of the binding sequence to some extent . However , other factors , such as the rigidity/plasticity of the protein interface and the flexibility of the ssDNA and/or the protein may also play a role in sequence-specific binding . As such , sequence-specific protein–ssDNA interactions are achieved through a subtle balance of intermolecular interactions and dynamics . To examine the role played by the electrostatic and aromatic interactions in the stability of the binding interface , we analyzed the energetics of the interfaces of the 12 studied ssDNA–ssDBP and ssRNA–ssRBP complexes . These structures bind their ssDNA/ssRNA ligands in three different ways that can be found in the following representative systems . i ) The Cold shock protein from Bacillus subtilis ( i . e . , Bs-Csp; an OB fold ) , in which the ssDNA binding is largely mediated by base–aromatic interactions and the ssDNA backbone remains solvent exposed . ii ) The human Poly ( rC ) -binding protein 1 ( a KH fold ) , in which the ssDNA binds solely by electrostatic interactions using its phosphate backbone with no known instances of intermolecular aromatic interactions . iii ) Telomere proteins ( an OB-fold ) , which are known for sequence-specific DNA binding and the human hn-RNP A1 ( RRM fold ) , where both electrostatic and aromatic energies are utilized to achieve specific binding . The contribution of the total electrostatic and aromatic energies is estimated by their ratio λ [= ( total electrostatic energy ) / ( total aromatic energy ) ] calculated for the near-native structures ( see Table 1 ) . A very low λ ( <<1 ) for the B protein of Bs-Csp ( Bs-CspB ) indicates the importance of aromatic interactions for this protein , with this also clear from the predicted structures in Fig 4 , where all ssDNA bases face toward the protein . By contrast , a very high value of λ ( >>1 ) is obtained for the KH domain , indicating the importance of its electrostatic interactions; again , the representative structures in Fig 4 demonstrate that most of the DNA bases face away from the protein . The experimental structures of three other OB-folds , as well as the RRM domain , reveal that the ssDNA is oriented such that most of the bases face toward the protein surface to participate in stacking interactions , whereas electrostatic interactions with the phosphate backbone make a smaller contribution . In the coarse-grained model , the contribution of the aromatic energy to the stability of the interface between ssDNA and the telomeric proteins was 2–4 times higher than the contribution of the electrostatic energy ( i . e . , 0 . 2>λ>0 . 4 , see Table 1 and S7 Fig ) . Depending on the function of the ssDBP , different ssDBPs utilize different proportions of interactions in order to bind sequence-specifically or indiscriminately to their ssDNA ligands . Some of them interact with ssDNA largely by contacting the bases , whereas others minimize sequence specificity by controlling base stacking and base-specific H-bond formations , both of which might confer specificity . Most of the ssRNA–ssRBP interfaces also reflect the importance of the aromatic interactions for their stability , as illustrated by the λ values being lower than 1 . In these cases , the electrostatic interactions between the phosphate backbone and positively charged residues make a modest contribution to binding affinities . Similarly to the interaction of the cold shock protein with ssDNA , its interaction with ssRNA is also characterized by a very low λ value , showing that it is mediated by stacking interactions only; the corresponding predicted structure in Fig 4 also shows all RNA bases facing towards the protein . For the Fab structure , the RNA is recognized mostly by base–aromatic interactions mediated by a number of Tyr residues from the CDR region , and the estimated value of λ for this structure is also low . By contrast , λ>1 in the case of RAMP protein indicates the importance of the electrostatic contribution for the corresponding RNA binding . Indeed , here the ssRNA binds to the positively charged groove on the protein surface and electrostatic interaction plays a major role in binding . In addition to the structural evaluation of the simulated binding of DBPs and RBPs with ssDNA or ssRNA , respectively , we were motivated to quantify the energetics of the predicted complexes . The binding energies , Ebind ( i . e . , EssDNA/ssRNA-Prot ) of the simulated complexes were compared with the experimentally measured equilibrium dissociation constants ( KD ) . Fig 7A shows a comparison between Ebind and ln ( KD ) for different oligonucleotide sequences that bind six ssDBP and three ssRBPs . For Pot1pc , we calculated Ebind for seven different ssDNA sequences for which structures are available[78] . Overall , KD values are in good agreement with Ebind ( r = 0 . 66 ) indicating that the model captures the energetics of interaction between various proteins and ssDNA as well as ssRNA . In each case , Ebind was calculated by considering only near-native conformations ( DConf1 and DConf2 ≤5 Å ) . Table 3 shows the KD and Ebind values for each system . To compare Ebind with KD for a particular protein that binds different ssDNA or ssRNA sequences , we analyzed the binding of Bs-CspB with various sequences of ssDNA and ssRNA . Two of its crystal structures were solved , one in complex with hexa-Thymine ( dT6 ) ( 2ES2 . pdb ) that binds with nM affinity , the other one with hexa-Uracil ( dU6 ) ( 3PF5 . pdb ) , whose binding is weaker than that of dT6 but nevertheless in the nM range . The Bs-CspB binding site can interact with six to seven nucleotides[37] . The nucleic acid strands bind at the same binding site in the two structures , but their conformation differs at the 3’ end . Further investigations were also made on the binding affinities of Bs-CspB to different hepta-nucleotide ssDNA and ssRNA sequences that bind with a 1:1 stoichiometry [38] . In the crystal structure , several aromatic and hydrophobic solvent-exposed residues surrounded by basic side-chains form an amphiphilic surface that associates with the ligand . On the opposite surface , the protein comprises several acidic residues that impart a negative potential to the surface , making it unfit to bind either ssDNA or ssRNA . Moreover , the 0 . 88 Å Cα RMSD of this structure from free Bs-CspB ( 1CSP . pdb ) shows a marginal conformational change of the protein due to ligand binding . Combining these observations , it is clear that Bs-CspB contains only a single binding site to which all the ssDNAs with variable sequences bind . Hence , in our analysis , we considered the bound conformation to be the same as that of Bs-CspB-dT6 for all ssDNA sequences . Starting from dT7 , T was progressively replaced by C to investigate their preferences at each position , as was tested experimentally . The binding constant KD of the resulting sequences varied in the μM to nM range , showing a preference for poly-Thymine over poly-Cytosine . Likewise , for all nine ssRNA sequences , the bound conformations were considered to be the same as in the Bs-CspB . dU6 complex . The binding constant of ssRNAs also varied in the μM to nM range , however the values were lower than for ssDNAs . The Ebind versus KD plot for 13 ssDNA ( solid red circles ) and nine ssRNA ( solid circles , blue ) that bind to Bs-CspB is shown in Fig 7B . Overall , they are in good agreement with a linear fit ( R = 0 . 76 , considering solid circles only ) . Our model captures the overall higher affinity of Bs-CspB for ssDNA compared with ssRNA . Similarly to the experimental data , the binding energies ( Ebind ) of the polythymine and polycytosine sequences in the coarse-grained model indicate that the former is more stable . However , the Ebind was less sensitive in predicting the effect on KD of a single mutation at different positions . This is expected , as achieving such accuracy is beyond the scope of any coarse-grained model . Nonetheless , results from our simulations agreed well with the experimental binding affinities when nucleotide content was taken into account , and thus such simulations can be used in binding specificity predictions . To understand the origin of the higher affinities of Bs-CspB for ssDNA than for ssRNA , we used the simulated binding events to estimate the association and dissociation rates for the interactions of the GTCTTTA ssDNA sequence and GUCUUUA ssRNA sequence with the cold shock protein , for which experimental kinetic results are available[38] . Computationally , the rate constant for association ( kon ) was estimated by the elapsed time for binding ( defined by Dconf <5Å ) when starting from an unbound state , and similarly the rate constant for dissociation ( koff ) was estimated by the elapsed time for dissociation ( defined by Dconf>5Å ) when starting from the bound complex . The association constant ratio kon ( ssDNA ) /kon ( ssRNA ) from the coarse-grained simulations is ~1 , in very good agreement with the experimental data . The dissociation constant ratio koff ( ssDNA ) /koff ( ssRNA ) from the simulations is ~0 . 2 . The value of this ratio based on the experimental results is 0 . 1[38] , yet both the simulations and the experimental data agree that the ratio is lower than unity . The higher dissociation rate for ssRNA compared with ssDNA is the main reason for the higher KD of Bs-CspB–ssRNA compared with Bs-CspB–ssDNA . The energy contribution from electrostatic and aromatic interactions plays a significant role in ssDNA/ssRNA binding with proteins . Nevertheless , it is not only the charged or aromatic side-chains that interact with the nucleic acid backbone or bases , respectively , to govern the protein–ssDNA/ssRNA assembly . For example , some charged residues that do not interact directly with DNA or RNA can still have a strong electrostatic effect on binding [80 , 81] . Unbound ssDNAs/ssRNAs are highly flexible in solution , without any definite shape . Prior to binding , they fluctuate in an ensemble whose length and shape match the size of the binding pocket . Their conformational flexibility usually leads to an induced fit of the ssDNA/ssRNA to the protein surface . Complexes between ssDNA/ssRNA and protein are therefore difficult to predict unless their backbone flexibility is properly modeled . In our model , we focused on incorporating the conformational flexibility of the ssDNA and ssRNA . The flexibility of ssDNA or ssRNA is often judged by their persistence length , where a lower persistence length value corresponds to greater flexibility . The persistence length of both ssDNA and ssRNA decreases with increasing salt concentration[12] . However , when their persistence lengths are compared , ssDNA was found to have lower averages compared with ssRNA , which indicates that , in solution , ssDNAs are more flexible than ssRNAs . In our coarse-grained model , we mimicked the effect of salt concentration by means of dihedral potentials ( see Methods ) , where the persistence length of ssDNA/ssRNA in solution increases with increasing dihedral potentials and decreases with increasing salt concentration[12] . To be consistent with the experimental finding , we set the dihedral potentials in the model such that ssDNA possess a lower persistence length ( greater flexibility ) than ssRNA . Further to compare the role of flexibility for ssDNA and ssRNA in their differential binding strengths , we used the Bs-CspB model system for which binding data for a number of ssDNA as well as ssRNA molecules are available . The dihedral parameters of ssDNAs and ssRNAs were interchanged so that ssRNAs became more flexible than ssDNAs . All other parameters including base–aromatic stacking strengths were unaltered . The resulting ssRNAs ( Fig 7B , empty red circles ) were found to bind Bs-CspB more tightly than ssDNAs do ( Fig 7B , empty blue circles ) . The correlation between the binding energy of the modified ssDNA and ssRNA , in which their degree of flexibility was switched , and the experimental KD values is much weaker ( Fig 7B ) . This observation indicates the major role that flexibility plays in their binding . It can further explain why ssDNAs–protein interactions can be stronger than ssRNA–protein interactions . Often , biomolecular affinity and specificity are linked and they can also be related to the degree of flexibility of the ligand[82–85] . Although , conventionally , high affinity is linked with high specificity , there are examples of flexibility resulting in reduced affinity while high specificity is retained . The interactions of ssDNA and ssRNA with their protein receptors are shown to differ with respect to their affinity ( Fig 7B ) . This , together with their different conformational flexibilities , may suggest that they may have different degrees of specificity [82–85] . Specificity is often defined as the binding affinity to one ligand relative to other ligands . Alternatively , one may define the intrinsic specificity , which is the binding affinity of a ligand to a receptor relative to the binding affinity of the ligand to other sites on the same receptor[86] . To quantify the decoupling between the affinity and specificity of ssDNA/ssRNA binding to proteins , and the link to their different intrinsic flexibilities , we analyzed the energy landscape for binding using the theory of energy landscape [87–90] . According to this theory , the native conformation of the binding complex is the conformation with the lowest binding energy and the energies of the non-native conformations follow a statistical Gaussian distribution . A dimensionless quantity termed the intrinsic specificity ratio ( ISR ) is defined to describe the magnitude of intrinsic specificity [86 , 91 , 92]: ISR=δE/ ( ΔE2S ) , where δE is the energy gap between the native binding state and the average non-native binding states , ΔE is the energy variance of the non-native states , and S is the configurational entropy . A large ISR value indicates that the protein strongly discriminates the native binding site from the non-native binding sites , which indicates a high binding specificity . The energy landscapes for the association of twelve ssDNA and nine ssRNA sequences with their corresponding protein receptors were analyzed by estimating the values of δE , ΔE , and S . Fig 8 shows that the complexes formed with ssDNA have smaller δE and ΔE values than the complexes with ssRNA . Namely , the native complexes of ssDNA–ssDBP are more distinguished energetically than the non-native conformation in comparison to the ssRNA–ssRBP complexes . Furthermore , on average , the non-native ssDNA–ssDBP complexes are less diverse than the ssRNA–ssRBP complexes . These two properties and their similar entropy , S , result in higher specificity for the ssDNA complexes than for the ssRNA complexes . In summary , the differences between the ssDNA–ssDBP and ssRNA–ssRBP complexes are due to the greater flexibility of ssDNA compared with ssRNA , which leads to higher affinity ( Fig 8B ) and higher specificity .
Predicting the complexes formed between proteins and either ssDNA or ssRNA is difficult because of their complex underlying energy landscapes , which originate mostly from the considerable flexibility of the ssDNA or ssRNA and their consequent lack of a defined structure . Effective factors that can be tuned to affect the interaction between single-stranded nucleic acids and receptor proteins include the extent of ligand flexibility and also the salt concentration , which may modulate the strength of the electrostatic interactions . The lack of an ordered structure in ssDNAs or ssRNAs allows them to interact with proteins not only through electrostatic interactions between their backbone phosphate groups and positively charged residues but also by stacking interactions between free nucleotide bases and aromatic side chains . These factors increase the degree of complexity and heterogeneity of these interfaces and thus computational modeling of specific ssDNA–ssDBP and ssRNA–ssRBP interactions becomes even more challenging compared with other specific macromolecular interactions . As a result , unlike protein–protein or protein–dsDNA interactions , the theoretical study of ssDNA/ssRNA–protein binding specificity from the structural and energetic points of view is not sufficiently advanced . In this study , we applied a physically based coarse-grained approach to construct a generalized model to study the recognition of ssDNA/ssRNA by ssDBPs and ssRBPs , respectively . A number of experimental studies showed that there are no obvious structural indicators for sequence-specific proteins [37 , 40 , 78] . Instead of strictly binding or not binding to particular sequences , a protein can bind different sequences with a range of affinities . From the perspective of structural properties , specific binding can be attributed to specific base–aromatic interactions and to the ssDNA/ssRNA dynamics . We incorporated binding specificity into the model by adding different base–aromatic stacking strengths as well as by adjusting the flexibility of the single-stranded nucleic acid . Accordingly , the model has only two free parameters . The developed transferable coarse-grained model was successfully applied to 12 complexes between ssDNAs or ssRNAs and binding proteins . The results demonstrated that single stranded nucleotide–protein recognition follows the binding energy model in which the predicted near-native structures correspond to minimum binding energies . The predicted complexes differ in the relative energetic contributions made to them by aromatic and electrostatic interactions . Few interfaces are governed solely by either electrostatic or aromatic interactions , rather , the majority of the interfaces are stabilized by both electrostatic and aromatic interactions , with the latter being more dominant . The model is sensitive to sequence-specific binding and the estimated interfacial binding energies of near-native conformations show good correlation with experimental dissociation constants . Our results suggest that the origin for the weaker stability of the complexes formed between proteins and ssRNA compared with ssDNA is the lower flexibility of ssRNA . The lower affinities of ssRNA–ssRBP compared with ssDNA–ssDBP are coupled with larger dissociation rate constants ( koff ) while their association rate constants ( kon ) are of similar values . The complexes of ssRNA are also found to be less specific than those of ssDNA , which might be linked to their greater stiffness . While the power of the developed coarse-grained model lies in its simplicity , which allows extensive sampling of several systems and thus enables the study of long timescale dynamic motions , it can be further advanced to address other molecular biophysical aspects of protein–ssDNA/ssRNA dynamics . For example , incorporating specific and explicit ion interactions with ssDNA and ssRNA and their interactions with the solvent may improve the accuracy of the predicted structures . Sequence specificity may also depend on base-specific hydrogen bonding networks that are formed between the single stranded nucleic acids and the proteins , implementation of which would enhance the efficiency of the model for specific recognition . Furthermore , the model deals with unstructured ssDNA and ssRNA and it may demand additional energetic terms to represent formations of more compact structures of ssDNA mediated by base-pairing and , in particular , the formation of secondary structures in ssRNAs . Nonetheless , the present model produces useful results for specific ssDNA–ssDBPs interactions , and thus this type of coarse-grained model can be further used to study other properties of these interactions ( e . g . , the sliding mechanism of ssDNA on ssDBPs; [93–95] ) , to complement experimental studies , and especially to elucidate how the molecular properties of the interfaces are linked to their function and dynamics . | Quantifying bimolecular self-assembly is pivotal to understanding cellular function . In recent years , a large progress has been made in understanding the structure and biophysics of protein-protein interactions . Particularly , various computational tools are available for predicting these structures and to estimate their stability and the driving forces of their formation . The understating of the interactions between proteins and nucleic acids , however , is still limited , presumably due to the involvement of non-specific interactions as well as the high conformational plasticity that may demand an induced-fit mechanism . In particular , the interactions between proteins and single-stranded nucleic acids ( i . e . , single-stranded DNA and RNA ) is very challenging due to their high flexibility . Furthermore , the interface between proteins and single-stranded nucleic acids is often chemically more heterogeneous than the interface between proteins and double-stranded DNA . In this study , we developed a coarse-grained computational model to predict the structure of complexes between proteins and single-stranded nucleic acids . The model was applied to estimate binding affinities and the estimated binding energies agreed well with the corresponding experimental binding affinities . The kinetics of association as well as the specificity of the complexes between proteins and ssDNA are different than those with ssRNA , mostly due to differences in their conformational flexibility . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [] | 2019 | Structure, stability and specificity of the binding of ssDNA and ssRNA with proteins |
Mucosal-Associated Invariant T ( MAIT ) cells , present in high frequency in airway and other mucosal tissues , have Th1 effector capacity positioning them to play a critical role in the early immune response to intracellular pathogens , including Mycobacterium tuberculosis ( Mtb ) . MR1 is a highly conserved Class I-like molecule that presents vitamin B metabolites to MAIT cells . The mechanisms for loading these ubiquitous small molecules are likely to be tightly regulated to prevent inappropriate MAIT cell activation . To define the intracellular localization of MR1 , we analyzed the distribution of an MR1-GFP fusion protein in antigen presenting cells . We found that MR1 localized to endosomes and was translocated to the cell surface upon addition of 6-formyl pterin ( 6-FP ) . To understand the mechanisms by which MR1 antigens are presented , we used a lentiviral shRNA screen to identify trafficking molecules that are required for the presentation of Mtb antigen to HLA-diverse T cells . We identified Stx18 , VAMP4 , and Rab6 as trafficking molecules regulating MR1-dependent MAIT cell recognition of Mtb-infected cells . Stx18 but not VAMP4 or Rab6 knockdown also resulted in decreased 6-FP-dependent surface translocation of MR1 suggesting distinct pathways for loading of exogenous ligands and intracellular mycobacterially-derived ligands . We postulate that endosome-mediated trafficking of MR1 allows for selective sampling of the intracellular environment .
Mucosal-Associated Invariant T ( MAIT ) cells are a class of CD8+ T cells that are unique in their use of a semi-invariant TCR , restriction by the highly conserved major histocompatibility complex , class I-related protein 1 ( MR1 ) , and their recognition of small molecule metabolites . In support of a role for these cells in host-defense to mucosal infection , human MAIT cells are present in high numbers in mucosal tissues and blood [1 , 2] and secrete pro-inflammatory molecules including TNFα and IFN-γ [3 , 4] . MAIT cells have the capacity to respond to intracellular pathogens such as Mycobacterium tuberculosis ( Mtb ) [3] and Francisella tularensis [5] and animal models demonstrate a requirement for MR1 , and by inference , MAIT cells in early control of certain pathogens [5–7] . Although MR1 is ubiquitously expressed in all mammalian tissues examined , surface expression is very low or undetectable in both phagocytic professional antigen presenting cells and non-hematopoetic cells [8 , 9] . Previous work demonstrated that MR1 surface expression in mouse cells overexpressing MR1 requires both the MHC-II chaperone , invariant chain ( Ii ) , and trafficking through late endosomal compartments [10] . This is in contrast to the constitutive surface expression of other Class I molecules , and may be key to understanding the regulation of MR1-restricted MAIT cell activation . The McCluskey and Rossjohn groups have identified Vitamin B metabolites as ligands that bind and stabilize MR1 [11 , 12] . With regard to MAIT cell activation , pterins , which are derived from folic acid ( 6-formylpterin ( 6-FP and ac-6-FP ) ) , are antagonistic , while the bacterially-derived riboflavin metabolites known as lumazines ( RL-6 , 7-diMe , RL-6-Me-7-OH , and rRL-6-CH2OH ) are agonists [11] . More recently an additional class of ligands , pyrimidines ( 5-OE-RU , 5-OP-RU ) , were identified as potent activators of MAIT cells . These pyrimidines are generated in a chemical reaction between bacterially-derived riboflavin precursor molecules and small metabolites derived from either the host or pathogen [13] . Currently , little is known about the intracellular localization of MR1 , how and where MR1 is loaded with these ligands , or the mechanism for MR1 translocation to the cell surface . The ubiquitous expression of MR1 in many tissues , the high frequency of MAIT cells in both the blood and mucosal tissues , and the prevalence of potentially activating ligands likely requires that MR1 loading and trafficking be tightly regulated to prevent indiscriminant MAIT cell activation . Although Ii was proposed as a required chaperone for MR1 [10] , we have demonstrated that airway epithelial cells ( AEC ) are highly efficient at presenting antigen to MAIT cells [14] even though they do not express MHC class II or Ii . The ability of AEC to efficiently present mycobacterial antigens and their proximity to MAIT cells in the lung suggests they may play an important role in early recognition of exposure to Mtb or other pathogens . In this study , we sought to define intracellular trafficking pathways necessary for the presentation of Mtb-derived MR1 ligands by AEC to human MAIT cells .
To determine the basal intracellular localization of MR1 in AEC , we expressed MR1 fused to green fluorescent protein ( GFP ) in the BEAS-2B airway epithelial cell line and primary normal human bronchial epithelial cells ( NHBE ) . To establish that the MR1-GFP protein was functional , cells expressing MR1-GFP were infected with Mtb or Mycobacterium smegmatis and used as antigen presenting cells ( APC ) in an IFNγ ELISPOT assay . The response of MR1-restricted T cells was increased with MR1-GFP over-expression with no impact on IFNγ production by HLA-E or HLA-B45 restricted T cells ( Fig 1A ) . We next sought to define the subcellular localization of MR1-GFP using high-resolution deconvolution microscopy . As expected , in both BEAS-2B and NHBE cells , the majority of MR1 observed was intracellular , with expression in an endoplasmic reticulum ( ER ) -like subcellular compartment as well as endosomal compartments ( EC ) ( Fig 1B , S1 Movie ) . This was in contrast to MHC-I , which was largely localized to the plasma membrane ( Fig 1B ) . Staining with antibodies against the ER-Golgi intermediate compartment ( ERGIC-53 ) the trans-Golgi network ( TGN-46 ) demonstrated that a proportion of the MR1-GFP localized to these compartments ( Fig 1C ) , while subsets of the MR1-GFP+ EC co-localized with markers that define late endosomes or lysosomes ( Rab7 , Lamp1 ) but not early endosomes ( Rab5 ) ( Fig 1D , S2 Movie ) . The majority of the MR1-GFP+ EC also co-localized with β-2-microglobulin ( β-2M ) ( Fig 1D ) . These MR1+ β-2M+ EC have a similar phenotype to the Mtb compartment in AEC [14] and may represent a pre-synthesized pool of MR1 that could be loaded in the context of intracellular infection . Recent reports indicate that incubation of cells with acetyl-6-FP results in cell surface stabilization of MR1 [12] . To define the impact of exogenously added ligand on the intracellular localization of MR1 in BEAS-2B and primary NHBE cells , MR1-GFP expressing cells were incubated with 6-FP . As expected , incubation with 6-FP resulted in translocation of MR1-GFP to the cell surface in both BEAS-2B and NHBE cells , detected by surface staining with the α-MR1 26 . 5 antibody ( red ) ( Fig 2A ) , which recognizes folded MR1 [13 , 15] . Flow cytometry confirmed 6-FP dependent cell-surface translocation in cells expressing MR1-GFP and native MR1 ( Fig 2B ) . Inhibiting ER-Golgi transport with brefeldin A ( BFA ) dramatically decreased the 6-FP-dependent cell surface stabilization of MR1 ( Fig 2B ) , indicating that the dominant processing and presentation pathway for 6-FP requires ER-Golgi transport . The 6-FP-dependent translocation of MR1 to the cell surface was accompanied by a decrease in the number of MR1-GFP+ EC in BEAS-2B and primary NHBE cells ( Fig 2C ) and MR1-GFP+ EC were further depleted by cycloheximide ( CHX ) treatment ( Fig 2C ) . When we analyzed the co-localization of β-2M with the MR1-GFP+ EC remaining following treatment with 6-FP , we found that the percentage of MR1-GFP+ β-2M+ EC decreased ( Fig 2D ) . Together these data indicate that the MRI+ β-2M+ EC are a source of intracellular MR1 molecules available for ligand-binding and translocation to the cell surface . The identification of vitamin B metabolites as MR1 ligands [11] , the observation that processing and presentation is both TAP and proteasome independent [3 , 10] , and apparent requirement for ligand in the cell surface stabilization of intracellular MR1 suggest a unique mechanism for MR1 antigen processing and presentation . To define this mechanism in the context of intracellular infection with a pathogen , we designed a screen to identify trafficking proteins that are essential for MR1-dependent MAIT cell activation in the setting of intracellular infection with Mtb ( S1 Fig ) . Here , shRNA knockdown of trafficking molecules in BEAS-2B cells expressing native MR1 was performed , and T-cell dependent release of IFN-γ was measured following Mtb infection . To identify MR1-specific mechanisms , human Mtb-reactive CD8+ T cell clones of known specificity ( MR1 ( D426 B1 ) , HLA-E ( D160 1–23 ) , and HLA-B45 ( D466 D6 ) ) were tested in parallel with the same Mtb-infected APC . Candidate proteins were identified by the selection criteria described in the Materials and Methods . Briefly , to reduce the impact of non-specific shRNA targeting , genes where at least two independent shRNAs resulted in reduced T cell recognition were considered candidates for subsequent siRNA validation . Based on these criteria , we identified candidate genes that fell into three categories with regard to recognition by MAIT cells: 1 ) Genes where silencing did not result in reduced recognition by any of the T cell clones; 2 ) Genes where silencing resulted in reduced recognition by all of the T cell clones; and 3 ) Genes where silencing resulted in reduced recognition specifically by MAIT cells . Examples are shown in Fig 3A , where silencing of Rab2b with multiple shRNAs in BEAS-2B cells did not inhibit Mtb-dependent IFN-γ release by any of the T cell clones tested , while silencing of Rab7L1 by multiple shRNAs inhibited responses by all T cell clones tested , including MAIT cells . In contrast , silencing of Stx18 with shRNA selectively inhibited only the response of the MAIT cell clone , D426B1 , according to the selection criteria ( Fig 3A ) . In total , we identified 27 candidate genes where shRNA knockdown specifically resulted in reduced MAIT cell recognition of Mtb-infected BEAS-2B cells . To date , we have evaluated ten of the 27 MAIT cell-specific candidates using siRNA knockdown . Gene knockdown was confirmed using RT-PCR , and T cell-dependent release of IFN-γ was responses were evaluated by ELISPOT assay . Silencing of eight of the ten candidates tested by siRNA resulted in reduced response by the MAIT cell clone , seven of which were confirmed to specifically impact only the MAIT cell clone . All candidates evaluated are listed in Table 1 and representative results are illustrated in Fig 3B . Here , Mtb-infected APC silenced with Stx18 siRNA had a substantially reduced ability to activate the MR1-restricted clone , with no effect on activation of the HLA-E and HLA-B45-restricted clones , validating the results of the shRNA screen . Furthermore , infection and intracellular viability of Mtb were not affected by Stx18 silencing as demonstrated by quantitative culture of intracellular Mtb ( Fig 3B ) . Similarly , Stx18 silencing in Mtb-infected NHBE cells resulted in diminished MR1-dependent recognition of infection with Mtb ( Fig 3C ) . Together , these data suggest that Stx18 is involved in a mechanism that selectively impacts MR1-dependent antigen processing and/or presentation of Mtb antigen by AEC . Because MR1 was observed in intracellular compartments in the endosomal pathway , we also focused on two endosome-associated trafficking proteins , the R-SNARE vesicle-associated membrane protein 4 ( VAMP4 ) and Rab6 , a small GTPase protein . Similar to what was observed for Stx18 , following VAMP4 or Rab6 silencing with siRNA , there was diminished ability of Mtb-infected APC to activate MR1-restricted T cells ( Fig 3D ) . Stx18 is an ER-localized soluble NSF attachment protein receptor ( SNARE ) protein of the Qa-SNARE family best known as a member of a SNARE complex regulating Golgi-ER retrograde transport consisting of Stx18 , BNIP1 , Use1 ( also called p31 , D12 , or MDS032 ) , and Sec22b [16] . Because Sec22b was also validated from the shRNA screen as an MR1-specific candidate , we analyzed the role of this SNARE complex in Mtb antigen presentation to MAIT cells . SNAREs known to partner with Stx18 were silenced by siRNA and tested for their ability to activate MAIT cells . As expected , knockdown of Sec22b resulted in a similar reduction in MAIT cell activation as was observed for Stx18 . Knockdown of BNIP1 also resulted in a reduction in MAIT cell activation , however knockdown of Use1 did not have any impact ( Fig 3E ) . The SNAREs in this complex can also play a role in ER-mediated phagocytosis through interactions with plasma membrane SNAREs [17] as well as post-Golgi trafficking pathways [18] . In fact , Stx18 has unusual properties for a Qa-SNARE , as it has a SNARE motif that is longer than conventional SNAREs [19] and undergoes a different mechanism of SNARE complex assembly compared to other ER SNAREs [20] that may allow it to function independently of this complex . These data suggest MR1 processing and presentation may require a Stx18-mediated trafficking pathway distinct from the previously characterized Golgi-ER retrograde transport pathway . To understand the mechanism by which these trafficking molecules regulate processing and presentation of Mtb antigens on MR1 , we looked for alteration in MR1 localization following siRNA treatment . In HeLa cells , knockdown of Stx18 has been reported to affect the ER , ER-Golgi intermediate compartment ( ERGIC ) , and Golgi including disruption of the Golgi stacks and ER exit sites and disorganization of ER membranes [21] . Knockdown of VAMP4 in HeLa cells also results in dispersion of the Golgi [22] whereas knockdown of Rab6 does not disperse the Golgi structure , but makes it more compact [23] . To confirm these observations , Stx18 , VAMP4 , and Rab6 were silenced in BEAS-2B cells . Knockdown of Stx18 and VAMP4 resulted in the dispersion of the Golgi , while the structure of the Golgi in Rab6 silenced cells was more compact ( Figs 4A and S2 ) . In the context of MR1-GFP expression , knockdown of Stx18 , VAMP4 , and Rab6 resulted in an increase in the number of MR1-GFP+ EC ( Fig 4B ) and in the amount of MR1 in those compartments , as measured by the mean intensity of GFP per EC ( Fig 4B ) . The functional defect observed by ELISPOT and the increase in MR1-GFP+ EC led us to hypothesize that silencing of these genes could result in diminished cell surface expression of MR1 in the presence of a ligand . Knockdown of Stx18 , VAMP4 , and Rab6 did not alter basal cell surface expression of MR1 in cells expressing endogenous MR1 or over-expressing MR1-GFP ( Fig 5A ) . However , Stx18 was required for ligand-dependent translocation of MR1 to the cell surface as Stx18 knockdown resulted in reduced translocation of MR1 to the cell surface after incubation with 6-FP ( Fig 5A ) . Knockdown of VAMP4 or Rab6 had no impact on translocation of MR1 to the surface with 6-FP incubation ( Fig 5A ) . Stx18 , VAMP4 , and Rab6 silencing all resulted in similar reduction in MR1-restricted T cell response to Mtb-infected cells . The discordant results with 6-FP dependent MR1 surface stabilization in Stx18- and VAMP4- or Rab6-silenced cells support a model in which the processing and presentation pathways for exogenously added 6-FP and intracellular bacteria are distinct . Consistent with this model , we observed colocalization of Stx18 with MR1 in the ER , while VAMP4 colocalized with MR1 in EC ( Fig 5B ) . To further distinguish exogenously delivered versus intracellular antigen , we analyzed the impact of pre-treatment with 6-FP , a MAIT cell antagonist , on the ability of MAIT cells to respond to exogenous or endogenous antigen in the context of native MR1 . Incubation of APC with M . smegmatis supernatant resulted in MAIT cell activation , which could be abrogated by pre-incubation of the APCs with 6-FP ( Fig 5C ) . In contrast , in cells infected with Mtb , pre-incubation with 6-FP did not significantly reduce MAIT cell activation ( Fig 5C ) . These data suggest that for processing and presentation of intracellular Mtb-derived antigen , there is an endosomal pool of MR1 that is distinct from the pool preferentially loaded with 6-FP .
MAIT cells are a unique class of MR1-restricted T cells capable of recognizing small molecule vitamin B metabolites . There is a growing appreciation for the importance of these cells in the mucosal response to intracellular microbial infection . At present , little is known about the mechanisms by which this novel class of antigens are processed and presented . As we have demonstrated previously , MAIT cells recognize Mtb-infected dendritic and epithelial cells and are present in high numbers in lung tissues [3 , 14 , 24] . We postulate that the circumstances by which MAIT cells become activated is tightly regulated to allow for the recognition of intracellular infection while avoiding tissue damage . Here we have demonstrated that MR1 is primarily localized to an endosomal compartment , and have identified trafficking molecules responsible for presentation of Mtb antigens on MR1 . These results provide a framework with which to understand the mechanisms regulating activation of MAIT cells . While we and others have shown that exogenous ligands such as the short-lived ribityllumazine metabolites can be presented in the context of MR1 in vitro , the extent to which exogenous ligands contribute to MAIT cell activation in vivo is unknown . Our data support the hypothesis that regulation of MAIT cell activation is critically dependent on intracellular infection . Cell-surface localization of classical Class I molecules is dependent on binding by a peptide that can be either endogenous or exogenous . This is in contrast to MR1 , which we demonstrate is primarily localized to endosomal compartments with features of late endosomes and lysosomes , consistent with previous findings in mouse embryonic fibroblasts [10] . The difference in intracellular localization between classical Class I molecules and MR1 suggest that translocation to the cell surface is a key event for MAIT cell recognition of intracellular infection . By maintaining a dynamic endosomal store of MR1 , ligands derived from intracellular pathogens can be rapidly loaded and presented on the cell surface . MR1 that does not encounter intracellular ligand may be rapidly recycled or degraded , keeping access to ligands from normal extracellular gut flora to a minimum . These findings led us to explore the role of endosomal trafficking in presentation of Mtb ligands on MR1 to MAIT cells . Using lentiviral shRNA knockdown in Mtb-infected APC , we identified a set of trafficking molecules specifically involved in the presentation of MR1-ligands to MAIT cells . Our results lead us to postulate a model where , in contrast to classical ER-associated Class I antigen processing and presentation , the expression of MR1 in endosomal compartments is critical to regulation of the processing and presentation of MR1 ligands generated in the context of intracellular infection with Mtb . In this model , we seek to explain 1 ) the proteins involved in generation of MR1+ EC , 2 ) the location where MR1 interacts with ligand , and 3 ) the ability of loaded MR1 to translocate to the cell surface . While we cannot yet say where MR1 is interacting with Mtb ligand based on our data , we can conclude that MR1 is likely interacting with exogenously added ligands such as 6-FP primarily in a pre-Golgi compartment such as the ER . We have identified proteins , Stx18 , VAMP4 , and Rab6 , which play a role in maintaining the structure of the TGN , and ultimately regulate the translocation of MR1 to the cell surface in the context of infection with Mtb ( Fig 6 ) . In the context of our model , 6-FP is loaded through a distinct pathway from those derived in the context of intracellular infection . The ER-associated SNARE , Stx18 , appears to play a role in both of these pathways , whereas the TGN-endosome localized Rab6 and VAMP4 uniquely contribute to the loading of Mtb ligands . These data support a role for the MR1+ EC in loading of Mtb ligands . While we have identified a preliminary framework by which these trafficking proteins participate in presentation of Mtb antigens on MR1 , there are still many unanswered questions about how MR1 samples the intracellular environment . At present , there are three ways in which the immune system is known to sample the endosomal environment , both of which involve exchange of ligand . The first is through MHC Class II antigen presentation , where peptides derived from the endosomal pathway are exchanged with the invariant chain ( Ii ) on MHC Class II molecules in a late endosomal compartment known as the MHC Class II compartment ( MIIC ) ( reviewed in [25] ) . The second is through the MHC Class I endosomal recycling pathway , where MHC Class I molecules are loaded with endosomally-derived peptides through the action of cathepsins and peptide exchange [26] . The third is through CD1 antigen presentation , where lipid antigens can be sampled in early and late endosomes , including the MIIC ( reviewed in [27 , 28] ) . Unlike MHC-II , classical MHC-I , and CD1 molecules , neither self-ligands nor chaperones have been defined for MR1 . Although invariant chain ( Ii ) was proposed as a required chaperone for MR1 [10] , we showed that BEAS-2B cells are highly efficient at presenting antigen to MAIT cells [14] even though they do not express MHC class II or Ii . Based on this , we hypothesize that there are chaperones in addition to Ii that are able to stabilize MR1 and contribute to endosomal loading of antigen . Peptide-loaded classical Class I molecules translocate to the cell surface regardless of whether they contain endogenous or exogenous peptides . CD8+ T cells sample all peptides that are presented , and the ability of T cells to discriminate self from non-self is determined by the specificity of the T cell receptor ( TCR ) . Here , constitutive cell surface localization of classical Class I molecules does not lead to indiscriminant T cell activation as thymic selection serves to remove T cells that are auto-reactive . For MAIT cells , the TCR appears to have broader reactivity from that of classical CD8+ T cells . Instead , recognition by MAIT cells may be dependent on the regulated intracellular loading of ligand that in turn leads to cell surface translocation . In support of the idea that stable expression of MR1 on the cell surface is tightly regulated , we observe that even the presence of intracellular infection with Mtb or incubation with soluble fractions from Mtb , there is little to no increase in the level of surface MR1 despite clear activation of MAIT cells . Similarly , translocation of MR1 to the cell surface in the context of bacterial ligands was not observed with soluble fractions from E . coli [29] , suggesting a common regulatory pathway . Mtb-derived MR1 ligands have not yet been identified , and we are not able to track the known small molecule ligands directly in cells . We show clear evidence , however , that trafficking of MR1 in endosomal compartments is key to Mtb antigen presentation , and hypothesize that the regulation of endosomal trafficking is key to generation of MR1+ EC that can encounter Mtb ligands and subsequently traffic to the cell surface . Further characterization of targets identified in this study will provide insight into the novel regulation of loading of ligand on MR1 and translocation of MR1 to the cell surface in the context of infection . The clear demonstration of endosomal trafficking as essential in the MR1-dependent activation of MAIT cells suggests a different paradigm for how MAIT cell recognition occurs and provides a critical first step in defining the fundamental mechanisms for MR1 antigen processing and presentation .
Mycobacterium tuberculosis H37Rv ( ATCC ) or H37Rv-expressing dsRED , a kind gift from Joel Ernst was grown in Middlebrook 7H9 broth supplemented with Middlebrook ADC ( Fisher ) , 0 . 05% Tween-80 , and 0 . 5% glycerol . Before infection , bacteria were passaged 15 times through a tuberculin syringe to obtain a single cell suspension . For culture of colony forming units , bacteria were plated on 7H10 agar supplemented with Middlebrook ADC . The bronchial epithelial cell line BEAS-2B ( CRL-9609 ) was obtained from ATCC and cultured in DMEM + 10% heat inactivated FBS . NHBE cells were obtained from Lonza and were cultured in BEGM as recommended . CD8+ T cell clones D160 1–23 , D466 H4 , and D426 B1 have been previously described [3 , 30 , 31] . Briefly , D160 1–23 is HLA-E restricted , D466 H4 is HLA-B45 restricted , and D426 B1 is MR1 restricted . T cell clones were expanded as previously described [32] . This study was conducted according to the principles expressed in the Declaration of Helsinki . Study participants , protocols , and consent forms were approved by the Oregon Health & Science University Institutional Review Board ( IRB00000186 ) . Written informed consent was obtained from all participants . Uninfected adults were recruited from employees at Oregon Health & Science University as previously described [33] . Uninfected individuals were defined as healthy individuals with a negative tuberculin skin test and no known risk factors for infection with Mtb . Enzymes used for molecular cloning were purchased from New England Biolabs . The following antibodies or reagents were used for fluorescence microscopy and flow cytometry: α-MR1 ( 26 . 5 , gift from Ted Hansen , biotinylated by Biolegend ) , α-Class I ( W6/32 , Serotec ) , α-TGN46 ( Abcam ) , α-ERGIC-53 ( Abcam ) , goat anti-mouse and anti-rabbit Alexafluor secondary antibodies ( Life Technologies ) , α-β-2-microglobulin ( SantaCruz ) streptavidin-Alexa647 ( Life Technologies ) , and CellLights reagents ( Life Technologies ) . 6-formyl pterin was obtained from Schirck’s Chemical . A subgenomic lentiviral short hairpin RNA ( shRNA ) library targeting 114 genes trafficking genes was obtained in collaboration with L . Moita through the RNAi Consortium ( Broad Institute ) . Genes were represented by at least five independent shRNA constructs arrayed in individual wells . The assay design is outlined in S1 Fig . Infection of BEAS-2B cells with lentivirus was performed as follows . BEAS-2B cells were plated on 96-well tissue culture plates at a concentration of 5 x 103 cells per well in 100ul DMEM+10%FBS . After 24 hours the media was removed and replaced with 10ul virus and 40ul media containing 8ug/ml polybrene and the plate was centrifuged at 2200 rpm for 90 min at 37°C . After centrifugation , the media was replaced with 100ul fresh DMEM+10%FBS . The cells were incubated for 48 hours , then selected with 5ug/ml puromycin for an additional 72 hours before analysis by ELISPOT assay as follows . Treated cells were infected with H37Rv Mtb ( MOI:30 ) for 18 hours . Following infection , cells were harvested and divided into four parts , one for counting and three for ELISPOT assays as described below . To count the relative number of cells per well , samples were fixed , combined with a fixed number of latex beads , and analyzed by flow cytometry . Wells that contained fewer than 20 cells per 10 , 000 latex beads were excluded from analysis as APCs from these wells did not elicit IFNγ response above background . The other three parts were plated as antigen presenting cells ( APC ) in IFNγ ELISPOT plates with MR1 ( D426B1 ) , HLA-E ( D160 1–23 ) , or HLA-B45 ( D466H4 ) -restricted T cells clones . Each well was given a rank order based on the relative number of cells , which was plotted against the number of IFNγ Spot Forming Units ( SFU ) . Each plate of shRNAs was assayed independently three times with each of the three T cell clones . Prism ( GraphPad ) was used for data analysis . Each well was given a rank value based on the relative number of cells and the rank was plotted in ascending order in relation to the number of spots per well . A linear regression analysis was performed to generate a best-fit line with 95% confidence intervals . Genes were considered candidates if two of the five independent shRNA-silenced wells resulted in T cell response that was reduced by at least 25% compared to the best-fit line in at least two of the three replicates . Selected candidates were confirmed using gene-specific siRNA as described below . BEAS-2B cells were plated in 24- , 12- , or 6-well tissue culture plates ( Corning ) , or 1 . 5mm glass-bottom chamber slides ( Nunc ) at 70% confluency and transfected with 50nM siRNA ( Life Technologies ) using HiPerfect ( Qiagen ) . After 72 hours , cells were used in assays as indicated below . Real-time PCR was used to analyze gene expression . Total RNA was isolated from samples treated with negative control or gene-specific siRNAs using the RNeasy Mini Kit following the manufacturer’s protocol ( Qiagen ) . cDNA was synthesized from total RNA using the High Capacity cDNA Reverse Transcription Kit following the manufacturer’s protocol ( Life Technologies ) . Real time PCR was carried out using TaqMan Universal PCR Master Mix ( Life Technologies ) on a Step One Plus Real-Time PCR System ( Applied Biosystems ) . The FAM-MGB TaqMan Gene Expression Assays for all targets were obtained from Life Technologies . Reactions were run in triplicate in three independent experiments . Expression data were normalized to the housekeeping gene GAPDH and relative expression levels were determined using the 2-ΔΔCT method described by Livak and Schmittgen [34] . pCI MR1-eGFP was constructed using a cDNA ORF clone for MR1 isoform 1 purchased through Origene ( RG220474 , Ref seq: MN_001531 . 1 ) utilizing a Mega-primer PCR protocol . MR1 was first amplified by PCR so that an EcoRI site was integrated into the N-terminal end of the amplicon and a 25 bp overlap with the sequence for the following eGFP amplicon was added to the C-terminus ( Primers: MR1_FW and MR1_RV ) . eGFP was then amplified out of a stock plasmid in our lab with a QGGGGFE flexible linker added to the N-terminus and a KpnI restriction site added to the C-terminal of the amplicon ( Primers: QGGGGFE eGFP_FW and QGGGGFE eGFP_RV ) . The gel-purified MR1 and eGFP amplicons were then amplified in a PCR reaction using the MR1 amplicon as a mega-primer for the reaction . Following PCR purification ( Qiagen cat . 28104 ) of the MR1 eGFP amplicon , an additional PCR reaction was run using primers initiating amplification from each terminal end of the template ( MR1_FW and QGGGGFE eGFP_RV ) . Following gel purification ( Qiagen cat . 28704 ) , the 1 . 8kb MR1 eGFP insert was cloned into the pCI mammalian expression vector ( Promega Cat . E1731 ) using EcoRI and KpnI restriction sites . Confirmatory sequencing of our construct revealed a missense mutation ( A116G ) within the Origene MR1 sequence that changed the coding for amino acid 39 from histidine to arginine . This mutation was repaired using mega-primer PCR utilizing a forward primer that amplified the N-terminal of pCI MR1 eGFP and a 30 bp reverse primer ( MR1 A116wt_ RV ) that spanned the region of the mutation and contained the correct A116 nucleotide . The resulting 145 bp amplicon was then used as a primer along with the reverse primer ( QGGGGFE eGFP_RV ) for a subsequent PCR reaction with pCI MR1 eGFP . The resulting PCR product was restriction cloned into pCI using EcoRI and KpnI sites . The completed plasmid was confirmed correct by sequencing through the entire MR1 eGFP insert . N-terminal Tag-RFP Stx18 and VAMP4 fusion proteins were produced as complete genes in a production plasmid ( Integrated DNA Technologies ) containing EcoRI and KpnI restriction sites . The Tag-RFP gene was cut from the production plasmid , gel purified and subcloned into the pCI mammalian expression vector ( Promega Cat . #E1731 ) using EcoRI and KpnI restriction sites . BEAS-2B cells were transfected with the MR1-GFP , Stx18-RFP , or VAMP4-RFP plasmid constructs using Lipofectamine2000 ( Life Technologies ) or by nucleofection with Amaxa kit T ( Lonza ) , program G-016 . NHBE cells were transfected with plasmid constructs using the Amaxa nucleofector , with kit VPI-1005 ( Lonza ) and program W-001 . For experiments with CellLights , reagent was added one hour after transfection with pCI:MR1-eGFP . Experiments with VAMP4-RFP co-transfection and Brefeldin A treatment were performed using BEAS-2B cells that were stably transduced with MR1-GFP lentiviral particles produced by The University of Pennsylvania Vector Core . BEAS-2B cells in 12- or 6-well tissue culture plates were transfected with siRNAs as indicated above for 72 hours . siRNA transfected cells were infected with Mtb-dsRED ( MOI:10 or as indicated ) for 18 hours , then harvested , counted , and used in equivalent numbers as APC ( 5e3/well ) in an IFNγ ELISPOT assay with 1e5 D426 B1 , D160 1–23 , and D466 H4 CD8+ T cell clones per well as previously described [31] . The data were compared using the student’s t-test , and all data presented are the average of a minimum of three independent replicates . For analysis of the functionality of MR1-GFP , BEAS-2B cells were transfected with MR1-GFP , or mock transfected . After 48 hours , cells were infected for 18 hours , then harvested , counted and plated at equivalent numbers in an IFNγ ELISPOT assay as described above . For the ELISPOT experiments with ‘Msm-sup’ , M . smegmatis was cultured with shaking for 24 hours , then pelleted . The supernatant was passaged over a 0 . 22uM filter , then used directly in the ELISPOT assay . Equivalent numbers of cells treated with siRNA and infected with Mtb as indicated above were lysed and serial dilutions were plated on 7H9 culture plates . Colonies were enumerated after 12–14 days . BEAS-2B cells plated in a 12-well tissue culture plate were transfected with siRNAs as indicated above for 72 hours . In the case of VAMP4 , the cells were incubated with siRNA for 48 hours . After 72 ( 48 ) hours , cells were transfected with pCI:MR1-GFP and incubated for 24 hours . Cells were then incubated at 37°C with 50uM 6-formyl pterin for 16 hours . Cells were harvested on ice and surface stained with a primary antibody against MR1 ( 26 . 5 , a kind gift from Ted Hansen , biotinylated , 1:100 ) for 40 min on ice in the presence of 2%HuS , 2% goat serum , and 0 . 5% FBS . After washing , streptavidin-Alexafluor 647 was added for 40 min on ice . Cells were washed and fixed , and subsequently analyzed with a BD FACSCanto II flow cytometer and FACS Diva software ( BD ) . All analyses were performed using FlowJo software ( TreeStar ) . BEAS-2B cells in 1 . 5mm glass bottom chamber slides ( Nunc ) were transfected with pCI:MR1-GFP ( 2ug/1e6 cells ) . Transfected cells were infected with H37Rv-dsRED , or treated with ligands as indicated above . Cells were washed , fixed with 1% paraformaldehyde for 18 hours , then stained as indicated above with the 26 . 5 antibody . Images were acquired on a high-resolution wide-field CoreDV system ( Applied Precision ) with a Nikon Coolsnap ES2 HQ . Each image was acquired as z-stacks in a 1–24x1024 format with a 60x 1 . 42 NA Plan Apo N objective . Images were deconvolved with an optical transfer function using an iterative algorithm of 10 iterations . Acquired images were analyzed using Imaris ( Bitplane ) . Analysis of MR1-eGFP EC colocalization with CellLights reagents or antibody staining was performed using the “Spots” module of Imaris and the “spots colocalization” MatLab Xtension module as described in S3 Fig . Data were analyzed and plotted using Prism 5 ( GraphPad Software ) . Statistical significance was determined using Student’s two-tailed t test , unless otherwise indicated . | Tuberculosis , caused by the bacterium Mycobacterium tuberculosis ( Mtb ) , remains a global health concern , with an estimated 9 million new cases and 1 . 5 million deaths each year . Mucosal-associated invariant T ( MAIT ) cells were recently identified as a non-classical CD8+ T cell subset that responds to intracellular infection with Mtb and other microbes . MAIT cells recognize vitamin B metabolites presented on the Class I like molecule , MR1 . MAIT cell recognition is likely to be tightly regulated to allow for the detection of intracellular infection while avoiding tissue damage . In this manuscript , we have characterized the intracellular localization and trafficking of MR1 at basal conditions and in the presence of a known ligand . Furthermore , we have investigated the role of intracellular trafficking in MR1 presentation of ligands to MAIT cells in the context of intracellular infection with Mtb . We show that , in contrast to other Class I molecules , MR1 resides in endosomal compartments and translocates to the cell surface in the presence of ligand . We went on to identify trafficking molecules that are required for the presentation of Mtb antigen to HLA-diverse T cells , and found at least seven trafficking molecules that are specifically involved in regulating MR1-dependent recognition of human MAIT cells . Among these , we show that Syntaxin 18 ( Stx18 ) , vesicle-associated membrane protein 4 ( VAMP4 ) , and Rab6 play distinct roles in the trafficking of MR1+ endosomal compartments . Furthermore , our results demonstrate that exogenously added ligands and those derived during intracellular infection are presented through different mechanisms . Our results provide a conceptual framework underlying the regulation of MAIT cell activation . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"blood",
"cells",
"innate",
"immune",
"system",
"medicine",
"and",
"health",
"sciences",
"antigen",
"presentation",
"immune",
"cells",
"enzyme-linked",
"immunoassays",
"pathology",
"and",
"laboratory",
"medicine",
"intracellular",
"pathogens",
"gene",
"regulation",
"ant... | 2016 | Endosomal MR1 Trafficking Plays a Key Role in Presentation of Mycobacterium tuberculosis Ligands to MAIT Cells |
Free-living amoebae are present worldwide . They can survive in different environment causing human diseases in some instances . Acanthamoeba sp . is known for causing sight-threatening keratitis in humans . Free-living amoeba keratitis is more common in developing countries . Amoebae of family Vahlkampfiidae are rarely reported to cause such affections . A new genus , Allovahlkampfia spelaea was recently identified from caves with no data about pathogenicity in humans . We tried to identify the causative free-living amoeba in a case of keratitis in an Egyptian patient using morphological and molecular techniques . Pathogenic amoebae were culture using monoxenic culture system . Identification through morphological features and 18S ribosomal RNA subunit DNA amplification and sequencing was done . Pathogenicity to laboratory rabbits and ability to produce keratitis were assessed experimentally . Allovahlkampfia spelaea was identified as a cause of human keratitis . Whole sequence of 18S ribosomal subunit DNA was sequenced and assembled . The Egyptian strain was closely related to SK1 strain isolated in Slovenia . The ability to induce keratitis was confirmed using animal model . This the first time to report Allovahlkampfia spelaea as a human pathogen . Combining both molecular and morphological identification is critical to correctly diagnose amoebae causing keratitis in humans . Use of different pairs of primers and sequencing amplified DNA is needed to prevent misdiagnosis .
Free-living amoebae ( FLA ) are present in different environments worldwide . They can survive in soil , surface water , other aquatic environments and even desert[1 , 2] . Some members of family Acanthamoebidae and Vahlkampfiidae are amphizoic , occurring as human parasite . The most commonly known genera are Acanthamoeba and Naegleria causing keratitis and primary amoebic meningoencephalitis[3 , 4] . Members of both families have vegetative form , the trophozoite , and quiescent form , the cyst , both can be used for morphological identification of different genera [5–7] . Cysts can survive for many years in environment as a potential source of infection[8] . Amoebic keratitis is an uncommon corneal disease that could eventually lead to loss of vision . It is usually associated with contact lens wearing or trauma[9 , 10] . The most common cause is the genus Acanthamoeba [4 , 11] . In India , Acanthamoeba keratitis was up to 2 . 5% of cases of non-viral keratitis and was more prevalent in rural poor areas[12 , 13] . Vahlkampfia was reported to cause keratitis with co-infection with Acanthamoeba or Candida [11 , 14 , 15] . Correct diagnosis is essential for treatment and prevention of vision loss . Diagnosis depends mainly on culture from corneal scraping [9 , 10] and molecular identification using polymerase chain reaction ( PCR ) . Primers designed to amplify 18S ribosomal subunit are widely used[16–18] . In 2009 , Allovahlkampfia spelaea was identified as a new genus and new species . It was reported as a FLA inhabiting cave in Slovenia [19] . Until now , there is no data about the ability of such amoeba to produce disease in human beings . In this work , we aimed to identify FLA causing keratitis in an Egyptian patient using morphological and molecular approaches .
Corneal scrapings from patient presented with keratitis were cultured on 1 . 5% non-nutrient agar made with Page’s saline and seeded with Escherichia coli kept in incubator at 30°C for 7 days . Cultures were examined using inverted microscope for presence of FLA every day and if FLA was detected , sub-culture was done every 10 to 14 days by inverting a slice on a new agar plate was done[9 , 20 , 21] . Morphology of trophozoites and cysts ( non-stained and Giemsa’s Stained ) was identified using light microscope and inverted microscope according to Smirnov and Goodkov ( 1999 ) and Smirnov and Brown ( 2004 ) [6 , 7] . Trial to axenize isolated FLA was done using Trypticase Soy Broth with Yeast Extract ( TSY ) . Medium was prepared using BD Bacto Tryptic Soy broth ( ref 211825 ) 30 grams , BD Bacto Yeast Extract ( ref 212750 ) 10 grams and distilled water up to 1000 ml , pH adjust to 7 . 3 and autoclaved at 121°C for 15 minutes . To start axenic culture , cysts were collected using phosphate buffered saline ( PBS ) containing 0 . 01N HCL for 15 minutes , then washed 3 times in PBS by centrifugation at 600xg for 4 minutes . Cyst then were suspended in 10 ml TSY and kept in 25cm2 Falcon culture flasks ( both vented and air tight flasks were tested ) . Cysts were collected and treated as previous in axenic trial . Cysts were monitored under inverted microscope for excystation and attachment of trophozoites to flask wall , then medium was decanted and replaced for 3 times to ensure only attached trophozoites were present . Preheated cell lysis solution ( 80°C ) from Gentra Puregene Yeast/Bact . Kit B ( Qiagen ) was added to flask to ensure rapid lysis of trophozoites . Then DNA was extracted according to kit protocol . We used 3 different pairs of primers to identify the FLA; JDP1 ( 5'GGCCCAGATCGTTTACCGTGAA ) and JDP2 ( 5'TCTCACAAGCTGCTAGGGAGTCA ) [16] as standard known primers to identify Acanthamoeba . Universal primers F-566 ( 5'CAG CAG CCG CGG TAA TTC C ) and R-1200 ( 5' CCC GTG TTG AGT CAA ATT AAG C ) as general primers for 18S ribosomal subunit [22] , and Naeg-F ( 5'GAACCTGCGTAGGGATCATTT ) and Naeg-R ( 5'TTTCTTTTCCTCCCCTTATTA ) as general primers for ribosomal internal transcribed spacers ( ITS ) [3 , 20] . PCR reactions were done using Thermo Scientific Phusion High-Fidelity PCR Master Mix ( ref F-531L ) according to product manual for 35–40 cycles . PCR products were identified using 1% Agarose gel stained with ethidium bromide and were purified for sequencing using Qiaquick spin columns ( Qiagen ) . Sequencing was done using Applied Biosystems 3730xl DNA Analyzer . Sequence data were retrieved and blasted using NCBI Blastn engine . After identification of isolated FLA as Allovahlkampfia spelaea , 3 pair of specific primers were designed using NCBI/Primer-BLAST using A . spelaea strain SK1 [GenBank:EU696948] as template ( shown in Table 1 ) . Amplification and sequencing were done as previously described . Phylogenetic trees were created using Mega6 platform [23] , choosing MUSCLE [24] for multiple alignment , and maximum likelihood tree using Tamura-Nei model for generation of tree with gaps removal . Sequences for other amoebae were downloaded from GenBank [accession numbers for 18S ribosomal RNA gene , GenBank: JQ271723 , M98052 , AJ224887 , M18732 , FJ169185 , GU230754 , U94740 , AF251938 , EU696948 , AY425009 , AY029409 , DQ388520 and for ITS , GenBank: KC820644 , AB330071 , AJ698838 , FJ169186 , AJ698839 , V00003 , AJ132032 , K00471 , EU696949 , KF547910] . In order to confirm the ability of A . spelaea to produce keratitis , cysts were collected and treated as previously described in axenic trials then suspended in sterile Page’s saline . They were allowed to excyst in culture flask and supernatant was discarded and replaced . The flask was chilled on ice for 3 minutes and shaken to collect trophozoite . Counting was done using hemocytometer and volume was adjusted to have 1x105 trophozoites/ml . Three laboratory rabbits weighing 1500–2000 gms about 3 months old were used for induction of keratitis . Each rabbit was anesthetized using ether , the left cornea was scratched using 27 gauge sterile needle and 10 μl were instilled in its eye . It was kept under anesthesia for 30 minutes to allow adherence of trophozoites . Right eye was only scratched with 27gauge sterile needle . Eyes were examined for the presence of keratitis grossly and with slit lamp . Corneal scrapings were obtained as routine investigation from patient with chronic keratitis . Patient made written consent for using his samples for both diagnostic and research purposes . Faculty of Medicine research ethics committee , Assiut University , approved this study . Animal experiments were done in Animal House , Faculty of Medicine , Assiut University . Animal House ethical committee , Faculty of Medicine , Assiut University , approved them . Animal handling protocols meet the standard international guidelines by the National Institutes of Health guide for the care and use of Laboratory animals and guideline used in other Egyptian universities and research centers .
Allovahlkampfia spelaea can cause keratitis in humans . This is the 1st time to report such parasite as a human parasite . The presence of FLA in coexistence with each other and with bacteria and fungi makes it necessary to combine both culture and molecular methods for correct diagnosis . For correct molecular diagnosis , use of different primers and sequencing of amplified DNA are important for correct identification of parasite . The close genetic relation between strain isolated in Slovenia and Egypt suggests that the genome of Allovahlkampfia spelaea is not much evolutionary separated but further analysis using full genome sequence is needed . | Free-living amoebae are present worldwide . Some species are known to cause chronic keratitis in human . Amoebic chronic keratitis is sight-threatening disease occurring in both developing and well-developed countries . Allovahlkampfia spelaea is a newly discovered free-living amoeba . We report the first human case of chronic keratitis due to that amoeba . For correct identification , both morphological and molecular techniques should be combined . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"and",
"Discussion"
] | [
"trophozoites",
"medicine",
"and",
"health",
"sciences",
"parasite",
"groups",
"keratitis",
"rabbits",
"vertebrates",
"parasitic",
"protozoans",
"animals",
"mammals",
"parasitology",
"acanthamoeba",
"animal",
"models",
"apicomplexa",
"model",
"organisms",
"protozoans",
"a... | 2016 | Allovahlkampfia spelaea Causing Keratitis in Humans |
The RNA-dependent RNA polymerase VP1 of infectious pancreatic necrosis virus ( IPNV ) is a single polypeptide responsible for both viral RNA transcription and genome replication . Sequence analysis identifies IPNV VP1 as having an unusual active site topology . We have purified , crystallized and solved the structure of IPNV VP1 to 2 . 3 Å resolution in its apo form and at 2 . 2 Å resolution bound to the catalytically-activating metal magnesium . We find that recombinantly-expressed VP1 is highly active for RNA transcription and replication , yielding both free and polymerase-attached RNA products . IPNV VP1 also possesses terminal ( deoxy ) nucleotide transferase , RNA-dependent DNA polymerase ( reverse transcriptase ) and template-independent self-guanylylation activity . The N-terminus of VP1 interacts with the active-site cleft and we show that the N-terminal serine residue is required for formation of covalent RNA∶polymerase complexes , providing a mechanism for the genesis of viral genome∶polymerase complexes observed in vivo .
Virus-encoded RNA-dependent RNA polymerases ( RdRPs ) are vital components in the life cycle of RNA viruses , being responsible for both genome replication and transcription . RdRPs from viruses that are considered unrelated [e . g . hepatitis C virus ( family Flaviviridae ) , Foot-and-Mouth disease virus ( family Picornaviridae ) and bacteriophage Φ6 ( family Cystoviridae ) ] share a similar structure and mechanism of catalysis and are therefore potential targets for generic antiviral drugs [1] . They all have the canonical “right-hand” polymerase fold with palm , fingers and thumb domains and contain a number of conserved sequence motifs ( A–G ) , the conserved residues GDD within motif C being essential for catalytic activity [2] , [3] . Members of the family Birnaviridae possess bi-segmented double-stranded RNA ( dsRNA ) genomes and form virions with non-enveloped single-shelled icosahedral capsids [4] . Infectious pancreatic necrosis virus ( IPNV ) , which infects salmonids , and infectious bursal disease virus ( IBDV ) , a poultry virus , are pathogens of significant economic importance and are the best characterized members of this virus family [5] , [6] . From sequence analysis it was proposed that the RdRPs of Birnaviridae possess a non-canonical active site where the polymerase sequence motifs are re-ordered C-A-B , motif C containing ADN in place of the conserved GDD catalytic residues [3] . The structure of the RdRP VP1 from IBDV confirmed this hypothesis [7] , revealing that the polymerase catalytic site was spatially similar to canonical viral RdRPs despite the different connectivity of the catalytic motifs . However , it remained unclear how substrate and metal ions bound at the active site during catalysis . An additional unusual feature of birnavirus RdRPs is their ability to self-guanylylate: auto-catalyzing the covalent addition of a GMP moiety to a serine residue of the polymerase in a template-independent manner to form VP1pG [8] , [9] , [10] . This self-guanylylation has been presumed to be involved in priming the initiation of replication [11] . VP1 is observed in the virion as two forms , free polymerase and covalently attached to the 5′ ends of the segments of the viral RNA genome [12] , and recent studies have shown that recombinant IBDV VP1 produces both free and protein-attached RNA products [9] . The exact site of self-guanylylation is unclear . Biochemical characterization mapped the self-guanylylation to residue S163 of IPNV VP1 [10] , but no evidence for guanylylation was observed at an equivalent site in structures of the IBDV enzyme [7] , [13] . Recent studies of IBDV VP1 suggest that the self-guanylylation reaction occurs in the first 175 residues of the enzyme at a locus distinct from the polymerase catalytic site [9] . We have cloned , expressed and solved the structure of the RdRP VP1 from IPNV in both its apo form and bound to a catalytic metal ( magnesium ) . We show that IPNV VP1 possesses significant terminal ( deoxy ) nucleotide transferase and reverse transcriptase activity in addition to its RNA replication/transcription activity . Unexpectedly our structures reveal that the first 30 residues of VP1 , not observed in the structure of IBDV VP1 , bind at the active site of the enzyme . Mutation of the N-terminal serine residue of the mature protein to alanine abolishes its ability to form covalent RNA∶polymerase complexes , suggesting a mechanism of VP1∶genome attachment in vivo .
Recombinantly expressed IPNV VP1 carrying a C-terminal hexahistidine affinity tag is predominantly monomeric in solution as determined by analytical gel filtration and multi-angle light scattering ( data not shown ) . Crystals of full-length VP1 with the C-terminal hexahistidine tag removed diffracted to 3 . 8 Å resolution but proved refractory to further optimization . A second construct lacking the C-terminal 55 residues ( ΔC55 ) was therefore cloned , expressed , purified and crystallised . These crystals diffracted well and the structure was solved by molecular replacement using the structure of IBDV VP1 [7] as a search model . The final structure contains 5 molecules in the asymmetric unit and has been refined to 2 . 3 Å resolution with residuals R = 0 . 166 , Rfree = 0 . 187 ( Table S1 ) . The stereochemical quality of the structure is excellent , with 3756 of 3835 residues ( 98% ) occupying favored regions of the Ramachandran plot ( Table S1 ) . The structure of ΔC55 IPNV VP1 comprises the polymerase domain ( residues 31–790 ) , an ordered helix of the N-terminus ( residues 11–19; see below ) and two residues of the C-terminal affinity tag . The IPNV polymerase closely resembles IBDV VP1 , with an r . m . s . deviation in Cα positions of 1 . 2 Å over 678 residues ( Figure 1 ) . The structure of the core polymerase , comprising the fingers , palm and thumb domains , is especially well conserved ( 1 . 0 Å Cα r . m . s . deviation over 463 residues ) , with significant structural rearrangements occurring only at the tips or back of the fingers and thumb , distal to the catalytic palm domain ( Figure 1 ) . The structure of IPNV VP1 confirms the previously-identified rearrangement of the catalytic palm domain [3] , [7] , with the non-canonical ADN catalytic motif ( residues 387–389 ) lying at the apex of a β-hairpin in the active site ( Figure 1 ) . As in IBDV VP1 , extensive N- and C-terminal extensions wrap around the core fingers , palm and thumb domains ( Figure 1 ) . The sequence and structure of the N-terminal extension is largely conserved , differences occurring only in the conformation of the loop between residues 82–91 . The general conformation of the C-terminal extension is similar to IBDV , passing from the inner-base of the thumb to the back of the fingers via the outer edge of the palm . However , many of the surface-exposed loops adopt different conformations and the helix-turn-helix between residues 736–779 is rotated by approximately 50° relative to the equivalent region of IBDV VP1 . Strong electron density was observed near the outer edge of the palm domain , between the side chains of N184 , N409 and N514 and the backbone carbonyl oxygen atoms of V177 , N182 and G512 ( Figure 2 ) . This feature , some 15 Å from the putative active site , was modeled as a K+ ion based on the strength of the electron density , the ligand-to-K+ distances of 2 . 6–2 . 8 Å [14] and the hard nature of the six ligands . To characterize binding of catalytic metals at the active site of the enzyme , crystals of ΔC55 VP1 were soaked in reservoir solution supplemented with 50 mM MgCl2 and 10 µM GTP and lacking citrate ( which chelated metals , frustrating soaking experiments ) . The Mg-bound ΔC55 VP1 structure was refined to 2 . 2 Å resolution with residuals R = 0 . 161 , Rfree = 0 . 181 ( Table S1 ) . A single Mg2+ ion was observed bound at the active site , coordinated by four solvent molecules and the side chain oxygen atoms of N389 and D402 in an octahedral geometry ( Figure 2 ) . The coordination geometry of the bound K+ ion does not change upon addition of Mg2+ , suggesting that Mg2+ does not replace K+ in this position . As in the structure of IBDV VP1 soaked with Mg2+ and GTP [13] , no significant structural rearrangements were observed between the structures of apo and Mg-bound ΔC55 VP1 and no electron density for a bound GTP molecule was observed . Additional electron density resembling a protein helix was observed in the active site cleft of both apo and Mg-bound ΔC55 VP1 . Based on the presence of strong density suggesting a sulfur-containing methionine side chain , this helix was tentatively assigned as residues 11–19 from the otherwise-disordered N-terminal tail of the protein . As the 48 Å separating the final residue of the helix ( M19 ) and the first residue of the polymerase domain ( I31 ) is too great to be spanned by the intervening 10 residues , the N-terminal helix presumably derives from an adjacent polymerase domain in the crystal . The identity of the N-terminal helix was confirmed by solving the structure of ΔC55 VP1 in an alternate crystal form at 3 . 0 Å resolution with residuals R = 0 . 188 , Rfree = 0 . 216 ( ‘large unit cell’ , Table S1 ) . Clear electron density connects the helix with residue 31 of an adjacent molecule in the crystal in 7 of the 8 molecules present in the asymmetric unit ( Figure 3 , Figure S1 ) . In addition , residues 3–10 are seen in this structure extending toward the catalytic active site ( Figure 3 , Figure S1 ) . As in the high-resolution structures , the helix formed by residues 13–19 lies in a hydrophobic pocket within the active site cleft formed by loops of the fingers domain and the N-terminal extension ( Figure 3 , Figure S1 ) . Residues 11–12 form a short isolated β strand which interacts with residue 241 of the fingers domain , while residue 10 spans the two sides of the active site cleft . Residues 7–10 interact with a loop between residues 594–598 of the thumb , this loop being partly mobile and modeled as having two conformations in the refined structure . Residues 3–6 interact with the thumb , the palm and the C-terminal extension , the side chain of F5 being buried in a hydrophobic pocket between the side chains of residues W563 , L578 , R582 and F662 ( Figure 3 ) . While the conformation of S2 ( the first residue of mature VP1 assuming removal of the initiator methionine by methionine aminopeptidase ) could not be assigned unambiguously , it will clearly lie extremely close to the catalytic residues of the active site ( Figure 3 ) . The structure of full-length IPNV VP1 was solved using ΔC55 VP1 as a search model and refined to 3 . 8 Å resolution with residuals R = 0 . 186 , Rfree = 0 . 213 ( Table S1 ) . As for ΔC55 VP1 in the large unit cell , electron density for residues 3–19 was present in the active site . Some residual electron density could be observed linking residue 19 to the first residue ( E27 ) of an adjacent polymerase molecule , but it was not sufficiently well-resolved to allow modeling of the intervening residues . Despite being crystallized in the presence of Mn2+ and ATP , no metal ions or nucleotides were observed at the active site . While additional electron density was seen connected to the C-terminus of the molecular replacement search model in three of the four copies of VP1 present in the asymmetric unit only eight additional residues ( 791–798 ) could be placed , suggesting that residues 799–845 are not ordered . Surprisingly the C-terminus of the fourth molecule in the asymmetric unit ( from residue 687 onwards ) is significantly refolded compared to the other IPNV VP1 molecules ( Figure S2 ) and IBDV VP1 . The strand that runs along the side of the palm domain is shifted ‘upward’ to lie at the outer rim of the active site cleft . Neither the helix that lay between the base of the palm and fingers ( residues 706–716 ) , the three-helix bundle ( residues 724–781 ) nor the intervening loop region are evident; instead the polypeptide has flipped up and projects away from the back of the fingers . A continuous stretch of density returning from this projection covers the back of the fingers and extends to cover the back base of the fingers and the N-terminal extension , but given the low resolution of the data we were unable to include this region in the final refined model . Full-length VP1 , the ΔC55 construct used to generate well-diffracting crystals , and an N- and C-terminally truncated construct ( ΔN27C55 ) that represents the well-folded VP1 polymerase domain ( residues 28–790 ) were biochemically characterized . All recombinant VP1 constructs are capable of initiating de novo synthesis with non-specific single-stranded ( ss ) and double-stranded ( ds ) RNA templates , but not with ssDNA or dsDNA molecules ( Figure 4 , Table 1 and data not shown ) . VP1-catalysed polymerization proceeds readily in the presence of divalent cations and nucleotides ( NTPs or dNTPs ) and produces full-length double-stranded products ( dsRNA or RNA/DNA hybrids , see below ) of all tested ssRNA templates . Full-length and truncated VP1 are less productive than the highly-active RdRP from bacteriophage Φ6 , which was used for comparison ( Figure 4 , Table 1 ) . VP1 can perform in vitro semi-conservative transcription with dsRNA templates , albeit less efficiently than replication ( Table 1 ) . VP1 is devoid of activity in the absence of divalent cations , whereas addition of Mg2+ or Mn2+ at concentrations exceeding 2 mM stimulates RNA synthesis ( Figure S3 ) . Adding Ca2+ as the sole divalent cation to the reaction mixture does not induce catalysis ( data not shown ) and supplementing the Mg2+- or Mn2+-containing reaction mixtures with Ca2+ completely inhibits polymerization ( data not shown ) . The addition of Zn2+ yields low levels of RNA synthesis at a concentration of 2 mM or less , whereas higher concentrations efficiently inhibit catalysis ( data not shown ) . Supplementing reaction mixtures with K+ does not stimulate RNA synthesis , and K+ on its own is unable to induce catalysis ( data not shown ) . Interestingly , adding both Mg2+ and Mn2+ to the reaction mixture at their respective optimal concentration has an adverse effect on RNA synthesis and RNA polymerization is most efficient when using half the optimal concentration of both ions ( 5 mM Mg2+ and 1 mM Mn2+; data not shown ) . VP1 is not able to initiate primer-dependent RNA synthesis but displaces various complementary RNA oligonucleotides annealed to ssRNA templates ( Figure S4 ) despite poor in vitro transcription activity ( Table 1 ) . Synthesis of a significant proportion of RNA strands is initiated by means of back-priming ( Table 1 ) , as observed previously for IBDV VP1 [15] . IPNV VP1 shows only a slight preference for cytidine as the terminal template nucleotide ( Table 1 ) . The elongation rate of IPNV VP1 is slow when compared with the Φ6 RdRP ( Table 1; Figure 4 ) and VP1 reaches a higher product yield with short ( <500 nucleotide ) ssRNA templates ( Figure 4 ) . Based on structural data , we designed a set of mutant polymerases to verify the identity of the catalytic center . Single point mutations were introduced at the putative active site ( D388N , D402N and S400A ) and at the K+ binding site ( N184S and N514H ) . The replication activity of all these mutant polymerases was below the detection limit of our assays ( <5% of wild-type VP1 activity , data not shown ) . IPNV VP1 is capable of utilizing free NTPs or dNTPs for the template-independent addition of one or more nucleotides to the 3′ terminus of single- or double-stranded RNA , so-called terminal ( deoxy ) nucleotide transferase ( TNTase and TdNTase ) activity ( Figure 4 ) . The TNTase activity of VP1 is very weak compared to that of Φ6 RdRP , although this activity is significantly enhanced in the truncated VP1 constructs ( Figure 4 , Table 1 ) . Somewhat surprisingly , ΔC55 VP1 has a 5-fold higher TdNTase activity than the full-length enzyme , surpassing the equivalent activity of the Φ6 RdRP by more than two-fold , while the TdNTase activity of ΔN27C55 VP1 is severely compromised ( Table 1 ) . In addition to the TdNTase activity of IPNV VP1 , the polymerase is capable of utilizing a ssRNA template and dNTPs for de novo RNA-directed DNA polymerization resulting in a RNA/DNA hybrid ( Figure 4 , Figure S5 ) . This reverse transcriptase activity is less efficient than in vitro RNA replication when utilizing the same ssRNA template . Both full-length and ΔC55 VP1 yield RNA/DNA hybrids of various heterologous ssRNA templates , ΔC55 being roughly 5-fold more active than the full-length enzyme , while ΔN27C55 VP1 reverse transcriptase activity is extremely low ( Table 1 , Figure 4 ) . Upon incubation with [α32P]-GTP both full-length and ΔC55 VP1 acquire radio-labels that do not dissociate during denaturing gel electrophoresis ( Figure 5 ) , presumably because the proteins have covalently bound [α32P]-GMP to form VP1pG ( so-called self-guanylylation activity ) [8] . A mutant form of VP1 where the N-terminal serine residue of the mature protein has been mutated to alanine ( S2A ) maintains the ability to form VP1pG , while ΔN27C55 VP1 is not radio-labeled under denaturing conditions and therefore lacks self-guanylylation activity ( Figure 5 ) . Replication of ssRNA by IPNV VP1 yields high molecular weight products in addition to the expected dsRNA , manifesting as ladders and smears of radioactivity on native agarose gel electrophoresis [8] . The formation of ladders is most marked when using short RNA templates and is consistent with the formation of ‘concatenated’ RNA products containing integral repeats of the template RNA ( Figure 4A and data not shown ) . RNA concatenation activity is strongest for ΔN27C55 VP1 , while formation of the larger heterogeneous smears is more pronounced for full-length and ΔC55 VP1 ( Figure 4A ) . Following denaturing sodium dodecyl sulfate polyacrylamide gel electrophoresis ( SDS PAGE; Figure 5 ) two major products are observed: free RNA product and RNA∶polymerase complexes that represent covalent association of VP1 with the nascent RNA [9] . However , while ΔN27C55 and S2A VP1 are capable of producing RNA product , they lack the ability to form covalent RNA∶polymerase complexes .
IPNV VP1 requires metal ions for catalysis , optimal activity being obtained in the presence of 5 mM Mg2+ and 1 mM Mn2+ ( Figure S3 and data not shown ) . However , when soaked with an excess of Mg2+ ions we observe only one Mg2+ ion at the active site , coordinated by the side chains of N389 and D402 ( Figure 2 ) . It was reported that three Mg2+ ions were bound in the 3 . 15 Å structure of Mg-bound IBDV VP1 [13] , however none of these occupy the Mg2+ site we observe in IPNV VP1 . We performed additional experiments , soaking ΔC55 IPNV VP1 crystals with up to 200 mM Mg2+ and 40 mM Mn2+ , but these failed to reveal any additional bound Mg2+ ( data not shown ) . In the light of these results and inspection of the IBDV electron density maps , we ascribe differences in the observed metal binding to difficulties in interpreting the low resolution electron density of the earlier IBDV VP1 structure . In initiation or elongation complexes of canonical viral polymerases two catalytic metal ions ( plus one structural metal in the case of Φ6 RdRP ) are observed at the active site [16] , [17] , [18] , [19] . In such complexes the catalytic metal ions bridge the conserved Asp residues of catalytic motifs A and C , lying above the tip of the β hairpin ( formed by motif C ) , and the triphosphate group of the substrate nucleotide . In the structure of reovirus RdRP solved in the absence of substrate nucleotides a single Mn2+ ion is observed in the active site in a position corresponding to that we observe for the single Mg2+ of IPNV VP1 [18] , being coordinated by the side chains of D735 , D585 and E780 ( equivalent to VP1 residues N389 , D402 and half-way between D559 and S400 , respectively; Figure 2 ) . Upon formation of an initiation complex two Mn2+ ions bind the reovirus RdRP active site ( Figure 2 ) . It is therefore likely that in the presence of template and nucleotide substrates a second metal ion will bind at the active site of IPNV VP1 and the resulting initiation complex will contain two metal ions coordinated by D388 , D402 and the nucleotide triphosphate moiety at the tip of the catalytic motif C β hairpin . In addition to the Mg2+ ion bound at the catalytic active site , in all structures of IPNV VP1 we observe a K+ ion bound at the junction of the palm and fingers domains ( Figure 2 ) . We find that density consistent with a K+ ion is also present in previously published structures of IBDV VP1 , although it was not identified as such at the time . The residues that interact with the K+ ion ( N184 , N409 and N514 ) are conserved across the Birnaviridae ( Figure S6 ) and we propose that binding of this K+ ion is a general feature of birnavirus polymerases . The region at which the K+ ion binds is not structurally conserved in canonical RdRPs and the K+ ion is not analogous to ‘structural’ metal ions such as the divalent cation bound on top of the palm near the catalytic residues in the Φ6 RdRP [20] . Mutations N184S or N514H ( both predicted to prevent K+ binding ) abolish polymerase activity , consistent with the K+ ion performing a role in structuring the palm and maintaining its orientation relative to the fingers . IPNV VP1 has significant reverse transcriptase activity , utilizing dNTPs and RNA to form DNA/RNA hybrids ( Figure 4 , Figure S5 ) . VP1 RNA-dependent RNA polymerization and reverse transcriptase activities share a requirement for RNA ( not DNA ) template , are optimally active under similar conditions , and are both abolished by mutations to the active site residues D388 , S400 and D402 , consistent with the two activities sharing a common catalytic mechanism . The molecular features of VP1 that allow it to utilize dNTPs in addition to NTPs are not obvious . Mutation of F155 to valine in Moloney murine lukemia virus reverse transcriptase removes its specificity for dNTPs [17] , [21] . Birnavirus RdRPs have a conserved glutamic acid at the equivalent position ( E407 in IPNV , Figure S6 ) , while in other viral RdRPs this residue is aspartic acid [7] . It had been suggested that this aspartate residue might recognize substrate NTPs by forming a hydrogen bond with the NTP O2′ atom [22] , but structures of Φ6 and Norwalk virus RdRP initiation complexes contradict this hypothesis [16] , [19] . The presence of asparagine in catalytic motif B near the tip of the motif C hairpin is unable to explain the preference of viral RdRPs for NTPs over dNTPs [23] , [24] since this residue is asparagine in IPNV VP1 ( N480 , Figure 1 ) . It is also unlikely that serine in the loop preceding the motif B helix ( IPNV S471 ) is crucial for ( d ) NTP selectivity [7] , as serine or threonine at this position is generally conserved across RdRPs not competent for reverse transcriptase activity . Removal of the N-terminal 27 residues of IPNV VP1 greatly diminishes the reverse transcriptase activity of the enzyme and ΔN27C55 VP1 has enhanced TNTase activity but severely attenuated TdNTase activity . Based on a structural alignment with the Φ6 RdRP we observe that the N-terminus of the polymerase domain is adjacent to the ( d ) NTP entry tunnel ( Figure S7 ) , suggesting that the N-terminal segment of VP1 plays a role in recruiting dNTPs for catalysis . Further work is required to understand the molecular features that define NTP versus dNTP selectivity in viral RdRPs . Residues 688 onwards are completely refolded in one of the four molecules present in the full-length VP1 structure ( Figure S2 ) . The functional implications of this reorganization are not obvious . Models of a VP1 elongation complex based on superposition of the reovirus RdRP elongation complex [18] or of HIV reverse transcriptase in complex with a DNA template [17] require refolding of a small segment of the VP1 C-terminal extension , residues 654–670 , in order to accommodate the nascent RNA duplex ( not shown ) . It is possible that the purpose of the C-terminal extension is to rigidify the polymerase during initiation , but that this extension must peel away to allow the growing nucleotide chain to exit the active site . The structure of VP1 where residues 688 onwards are refolded might thus represent an intermediate unfolding step rather than a conformation of biological significance in its own right . Birnavirus RdRPs hydrolyze GTP to form a covalent VP1pG complex in a template-independent manner , so-called self-guanylylation [8] . We see no evidence of covalent GMP modification at S163 , the IPNV VP1 residue previously identified as being guanylylated [10] , nor at any other position in the structures presented here or in crystals of VP1 co-crystallized with GTP ( data not shown ) . Removal of the N-terminal 27 residues abolishes VP1 self-guanylylation activity ( ΔN27C55 VP1 , Figure 5 ) , consistent with recent studies of IBDV VP1 that identified the site of self-guanylylation as residing within the first 175 residues of the polymerase [9] . We observe that ΔN27C55 VP1 maintains significant replication and transcription activity , demonstrating that self-guanylylation is not required for general “protein priming” of RNA polymerase activity . Further , as S2A VP1 self-guanylylates but lacks the ability to form covalent RNA∶polymerase complexes , the self-guanylylation activity is not sufficient for covalent attachment of RNA product to the enzyme ( see below ) . While it is therefore clear that the N-terminal 27 residues of VP1 are required for self-guanylylation , the precise site and mechanism of this activity remains enigmatic . In 16 of the 17 independent molecular views of the IPNV VP1 structure presented here the flexible N-terminal tail of an adjacent VP1 molecule in the crystalline lattice lies anchored in the active site cleft ( Figure 3 ) . The affinity of this self-interaction is obviously not high as VP1 behaves predominantly as a monomer in solution ( data not shown ) . However , the high concentration of VP1 in crystal structures may well reflect the environment of the viral capsid more closely than biochemical assays performed in dilute solution . There are two main sites of interaction between the N-terminal tail and the active site cleft of IPNV VP1: F5 binds in a pocket formed by residues from the thumb and palm , and a helix ( S13–M19 ) plus the two residues that precede it ( K11–A12 ) bind a hydrophobic pocket formed by the fingers and N-terminal extension . F5 and the residues with which it interacts ( W563 , L578 , R582 and F662 ) are highly conserved across birnavirus RdRPs ( Figure S6 ) , the conformations of these residues being identical in IBDV and IPNV VP1 . A hydrophobic helix-binding pocket is also evident in IBDV VP1 . The edge of the pocket proximal to the active site ( IPNV residues 238–242 ) is similarly positioned in IPNV and IBDV VP1 . The loops which form the edges of the pocket distal from the active site ( IPNV residues 86–91 and 227–234 ) are very poorly ordered in IBDV VP1 structures , indicating significant flexibility , and they would require only modest rearrangement to accommodate a bound N-terminal helix . In crystal structures of IBDV VP1 there is no evidence for the N-terminal tail binding the active site [13]; however , the molecular packing does not bring N-termini of adjacent molecules into close enough proximity to interact with the active site , precluding the interaction . The observed self-interaction brings the N-terminal residue of VP1 ( S2 , assuming removal of the initiator methionine ) within approximately 5 Å of the catalytic site ( Figure 3 ) , and removal of the N-terminal tail or mutation of S2 to alanine abolishes the ability of VP1 to form covalent RNA∶polymerase complexes ( Figure 5 ) . Superposition of VP1 onto the initiation complex of the Φ6 RdRP shows the VP1 N-terminal tail to be located in approximately the same position as the 5′ nucleotide of a nascent RNA daughter strand ( Figure 6 ) . The N-terminal interaction is not required for “protein priming” since ΔN27C55 VP1 maintains significant replication activity ( Figure 5 ) . Instead , we propose that residue S2 at the N-terminus of VP1 represents the site of covalent attachment of RNA product to the polymerase . As there is no strong preference for templates with 3′ cytidine bases it is unlikely that the covalently-attached RNA is formed by extension of a self-guanylylation guanosine that Watson-Crick base pairs with template . Further , S2A VP1 is incapable of forming covalent RNA∶polymerase complexes despite maintaining self-guanylylation activity . This suggests that nascent RNA is directly ligated to S2 , independently of initiation or of any self-guanylylation activity . Such a hypothesis , based on the observed interaction of the VP1 N-terminus with the active site and requirement for an N-terminal serine to generate covalent RNA∶polymerase complexes , provides an elegant molecular mechanism for the birnavirus VP1∶genome association observed in vivo .
A pET-21a ( + ) –derived plasmid encoding full-length VP1 from IPNV strain Jasper ( UniProt ID P22173 ) plus a C-terminal VEH6 tag [10] was the generous gift of Dr P . Dobos ( University of Guelph ) . VP1 constructs lacking the C-terminal 55 residues ( ΔC55 ) and N-terminal 27 plus C-terminal 55 residues ( ΔN27C55 VP1 ) were cloned as described in Text S1 . Site-directed mutagenesis of full-length IPNV VP1 was performed using the QuikChange site-directed mutagenesis kit ( Stratagene ) . Full-length , ΔC55 , ΔN27C55 and mutant VP1 were expressed and purified by Ni-immobilization affinity and gel-filtration chromatography using standard protocols ( see Text S1 ) . Pure VP1 was used immediately for crystallization or stored at −20°C in 50% v/v glycerol for kinetic analysis ( conditions under which the enzyme remained stable and active for several months ) . Crystals of ΔC55 VP1 were grown in sitting drops containing 200 nL protein ( 5 . 8–6 . 8 mg/mL ) and 100 nL reservoir solution ( 20–18% w/v PEG3350 , 0 . 10–0 . 09 M bis-Tris propane pH 7 . 5 , 0 . 2–0 . 18 M sodium citrate ) equilibrated against 95 µL reservoirs at 20 . 5°C . Crystals were cryoprotected by soaking in mother-liquor supplemented with 20–25% glycerol for 1–10 min . Crystals of Mg-soaked ΔC55 VP1 were prepared by diluting the mother-liquor with ∼0 . 5 µL 20% w/v PEG3350 , 0 . 1 M bis-Tris propane pH 8 . 25 and 20% v/v glycerol , transferring the crystals to a fresh drop containing 20% w/v PEG3350 , 0 . 1 M bis-Tris propane pH 8 . 25 , 20% v/v glycerol , 50 mM MgCl2 and 10 µM GTP , and incubating for 10 min . Crystals of full-length VP1 were grown in sitting drops containing 100 nL protein ( 5 . 3 mg/mL ) plus 5 mM MnCl2 and 100 nL reservoir solution ( 20% w/v PEG3350 , 0 . 1 M bis-Tris propane pH 7 . 5 , 0 . 2 M sodium citrate , 20 mM ATP , 5% v/v MPD , 10 mM NaOH ) equilibrated against 95 µL reservoirs at 20 . 5°C . All crystals were dipped into perfluoropolyether oil ( PFO-X125/03 , Lancester Synthesis ) prior to cryocooling in a 100 K stream of N2 gas . Initial diffraction data collected on Diamond beam line I24 guided crystal optimization , from which diffraction data were recorded at Diamond beam line I02 ( ΔC55 VP1 ) or ESRF beam line ID14-2 ( Full-length VP1 ) . Diffraction data were processed using MOSFLM [25] and SCALA [26] as implemented by xia2 [27] . Data processing statistics are shown in Table S1 . The structure of Mg-soaked ΔC55 VP1 was solved by molecular replacement with PHASER [28] using the structure of IBDV VP1 , PDB ID 2PGG [7] , as a starting model . The structures of apo ΔC55 VP1 , apo ΔC55 VP1 in an alternate ( large ) unit cell and of full-length VP1 were solved by molecular replacement using the polymerase domain of Mg-soaked ΔC55 VP1 ( residues 31–792 ) as a starting model . Manual building was performed using COOT [29] and structures were refined using BUSTER-TNT [30] in consultation with the MolProbity web server [31] . Non-crystallographic local structure similarity restraints [32] were used during the refinement of all structures . For full-length VP1 and apo ΔC55 VP1 in the large unit cell local structure similarity restraints [32] were also used to deter deviation of the models from the high-resolution apo ΔC55 VP1 structure unless compelling evidence for a difference was present in the electron density . As the test sets for all structures were chosen randomly ( apo and Mg-bound ΔC55 VP1 sharing the same random set ) the presence of non-crystallographic symmetry may artificially lower the value of Rfree by a small amount but will not render the metric invalid [33] . Structural superpositions were performed using SSM [34] and Theseus [35] . Molecular graphics were prepared using PyMOL ( DeLano Scientific ) and sequence alignments with ALINE [36] . Structure factors and final refined coordinates have been deposited in the PDB with accession codes 2yi8 ( apo ΔC55 VP1 ) , 2yi9 ( Mg-bound ΔC55 VP1 ) , 2yia ( apo ΔC55 VP1 , large unit cell ) and 2yib ( full-length VP1 ) . The RdRP from bacteriophage Φ6 was expressed and purified as described previously [37] and RNA templates were generated as described in Text S1 . Please refer to Table S2 for an overview of the RNA templates . The reaction conditions for RNA polymerization by IPNV VP1 were optimized using the 723 nucleotide ( nt ) ssRNA sΔ+13 template ( Figure S3 ) and found to be similar to the ‘standard’ Φ6 RdRP reaction conditions ( 50 mM HEPES-KOH pH 7 . 5 , 20 mM NH4Ac , 6% w/v PEG 4000 , 5 mM MgCl2 , 1 mM MnCl2 , 0 . 1 mM EDTA , 0 . 1% v/v Triton X-100 ) [38]; these conditions were hence used for all subsequent experiments unless otherwise noted . Under optimized reaction conditions all VP1-catalysed reactions except reverse transcription were readily observable by ethidium bromide staining of agarose gel electrophoresis , indicating a high level of polymerase activity . Replication ( ssRNA template ) and transcription ( dsRNA template ) assays were carried out in the presence of 1 mM ATP and GTP and 0 . 2 mM CTP and UTP supplemented with labeled [α32P]-UTP ( replication , transcription ) or [α32P]-GTP ( replication ) . TNTase , TdNTase and self-guanylylation assays contained only labeled nucleotides of [α32P]-UTP ( TNTase ) , [α32P]-dTTP ( TdNTase ) or [α32P]-GTP ( self-guanylylation ) . Reverse transcription was performed with 1 mM dATP and dGTP and 0 . 2 mM dCTP supplemented with 0 . 3 µM ( 0 . 1 mCi/mL ) labeled [α32P]-dTTP . After 2 h incubation , reactions were stopped by the addition of 2× or 10× loading buffer [39] , [40] and analysis of the reaction products was performed using 0 . 8% w/v agarose gel ( TBE ) electrophoresis [40] . [α32P]-GTP labeled self-guanylylation and replication assays were terminated with 7× SB loading buffer [40] , boiled for 3 min and analyzed in 8% SDS polyacrylamide gels . Oligonucleotide displacement was assayed as previously described [41] . Signals were collected by autoradiography on BAS1500 image plates ( Fujifilm ) , which were scanned using a Fuji BAS-1500 phosphorimager ( Fujifilm ) . Digital image analysis ( densitometry ) was performed using AIDA Image Analyzer software ( version 3 . 44; Raytest Isotopenmeβgeräte ) , measuring the band intensities in 1D Evaluation mode using Lane and Peak Determination . | Infectious pancreatic necrosis virus ( IPNV ) is highly contagious and causes severe disease in fish . As a result of intensive rearing conditions it has become a serious problem for the salmon and trout farming industries . IPNV , like many other viruses , replicates its genome using a protein ( a ‘polymerase’ ) that is itself encoded by the viral genome . Unusually , in infectious IPNV particles the polymerase is found chemically linked to the viral genome . We have determined the atomic structure of IPNV polymerase using X-ray crystallography , revealing some significant differences in the fold of the protein chain compared to other well-characterized viral polymerases . By mutating an amino acid residue at the beginning of the protein we show how the chemical linkage to the viral genome can be disrupted . This provides an elegant mechanism for the attachment of the viral genome to the polymerase observed in vivo . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biomacromolecule-ligand",
"interactions",
"protein",
"interactions",
"enzymes",
"viral",
"enzymes",
"microbiology",
"fish",
"farming",
"aquaculture",
"protein",
"structure",
"dna",
"rna",
"synthesis",
"enzyme",
"kinetics",
"dna",
"synthesis",
"proteins",
"marine",
"and",... | 2011 | The N-Terminus of the RNA Polymerase from Infectious Pancreatic Necrosis Virus Is the Determinant of Genome Attachment |
Enterovirus 71 ( EV71 ) , a major causative agent of hand , foot , and mouth disease , occasionally causes severe neurological symptoms . We identified P-selectin glycoprotein ligand-1 ( PSGL-1 ) as an EV71 receptor and found that an amino acid residue 145 in the capsid protein VP1 ( VP1-145 ) defined PSGL-1-binding ( PB ) and PSGL-1-nonbinding ( non-PB ) phenotypes of EV71 . However , the role of PSGL-1-dependent EV71 replication in neuropathogenesis remains poorly understood . In this study , we investigated viral replication , genetic stability , and the pathogenicity of PB and non-PB strains of EV71 in a cynomolgus monkey model . Monkeys were intravenously inoculated with cDNA-derived PB and non-PB strains of EV71 , EV71-02363-EG and EV71-02363-KE strains , respectively , with two amino acid differences at VP1-98 and VP1-145 . Mild neurological symptoms , transient lymphocytopenia , and inflammatory cytokine responses , were found predominantly in the 02363-KE-inoculated monkeys . During the early stage of infection , viruses were frequently detected in clinical samples from 02363-KE-inoculated monkeys but rarely in samples from 02363-EG-inoculated monkeys . Histopathological analysis of central nervous system ( CNS ) tissues at 10 days postinfection revealed that 02363-KE induced neuropathogenesis more efficiently than that induced by 02363-EG . After inoculation with 02363-EG , almost all EV71 variants detected in clinical samples , CNS , and non-CNS tissues , possessed a G to E amino acid substitution at VP1-145 , suggesting a strong in vivo selection of VP1-145E variants and CNS spread presumably in a PSGL-1-independent manner . EV71 variants with VP1-145G were identified only in peripheral blood mononuclear cells in two out of four 02363-EG-inoculated monkeys . Thus , VP1-145E variants are mainly responsible for the development of viremia and neuropathogenesis in a non-human primate model , further suggesting the in vivo involvement of amino acid polymorphism at VP1-145 in cell-specific viral replication , in vivo fitness , and pathogenesis in EV71-infected individuals .
Enterovirus 71 ( EV71 ) is a non-enveloped positive-stranded RNA virus belonging to the species Enterovirus A of the genus Enterovirus in the family Picornaviridae . The EV71 RNA genome is enclosed within an icosahedral capsid comprising 60 structural protein subunits ( protomers ) , each containing four viral structural proteins , VP1-VP4 [1 , 2] . According to the molecular epidemiological analysis of the capsid VP1 sequence , EV71 was previously classified into the three genogroups A , B ( subgenogroups B1–B5 ) , and C ( subgenogroups C1–C5 ) [3 , 4] . Recently , several additional genogroups of EV71 were identified as genogroups D , E , and F [5] . EV71 is a major causative agent of hand , foot , and mouth disease ( HFMD ) along with coxsackievirus A16 ( CVA16 ) and coxsackievirus A6 ( CVA6 ) [6 , 7] . EV71 usually causes mild or subclinical infection . However , in some patients , EV71 may cause severe neurological symptoms , including meningitis , brainstem encephalitis , poliomyelitis-like paralysis , pulmonary edema , and death . As recent EV71 outbreaks in the Asia-Pacific region , including Malaysia , Taiwan , China , Cambodia , and Vietnam , have involved millions of children ( almost all under 5 years old ) including thousands of fatal cases , EV71 poses a threat to global public health [4 , 8–10] . Recently , a growing number of viral and host factors associated with EV71 infection have been reported [11–14]; however , no conclusive risk factors for neuropathogenesis have yet been identified . In the previous studies , two molecules , scavenger receptor class B , member 2 ( SCARB2 ) [15] and P-selectin glycoprotein ligand-1 ( PSGL-1 ) [16] were identified as functional receptors for EV71 . Subsequently , other cell surface molecules , including heparan sulfate [17] , annexin II [18] , sialic acid [19] , dendritic cell-specific ICAM3-grabbing non-integrin ( DC-SIGN ) [20] , and vimentin [21] have been identified as EV71-binding molecules involved in the early stages of EV71 infection . SCARB2 is expressed on the membrane of various cells and tissues , and involved in the endocytic transport mechanism from the ER to lysosomes [22 , 23] . Expression of human SCARB2 allows non-susceptible mouse L929 cells ( L-SCARB2 ) to support EV71 replication and development of cytopathic effects [15] . All EV71 strains examined and coxsackievirus A7 , A14 , and A16 replicate in L-SCARB2 and SCARB2-positive RD cells in a SCARB2-dependent manner [15 , 24] . Furthermore , transgenic mice expressing human SCARB2 ( SCARB2-Tg ) are more susceptible to EV71 and developed neurological disorders following EV71 infection [25 , 26] . We have identified another functional cellular receptor , PSGL-1 [16] , expressed on lymphocytes that plays a critical role in tethering and rolling during recruitment of leukocytes from blood vessels into inflamed tissues [27 , 28] . In contrast to SCARB2-Tg mice , human PSGL-1 expressing transgenic mice are not susceptible to EV71 infection [29] . Independent of genogroup/subgenogroup , some EV71 strains have been found to bind PSGL-1 and infect Jurkat T cells in a PSGL-1-dependent manner whilst other strains do not . Thus , EV71 strains are classified into two distinct phenotypes according to PSGL-1-binding capability; PSGL-1-binding ( PB ) and PSGL-1-nonbinding ( non-PB ) strains [16] . Through molecular epidemiological , structural , and mutational analyses of EV71 , we recently demonstrated an amino acid residue 145 of the capsid protein VP1 ( VP1-145 ) as a critical molecular determinant of the binding of PB EV71 strains to PSGL-1 [30] . VP1-145 is located within the DE loop at the center of the 5-fold mesa of EV71 virions and functions as a molecular switch to change the binding to PSGL-1 by regulating the orientation of the side chain of lysine at VP1-244 [30] . Accordingly , EV71 strains with either G or Q at VP1-145 ( VP1-145G or VP1-145Q ) exhibit a PB phenotype and those with E at VP1-145 ( VP1-145E ) have a non-PB phenotype regardless of EV71 genogroup . VP1-145 is a variable amino acid residue among the field EV71 isolates and has been identified as a major site of positive selection in the molecular evolution of EV71 [31–33] . Based on the available VP1 sequences of EV71 in the GenBank database , VP1-145E isolates were most predominant ( 81% ) , and VP1-145G ( 9% ) and VP1-145Q ( 9% ) isolates were less common than VP1-145E [30] . Furthermore , substitution of VP1-145G or VP1-145Q to VP1-145E was found to be responsible for murine adaptation and/or virulence , either alone or in combination with other amino acids , demonstrating VP1-145E strains of EV71 are more virulent than either VP1-145G or VP1-145Q ( VP1-145G/Q ) strains in different mouse models [34–37] . On the other hand , recent molecular epidemiological studies have suggested that VP1-145G/Q isolates are more frequently detected in cases with severe neurological disease in humans than VP1-145E isolates [38–41] . These apparently contradictory findings in humans and mouse models are yet to be resolved . To elucidate the in vivo involvement of PSGL-1-dependent replication and pathogenesis , and the role of amino acid polymorphism at VP1-145 , we investigated viral replication , pathogenicity , and genetic stability of PB ( VP1-145G ) and non-PB ( VP1-145E ) strains of EV71 in a cynomolgus monkey model , a more reliable animal model than mouse models due to greater homology between primate and human PSGL-1 molecules than mouse . We found that , following inoculation of monkeys with the PB strain of EV71 , the PB strain with VP1-145G frequently underwent mutation at VP1-145 from G to E ( VP1-145E ) , and the resultant VP1-145E variants were capable of inducing viremia and neuropathology , presumably in a PSGL-1-independent manner . Conversely , PB variants with VP1-145G were identified only in peripheral blood mononuclear cells ( PBMCs ) in two out of four PB-inoculated monkeys in the later stages of infection , suggesting potential involvement of PSGL-1-dependent EV71 replication of PB variants in cell-specific viral replication and pathogenesis in EV71-infected individuals .
To prepare cDNA-derived PB and non-PB strains of EV71 , we used an infectious molecular clone of the 02363 strain of EV71 ( subgenogroup C1; GenBank accession No . AB747375 ) . The cDNA-derived 02363 strain contains a combination of VP1-98K and VP1-145E ( EV71-02363-KE strain ) . VP1-145E has been identified as a single determinant of the non-PB phenotype by using a series of 02363-derived EV71 mutants in which amino acid substitutions were introduced at VP1-145 and/or VP1-98 [30] . VP1-98 ( 98E or 98K ) was not directly responsible for the PB phenotype . In the previous studies , VP1-145E of EV71 mutants was found to be unstable and rapidly reverted to VP1-145G during virus passage in RD cells [30 , 35] . However , no apparent reversions at VP1-145 were observed in RD cells for the 02363-KE strain or the 02363-derived strain with VP1-98E and VP1-145G mutations ( EV71-02363-EG strain ) . Therefore , we choose the cDNA-derived EV71-02363-KE and EV71-02363-EG strains , with two amino acid differences at VP1-98 and VP1-145 , as representative non-PB and PB strains , respectively ( Fig 1A ) for use in a cynomolgus monkey model . To minimize the emergence of revertants and quasi-species during virus passages in cell culture , viral stocks were prepared from a single passage in RD cells following RNA transfection of RD cells . Furthermore , we confirmed 100% sequence identity of virus stocks against original cDNA clones by direct sequencing of RT-PCR products covering the full-capsid sequence of EV71 ( GenBank accession No . AB747375 ) . Both cDNA-derived PB and non-PB strains ( 02363-EG and 02363-KE , respectively ) had similar growth kinetics in RD cells [30] . The 02363-EG strain was capable of replication in Jurkat T cells in a PSGL-1-dependent manner but 02363-KE did not replicate in T cells , confirming the PB and non-PB phenotypes [30] . We previously established a cynomolgus monkey model of acute viremia by using intravenous inoculation of EV71 to investigate the neuropathology of different EV71 strains in non-human primates [42–44] . Using this model in the present study , we compared the neuropathogenicity of EV71-02363-EG ( PB ) and EV71-02363-KE ( non-PB ) strains . As shown in Fig 1B , four monkeys were intravenously inoculated with 106 . 3 CCID50 of 02363-EG or 02363-KE strain and clinical manifestations of the monkeys were observed daily for 10 days . During the observation period , no monkeys in either group demonstrated severe clinical manifestations . In the 02363-KE-inoculated group , three out of four monkeys ( #5132 , #5133 , and #5137 ) exhibited tremor and/or ataxia in the later stages of infection ( Fig 2 ) . None of the 02363-EG-inoculated monkeys had apparent neurological manifestations . Neither exanthema ( HFMD-like symptoms ) nor pulmonary edema was observed in either group . Thus , mild neurological symptoms were found predominantly in the 02363-KE-inoculated monkeys but not in the EG-inoculated ones . Serum samples were collected at 3 , 7 , and 10 days p . i . , from inoculated monkeys to determine the serum neutralizing antibody titers against the inoculated EV71 strain . Neutralizing antibody titers were induced at 7–10 days p . i . in both groups , even in the four 02363-EG-inoculated monkeys that exhibited no apparent clinical symptoms with asymptomatic EV71 infection ( S1 Fig ) . In a cell culture system , the 02363-EG strain is capable of replication in Jurkat T cells in a PSGL-1-dependent manner whilst the 02363-KE strain is not [30] . We therefore investigated the impact of intravenous infection of 02363-EG and 02363-KE on lymphocyte subsets in infected monkeys . Peripheral blood was collected at 3 , 7 , and 10 days p . i . from each monkey and analyzed by flow cytometry to detect lymphocyte subsets: CD3+CD4+ ( CD4+ T cell ) , CD3+CD8+ ( CD8+ T cell ) , CD3-CD16+ ( NK cell ) , and CD3-CD20+ ( B lymphocyte ) cells ( Fig 3 and S2 Fig ) . The average pre-inoculation number of lymphocytes in PBMC from eight monkeys in both groups was used as an uninfected control ( Fig 3 , Day 0 ) . In four 02363-KE-inoculated monkeys , lymphocyte numbers decreased overall at 3 days p . i . , and then recovered in later stages of infection across all lymphocyte subsets demonstrating a transient lymphocytopenia at 3 days p . i . A significant difference in the number of CD3-CD16+ cells was observed between 02363-KE-inoculated and uninfected control groups 3 days p . i . ( Fig 3C ) and a trend towards decreased numbers of CD3+CD4+ ( P = 0 . 08 , Fig 3A ) CD3+CD8+ ( P = 0 . 07 , Fig 3B ) , and CD3-CD20+ ( P = 0 . 09 , Fig 3D ) , was observed in the KE-inoculated group compared to the uninfected group 3 days p . i . The number of CD3-CD16+ ( Fig 3C ) and CD3-CD20+ ( Fig 3D ) cells increased at 7 days p . i . in the 02363-KE-inoculated monkeys with a significant difference compared to uninfected controls . In contrast , the 02363-EG-infected group demonstrated almost entirely stable lymphocyte kinetics ( Fig 3 and S2 Fig ) , indicating no apparent lymphocytopenia with the exception of monkey #5134 which exhibited a decline in CD3+CD4+ , CD3+CD8+ , and CD3-CD20+ lymphocytes in the later stages of infection , particularly at 10 days p . i . ( Fig 3A , 3B , and 3D , right panel ) . Thus , transient lymphocytopenia was observed apparently in the 02363-KE-infected group during the early stages of infection but not in the 02363-EG-infected group with the exception of monkey #5134 in the later stages of infection . Increased serum levels of pro-inflammatory cytokines , IFN-γ , IL-6 , and TNFα are observed in patients with EV71-associated encephalitis and pulmonary edema compared to patients with uncomplicated HFMD [4 , 45–48] . Moreover , increased serum levels of other cytokines , including IL-1β , IL-1RA , and G-CSF , are associated with poor prognosis in EV71 infection and therefore considered possible markers of severe EV71 infections in humans [4 , 45–48] . We measured the serum cytokine levels during EV71 infection to investigate the immune response following inoculation with 02363-EG and 02363-KE strains . The average pre-inoculation serum cytokine level of all eight monkeys was used as an uninfected control ( Fig 4 , Day 0 ) . As shown in Fig 4A and 4B , and S3 Fig , a significant increase in serum levels of the pro-inflammatory cytokines IL-1β and TNF-α was observed in both groups 3–10 days p . i . A non-significant increase in serum IL-6 levels compared to control levels was observed in 02363-KE-inoculated monkeys but not in 02363-EG-inoculated monkeys ( Fig 4C and S3 Fig ) . Increased levels of other cytokines ( G-CSF , IL-1RA , and IFN-γ ) were also observed , particularly IFN-γ , in the 02363-KE-inoculated monkeys except #5132 at 7–10 days p . i . , but not in the EG-inoculated monkeys ( Figs 4D–4F ) . In general , cytokine responses were more evident in 02363-KE-inoculated monkeys than those in EG-inoculated monkeys . To monitor viral spread following intravenous inoculation with 02363-EG and 02363-KE strains , and identify primary and secondary sites for viral replication , clinical samples ( throat swabs , rectal swabs , serum , and PBMC ) were collected at 3 , 7 , and 10 days p . i . and autopsy samples obtained at 10 days p . i . were used for inoculation of RD cells . Almost all cytopathic effects ( CPEs ) appeared during the second blind passage following inoculation of RD cells except for a spleen sample from #5134 where CPE was observed after the first inoculation ( 104 . 1 CCID50/g tissue ) . As viral titers in clinical and autopsy samples were generally low , we also performed direct detection of viral RNA from clinical and tissue samples . The sensitivity of consensus-degenerate hybrid oligonucleotide primer ( CODEHOP ) RT-PCR ( semi-nested RT-PCR for enteroviruses ) [49] is higher than that for virus isolation using RD cells . All samples found to be positive by viral isolation were found to be positive by molecular detection and a number of samples found to be negative by viral isolation were found to be positive only by molecular detection ( denoted by “+” and “± , ” respectively , in Tables 1 and 2 ) . As shown in Table 1 and summarized in Fig 5 , EV71 was detected in throat swabs , rectal swabs , serum samples , and PBMC samples from all four 02363-KE-inoculated monkeys at 3 days p . i . , demonstrating efficient induction of acute viremia . In 02363-KE-inoculated monkeys , EV71 was detected in some throat and rectal swab samples , but not in serum and PBMC samples , in the later stages of infection at 7 and 10 days p . i . On the other hand , EV71 was rarely detected in throat and rectal swab samples from 02363-EG-inoculated monkeys throughout the experimental period but was detected in some serum samples mainly in the later stages of infection , except in monkey #5131 . EV71 was detected in some PBMC samples of 02363-EG-inoculated monkeys during the early stages of infection with a faint signal ( denoted by “±” ) ; however , three out of four 02363-EG-inoculated monkeys had apparently CODEHOP-RT-PCR-positive PBMC samples at 10 days p . i . In general , EV71 was frequently detected in various clinical samples from 02363-KE-inoculated monkeys during the period of acute viremia . At 10 days p . i . , EV71 was detected in some central nervous system ( CNS ) and non-CNS postmortem tissue samples in both groups , except monkey #5131 ( Table 2 , Fig 5 ) . In general , EV71 was identified at a higher frequency in the lumbar and cervical spinal cord samples in CNS tissues and in cervical lymph nodes in non-CNS tissues in both groups ( Table 2 ) . In 02363-KE-inoculated monkeys , EV71 was also detected in kidney ( 3/4 monkeys ) , dorsal root ganglion ( 3/4 monkeys ) , and trigeminal nerve ( 2/4 monkeys ) samples . In one 02363-EG-inoculated monkey #5134 , EV71 was detected mainly in non-CNS tissues , serum , and PBMC samples , suggesting a viremia peak at 10 days p . i . On the other hand , in two 02363-EG-inoculated monkeys ( #5135 and #5136 ) , EV71 was extensively detected in CNS tissues , including medulla oblongata , cerebrum , and pons ( Table 2 and Fig 5 ) . To investigate the genetic and phenotypic stabilities of PB and non-PB strains of EV71 following inoculation into monkeys , we determined the partial and/or entire VP1 sequences , which are mainly responsible for the PSGL-1-binding phenotype . To minimize selection and adaptation bias of EV71 variants during cell culture , we attempted to amplify the entire VP1 region of EV71 directly from the clinical and autopsy samples by RT-PCR without virus isolation ( Tables 1 and 2 , Fig 5 ) . When entire VP1 fragments could not be amplified from samples by RT-PCR , we applied a highly sensitive CODEHOP RT-PCR to amplify the partial VP1 region ( approximately from VP1-45 to VP1-162 ) , including amino acid residues at VP1-98 and VP1-145 . As shown in Table 1 , amino acid combinations at VP1-98 and VP1-145 , identified from clinical samples of 02363-KE-inoculated monkeys were VP1-98K and VP1-145E ( KE ) or VP1-98E and VP1-145E ( EE ) . The EE variants of EV71 were detected in serum and PBMC samples from all four 02363-KE-inoculated monkeys at 3 days p . i . ( Table 1 ) , suggesting rapid amino acid substitution of VP1-98K by VP1-98E following 02363-KE-inoculation ( Fig 5 ) . The EE variants of EV71 were frequently detected in throat and rectal samples throughout the experimental period , and also detected in CNS and non-CNS tissues at 10 days p . i . ( Table 2 ) . In postmortem tissues ( Table 1 ) , VP1-98 was heterogeneous ( K , E , or Q ) at VP1-98 and viral quasi-species was observed in a number of clinical samples ( Table 1 and S1 Table ) . However , the major determinant of the PSGL-1-nonbinding phenotype , VP1-145E , was consistently stable in the 02363-KE-inoculated monkeys , suggesting that the original KE strain and EE variants , both with a non-PB phenotype , efficiently induced acute viremia with further spread to the CNS tissues , and excretion in throat and stool samples in 02363-KE-inoculated monkeys , presumably by a PSGL-1-independent mechanism . Although there was a low detection rate in clinical samples from 02363-EG-inoculated monkeys , EE variants with a VP1-G145E substitution were detected in several serum samples from three monkeys ( Table 1 ) . The EG strain was not identified in any swab or serum samples from any of the four monkeys during the experimental period . However , EG variants were identified in PBMC samples in two 02363-EG-inoculated monkeys ( #5135 and #5136 ) by CODEHOP RT-PCR ( Table 1 , Fig 5 ) . In 02363-EG-inoculated monkeys , all EV71 variants identified from the CNS tissues possessed a VP1-G145E substitution ( EE variant , Table 2 ) . EE variants were also detected in superficial cervical lymph nodes ( 3/4 monkeys ) and extensively detected in various non-CNS tissues in one 02363-EG-infected monkey ( #5134 ) . No EG strains were detected in CNS or non-CNS tissues in the 02363-EG-infected monkeys , revealing a strong mutation/selection bias from VP1-145G to VP1-145E during the in vivo replication of the original 02363-EG strain in cynomolgus monkeys . Thus , different phenotypic variants of EV71 ( EE and EG variants ) were detected in both monkeys #5135 and #5136 at 10 days p . i . ( Fig 5 ) . No detectable amino acid substitutions or viral quasi-species were identified at VP1-98E in clinical or tissue samples from EG-infected monkeys ( Tables 1 and 2 , Fig 5 , and S1 Table ) . With the exception of two amino acids , VP1-98 in the 02363-KE-infected monkeys and VP1-145 in 02363-EG-infected monkeys , no common amino acid substitutions were identified in the VP1 region among EV71 variants from clinical and tissue samples from both groups ( Tables 1 and 2 ) . However , in some tissue and clinical samples from three 02363-KE-infected monkeys , unique amino acid substitutions ( K242Q , K242T , and K242E ) were found at a conserved lysine residue at VP1-242 , which was identified as a minor determinant of PSGL-1-binding phenotype in combination with VP1-145G [30] . Other amino acid substitutions were found at A58S and L95I in the VP1 region of EV71; however , amino acid residues in the VP1 region are generally stable except three unique amino acids at VP1-98 , VP1-145 , and VP1-242 , during in vivo replication of EV71 in monkeys . To analyze viral quasi-species with variable amino acid residues at VP1-98 and VP1-145 in EV71 variants from clinical and tissue samples , partial viral cDNA was amplified directly from the samples by RT-PCR and cloned into bacterial plasmid vectors . Independent plasmid clones were then sequenced . In four clinical samples from the 02363-KE-inoculated monkeys , VP1-98 was highly heterogenous ( K , Q , N , or E ) , suggesting high quasi-species generation and low selection at VP1-98 , but VP1-145E was identical in 30 independent clones ( S1 Table ) . Among six tissues samples examined ( two from 02363-KE-inoculated and four from 02363-EG-inoculated monkeys ) , which were identified as EE variants by direct VP1 sequencing , all of the 44 plasmid clones were found to possess VP1-98E ( GAA ) and VP1-145E ( GAG ) sequences , representing no apparent quasi-species variants ( S1 Table ) . As expected , in a PBMC sample from a 02363-EG-inoculated monkey , #5136 , ( 10 days p . i . ) , an amino acid residue at VP1-145 was heterogenous and VP1-145G was dominant ( eight out of nine clones ) . The results suggest that the rate of quasi-species generation and selection at VP1-98 and VP1-145 was varied in different clinical and tissue samples . The CNS tissue samples from the 02363-KE- or 02363-EG-inoculated monkeys were histopathologically examined to compare the neuropathogenesis of EV71 strains . Inflammation and neuronal degeneration were extensively found in different CNS tissues from all four 02363-KE-inoculated monkeys at 10 days p . i . ( Fig 6A ) . Inflammation in the spinal cord and gray matter of cerebrum ( Fig 6A ) and the induction of lymphocytes in cerebrospinal fluid ( CSF ) samples ( S2 Table , S4 Fig ) indicate myelitis , encephalitis , and meningitis occurred in three monkeys , #5061 , #5133 , and #5137 at 10 days p . i . Neuronal degeneration was observed in the spinal cord anterior horn of two monkeys , #5061 and #5133 ( Fig 6A ) . Inflammation was observed in the brainstem of one monkey #5132 . Restriction of inflammation to the spinal cord gray matter and a decline the number of lymphocytes present in CSF samples between 7 days p . i . and 10 days p . i . , suggested monkey #5132 might have a peak in meningoencephalitis at 7 days p . i . , and was presumably within the convalescent phase at 10 days p . i . ( Fig 6A and S2 Table ) . Inflammation and neuronal degeneration were observed in two 02363-EG-inoculated monkeys , #5135 and #5136 . Although lymphocytes were failed to investigate in some CSF samples due to blood contamination ( S2 Table ) , these monkeys had histopathological evidence of myelitis , encephalitis , and meningitis at 10 days p . i . ( Fig 6B ) . However , other two monkeys , #5131 and #5134 , had no apparent CNS lesions ( Fig 6B ) . Histopathological results in 02363-EG-inoculated monkeys were mostly consistent with the distribution of viral infection , shown in Table 2 . There was no significant difference in typical neuronal lesions of the spinal cord between 02363-KE- and 02363-EG-inoculated monkeys; there was mild neurodegeneration with perivascular mononuclear cuffing and central chromatolysis and gliosis of the anterior horn of the spinal cord in both groups ( Fig 7 ) . Taken together with virus detection ( Table 2 ) and histopathological ( Fig 6 ) analyses of the CNS tissues , all four KE-inoculated monkeys were likely in the late or convalescent stages of meningoencephalitis at 10 days p . i . Two out of four 02363-EG-inoculated monkeys ( #5135 and #5136 ) demonstrated a peak of meningoencephalitis and monkey #5134 developed viremia without apparent CNS symptoms at 10 days p . i . In conclusion , histopathological analysis of CNS tissues at 10 days postinfection revealed that 02363-KE induced neuropathogenesis ( inflammation and neuronal damage ) more efficiently than that induced by 02363-EG .
To elucidate the in vivo effects of PSGL-1-dependent replication and pathogenesis , in this study , we used a cynomolgus monkey model to assess viral replication , pathogenicity , and the genetic stability of PB and non-PB strains of EV71 . Mouse models have been extensively used to study the pathogenesis of EV71 infection . However , amino acid identity between human and murine PSGL-1 molecules is not high , particularly in the N-terminal region of PSGL-1 , which is responsible for the binding of human PSGL-1 to EV71 . Accordingly , a PB strain of EV71-1095 was shown not to bind to mouse PSGL-1 [16] . Therefore , to investigate PSGL-1-dependent viral replication and pathogenesis , mouse models would be unreliable as the PB strains of EV71 may not replicate in mice in a PSGL-1-dependent manner . On the other hand , the amino acid identity of the EV71-binding region of PSGL-1 ( amino acids 42–61; QATEYEYLDYDFLPETEPPE ) between human and cynomolgus monkey PSGL-1 molecules is 100% , thus indicating that a monkey model , more closely resembled the human , would be more reliable than mouse models in investigating PSGL-1-dependent EV71 replication . In our previous studies , cynomolgus monkeys infected via intravenous inoculation with EV71 demonstrated similar neurological manifestations to those observed in humans including paralysis , ataxia , tremor , and development of CNS lesions in the cerebrum , medulla oblongata , and spinal cord [42–44] . In addition , this monkey model allows the identification of EV71 variants in various clinical and tissue samples , including CNS and non-CNS tissues , to monitor viral spread and genetic stability of EV71 in infected individuals in more detail than smaller mouse models . After the intravenous inoculation with the 02363-KE ( non-PB ) strain , KE or EE variants were detected in clinical samples from different locations ( throat and rectal swabs , serum , and PBMC ) in all four monkeys at 3 days p . i . , demonstrated acute viremia associated with transient lymphocytopenia ( Fig 3 ) and cytokine induction ( Fig 4 ) in the early stage of infection . In the later stages of infection at 10 days p . i . , KE or EE variants were detected from lumbar and cervical spinal cord samples of 02363-KE-inoculated monkeys ( Fig 5 ) that developed typical neuropathogenesis ( inflammation and neuronal damage ) ( Fig 6 ) . Thus , the KE or EE variants efficiently induced viremia and CNS involvement with apparent neurological manifestations ( Fig 2 ) , presumably in a PSGL-1-independent manner as a major determinant of the non-PB phenotype , VP1-145E , was genetically stable during the in vivo replication and spread to CNS and non-CNS tissues in 02363-KE-inoculated monkeys ( Fig 5 , Table 2 , and S1 Table ) . The emergence of EE variants with a substitution of glycine from glutamic acid at VP1-145 in serum samples ( Table 1 ) and their wide distribution throughout CNS and non-CNS tissues in 02363-EG-infected monkeys ( Table 2 ) again suggested EE variants may confer better in vivo fitness than the inoculated EG strain during the acute viremia phase . Following viremia , resultant EE variants were mainly responsible for the CNS involvement in monkeys #5135 and #5136 in a PSGL-1-independent manner . SCARB2 was histopathologically identified in a number of CNS and non-CNS tissues , including EV71 antigen-positive neuronal cells in human encephalitis cases [50] , and neuronal cells in EV71-infected SCARB2-transgenic mice [25] . Therefore , SCARB2-expressing neuronal cells in the CNS may play a critical role in direct CNS involvement during EV71 infection in a PSGL-1-independent manner [13 , 51] . It is likely that some time is required for mutation and selection of EE variants following 02363-EG-inoculation , possibly resulting in slower development of pathogenesis in the 02363-EG-inoculated group than the 02363-KE-inoculated group . Mutation and selection of EE variants may also contribute to individual variability in the pathogenesis observed in the four EG-inoculated monkeys at 10 days p . i . ; inefficient infection in monkey #5131 , acute viremia in monkey #5134 , and meningoencephalitis in monkeys #5135 and #5136 ( Figs 5 and 6 ) . In monkey #5134 , EE variants were detected in some clinical samples , various non-CNS tissues , and some CNS tissues ( lumbar and cervical spinal cord ) only in the later stages of infection ( Tables 1 and 2 ) . At the same time , this monkey developed typical lymphocytopenia at 10 days p . i . ( Fig 3 ) , observed in four 02363-KE-inoculated monkeys at 3 days p . i . , consistent with slower development of viremia in monkey #5134 than in 02363-KE-inoculated monkeys . EG variants were detected in PBMC samples from two 02363-EG-inoculated monkeys ( #5135 and #5136 ) during the meningoencephalitis phase at 10 days p . i . ( Table 1 , Fig 5 ) indicating possible involvement of PSGL-1-dependent EV71 replication in local , specific immune cell types . Viral quasi-species variants with VP1-145G and VP1-145E were identified in a PBMC sample from monkey #5136 ( S1 Table ) . It is uncertain whether the inoculated 02363-EG strain was survived in minor PSGL-1-positive immune cells without a VP1-G145E substitution or EG variants were further mutated from circulating EE variants by acquiring viral fitness in the local cell microenvironments in the later stages of infection . We cannot exclude the possibility that additional mutations outside of the capsid VP1 region may compensate for the in vivo fitness of EG variants . However , we were unable to isolate infectious EG variants in cell culture , and detected viral RNA only by RT-PCR directly from PBMC samples , presumably due to low virus titers in the samples ( Table 1 ) . Recent studies have demonstrated that rhesus monkeys exhibit neuropathogenesis following the EV71 infection [52] and CD14-positive lymphocytes in peripheral blood are responsible for the efficient EV71 replication [53] . Therefore , to confirm replication competency of 02363-EG and 02363-KE strains in monkey PBMCs and CD14-positive lymphocytes , we isolated PBMCs and CD14-positive lymphocytes from four healthy monkeys and infected them with the EV71 strains ( 1 CCID50/cell ) . No apparent virus replication was observed in the monkey PBMCs or CD14-positive lymphocytes with either of the strains ( S5 Fig ) in-keeping with results from our previous study using human PBMCs [54] . Thus , we have yet to identify the specific immune cell population in peripheral blood responsible for PSGL-1-dependent replication of EV71 . In this study , we demonstrated potent in vivo selection of VP1-145E variants in a cynomolgus monkey model . Similar in vivo mutation and selection from G/Q to E at VP1-145 has been reported in several mouse models using mouse-adapted EV71 strains [34–37] and VP1-145E has been identified as a critical virulence determinant in mice alone or in combination with other amino acids . As mentioned above regarding the structural differences between human and mouse PSGL-1 , it is unlikely that a phenotypic change from PB to non-PB by VP1-145 substitution also contributes to the improved mouse PSGL-1-binding of VP1-145E mutants . Instead , VP1-145E may be responsible for increased virulence in mice thorough unknown mechanisms , such as the involvement of the other receptors and/or cellular proteins in viral entry , uncoating , or host immune responses to EV71 , regardless of the PSGL-1-binding capability of EV71 . We speculate the same mechanism underlies in vivo fitness and virulence associated with VP1-145E in both mouse and monkey models , however , the viral and cellular factors determining in vivo phenotypes of EV71 have yet to be elucidated . One possible explanation for the rapid selection of EV71 variants with amino acid substitutions at VP1-98 or VP1-145 is that escape mutants were selected under immunological pressure as VP1-98 and VP1-145 are located at possible neutralizing epitopes of EV71 [55] . However , the EE variants with substitution of K by E at VP1-98 were frequently identified in the serum samples from 02363-KE-inoculated monkeys , even at 3 days p . i . before the induction of neutralizing antibodies ( Table 1 , S1 Fig ) . Moreover , cross-neutralizing antibodies against both strains were induced in both groups ( S6 Fig ) , suggesting neutralization escape is not the most critical factor for rapid selection of EV71 variants in this monkey model . We have not established an efficient oral infection model in non-human primates [56]; therefore , the route of virus inoculation in this study ( intravenous inoculation ) does not reflect the natural EV71 infection in humans . After intravenous inoculation with 02363-KE , following viremia , infectious KE and EE variants were excreted in throat and rectal swab samples . EE variants are commonest among EV71 isolates available in GenBank; therefore , excreted EE variants in the monkeys may be transmissible . However , we cannot exclude the possible involvement of PSGL-1-dependent replication following the primary mucosal infection of EV71 as various PSGL-1-expressing leukocytes are located in pharyngeal and intestinal mucosal tissues [27 , 57] . In this monkey model several typical clinical manifestations associated with EV71 infections in humans , including HFMD-like skin and oral rashes ( mild local inflammation ) , polio-like paralysis , and fatal pulmonary edema ( more severe clinical manifestations of CNS disease ) , were not observed . Moreover , we have not investigated the in vivo phenotypes of other PB strains such as the 02363-EQ strain [30] . Thus , we were unable to investigate all aspects of PSGL-1-dependent and -independent pathogenesis in EV71 infections in this cynomolgus monkey model . To our knowledge , this is the first comprehensive study to describe the evolutionary dynamics and neuropathogenesis of EV71 within infected individuals in a non-human primate model . In a recent study , Cordey et al . reported virus detection and characterization in several different clinical samples from an EV71-infected immunocompromised patient and identified an amino acid at VP1-97 as a potential determinant of host adaptation and neurovirulence in humans by comparing full-length genome sequences and in vitro phenotypes of EV71 variants [58] . In the present study , we used cDNA-derived EV71 strains to minimize in vitro quasi-species and examined the dynamics of viral distribution in clinical and tissue samples , genetic stabilities , viral fitness , and the neuropathogenesis in various CNS tissues in EV71-infected monkeys . We identified two amino acids in the VP1 region , VP1-98 and VP1-145 , as variable sites during the in vivo viral replication in the monkeys , which may be associated with the apparent amino acid polymorphism observed among the field EV71 isolates . Moreover , unique amino acid substitutions were identified at VP1-242 in some clinical and tissue samples from 02363-KE-inoculated monkeys ( Tables 1 and 2 ) . Among them , VP1-145 is a major determinant of neuroinvasion to the CNS and neuropathogenesis as all EV71 variants detected in CNS tissues consistently possessed VP1-145E in both 02363-KE- and 02363-EG-inoculated groups ( Table 2 and Fig 5 ) . On the other hand , amino acid residues among EV71 variants from CNS tissues were variable and heterogenous at VP1-98 ( K , E , or Q ) and VP1-242 ( K , Q , or T ) ( Tables 1 and 2 , S1 Table ) . Recent structural analyses of EV71 have revealed VP1-98 , VP1-145 , and VP1-242 are located within surface loops ( BC , DE , and HI , respectively ) surrounding the center of the 5-fold mesa of EV71 virions [2 , 59] . These three amino acid residues have previously been identified as critical determinants of several distinct EV71 phenotypes , including PSGL-1-binding ( VP1-145 and VP1-242 ) [30] , putative heparan sulfate-binding ( VP1-242 ) [17] , neutralizing antibody-binding ( VP1-98 , VP1-145 , and VP1-242 ) [55] , adaptation and virulence in mouse models ( VP1-145 ) [34–37] , and positive selection in humans ( VP1-98 and VP1-145 ) [31–33] . In addition , other amino acids , adjacent to these three amino acids , are also involved in certain EV71 phenotypes alone or in combination with these three amino acids; VP1-97 as a possible virulence determinant in humans [58] , VP1-243 as a cyclophilin A-binding site [60] , and VP1-244 as a determinant of PSGL-1-binding [30] and mouse adaptation/virulence [61] . A growing number of molecular epidemiological studies have demonstrated that some of the critical residues in the VP1 region are variable ( VP1-98K/E/Q/N/G and VP1-145E/G/Q/A/K/R ) but that others are apparently conserved ( VP1-97L , VP-242K , VP1-243S , and VP1-244K ) among EV71 isolates . Recent studies have also suggested VP1-145G/Q isolates were more frequently detected from cases with severe neurological disease in humans than VP1-145E isolates [38–41] . As we have demonstrated for the PSGL-1-binding phenotype [30] , combinations and mutations in several critical amino acid residues , including VP1-98 , VP1-145 , and VP1-242 , may be responsible for the distinct in vitro and in vivo phenotypes of EV71 and possibly associated with the overall pathogenesis in humans . In this study , we examined the in vivo evolutionary dynamics and neuropathogenesis of PB and non-PB strains of EV71 in a non-human primate model . We demonstrated potent in vivo selection of VP1-145E variants . VP1-145E was identified as a major determinant of viremia and neuropathogenesis , and may play a key role in the in vivo fitness of EV71 . On the other hand , VP1-145G variants were identified only in PBMCs as viral quasi-species , indicating the possible involvement of PSGL-1-dependent EV71 replication in the local cell microenvironment . In this regard , our study provides new insights into the interplay between virus ( tissue- and cell-tropic EV71 variants ) , receptors ( PSGL-1 , SCARB2 , and other EV71 binding molecules ) , and host ( immune responses and pathogenesis ) in EV71-infected individuals .
Animal experiments were conducted in compliance with Japanese legislation ( Act on Welfare and Management of Animals , 1973 , revised in 2012 ) and guidelines under the jurisdiction of the Ministry of Education , Culture , Sports , Science and Technology , Japan ( Fundamental Guidelines for Proper Conduct of Animal Experiment and Related Activities in Academic Research Institutions , 2006 ) . Animal care , housing , feeding , sampling , observation , and environmental enrichment were performed in accordance with the guidelines . Every possible effort was made to minimize animal suffering . The protocols of animal experiments were approved by the committee of biosafety and animal handling and the committee of ethical regulation of the National Institute of Infectious Diseases , Japan ( Authorization Number; 512001 ) . Each animal was housed in a separated cage , received standard primate feed and fresh fruit daily , and had free access to water at the National Institute of Infectious Diseases , Japan , and animal welfare was observed on a daily basis . Sampling procedures were conducted under anesthesia ( 10 mg/kg ketamine intramuscularly ) . Animals were sacrificed under excess anesthesia . Jurkat T cells were obtained from the Riken Cell Bank and cultured in RPMI-1640 medium ( Sigma , St . Louis , MO ) supplemented with 10% fetal bovine serum ( FBS; CCB Nichirei Bioscience , Japan ) . RD cells were obtained from the US Centers for Disease Control and maintained in MEM ( Sigma , St . Louis , MO ) supplemented with 10% FBS . We used the genomic RNA of the 02363 strain of EV71 ( non-PB; subgenogroup C1 ) as the template for RT-PCR . Infectious viral cDNA clones were constructed and cDNA-derived mutant viruses were prepared as previously described [30] . The RNA-transfected cells and supernatants were freeze-thawed thrice at 24 h post-transfection . Before use in experiments , recovered viruses were amplified once in fresh RD cells , and the sequence of the whole capsid region was confirmed by direct sequencing of RT-PCR products . Sequence identity between the virus stock and original cDNA was confirmed . Viral titers were determined by a microtitration assay using 96-well plates and RD cells , as previously described [62] . Briefly , 10 wells were used for each viral dilution and viral titers were expressed as 50% cell culture infectious dose ( CCID50 ) . Eight male cynomolgus macaques ( Macaca fascicularis ) imported from the Philippines were used in this study . Seven out of eight monkeys were 4 years old and the other was 6 years old ( #5061 ) . The average weight was 3 . 0 kg ( range , 2 . 4 to 4 . 3 kg ) . All procedures were performed in a biosafety level 2 containment facility . Serological testing revealed all animals used in this study were free from infection with EV71 and poliovirus . Eight monkeys were divided into two groups; one group received EV71-02363-EG ( PB ) inoculations ( monkeys No . #5131 , #5134 , #5135 , and #5136 ) and the other group received EV71-02363-KE ( non-PB ) inoculations ( monkey No . #5061 , #5132 , #5133 , and #5137 ) . Virus inoculation and sampling procedures were conducted under anesthesia ( 10 mg/kg ketamine intramuscularly ) . One ml of EV71 virus solution containing either 106 . 3 CCID50 of 02363-KE ( non-PB ) or 02363-EG ( PB ) strain was intravenously inoculated into the right tibial vein . Following inoculation , monkeys were observed daily for 10 days to assess neurological manifestations . Clinical manifestations were observed in more detail under anesthesia , including blisters within the oral cavity , palm , and sole of the foot , at days 0 , 3 , 7 and 10 p . i . Whole blood , throat , and rectal swab samples were collected at these times . CSF samples were collected at 7 and 10 days p . i . Blood-contaminated CSF samples were excluded from the analysis ( S2 Table ) . Isolation of PBMC and serum from whole blood was carried out within one day of collection . Monkeys were sacrificed under deep anesthesia at 10 days p . i . and various parts of the CNS , main organs , lymph nodes , swabs and blood samples were collected for histopathological and virological analyses . Ten percent [wt/vol] homogenates of collected tissues were prepared in Eagle’s minimal essential medium ( MEM ) containing 2% FBS with MagNA Lyzer ( Roche , Basel , CH ) before centrifugation at 10 , 000 × g for 5 min at 4°C to remove tissue debris . The supernatants were used for virus isolation in RD cells and RNA extraction for molecular detection of EV71 . To investigate the genetic stability of EV71 strains following inoculation , we determined the entire and/or partial VP1 sequence of EV71 from clinical and postmortem samples . To minimize selection and adaptation during cell culture , the entire VP1 region of EV71 was directly amplified from the samples by RT-PCR without virus isolation . Briefly , before RNA extraction , swab samples were centrifuged at 2000 rpm for 3 min and the supernatants were used for RNA extraction . Viral genomic RNA was extracted from 10% tissue homogenates of autopsy or clinical samples using HighPure viral RNA purification kit ( Roche , Penzberg , DE ) . Reverse transcription-PCR ( RT-PCR ) was performed using PrimeScript II High Fidelity RT-PCR Kits ( Takara , Japan ) . PCR products were purified using Wizard SV Gel and PCR Clean-Up System PCR purification kits ( Promega , Madison , USA ) . The full length of the VP1 sequence was analyzed using the following primers; VP1-1: 5′-TAATAGCACTAGCGGCAGCC-3′ , VP1-2: 5′-AGCTGAGACCACTCTCGATAG-3′ , VP1-3R: 5′-TGGGGTATCCGTCATAGAACC-3′ VP1-4R: 5′-TGGTGGATGACACGAGCAAG-3′ The full length of the VP1 sequence was amplified as a series of overlapping fragments using primer sets VP1-1/VP1-3R ( 714 nt ) and VP1-2/VP1-4R ( 803 nt ) . All primers were used in sequencing analysis . When the entire VP1 fragment was not amplified by RT-PCR directly from samples , a highly sensitive CODEHOP RT-PCR was used to amplify the partial VP1 region [49] . Briefly , cDNA was synthesized from total RNA using four primers ( AN32–AN35 ) . The partial VP1 region was amplified by semi-nested PCR: primers 222 and 224 were used for the first-round PCR and primers AN88 and AN89 for the second amplification . DNA sequencing was performed using BigDye Terminator v3 . 0 cycle sequencing ready reaction kits ( Applied Biosystems , Foster city , CA ) and results were analyzed using an ABI 3130 genetic analyzer ( Applied Biosystems , Foster City , CA ) and Sequencher ver . 5 . 0 . 1 sequence analysis software ( Gene Codes Corporation , Ann Arbor , MI ) . To assess quasi-species at VP1-98 and VP-145 , partial capsid cDNA ( about 700 nt covering VP3-214 to VP1-210 ) was amplified by VP1-1 and VP1-3R primers directly from tissue or clinical samples and cloned into a pCR-TOPO vector using TOPO-TA cloning kit ( Life technologies , Carlsbad , CA ) according to the manufacturer’s instructions . Resultant colonies were randomly selected and plasmids were purified using QIAprep Miniprep kits ( QIAGEN , Hilden , Germany ) . In total , forty-four independent plasmids carrying viral cDNA from 7 postmortem samples ( 3 CNS , 3 non-CNS , and 1 PBMC samples ) from 02363-KE- or 02363-EG-inoculated monkeys were sequenced . Similarly , thirty independent plasmids derived from 4 clinical samples ( 2 rectal and 2 throat swab samples ) from two 02363-KE-inoculated monkeys were analyzed . Viral quasi-species variants were identified by nucleotide sequence alignment of plasmid clones derived from each clinical or tissue sample . For virus isolation , RD cells inoculated with tissue homogenates or clinical sample preparations were observed for cytopathic effect ( CPE ) for one week and then blind passage was performed for CPE-negative samples after freeze-thawing of the first-round passage . If no CPE were observed during first and second round cultures , the result of virus isolation was recorded as negative . Serum samples were diluted serially two-fold ( 1:4–1:4096 ) in MEM containing 2% FBS . The EV71 stock was diluted to a concentration of 100 CCID50/50 μl . The 50 μl of diluted serum was mixed with 50 μl of EV71 solution on 96-well plates in duplicate and incubated for 2 h at 35°C . Following incubation , 100 μl of the cell suspension containing 104 RD cells was added to the wells followed by incubation at 35°C with 5% CO2 for 7 days to observe CPE . Neutralization titers were determined as the highest serum dilutions that could completely protect from CPE . Neutralization titers against the 02363-KE strain were difficult to determine , presumably due to non-specific viral aggregation . Therefore , to minimize aggregation , virus solutions were pre-treated with an equal volume of chloroform , shook vigorously for 20 min , and then centrifuged at 10 , 000 × g for 10 min at 4°C . Supernatants were used as the challenge virus for the neutralization assay . Cross neutralization titers were measured using serum samples collected at 10 days p . i . PBMCs were isolated from 5 ml of whole blood from cynomolgus monkeys mixed with EDTA for anti-coagulation by Percoll density-gradient centrifugation . Briefly , whole blood was diluted with PBS and carefully layered over 5 ml of Percoll ( specific gravity 1 . 070 g/ml ) in 15 ml conical tubes . Tubes were centrifuged at 400 × g for 30 min at 20°C in a swinging-bucket rotor . Mononuclear cell layers were carefully transferred to a new tube , mixed with PBS , and centrifuged at 300 × g for 10 min at 20°C before complete removal of supernatants . To remove platelets , cells were resuspended in PBS and centrifuged at 200 × g for 10 min at 20°C before complete removal of supernatants . This procedure was repeated once . Precipitated cells were used for RNA extraction , infection assays , and CD14-positive cell isolation . CD14-positive cells were isolated from freshly isolated PBMC using CD14 microbeads and MACS MS columns according to the manufacturer’s instructions ( Miltenyi Biotec , Bergisch Gladbach , Germany ) . Isolated CD14-positive cells were cultured in RPMI-1640 with 10% FBS . Monkey PBMC and CD14-positive cells were inoculated with 02363-KE or 02363-EG strain at 1 CCID50/cell and incubated at 35°C with 5% CO2 for 2 h . After incubation , cells were washed with RPMI-1640 without FBS twice and then resuspended with RPMI-1640 containing 10% FBS . Cells were cultured in 96-well plates at 35°C with 5% CO2 . CD14-positive cells were harvested at 0 , 1 and 3 days p . i . PBMCs were harvested at 0 , 1 , 3 and 7 days after infection . Cells and supernatants were collected together for RNA extraction . To monitor EV71 viral replication in PBMC and CD14-positive cells , viral genomic RNA in each cell preparation was measured by real-time PCR using the following primers: EnteroFw , 5′-GCCCCTGAATGCGGCTAAT-3′ and EnteroRev , 5′-ATTGTCACCATAAGCAGCCA-3′ , as described previously [63] . Brain , spinal cord , lung , heart , liver , spleen , kidney , lymph nodes , tonsil , and gastrointestinal tract samples were fixed in 10% formalin in PBS and embedded in paraffin . Paraffin-embedded sections were stained with hematoxylin and eosin ( H&E ) . Immunohistochemical detection of the EV71 antigen was performed on paraffin-embedded sections using a labeled streptavidin-biotin method ( Dako Denmark A/S , Glostrup , Denmark ) . For antigen retrieval , sections were heated to 121°C for 10 min in an autoclave with 10 mM citrate buffer solution ( pH 6 . 0 ) . A polyclonal antibody raised against denatured virus particles from the 1095/Japan 97 isolate of EV71 was used as the primary antibody [44] . Peroxidase activity was detected with 3 , 3-diaminobenzidine ( Sigma-Aldrich , St . Louis , MO ) and sections were counterstained with hematoxylin . Lymphocytes in CSF were detected by VetScan HM2 ( ABAXIS , Union City , CA ) according to manufacturer’s instructions . One hundred μl of each CSF sample was used for Giemsa staining as following: cells were attached to glass slides using a Shandon cytocentrifuge ( Thermo Fisher Scientific Inc . , Waltham , MA ) at 1 , 000 rpm for 10 min and then stained with Giemsa’s Azure Eosin Methylene Blue solution ( Merck Millipore , Billerica , MA ) . Stained cells were analyzed by microscopy . The number of lymphocytes in peripheral blood samples from infected monkeys was calculated by flow cytometry . Both CD3+CD4+ and CD3+CD8+ T cells were stained using NHP T Lymphocyte Cocktail ( BD Pharmingen , San Diego , CA ) and CD3-CD16+ NK cells and CD3-CD20+ B cells were stained using NHP T/B/NK Cell Cocktail ( BD Pharmingen , San Diego , CA ) . Analyses were performed using a BD FACSCanto II Flow Cytometer . Serum cytokine concentrations were analyzed using Cytokine Monkey Magnetic 28-Plex Panels ( Life technologies , Carlsbad , CA ) on a Luminex 200 ( Luminex , Austin , TX ) . We carried out flow cytometric ( Fig 3 and S3 Fig ) and cytokine/chemokine ( Fig 4 and S4 Fig ) assays in triplicate and compared mean values using Student’s t-test . P < 0 . 05 were considered statistically significant . | Recently , large outbreaks of hand , foot , and mouth disease , including fatal neurological cases in young children primarily because of enterovirus 71 ( EV71 ) have been reported , particularly in the Asia Pacific regions where the disease poses a serious threat to public health . Based on mutational and structural analyses of EV71 , we identified amino acid residue 145 of the capsid protein VP1 ( VP1-145 ) as a critical molecular determinant for the binding of EV71 to a specific cellular receptor , human P-selectin glycoprotein ligand-1 ( PSGL-1 ) . VP1-145 is highly variable among EV71 isolates and has been identified as a potential neurovirulence determinant in humans and experimental mouse models . To elucidate the in vivo involvement of PSGL-1-depentent replication and pathogenesis , we investigated viral replication , genetic stability , and the pathogenicity of the PSGL-1-binding ( PB ) and PSGL-1-nonbinding ( non-PB ) strains of EV71 in a cynomolgus monkey model . After the intravenous inoculation with the PB strain , viruses found to be highly mutated at VP1-145 with resultant VP1-145E variants ( non-PB ) inducing viremia and neuropathogenesis , presumably in a PSGL-1-independent manner . VP1-145G variants were identified only in peripheral blood mononuclear cells from two PB-inoculated monkeys . Our study provides new insights into the interplay between virus , receptors , and host in EV71-infected individuals . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | The Role of VP1 Amino Acid Residue 145 of Enterovirus 71 in Viral Fitness and Pathogenesis in a Cynomolgus Monkey Model |
Massive technological advances enabled high-throughput measurements of proteomic changes in biological processes . However , retrieving biological insights from large-scale protein dynamics data remains a challenging task . Here we used the mating differentiation in yeast Saccharomyces cerevisiae as a model and developed integrated experimental and computational approaches to analyze the proteomic dynamics during the process of cell fate determination . When exposed to a high dose of mating pheromone , the yeast cell undergoes growth arrest and forms a shmoo-like morphology; however , at intermediate doses , chemotropic elongated growth is initialized . To understand the gene regulatory networks that control this differentiation switch , we employed a high-throughput microfluidic imaging system that allows real-time and simultaneous measurements of cell growth and protein expression . Using kinetic modeling of protein dynamics , we classified the stimulus-dependent changes in protein abundance into two sources: global changes due to physiological alterations and gene-specific changes . A quantitative framework was proposed to decouple gene-specific regulatory modes from the growth-dependent global modulation of protein abundance . Based on the temporal patterns of gene-specific regulation , we established the network architectures underlying distinct cell fates using a reverse engineering method and uncovered the dose-dependent rewiring of gene regulatory network during mating differentiation . Furthermore , our results suggested a potential crosstalk between the pheromone response pathway and the target of rapamycin ( TOR ) -regulated ribosomal biogenesis pathway , which might underlie a cell differentiation switch in yeast mating response . In summary , our modeling approach addresses the distinct impacts of the global and gene-specific regulation on the control of protein dynamics and provides new insights into the mechanisms of cell fate determination . We anticipate that our integrated experimental and modeling strategies could be widely applicable to other biological systems .
Retrieving gene regulatory networks from experimental measurements lies at the foundation for deciphering the mechanistic basis of cellular responses . To date , several strategies have been proposed to reconstruct biological networks . It is possible to infer network connectivity directly from genome-wide localization analysis , which takes advantages of high-throughput techniques to identify genomic sites bound by transcription factors ( TFs ) [1–4] . However , uncovering the physical interactions is insufficient to bring insight into the underlying regulatory mechanisms and recapitulate the dynamics of the system . Another strategy makes use of the simultaneous measurements of network elements and requires reverse engineering methods to deduce the network structure . A plethora of algorithms have been proposed to reconstruct gene regulatory networks in different organisms , and their performance has been quantitatively assessed [5–8] . Well-established methods include statistical methods based on correlation and mutual information [9 , 10] , ordinary differential equation ( ODE ) model [11] , Bayesian networks [12] and Boolean network models [13 , 14] . Prior knowledge about the organization of the biological network can be further incorporated into the workflow to facilitate the reverse engineering process [15 , 16] . Despite these research achievements , several challenges exist in the reconstruction of biological networks . Gene expression profiles are widely used to retrieve transcriptional regulatory networks [8] with the implicit assumption that the activity of a TF is proportional to its mRNA level . However , the expression level of TFs is also subject to post-transcriptional regulations . Earlier studies employing simultaneous measurements of the transcriptome and proteome showed substantial differences between the mRNA and protein abundance either at the population level or the single-cell level [17–19] . On the other hand , although proteomic data provides a more reliable estimation of gene activity , it is not a good indicator of gene regulatory events . Physiological changes that involve global variations in ribosome number , metabolite concentration and growth rate can also affect protein synthesis and dilution , contributing to a layer of regulation that is independent of gene-gene interactions [18 , 20–22] . This is especially important for investigating the gene regulatory networks underlying cell fate switches , in which distinct cell fates are often associated with very different physiological parameters ( such as growth rate and biogenesis ) . Recently , several analyses applied dynamic modeling of protein life cycles to characterize the effect of different factors on the variations in protein abundance [23 , 24] , and their results indicated the critical role of kinetic modeling in decoupling the influence from different levels of regulation . In this work , we incorporated high-resolution gene expression and cell growth profiling into kinetic modeling to study the cell fate determination in yeast mating response . The yeast mating response pathway is among the best-characterized models in the study of signal transduction , in which the external signal is transmitted through a mitogen-activated protein kinase ( MAPK ) cascade . This signal eventually activates transcription factor Ste12 , which initiates downstream gene regulatory programs ( Fig 1A ) . Despite a wealth of detailed information revealed by past studies [25] , a less-studied perspective of the mating response is the cell fate decision governed by changes in the pheromone concentration [26–28] . While growth arrest and shmoo-like morphology is triggered when cells sense a high concentration of pheromone , yeast cells initiate a chemotrophic elongated growth along the pheromone gradient in response to a lower dose of pheromone . Due to the complexity of gene expression programs induced by pheromone stimulation , the mechanism underlying the mating differentiation switch remains elusive . Therefore , it represents a unique opportunity for quantitative investigations into whether and how divergent gene expression networks leading to distinct cell fates can be stimulated in a dose-dependent way . Through kinetic modeling of protein abundance , we found that the observed protein dynamics in the mating response were partially determined by changes in the physiological state of the cell . Therefore , a model-driven framework named protein synthesis decoupling analysis ( PSDA ) was developed to decouple gene-specific regulation information from the influence of global physiological regulation . Based on the temporal order of gene-specific changes from PSDA , the putative regulatory networks were then reconstructed using a Boolean network model [14 , 29] . These model analysis revealed network rewiring during cell differentiation and suggested a pheromone-dependent regulation of the TOR signaling pathway [30] . In summary , our results highlight the importance of considering the global physiological effects on gene expression control and provide mechanistic insights into the cell fate decisions triggered by different doses of pheromone .
To quantify the effect of physiological constraints on protein dynamic changes and reconstruct the gene regulatory networks in the yeast mating response , we developed a high-throughput microfluidic device that features a throughput of 96 experiments in one single run , continuous control of the medium and an automatic image processing pipeline . The system allows for simultaneous measurements of cell mass accumulation and protein expression level ( Fig 1B ) . We used our platform to track the expression of ~200 fluorescently tagged genes as well as the growth dynamics of yeast cells in response to high and intermediate levels of pheromone ( 0 . 59 μM and 5 . 9 μM ) . These data offered a comprehensive view of the downstream gene regulatory response with unprecedented temporal resolution . In our experiment , the yeast strains from a green fluorescent protein ( GFP ) -tagged library [31] with BAR1+ background were cultured and loaded into a 96-well microfluidic device ( S1A Fig ) , in which each strain was confined within the observation region of an insulated chamber for time-resolved analysis . The chemical condition of the chamber was controlled in an accurate and continuous way , and the concentration of pheromone was set to a high or intermediate level about 1 h after cell loading . In each experiment , phase contrast and fluorescence images of cells under pheromone stimulation were acquired at an interval of 5-min for 6 hours , producing >20000 images with single-cell resolution from a single chip . An image processing pipeline was adopted to automatically track the fluorescence intensity of each cell and estimate the growth rate via quantification of the accumulation of cell mass . To study the underlying mechanism of mating differentiation switching , we focused on a set of 195 selected genes including 79 Ste12 regulated genes [3] , 52 transcriptional factors that are critical in the regulation of protein expression , and 64 manually selected genes that are representative of different functional groups relating to mating response . We monitored the protein expression and growth dynamics of the selected yeast strains from the GFP library in response to two different concentrations of pheromone ( first two columns of Fig 2A and 2B , S1 Dataset ) . Consistent with previous studies [26 , 27] , different cell fate responses , characterized by distinct morphological changes were induced depending on pheromone levels ( S1B Fig ) . When exposed to a high concentration of pheromone , cells underwent a sudden cell cycle arrest within 60 min and formed multiple short projections . Intriguingly , the protein abundances of most examined genes were first up-regulated and then relaxed to their original levels towards the end of the experiment ( Fig 2A , left column ) . These results exhibit a substantial deviation from the transcriptional regulation reported before , because previous microarray data showed that a reduction in the transcript levels of at least 48% of examined genes under the same condition [32] . In addition , we observed the growth rate of shmooing cells underwent a sudden arrest , falling to about half of its normal value within 60 min ( Fig 2A , middle column ) . Under an intermediate level of pheromone stimulation , yeast cells arrested in cell cycle synchronously and initiated a chemotrophic elongated growth . Accordingly we observed a gradual slowdown of growth rate , in accord with the linear mode of cell mass accumulation in elongated cells ( Fig 2B , middle column ) . The protein abundances of most examined genes were also up-regulated , but to a lesser extent and were more fluctuating over time ( Fig 2B , left column ) . These results suggest that cell cycle progression and cell growth may influence the protein dynamics , in accordance with previous studies [33] . To calculate the actual protein synthesis rate from our data , we next employed a mass-action kinetic model of protein abundance and investigated the interdependence between the measured fluorescence and the growth dynamics . In this model , the changes in protein concentration were considered to be due to protein synthesis and decay; the latter was attributed to protein degradation and cell growth dilution ( see Methods ) . To calculate the synthesis rate for each protein , we took into account the expression P ( t ) and growth profiles γ ( t ) generated from the time-resolved measurements ( first two columns of Fig 2A and 2B ) and the protein degradation rate d obtained from a genome-wide analysis of protein half-lives [34] . The kinetic model is solved in a discrete manner so that the protein concentration change between two time points can be expressed as ΔP ( t ) = Δtα ( t ) − Δt ( d + γ ( t ) ) P ( t ) . The calculated results of protein synthesis rate α ( t ) are presented in the third columns of Fig 2A and 2B . Under the high pheromone level , the synthesis rates of most examined proteins were slightly up-regulated followed by a significant decrease to a much lower level . Because the average mRNA level of examined genes is not significantly reduced throughout the same time period [32] , our results suggest that the genome-wide regulation of translation rate is responsible for the global decrease in protein synthesis rate . This global reduction in translation rate may compensate the decrease in protein dilution rate caused by cell cycle arrest and lead to an overall decline of protein abundance after the initial induction ( Fig 2A , right column ) . In contrast , in cells responding to an intermediate level of pheromone , the protein synthesis rate was more fluctuating with a general trend of a more prolonged increase ( Fig 2B , right column ) . These results showed that distinct cell fates are associated with different dynamic changes in protein synthesis rate , as protein biogenesis might be strongly inhibited in the shmooing cells but not in the elongated growing cells . Since gene-gene interactions and variations in physiological parameters can affect protein dynamics at the same time , we investigated their impacts on protein expression control individually through kinetic modeling . We first examined to what extent the global physiological regulation inherent to each cell strain accounted for the dynamic changes in protein expression . We used a similar kinetic model as that used in the estimation of protein synthesis rate . The only difference is that the synthesis rate of each protein was replaced by a rescaled control rate S ( t ) generated as follows: ( 1 ) we estimated the dynamic changes of global protein synthesis rate for two phenotypes , which revealed the physiological constraints on protein expression control in the pheromone response ( S3 Fig ) ; ( 2 ) the global synthesis rate of each protein was generated by rescaling the normalized control rate to model the differences in their basal expressions . The rescaling factor is proportional to the steady state protein abundance prior to pheromone stimulation . The protein dilution and degradation rates were set to the per-strain values . Therefore , by estimating changes in protein synthesis and dilution that are independent of gene-specific interactions , we were able to obtain the protein dynamic changes caused by global regulation . We found the simulated protein abundance profiles showed a trend of transient up-regulation even though there is no gene-gene interactions ( Fig 3A , S3 Fig ) . The Pearson correlation between the predicted and observed fold change was 0 . 37 ( p-value = 8 . 5 x 10−6 ) , indicating that about 14% of the variance in the protein level can be explained by the influence of physiological factors . Thus our results suggest that although physiological regulation is responsible for the global trend of protein dynamics , the protein dynamics of different genes are mainly determined by gene-specific regulations . We assume that the discrepancy between the measured expression and the above-mentioned calculation stems from gene-specific regulation . To quantitatively dissect the gene-specific transcriptional regulation , we developed a computational framework named Protein Synthesis Decoupling Analysis ( PSDA ) that employs a robust and cross-grain estimation of the time window and fold change of gene regulation events ( Fig 3A ) . We used a modified kinetic model of protein life cycle , in which protein synthesis is the product of the gene-specific regulatory term R ( t ) and the global synthesis rate S ( t ) related to the physiological status . Therefore , rate parameters except S ( t ) contributed to the global regulation of protein abundance and R ( t ) was used to quantitatively assess the protein dynamic changes that could not be explained by the global regulation for each gene , thereby identifying gene-specific regulation dynamics . To filter out the noise inherent to experimental data and capture the main features of gene regulation , the gene-specific regulation term R ( t ) was parameterized as a pulse-like function [18 , 24] . The parameters were estimated utilizing a global optimization algorithm , namely differential simulated annealing ( DSA ) [35] . In each round of parameter evaluation , we simulated the dynamics of protein concentrations and set the value of the objective function to the sum of squared errors between the simulated and observed trajectory . The Markov chain length was set to 100 and the maximum round of function evaluation was set to 2000 for each gene . Distinct modes of regulation were deconvoluted from similar dynamic patterns of protein abundance in different cell fate responses ( S2 Dataset ) . Notably , our algorithm is capable of identifying the transcriptional down-regulation of genes despite of the induction of overall protein expression that is purely due to the alterations of physiological parameters . For instance , although the expression of FKH1 showed a 1 . 5-fold induction in cells undergoing shmoo formation , PSDA revealed a reduction in its protein synthesis in the last 4 h . This reduction in the protein abundance of Fkh1 , which is a cell cycle regulator , could account for the cell cycle arrest observed under this condition ( Fig 3B ) . Our PSDA approach yielded the gene-specific dynamics that are consistent with previous microarray results [32] ( S4 Fig ) , for most genes examined in this study . For example , synthesis of BAR1 is up-regulated after 1 . 5 h ( Fig 3B , gray shaded window in the panel labeled “Bar1” ) , contributing to a delayed negative feedback that modulates the response duration . Our analysis also revealed up-regulation of genes in the MAPK cascade , such as FUS3 and STE5 , consistent with previous microarray data [32] ( Fig 3B , gray shaded window in the panels labeled “Fus3” and “Ste5” ) . More importantly , the PSDA approach can identify time intervals of regulatory events that are otherwise invisible from the protein expression profiles . For instance , we found synthesis of FUS3 was up-regulated during the intermediate phase , while several typical mating genes , such as FAR1 , were induced in the late phase of experiment ( Fig 3B , gray shaded window in the panels labeled “Fus3” and “Far1” ) . Therefore , PSDA can identify gene-specific regulatory events from protein expression profiles and reveal the temporal patterns of regulatory events in a quantitative manner . To further characterize the regulatory modes of cells committed to the shmooing fate , we clustered genes into 6 groups according to their temporal patterns of gene-specific regulation , with 2 groups inhibited and 4 groups activated ( Fig 3C; S5A Fig ) . Using a smaller cluster number makes it hard to detect the sequential gene regulatory steps . For example , reducing the cluster number could merge C3 and C4 into a new cluster and we cannot distinguish genes that are activated in the early phase form that activated in the intermediate phase . On the other hand , subdividing the genes into more clusters would increase computation complexity and the number of possible networks , but does not alter the general network topology ( see Discussion below ) . The first cluster includes RP genes such as RPL1B and TFs , such as ABF1 and RAP1 , which are important regulators of RP synthesis [36] . The synthesis rate of these genes is rapidly down-regulated before recovering to the pre-treatment value , which might lead to decreases in ribosome synthesis ( Fig 3C , C1 ) . The second cluster contains genes that are inhibited with a time lag of ~2 h . This cluster is enriched with cell cycle genes ( S1 Table ) , including FKH1/2 , which are TFs that mediate the expression of genes in M phase ( Fig 3C , C2 ) . Genes in cluster 3 show a rapid and transient induction of expression and are primarily involved in stress-activated signaling responses , e . g . , OPY2 in the high-osmolarity glycerol ( HOG ) pathway and MDS3 that associates with the TOR pathway . DIG1 , an important regulator of the pheromone response pathway , is also included in this cluster ( Fig 3C , C3 ) . Genes encoding the components of MAPK pathway , such as KSS1 and FUS3 are enriched in cluster 4 and are transiently induced after cluster 3 genes ( Fig 3C , C4 ) . Cluster 5 includes the genes participate in execution of yeast mating and fusion , such as FUS1 and FAR1 . These genes are up-regulated slightly later than cluster 4 genes and exhibit a more prolonged induction patterns ( Fig 3C , C5 ) . Finally , cluster 6 genes , such as ISW1 and SNF2 , are up-regulated very late in the response and are primarily involved in chromatin remodeling ( Fig 3C , C6 ) . Because genes in the same cluster show similar temporal patterns , we hypothesized that they might share same upstream regulators . So we treated each cluster as a ‘meta-gene’ and generated its activation/inhibition time sequence via a threshold model ( S5B Fig ) , which allows us to reconstruct the putative interactions among the clusters by analyzing the time trajectory using a Boolean network model [14 , 16] . A wide range of parameter values was used in the threshold model to investigate the robustness of our results ( S5B Fig ) . The discrete trajectory resulted in 4 . 1 x 108 possible networks , including 96 minimal networks without redundant edges . We further incorporated prior knowledge to further confine the network structure ( Fig 4A ) . The regulatory network and representative genes in each cluster are illustrated in Fig 4B . The “input” of the network is the TFs activated by pheromone that can directly regulate gene expression , such as Ste12 and Tec1 . The reconstructed network successfully captures the core structure of pheromone response pathway and reveals genetic interactions in accordance with previous knowledge of the system ( Fig 4B , solid arrows ) . The network structure is robust to the choice of the cluster number . Subdivision of the largest cluster ( C6 ) into smaller clusters would increase the number of possible networks , while leaving the general network structure unchanged ( S6 Fig ) . More importantly , our results also suggested novel putative interactions in the gene expression programs induced by pheromone ( Fig 4B , dashed arrows ) . For example , the stress response genes in cluster 3 might repress translation by inhibiting cluster 1 genes . Additionally , in the early phase of the response , cluster 2 cell cycle genes may repress the expression of cluster 6 genes , implying a coordinated regulation of cell cycle and chromatin remodeling during the mating response . In summary , these results support the use of our experimental and computational approaches in dynamic network analysis and imply the potential interactions of the canonical mating pathway with various signaling and cellular processes , offering a more comprehensive picture of the yeast mating response . To investigate the mechanisms underlying dose-dependent mating differentiation , we next reconstructed the gene regulatory network for cells committed to chemotropic elongated growth in response to the intermediate pheromone level . To this end , we examined the deconvoluted gene-specific regulations of the six meta-genes ( classified in Fig 3C ) for elongated growth and compared their time trajectories with those of the cells committed to shmoo formation upon a high pheromone dose . Interestingly , although we observed striking differences in protein dynamics of all six meta-genes between the two cell fates , only the genes in cluster 1 and 3 exhibited a qualitative difference . Cluster 1 genes are down-regulated during shmoo formation in response to the high level of pheromone stimulation , but are up-regulated during elongated growth in response to the intermediate pheromone dose . In contrast , cluster 3 genes are transiently induced during shmoo formation , but are repressed during elongated growth . All the other clusters showed quantitative differences between the responses to different doses of pheromone , in which the gene regulation are towards the same direction and only differ in the extent of changes ( Fig 5A ) . Based on the dynamic changes of the meta-genes in response to the intermediate level of pheromone , we generated a putative regulatory network that can recapitulate the protein dynamics during elongated growth ( S7A and S7B Fig ) . We then compared this network structure with that of shmoo formation response ( Fig 5C–5D ) . Consistent with the analysis of protein dynamics in Fig 5A , we found that the core network structure consisting of the interactions between clusters 2 , 4 and 5 remained unchanged in the two cell fates . The major difference lies in cluster 1 and 3 . During elongated growth in response to an intermediate pheromone level , the RP genes in cluster 1 are induced but the stress response genes in cluster 3 are repressed . In contrast , during shmoo formation triggered by a high pheromone dose , cluster 1 genes are inhibited whereas cluster 3 genes are induced . Furthermore , since cluster 3 undergoes a slightly faster response than cluster 1 during shmoo formation , it is possible that the stress response genes in cluster 3 may contribute to the repression of RP genes in cluster 1 . To examine the robustness of the uncovered networks , we further investigated the attractor landscape for cell fate determination in the mating response ( S7C and S7D Fig ) . Although re-wiring of the regulatory network alters the attractor landscape , the two biological phenotypes emerge as the largest attractors , revealing the dynamic robustness of the cell fate commitment . Taken together , our modeling analysis suggested that the dose-dependent regulation of stress response genes might lead to the phenotypic differences associated with distinct cell fates , in which the cells undergoing elongated growth have an enhanced cell growth and biogenesis capacity , whereas the cells committed to shmoo formation feature a reduced biogenesis capacity . To further validate the model results , we experimentally explored the molecular connections between the pheromone response pathway and the RP or stress response genes . In S . cerevisiae , induction of stress response genes and repression of RP genes are simultaneously associated with the responses to nutrient limitation or stresses and are mediated by the general stress responsive pathways , such as TOR or protein kinase A ( PKA ) pathways [37 , 38] . Thus , we examined major stress responsive regulators in yeast , including Sfp1—a stress responsive transcription factor [30] , Hog1 –a stress-activated protein kinase [39] , Msn2 –a general stress responsive TF in the PKA pathway [40] , Yap1 –a TF involved in transcriptional response to various stresses [41] , and Tpk1 –the catalytic subunit of PKA [42] . These regulators can govern the expression of RP and/or stress response genes and their activities are primarily controlled via nucleocytoplasmic translocation . We found that pheromone stimulation had no effect on the localization of Hog1 , Msn2 , Yap1 or Tpk1 ( S7 Fig ) . In contrast , whereas localized in the nucleus under vegetative or elongated growth conditions , Sfp1 translocated from the nucleus to the cytoplasm in response to a high level of pheromone stimulation leading to shmoo formation ( Fig 6; S8 Fig; S1 Video ) . Intriguingly , Sfp1 has been best known as a stress responsive TF primarily involved in the TOR signaling pathway [30 , 43] . Under optimal growth conditions , the Tor kinases are active and Sfp1 localizes in the nucleus where it activates the expression of RP and ribosomal biogenesis genes; in response to nutrient limitation or chemical stress , Tor kinases are inhibited and Sfp1 is translocated from the nucleus to the cytoplasm . Therefore , our results suggest a potential cross-regulation of the TOR-dependent ribosomal biogenesis pathway that only occurs in shmooing cells . This pathway crosstalk could serve as a molecular switch that mediates the dose-dependent network rewiring and cell fate determination during mating differentiation .
It is a challenging task to resolve regulatory networks from experimental observations in eukaryotic organisms , despite continuous advances in reverse engineering methods in the past decades [8] . One reason for this is that regulation of protein abundance is not only determined by the gene-gene interactions , but also subject to the availability of cellular resources and changes in growth dynamics . In this paper , we employed an integrated approach that combines high-throughput experiments , dynamic modeling of the protein life cycle and a reverse engineering method to investigate the regulatory mechanisms underlying distinct cell fates in the yeast mating response . Based on the high-resolution profiling of the protein abundance and growth dynamics of yeast cells , we found that gene expression can be strongly affected by the global changes in cell physiological state , hampering the identification of gene-specific regulation events that are critical for network discovery . Here , we solved this problem by taking a model-driven approach ( PSDA ) to assess the influence of global regulation and deconvolute the gene-specific regulation , which enables us to further reconstruct regulatory networks underlying different mating responses . Our PSDA method confers advantages of robust detection of regulation events and little data requirement , but is not without limitations . For example , we assumed the regulation of protein level mainly takes place in the synthesis part . This assumption , supported by previous studies [23 , 44] , helps to reduce the number of free parameters while preserving the capacity to identify regulatory events . However , it may overlook dramatic changes in protein degradation that is crucial for some specific responses . Also PSDA quantitates the summarized effect of regulations in transcriptional and translational levels and therefore it cannot distinguish between the two effects . Further incorporation of time-resolved transcriptome data may enable more accurate quantification of transcriptional and translational regulation and comparison with relevant methods [24] . The putative regulatory networks were generated without empirical parameterization of the regulation functions , yet successfully recapitulated the dynamics of the system underlying different phenotypic responses . Our reconstructed network includes genes encoding the components of canonical pheromone response pathway that regulate signal transduction , cell cycle arrest and cell fusion . In addition , our network analysis suggests a conditional regulation of RP and stress response genes , which guided our identification of the dose-dependent translocation of a stress responsive transcriptional factor , Sfp1 . Since the localization of Sfp1 is directly regulated by the activity of Tor kinases [30 , 45] , it is possible that Tor activity is only diminished in shmooing cells , but not in elongated cells . Intriguingly , a previous study [33] showed using microarray time course analysis that a high dose of pheromone down-regulates 54 ribosomal proteins . This down-regulation is independent of pheromone-induced cell cycle arrest and the kinetics is similar to that observed in response to rapamycin treatment . Their findings are consistent with our modeling prediction and experimental results , pointing to a possible cross talk between the mating pathway and the TOR pathway . A systematic epistasis analysis of the pheromone response pathway would provide us further insights into the mechanisms of this pathway crosstalk . Taken together , our experimental and computational approaches represent an integrated framework for analyzing the proteomic dynamics during cell differentiation . Given the growing interests in large-scale protein dynamics and network analysis , we anticipate that our strategies would be widely applied to enable systems-level understanding of other biological systems .
The yeast cell strains used in this study were selected from a chromosomally GFP-tagged library . Strains were grown to saturation at 30°C and further diluted and cultured for another 8 hours to reach the exponential growth phase before the microfluidic experiments . The alpha factor ( Sigma-Aldrich , St . Louis , MO ) level was set to 5 . 9μM/L for the high concentration and 0 . 59μM/L for the intermediate concentration . A high-throughput microfluidic chip was used in our fluorescence experiment , which allows a maximum of 96 parallel experiments across 2 different conditions ( S1A Fig ) . Our chip was fabricated with PDMS ( polydimethylsiloxane , RTV615 , Momentive Performance Materials Inc . ) using standard soft lithography technology . Each strain was loaded into an individual channel and a constant flow rate of 400 μl/h for fresh media was achieved . Phase contrast and fluorescence images of yeast cells were generated via a Nikon Ti-E microscope in combination with NIS-Elements software every 5 min . A cell culture incubator around the microscope was used to maintain a temperature of 30°C . We developed a kinetic model of protein dynamic change in which: dP ( t ) dt=α ( t ) − ( d+γ ( t ) ) P ( t ) where P ( t ) is the protein concentration at time t , α ( t ) represents the protein synthesis rate and γ ( t ) denotes protein dilution rate , which equals the exponential cell mass accumulation rate . d is the protein degradation rate estimated from a genome-wide measurement of protein half-lives [34] . We suspected that the per-gene regulations in the response mainly took place in the protein synthesis term . Our kinetic model for protein concentration was thus modified to the following form: dP ( t ) dt=R ( t ) S ( t ) − ( d+γ ( t ) ) P ( t ) in which S ( t ) is the global synthesis rate that reflects the changes of cellular resources related to protein synthesis , and R ( t ) is the temporal regulation term of the protein . We adopted an efficient parameter optimization algorithm named differential stimulated annealing ( DSA ) to estimate the regulation parameters for each protein [35] . The calculation was based on a customized MATLAB version of the algorithm . For detailed information , see S1 Text . In the Boolean network model , each node represents a biological species . Si ( t ) ∊{0 , 1} is used to denote the state of node i at time t . Regulation from node j to node i is represented by the coefficient aji , which is positive for activation and negative for inhibition . Update of the node states is described by the following Boolean functions: {Si ( t+1 ) =θ ( ∑jSj ( t ) aji ) , ∑jSj ( t ) aji≠0Si ( t+1 ) =Si ( t ) , ∑jSj ( t ) aji=0 where θ is the Heaviside step function as follows: θ ( x ) = 1 when x > 0 and θ ( x ) = 0 when x < 0 . From a given initial state , the state of the system is updated until it reaches a steady state known as an attractor . | A systematic characterization of the proteomic changes during the process of cell differentiation is critical for understanding the underlying molecular mechanisms . However , protein expression can be largely affected by changes in cell physiological state , which hampers the detection of regulatory interactions . Here we proposed an integrated experimental and computational framework to reconstruct regulatory circuits in mating differentiation of budding yeast Saccharomyces cerevisiae , in which distinct cell fates are triggered by alteration in pheromone concentration . A modeling approach was developed to decouple gene-specific regulation from growth-dependent global regulation of protein expression , allowing us to reverse engineering the gene regulatory circuits underlying distinct cell fates . Our work highlights the importance of model-based analysis of proteomic data and delivers new insight into dose-dependent differentiation behavior of budding yeast . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"cellular",
"stress",
"responses",
"gene",
"regulation",
"cell",
"processes",
"protein",
"abundance",
"protein",
"expression",
"protein",
"synthesis",
"molecular",
"biology",
"techniques",
"chemical",
"synthesis",
"research",
"and",
"analysis",
"methods",
"proteins",
"g... | 2017 | Reconstructing the regulatory circuit of cell fate determination in yeast mating response |
To develop more detailed models of EBV persistence we have studied the dynamics of virus shedding in healthy carriers . We demonstrate that EBV shedding into saliva is continuous and rapid such that the virus level is replaced in ≤2 minutes , the average time that a normal individual swallows . Thus , the mouth is not a reservoir of virus but a conduit through which a continuous flow stream of virus passes in saliva . Consequently , virus is being shed at a much higher rate than previously thought , a level too high to be accounted for by replication in B cells in Waldeyer's ring alone . Virus shedding is relatively stable over short periods ( hours-days ) but varies through 3 . 5 to 5 . 5 logs over longer periods , a degree of variation that also cannot be accounted for solely by replication in B cells . This variation means , contrary to what is generally believed , that the definition of high and low shedder is not so much a function of variation between individuals but within individuals over time . The dynamics of shedding describe a process governing virus production that is occurring independently ≤3 times at any moment . This process grows exponentially and is then randomly terminated . We propose that these dynamics are best explained by a model where single B cells sporadically release virus that infects anywhere from 1 to 5 epithelial cells . This infection spreads at a constant exponential rate and is terminated randomly , resulting in infected plaques of epithelial cells ranging in size from 1 to 105 cells . At any one time there are a very small number ( ≤3 ) of plaques . We suggest that the final size of these plaques is a function of the rate of infectious spread within the lymphoepithelium which may be governed by the structural complexity of the tissue but is ultimately limited by the immune response .
Epstein-Barr virus is a human herpes virus that has the capacity to transform B lymphocytes in culture into continuously proliferating lymphoblasts [1] , [2] . However , it persists in vivo in resting , germinal center derived memory B cells [2] , [3] that we refer to as mBLat [4] . The generally accepted model of viral persistence holds that the virus uses its transforming abilities to activate newly infected resting B cells and then latency associated viral proteins provide surrogate signals to allow the newly activated cells to differentiate through the germinal center into the memory compartment [1]–[3] . EBV is clinically important because of its involvement in several human cancers including Hodgkin's disease , Burkitt's lymphoma , immunoblastic lymphoma in the immunosuppressed and nasopharyngeal and gastric carcinomas [5] , [6] . The ability to immortalize B cells and the fact that EBV encodes for several potential oncogenes [7] , [8] provides one source of risk for EBV associated tumorigenesis . A second factor in EBV associated cancer susceptibility is that the virus establishes a persistent infection resulting in lifetime exposure of the host to infectious virus and viral gene products . The tissue that is most exposed to infectious virus is the epithelium of the tonsils and nasopharynx since this is where the virus is replicated and shed into saliva for horizontal spread of the infection to new hosts [9] , [10] . Long term exposure to infectious virus in this anatomical region may be an important risk factor in the development of the EBV associated cancer nasopharyngeal carcinoma . Although virus shedding is a hallmark of persistent infection it has not , to our knowledge , been systematically studied in healthy carriers of the virus by sensitive , quantitative PCR techniques . Currently , therefore , there is a dearth of systematic quantitative data on viral shedding that would allow the formulation and testing of models of virus production . Consequently it remains unclear what regulates the levels of virus shedding in saliva . Some previous studies have used a biological assay to detect EBV [10] , [11] , based on its ability to transform B cells in culture . However , in our hands this assay is unreliable , insensitive and not quantitative . Others have used relatively insensitive and therefore non-quantitative PCR approaches to assess virus shedding particularly during acute infections e . g . [12]–[15] . Perhaps not surprisingly these studies have produced conflicting results . For example one study concluded that virus shedding was correlated with the levels of infected B cells in the blood whereas a second study found no such correlation [11] , [13] . Generally however , it appears to be the view in the field that individuals can be classified by relative levels of shedding into high , intermediate and low shedders where virus shedding is sporadic and often undetectable [5] , [11] . Since these conclusions are based on non-quantitative and insensitive assays it is difficult to assess and reconcile these conclusions . For example , it is unclear if failure to detect virus is due to its absence or simply a function of the insensitivity of the assays used . We have previously developed sensitive , quantitative PCR techniques for detecting EBV DNA [4] , [16] and EBV gene products [17] , [18] . Using these assays we showed for the first time that the levels of virus infected cells in the blood of healthy carriers were individual specific and highly stable over long periods of time ( years ) [19] . In this study we have set out to apply this same approach to the dynamics of virus shedding into saliva by healthy carriers of the virus and use this data to develop and test potential models of EBV shedding . We find that the overall dynamics strongly implicate the epithelium as a site where virus is amplified prior to shedding .
Traditionally the levels of virus in saliva have been assessed at a single point in time by having subjects gargle with a defined volume of fluid . The amount of virus in the gargle is then measured either with a biological assay [10] , [11] , based on transformation of B cells in culture or by PCR for the copy number of viral genomes [12]–[15] . However , this type of analysis is static and provides no kinetic information about the rate of virus production . Therefore in our first series of studies we sought to measure the rate of virus production by first rinsing out all of the virus in saliva and then measuring the rate at which the virus returned . To this end volunteers were asked to rinse and gargle sequentially with 5 cc of fluid and the levels of virus measured with a quantitative DNA PCR technique that can detect a single copy of the viral genome . To our surprise sequential rinses failed to deplete the levels of virus in the saliva even when up to 8 rinses in a row were tested ( Figure 1 shows the results for two subjects who were at the lower and upper levels respectively of shedding at the time of the experiment ) . This held true even when the subjects attempted to wash out the virus first with 4 sequential 25 ml rinses and then with 2 more 25 ml rinses . After each attempt a subsequent 5 ml rinse had essentially the same amount of virus as the initial rinse ( see Figure S1 ) . We reached the unexpected conclusion therefore that virus is continuously being shed into saliva at such a high rate that the virus removed by rinsing was already replaced by the time we retested the saliva ( ∼2 minutes ) . To our knowledge this is the first time that this simple technique of sequential sampling has been performed and leads us to the conclusion that the rate of virus production is much greater than had been previously thought . Included in Figure 1 is the estimated virus shedding per day for the two subjects analyzed which lead to the conclusion that 5×105 and 108 genomes are being produced by Subject 1 and 2 respectively per rinse or 3 . 2×108 and 6×1010 per day . The time course of virus production by individual cells in culture has been estimated to be approximately 24 hours from initiation of lytic gene transcription to the release of infectious virus [20] . However , to our knowledge no-one has attempted to accurately estimate the viral burst size . We have done this multiple times for both B cell lines ( Akata and B95-8 ) and epithelial cell lines ( HONEAkata and AGSBX1 ) . The fraction of cells replicating the virus was assessed by flow cytometry after staining for the viral capsid antigen 48 hours post stimulation . The amount of virus released was measured by performing EBV specific DNA PCR on the cell culture supernatants at 72 hours . The results of this analysis are summarized in Table 1 . For the B cell lines the average production was quite similar despite their very different origins or whether or not the virus was chemically induced . Similarly for the two epithelial cell lines . However , production from the epithelial cells ( 6 . 2×105genome copies per cell ) was about ten fold higher than from the B cells ( 7 . 3×104 genome copies per cell ) However less than 10% of these genomes appeared to be intact virions based on resistance to DNase digestion . ( Table 1 ) . Using a burst size of 7 . 3×104 genome copies per B cell and assuming a 24 hour lytic cycle we can estimate that the shedding by Subjects 1 and 2 would require viral production from 4 . 4×103 and 8 . 2×105 B cells per day , respectively . This is several logs higher than the value we have measured previously for the number of lymphocytes in Waldeyer's ring that complete virus replication which we estimate to be extremely small averaging 0 . 5 cells per day with an upper bound <100 ( Box 1 ) . Therefore , we may conclude that there are insufficient B cells replicating EBV in Waldeyer's ring to account for the levels of virus being shed into the saliva ( see also Box 1 and Discussion ) . Figure 1 demonstrates that virus shedding into saliva was quite stable at least over short time periods . When we measured shedding in the same individuals over hours or days we saw a similar degree of stability ( Figure 2 ) . However , when we tracked them for longer periods ( up to one and a half years ) we saw a strikingly different result ( Figure 3 ) with fluctuations over 4–5 logs . To test if this variation was a consequence of variable “aggressiveness” in the sampling procedure over time we also performed PCR for an internal cellular control gene ( bcl-2 ) . When the levels of cellular DNA in saliva were tracked over the course of three months in three individuals the largest variation in recovery we obtained was 40 fold compared to 1 . 5×104 fold in virus shedding in the same samples ( not shown ) This indicates that sampling variation was not the source of the variation in virus shedding that we observed . Unlike virus shedding we have shown previously that the level of latently infected memory B cells ( mBLat ) in the blood of healthy carriers is extremely stable over years [19] and even decades ( DTL unpublished observations ) in healthy carriers . To confirm that this was true for the subjects and times studied here we measured the levels of mBLat in their blood at the same times we measured virus shedding . As expected the levels of mBLat in their blood were stable over the entire time period ( Figure 3 ) . We conclude that although virus shedding within an individual is quite stable over short time periods ( hours/days ) it is highly variable over the course of months and years . As described above we have previously shown that the number of B cells completing viral replication per day in Waldeyer's ring is small therefore it is also not possible to account for the variation in shedding within an individual simply in terms of variation in the number of B lymphocytes replicating the virus in Waldeyer's ring at any one time ( see also Discussion ) . We have analyzed the data from Figure 3 to see if it contained an underlying structure that might shed light on the dynamics and mechanism of virus shedding . Most revealing was the cumulative distribution function ( CDF ) . The CDF ( http://en . wikipedia . org/wiki/Cumulative_distribution_function ) describes the probability distribution of a real-valued random variable X , in this case the log value of virus shedding . The CDF of a given value of X is the proportion of the measurements whose value is equal to or less than X . As the value of X increases , the CDF of X increases from 0 to 1 . So for example the median for a set of measurements has a CDF of 0 . 5 . The CDF for the three subjects in Figure 3 are presented in Figure 4A . Here the log values for virus shedding are plotted cumulatively from lowest to highest and thus appear as steps , and the number of data points are normalized by setting the total to equal 1 . Strikingly the distributions were essentially linear for all three data sets . To see if this was a coincidence unique to this particular experiment or a general property of virus shedding we analyzed a large data set collected from 8 different subjects over an extended period of time ( up to two years ) which also included a much larger number of data points for subjects 1–3 ( essentially additional points for which we did not have matching blood level measurements ) . A summary of all this data , which includes a total of 165 positive data points and 8 time points where no virus shedding was detected , is presented in box plots in Figure 5A and in Table S1 . The time points where no virus shedding was detected are not included in the box plots but are indicated by the asterisks in Figure 5A . We are confident that there was no viral DNA present in the saliva samples at these times because our PCR technique can reproducibly detect a single copy of the viral genome and positive detection of a control cellular sequence ( bcl-2 ) confirmed the sampling process was normal . From Figure 5A it is apparent that a large range of variation is a ubiquitous feature of virus shedding with most subjects demonstrating around a 4 log range , the largest being 5 . 5 logs ( subject #9 ) and the smallest 3 . 3 logs ( subject #8 ) . Table S1 shows the mean , median and standard error of the mean ( SEM ) for the measurements on all 8 subjects Although the mean and median values only varied by about two orders of magnitude between the 8 subjects , the SEM within each individual ranged over 4–6 orders of magnitude indicating the extent to which the range of virus shedding overlapped between individuals and that the shedding profiles are not statistically different between different individuals . Most revealing though was the CDF analysis of this data shown in Figure 4B which confirmed the original result with the limited data set from subjects 1–3 . For all 8 subjects , a linear relationship was observed between the log value of the shedding and its cumulative probability of occurring . Put another way the probability of ( log ) shedding was equal for all levels over a several log range . This is the expected result when a simple exponential process , such as growth or infectious spread , is sampled randomly over time . In Figure 6 we show the best fit lines to the data from all 8 Subjects ( from Figure 4B ) and in Table 2 we describe the properties of these lines including the slopes and the results of linear regression analysis ( adjusted R2 ) . The R2 value for all 8 subjects was extremely high ( ≥0 . 88 ) indicating that the data closely approximates a linear relationship . In addition it is apparent that the slopes of the lines are very similar for all 8 subjects although the lines were spread over a wide range encompassing >2 logs ( compare Subject 2 and 9 in Figure 4B and the x intercepts for all 8 subjects in Figure 6 ) . These results provide several important conclusions . Contrary to what has been claimed previously [11] , there are no distinct high , intermediate or low virus shedders . So although on average for example subject #9 is low , subject #3 medium and subject #2 high , multiple measurements over extended time show these assignments to be relatively meaningless . Rather at any one time an individual might be high , low or intermediate . Furthermore the underlying process behind the variation in shedding is a simple exponential that is either arrested randomly , and then held stable , or observed at random times . This exponential is initiated over a range of shedding at levels that are subject specific ( lines are displaced over >2 logs in the CDF plots ) but whose subsequent growth rate is subject independent ( slopes are similar in the CDF plots ) . We suggest that these events are the exponential expansion of infected epithelial cell plaques ( see Discussion ) . A straight line relationship between the log of shedding and the cumulative number of observations ( the CDF ) implies that the number of these putative plaques must be very small otherwise the shedding from the largest plaques would always dominate and the plot would become non-linear . This is demonstrated in Figure 6 and Figure S2 where we have performed a simulation of this process for various numbers of plaques ( For details of the simulation see Text S1 – “Empirical CDF of shedding suggests there are at most 3 plaques at any one time . ” ) . In this simulation various numbers of plaques are seeded randomly , allowed to grow exponentially and then sampled at 20 random points in time . This produces a simulated CDF of log shedding from which an adjusted R2 can be calculated . As the number of plaques increases the plots tend to become steeper and increasingly skewed away from linear . We next used the simulation to ask: what is the probability of obtaining our actual data set with different numbers of plaques ? To achieve this we simply repeated the simulations shown in Figure S2 100 times for each plaque number , performed a linear regression analysis to calculate the adjusted R2 for each run and then plotted these values as a CDF for the adjusted R2′s ( Figure 7 ) . We can then ask: what is the probability of obtaining the adjusted R2 values we have observed for our 8 subjects with simulations involving different numbers of plaques . We have used three different criteria . First , what is the probability of obtaining an adjusted R2 value 7 out of 8 times in a row that is higher than the lowest value we have observed with our 8 subjects ? Second what is the probability of obtaining 8 values with a median value equal to or higher than the median value for our 8 subjects ? Third , what is the probability of obtaining 8 values as high as or higher than the lowest value we have observed with our subjects . This calculation revealed p values of <0 . 05 for ≥2 plaques , ≥3 plaques and ≥4 plaques for the three methods respectively . Taken together these analyses suggest that the shedding of EBV in healthy carriers is the result of a very small number , ≤3 , of these putative epithelial plaques . We can also derive an estimate of the average plaque number in a completely different and independent way since we have observed that virus was not shed on 8 occasions out of the total 173 measurements we have taken . From simple Poisson statistics we can calculate that if we find no evidence of infectious plaques 8/173 times then on average there must be ∼3 such plaques , a value which is in very close agreement with the prediction ( ≤3 ) of the CDF analysis . In conclusion it appears that EBV shedding is the consequence of an exponential process that is either terminating or being observed randomly and that there are ≤3 of these events occurring at any given time . As noted above the levels of mBLat in the blood of healthy carriers of EBV is stable over long periods of time ( [19] and Figure 3 ) and we have now shown conversely , that virus shedding is extremely variable over similar time spans . Therefore there is no correlation within individuals between virus shedding and levels of infected cells . In Figure 5B we have plotted the mean values from Table S1 versus the frequency of mBLat in the circulation for all subjects studied and it is apparent that there is also no correlation between the levels of mBLat in the blood and the amount of virus shed into saliva between individuals . ( The same result holds true if the median is plotted , not shown ) . One possible explanation for the lack of correlation between shed virus and the levels of mBLat in the blood is that PCR does not distinguish intact virions from non-infectious viral DNA . To evaluate this we assessed the DNase sensitivity of viral DNA in saliva ( see Figure S3 ) for an example of such an experiment ) . We found that most of it was resistant; consistent with being in intact virions , whether we tested raw saliva or the supernatant after clarification by centrifugation ( Table 3 ) . This was much higher than for the 2 cell lines we tested Akata and B95-8 ( only 10–20% resistant ) suggesting that virions shed into saliva are more efficiently packaged than in the cell lines . When we tracked the levels of DNase resistant material in raw or clarified saliva we again saw a 4 to 5 log variation over extended periods of time ( Figure 8 ) and no correlation with the level of mBLat ( Figure S4 ) . The other possible explanation for our failure to find a correlation between shed virus and mBLat was that PCR measures all virions irrespective of their infectivity and it has been suggested that a large fraction of virions in saliva are bound by neutralizing antibody [21] . To test this possibility we assayed the amount of viral DNA that would bind specifically ( as judged by the failure to bind the T cell line Jurkat and the ability to be blocked with anti-gp350 neutralizing antibody 72A1 – not shown ) to a B cell line . As shown in Table 4 in two separate experiments despite repeated absorptions ( up to 6 times ) only <5% of the viral DNA in saliva could bind the B cell line Akata . However , when we tested the amount of Akata binding activity in saliva over an extended time period we again found that the levels varied by approximately 4 to 5 logs and that there was no correlation with the levels of mBLat in the peripheral blood ( Figure S5 ) .
In this paper we have described detailed kinetics of EBV shedding in the saliva of healthy carriers of the virus . We have employed a sensitive and quantitative DNA PCR assay that can reliably detect a single copy of the viral genome . Using this assay we show for the first time that EBV positive individuals continuously shed EBV such that the saliva reservoir is completely replaced in ∼2 minutes . This is consistent with studies showing that a normal individual swallows on average once every 2 minutes ( www . uiowa . edu/~c003236/indexS1L1a . html ) . Thus the mouth is not a reservoir of virus but a conduit through which a continuous flow stream of virus passes in saliva ( Figure 9 ) . Over the course of several months to a year the level of shedding by a given individual varies by 3 . 5 to 5 . 5 logs . Thus contrary to what is currently thought [11] there is no clear distinction between low , intermediate or high shedders of EBV . In fact shedding changes so much over time that most of the variation between high and low occurs within an individual not between individuals . If a cohort of individuals are only measured over the course of a few days some will be low , some intermediate and some high , but the classifications will change over time . For example in Figure 3 Subject 1 might be classified as high ( 106 to 108 per ml ) on day 300 or low ( 102 to 104 per ml ) on day 325 and intermediate ( 104 to 106 per ml ) on day 500 whereas Subject 2 would be high on day 50 , low on day 300 and intermediate on day 325 . This point is confirmed by the CDF analysis which demonstrates that an individual has an equal probability of shedding at any level ( log values ) from high to low . Previous studies using either a biological assay [22] , [23] or PCR [24] , [25] to detect the virus have reported that 10 to 70 percent of EBV positive individuals shed detectable virus . In this study we took 173 measurements over time on 8 different EBV positive individuals and consistently found all 8 to be positive with only 8/173 negative readings . Unlike previous studies we can be confident that there was no viral DNA present in the saliva samples at these times because our PCR technique can reproducibly detect a single copy of the viral genome . Therefore , we did not detect an EBV positive individual who was consistently a non-shedder . The much higher rate of success we obtained no doubt reflects that the assays used before were considerably less sensitive . We observed similar discrepancies previously when assessing the prevalence of EBV in peripheral blood B cells . Earlier reports had suggested a wide range of detection based on PCR or a biological assay . However , application of our DNA PCR assay revealed that 100% of EBV seropositive individuals had infected cells in their blood 100% of the time and indeed the levels were very stable and specific to each individual [19] . These studies on viral loads in saliva and blood B cells demonstrate that performing precise , sensitive and systematic quantitative studies can provide consistent and novel insights into EBV biology . Surprisingly the main conclusions of our studies on viral shedding point to the potential importance of epithelial cells in EBV replication and shedding . This is because neither the absolute levels nor the variation over time of the virus shedding we have observed can be accounted for by B cells alone replicating the virus in Waldeyer's ring [26] . In Box 1 we have attempted to estimate the amount of virus being shed from Waldeyer's ring by B cells . This calculation is based on the number of infected B cells in Waldeyer's ring from measurements on a large cohort of normal subjects [27] , our previously published estimates of the fraction of infected B cells that completes viral replication at any time in Waldeyer's ring [26] and the viral burst size as estimated in the current study . From this calculation we can see that there is ∼3 log discrepancy between the absolute maximum amount of virus that could be produced by B cells in Waldeyer's ring and the actual levels we have measured in saliva . Therefore it is highly unlikely that B cells alone can account for the levels of virus shedding we have observed . Besides the absolute levels of virus shedding , the range ( 3 . 5 to 5 . 5 logs ) we have observed over time in any given healthy carrier is also too large to be accounted for by B cells alone replicating the virus since we estimated previously that at most ∼100 B cells complete EBV replication in Waldeyer's ring at any one time ( [26] and Box 1 ) . This could only yield at best ∼2 logs of variation , again implicating a second site of EBV replication . What/where could this cell type be ? Recently the possible role of epithelial cell infection in EBV biology has again been raised . Infected epithelial cells have been detected in cultures of primary tonsil epithelium [28] , EBV has been shown to have a cluster of virion glycoproteins ( gH and gL ) that mediate epithelial cell infection [29] , [30] and EBV is found in the epithelial cells of nasopharyngeal carcinoma [31] ( where the virus is latent ) and oral hairy leukoplakia [32] ( where the virus is lytic ) . Furthermore , the virus in saliva is high in the glycoprotein gp42 . This glycoprotein binds to MHC Class II and is usually depleted in virions that emerge from B cells suggesting that the virus in saliva originates from a non-B cell [33] . This is consistent with a model whereby virus released from B cells infects the tonsil epithelium where it replicates and becomes amplified prior to shedding into saliva . To investigate this possibility we have estimated the potential amount of virus produced from epithelial cells in Waldeyer's ring ( Box 2 ) . This calculation was based on published values for the surface area of epithelial cells in Waldeyer's ring [34] , our estimate of the density of an epithelial cell monolayer [28] , our published estimates of the frequency of epithelial cells replicating EBV in cultures of primary tonsil epithelium [28] and the viral burst size of epithelial cells as estimated in this study ( Table 1 ) . From this calculation it is apparent that epithelial cell infection could account for the levels of shedding we have seen . The CDF analysis provides considerable insight into how epithelial infection could account for the shedding we have observed . We have seen that the CDF for the log of shedding is close to linear for each subject . A linear CDF occurs when a process grows exponentially with respect to time over a fixed life span and is either sampled at random or arrested and held constant at random times . We can discount the first hypothesis ( sampling ) because virus shedding is stable over short periods of time and is therefore not simply an exponentially growing function . Rather it appears to be an exponentially growing function that is stopped at random , stays stable for several hours/days and then dissipates . Since virus shedding cannot be accounted for by B cells we propose that this function is the growth of infected epithelial plaques that are seeded by virus released from B cells and that these plaques become arrested in their spread randomly over time . This would be the EBV equivalent of HSV reactivating in a ganglia traveling to the skin and infecting the surrounding fibroblasts leading to the production of a herpes lesion . Assuming a burst size of 6 . 2×105 for epithelial cells ( Table 1 ) the range of virus shedding/ml saliva rinse we have observed in Figure 5A and Table S1 ( 5−2×107 genomes yields [5−2 . 107]×5 ml/rinse×30 rinses/hour×24 hrs = 1 . 8×104 to 7 . 2×1010 genomes per day ) would derive from <1 to 105 epithelial cells releasing virus per day . This implies that the plaques range in size anywhere from 1 cell ( which could simply be the original seeding B cell ) up to 105 epithelial cells . As demonstrated in the results section the linear nature of the CDF for our subjects constrains the number of these plaques to be ≤3 at any time . However if this is true why do we not see periods of simple exponential growth in our measurements ? When we perform measurements at ever shorter time periods we eventually encounter stability not exponential growth . This observation puts two constraints on the model , The first is that the number of plaques is usually >1 such that a plaque in its stable state is usually masking part or all of the growth phase of a new plaque ( s ) and second that the maximum time of growth for the plaque must be less than the length of time they are stable which we have seen is up to 7 days ( For a detailed discussion of this issue see Text S2 – “Why we have not observed the growth phase . ” . This places the maximum life time for a plaque at 14 days: at most 7 days of expansion followed by <7 days of stability . The two obvious candidates for the exponential function predicted by the CDF are growth of infected epithelial cells or infectious spread of virus . If growth were responsible it assumes a model where new infection produces latently infected epithelial cells that continue to proliferate and generate cells producing virus at the same time until some event , presumably the immune response halts the growth . Latent infection of tonsil epithelium has been documented [28] and this type of scenario was proposed many years ago based on analogy to the HPV system [35] where the differentiation of latently infected cells triggers viral replication . In support of this idea , differentiation dependent replication of EBV has been documented in both epithelial [36] and B cells [26] . However , we do not favor this model because it would take two to three weeks to achieve the maximal plaques size of 105 cells ( assuming the cells divide once every 24 hours ) . As we have seen the data constrains the growth phase to be no more than a week and in addition it seems highly unlikely that such a structure could evade the immune response for such a long period of time . The alternative hypothesis is that the plaques simply grow by infectious spread . Precedence for this idea comes from in vitro infection of polarized tonsil epithelium which has been shown to readily generate plaques of cells replicating the virus [37] , [38] and in vivo from the pathological entity oral hairy leukoplakia which arises in the immunosuppressed and essentially consists of very large plaques of epithelial cells replicating EBV [32] . In this model a single epithelial cell releasing virions sufficient to infect on average ∼10 or more new cells after each round would reach the maximum plaque size in less than a week . However , there is a geometric constraint which must be considered . The tonsil surface exhibits complex geometry but can be thought of effectively as a large 2-dimensional surface . If infection can only take place between neighboring epithelial cells , then infection is confined to a growing 2-dimensional disc . The radius of this disc depends on the number of rounds of neighbor to neighbor infections , i . e . the length of time the infection has been spreading . Since the area of a disc grows with the square of its radius , then the number of infected cells should only grow in proportion to the time squared . This means that strict neighbor-to-neighbor contact infection will not support exponential growth . To maintain exponential expansion virus would have to spread to targets beyond those in proximal contact but , this virus would normally be removed by neutralizing antibody . However , Tugizov et al have shown the in vitro development of epithelial plaques through the cell to cell spread of infectious virus via a mechanism involving tightly connected lateral membranes , which is time dependent [37] and resistant to neutralizing antibody [38] . In addition they have shown in vitro that EBV can rapidly spread in epithelium by transcellular transcytosis potentially allowing the virus to spread rapidly beyond the adjacent cells allowing exponential expansion of the plaque ( Sharof Tugizov personal communication ) . A further result from the CDF analysis is that the slopes are similar but the range of shedding is different between subjects . This means that the exponential expansion of the plaques is being initiated at a different size for each subject but once started is constant between subjects . This fits nicely with the idea that very small numbers of B cells are releasing virus that initiates plaque formation and the number of epithelial cells that initially become infected is dependent on the particular subject . This variation could be introduced simply based for example on the efficacy of the neutralizing antibody each subject produces . A soon as epithelial cell infection is achieved the growth of the resulting plaque becomes subject independent again consistent with the results of Tugizov et al that epithelial infection is not affected by neutralizing antibody . If we assume that the lowest value for each subject approximates the initial size of the plaques then we can estimate ( assuming an epithelial cell burst rate of 6 . 2×105 per cell per day ) that the initial infection ranges between ∼0 and 5 epithelial cells . We have estimated above that the plaques grow for a maximum of one week and then survive stably for maximally another week and that there are on average about 3 plaques . If each plaque is seeded by a single B cell releasing virus then on average only about 3 such B cells are generated in the Waldeyer's ring every 1 to 2 weeks or 1 cell every 2 to 5 days . Although this seems extremely low it is consistent with our previous findings ( [26] and Box 1 ) . As described in Box 1 a screen of tonsils for expression of the immediate early lytic gene BZLF1 suggested that on average around 1 B cell every two days completes the lytic cycle ( see Box 1 for the calculation ) a result in remarkable concordance with the estimate from the CDF analysis . Another insight from the CDF analysis is that the exponential expansion is arrested at random time intervals to generate the simple linear step ladders observed in Figure 4 . Two explanations seem possible/likely . The first is that the heterogeneous structure of the epithelium imposes a random sized barrier on the expansion . This barrier could take the form of cells that were not infectable by EBV or otherwise prevented the virus from spreading . However this would require the epithelial cells to continue shedding relatively stable quantities of virus for several hours/days . Continuous virus production for several days without dying has not been documented for EBV in vitro however , we have virtually no information about EBV replication in vivo and such behavior has been described for another herpesvirus CMV [39] , [40] . The other process that could interrupt exponential expansion randomly over time could be the cytotoxic T cell response . It has been shown that tonsils contain large numbers of EBV specific CTL [14] . Often >1% of CD8 T cells in the tonsil are specific for EBV lytic antigens . That study suggested ∼10% of these CTL were activated at any given time indicating an ongoing immune response . The random log size distribution of the plaques could then simply reflect the random probability that a CTL will encounter the exponentially growing plaque over time and destroy it . This , at least in part , answers the question of why the virus is able to continuously replicate in Waldeyer's ring with impunity despite a persistent ongoing CTL response . A further contributing factor could be that the virus is known to posses immune evasion mechanisms that could delay or help evade the CTL response . These include an IL-10 homologue that suppresses the CTL response and gene products that interfere with antigen presentation ( reviewed in [41] ) . Presumably if small numbers of plaques are regularly being seeded from B cells these evasion mechanisms must be effective enough to allow the plaques to grow and survive for several days . Indeed the relatively stable short term shedding we have observed may simply reflect an extended period of time where the CTL response is only able to contain the spread of the plaque , it taking several days to finally eliminate it . If we are predicting that EBV is shed via epithelial plaques that range in size up to 105 cells why have they not been observed in immunohistochemical studies of tonsils ? The answer almost certainly relies on chance since our analysis predicts that plaques in the range of 104 to 105 cells are only present ∼10% of the time in a random sampling of subjects . Furthermore , these cells are likely contained within a single plaque in the entire Waldeyer's ring . Therefore our results predict that it would take exhaustive and comprehensive analysis of multiple tonsils to have a chance of seeing an infected plaque of epithelial cells . In summary , the dynamics of EBV shedding seem to favor a model ( Figure 9 ) where single B cells sporadically initiate EBV replication in Waldeyer's ring . The resultant virus seeds a local infection of epithelial cells that grows exponentially , for no more than 7 days , into an infected plaque . Plaque creation occurs at a rate such that there are 1 to 3 plaques of lytic epithelial cell infection at any one time . The size of these plaques depends on the initial number of epithelial cells infected and the ability of the infection to spread and grow . The size is ultimately bounded randomly either by the structural constraints of the epithelium and/or the efficacy of the immune response . Once the plaque has reached its mature size it produces , for a brief time ( up to 7 days ) , a constant rate of cells that initiate lytic replication probably representing a balance between infectious spread and immune containment . This continues for a few days until the plaque ceases production presumably due to destruction by the cellular immune response . Our results also relate to conclusions drawn previously by other groups . A long standing claim in the EBV field is that the level of infected cells in the blood correlates with the level of virus shedding in saliva [11] . We failed to confirm this finding both between and within individuals . Furthermore this discrepancy cannot be accounted for by the fact that we have used PCR to detect the virus whereas the previous study used a biological assay . We found no correlation whether we measured total shed genomes , DNase resistant genomes ( virions ) or viral binding . During the course of these studies we observed that , unlike virus produced by cell lines , the majority of saliva virus is encapsidated ( 10% versus 60% , respectively ) but most of it does not bind to cells ( <5% ) consistent with previous observations that most of the virus may be coated with neutralizing antibodies [21] . In conclusion , detailed studies of virus shedding indicate that healthy carriers of EBV consistently shed large amounts of the virus into saliva . Furthermore the dynamics of shedding suggest a crucial role for epithelial cells in amplifying the amount of shed virus . However in healthy carriers , although most of the shed virus is intact , very little of it is infectious .
This study was conducted according to the principles expressed in the Declaration of Helsinki . The study was approved by the Institutional Review Board of Tufts Medical School and Tufts Medical Center ( IRB study #731 ) . All subjects provided written informed consent for the collection of samples and subsequent analysis . Whole blood samples were obtained from health volunteers . PBMCs were isolated by Ficoll Hypaque centrifugation and B cells were purified by negative selection with a StemSep column following the manufactures instructions . The efficiency of purification was checked by staining for CD20 and FACS analysis . B cells were always 90–96% pure . Memory B cells were isolated from purified B cells using a MoFlo FACS sorter , based on staining with anti-CD27-FITC ( memory B cell marker ) and anti-CD20-PE ( pan B cell marker ) . The EBV positive lymphoblastoid line IB4 ( gift from Dr Elliot Kieff ) was used as a positive control and the murine CB59 T cell line ( gift from Dr Miguel Stadecker ) as negative control for EBV DNA PCR . The EBV negative Burkitt lymphoma line Akata and the gastric carcinoma cell line AGS ( gifts from Dr Lindsey Hutt-Fletcher ) were used as target cells for binding assays . The EBV positive B95-8 marmoset B cell ( gift of Dr Elliot Kieff ) , Akata 2A8 . 1 ( gift of Dr . Jeff Sample ) , AGSBX1 gastric carcinoma ( gift from Dr Lindsey Hutt-Fletcher ) and Hone1Akata nasopharyngeal carcinoma ( gift from Dr Ron Glaser ) cell lines were used for measuring the viral burst size . EBV genomes were detected by a Real time Taqman DNA PCR assay for the W repeat region as described previously [4] . This assay can detect a single copy of the EBV genome in a background of 106 EBV-negative cells and a single virion in 0 . 1 ml of saliva ( not shown ) . PCR was performed in a volume of 25 µl on the ABI Prism 5700 ( Perkin Elmer ) or MyIQ System ( BioRad ) . Thermal cycling was initiated with a denaturation step at 95°C for 3 min , and continued with 50 cycles of amplification at 95°C for 15 seconds and 60°C for 1 minute . For each reaction , 5 µl of DNA template was added to a master mix containing 12 . 5 µl of Universal TaqMan Master Mix ( Applied Biosystems or BioRad ) and 2 . 5 µl each of the forward primer , reverse primer and flurogenic probe . The primers used were AGTGGGCTTGTTTGTGACTTCA ( forward ) and GGACTCCTGGCGCTCTGAT ( reverse ) and the flurogenic probe was 6-FAM-TTACGTAAGCCAGACAGCAGCCAATTGTC-TAMRA . To control for variation in sampling over time , the levels of viral DNA in saliva samples were normalized to the levels of endogenous cellular DNA based on detection of the bcl-2 gene . Bcl-2 Sybr Green Real time PCR was performed in a volume of 25 µl . For each reaction , 5 µl of DNA template from shedding assay was added to a master mix containing 12 . 5 µl of Sybr Green Master Mix ( BioRad ) and 2 . 5 µl each of the forward , reverse primers and water . The primers used were CTTTAGAGAGTTGCTTTACGTG ( forward ) and TCCATATTCATCACTTTGACAA ( reverse ) . Thermal cycling was initiated with a denaturation step at 95°C for 3 minutes and continued with 45 cycles of amplification at 95°C for 15 seconds , 55°C for 15 seconds , 72°C for 30 seconds , and 83°C for 10 seconds . This was followed by melting curve analysis at 95°C for 1 minute , 55°C for 1 minute , and 80 cycles of 55°C for 30 seconds . To assess the efficiency of DNase treatment samples were assayed for the presence of cellular DNA by Real time Sybr Green PCR for the β-actin gene . PCR was performed in a volume of 25 µl . For each reaction , 5 µl of DNA template from saliva samples was added to a master mix containing 12 . 5 µl of Sybr Green Master Mix ( BioRad ) and 2 . 5 µl each of the forward , reverse primers and water . The primers used were GCGGGAAATCGTGCGTGACATT and GATGGAGTTGAAGGTAGTTTCGTG . Thermal cycling was initiated with a denaturation step at 95°C for 3 minutes and continued with 40 cycles of amplification at 95°C for 15 seconds , 60°C for 20 seconds , and 72°C for 25 seconds . This was followed by melting curve analysis at 95°C for 1 minute , and 80 cycles of 58°C for 20 seconds . The frequency of latently infected cells in purified whole B or memory B cell populations was measured by limiting dilution analysis of the cell population followed by detection of EBV infected cells by EBV specific Real time Taqman DNA PCR for the W repeat sequence as described above and in [4] . This PCR can reproducibly detect a single copy of the viral genome . Saliva sample were obtained from health volunteers . Except where noted , saliva samples were collected by mouth rinse and gargling with 5 ml of fluid . We have compared the efficiency of recovery using isotonic saline , PBS , RPMI 1640 and simple spring drinking water . We found that the recovery of viral DNA was the same regardless of the fluid used . Thereafter rinses were always performed with water and the salinity and pH adjusted by immediate addition of 10× PBS . For some experiments the saliva was clarified by centrifugation at 1500 rpm for 5 minutes at 4°C in an Allegra 6 bench top microfuge ( Beckman Coulter ) and/or treated with DNase ( see below ) . Samples were stored at −80°C until all time points were collected and then analyzed simultaneously . In control experiments we found no detectable change in recovery of viral DNA after such storage periods . 100 ul of the saliva sample were digested overnight with proteinase K ( PK ) ( Invitrogen , 0 . 5 mg/ml final concentration ) at 55°C in a 96-well plate ( 8–10 replicate wells per subject per time point ) . 5 ul of the digested sample was used as template for EBV DNA PCR as described after deactivating the PK by heating to 95°C for 10 minutes . The amount of viral DNA in each replicate was then estimated from a standard curve generated by serial dilution ( in duplicate ) of an aliquot from a defined B958 culture supernatant . The amount of viral DNA in this supernatant was previously defined by limiting dilution analysis and the supernatant aliquoted and stored frozen The average value from the 8–10 replicates was then used to calculate the absolute number of virions in the 5 ml saliva sample . DNase mix was pre-made and contained DNase ( Invitrogen ) , 10× DNase Buffer and water . 10 ul of DNase mix was added per well of 100 ul saliva sample to a final concentration of 1 U/ul unless otherwise noted . DNase treatment was performed at room temperature for 30 minutes . DNase was deactivated by adding EDTA ( 2 mM ) followed by incubation at 70–80°C for 30 minutes . Under these conditions virion DNA , defined as DNase resistant viral sequences , were stable for at least 1 hour whereas essentially all free cellular DNA ( >99% of β-actin ) was digested within 15 minutes ( see Figure S3 ) . The EBV negative Akata cell line and gastric carcinoma cell line AGS were used as target cells at 106 cells per ml . AGS cells were placed in suspension by brief treatment with trypsin . Eight replicates of each binding condition were set up in a 96-well plate . Saliva samples with or without prior clarification by centrifugation and with or without DNase treatment were added into wells of target cells . The binding reaction was then performed at 37°C for 30 minutes . Cells were then pelleted by centrifugation at 1500 rpm at 4°C for 10 minutes . Cell pellets were washed 3 times with 100 ul of PBSA ( 0 . 5% BSA in PBS ) per well and subsequently PK-digested at 55°C overnight . A serial dilution of B958 supernatant ( EBV-producing cell line ) with known numbers of virions was used as standard for the binding assay . Saliva of an EBV ( - ) subject and PBSA added onto target cells were used as negative controls . For induction of lytic replication , the B95-8 B cell line was stimulated with PMA at 20 ng/ml , the Akata 2A8 . 1 B cell line was stimulated with a combination of PMA ( 20 ng/ml ) and Ionomycin ( 1 or 10 ug/ml ) [42] , the AGSBX1 epithelial line was stimulated with a combination of PMA ( 30 ng/mL ) and Butryic Acid ( 2 . 5 mM ) [29] and the Hone1Akata epithelial line with PMA ( 40 ng/mL ) and Butryic Acid ( 3 mM ) [43] . Prior to stimulation , the cell lines were synchronized by reseeding in media overnight at 5×105/ml for B cells and 3–5×106/ml for epithelial cells . B cells were harvested at 48 hours for intracellular VCA ( viral capsid antigen ) staining and FACS analysis and culture supernatants were harvested for EBV specific DNA PCR at 72 hours . Epithelial cells were harvested for VCA staining after 96 hours and for PCR from 96–174 hours . For intracellular VCA staining , harvested cells were washed once with PBS and pelleted by centrifugation at 1300 rpm for 7 minutes at 4°C . Cell pellets were resuspended at 106 cells per 200 ul of fixing buffer ( 4% formaldehyde in PBS ) and incubated at room temperature for 10 minutes . Once the cells were fixed , 1 ml of PBS was added per 106 cells and the cell pellets were dissociated by vortexing , and then re-pelleted by centrifugation . The PBS wash step was repeated again before resuspension into 200 ul of permeabilization buffer ( 8% FBS , 0 . 02% NaN3 , 0 . 04% Saponin , 2% Goat Serum in PBS ) per 106 cells , followed by incubation at 4°C for 30 minutes . All staining was performed in FACS tubes at 106 cells per tube . Cells were stained with either 200 ul of mouse IgG2a isotype control ( Caltag ) or 200 ul of mouse anti-VCA-IgG2a antibody ( Capricorn ) per 106 cells diluted 1∶1000 in permeabilization buffer and incubated at 37°C for 30 minutes . The cells were then washed twice with 1 ml wash buffer ( 8% FBS , 0 . 02% NaN3 , 0 . 04% Saponin in PBS ) and pelleted . Next , cells were resuspended in 100 ul of a 1∶10 , 000 dilution of Alexa Fluor 488 or 647 conjugated goat anti-mouse IgG ( H+L ) ( Molecular Probes ) diluted in permeabilization buffer and incubated at 37°C for 30 min in the dark . Then , the cells were washed twice by vortexing in 2 ml of wash buffer and pelleted by centrifugation . Finally , cells were resuspended in 300 ul of FACS buffer ( 8% FBS , 0 . 02% NaN3 in PBS ) for analysis on a FACS Caliber flow cytometer or 60 ul of FACS buffer for analysis on an Amnis ImageStream with IDEA analytical software to confirm the authenticity of the staining . EBV DNA content of the culture supernatants was measured by DNA PCR with or without DNase treatment as described for the saliva samples . Simulations and statistical analysis ( CDF and adjusted R2 linear regression analysis ) were performed with MatLab software . To study linearity in the expected CDF for a limited number of data points with various numbers of hypothetical plaques a simulation was run . In this simulation plaques were arbitrarily assigned to undergo exponential growth . Growth of the plaques was initiated randomly and the sum of infected cells assessed at 20 random time points picked using a Monte Carlo algorithm . For details see Figure S2 and associated text . | Epstein-Barr virus is a human pathogen associated with several human cancers that nevertheless persists benignly as a latent infection in the majority of adults . EBV persistence is characterized by the presence of latently infected cells in the blood and the shedding of virus into saliva . We present the first systematic quantitative analysis of virus shedding . We show , contrary to what was previously thought , that shedding is continuous and at a high level for all subjects tested . This constant presence of infectious virus may be a crucial risk factor in the development of the EBV-associated tumor nasopharyngeal carcinoma . Unlike infected cells in the blood , which are maintained at very stable levels for years , we show that virus shedding is highly variable such that at any time any individual may be a relatively high or low shedder . We have analyzed these dynamics mathematically and with a simple simulation model . We find that they can be explained by a simple exponential function which we hypothesize is the expansion of 1–3 plaques of epithelial cells . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"virology/persistence",
"and",
"latency"
] | 2009 | The Dynamics of EBV Shedding Implicate a Central Role for Epithelial Cells in Amplifying Viral Output |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.