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 |
|---|---|---|---|---|---|
RNA virus populations will undergo processes of mutation and selection resulting in a mixed population of viral particles . High throughput sequencing of a viral population subsequently contains a mixed signal of the underlying clones . We would like to identify the underlying evolutionary structures . We utilize two sources of information to attempt this; within segment linkage information , and mutation prevalence . We demonstrate that clone haplotypes , their prevalence , and maximum parsimony reticulate evolutionary structures can be identified , although the solutions may not be unique , even for complete sets of information . This is applied to a chain of influenza infection , where we infer evolutionary structures , including reassortment , and demonstrate some of the difficulties of interpretation that arise from deep sequencing due to artifacts such as template switching during PCR amplification .
RNA viruses have evolutionary dynamics characterized by high turnover rates , large population sizes and very high mutation rates [1 , 2] , resulting in a genetically diverse mixed viral population [1 , 3–5] . Subpopulations in these mixtures containing specific sets of mutations are referred to as clones and their corresponding mutation sets as haplotypes . Unveiling the diversity , evolution and clonal composition of a viral population will be key to understanding factors such as infectiousness , virulence and drug resistance [6] . High throughput sequencing technologies have resulted in the generation of rapid , cost-effective , large sequencing datasets [7] . When applied to viruses , the set of reads obtained from a deep sequencing experiment represents a sample of the viral population which can be used to infer the underlying structure of that population at an unprecedented level of detail [1] . In this study , we aim to identify the haplotypes of clones and quantify their prevalence within a viral population . The method also constructs evolutionary histories of the process consistent with the data . Reconstructing the structure of a mixed viral population from sequencing data is a challenging problem [8] . Only a few works address the issue of viral mixed population haplotype reconstruction which infer both the genomes of sub-populations and their prevalence . Reviews of the methods and approaches dealing with these issues can be found in [1 , 9–11] and [12] . These works frequently make use of read graphs , which consist of a graph representation of pairs of mutations linked into haplotypes [13] . Haplotypes then correspond to paths through these graphs , although not every path will necessarily be realized as a genuine haplotype , which can lead to over-calling haplotypes . Different formalizations of this problem has led to different optimization problems in the literature [11] , including minimum-cost flows [14] , minimum sets of paths [13 , 15] , probabilistic and statistical methods [8] , network flow problems [3 , 16] , minimum path cover problems [17] , maximizing bandwidth [18] , graph coloring problems [19] or K-mean clustering approaches [13] . After the haplotypes are constructed , in many cases an expectation-maximisation ( EM ) algorithm is used to estimate their prevalence in the sampled population . Some other works [6 , 20] use a probabilistic approach instead of a graph-based method . In this work we take an integrative approach to address both the genetic diversity and the evolutionary trajectory of the viral population . The method presented is not read graph based and constructs evolutionary trees and recombination networks weighted by clone prevalences . This reduces the size of the solution set of haplotypes . The method does not rely exclusively on reads physically linking mutations so is applicable to longer segments . The method will also be shown to have particular utility with time series data and is highlighted on a chain of infections by influenza ( H3N8 ) . The question of the influenza genome diversity has been addressed in the literature largely between strains or samples from different hosts , considering one single dominant genome for each host [21] . Within-host evolution is a source of genetic diversity the understanding of which may lead to the development of models that link different evolutionary scales [10] . Kuroda et al . [22] addressed the question of evolution within a single host of influenza extracted on a patient who died of an A/H1N1/2009 infection , but with a focus on HA segment using a de-novo approach . Our approach provides a method to further understand within host evolution of such viruses . The next section highlights the approach with an overview of the methods used , examples of the tree and network construction methods with simulated data , followed by an application of the method to a daily sequence of real influenza data . The Methods section describes the construction of the trees and networks in more detail .
Here we provide a brief outline of our general approach , which is explained in more detail in subsequent sections . The aim of our approach was twofold . Firstly , identify mutations arising in an infection chain of viral hosts . Secondly , provide a phylogenetic tree or recombination network that best explains the evolution structure of the most prevalent mutations . A flow chart of the process is provided in Fig 1 . The first step was to sequence a time series of virus samples . Next , we took the initial sample in the time series ( such as an inoculum , for example ) and obtained the nucleotide with highest prevalence at each position . This defined a reference ( concensus ) sequence , which was then compared to the remaining samples enabling the identification of de novo mutations . The proportion of reads containing each mutation then represented its prevalence in the viral population ( paired reads with identical start positions , end positions and sequences were counted once in this process , being assumed to derive from a single PCR product ) . Mutations with a prevalence above a ( user defined ) threshold ( 10% ) on at least one sample were included for further analysis . Next we attempted to construct an evolutionary tree consistent with the prevalence information and paired end linkage information across all samples . If no consistent tree could be identified , we attempted to construct a recombination network with one reticulation event . If this failed we iteratively increase the number of reticulation events permissable to find a recombination network consistent with the data . Any solutions obtained were outputted in either the Newick format ( for trees ) or the extended Newick format ( for networks ) . Our method used two main sources of information . Firstly , a pigeon hole principle was utilized , restricting how different sub-populations of viruses , each containing a certain combination of mutations , can fit within a tree or network structure . Secondly , linkage information was harnessed , describing how pairs of mutations co-exist in sub-populations . This information is obtained from single paired-end reads ( likely to derive from a single viral particle ) that contained two ( or more ) mutations . The pigeon hole principle worked best with a set of mutation prevalences that vary significantly across the samples collected . More specifically , a subset of mutations undergoing either rapid selection or drift were found to provide the most informative datasets ( RNA viruses undergoing drug treatments , or the bottleneck arising when a small number of viral particles infects an animal are examples of where this might happen ) . Mutations that have consistently low or high prevalence contain information that is harder to leverage , and the underlying evolutionary structures are harder to infer . Such mutations were not included in the analysis . Slowly mutating viruses ( DNA viruses for example ) are also less likely to be sufficient mutagenic for our approach . The linkage information worked best when recombination events were relatively rare . Viruses with high rates of recombination ( such as HIV ) will rapidly lose linkage information making the evolutionary structure harder to identify . We next outline the tree and network construction methods in more detail . Consider the pedagogic simulation in Fig 2 , where we have a region of interest ( such as an influenza segment , for example ) that has undergone mutational and selective processes encapsulated by the evolution tree in Fig 2A . This tree contains five mutations M1 , M2 , M3 , M4 , M5 that lie on various branches of the tree . These combine into the six clones that are the leaves of the tree . For example , the second leaf is labeled C11000 , indicating a clone with haplotype consisting of mutations M1 , M2 but not M3 , M4 , M5 . Note that the path from the root of the tree to this leaf crosses the two branches corresponding to mutations M1 and M2 . The number 20 at the leaf indicates that this clone makes up 20% of the viral population , and is termed the prevalence . Note that these prevalences form a conserved flow network through the tree [23] . For example , the prevalence of mutation M1 is 55% , which accounts for the two haplotypes C11001 and C11000 , with prevalences 35% and 20% , respectively . In general , we find that the prevalence flowing into a node of the tree must equal the sum of the exiting prevalences . This represents conservation of the viral sub-populations . The total prevalence across all the leaves is therefore 100% . In reality we are not privy to this information and perform a sequencing experiment to investigate the structure . This takes the form of molecular sequencing , where we detect the five mutations , which each have a different depth of sequencing , as portrayed in Fig 2B . We will later see with real influenza data that percentage depth can be reasonably interpreted as prevalence . Furthermore , we can group mutations arising on individual sequencing reads into clusters . For our example , mutations M1 , M2 and M3 are positioned such that there are paired end reads ( exemplified in Fig 2Bi ) where M1 and M2 will lie on the sequence at one end , and M3 in the sequence at the other end of the read . Mutations M1 , M2 and M3 thus form one cluster . Similarly , mutations M4 and M5 can be found at the two ends of paired reads and form a distinct cluster . We then find the mutations are grouped into two clusters , giving the two corresponding haplotype tables in Fig 2C . We first construct an evolution tree for each of these tables . Our approach is based upon two sources of information; one utilizes mutation sequencing depth with a pigeon hole principle , the other utilizes linkage information from haplotype tables . Now we have mutation M2 present in 80% of viruses and mutation M1 present in 55% of viruses . If these mutations are not both simultaneously present in a sub population of viruses , then the mutations are exclusive . This implies the two populations of size 80% and 55% do not overlap . However , the total population of viruses containing either of these viruses would then be greater than 100% < 80% + 55% . This is not possible , and the only explanation is that a subpopulation of viruses contain both mutations; the pigeon hole principle . The only tree-like evolutionary structure possible is that M1 is a descendant of M2 , as indicated by the rooted , directed tree in Fig 2Di . Note that we have not utilized any haplotype information to infer this , just the mutation prevalence of the two mutations and a pigeon hole principle . Mutation M3 has a prevalence that is too low to repeat a prevalence based argument . However , we have a second source of information; the paired read data that can link together mutations into the haplotypes in Fig 2Ci . This table is based on three mutations , which group into 23 = 8 possible haplotypes . However , a tree structure with three mutations will only contain four leaves [24] and we see that four of the halpotypes ( emboldened ) have notably larger counts of reads and are likely to be genuine . The four haplotypes with a notably lower read counts are likely to be the result of sequencing error at the mutant base positions , or template switching from a cycle of rtPCR , and are ignored . The presence of genuine haplotypes C011 and C010 , lead us to conclude that M3 is descendant from M2 but not M1 , resulting in the tree of Fig 2Di . From the mutation prevalences 55% , 80% and 15% of M1 , M2 and M3 , we can also use the conserved network flow to measure the haplotypes prevalence . For example , the leaf descending from M2 ( 80% ) , but not M1 ( 55% ) or M3 ( 15% ) ( clone C010 of Fig 2Di ) must represent the remaining 10% = 80% − 55% − 15% of the population . This provides us with two sources of information ( sequencing depth and linkage information ) we can utilize to reconstruct the clone haplotypes , prevalence , and evolution . However , not all mutations can be connected by sequencing reads . They may be either separated by a distance beyond the library insert size , or may lie on distinct ( unlinked ) segments . Our approach is then as follows . We first construct a tree for each cluster of linked mutations . This will be a subtree of the full evolutionary structure . We then construct a supertree from this set of subtrees . Now both of the trees in Fig 2D must be subtrees of a full evolutionary tree for the collective mutation set so we need to construct a supertree of these two trees . We can do this recursively as follows . We take the mutations and place them in decreasing order according to their prevalence , as given in Fig 2E . We then attach branches corresponding to these mutations to the supertree in turn , checking firstly network flow conservation , and secondly that the haplotype information in the subtrees is preserved . The steps for this example can be seen in Fig 2F . We start with a single incoming edge with prevalence 100%; the entire viral population . We next place an edge corresponding to M2 , the mutation with maximum prevalence of 80% . The next mutation in the tree either descends from the root or this new node . Any descendants of M2 must have a prevalence less than this 80% . Any other branches must descend from the top node but can only account for up to 20% of the remaining population . These two values are the capacities indicated in square brackets . The next value we place is M1 with prevalence 55% . This is beyond the capacity 20% of the top node , so M1 is descendant to M2 , accounting for 55% of the 80% , leaving 25% . We thus have a three node tree with capacities 20% , 25% and 55% . The third ordered mutation M5 has prevalence 35% , which can only be placed at the bottom node with maximum capacity . Our next mutation M3 has a prevalence 15% that is less than any of the four capacities available , and no useful information on the supertree structure is obtained . This branch is the first to use haplotype information . We know from the first subtree that the corresponding branch is a descendant of M2 and not M1 . The only node we can use ( in red ) has capacity 25% and we place the branch . For the final branch corresponding to mutation M4 , the prevalence 15% is less than four available capacities . The second subtree tells us that M4 is not a descendant of M5 . This only rules out one of the four choices , and any of the three ( red ) nodes will result in a tree consistent with the data . The top node selected results in a tree equivalent to that in Fig 2A . To see this tree equivalence , the internal nodes in the last tree of Fig 2F have additional leaves attached ( dotted lines ) to obtain Fig 2A . We thus find that a single dataset can result in several trees that are consistent with the data . However , having a time series of samples means a tree consistent with all days of data is required , which will substantially reduces the solution space . Note that the prevalences of the clones at the leaves of the tree results from this recursive process . We thus find that supertree construction is relatively straightforward with the aid of prevalence . However , trees do not always fit the data . This can be due to recombination occurring within segments , or re-assortment occurring between segments . In the next section we construct recombination networks to cater for this , although we will see that they cannot be constructed as efficiently as trees . In Fig 3A we see another simulated evolution based upon the two segments in Fig 3Ci that accumulate four mutations , M1 , M2 , M3 and M4 . First we have mutations M1 and M3 . Then we have the first of two recombination events , r1 , where we have recombination within the first segment as described in Fig 3Cii . We then have mutations M2 and M4 , followed by the second recombination event r2 in Fig 3Ciii , a re-assortment between the two segments . This results in the seven clones given at the leaves of Fig 3A . The prevalences of the four mutations across five time points are given in Fig 3E . Note that we no longer have the conservation of prevalence observed in trees . For example , mutations M1 and M3 are on distinct branches extending from the root , yet their total prevalence is in excess of 100% ( on Day 5 for example ) . This is due to recombination r1 resulting in the presence of a clone containing both mutations . The use of the prevalence to reconstruct this structure from observable data thus requires more care . Now we see in Fig 3Ci that the four mutations cluster into two groups of mutations each bridged by a set of paired reads , resulting in two tables of read counts in Fig 3Bi and 3Bii . We would like to reconstruct the evolution in Fig 3A from these data . Firstly , we need to decide which of the haplotypes in Fig 3B are real . The haplotypes with consistently low entries are classified as artifact ( in opaque ) . We next use a standard approach ( such as a canonical splits network [25] ) to construct sub-networks from the real haplotypes in each of these tables , such as those given in Fig 3Dii . We then build super-networks ensuring that all sub-networks are contained as a sub-graph . There does not appear to be an efficient way of doing this ( such as ordering by prevalence which works so well with trees ) so a brute force approach is taken , where we construct all possible networks that contain four mutations and the haplotypes observed in Fig 3Bi and 3Bii . This results in many candidate super-networks . We now find that the prevalence information can be used to reject many cases . For example , the super-network in Fig 3Hi contains both sub-networks of Fig 3Di and 3Dii as subgraphs . Note that the root node , representing the entire 100% of the population , has daughter branches containing mutations M4 , M1 and M3 . However , from the prevalences on Day 5 we see that M4 has prevalence 66% and M1 and M3 ( which recombine ) have a collective prevalence ( from clones C001 , C100 , C101 , and C111 in Fig 3Bi ) of 93% . This is in excess of the possible 100% available and the network is rejected . Application of filtering by prevalence ( see Methods section for full details ) rejects all networks with one recombination event , so we try all networks with two recombination events , resulting in just seven possible recombination networks . These all contain the same set of clones , all of which correspond to the single phylogenetic network in Fig 3F . Although only one recombination event is present across the subnetworks , all super-networks with one recombination event were filtered out and two recombination events were required . Lastly we require estimates of the prevalences of each of the seven clones . We would like to match these to the prevalences in the tables of Fig 3B . This is a linear programming problem , the full details of which are given in the Methods section . The resulting estimates are given in Fig 3G where we see that some clones have point estimates , whereas others have ranges . For example , we see that clone C0010 has a point estimate for each day . This is because it is the only clone of the super-network that corresponds to clone C001 of Fig 3Bi and their prevalences can be matched . Conversely , we see ranges for the prevalences of clones C1110 and C1111 . This is because both clones correspond to clone C111 of Fig 3Bi and prevalence estimates for each clone cannot be uniquely specified . Full details of this approach can be found in the Methods section . In the next section we describe the results obtained when applying these methods to a time series of influenza samples . The data used in this study were generated from a chain of horse infections with influenza A H3N8 virus ( sample processing details can be found in the methods section ) . An inoculum was used to infect two horses labeled 2761 and 6652 . These two animals then infected horses labeled 6292 and 9476 . This latter pair then infected 1420 and 6273 . The chain continued and daily samples were collected from the horses resulting in 50 samples in total . For the present study we used 16 samples; the inoculum and hosts 2761 ( days 2 to 6 ) , 6652 ( days 2 , 3 and 5 ) , 6292 ( days 3 to 6 ) and 1420 ( days 3 , 5 and 6 ) . Influenza A virus is a member of the family Orthomyxoviridae which contains eight segmented , negative-stranded genomic RNAs commonly referred to as segments and numbered by their lengths from the longest 2341 to the shortest 890 bps [21] , as summarized in Fig 4A . Daily samples were collected from each host and paired end sequenced was performed with Hi-seq and Mi-seq machines . The samples sequences were aligned with Bowtie2 [26] with default parameters . We obtain for each sample a SAM file containing mapping information of all the different reads in the sample . Any mapped read whose average Phred-quality per base was less than 30 were discarded . In order to identify mutations from real data we need a reference sequence to compare the read sequences to . Consistent differences between the two can then be classified as a mutation . We constructed a majority consensus sequence from the inoculum sample . This consensus sequence was then used as a starting reference for the chain of infected animals . To produce DNA for sequencing , viral RNA was reverse-transcribed and amplified ( RT-PCR ) . The reverse transcription step can result in the introduction of artefact mutations that in turn would be further amplified in the PCR step , resulting in different levels of amplification and mutation . This in turn is likely to introduce significant differences between the sequencing depth and prevalence . To combat this , all identical paired end reads ( with equal beginning and endpoints , and identical sequences ) were grouped , classified as a single PCR product , deriving from a single molecule and only counted once . We tested this assumption by using the observed insert size distribution to randomly simulate reads with a number equal to the observed depth ( ∼ 1 . 8e6 reads ) , assuming a single mutation with prevalence of 50% exists , to determine how often two distinct events would produce identical reads . This produced a surprisingly high figure of 7% which will get worse as the depth of sequencing increases and some care is needed ( see [27] for further discussion on these kind of ‘birthday paradoxes’ ) . However , many reads contain more than one mutation making identical sequences less likely and the real figure will be somewhat lower . The depth of sequencing with these adjusted counts should then provide an improved measurement of the prevalence of viral subpopulations . We compared an identical sample that was sequenced separately ( following the RT-PCR step ) , the results of which can be seen across two samples in Fig 4Bi , 4Bii , 4Ci and 4Cii . Both the position and prevalence of mutations were reproducible to good accuracy suggesting proportional sequencing depth is a good surrogate for prevalence . We note that although there was variation in the depth of sequencing across the genome , the expected proportion of reads containing any given mutation will not change , and the depth of sequencing will not be a large source of bias on the estimated prevalence . However , we cannot rule out the possibility that some mutations are preferentially amplified , which would cause some systematic bias . We thus make the cautionary observation that some biases may exist in the prevalence , and that spike-in experiments to systematically examine the strength of correlation between sequence depth and viral prevalence are needed . Such experiments are beyond the scope of the present study and proportional sequence depth was taken as a suitable proxy for proportional viral prevalence . We then applied the methods to sets of high prevalence mutations in each of the eight segments individually , and also to a set of three mutations from distinct segments . The main observations are below . For segments 1 , 3 , 5 , 6 , 7 and 8 we obtained tree like evolutions for the segments . In all cases the mutations involved lay on distinct branches and were indicative of mutations arising in independent clones . Segment 6 can be seen in Fig 5A , where we see five mutations on six branches . We also see from the stacked bar chart in Fig 5B that many of the mutations arose during different periods in the infection chain . However , the evolution structure of mutations within segments did not always appear to be tree like , with segment 2 containing one putative recombination event and seven mutations , and segment 4 containing three putative recombination events and six mutations . This latter case arose because we found three pairs of mutations in putative recombination . Using nucleotide positions as labels , these were ( 431 , 674 ) , ( 431 , 709 ) and ( 709 , 1401 ) . That is , we found significant counts of all four combinations of mutations , labeled C00 , C01 , C10 and C11 , lying on paired reads . Examples of typical counts for three ( out of sixteen ) samples are given in the top table in Fig 6B ( see Supplementary Information for full details ) . If the evolution is tree-like , reads from one of the types C01 , C10 or C11 should only arise as an artifact . Note that we have high read counts of all four categories , which is indicative of recombination . However , various studies have shown that there is very little evidence of genuine recombination that occurs within segments of influenza [28–30] , and these kind of observations can arise from template switching across different copies of segments during the rtPCR sequencing cycle [31] . We developed an analytic approach to consider this possibility in more detail . Now if the true underlying structure is tree-like , it suggests that one of C11 , C01 or C10 arises purely from template switching ( the wild type C00 is assumed to always occur ) . This gives us the three models ( labeled i-iii in Fig 6A and 6B ) to consider ( see also Template Switching subsection in Methods section ) . We let a , b and c be the population proportions of the three real genotypes . We let n be the probability that a cycle of rtPCR causes template switching . We then treat template switching as a continuous time three state random process . This allows us to derive probabilities that genotypes C00 , C01 , C10 and C11 arise on paired end reads , as given in Fig 6A ( see Methods for derivations ) . The counts of the four classes of read then follow a corresponding multinomial distribution . Maximum likelihood was used to estimate parameters , obtain log-likelihood scores , and a chi-squared measure of fit was obtained for each of the three models . For the pair ( 431 , 674 ) we found that the best log-likelihood , on all sixteen sampled days , was Model 1 ( Fig 6i ) , where reads of type C11 are artifacts arising from template switching alone . The parameters obtained provided an almost perfect fit; the expected counts were almost equal to the observed counts and the goodness of fit significance values were close to 1 . Models 2 and 3 ( Fig 6ii and 6iii ) had substantially lower likelihoods and significantly bad fits . This tells us that if the underlying structure is a tree , it involves the three genotypes C00 , C01 and C10 and mutations 431 and 674 lie on distinct branches . For the pair ( 431 , 709 ) we found that the best log-likelihood , over the sixteen sampled days , was Model 3 ( Fig 6iii ) , where reads of type C10 are artifacts . The parameters obtained provided an almost perfect fit on most days with goodness of fit significance values close to 1 . A couple of days had relatively poor fits , but were not significant when multiple testing across all sixteen days was considered . Model 1 ( with C11 as an artifact , Fig 6i ) had very similar likelihoods , but the data exhibited significantly poor fits on multiple days . Model 2 ( with C01 as an artifact , Fig 6ii ) performed very badly . This tells us that if the underlying structure is a tree , it involves the three genotypes C00 , C01 and C11 and mutation 431 is a descendant of 709 . For the pair ( 709 , 1401 ) we found that the best log-likelihood , on all sixteen sampled days , was Model 2 ( Fig 6ii ) , where reads of type C01 are artifacts . The parameters obtained provided an almost perfect fit on all days with goodness of fit significance values close to 1 . Models 1 and 3 ( Fig 6i and 6iii ) performed very badly . This tells us that if the underlying structure is a tree , it involves the three genotypes C00 , C10 and C11 and mutation 1401 is a descendant of 709 . Thus the three cases where data are indicative of recombination can be explained purely by template switching during rtPCR . This is reinforced somewhat by the fact that the same model emerged across all sampled days for each mutation pair . However , this does not definitively rule out recombination , which could also exhibit these consistent patterns across sampled days , and so care is needed when interpreting data . Furthermore , the rates of template switching required to explain the data without recombination were not always consistent . For example , in the sample from host 2761 Day 4 , the estimated template switching between mutations ( 431 , 674 ) was 43 . 6% ( 95% c . i 38 . 2%—49 . 2% ) . Between mutations ( 431 , 709 ) it was 48 . 8% ( 95% c . i 46 . 3%—51 . 4% ) , giving reasonable agreement . Between mutations ( 709 , 1401 ) it was somewhat higher , at 58 . 6% ( 95% c . i 68 . 3%—78 . 7% ) , although this may be expected due to the greater distance between the mutations . However , in sample 1420 Day 3 , the template switching rate for the pair ( 431 , 674 ) , at 99 . 6% ( 95% c . i 76 . 6%—99 . 3% ) , was notably higher than both the mutation pair ( 431 , 709 ) , at 43 . 2% ( 95% c . i 39 . 5%—47 . 1% ) , and mutation pair ( 709 , 1401 ) , at 49 . 3% ( 95% c . i 36 . 9%—64 . 1% ) . Although differences between samples ( and so library preparations ) may be expected , differences such as this in the same library are harder to explain without implicating genuine recombination . We thus have two explanations of the data; genuine recombination or template switching artifacts . We consider both cases and then draw comparisons . Firstly we consider segment 4 assuming recombination has taken place . The results can be seen in Fig 7 . The prevalences of six mutations of interest are given in Fig 7A . Reasonable linkage information was available across the segment , including the two haplotype tables in Fig 7C . The first is linkage information between mutations 709 and 1401 , where all four combinations of mutation occur to reasonable depth , implying recombination between the mutations . The second is between mutations 1387 and 1401 , where we see only three haplotypes occur to significant depth , suggesting a tree like evolutionary structure between the two mutations . The full set of tables is in Supplementary Information . Although the sequencing depth in the first table is lower , due to the rarer occurrence of sufficiently large insert sizes , the information gleaned is just as crucial . The most parsimonious evolution found involved three recombination events , resulting in the single cloneset contained in the phylogenetic network given in Fig 7D . There were 22 possible recombination networks that fit this phylogenetic network , one example of which is given in Fig 7E . The relatively complete linkage information resulted in point estimates for the clone prevalences ( rather than ranges ) , as given in Fig 7B . If we now assume that the recombination like events are template switching during rtPCR , then from above , we observed that mutations 431 and 674 are on distinct branches , mutation 431 is a descendant of 709 , and 1401 is also a descendant of 709 . This resolves all three reticulation events in the network of Fig 7E and we end up with the tree given in Fig 7F . However , this structure still has two minor conflicts . Firstly , the tree like structure suggests that mutation 431 should have a lower prevalence than 1401 , and on most days it does . However , the sample from host 2761 Day 4 has prevalences 54 . 3% and 51 . 2% for mutations 431 and 1401 , respectively . Similarly , the samples from host 1420 Day 3 are 66 . 1% and 70 . 3% , respectively . Secondly , the four mutations 674 , 709 , 1013 and 1401 all descend from the root on separate branches and should have a total prevalence that is less than 100% and on fifteen of sixteen samples this is true . However , on sample 6292 Day 3 the prevalences are 9 . 9% , 52 . 8% , 18 . 0% and 26 . 0% , which combine to 106 . 7% . Although the conflicts are relatively small , these differences are larger than would be expected from Poisson sampling of such deep data . However , this is the most plausible tree structure we found . Re-assortments occur when progeny segments from distinct viral parents are partnered into the same viral particle , resulting in a recombinant evolutionary network . Now re-assortment is a form of recombination . This is usually possible to detect in diploid species such as human because linkage information is available across a region of interest , such as a chromosome , and recombination can be inferred . Furthermore human samples have distinct sequencing samples for each member of the species . Inferring re-assortment across distinct viral samples is more difficult because firstly we do not have linkage information across distinct segments , and secondly , we have mixed populations within each sample . However , we show that re-assortment can still be detected within mixed population viral samples with the aid of information provided by prevalence . Consider Fig 8 . We have three mutations in segments 2 , 3 and 4 , along with their mutation nucleotide positions 2037 , 201 and 709 , respectively . We refer to the mutations as S 2_2037 , S 3_201 and S 4_709 accordingly . We see in Fig 8A that S 2_2037 and S 3_201 have prevalences that alternate in magnitude across the 16 days sampled . If we assume a tree like structure , these two mutations cannot lie on a single branch , because one prevalence would have to be consistently lower than the other; they must therefore lie on distinct branches . Now mutation S 4_709 can; i ) be on a distinct third branch , ii ) be a descendant of S 2_2037 , iii ) be a descendant of S 3_201 , iv ) be an ancestor of S 2_2037 , v ) be an ancestor of S 3_201 , or vi ) be an ancestor of both . We can rule out all of these choices as follows . Firstly we note that S 4_709 has a prevalence that is consistently larger than that of S 2_2037 or S 3_201 , so cannot be a descendant of either mutation , ruling out ii ) and iii ) . We see from sample 6292 Day 3 that S 2_2037 and S 3_201 have a total prevalence greater than S 4_709 , meaning S 4_709 cannot be an ancestor of both mutations , ruling out vi ) . In this sample , the total prevalence of all three mutations is in excess of 100% , ruling out i ) . Now if S 4_709 and S 2_2037 lie on distinct branches , we see from 2761 Day 4 that their combined prevalence is in excess of 100% , ruling out v ) . Finally , if S 4_709 and S 3_201 lie on distinct branches , we see from 6292 Day 4 that their combined prevalence is in excess of 100% , ruling out iv ) . No tree structure is possible and we conclude the presence of re-assortment as the most likely explanation . In fact , application of the full method reveals that two re-assortment events are required to explain the data . This results in 51 possible recombination networks , one such example is given in Fig 8B . These correspond to the four clonesets given in Fig 8C , arising from two possible phylogentic networks . The four clonesets have prevalences that could not be uniquely resolved; their possible ranges are shown in Fig 8D . Although we cannot uniquely identify the network or the prevalences , all solutions involved two re-assortments , one involving mutations S 4_709 and S 2_2037 , the other involving S 4_709 and S 3_201 . This observation was only possible because of inferences made with the prevalence .
We have introduced a methodology to analyze time series viral sequencing data . This has three aims; to identify the presence of clones in mixed viral populations , to quantify the relative population sizes of the clones , and to describe underlying evolutionary structures , including reticulated evolution . We have demonstrated the applicability of these methods with paired end sequencing from a chain of infections of the H3N8 influenza virus . Although we could identify underlying evolutionary structures , some properties of the viruses and the resulting data make interpretation difficult . In particular , template switching during the rtPCR cycle of sequencing an RNA virus is known to occur , and can result in paired reads that imply the presence of recombination . Although any underlying tree like evolutions can still be detected , these artifacts confound the signal of any genuine recombination that may be taking place , making it harder to identify . The prevalence of mutations , measured as sequencing depth proportion , offers an alternative source of information that can help resolve these conflicts in theory , although more work is needed to evaluate how robust this metric is in practice . For example , although tree like evolutions were identified in six of the segments , in the two remaining segments the approach found reticulated networks , with three distinct reticulated nodes in the hemagglutinin segments network . Although each of these nodes were consistent with template switching artifacts , the resultant tree structure could not quite be fitted to the mutation prevalences . Although this conflict implies the original network is correct and recombination has taken place , within segment recombination in influenza is rare [28–30] and other explanations may be required . In particular , we note in Fig 4B that there are slight differences between the prevalences obtained from independent Mi-seq and Hi-seq runs . Although some of this will be due to Poisson variation of depth , there could be some biases in PCR over certain mutations , for example . The application of prevalence thus needs to be used with caution , and further studies are needed to fine tune this type of approach . When the approach was applied to mutations in distinct segments , two re-assortment events were inferred . The differences in mutation prevalences were more marked in this case suggesting the inference is more robust and re-assortment more likely to have taken place . This is also biologically more plausible , with events such as this accounting for the emergence of new strains . We note that although re-assortment may have genuinely taken place , only one of the original clones ( containing just mutation 709 on segment 4 ) survived the infection chain and a longitudinal study would not have picked up such transient clonal activity . These methods utilized paired end sequencing data and showed that even when paired reads do not extend the full length of segments , or bridge distinct segments , we can still make useful inferences on the underlying evolutionary structures . The two main sources of information are the linkage offered by two or more mutations lying on the same paired reads , and the prevalence information . We note , firstly , that utilising the full range of insert sizes produced in the sequencing library provides linkage information that covered most distances across segments . Filtering paired reads to remove inserts with larger insert sizes can lose useful linkage information . Indeed , it is likely to be profitable to produce libraries with different insert sizes . Secondly , we note that it is by utilizing the variability of the prevalence in a time series dataset that we can narrow down the predictions to a useful degree; application of this method to individuals days will likely result in too many predictions to be useful . Furthermore , this has greatest application to mutations of higher prevalence; this places more restrictions on possible evolutions consistent with the data . Subsequently , this kind of variability is most likely to manifest itself under conditions of differing selectional forces . A stable population is less likely to contain mutations moving to fixation under selective forces . Lower prevalence mutations will result , meaning less predictive power . Simulations also suggest that although clone-sets may be uniquely identified , prediction of the underlying reticulation network is difficult , with many networks explaining the same dataset . As we lower the minimum prevalence of analyzed mutations , their number will increase . The number of networks will likely explode and raise significant challenges . Furthermore , single strand RNA viruses such as influenza mutate quickly , suggesting a preponderance of low prevalence mutations likely exist . This is further exacerbated by the fact that sequencing uses rt-PCR , introducing point mutations and template switching artifacts that create noise in the data . These processes are likely responsible for the grass-like distribution of low prevalence mutations visible in Fig 4B and 4C . Thus as we consider lower prevalence mutations we are likely to get a rapidly growing evolution structure of increasingly complex topology . The methods we have introduced , however , can provide useful information at the upper-portions of these structures . The software ViralNet is available at www . uea . ac . uk/computing/software . The raw data is available from the NCBI ( project accession number SRP044631 ) . More detailed outputs from the algorithm are available in Supplementary Information .
Viral RNA was isolated from 280 μl-aliquots of nasal swabs using the QIAamp viral RNA mini kit ( Qiagen ) following the manufacturer’s instructions . To quantify virus shedding , a real-time RT-PCR as described in [5] and [32] was performed . Virus copy numbers are available in supplementary information . Full genome amplification was performed as described in [33] . Each DNA sample was then processed for paired end sequencing on both a Hi-Seq and Mi-Seq machine , producing ends with 101 bases . The reads from the innoculum sample were then used to construct a majority reference sequence . Reads from samples further down the infection chain were compared against this reference for variant calling . Paired reads with identical start and end reference points , and identical sequences , were deemed to arise from a single PCR product . Duplicates were thus removed and only one paired read is selected for the final dataset . The construction of phylogenetic trees is a well established area [24] . Trees are frequently constructed from tables of haplotypes of different species . However , we have two properties that change the situation . Firstly , if we have a set of n mutations linked by reads , we can have up to 2n distinct haplotypes . However , a consistent set of splits from such a table should only have up to n + 1 distinct haplotypes , in a split-compatible configuration [24] . To construct a phylogenetic tree we thus need to classify the genotypes as real or artifact . Secondly , we have prevalence information , in the form of a conserved network flow through the tree . This can help us to both decide which haplotypes to believe and to construct a corresponding tree . To describe the algorithm we first introduce some notation . Now , the evolutionary structure is represented by two types of rooted directed tree; one where each edge represents a mutation , such as in Fig 2F , and one where all leaves represent clones in the population , such as in Fig 2A and 2D . The first is a subtree of the latter . The latter has a conserved flow network . These will be termed the Compact Prevalence Tree and Complete Prevalence Tree respectively . Now to each edge e in the compact prevalence tree , we assign prevalence ρ ( e ) . This represents the proportion of population containing the mutation represented by the edge e . The single directed edge ein ( v ) pointing toward a vertex v ( away from the root ) represents a viral population of prevalence ρ ( ein ( v ) ) , all containing the mutation corresponding to edge ein ( v ) , along with its predecessor mutations . The set of daughter edges Eout ( v ) leading away from node v represent populations containing subsequent mutations , each with prevalence ρ ( e ) , e ∈ Eout ( v ) . The remaining population from ρ ( ein ( v ) ) contains just the original mutation set , having a prevalence described by the capacity ζ ( v ) . The conservation of prevalence satisfied by each vertex v ∈ T is then represented by the condition: ρ ( e i n ( v ) ) = ζ ( v ) + ∑ e ∈ E o u t ( v ) ρ ( e ( v ) ) ( 1 ) The root node has total prevalence of 1 , representing the entire population of interest . This describes the mutation based trees such as that in Fig 2F . To obtain a complete tree containing all the clones , we need to extend an edge from each internal node to represent the associated clone ( these are the dashed lines in Fig 2A ) . The prevalence of the additional edges are equal to the capacities of the parental nodes . We saw in Fig 2B that mutations can be clustered together , and evolution trees constructed for each cluster . We define a cluster to be any subset of mutations with reference positions all lying within the two sequences of individual paired end reads . We restricted clusters to cases where the total number of paired end reads containing the mutation sites numbered at least 200 and greater reliability in the linkage information exists . The evolutionary tree corresponding to each mutation cluster is referred to as a Subtree . We then look for a tree that contains all such subtrees as a subset of edges . We refer to these as Supertrees . The algorithm is broken into two steps . The first calculates subtrees . The second calculates supertrees . We would like to use data such as Fig 3B to reconstruct the evolutionary structure . The splits method [25] is used to construct phylogenetic networks such as Fig 3G . There are many recombination networks that correspond to any given phylogenetic network . A standard method to identify recombination networks is to look for an optimal path of trees across the recombination sites [36] . These methods generally have the full mutation profile of a set of species of interest to compare . Our problem is exacerbated by missing data and the full haplotypes of distinct species ( clones in our case ) are not available . However , we have prevalence information which can help identify structures consistent with the data . We construct recombination networks in five steps; haplotype classification , super-network construction , super-network filtering , prevalence maximum likelihood estimation , and prevalence range estimation . We describe these steps in detail . We model template switching during rtPCR as follows . Suppose we have two mutations of interest and four possible genotypes , labeled C00 , C01 , C10 and C11 . We have corresponding read depth counts n00 , n01 , n10 and n11 . Now , if tree like evolution exists , one of C01 , C10 or C11 is an artifact arising from template switching during rtPCR ( the wild type C00 is assumed to always occur ) . We demonstrate the case where C01 is an artifact ( model 2 in Fig 6Aii ) . The derivation for the other two models is similar . Then we assume that the real clones C00 , C10 and C11 have prevalences of a , b and c , respectively , so that a + b + c = 1 . We model rtPCR as a time continuous three state process , where template switching occurs at a rate λ , jumping to any of the three templates C00 , C10 or C11 with probabilities a , b and c , respectively . We also refer to the states as a , b and c . The template switching rate λ is taken to be a constant , which is equivalent to assuming template switching occurs with uniform probability along the segments . Any sequence context effects along segments are ignored . We let pa ( t ) , pb ( t ) and pc ( t ) be the probabilities of occupying a copy of the corresponding templates at position t . Then conditioning pa ( t ) over a time interval ( t , t + dt ) results in the following expression ( see [38] for typical derivations ) : p a ( t + d t ) = p a ( t ) ( 1 - λ d t ) + p a ( t ) a λ d t + p b ( t ) a λ d t + p c ( t ) a λ d t This gives us the following differential equation and solution: d p a d t = λ ( a - p a ) ⇔ p a ( t ) = a + ( p a ( 0 ) - a ) e - λ t We rescale time so that t = 1 represents one rtPCR cycle . We then have the following transition matrix between states: T = a + ( 1 - a ) e - λ b - b e - λ c - c e - λ a - a e - λ b + ( 1 - b ) e - λ c - c e - λ a - a e - λ b - b e - λ c + ( 1 - c ) e - λ Probabilities for all types C00 , C01 , C10 and C11 can now be defined , which we demonstrate for C10 . Derivations for the other terms can be obtained in a similar manner . From Fig 6Aii we see that to obtain a read of the form C10 , we can start in either state b or c and end in either state a or b . This gives us four terms to add: P r ( C 10 ) = b T b a + b T b b + c T c a + c T c b = b ( a - a e - λ + b + ( 1 - b ) e - λ ) + c ( a - a e - λ + b - b e - λ ) = b + a c n Here n = 1 − e−λ is the probability a template switch occurs . The formulas in Fig 6A are obtained similarly . The counts n00 , n01 , n10 and n11 then follow a multinomial distribution , from which log-likelihoods can be derived . A chi-squared goodness of fit can then be obtained . We note that in many cases , solutions for the four terms Pr ( C00 ) , Pr ( C01 ) , Pr ( C10 ) and Pr ( C11 ) in terms of a , b , c and n can be obtained , resulting in a perfect fit . When this is not possible , one or more of the three models can be rejected if the fit is sufficiently bad . Note that none of these three models necessarily explain the data . In the last column of Fig 6D , for example , we have four artificial counts 50 , 1000 , 1000 and 1000 corresponding to genotypes C00 , C01 , C10 and C11 . All three models are a bad fit suggesting recombination is present . However , this relies on small counts for C00 , which were not observed in the real data that was examined . Note that template switching has no effect on the prevalence of individual mutations . For example , considering Fig 5Ciii , if we add Pr ( C01 ) and Pr ( C11 ) , we get b + c , which is precisely the prevalence of mutation M2 . The validation of the method is based upon simulated data . This will give some idea of the reconstruction capabilities of the methods and allow benchmarking with other existing approaches . In particular , we compared our tree construction algorithm to the benchmark software Shorah using the same simulation approaches as Zagordi et al [12 , 15] and Astrovskaya et al . [16] . To measure the performance of the mixed population estimation , we computed the Precision , the Recall , and the Accuracy of prevalence estimation for the methods of interest . The recall ( or sensitivity ) gives the ratio T P T P + F N of correctly reconstructed haplotypes to the total number of true haplotypes , where we have true positives ( TP ) , false negatives ( FN ) and false positives ( FP ) . The precision gives their ratio to the total number of generated haplotypes , T P T P + F P . The accuracy measures the ability of the method to recover the true mixture of haplotypes , and was defined as measuring the mean absolute error of the prevalence estimate . Where a range estimate is obtained for the prevalence , we calculate the shortest distance from the true value to the range . Comparison with Shorah was done on simulated deep sequencing data from a 1 . 5 kb-long region of HIV-1 . Simulated reads have been generated by MetaSim [39] , a meta-genomic simulator which generates collections of reads reproducing the error model of some given technologies such as Sanger and 454 Roche . It takes as input a set of genome sequences and an abundancy profile and generates a collection of reads sampling the inputted genomic population . For up to 12 haplotypes and 3 reticulations we performed 100 runs as follows . We randomly constructed a network by attaching each new branch to a random selected node . Reticulations were also randomized . The prevalences of the resulting clones ( at the leaves ) were randomly selected from a Dirichlet distribution . This is repeated for 10 time points of data . We used MetaSim to generate a collection of 5 , 000 reads having an average length of 500bp and replicating the error process of Roche 454 sequencing . The methods were then applied to the resulting data . Shorah output can display mismatches or gaps in the outputted genomes , with increasing frequency at the segment edges . We applied a modification on Shorah output by trimming the edge and we then corrected one or two mismatches or gaps on all the genomes before addressing the comparison . Fig 10A–10C provide the comparison for recall , precision and error indicators . We found slight improvements for recall , especially for tree like evolution . The precision and error also had improved results . We acknowledge that the simulations were based upon evolutionary structures that the models are designed to fit so such improvement might be expected . Furthermore , Shorah likely have better performance on low prevalence clones . However , these simulations demonstrate that reasonable results can be obtained from the techniques we have introduced . | Any functional influenza virus particle is made up of eight distinct RNA segments . There can be in the order of 106 such particles per mL of infected tissue . Furthermore , on average , each new virus particle has a single mutation distinguishing the virus from its parent particle . The population of viruses thus contains a diverse mix of mutations . Modern sequencing experiments produce a signal that represents this mixed population . Untangling this signal to describe the evolutionary processes at work is an important part of virus biology . Furthermore , if an individual is infected with two different strains that both infect a single cell , new particles can form that contain a mixture of the two parents segments . This is known as reassortment and can result in the emergence of new virus strains . These events are hard to identify from sequencing experiments . Here we introduce a statistical method that can infer the evolutionary structure from a time series of sequencing experiments , which can also detect reassortment events , thus providing a method to help improve the understanding of within host evolution of viruses . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2015 | Inferring the Clonal Structure of Viral Populations from Time Series Sequencing |
Phosphatidylinositol-3-phosphate 5-kinase ( PIKfyve ) is a lipid kinase involved in endosome maturation that emerged from a haploid genetic screen as being required for Ebola virus ( EBOV ) infection . Here we analyzed the effects of apilimod , a PIKfyve inhibitor that was reported to be well tolerated in humans in phase 2 clinical trials , for its effects on entry and infection of EBOV and Marburg virus ( MARV ) . We first found that apilimod blocks infections by EBOV and MARV in Huh 7 , Vero E6 and primary human macrophage cells , with notable potency in the macrophages ( IC50 , 10 nM ) . We next observed that similar doses of apilimod block EBOV-glycoprotein-virus like particle ( VLP ) entry and transcription-replication competent VLP infection , suggesting that the primary mode of action of apilimod is as an entry inhibitor , preventing release of the viral genome into the cytoplasm to initiate replication . After providing evidence that the anti-EBOV action of apilimod is via PIKfyve , we showed that it blocks trafficking of EBOV VLPs to endolysosomes containing Niemann-Pick C1 ( NPC1 ) , the intracellular receptor for EBOV . Concurrently apilimod caused VLPs to accumulate in early endosome antigen 1-positive endosomes . We did not detect any effects of apilimod on bulk endosome acidification , on the activity of cathepsins B and L , or on cholesterol export from endolysosomes . Hence by antagonizing PIKfyve , apilimod appears to block EBOV trafficking to its site of fusion and entry into the cytoplasm . Given the drug’s observed anti-filoviral activity , relatively unexplored mechanism of entry inhibition , and reported tolerability in humans , we propose that apilimod be further explored as part of a therapeutic regimen to treat filoviral infections .
The epidemic of Ebola virus disease ( EVD ) that raged through Western Africa between 2013 and 2016 was the most severe filovirus disease epidemic in recorded history [1 , 2] . While several promising therapeutic antibodies [3–11] and novel small molecules [12–19] remain in development , no therapeutic is yet approved to treat patients with EVD . In the continuing pursuit of an anti- Ebola virus ( EBOV ) therapeutic , one strategy is to identify approved drugs that show anti-EBOV activity [20–28] , with the goal of repurposing them for an anti-EBOV therapeutic , either alone or as part of a multi-component regimen [29–34] . Most of the approved drugs that have been identified as blocking EBOV infection inhibit the entry phase of the viral lifecycle [19–25 , 27 , 28] . Cell entry by EBOV is a complex process [35 , 36] entailing virus binding to cell surface attachment factors , internalization by macropinocytosis , processing by endosomal proteases , and transport to endolysosomes containing Niemann-Pick C1 ( NPC1 ) [14 , 37] , the intracellular receptor for EBOV [38] . Finally , EBOV fuses with the limiting membrane of NPC1+ endolysosomes [39–41] , liberating its genome and associated proteins into the cytoplasm to begin replication . The essential role of NPC1 in EBOV entry and infection was powerfully illuminated in a haploid genetic screen [37] . The same screen revealed other gene products critical for EBOV entry [42 , 43] including many involved in endosome and lysosome biogenesis and maturation . One of the latter proteins was phosphatidylinositol-3-phosphate 5-kinase ( PIKfyve ) [37] , a lipid kinase that phosphorylates phosphatidylinositol-3-phosphate ( PI3P ) to generate phosphatidylinositol-3 , 5-bisphosphate ( PI ( 3 , 5 ) P2 ) . PIKfyve and PI ( 3 , 5 ) P2 are known to be critical for endosome maturation [44–53] . Apilimod is a small molecule that binds to and inhibits the phosphotransferase activity of PIKfyve [54] . The drug was developed as a suppressor of interleukin 12 and 23 production [55] , and was tested in phase 2 clinical trials for treatment of Crohn’s disease [56 , 57] , psoriasis [58] , and rheumatoid arthritis [59] . Although no clinical benefit has yet been reported , apilimod is deemed to be well tolerated in humans . We chose to test whether apilimod could inhibit infections by EBOV and Marburg virus ( MARV ) for three reasons . The first was that apilimod binds [54] to the EBOV entry factor PIKfyve [37] . The second was because apilimod emerged from a blinded screen of 35 drugs ( S1 Fig; S1 Table ) , which were selected as potential inhibitors based upon hypotheses of drugable targets and from theoretical considerations of pathways possibly involved in the EBOV life cycle . The third reason was because apilimod is well tolerated in humans . We find that apilimod inhibits infection by both EBOV and MARV , being notably effective in primary human macrophages , which are initial targets of filoviral infection [60 , 61] . Mechanistic studies revealed that apilimod blocks EBOV entry into the cell cytoplasm by working through PIKfyve and that its effect is to block viral particle trafficking to NPC1+ endolysosomes , the site of EBOV fusion [39–41] . Hence we propose that apilimod be further explored as part of a cocktail of small molecules to combat EVD .
Vero E6 ( African green monkey kidney; ATCC 1586 ) cells were obtained from the American Type Culture Collection ( Manassas , VA ) . Huh 7 ( human hepatocellular carcinoma ) cells were obtained from Dr . Hideki Ebihara ( National Institute of Allergy and Infectious Diseases ( NIAID ) , Rocky Mountain Laboratories , Hamilton , MT ) . Peripheral blood mononuclear cells ( PBMCs ) were prepared from human whole blood ( Biological Specialty Corporation; Colmar , PA; Cat # 3100-03-04 ) and human monocyte-derived macrophages ( hMDM ) were generated from peripheral blood mononuclear cells at the Integrated Research Facility ( IRF ) immunology core laboratories as described previously [62 , 63] . hMDM were characterized by flow cytometric analysis for expression of major macrophage markers , including human leukocyte antigen-D related , CD11b , CD14 , CD163 , and CD86 , to confirm that the hMDM population was mature and highly purified [63] . HEK 293T/17 ( Human embryonic kidney; ATCC CRL-11268 via University of Virginia Tissue Culture Facility ) and BSC-1 ( Grivet monkey kidney; gift from Dr . Xiaowei Zhuang , Harvard University , Cambridge , MA ) cells were maintained in growth medium: high glucose Dulbecco's Modified Eagle Medium ( DMEM ) supplemented with 1% L-glutamine , 1% sodium pyruvate , and 1% antibiotic/antimycotic , all from Gibco Life Technologies ( Carlsbad , CA ) , and either 10% supplemented calf serum ( SCS; Hyclone , GE Healthcare Bio-Sciences , Pittsburgh , PA ) for HEK 293T/17 cells or 10% fetal bovine serum ( FBS , Seradigm , VWR International , Radnor , PA ) or 10% cosmic calf serum ( CCS , Hyclone ) for BSC-1 cells . Toremifene citrate ( CAS 89778-27-8 ) was purchased from Sigma-Aldrich ( St . Louis , MO; Cat# T7204-25MG ) and apilimod ( CAS 541550-19-0 ) was purchased from Axon MedChem ( Groningen , NL; Cat# 1369 ) . Nocodazole ( CAS 31430-18-9 ) was purchased from Sigma-Aldrich ( St . Louis , MO; Cat # M1404-2MG ) . All procedures using infectious EBOV/Mak or MARV were performed under biosafety level 4 ( BSL-4 ) conditions at the IRF . The C05 isolate of the Makona variant of EBOV ( EBOV/Mak; full designation: Ebola virus/H . sapiens-tc/GIN/2014/Makona-C05; GenBank: KX000398 ) and Marburg Angola virus ( MARV; full name: Marburg virus/H . sapiens-tc/AGO/2005/Ang-1379v; GenBank: N/A ) were propagated in BEI NR-596 Vero E6 cells and used after one or two passages . The drug screen method was performed as described previously [63] . Briefly , Vero E6 and Huh 7 cells were seeded in 96-well plates at a density of 3 x 104 cells/well , and hMDMs were plated at a density of 1 x 105 cells/well 24 h prior to the addition of drugs . For each cell type , cells were plated in 1 black opaque 96-well plate , for the evaluation of drug cytotoxicity , and 2 clear bottom , 96-well Operetta plates , for the evaluation of drug efficacy . Drugs dissolved in dimethyl sulfoxide ( DMSO; Sigma-Aldrich , St . Louis , MO ) were diluted in DMEM with 10% FBS with the final DMSO concentration not exceeding 0 . 05% . The drug solutions were diluted two-fold in an 8-point dilution series and transferred to cell plates 1 h prior to virus infection . Efficacy plates for each cell type were infected with EBOV or MARV at a multiplicity of infection ( MOI ) of 0 . 5 . After 48 h , cells were fixed with 10% neutral-buffered formalin . Chemiluminescent enzyme-linked immunosorbent assay was used to determine virus activity . EBOV was detected with a mouse antibody against the EBOV VP40 matrix protein ( B-MD04-BD07-AE11 , made by US Army Medical Research Institute of Infectious Diseases , Frederick MD under Centers for Disease Control and Prevention contract ) [3] and MARV was detected with a mouse antibody against the MARV VP40 protein ( Cat# IBT 0203–012 , IBT Bioservices , Rockville , MD ) for 1–2 h at 37°C . Cells were stained with a secondary antibody , anti-mouse IgG , peroxidase labeled antibody ( Cat# 074–1802 , KPL Inc . , Gaithersburg , MD ) . Luminescence was detected using Pico chemiluminescent Substrate ( Thermo Fisher Scientific Inc . , Rockford , IL ) and an Infinite M1000 Pro plate reader ( Tecan , Morrisville NC ) . For quantitation of drug toxicity , 1 black opaque cell plate for each cell type was mock infected ( no virus ) and treated with drug dilutions under the same conditions as the infected cells . After 48 h , cell viability was measured using the CellTiter Glo Luminescent Cell Viability Assay kit according to the manufacturer’s protocol ( Promega , Madison , WI ) . Luminescence was read on an Infinite M1000 Pro plate reader . Following background subtraction , inhibition was measured as percent relative to untreated infected cells . Non-linear regression analysis was performed , and IC50s were calculated from fitted curves ( log [agonist] vs response [variable slope] with constraint to remain above 0; GraphPad Software , La Jolla , CA ) . Error bars of dose-response curves represent the standard deviation of three replicates . Entry reporter viral-like particles ( VLPs ) bearing GP from the Yambuku-Mayinga isolate of EBOV were prepared as described previously [24 , 25 , 41] . In brief , HEK 293T/17 cells ( ~80% confluent ) were transfected with cDNAs encoding EBOV GP , VP40 , mCherry-VP40 , and β-lactamase-VP40 ( βlam-VP40 ) . The cell medium was collected 24 and 48 h post-transfection and cleared of debris . VLPs in the cleared medium were then pelleted through a 20% sucrose cushion by centrifugation , resuspended in HM buffer ( 20 mM HEPES , 20 mM MES , 130 mM NaCl , pH 7 . 4 ) , and repelleted . The final VLP pellet was resuspended ( 1:100 starting volume of medium ) in 10% sucrose-HM . The total protein concentration of the VLPs was determined by bicinchoninic acid ( BCA ) assay . All entry-reporter VLP preparations were assessed by western blot analyses ( for the presence of GP as well as EBOV VP40 ) and titered on HEK 293T/17 cells to confirm entry competency . The VLP entry assay scores the ability of βlam-VP40 ( from incoming entry reporter VLPs ) to cleave a βlam substrate preloaded into the target cell cytoplasm; this only occurs if the VLP fuses with an endosome . The assay was performed as described previously [24 , 25 , 41] . In brief , 30 , 000 HEK 293T/17 cells or BSC-1 cells were seeded per well in a clear 96-well plate . 18–24 h post seeding , the cells ( ~80%–90% confluent ) were treated with the indicated concentration of apilimod ( Axon MedChem; DMSO for mock ) diluted in Opti-MEM I ( OMEM , Gibco Life Technologies , Thermo Fisher Scientific ) for 1 h at 37°C in a 5% CO2 incubator . VLPs diluted in OMEM ( with DMSO or the same concentration of apilimod ) were bound to the cells by spinfection ( 250× g ) for 1 h at 4°C . After 3 h in a 37°C , 5% CO2 incubator , the βlam substrate CCF2-AM ( Life Technologies , via ThermoFisher Scientific , Waltham , MA , USA ) was loaded into the cells using 20 or 5 mM Probenecid ( MP Biomedicals via ThermoFisher Scientific , Waltham , MA , USA ) , for BSC-1 or HEK 293T/17 cells , respectively . The cells were incubated overnight at RT and then fixed and analyzed by flow cytometry . To measure corresponding cell viability , 3 x 104 HEK 293T/17 cells , seeded and grown as above but in 96-well opaque white plates were treated as above for VLP entry , but without addition of VLPs or CCF2-AM . Following overnight incubation at RT ( as above ) , the medium was removed and replaced with 50 μL of fresh medium per well . Fifty microliters ( per well ) of CellTiter-Glo 2 . 0 ( Promega , Madison WI , USA ) was then added . After shaking for 2 min at RT at 575 rpm on a Jitterbug orbital shaker ( Boekel Scientific , via ThermoFisher Scientific , Waltham , MA , USA ) , the plate was incubated at RT for 10 min , after which the luminescent signal was detected using a BioTek Synergy HT plate reader ( BioTek , Winooski , VT , USA ) . Transcription/replication-competent virus-like particles ( trVLPs ) were prepared as described [25 , 64 , 65] . In brief , HEK 293T/17 cells were seeded in six well plates and transfected 24 h later ( when ~50% confluent ) using TransIT-LT1 ( Mirus , Madison , WI , USA ) with pCAGGS-NP , pCAGGS-VP35 , pCAGGS-VP30 , pCAGGS-L , a tetracistronic minigenome plasmid , and pCAGGS-T7 polymerase . The minigenome plasmid encodes Renilla luciferase , as well as the matrix protein VP40 , the nucleocapsid associated protein VP24 , and the GP from EBOV . 24 h post transfection , the medium in each well was replaced with 4 mL fresh growth medium containing 5% FBS . 96 h after transfection , the medium ( containing trVLPs harboring the Renilla luciferase-containing minigenome ) was harvested , pooled , and cleared of cellular debris by centrifugation for 5 min at 800× g and used for trVLP assays as described below . The trVLP assay measures the ability of trVLPs containing a Renilla luciferase-encoding tetracistronic EBOV minigenome to infect target cells pretransfected with plasmids encoding proteins to enhance trVLP entry ( the adhesion factor Tim-1 ) and ( other plasmids ) to support replication of the minigenome . If trVLPs enter target cells , the minigenome is replicated and transcribed , leading to Renilla luciferase reporter activity [64 , 65] . In brief: Cells were pretreated with apilimod ( Axon MedChem; DMSO for mock ) as described above . The pretreatment solution was then removed and replaced with 100 μL trVLPs diluted to 200 μL in growth medium containing 10% SCS and the indicated concentration ( s ) of apilimod ( DMSO for mock ) . The cells were then incubated for 48 h at 37°C in a 5% CO2 incubator , after which the medium was replaced with 40 μL of fresh growth medium containing 10% SCS . 40 μL of RenillaGlo substrate ( Promega , Madison , WI , USA ) was then added to each well and the plate immediately analyzed on a GloMax plate reader ( Promega , Madison , WI , USA ) . To assess cell viability in corresponding samples without trVLPs , the pretreatment solution was removed and replaced with 200 μL fresh growth medium containing 10% SCS and the indicated concentrations of apilimod ( DMSO for mock ) . The cells were then incubated for 48 h at 37°C in a 5% CO2 incubator , after which the medium was replaced with 40 μL of fresh growth medium containing 10% SCS . 40μL of CellTiter-Glo 2 . 0 ( Promega ) was then added to each well and the plate placed on a Jitterbug orbital shaker ( 575 rpm ) for 2 min at RT . The plate was then incubated at RT for 10 min , after which the luminescent signal was detected using a Synergy HT ( BioTek , Winooski , VT , USA ) plate reader . BSC-1 cells were seeded in 35mm glass bottom dishes ( MatTek , Ashland , MA ) that were coated with 20 μg/mL fibronectin ( Sigma-Aldrich , St . Louis , MO , USA ) . The next day , when the cells were 90–100% confluent , the cells were treated with the indicated drug at the indicated concentration , diluted in growth medium containing 10% cosmic calf serum , for 3 h at 37°C in a 5% CO2 incubator . Acridine Orange ( Life Technologies , Thermofisher Scientific , Waltham , MA , USA ) was added directly to each dish to reach a final concentration of 6 . 6 μg/mL . The cells were incubated at 37°C in a 5% CO2 incubator for 20 min and then were washed 3 times with phosphate buffered saline ( PBS ) , 5 min per wash . Cell imaging medium [Live cell imaging solution ( Molecular Probes , Cat# A14291DJ , Thermo Fisher Scientific , Waltham , MA ) containing 10% FBS and 4 . 5 g/L glucose] was added to the dishes and images were taken using a Nikon C1 laser scanning confocal unit attached to a Nikon Eclipse TE2000-E microscope with a 100X , 1 . 45-numerical-aperature ( NA ) Plan Apochromat objective ( Nikon , Melville , NY ) . BSC-1 cells were seeded in 35mm glass bottom dishes ( MatTek , Ashland , MD , USA ) that had been coated with 20 μg/mL fibronectin ( Sigma-Aldrich ) . The next day , when the cells were 90–100% confluent , the cells were treated with the indicated drug at the indicated concentration plus 0 . 05 μM TopFluor Cholesterol ( Avanti Polar Lipids , Alabaster , AL ) , diluted in serum-free growth medium , for 18 h at 37°C in a 5% CO2 incubator . Following incubation , the cells were gently rinsed once with PBS and cell imaging medium ( Live cell imaging solution ( Molecular Probes ) containing 10% FBS and 4 . 5 g/L glucose ) was added to the dishes . The cells were incubated at 37°C in a 5% CO2 incubator for 30 min . Images were then taken using a 60X /1 . 45 numerical aperture ( NA ) Nikon Plan Apo total internal reflection fluorescence oil immersion objective attached to a Nikon Eclipse TE2000-E microscope equipped with a Yokogawa CSU 10 spinning-disk confocal unit , a 512-by-512 Hamamatsu 9100c-13 EM-BT camera , a motorized stage maintained at 37°C , and a Nikon Perfect Focus system . VLP trafficking experiments were performed in BSC-1 cells essentially as described previously [41] with the following minor modifications . Cells were pretreated with apilimod or nocodazole ( indicated concentrations ) diluted in OMEM for 1h at 37°C prior to VLP addition . VLPs at 0 . 5 μg/well were bound to the cells by spinfection ( 250 x g ) for 1 h at 4°C . After incubation at 37°C ( CO2 incubator ) for the indicated times , the cells were fixed and washed . Next , primary antibodies ( 1:1000 rabbit α-NPC1 , ( Abcam ) or 1:1000 mouse α-early endosome antigen 1 ( EEA1 ) , BD Biosciences , San Jose , CA ) were added for 45 min at room temperature ( RT ) and , following washing , secondary antibodies ( 1:1500 α-mouse or α-rabbit AlexaFluor 488 , Life Technologies , Thermo Fisher Scientific ) were added for 30 min at RT . The cells were washed and the coverslips were mounted overnight on glass slides using ProLong Gold Antifade reagent ( Life Technologies , Thermo Fisher Scientific ) . The coverslips were then sealed and images were taken using a Nikon C1 laser scanning confocal unit attached to a Nikon Eclipse TE2000-E microscope with a 100X , 1 . 45-NA Plan Apochromat objective . Colocalization of VLPs ( red , mCherry-VP40 ) and endosomal markers ( green , NPC1 or EEA1 ) was assessed as Manders coefficients . Statistics were analyzed using GraphPad Prism 7 . Normality of the data was assessed using the D’Agostino & Pearson normality test . Significance of normally distributed data was determined by T-test , and significance of non-normally distributed data was determined by Mann-Whitney test . Cathepsin B+L activity was assayed as described previously [20 , 21 , 24] . ( 2S , 3S ) -trans-epoxysuccinyl-L-leucylamido-3-methylbutane ethyl ester ( EST , Cat # 330005 , Calbiochem , EMD Millipore , Billerica , MA ) , an inhibitor of cathepsin B , H , and L , was used as a positive control for inhibition at the indicated concentration . Data are displayed as fluorescence units ( Ex 360/Em 460 ) . Thirty-five drugs obtained from the National Center for Advancing Translational Sciences ( NCATS ) were dissolved in DMSO at 500 μM . Drugs were diluted in DMEM ( Life Technologies , Thermo Fisher Scientific ) supplemented with 2 mM L-Glutamine ( Q; Life Technologies , Thermo Fisher Scientific ) and 100 U/ml penicillin and 100 μg/ml streptomycin ( PS; Life Technologies , Thermo Fisher Scientific ) . Drugs were added to confluent Vero E6 cells . Drugs and cells were then incubated at 37°C and 5% CO2 in a humidified incubator in 96-well plates for final concentrations of 10 , 1 , or 0 . 1 μM in a final volume of 100 μl DMEM/PS/Q with 2% FBS ( Life Technologies , Thermo Fisher Scientific ) . Cells were returned to the incubator for 2 h . For efficacy studies , 50 μl DMEM/PS/Q containing 1x103 TCID50 of recombinant EBOV expressing firefly luciferase from an additional transcriptional unit ( rgEBOV-luc2 , Genbank Accession number KF990214 . 1 ) [66] was added to the cells . At 48 h post-inoculation the supernatant was removed and 100 μl GloLysis buffer ( Promega ) was added to the cells and incubated for 10 min at RT . Afterwards , 40 μl lysate was added to 40 μl BrightGlo reagent ( Promega ) in white opaque 96 well plates , and reporter activity was measured using a GloMax luminometer . For cytotoxicity studies , 50 μl of DMEM/PS/Q without virus was added to the cells following the 2 h pre-incubation with drugs , and cells were returned to the incubator . At 48 h , 100 μl of supernatant was removed , and 50 μl of CellTiterGlo reagent ( Promega ) was added to the cells . Cells were incubated for 2 min on an orbital shaker at 60 RPM , and then for an additional 10 min without shaking at RT . Supernatants were transferred to white opaque 96-well plates , and reporter activity was measured using a GloMax luminometer ( Promega ) . Ribavirin at final concentrations of 1 mg/ml , 100 μg/ml , and 10 μg/ml , as well as DMSO at concentrations corresponding to the DMSO concentrations found in the drug dilutions served as controls . All experiments involving infectious rgEBOV-luc2 were performed in the maximum containment laboratory of the Rocky Mountain Laboratories , National Institutes of Health , Hamilton , MT , following approved protocols . HEK 293T/17 cells were seeded at a density of 3 x 106 cells per 10 cm plate . The next day , when the cells were approximately 60% confluent , the media above the cells was replaced with 6mL OMEM and the cells were transfected with 2 . 4 μg pTG-luc , 1 . 2 μg pCMV-MLVgag-pol , 1 . 2 μg pGPΔmucin ( encoding Ebola GP deleted for its mucin domain ) , and 1 . 2 μg of MLV-gag-βlam diluted to 300 μL in OMEM ( per plate ) , using 18 μL Lipofectamine 2000 ( Invitrogen , ThermoFisher Scientific , Walthan , MA ) diluted to 300 μL in OMEM ( per plate ) . 4h post transfection , 6 mL of antibiotic-free growth medium containing 10% SCS was added to each plate , and the cells were incubated for 48 h at 37°C in a CO2 incubator . Cell medium containing pseudovirus was then collected , pooled , and cleared of cellular debris by centrifugation at 250 x g for 7 min . The clarified supernatant containing pseudovirus was then passed through an 0 . 45 μm filter and the pseudoviruses were concentrated 100-fold by high-speed centrifugation through a 25% sucrose cushion in HM buffer ( 20mM HEPES , 20mM MES , 130mM NaCl , pH7 . 4 ) for 75 min at 103 , 745 x g . The final pseudovirus pellet was resuspended in growth medium ( 100-fold concentrated from harvest supernatant ) . HEK 293T/17 cells were seeded at a density of 5 x 105 cells per well in 6-well plates . When the cells were ~50% confluent ( ~18–24 h post seeding ) , they were transfected with plasmids encoding GFP-PIKfyve or pEGFP-Cl using TransIT LT1 transfection reagent ( Mirus , Madison , WI ) following the manufacturer’s instructions . 18 h post transfection , the cells were re-seeded in 96 well opaque white plates ( BD Falcon , ThermoFisher Scientific , Waltham , MA ) at a density of 3 x 104 cells per well . Transfection was confirmed by fluorescence microscopy . 18 h post re-seeding , the cells were pretreated for 1 h at 37°C with apilimod . MLV-luciferase particles pseudotyped with EBOV GPΔmucin were added to the cells in the presence of apilimod , and infection was allowed to proceed for 48 h at 37°C . The cells were then washed once with PBS and overlaid with 50μL PBS . Luciferase activity was then immediately assayed by adding 50μL of Britelite plus ( Perkin Elmer , Waltham , MA ) and reading on a Glomax plate reader ( Promega , Madison , WI , USA ) following the manufacturer’s instructions .
We first tested whether apilimod blocks EBOV infection in cell cultures . Apilimod blocked EBOV infection of Huh 7 ( liver ) cells , Vero E6 ( kidney ) cells , and primary human monocyte-derived macrophages ( hMDMs ) ( Fig 1 ) with 6- to 247-fold higher activity than the positive control , toremifene citrate [20 , 22] . Apilimod also blocked MARV infection of the same cell types ( Fig 2 ) with 38- to 1160-fold higher activity than the positive control . Apilimod was notably potent ( IC50 , 10 nM ) against both filoviruses in hMDMs ( Figs 1 and 2 , Table 1 ) . While similar potency ( IC50 , 15–25 nM ) was seen in Vero E6 cells , apilimod was ~10-fold less potent ( IC50 , 140 nM ) in Huh 7 cells ( Fig 1 , Fig 2 and Table 1 ) . To begin to probe the mechanism by which apilimod blocks EBOV infection , we directly compared dose-response profiles for blocking EBOV entry and replication using entry reporter VLPs [24] and trVLPs [64] , respectively . Both sets of VLPs bore the GP from the Mayinga isolate of EBOV . Apilimod blocked EBOV particle entry ( Fig 3A ) and replication ( Fig 3B ) with similar dose-response profiles ( Fig 3C ) . This finding suggested that apilimod blocks the entry phase of the filoviral lifecycle . Since apilimod targets PIKfyve [54] , since PIKfyve is required for EBOV entry and infection [37] , and since apilimod blocks EBOV entry and infection ( Figs 1–3 ) , we reasoned that apilimod blocks EBOV entry and infection by targeting PIKfyve . To test this hypothesis we over-expressed PIKfyve ( GFP-PIKfyve ) and compared the dose response needed for apilimod to block EBOV-GP mediated pseudovirus infection . As predicted , and as seen in Fig 4 , higher doses of apilimod were needed to achieve similar levels of inhibition of EBOV GP-mediated infection in GFP-PIKfyve vs . GFP expressing cells . This supports our proposal that apilimod blocks EBOV entry and infection through a PIKfyve-dependent pathway . Recent work has shown that EBOV traffics deep in the endocytic pathway , to NPC1+ endolysosomes , for fusion and entry [39–41] . Therefore we asked whether apilimod prevents EBOV VLPs from reaching NPC1+ endolysosomes . We used BSC-1 cells for these experiments as they are more suitable ( flatter and more adherent ) for immunofluorescence analysis than the HEK 293T/17 cells used in previous experiments ( Figs 3 and 4 ) . We first demonstrated that apilimod blocks EBOV VLP entry into BSC-1 cells with the same approximate dose-dependency as its effects in HEK 293T/17 cells ( Fig 3D ) . Given that , we next asked if apilimod blocks trafficking of EBOV GP VLPs to NPC1+ endolysosomes in BSC-1 cells . As seen in Fig 5 , this was , indeed , the case . Apilimod blocked EBOV VLP trafficking to NPC1+ endolysosomes to a similar extent as nocodazole , a microtubule destabilizer that is known to block traffic between early and late endosomes [67] . The findings presented in Fig 5A–5D were obtained after allowing VLPs pre-bound to the cell surface to enter cells for 90 min at 37°C . This time point was chosen based on our extensive prior analysis of the time courses of EBOV VLP co-localization with NPC1+ endolysosomes and entry into BSC-1 cells [41] . To assure that apilimod did not accelerate VLP trafficking to NPC1+ endolysosomes , we analyzed co-localization of VLPs at various times up to 90 min in cells treated or not treated with apilimod . As seen in Fig 5E , at no point during this time course were VLPs seen to associate with NPC1+ endolysosomes in apilimod-treated cells , supporting our conclusion that apilimod blocks trafficking of EBOV particles to NPC1+ endolysosomes . Concomitant with decreased trafficking of EBOV GP VLPs to NPC1+ endolysosomes , apilimod caused EBOV VLPs to accumulate in EEA1+ endosomes ( Fig 6 ) . In apilimod-treated cells , EEA1+ endosomes appeared larger than those in mock-treated cells , consistent with previous reports showing enlarged endosomes in cells genetically deficient for PIKfyve or treated with PIKfyve inhibitors [44 , 45 , 50 , 53] . The findings presented in Figs 5 and 6 indicate that the primary mechanism by which apilimod blocks EBOV entry and infection ( Figs 1–4 ) is by blocking virus transport from early ( EEA1+ ) endosomes to their site of fusion in NPC1+ endolysosomes . To further test this model , we asked whether apilimod affects other attributes of the endosomal pathway needed for EBOV entry , either endosome acidification or the activity of cathepsin B and L [42 , 43] . At a concentration that strongly blocked EBOV entry and infection , apilimod had no detectable effect on endosome acidification ( Fig 7 ) . Bafilomycin , an inhibitor of EBOV infection that blocks endosome acidification , was used as a positive control . Apilimod also had no apparent direct effect on the activity of cathepsin B and L ( Fig 8 ) , in contrast to EST , a known inhibitor of cathepsin B , H , and L . Several cationic amphiphilic drugs such as U18666A that block EBOV entry and infection [20–22 , 24 , 25 , 68] induce cholesterol accumulation in endolysosomes [20 , 24] . In contrast , apilimod did not cause a detectable increase in cholesterol levels in endolysosomes ( Fig 9 ) . Hence apilimod appears to block filoviral entry and infection by inhibiting virus particle trafficking to NPC1+ endolysosomes , the portal for entry of the EBOV genome into the host cell cytoplasm [39–41] .
EBOV journeys deep into the cellular endosomal system , entering the cytoplasm through endolysosomes that are positive for NPC1 and two-pore channel 2 ( TPC2 ) [39] . In addition to NPC1 , its intracellular receptor [14 , 37 , 38] , EBOV requires multiple factors involved in endosome and lysosome biogenesis and maturation for entry [37] . One of the latter factors is PIKfyve [37] , which phosphorylates PI3P to generate PI ( 3 , 5 ) P2 . Here we have shown that apilimod , which binds to PIKfyve [54] , blocks EBOV entry and infection in a PIKfyve-dependent manner . The inhibitory effect of apilimod on EBOV entry is likely due to a defect in the maturation of endolysosomes , as extensive evidence indicates the importance of PIKfyve and PI ( 3 , 5 ) P2 in this process [44–53] . Although the exact mechanism by which PIKfyve and PI ( 3 , 5 ) P2 orchestrate endosome maturation is not known , several mechanisms have been postulated . Considered in these mechanisms are the observations that two Ca++ channels found in ( endo ) lysosomes—transient receptor potential cation channel , mucolipin 1 ( TRPML1 ) [50 , 69] and TPC2 [70–72]—are downstream effectors of PIKfyve and PI ( 3 , 5 ) P2 . Through its action on TRPML1 , PIKfyve has been shown to regulate the fission and consequent remodeling and maturation that reduces the size of macropinosomes containing endocytosed material from the cell surface and exterior [50] . In addition , the TPC2 channel has been reported to be activated by PI ( 3 , 5 ) P2 [71 , 72] . Intriguingly , both macropinocytosis [35 , 73 , 74] and TPC2 [75] are involved in EBOV entry and infection . Although we do not yet know all of the endolysosomal factors needed to trigger EBOV GP for fusion [36 , 39 , 43 , 76] , it appears clear that proper endosomal maturation is required . These findings are consistent with the mounting evidence for a role of PIKfyve in EBOV entry and our observation that the PIKfyve inhibitor , apilimod , blocks transport of EBOV particles to NPC1+ endolysosomes . It is therefore likely that interconnected effects of apilimod on PIKfyve [37 , 54] , TPC2 [75] , and endolysosome maturation culminate in its blockade of EBOV entry and infection . Several approved drugs that function as EBOV entry inhibitors ( e . g . , clomiphene , toremifene , and sertraline ) block EBOV entry into the cytoplasm after EBOV particles have been delivered to NPC1+ endolysosomes [20 , 21 , 24 , 25] . Hence they likely interfere with some aspect of the virus-endolysosome membrane fusion process , per se . Other approved drugs , including chloroquine , niclosamide , atovaquone , amodiaquine and quinacrine [21–23] , block endosomal acidification . Hence these drugs likely interfere with the processing of EBOV GP by acid-optimal endosomal cathepsins [42 , 43] and/or low pH-induced conformational changes required for fusion activity of cleaved GP [77 , 78] . In contrast to these mechanisms , our findings indicate that apilimod blocks EBOV entry by blocking particle delivery into NPC1+ endolysosomes . The only other approved drugs that we know of with anti-EBOV activity that are expected to have this mode of action are microtubule-disrupting agents , including colchicine , nocodazole , vinblastine , and vinorelbine [21–23] . Hence the mode of action of apilimod as an anti-filoviral agent is novel . Rather than blocking EBOV trafficking to NPC1+ endosomes by interfering with microtubules , apilimod blocks EBOV trafficking by inhibiting PIKfyve . Our findings indicate that apilimod has similar anti-viral activity against EBOV and MARV , consistent with the need for NPC1 in endolysosomes for the entry of these and other filoviruses [37 , 79 , 80] . We therefore consider it likely that apilimod , a host-directed small molecule , will have broad or even pan-filoviral activity . Furthermore , since many other viruses , so-called late penetrating viruses [36 , 81] , traffic beyond early endosomes for entry , it is possible that apilimod will block entry and infection by members of other virus families . Apilimod is an investigational drug . Although it has been tested in phase 2 clinical trials for the treatment of Crohn’s disease , psoriasis , and rheumatoid arthritis , the drug has not yet been approved for any indication . Nonetheless apilimod was well-tolerated in humans in the reported phase 2 trials [56–59] . We found that intraperitoneal delivery of 10 mg/kg of apilimod to mice resulted in a Cmax of 2 . 53 μM . This is well above the IC50 for apilimod inhibition of EBOV infection in the three cell lines tested , ~250 times greater than the IC50 in hMDMs ( 10 nM ) , initial major targets of filoviral infections [60 , 61] . We therefore consider it plausible that apilimod be used in the treatment of EVD . And , while apilimod may not function as a single agent , it may perform well as a component in an anti-filoviral small molecule cocktail . In summary , we introduce apilimod , a small molecule PIKfyve inhibitor that has proven safe in phase 2 clinical trials , as a potential anti-filoviral agent . | The recent outbreak of Ebola virus ( EBOV ) disease in Western Africa highlights the urgent need to develop therapeutics to help quell this devastating hemorrhagic fever virus , especially in resource-limited areas around the globe . Here we show that apilimod , an investigational drug that was well-tolerated in phase 2 clinical trials for rheumatoid arthritis , Crohn’s disease , and psoriasis , is a strong inhibitor of both EBOV and Marburgvirus infections in multiple cell types . Further work shows that apilimod blocks the entry of EBOV particles into the host cell cytoplasm and that it does so by blocking the particles from reaching their normal portal of entry , in Niemann-Pick C1-positive endolysosomes . Our findings are consistent with the identity of phosphatidylinositol-3-phosphate 5-kinase as the molecular target of apilimod , as the kinase and its product phosphatidylinositol 3 , 5-bisphosphate are required for the proper maturation of late endocytic organelles . Hence we propose that apilimod be further explored for repositioning as part of a therapeutic regimen to help ameliorate the sequelae of filoviral infections . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"blood",
"cells",
"medicine",
"and",
"health",
"sciences",
"vesicles",
"immune",
"cells",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"immunology",
"microbiology",
"drug",
"screening",
"viruses",
"filoviruses",
"clinical",
"medicine",
"rna",
"viruses",
... | 2017 | The phosphatidylinositol-3-phosphate 5-kinase inhibitor apilimod blocks filoviral entry and infection |
In nature , rabies virus ( RABV; genus Lyssavirus , family Rhabdoviridae ) represents an assemblage of phylogenetic lineages , associated with specific mammalian host species . Although it is generally accepted that RABV evolved originally in bats and further shifted to carnivores , mechanisms of such host shifts are poorly understood , and examples are rarely present in surveillance data . Outbreaks in carnivores caused by a RABV variant , associated with big brown bats , occurred repeatedly during 2001–2009 in the Flagstaff area of Arizona . After each outbreak , extensive control campaigns were undertaken , with no reports of further rabies cases in carnivores for the next several years . However , questions remained whether all outbreaks were caused by a single introduction and further perpetuation of bat RABV in carnivore populations , or each outbreak was caused by an independent introduction of a bat virus . Another question of concern was related to adaptive changes in the RABV genome associated with host shifts . To address these questions , we sequenced and analyzed 66 complete and 20 nearly complete RABV genomes , including those from the Flagstaff area and other similar outbreaks in carnivores , caused by bat RABVs , and representatives of the major RABV lineages circulating in North America and worldwide . Phylogenetic analysis demonstrated that each Flagstaff outbreak was caused by an independent introduction of bat RABV into populations of carnivores . Positive selection analysis confirmed the absence of post-shift changes in RABV genes . In contrast , convergent evolution analysis demonstrated several amino acids in the N , P , G and L proteins , which might be significant for pre-adaptation of bat viruses to cause effective infection in carnivores . The substitution S/T242 in the viral glycoprotein is of particular merit , as a similar substitution was suggested for pathogenicity of Nishigahara RABV strain . Roles of the amino acid changes , detected in our study , require additional investigations , using reverse genetics and other approaches .
Rabies virus ( RABV; genus Lyssavirus , Family Rhabdoviridae ) circulates worldwide ( except Australia , Antarctica and several insular territories ) in a variety of carnivores , and in New World bats [1] . Circulation of RABV in Old World bats was suggested several times but not corroborated [2] . Bats are principal reservoir hosts of all lyssaviruses except Mokola virus ( for which the principal hosts remain unknown ) . Generally , lyssaviruses are believed to have evolved originally in bats and later switched to carnivores [3] . Two such historical switches were inferred from viral phylogeny: one between lyssavirus genotypes ( or species , following the present virus taxonomy ) , and the other within the rabies virus [4] . The latter suggestion was inferred from the observation that the raccoon RABV lineage is related phylogenetically to bat RABV lineages . As shown later , the south-central skunk RABV variant is monophyletic with the raccoon and bat RABV variants as well [5] . However , the major questions regarding these two switches remain unanswered: where , from which bat species , and into which terrestrial mammalian species the first switch occurred , given that bat-associated RABV has only been found in the New World , and non-RABV lyssaviruses only in the Old World , always in bats ( except Mokola virus ) [2] , [3] . Moreover , there is no evidence that the raccoon or skunk RABV lineages facilitated transition of the virus to other carnivores , and the origins of viruses that represent multiple lineages of dog RABVs remain unknown . The RABV genome consists of five genes , coding for structural proteins including N ( nucleoprotein ) , P ( phosphoprotein ) , M ( matrix protein ) , G ( glycoprotein ) , and L ( large protein , or RNA-dependent RNA-polymerase ) . Each viral protein is multifunctional and significant for pathogenicity . The G protein is particularly important from the adaptive perspective , because it is responsible for recognition of host cell receptors and membrane fusion , promotes virus dissemination between infected cells , and stimulates host immune responses [6] , [7] . The major forces of RABV evolution include point mutations , introduced by the viral polymerase during genome replication due to the lack of proofreading mechanisms , and purifying selection [8] , [9] . The RABV populations demonstrate geographical variability , dominated by spatio-temporal separation ( most notable in the widely distributed dog RABV lineages [10] ) , and compartmentalization of the diversity based on the particular reservoir species [11] , [12] . Host adaptation mechanisms and host shifts/switches of lyssaviruses have not been explained . There was a suggestion that successful spill-over infections and host shifts within bats significantly depend on the host phylogeny ( i . e . , such events happen more frequently in evolutionary-related host species [12] ) . In addition , spill-over infections of bat RABVs in carnivores have been reported in the US often [13] , [14] , but these usually were single fatal events without perpetuation of transmission . Rare exceptions included a limited rabies outbreak in red foxes ( Vulpes vulpes ) on Prince Edward Island ( Canada ) , previously free of carnivore rabies . Monoclonal antibody typing of the viruses , isolated from rabid foxes , suggested their likely origin from mouse-eared bats , Myotis lucifugus or M . septentrionalis [15] . More recently , a limited local outbreak , caused by a RABV variant associated with T . brasiliensis bats was documented in white-nosed coatis ( Nasua narica ) in Mexico [16] . Our study is dedicated to the series of rabies outbreaks in mesocarnivores caused by bat rabies viruses in the Flagstaff area ( Coconino County ) of northern Arizona , USA . The series was believed to have begun during 2001: 19 rabid striped skunks ( Mephitis mephitis ) were encountered during January–July . They all were infected with a RABV variant associated with big brown bats ( Eptesicus fuscus ) [17] . Before 2001 , rabies in carnivores was documented only in the south of Arizona , where skunks and gray foxes ( Urocyon cinereoargenteus ) maintain circulation of their own specific RABV variants ( the south-central skunk and the Arizona gray fox variants , respectively ) , but not in the northern Arizona . The Flagstaff outbreak triggered significant public health attention . Control measures included prohibiting relocation of nuisance skunks by pest control companies , comprehensive public education , pet rabies vaccine clinics , and a 90-day emergency quarantine requiring pets to be leashed or confined , and vaccinated . Additionally , 217 urban skunks were vaccinated parenterally with inactivated rabies vaccine during a 6-month phased program of trap , vaccinate , and release ( TVR ) . The epizootic was eliminated ( due to the implemented efforts or spontaneously ) , and during the following two years no rabies cases in carnivores were registered in the Flagstaff area . However , additional outbreaks caused by the same RABV variant occurred during 2004–2005: 6 striped skunks , 2 gray foxes , and a domestic cat ( Felis catus ) , and during 2008–2009: 16 foxes , 1 skunk , and 1 ringtail cat ( Bassariscus astutus ) . Extensive TVR programs for skunks and oral rabies vaccination ( ORV ) for foxes were implemented each time . Both outbreaks ceased , but the question about possible continuous perpetuation and evolution of bat RABV in populations of mesocarnivores re-arose with new magnitude . The purpose of this study was to identify whether the Flagstaff outbreaks originated from a single introduction of bat RABV into mesocarnivores , with perpetuation in their populations , or each outbreak was caused by independent introduction of RABV . We also sought to identify whether the bat RABV variant , that caused Flagstaff outbreaks was different from other bat RABV variants , which had not caused such outbreaks in carnivores , and whether this RABV underwent recent genetic changes that might facilitate a host shift to carnivores ( given that no similar outbreaks were documented in the Flagstaff area before 2001 ) . To answer these questions , we undertook a large-scale genetic investigation of bat RABVs , isolated in the Flagstaff area during 2001–2009 from carnivores , via sequencing of their complete genomes , evolutionary analysis , and comparison to the genomes of viruses isolated from big brown bats during the same time frame and historically . Furthermore , we included into genetic analysis representative RABVs from all major phylogenetic lineages circulating in North American bats and carnivores , and other complete and partial genomes from RABV variants circulating in other parts of the world .
Brain specimens from carnivores and bats from the Flagstaff area , where the presence of RABV antigens were demonstrated by the direct fluorescent antibody ( DFA ) test [18] , were provided by the Arizona Department of Health Services , Bureau of State Laboratory Services , Phoenix , Arizona , USA , and the US Department of Agriculture ( USDA ) /Animal and Plant Health Inspection Service ( APHIS ) /Wildlife Services , Phoenix , Arizona , USA . At the Centers for Disease Control and Prevention ( CDC ) , Atlanta , Georgia , USA , the samples were subjected to typing with antinucleocapsid monoclonal antibodies [13] , and stored for further investigation at −80°C; . In total , we were able to collect brain specimens from 29 carnivores involved in the outbreaks caused by bat RABVs in the Flagstaff area during 2001–2009 ( 15 skunks from 2001; a skunk and a fox from 2004; a skunk and a domestic cat from 2005; 9 foxes and a ringtail cat from 2009 ) . For two specimens that were unavailable in our study , N gene sequences were collected from GenBank where they were present from the previous investigation [17] . In addition , we obtained brain material from 9 big brown bats , infected with the same RABV variant , for the period of 2001–2010 . Six historical big brown bat RABV isolates ( obtained during 1975–1999 ) , representing the same virus variant , and one isolate from a sister phylogenetic RABV lineage , were retrieved from CDC archives . For comparative purposes , we included representatives of the major RABV lineages circulating in North America which had been sent to CDC via routine surveillance or in the framework of previous projects . Total RNA was extracted from brain tissue using TRIZol reagent ( Invitrogen , Carlsbad , CA , USA ) . Amplification of RABV genes and sequencing of the complete genomes was performed as described previously [19] . The primers are present in Table S2 . The leader and trailer sequences were elucidated via RNA circularization , PCR-amplification across the ligated termini , and cloning of the PCR products [19] . For all carnivore specimens the complete genomes were generated from the primary field samples of brain . For several outbreak specimens which demonstrated over 99% nucleotide identity , the genome termini ( parts of the leader and trailer sequences , which demonstrated 100% identity or 1–2 random nucleotide substitutions ) were not generated , because this cumbersome operation was not expected to provide new valuable information . For the majority of bat samples , limited amounts of brain tissues were available . Therefore , amplification and sequencing of the N , P , M and G genes was performed from the field samples , whereas amplification and sequencing of the L gene and the genome termini was performed after one intracerebral passage in suckling mice . The reliability of this approach was tested on 17 randomly selected samples , for which sequences of the G mRNA ( about 2000 nucleotides ) were determined from the field samples and after one intracerebral passage in suckling mice . No nucleotide substitutions were detected after the passage when sequences were compared . For the 7 historical big brown bat RABV isolates , two vampire bat isolates from Mexico ( originating from cattle ) , and a white-nosed coati isolate from an outbreak documented in Mexico , the complete viral genomes were determined after one intracerebral passage in suckling mice , because no primary brain tissues were available . In total , we sequenced 66 complete RABV genomes , and 20 additional genomes with truncated leader and trailer sequences . We complemented the dataset with representative genomes of RABV lineages circulating worldwide , retrieved from GenBank . Given the limited number of complete RABV genomes available in GenBank and our focus on the New World RABV lineages , we also included in the dataset a greater number of sequences of separate viral genes ( with the complete coding region as the minimum requirement ) from RABV lineages circulating in the New World , either available from GenBank or generated during our study . Information on the samples is provided in the Table S1 and Figure 1 . The datasets included 168 , 157 , 103 , 172 and 103 sequences for the N , P , M , G and L genes , respectively . The assembly of viral genomes was performed either in MUSCLE [20] or in BioEdit , version 7 . 0 . 5 [21] . Translation of nucleotide sequences into deduced amino acid sequences , as well as calculation of identity values , was performed in BioEdit . Multiple alignments of complete genomes , separate genes , and non-coding genome regions were built by the ClustalW method , implemented in BioEdit and in ClustalX , version 2 . 0 . 17 [22] . The alignments were tested for recombinations using the Recombination Detection Program , version 2 [23] . Phylogenetic analysis was performed by the maximum likelihood method using PhyML , version 3 . 0 [24] . Statistical support of tree topology was derived from 500–1000 bootstrap replicates . Bayesian analyses were performed in MrBayes , version 3 . 2 [25] or in BEAST , version 1 . 3 [26] . Appropriate nucleotide substitution models were evaluated in MEGA , version 5 . 01 [27] . The general time-reversible model incorporating both invariant sites and a gamma distribution ( GTR+I+G ) was favored for all datasets . The base frequencies were estimated from the model , with the first two codon positions partitioned separately from codon position 3 , and expansion population growth parameterized for growth rate was used as the coalescent tree prior . Two simultaneous runs , each with four Markov chains , were performed for 10 , 000 , 000 generations and sampled every 1 , 000 generations . The final inference of the tree was summarized from both runs with the initial 10% of samples discarded as burn-in . The trees were visualized using FigTree program , version 1 . 3 . 1 [26] . To examine if there was episodic positive selection following putative host shifts , we performed branch-specific selection analyses using the CODEML program implemented in the PAML package , version 4 . 4 [28] . In these analyses , branch ( es ) expected to be under positive selection were specified as “foreground” branch ( es ) while the rest were specified as “background” branches . We separated the ω ( the ratios of non-synonymous to synonymous mutations , dN/dS ) of foreground and background branches , and estimated these two parameters under a branch-specific model . Positive selection was indicated if the ω ratio was significantly higher in foreground branches than in background branches . The significance of the result was evaluated through a likelihood ratio test ( LRT ) against a null model assuming a single ω ratio for all branches . To test whether RABV underwent positive selection following a host shift from bats to carnivores , we specified the branches leading to outbreak clusters and branches within the outbreak clusters as foreground , and other branches within the EF-W1 lineage as background branches ( Figures 2 , 3 ) . To account for potential difference of selection in different RABV genes , the tests were performed for each gene alignment separately . We performed amino acid ancestral reconstructions to find convergent or parallel changes for different clusters or lineages of RABVs . Based on the obtained phylogeny ( Figures 1 , S1 ) and the five gene datasets , we estimated the most likely amino acids at all nodes using an empirical Bayes method ( WAG+F as substitution model ) implemented in PAML . The estimated amino acids at ancestral nodes were then used to infer the changes that occurred along all branches in the tree . A Perl script was used to identify convergent amino acid changes ( i . e . changes from different amino acids that resulted in the same amino acid ) and parallel amino acid changes ( i . e . changes that start and result in the same amino acid ) between branch pairs , both as potential signal for convergent evolution .
The genomes varied in length from 11 , 899 nt to 11 , 930 nt . The shortest genome ( vampire bat isolate 3645DR ) had a long continuous deletion ( 23 nt ) within the G-L intergenic region ( the 3′ non-coding tail of the G mRNA ) , compared to the alignment of other RABV genomes . The second vampire RABV isolate in our study , 3634DR , did not have such a deletion . Another feature was detected in the leader regions of several bat RABV genomes . All RABV genomes described to date harbored A in the position 10 of their leader regions ( A10 ) . Furthermore , the first 12 nt of their leader regions were strictly complementary to the terminal 12 nt of their trailer regions [29] . However , all but one available in our study RABV genomes , associated with lasiurine bats and Perimyotis subflavus ( seven viruses from Lasiurus spp . , one from Lasionycteris noctivagans , and one from P . subflavus ) harbored G10 in their leader region ( introducing non-complementarity ) like it was documented previously in several non-RABV bat lyssaviruses [19] , [30] , [31] . The only virus from this cluster that did not have such a substitution was SHBRV-18 . Besides this cluster , we documented G10 in the genome of RABV associated with the pallid bat ( Antrozous pallidus ) . Another feature was observed in one of the several groups of RABV , recovered from skunks during the Flagstaff outbreaks ( Figure 2 ) . All nine viruses from the cluster 2001a had insertion of an additional incomplete transcription termination and polyadenylation signal ( TTP ) between the regular TTP of the M gene and the standard intergene dinucleotide CT: GAAAAAAACAAAAAACT instead of GAAAAAAACT . This feature was not observed in other RABV genomes . No evidence of recombination was observed in RABV genomes , generated during our study . The ML and Bayesian phylogenetic analyses produced trees of identical topology for each gene dataset . The tree in the Figure 1 is based on the G gene sequences , as this dataset was most representative in our study , and it indicates that we incorporated into analysis representatives from all major RABV lineages , described elsewhere [1] , [4] , [5] , [11] , [12] , [32] . The big brown bat viruses were segregated into two major clusters ( EF-W , EF-E ) , with two lineages in each . The lineages EF-W1 and EF-W2 represented the viruses circulating predominantly in the western part of North America , whereas lineages EF-E1 and EF-E2 represented the viruses of eastern and central distribution . Geographic overlaps in distribution areas of these lineages occur , as was described , for example , in Colorado [14] . If compared to the recent publication on the distribution and dynamics of big brown bat RABV in Canada [33] , our lineage EF-W2 corresponds to the lineage BB1 of that paper , whereas viruses from our lineage EF-W1 were not shown in that publication , perhaps because they are not present in Canada or their circulation is limited . Further , our lineage EF-E1 corresponds to the lineage BB5 ( that also includes sub-lineages BB3 and BB4 ) , and our lineage EF-E2 corresponds to the lineage BB2 from the same report [33] . All bat viruses that caused outbreaks in carnivores in the Flagstaff area during 2001–2009 belonged to the lineage EF-W1 ( Figures 1 , 2 ) . The nucleotide identity of viral gene sequences derived from bats in this lineage was relatively high: 98 . 0–99 . 0% , 97 . 1–99 . 8% , 98 . 1–100% , 97 . 7–99 . 6% and 98 . 2–99 . 7% for N , P , M , G and L genes , respectively ( amino acid identities 98 . 8–100% , 98 . 3–100% , 99 . 0–100% , 98 . 0–100% and 99 . 2–100% , respectively ) . Genomes of the oldest available isolates , such as AZBAT-7453 ( 1975 ) and AZBAT-6763 ( 1985 ) might be considered ancestral to minor sublineages , but in general fell within the diversity of viruses recovered from bats during 2001–2010 . The viruses recovered from skunks during 2001 were subdivided clearly into two lineages , 2001a and 2001b , based on the N , G , and L gene sequences . Despite the high identity values between the clades ( 99 . 5–99 . 8% ) , these clades were separated by 2 fixed synonymous substitutions within the N gene , 3 synonymous and 1 non-synonymous substitutions within the G gene , 7 synonymous and 1 non-synonymous substitutions within the L gene , and 2 substitutions within the G-L intergenic region . Moreover , all viruses from the lineage 2001a had the insertion of an additional incomplete TTP after the regular TTP of the M gene as described above . The bat virus SM4871 was more similar phylogenetically to the skunk lineage 2001a than to the skunk lineage 2001b ( although it did not have the insertion of incomplete TTP after the M gene ) . The P and M genes of viruses from clades 2001a , 2001b and from the bat SM4871 were identical , without any nucleotide substitutions . By comparison , the majority of viruses in each clade had 1–4 random nucleotide substitutions in the N , G , and L genes . The outbreak isolates from 2004–2005 available in our study were not closely related to the viruses from 2001 or to each other , except two skunk samples , SM5596 and SM1545 , which could have originated from one introduction of a bat RABV . The cat sample SM6709 was related to another group of bat viruses , and the fox sample SM5950 was different from all the above , but grouped with the bat sample CA100 . Therefore , these 4 viruses represent at least 3 independent introductions of bat RABV into carnivores . In the absence of other outbreak samples from 2004–2005 we cannot establish whether any of these sequenced viruses could be a part of the major outbreak ( potentially more introductions from bats could occur ) . Nevertheless , it is clear that neither of these 4 available viruses were related phylogenetically to the outbreaks that occurred during 2001 and 2008–2009 . The fox and ringtail cat viruses from 2009 constituted another monophyletic cluster of closely related viruses ( identity values 99 . 7–100% , with the most conserved P and M genes , as was also observed in the skunk viruses from 2001 , described above ) . The 2009 outbreak cluster was more similar phylogenetically to the bat virus A093500 than to the viruses recovered from previous outbreaks in carnivores . A single case of spill-over infection in a coyote ( Canis latrans ) with a virus from the same lineage EF-W1 was reported from Colorado during 2010 ( sample CO-coyot-2010; Figure 2 ) . Of the five genes tested for positive selection , two ( M and L genes ) were indicated to have significantly greater dN/dS following the host shift ( i . e . at foreground branches ) ( Table 1 ) . However , these positive signals were unlikely caused by adaptive evolution . In the M gene , the dN/dS ratio was estimated to be infinity within foreground branches . But the large value was due to the absence of synonymous changes rather than to excessive non-synonymous changes ( only 1 non-synonymous change was observed ) . In the L gene , the greater dN/dS ratio was due to a transient ( unfixed ) amino acid polymorphism , because most of the non-synonymous changes were distributed at terminal branches rather than at internal branches ( Table 1 , Figure 3 ) . Based on the ancestral reconstruction results , the ratio of non-synonymous mutation at terminal branches was 2 times greater than at internal branches . In one extreme case , 5 non-synonymous changes were observed in the branch leading to one virus , 2401 ( Table 1 ) . The overall dN/dS ratio dropped significantly at foreground branches after removing the sample 2401 from our dataset . The ancestral reconstruction results also indicated that RABV sequences were highly similar before and after the host shifts . This observation can be extended to noncoding regions , where almost identical sequences were observed between host shift samples and background ( bat ) samples . In other words , except for a few stochastic changes , the virus sequences remained the same following the shift into another host . No convergent or parallel changes were observed among the three independent host shift events ( lineages 2001a , 2001b , and 2009; Figure 3 ) , suggesting there was no detectable common mechanism that drove the adaptive evolution of viruses after each host shift . Therefore , we examined the hypothesis of pre-host shift adaptation . Under this hypothesis , we assumed that the ability to circulate in carnivores has been determined by neutral mutation in the RABV populations circulating in bats . We evaluated convergent and parallel changes between branches leading to EF-W lineages and those leading to carnivore-associated RABVs or bat RABVs associated with Desmodus rotundus ( DR ) and Tadarida brasiliensis ( TB ) ( the latter involved into host shift in the white-nosed coatis [16]; Figure S1 ) . Six parallel changes were observed between EF-W and carnivore RABV ancestral branches , whereas 3 parallel changes were observed between EF-W and DR&TB ancestral branches ( Table 2 ) . These sites detected to undergo convergent or parallel changes , except for site 464 in G gene , were all relatively conserved , with 2–5 substitutions across the entire phylogeny of RABV . In general , there were multiple fixed ( conserved within specific lineages ) amino acids in each protein that differentiated bat viruses from the “terrestrial” viruses , part of which were shared by viruses from the “bat-associated” RABVs from the raccoon ( RAC ) , south-central skunk ( SCSK ) , and one of Mexican skunk ( MexSK-1 ) lineages . Based on the amino acid patterns , all viruses from the Flagstaff outbreaks were typical bat viruses . Moreover , no conservative substitutions were detected when historical bat isolates from the EF-W1 lineage ( 1975–1999 ) were compared to the recent isolates ( 2001–2010 ) , either from bats or mesocarnivores . The number of substitutions was limited , and they all were randomly scattered among the sequences compared . For the amino acid sites detected to undergo convergent or parallel changes , we reconstructed their evolutionary histories throughout the entire RABV tree . Although 6 parallel changes were observed between EF-W and carnivore RABV lineages , they were not specific between these two groups . For five of the sites detected , more than one bat RABV lineages shared the same convergence between EF-W and carnivore-associated RABV ( Figure S3 , A , B , C , D , and E ) , whereas no host shifts into mesocarnivores have been reported for those bat RABV lineages . One parallel change ( P→S485 in the G ) was unique for EF-W among all bats lineages . However , this site experienced multiple S→P substitutions within the carnivore RABV clades ( Figure S3 , F ) . On the other hand , the convergence between EF-W and DR&TB lineages involved three sites which did not overlap with carnivore RABVs . Among these , position 464 of the G was hyper-variable and thus was not likely to be involved in pre- shift adaptation ( Figure S3 , G ) ; position 107 of the L had only two variants: H and Y , but the H variant was shared by both bat and carnivore RABV clades ( Figure S3 , H ) , and therefore this site is not likely to contribute to the adaptive differences of RABV between bats and terrestrial mammals . Only the substitution S→T242 in the G ectodomain merits additional attention ( Figure S2 ) . All but one representatives of carnivore RABV lineages , available in our study , have A242 . These include such “bat-associated” carnivore RABV lineages as RAC and SCSK but exclude MexSK-1 , which has T242 . In contrast , the majority of bat viruses have S242 . The exceptions include all big brown bat RABVs ( viruses from lineages EF-E1 and EF-E2 have A242 , whereas viruses from lineages EF-W1 and EF-W2 have T242 ) . In addition , T242 was found in RABVs , associated with T . brasiliensis ( lineage TB ) and in several bat viruses from Brazil , predominantly associated with Eptesicus furinalis ( lineage EFu; Figure S2 ) . Bat viruses with T242 have caused host shifts to carnivores more frequently than viruses with S242 . These included 3 outbreaks in the Flagstaff area ( or even 4 , if we consider that viruses from the outbreak of 2004–2005 , unavailable in our study , were the result of one introduction from bats ) ; an outbreak in gray foxes in Oregon during 2010 , that was caused by an introduction of a big brown bat RABV from lineage EF-W2 [34]; and an outbreak in coatis in Mexico during 2008 [16] , caused by a bat virus from lineage TB . In contrast , viruses with S242 , despite frequently detected spillover infections , caused only one documented outbreak in gray foxes in Oregon during 2009 [34] , [35]; Figure S2 .
One of the major questions we addressed in this study was whether bat RABV was re-introduced repeatedly into populations of mesocarnivores in the Flagstaff area during 2001–2009 , or perpetuated in their populations after a single introduction in 2001 . Our phylogenetic reconstructions demonstrated clearly that multiple introductions of bat viruses from the EF-W1 clade occurred into the skunk and fox populations independently . Moreover , the outbreak of 2001 most likely resulted from two virus incursions , each causing secondary transmission within the skunk population . The differences between viruses from these two clusters are supported by molecular and epidemiological data . Viruses from clade 2001a were collected during January–April from the northeastern part of the Flagstaff area , whereas viruses from clade 2001b were collected during March–July from the southern and western parts [17] . An alternative hypothesis , that each rabies case in carnivores could be an individual spill-over infection from a bat , was rejected as well because the sequence similarity of viruses within each outbreak cluster was much greater than the similarity between bat viruses from the EF-W1 lineage , and between viruses from different outbreaks . Each host shift of bat viruses into carnivores in the Flagstaff outbreaks was transient , likely due to control efforts of state and federal services , including TVR programs for skunks and ORV for gray foxes . However , it is unknown what would have happened if no such significant efforts for outbreak containment had been implemented: whether outbreaks would self-eliminate , or the virus would perpetuate in populations of mesocarnivores . Virus circulation is a balanced equilibrium of multiple components , including genetic and antigenic properties of the pathogen , pathobiology of infection on the individual host level , and ecological properties of the host on the population level [3] . Although significant knowledge has been generated during a long history of studies on rabies in wildlife [36] , [37] , the evolutionary genetics and host shifts of RABV and other emerging viruses are poorly understood [12] , [38] . Putative adaptation at the sequence level , if present , can take place at two stages of host shift . First , it may take place after cross-species transmission , for the virus to adjust to the new host environment ( post-shift adaptation ) . Under this scenario , the virus may undergo a series of active changes in the genome under selective pressures to adapt to the new host , which can be reflected by higher dN/dS values or non-neutral convergent evolution . We performed positive selection analyses on branches that represent the post-host shift evolution . Although the test yielded positive results in M and L genes , they were most likely false positives . In fact , the EF-W1 viruses isolated from mesocarnivores during the Flagstaff outbreaks were closely related and could not be distinguished from those isolated from bats . Only a limited number of amino acid changes occurred at the post-shift stage , and these were found on terminal branches instead of on internal branches , indicating a lack of amino acid fixation after the host shift . Moreover , no convergence was detected in these amino acid changes among the three independent host shift events . Therefore , the RABVs in this study are unlikely to have undergone post-shift adaptation . As we established that each outbreak was caused by a separate virus introduction and continued only for several months , it is not surprising that adaptive changes in viral genomes did not accumulate during such a short time span . We addressed whether viruses from the EF-W1 lineage that caused Flagstaff outbreaks were different from those circulating in bats historically . The phylogenetic analyses indicated that the historical bat isolates from the EF-W1 lineage were closely related to those causing multiple host-shift events , which suggested that viruses of this lineage have been capable of causing host shifts at least since 1975 . Previous presumptive outbreaks in carnivores might have been missed because of the inherited limitations of the passive surveillance system . Alternatively , ecological factors may have favored outbreak conditions and multiple spill-over incidents during 2001 , 2004–2005 and 2008–2009 . In general , a virus circulating in a certain reservoir host may already be competent for circulation in a new host ( pre-shift adaptation ) . We examined this scenario by looking for neutral convergence between the EF-W and carnivore RABV lineages , and between the EF-W and DR+TB lineages . A number of sites were discovered to have converged between these lineages/clades . However , it is still unclear whether these convergences rendered a higher chance for successful host shift events . One amino acid substitution which can be considered potentially as a marker of pre-adaptation of bat RABV to cause effective infection in carnivores is the S→T242 in the G ectodomain . The amino acid in this position was significant for pathogenicity in the Nishigahara strain: the parental virus with A242 was pathogenic in a mouse model , whereas a mutant with S242 was attenuated [39] . Another recent study demonstrated that A242 increases virus spread between infected cells [40] . No studies on other amino acids in this position , including T242 , have been described to date . The RABV variants associated with carnivores have A242 , and one lineage associated with Mexican skunks ( MexSK-1 ) has T242 . In contrast , the majority of bat RABV variants have S242 . We know only one outbreak , caused by a virus with S242 ( of a Myotis bat origin ) in gray foxes in Oregon , during 2009 [35] . That outbreak was self-eliminating without implementation of any control actions . It is notable that lyssaviruses of other species ( which all except Mokola virus are associated with bats ) also have S242 , and infect carnivores very infrequently [41] . This may suggest that lyssaviruses with S242 are less pathogenic to carnivores . By comparison , bat viruses with T242 caused at least five outbreaks in carnivores , documented during the last decade , and were transmitted efficiently among skunks , foxes and coatis . However , the significance of T242 for the ability of bat viruses to switch to carnivores should be appreciated with caveats . For example , T242 is present in viruses of the TB lineage . Only one outbreak caused by a virus from this lineage was documented in coatis in Mexico [16] . However , RABV of the TB lineage is prevalent in large populations of T . brasiliensis bats in the southern US , but no outbreaks were documented in mesocarnivores from these areas , despite their frequent chances of exposure to rabid bats . Moreover , the viruses from EF-E1 and EF-E2 lineages broadly distributed in North America have A242 , similarly to the viruses circulating in carnivores . Nevertheless , no outbreaks in carnivores caused by these viruses were documented in the US and Canada . Additional reverse genetics studies coupled with animal experiments in captivity are needed to elucidate the significance of T242 in the RABV G ectodomain for adaptation of bat viruses to circulation in carnivores . In general , our study resolved several questions on the origin of Flagstaff rabies outbreaks , phylogenetic relationships of viruses involved in these outbreaks , and stability of their genomes over a limited period of time . Concurrently , our data raise questions on different aspects of RABV shifts from bats to carnivores . These include the significance of substitutions detected in viral genomes under different evolutionary models; likelihood of host shift to carnivores for different bat RABV variants; and outcomes of outbreaks in carnivores , caused by bat viruses , if no containment measures are implemented by humans . In addition , we generated a significant dataset of complete and near-complete genomes of RABV from different lineages which warrant further extensive investigations . Our findings pose important questions on the molecular biology of RABV . For example , what is the significance of A→G10 substitution in the leader region , which previously had been observed only in several non-RABV bat lyssaviruses ? This substitution and the following lack of complementarity may be important for virus replication , as the termini are involved in genome encapsidation and polymerase recognition [42] . It is remarkable that to date this substitution has been found only in genomes of bat lyssaviruses from different species [19] , [30] , [31] . On the other hand , this finding demonstrates inappropriateness of generation of “complete” viral genomes via RT-PCR with primers complementary to the genome termini , and substituting sequences of the real viral genome termini by sequences of the primers used for their amplification [43] . In this context , the termini of the SHBRV-18 genome should also be confirmed , as this was the only virus from the lasiurine RABV cluster that had A10 in the leader region . The SHBRV-18 was isolated from a human , presumably infected by L . noctivagans , and was subjected to a passage in suckling mice before genome sequencing [44] . Further phylogenetic refinement demonstrated that SHBRV-18 belongs to the RABV variant associated with P . subflavus , rather than with L . noctivagans . Potentially , alteration of the leader sequence could occur during these passages in heterologous models . Another interesting observation was the insertion of an additional incomplete TTP after the regular TTP of the M gene in one cluster of viruses , recovered from skunks during an outbreak of 2001 . Previously , similar insertions were observed in the non-coding region of the N gene , right before the regular TTP in several specimens of European bat lyssavirus , type 1 [45] . The significance of such insertions is unknown . Perhaps , they appear randomly in RABV genomes as a result of TTP duplication , but are eliminated during further virus passages because of redundancy . At least during the Flagstaff outbreak , such viruses were efficiently transmitted between skunks . These and other observations made in our study warrant further investigations on multiple levels , including evolutionary and functional studies . The gene sequences , generated during this study , were deposited in GenBank , with accession numbers JQ685892-JQ686013 . | Host shifts of the rabies virus ( RABV ) from bats to carnivores are important for our understanding of viral evolution and emergence , and have significant public health implications , particularly for the areas where “terrestrial” rabies has been eliminated . In this study we addressed several rabies outbreaks in carnivores that occurred in the Flagstaff area of Arizona during 2001–2009 , and caused by the RABV variant associated with big brown bats ( Eptesicus fuscus ) . Based on phylogenetic analysis we demonstrated that each outbreak resulted from a separate introduction of bat RABV into populations of carnivores . No post-shift changes in viral genomes were detected under the positive selection analysis . Trying to answer the question why certain bat RABV variants are capable for host shifts to carnivores and other variants are not , we developed a convergent evolution analysis , and implemented it for multiple RABV lineages circulating worldwide . This analysis identified several amino acids in RABV proteins which may facilitate host shifts from bats to carnivores . Precise roles of these amino acids require additional investigations , using reverse genetics and animal experimentation . In general , our approach and the results obtained can be used for prediction of host shifts and emergence of other zoonotic pathogens . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"public",
"health",
"and",
"epidemiology",
"genome",
"evolution",
"microbiology",
"convergent",
"evolution",
"veterinary",
"science",
"infectious",
"diseases",
"forms",
"of",
"evolution",
"veterinary",
"diseases",
"epidemiology",
"biology",
"viral",
"evolution... | 2012 | Molecular Inferences Suggest Multiple Host Shifts of Rabies Viruses from Bats to Mesocarnivores in Arizona during 2001–2009 |
σ factors provide RNA polymerase with promoter specificity in bacteria . Some σ factors require activation in order to interact with RNA polymerase and transcribe target genes . The Extra-Cytoplasmic Function ( ECF ) σ factor , σV , is encoded by several Gram-positive bacteria and is specifically activated by lysozyme . This activation requires the proteolytic destruction of the anti-σ factor RsiV via a process of regulated intramembrane proteolysis ( RIP ) . In many cases proteases that cleave at site-1 are thought to directly sense a signal and initiate the RIP process . We previously suggested binding of lysozyme to RsiV initiated the proteolytic destruction of RsiV and activation of σV . Here we determined the X-ray crystal structure of the RsiV-lysozyme complex at 2 . 3 Å which revealed that RsiV and lysozyme make extensive contacts . We constructed RsiV mutants with altered abilities to bind lysozyme . We find that mutants that are unable to bind lysozyme block site-1 cleavage of RsiV and σV activation in response to lysozyme . Taken together these data demonstrate that RsiV is a receptor for lysozyme and binding of RsiV to lysozyme is required for σV activation . In addition , the co-structure revealed that RsiV binds to the lysozyme active site pocket . We provide evidence that in addition to acting as a sensor for the presence of lysozyme , RsiV also inhibits lysozyme activity . Thus we have demonstrated that RsiV is a protein with multiple functions . RsiV inhibits σV activity in the absence of lysozyme , RsiV binds lysozyme triggering σV activation and RsiV inhibits the enzymatic activity of lysozyme .
In order to survive in rapidly changing environmental conditions , bacteria use signal transduction systems to transmit information from outside the cell across the membrane to alter transcriptional responses . In bacteria , Extra-Cytoplasmic Function ( ECF ) σ factors are one class of signal transduction system capable of responding to extracellular signals . ECF σ factors represent the largest and most diverse group of σ factors [1] . However , one common feature of many ECF σ factors is that they are sequestered in an inactive state by an anti-σ factor and must be activated in order to interact with RNA polymerase . In many cases the signals that induce activity of these ECF σ factors and the molecular mechanisms controlling activation are not well understood . The anti-σ factor is responsible for inhibiting ECF σ factor activity by blocking its association with RNA polymerase in the absence of signal . Activation of ECF σ factors occurs via modification of the anti-σ factor , leading to release of the ECF σ factor or transition the ECF σ factor to an active state , allowing interaction with RNA polymerase . The activation of several ECF σ factors occurs via a mechanism termed Regulated Intramembrane Proteolysis ( RIP ) , which results in the sequential cleavage of the anti-σ factor in response to extracellular stress [2–4] . RIP is initiated by a cleavage event at site-1 of the anti-σ factor and this initial cleavage event usually occurs on an extracellular domain of the anti-σ factor . Following site-1 cleavage a second protease cuts within the transmembrane domain of the anti-σ factor at site-2 . The remaining cytosolic portion of the anti-σ factor is then destroyed by cytosolic proteases [2–4] . The B . subtilis ECF σ factor , σV , belongs to the ECF30 subfamily of ECF σ factors , which are found almost exclusively in Firmicutes ( low GC Gram-positive bacteria ) [1] . The activity of a subset of the ECF30 σ factor homologs are inhibited by anti-σ factors homologous to RsiV . C-type lysozyme activates σV in B . subtilis [5 , 6] and in other bacteria encoding homologous systems including C . difficile and E . faecalis [7–10] . σV is activated by RIP mediated degradation of the transmembrane anti-σ factor RsiV in response to C-type lysozyme [11 , 12] . In each of these organisms free σV then interacts with RNA polymerase to transcribe genes required for lysozyme resistance [5 , 6 , 8 , 9 , 13 , 14] . In B . subtilis this includes oatA which encodes a peptidoglycan O-acetylase that adds an acetyl group to peptidoglycan [15] which increases resistance to lysozyme [5 , 6] . σV is also required for lysozyme inducible expression of dltABCDE which encode enzymes responsible for the addition of D-alanine ( D-ala ) to teichoic acids of the cell wall [6 , 16] . Increasing the positive charge of the teichoic acids presumably repels the positively charged lysozyme [14 , 17–22] . Activation of σV in C . difficile results in a large increase in expression of the operon encoding σV and RsiV ( pdaVprsA2csfVrsiVcd1560 ) [8] and to a lesser degree the dltABCDE operon [14] . The activity of σV is inhibited by RsiV . In order to activate σV RsiV must be degraded in a RIP-dependent manner . Previous work from our laboratory has shown signal peptidase is responsible for the initial site-1 cleavage of RsiV , which removes the extracellular domain of RsiV [12] . Following site-1 cleavage , the truncated form of RsiV is cleaved by the site-2 protease RasP [11] . In other ECF σ factor systems RIP is initiated by the site-1 protease which is thought to be responsible for sensing the inducing signal and initiating RIP of the anti-σ [2 , 3] . In the case of E . coli , σE activation is initiated by the binding of unfolded outer membrane proteins to the site-1 protease DegS [23 , 24] . Interestingly RseB a negative regulator of site-1 cleavage has also been implicated in sensing envelope stress by directly binding lipid A fragments [25] . In the case of B . subtilis σW activation , it is less clear how PrsW controls site-1 cleavage in response to cell stress , but evidence suggests that it can act as a sensor since constitutively active mutants have been isolated that alter PrsW activity [26] . Since the activity of signal peptidase is not known to be controlled by extracellular signals , it raised the question of how site-1 cleavage of RsiV is initiated in the presence of lysozyme [27 , 28] . Interestingly signal peptidase could cleave RsiV at site-1 in vitro but only in the presence of lysozyme [12] . We found the anti-σ factor RsiV directly binds to lysozyme [12] . In addition , we found that lysozyme but not the structurally distinct muramidase mutanolysin could activate and bind RsiV . Thus we proposed a model where activation of σV is controlled by the binding of lysozyme to the anti-σ factor RsiV . Here we demonstrate that the binding of RsiV to lysozyme is required for activation of the B . subtilis ECF σ factor σV in response to lysozyme . We determined the x-ray co-crystal structure of the RsiV-hen egg white ( HEW ) lysozyme at 2 . 3 Å . Using the structure as a guide we identify residues required for the binding of RsiV to lysozyme and demonstrate that this binding is essential to induce proteolysis of RsiV and thus σV activation . Taken together this demonstrates that RsiV is a receptor for lysozyme and that binding of RsiV to lysozyme is required for initiating RIP and σV activation in B . subtilis . The structure also revealed RsiV binds to the active site of lysozyme and we show that RsiV can inhibit activity of lysozyme in vitro and RsiV mutants that are unable to bind lysozyme are also unable to inhibit its activity .
We previously found the extracellular domain of RsiV ( RsiV58-285 ) was able to bind C-type lysozyme both in vitro and in vivo [12] . To better understand this interaction we determined the x-ray crystal structure of the RsiV-lysozyme complex at 2 . 3 Å ( PDB ID: 5JEN; Fig 1A and Table 1 ) . The RsiV-lysozyme complex crystallized as a heterodimer and the crystal structure revealed extensive contacts between RsiV and lysozyme ( Fig 1A ) . Based on PDBePISA analysis RsiV has a surface area of 12184 Å2 and lysozyme has a surface area of 6503 Å2 and they share an interface of 1407 . 3 Å2 [29] further demonstrating the large interaction interface between the two proteins . RsiV is made up of two domains of unknown function ( DUF4163 and DUF3298 ) that are often located within the same protein ( S1 Fig ) . The crystal structure revealed that the two DUF domains are tightly associated ( S1 Fig ) . The amino acids ( 79–162 ) of RsiV make up DUF4163 and the amino acids ( 188–270 ) encompass DUF3298 ( S1 Fig ) [30] . The N-terminal domain of the RsiV structure which encompasses DUF4163 shows structural similarity to β-grasp domain containing protein Rv1980c from M . tuberculosis ( PDB:2HHI ) ( S2 Fig ) [31–33] . Using Vast+ we determined the best structural match to RsiV in the PDB is 3CYG an uncharacterized protein from Fervidobacterium nodosum rt17 ( S3 Fig ) [34] . There is also similarity to two other proteins of unknown function BF2082 from Bacteroides fragilis NCTC 9343 ( PDB:3S5T ) and PA4972 from Pseudomonas aeruginosa PA01 ( PDB:4E72 ) ( S2 Fig ) [34] . However , none of these proteins are encoded near ECF σ factors and in some cases ( PA4972 ) do not appear to contain a transmembrane domain . In addition the amino acid homology between these structures is very limited ( S3 Fig ) . To determine if lysozyme binding was a feature retained by this entire class of proteins we obtained a clone of 6xHis-Rv1980c used to determine the structure of Rv1980c [31] . We tested the ability of Rv1980c to bind lysozyme using co-purification . We found that when 6xHis-Rv1980c was bound to a nickel column lysozyme did not co-elute with Rv1980c ( S4 Fig ) . This suggests that not all proteins containing DUF4163 and DUFM3298 bind lysozyme . In order to determine the biological significance of RsiV binding lysozyme we sought to disrupt the interaction . To accomplish this , we used the co-crystal structure to identify residues involved in the interaction between RsiV and lysozyme ( Table 2 ) . We identified multiple residues that were not grouped together in a single linear segment but were spread over three distinct loops that come together when RsiV is folded ( Fig 1C ) . RsiV homologs are present in a number of different Firmicutes [1] and we demonstrated RsiV homologs from C . difficile and E . faecalis bind lysozyme [12] . This suggests that lysozyme binding is likely conserved amongst RsiV homologs . BLAST was used to identify the sequences of 898 RsiV homologs ( S1 Table ) [35] . A multiple sequence alignment of these homologs was then generated using ClustalW [36] . Using this alignment , we mapped sequence conservation onto the structure of RsiV using ConSurf [37 , 38] . This analysis revealed a pocket of high conservation in the “saddle” of RsiV of which numerous residues directly contact lysozyme ( Fig 2A and Table 2 ) . The most highly conserved residues aligned with those that interact with lysozyme , as determined by our co-crystal structure . We performed similar analysis on 400 C-type lysozymes . This analysis revealed that the highest conservation is located in the peptidoglycan binding pocket and active site of lysozyme which is where RsiV binds to lysozyme ( S5 Fig ) . Thus as expected the most highly conserved regions of both RsiV homologs and lysozyme are involved in the interaction between the two proteins . Previous work showed that the extracellular domain of RsiV binds lysozyme and in vitro site-1 cleavage of RsiV by signal peptidase required the presence of lysozyme [12] . We hypothesized that activation of σV is dependent upon binding of RsiV to lysozyme . To directly test this model , we sought to identify mutants of RsiV that are unable to bind lysozyme and determine if they are also unable to activate σV . We began by constructing mutations of highly conserved residues of RsiV which interacted with lysozyme . We then purified each of these mutant proteins and tested their ability to bind lysozyme in vitro . As previously reported we found that wild type RsiV59-285 binds lysozyme with a Kd of 70 nM ( Table 3 ) [12] . We found the mutant proteins RsiV59-285 , P259A and RsiV59-285 , Y261A were able to bind lysozyme as well as wild type ( Table 3 and S6 Fig ) . In contrast we found that RsiV59-285 , S169W showed a reduced affinity for lysozyme , but not a complete loss of binding as compared to the lysozyme only control . ( Table 3 , S6 Fig , and Fig 3C ) . Since RsiV makes multiple contacts with lysozyme ( Table 2 ) , we hypothesized that a single mutation may not be sufficient to completely block lysozyme binding , thus we analyzed the ability of RsiV double and triple mutants to bind lysozyme . The double mutants RsiV59-285 , P259A Y261A and RsiV59-285 , S169W P259A were able to bind lysozyme , but with a ~100 fold and ~3 fold less affinity , respectively ( Table 3 and S6 Fig ) . The RsiV59-285 , S169W , P259A , Y261A triple mutant behaved similar to a lysozyme only injected control , suggesting a near complete loss of lysozyme binding activity ( Table 3 and Fig 3B and 3C ) . Circular dichroism ( CD ) analysis was used to determine if the RsiV59-285 , S169W , P259A , Y261A triple mutant was simply unable to properly fold . We determined the RsiV59-285 , S169W , P259A , Y261A triple mutant retains similar α-helix and β-sheet conformation as wild type suggesting the overall secondary structure is only slightly altered ( S7 Fig ) . Taken together these data suggest that multiple regions are involved in RsiV binding to lysozyme and these contribute to the high affinity this receptor displays for its ligand . We hypothesize that RsiV binding to lysozyme is required for σV activation . Therefore we analyzed the effect of RsiV mutants defective in lysozyme binding on σV activation . The single , double and triple mutant rsiV alleles were integrated into the native chromosomal site by homologous recombination to allow for σV dependent regulation of rsiV expression . We confirmed that all of these strains produced similar basal amounts of RsiV , in the absence of lysozyme , by immunoblot ( Fig 4B ) . This demonstrates that these mutant proteins are stable and are not altered for σV activation in the absence of lysozyme . The activation of σV in the resulting strains was determined by exposure to increasing concentrations of lysozyme ( 0 , 2 , 5 , 10 , 20 μg/ml ) by measuring expression from the PsigV-lacZ reporter . We found that the triple mutant rsiVS169W , P259A , Y261A abolished σV activation ( Fig 4A ) . As previously mentioned , immunoblot analysis shows that RsiVS169W , P259A , Y261A was produced at similar levels to wild type RsiV ( Fig 4B ) . So this lack of activation is not simply due to insufficient or unstable protein . CD analysis suggests that RsiVS169W , P259A , Y261A is not simply improperly unfolded ( S7 Fig ) . In addition , if RsiVS169W , P259A , Y261A were unfolded we would predict that it would be constitutively degraded via RIP . We also tested the role of the single and double mutants of RsiV and found they fell into two distinct groups: 1 ) mutants that had near wild type levels of σV activation; RsiVP259A and RsiVY261A ( Fig 4A ) , and 2 ) mutants that reduce , but do not abolish , σV activation; RsiVS169W , RsiVP259A , Y261A , RsiVS169W , P259A , and RsiVS169W , Y261A ( Fig 4A ) . We found that the reduced activation of σV in RsiV mutants , RsiVS169W , RsiVP259A , Y261A , RsiVS169W , P259A , and RsiVS169W , Y261A , was dependent upon lysozyme concentration ( Fig 4 ) . At higher lysozyme concentrations these mutants were still able to activate σV . All of these data are consistent with the model in which these mutants show lower affinity for lysozyme and thus activate σV only at higher lysozyme concentrations . Thus , we have demonstrated the overall the ability of the RsiV mutants to activate σV correlates with their ability to bind lysozyme . Activation of σV requires the degradation of RsiV via RIP [11 , 12] . Thus we sought to determine if RsiV mutants which blocked or decreased binding of RsiV to lysozyme also blocked RIP of RsiV in response to lysozyme . We expressed wild type or mutant forms of RsiV , using an IPTG inducible promoter ( Phs ) in a strain of B . subtilis with a ΔsigVrsiV::kan mutation to break the auto-regulation controlling RsiV production . Control experiments revealed that when all strains were grown in 1 mM IPTG RsiVP259A and RsiVS169W Y261A were produced at about 6-fold higher levels compared to the other strains ( S8A Fig ) . We found strains producing RsiVP259A and RsiVS169W Y261A when grown in media with 0 . 1 mM IPTG resulted in production of similar levels of RsiV compared to other strains grown in the presence of 1 mM IPTG ( S8B Fig ) . Thus for these experiments the concentration of IPTG used differed in order to produce the RsiV mutant proteins at similar levels . Cells producing wild type or mutant forms of RsiV were exposed to a gradient of lysozyme concentrations ( 0 , 0 . 01 , 0 . 1 , 1 , 2 μg/ml ) for 10 minutes and cleavage of RsiV was monitored by immunoblot using an RsiV specific antibody . As previously observed , we found that wild type RsiV is nearly completely cleaved at site-1 in 10 minutes in the presence of as little as 0 . 1 μg/ml of lysozyme ( Fig 5 ) [11 , 12] . In contrast the RsiV triple mutant ( RsiVS169W , P259A , Y261A ) , which was unable to bind lysozyme and activate σV , was not degraded even at the highest concentration of lysozyme ( 2 μg/ml ) ( Fig 5 ) . This strongly suggests it can no longer sense lysozyme . Similar to the effects on σV activation , we observed the RsiV single and double mutants displayed intermediate phenotypes compared to wild type RsiV ( Fig 5 ) . RsiVP259A and RsiVY261A were degraded similar to wild type RsiV ( Fig 5 ) . Mutants RsiVS169W , RsiVP259A , Y261A , RsiVS169W P259A , and RsiVS169W Y261A are degraded in the presence of the highest concentration of lysozyme ( Fig 5 ) . However at lower concentrations ( 1 μg/ml and 0 . 1 μg/ml ) the mutant RsiV remains intact , while wild type RsiV is rapidly degraded . Taken together these data support a true receptor-ligand signaling model that requires lysozyme binding to RsiV for activation of σV . To determine if binding of RsiV to lysozyme was required for lysozyme resistance we performed assays to measure the minimum inhibitory concentration ( MIC ) by lysozyme on strains with the single , double and triple mutants integrated into the native chromosomal locus . As previously reported , we found the ΔsigVrsiV mutant showed increased sensitivity to lysozyme compared to wild type B . subtilis ( Table 4 ) [5] . As predicted from the data above , we found that that the cells producing the RsiV triple mutant ( RsiVS169W , P259A , Y261A ) were nearly as sensitive to lysozyme as the ΔsigVrsiV ( Table 4 ) . Similarly , cells producing either single or double mutant RsiV proteins showed intermediate resistance to lysozyme , correlating with their ability to activate σV and bind lysozyme ( Table 4 ) . Taken together this demonstrates that an inability of RsiV to bind lysozyme results in increased sensitivity to lysozyme . Analysis of the RsiV-lysozyme co-crystal structure revealed RsiV binds in the enzymatic pocket of lysozyme containing the active site residues D52 and E35 as well as several residues required for binding peptidoglycan [39 , 40] . Thus we hypothesized RsiV could act as an inhibitor of lysozyme activity . We assayed the activity of lysozyme in the presence and absence of RsiV using a standard lysozyme activity assay [41] . Wild type RsiV59-285 almost completely inhibits lysozyme activity at a 1:1 molar ratio , suggesting RsiV is an effective lysozyme inhibitor ( Fig 6 ) . We then compared the ability each of the RsiV mutants to inhibit lysozyme activity . The mutants RsiV59-285 , P259A and RsiV59-285 , Y261A retain the ability to inhibit lysozyme to the same degree as wild type RsiV ( Fig 6 ) . The mutants RsiV59-285 , S169W , RsiV59-285 , S169W Y261A , RsiV59-285 , S169W P269A , RsiV59-285 , P259A and Y261A , and RsiV59-285 , S169W , P259A , Y261 were not able to inhibit lysozyme as well as wild type ( Fig 6 ) . We found that at higher concentrations , 2 and 4 molar ratios , the mutants RsiV59-285 , S169W and RsiV59-285 , S169W Y261A almost completely abolish lysozyme activity ( S9 Fig ) . However , the mutants RsiV59-285 , S169W P269A , RsiV59-285 , P259A and Y261A , and RsiV59-285 , S169W , P259A , Y261 showed significant decreases in the ability to inhibit lysozyme activity even at a molar ratio of 4:1 ( S9 Fig ) . Together the data suggest lysozyme inhibition correlates with ability of RsiV to bind lysozyme ( Fig 6 and S9 Fig ) . To determine if RsiV was produced at a level sufficient to inhibit lysozyme activity we determined the levels of RsiV in the absence of lysozyme and estimated RsiV levels in the presence of lysozyme . In wild type B . subtilis there are approximately 220 molecules of RsiV per cell ( S10 Fig ) . We found that the cells harboring the IPTG inducible construct of rsiV produce ~7420 molecules per cell . Since RsiV is cleaved upon binding lysozyme and released into the supernatant it was more difficult to determine RsiV protein levels in the presence of lysozyme . Therefore , we estimated the levels of RsiV protein by comparing the levels of rsiV transcript in uninduced and induced wild type cells to cells harboring the IPTG inducible construct of rsiV . The rsiV transcript was induced ~100 fold upon addition of 1 . 25 μg/ml of lysozyme in wild type cells ( S2 Table , [5 , 6] ) . We found that rsiV transcript was ~44 fold higher in cells harboring the IPTG inducible construct of rsiV compared to uninduced wild type cells ( S2 Table ) . Thus we estimate there are approximately 16700 molecules of RsiV per cell when induced with 1 . 25 μg/ml of lysozyme which is the equivalent of ~52600 molecules of lysozyme per cell ( S2 Table ) . This suggests RsiV is produced at levels high enough to function as a competitive inhibitor of lysozyme . We sought to determine if RsiV contributes to lysozyme resistance in vivo by identifying the minimum inhibitory concentration ( MIC ) of cells producing RsiV from an exogenous promoter integrated at amyE in the absence of σV . We observed that the ΔsigVrsiV mutant strain was 3-fold more sensitive to lysozyme compared to wild type B . subtilis ( Table 5 ) . Cells producing RsiV in a ΔsigVrsiV are 1 . 5 fold more resistant to lysozyme than the ΔsigVrsiV parent strain ( Table 5 ) , suggesting RsiV inhibition of lysozyme activity contributes to σV-induced lysozyme resistance .
We determined the x-ray structure of RsiV and lysozyme complex at 2 . 3 Å . Using the co-structure we demonstrate that RsiV is a receptor for lysozyme and this interaction is required for σV activation . Our results indicated that due to the extensive contacts between RsiV and lysozyme a single mutation was not sufficient to block binding or σV activation . We found that several amino acid residues , S169 and P259 Y261 , are critical for RsiV binding to lysozyme . These residues are located in two distinct loops of RsiV that protrude from the top of the structure and interact with lysozyme . Interestingly these residues are highly conserved amongst RsiV homologs . Our data indicates that S169 has a greater impact on the binding than P259 and Y261 , but all contribute to binding . While the majority of the phenotypes are consistent with the each RsiV mutant there are some outliers . For example , the Y261A mutation has a more dramatic effect on σV activity than it appears to have on lysozyme binding or RsiV cleavage . The reason for this remains unclear . Overall however , site directed mutagenesis of residues S169 , P259 , and Y261 showed a correlation between decreased σV activity , loss of RsiV degradation , and an inability to bind lysozyme . Our model that RsiV binding to lysozyme is required for σV activation is supported by multiple pieces of evidence . First , the solved co-structure of RsiV and lysozyme , revealed an intimate interaction between these proteins coordinated by contacts through multiple amino acid residues ( Fig 1 ) . Second , RsiV and lysozyme bind with high affinity ( Table 2 and [12] ) . Third , we previously found site-1 cleavage of RsiV by signal peptidase occurred only in the presence of lysozyme [12] . Fourth , here we demonstrate that mutants of RsiV deficient in lysozyme binding also fail to activate σV . Our data indicate that lysozyme binds RsiV some distance from the site-1 cleavage site ( Fig 7 ) . Thus we propose the following model for how binding of RsiV to lysozyme controls σV activation . In the absence of lysozyme , the cell normally makes a low basal level of σV and RsiV . Binding of lysozyme to RsiV results in a conformational change in RsiV which allows signal peptidase to recognize the previously masked cleavage site . Once site-1 cleavage of RsiV occurs the truncated RsiV can be cleaved by the site-2 protease RasP , leading to activation of σV . Once activated , σV is able increase transcription of the sigV-rsiV-oatA operon . This leads to increased production of RsiV which continues to bind free lysozyme and increases the level of free σV . This auto-regulatory loop allows the bacteria to rapidly respond and to produce more RsiV until all the lysozyme in the environment is sequestered by RsiV . At which point unbound RsiV builds up and inhibits further σV activation . This is analogous to the recently proposed model for how the ABC transporter BceAB and the two component signal transduction system BceSR control bacitracin resistance in B . subtilis [42] . In this system BceAB is required for resistance to bacitracin and acts as the sensor for bacitracin presence by controlling BceS activity [42] . Similarly , a “Just-in-time” model of regulation was proposed to explain how the immunity protein SdpI was involved in controlling expression of resistance to the cannibalism peptide [43] . A key similarity in each of these systems is the levels of the sensor are controlled by the regulator as well . The co-crystal structure revealed that RsiV binds lysozyme by interacting with the enzymatic pocket required for muramidase activity . Several lysozyme inhibitors have been identified in Gram-negative bacteria , however , few if any dedicated lysozyme inhibitors have been identified in Gram-positive bacteria [44] . A few virulent strains of group A streptococci produce SIC ( Streptococcal inhibitor of complement ) , which inhibits antimicrobial peptides and proteins of the innate immune response [45] . However , SIC only binds lysozyme relatively weakly , Kd = 85 μM , while binding to several other antimicrobial factors with higher affinity [45 , 46] . In contrast , the interaction between RsiV and lysozyme has a high affinity in the same range as Gram-negative inhibitors ( nm range ) [47] . Thus RsiV may represent the first specific inhibitor of lysozyme activity in Gram-positive bacteria . Lysozyme inhibitors from Gram-negative bacteria have been easier to isolate because they are soluble and located in the periplasm . It may be that other Gram-positive bacteria encode lysozyme inhibitors however association with the membrane has made them more difficult to identify . Known lysozyme inhibitors include: Ivy ( inhibitor of vertebrate lysozyme ) identified from E . coli [48] , MliC ( membrane lysozyme inhibitor ) from E . coli , Pseudomonas aeruginosa , and Salmonella enteritidis [47] and PliC ( periplasmic lysozyme inhibitor ) from Proteus mirabilis [49] . Ivy and MliC , which inhibit C-type lysozymes have been co-crystalized with lysozyme , revealing specific loops that interact with the catalytic domains [44] . Ivy inserts a 5 residue loop into the active site of lysozyme , which allows residue H60 to makes contact with E35 and D52 of lysozyme [50] MliC inserts two conserved loops into the active site of lysozyme . This allows S89 to interact with lysozyme D52 , and K103 to interact with lysozyme E35 and D52 [51] . RsiV appears to utilize a similar mechanism of inhibition by inserting two loops into the active site of lysozyme , which allows S169 to interact with D52 , and Y261 to interact with E35 ( Table 2 ) . This suggests that proteins with very different structures use similar mechanisms to occlude the active site of lysozyme . We have shown the anti-σ factor RsiV is a multifunctional protein . RsiV inhibits σV activity in the absence of stress , senses the inducing signal lysozyme , and inhibits lysozyme function by binding in the active site . The main role of most anti-σ factors is to inhibit the activity of the ECF σ factor . However some anti-σ factors are involved in sensing the inducing signal [3] . For example the anti-σ factors in several organisms respond to oxidative stress which is thought to occur via modification of cysteine residues in the anti-σ factor [3 , 52–54] . In Clostridium thermocellum several anti-σ factors contain carbohydrate binding domains [55–58] . However it is not known how binding of the carbohydrates to the anti-σ factor leads to activation of the σ factor . Another example of polysaccharide sensing by anti-σ factors is hypothesized to occur in Bacteroides thetaiotaomicron . In this system it is thought the ECF σ factors are responsible for responding to polysaccharides in the intestinal tract [4 , 59 , 60] . Most bacteria respond to lysozyme stress by either modifying their peptidoglycan or making a lysozyme inhibitor [44 , 61] . Interestingly our data indicate that activation of σV induces both activities . RNA polymerase containing σV transcribes the sigV-rsiV-oatA operon . The increased production of RsiV allows the cell to bind free lysozyme and inhibit its activity . In addition , increased production of OatA , an O-acetyltransferase that adds an acetyl group to MurNac of peptidoglycan , increases resistance to lysozyme [5 , 15] . Similarly σV homologs in C . difficile and E . faecalis are also required for transcription of a gene encoding a peptidoglycan modifying enzyme that removes an acetyl group from GlcNac [8 , 62] . The deacetylation of GlcNac also increases resistance to lysozyme [8 , 10 , 63] . Thus it appears that the σV-RsiV signal transduction system is able to increase lysozyme resistance using two distinct mechanisms . RsiV responds by inhibiting lysozyme activity via direct binding to the lysozyme active site . At the same time this binding event triggers the proteolytic cascade that results in increased levels of free σV . Activation of σV leads to production of RsiV which continues to bind free lysozyme and production of a peptidoglycan modification enzyme which increases lysozyme resistance . Thus σV can induce immediate lysozyme resistance by inhibiting lysozyme activity through RsiV and a longer term stress response by modifying peptidoglycan . This could provide both transient protection and prime the cells for continued growth in an environment where future lysozyme stress on their cell wall might be likely . It is not known when or if vegetative B . subtilis encounters lysozyme outside of the laboratory . B . subtilis is widely considered a soil organism and is often found associated with plant roots [64–67] . Interestingly , as a generally regarded as safe ( GRAS ) organism B . subtilis is also used in a variety of industrial and agricultural processes which may contribute to diverse environmental exposure . Recent studies have found B . subtilis can be isolated from the intestinal tracts of a variety of organisms which produce C-type lysozyme including Drosophila melagnoster , chickens , mice and humans [61 , 68–71] . It is unclear how association with the intestinal tract of a diverse number of organisms developed , likely consumption of soil , but perhaps B . subtilis has additional environmental niches beyond life in the soil . It is tempting to hypothesize that lysozyme resistance could be an important trait required for colonization of or survival though the intestinal tract of higher organisms . It is also not known what types of lysozymes B . subtilis encounters in the environment . Another possibility is that σV is required for resistance to multiple types of lysozyme . Previous work demonstrated that the muramidase mutanolysin was unable to activate σV [6] and RsiV did not bind mutanolysin [12] . The structure of the bacterial produced mutanolysin is quite distinct from C-type lysozymes like hen egg white lysozyme [72 , 73] . In contrast G-type and I-type lysozymes produced by eukaryotes are structurally similar to C-type lysozyme and likely the result of divergent evolution [74 , 75] . However , at this time it is not known if RsiV can bind to G-type or I-type lysozymes . One interesting point to note however is many inhibitors of one class of lysozyme fail to inhibit other classes of lysozyme suggesting that they are distinct enough to not be recognized by the same inhibitor [44] .
All plasmid constructs are listed in Table 6 were confirmed by DNA sequencing ( Iowa State DNA sequencing facility ) . All B . subtilis strains are isogenic derivatives of PY79 , a prototrophic derivative of B . subtilis strain 168 [76] and are listed in Table 7 . B . subtilis competent cells were prepared by the one-step method previously described [77] . Site-directed mutants of rsiV were constructed using the QuickChange site-directed mutagenesis kit ( Agilent Technologies ) and Isothermal Assembly ( Gibson et al . , 2009 ) . The sequences of the oligonucleotide primers used are listed in S2 Table . The IPTG-inducible hyper-spank ( Phs ) promoter was placed upstream of rsiV by assembling two rsiV PCR products into SphI digested pDR111 [78] . The two rsiV PCR products for the rsiVS169W mutant were PCR amplified with CDEP1859/CDEP3092 , and CDEP3093/CDEP1860 . Each mutation was generated the same way using the appropriate primer pair with CDEP1850 and CDEP1860: S169W ( CDEP3092/CDEP3093 ) , P259A ( CDEP3108/3109 ) , Y261A ( CDEP3110/CDEP3111 ) , P259A and Y261A ( CDEP3098/CDEP3099 ) . S169W P259A Y261A , S169W P259A and S169W , Y261A were made using rsiVS169W ( pJH375 ) as template and primers for P259A Y261A , P259A , or Y261A . The resulting plasmids: S169W ( pJH375 ) P259A ( pJH436 ) , Y261A ( pJH397 ) , P259A Y261A ( pJH398 ) , S169W P259A ( pJH405 ) , S169W Y261A ( pJH439 ) , S169W P259A Y261A ( pJH386 ) were transformed into CDE1563 sigVrsiV::kan , resulting in Phs-rsiVS169W sigVrsiV::kan ( JLH1271 ) , Phs-rsiVP259A sigVrsiV::kan ( JLH1481 ) , Phs-rsiVY261A sigVrsiV::kan ( JLH1342 ) , Phs-rsiVP259A Y261A sigVrsiV::kan ( JLH1343 ) , Phs-rsiVS169W P259A sigVrsiV::kan ( JLH1326 ) , Phs-rsiVS169W Y261A sigVrsiV::kan ( JLH1504 ) , and Phs-rsiVS169W P259A Y261A sigVrsiV::kan ( JLH1312 ) . The site directed mutants were introduced onto the chromosome of PY79 by homologous recombination using the temperature sensitive plasmid pMAD ( Arnaud et al . , 2004 ) . PCR products for rsiVS169W were amplified with rsiV + 1 kb upstream using CDEP1892/CDEP3092 and rsiV + 1 kb downstream using CDEP1893/CDEP3093 . The resulting PCR products were moved into SmaI digested pMAD using isothermal assembly . Each mutation was generated using the appropriate primer pair with CDEP1892/CDEP1893 or CDEP1892/CDEP3136 ( adds 1 . 5 kb downstream rsiV to increase recombination efficiency ) : S169W ( CDEP3092/CDEP3093 ) , P259A ( CDEP3108/3109 ) , Y261A ( CDEP3110/CDEP3111 ) , P259A and Y261A ( CDEP3098/CDEP3099 ) . The triple and double mutants ( S169W P259A Y261A , S169W P259A , and S169W , Y261A ) were made using rsiVS169W ( JLH1413 ) as template with primers for P259A Y261A , P259A , or Y261A respectively . Each plasmid ( S169W ( pJH381 ) P259A ( pJH403 ) , Y261A ( pJH438 ) , P259A Y261A ( pJH404 ) , S169W P259A ( pJH405 ) , S169W Y261A ( pJH437 ) , and S169W P259A Y261A ( pJH406 ) ) was transformed into CDE1546 PsigV-lacZ ( cat ) resulting in rsiVS169W PsigV-lacZ ( JLH1413 ) , rsiVP259A PsigV-lacZ ( JLH1380 ) , rsiVY261A PsigV-lacZ ( JLH1510 ) , rsiVP259A Y261A PsigV-lacZ ( JLH1372 ) , and rsiVS169W Y261A PsigV-lacZ ( JLH1507 ) , and rsiVS169W P259A Y261A PsigV-lacZ ( JLH1412 ) . To generate constructs for purification the extracellular domain of RsiV rsiV59-285 was PCR amplified using primers CDEP1139 and CDEP952 and the already constructed plasmids as template: S169W ( pJH375 ) P259A ( pJH436 ) , Y261A ( pJH397 ) , P259A Y261A ( pJH398 ) , S169W P259A Y261A ( pJH386 ) . This product was PCR amplified again with CDEP1140 and CDEP952 to add 2x-flag and cloned into pEntrD-TOPO , resulting in plasmids: S169W ( pJH421 ) P259A ( pJH426 ) , Y261A ( pJH424 ) , P259A Y261A ( pJH425 ) , S169W P259A ( pJH446 ) , S169W Y261A ( pJH448 ) , and S169W P259A Y261A ( pJH423 ) . To construct recombinant 6x-His-2x-Flag-RsiV ( 59–285 ) + site directed mutant , 2x-flag-rsiV ( 59–285 ) was moved from each pEntrD-TOPO vector into the T7-inducible 6x-His destination vector pDEST17 ( Invitrogen ) using LR Clonase II . The following plasmids were moved into BL21λDE3 cells for expression: S169W ( pJH427 ) , P259A ( pJH450 ) , Y261A ( pJH430 ) , P259A Y261A ( pJH431 ) , S169W P259A ( pJH447 ) , and S169W P259A Y261A ( pJH429 ) . Antibiotics were used at the following concentrations: chloramphenicol , 5 μg/ml; erythromycin plus lincomycin , 1 μg/ml and 25 μg/ml; kanamycin , 5 μg/ml; spectinomycin , 100 μg/ml; tetracycline , 10 μg/ml; ampicillin 100 μg/ml . The β-galactosidase chromogenic indicator 5-bromo-4-chloro-3-indolyl β-D-galactopyranoside ( X-Gal ) was used at a concentration 100 μg/ml . Isopropyl β-D-1-thiogalactopyranoside ( IPTG ) was used at a final concentration of 1 mM unless otherwise noted . Strains were grown for 16 hours in LB at 37°C . The cells were subcultured 1:100 in LB+1 mM IPTG at 37°C and grown to an OD600 of 0 . 8–1 . The cells were pelleted by centrifugation and resuspended in 100 μl of 2X Laemmli sample buffer and lysed by repeated sonication . Samples were electrophoresed on a 15% SDS polyacrylamide gel ( BioRad ) . The proteins were then blotted onto nitrocellulose . The nitrocellulose was blocked with 5% milk for 30 minutes . The proteins were detected by incubating with a 1:10 , 000 dilution of anti-RsiV59–285 antibodies ( Hastie et al . , 2013 ) or 1:15 , 000 dilution of anti-σA antibodies followed by 3 washes and incubation in a 1:10 , 000 dilution of goat anti-rabbit IgG ( H+L ) IRDye 800CW ( Li-Cor ) and imaged on an Odyssey CLx ( Li-Cor ) . Quantification of band intensities was performed using Image Studio software ( Li-Cor ) . Cultures were grown overnight in LB broth at 30°C and 20 μl were spotted onto LB agar + 0 , 2 , 5 , 10 or 20 μg/ml lysozyme . Plates were incubated at 37°C for 6 hours . Cells were harvested and resuspended in 500 μl of Z buffer ( 60 mM Na2HPO4 , 40 mM NaH2PO4 , 10 mM KCl , 1 mM MgSO4 , 50 mM β-mercaptoethanol pH 7 . 0 ) . Cells were transferred to a 96 well plate and optical density ( OD600 ) determined . Cells were permeabilized by mixing with chloroform and 2% sarkosyl [5 , 79] . Permeabilized cells ( 100 μl ) were mixed with 10 mg/ml ortho-Nitrophenyl-β-galactoside ( ONPG , RPI , 50 μl ) and OD405 was measured over time . β-galactosidase activity units ( μmol of ONP formed min−1 ) X 103/ ( OD600 X ml of cell suspension ) were calculated as previously described [80] . Experiments were performed in triplicate with the mean and standard deviation shown . For each sample , a single colony of the appropriate B . subtilis strain was inoculated in LB medium and grown overnight . The overnight cultures were subcultured into LB or LB + 1 mM IPTG and grown to -OD600 of 0 . 8 at which point lysozyme when necessary was added to a final concentration of 1 . 25 μg/ml and grown for 1 hour . RNA was extracted using the RNeasy RNA Isolation kit ( Qiagen ) . Contaminating DNA was removed using the Turbo DNA-free kit protocol ( Ambion ) . Samples were tested for DNA contamination by PCR amplification ( Thermo Taq polymerase , NEB ) using primers CDEP1017 and CDEP1018 ( S3 Table ) . To generate cDNA from RNA samples , we used Superscript II ( Invitrogen ) according to manufacturer’s protocols . The resulting reverse transcription reactions were diluted 1:5 in DEPC-treated water . For each quantitative RT PCR reaction , 5 μl of sample was added to 10 μl of power Sybr green master mix ( Applied Biosystems ) and 5 μl gene-specific primers ( 2 x 2 . 5 μM ) . The list of primers used to quantitate cDNA levels of different samples is provided in S3 Table . Experiments were performed on three biologically independent replicates . Data were normalized to RNA levels of the housekeeping gene rpoB . Samples were prepared the same as previously described [12] . Briefly , overnight cultures of E . coli BL21λDE3 containing WT RsiV pKBW201 ( pDEST17-6xhis-2xflag-rsiV59–285 ) or mutant RsiV S169W ( pJH427 ) , P259A ( pJH450 ) , Y261A ( pJH430 ) , P259A Y261A ( pJH431 ) , S169W P259A ( pJH447 ) , and S169W P259A Y261A ( pJH429 ) were grown at 30°C in LB + ampicillin . The cell cultures were diluted 1:100 into 500 ml of LB + ampicillin in 2 L baffled flasks and incubated at 30°C to an OD600 of 0 . 5–0 . 6 . IPTG was added to a final concentration of 1 mM to induce protein production and the cultures incubated for an additional 4 hours . Cells were chilled on ice and collected by centrifugation at 5000xg . Cell pellets were stored at −80°C until time for purification . Cell pellets were thawed on ice and resuspended in 5 ml lysis buffer ( 50 mM NaH2PO4 , 250 mM NaCl , 10 mM imidazole , pH 8 . 0 ) per 500 ml of initial culture volume . Cells were lysed by passaging through a Microfluidics LV1 high shear microfluidizer ( Newton , MA ) twice . Lysate was centrifuged at 15 , 000xg , for 30 minutes at 4°C to pellet cellular debris . Cleared lysate was applied to a nickel affinity column to bind 6xHis-tagged protein ( Qiagen ) . The column was washed with 10 column volumes of wash buffer ( 50 mM NaH2PO4 , 250 mM NaCl , 20 mM imidazole , pH 8 . 0 ) . Protein was eluted with elution buffer ( 50 mM NaH2PO4 , 250 mM NaCl , 250 mM imidazole , pH 8 . 0 ) and collected in 0 . 5 ml fractions . Samples from each fraction were analyzed by SDS-PAGE and elution fractions containing the desired protein were combined . Combined fractions were then dialyzed into lysis buffer to remove the excess imidazole . RsiV and HEW-lysozyme mixed in a 1:3 molar ratio were allowed to incubate for 60 minutes , at 4°C , and run through a Superdex-75 size exclusion column ( GE ) . Fractions representing the complex confirmed by running a SDS-PAGE gel stained with Coomassie Blue were pooled and concentrated using a Amicon 30kDa filter to 8 . 9 mg/ml . Crystallization drops with commercially available screens using a TTP LabTech Mosquito robot were set-up via the hanging-drop vapor diffusion method at 18°C with the drop containing 0 . 4 μl each of protein and crystallization solutions . Crystals were obtained in a variety of PEG containing conditions within 14 days . In order to perform experimental phasing selenomethionine labeled RsiV protein was prepared . For production for selenomethionine labeled protein , cells were grown at 35°C overnight in M9 minimal media supplemented with 100 μg/ml ampicillin . The following morning cells were subcultured 1:50 into 1L flasks of M9 minimal media with ampicillin and incubated at 35°C . When the cultures reached O . D . 600 of 0 . 5 , 100 mg/L of lysine , phenylalanine , threonine , and 50 mg/L of isoleucine , leucine , and valine were added to inhibit methionine synthesis as described [81] . Selenomethionine was added to the media at 60 mg/L [81] and the temperature shifted to 28°C . 15 minutes after addition of amino acid cocktail , IPTG was added to a final concentration of 1mM to induce protein expression . Purification of the selenomethionine labeled protein proceeded as described for unlabeled RsiV . Selenomethionine labeled RsiV was purified as described above complexed with lysozyme and crystallization plates set up similar to the wild type protein . Crystals obtained in 0 . 2 M sodium nitrate , 20% w/v PEG 3350 were harvested for diffraction data collection . Crystals were flash-cooled in liquid nitrogen and shipped to the 4 . 2 . 2 synchotron beamline at the Advanced Light Source ( Berkeley , CA , USA ) for remote data collection . The data were processed using XDS [82] . Pointless [83 , 84] and Aimless [84–89] from the CCP4 [90] software suite were used for conversion of intensities to structure factors and scaling . Structure solution , initial model building and refinements were performed using AutoSol [91] , AutoBuild [92] and Phenix . refine [93] from the Phenix suite [93] . Model building was performed in Coot [94] and all structural figures were generated using PyMOL [95] . All crystallography software used were configured and deployed using SBGrid [96] . The surface interface of RsiV and lysozyme was determined using Protein interfaces , surfaces and assemblies' service PISA at the European Bioinformatics Institute ( http://www . ebi . ac . uk/pdbe/prot_int/pistart . html ) [29] . ITC analysis was performed as previously described [12] . Briefly , 6xHis-2xFlag-RsiV59–285 and mutants were purified as described above and buffer matched with HEW lysozyme ( ≥98% pure , Sigma Aldrich ) by dialysis into 50 mM Na2HPO4 , 200 mM NaCl , and pH 7 . 0 for 24 h at 4°C . Final protein concentrations as determined by absorbance at OD280 were adjusted to 6xHis-2xFlag-RsiV59–285 ( 0 . 01 mM ) and HEW lysozyme ( 0 . 1 mM ) with filtered dialysate . The protein samples were degassed and ITC measurements recorded using a MicroCal VP-ITC System ( GE Healthcare ) with HEW lysozyme as the injected sample and 6XHis-2XFlag-RsiV59–285 as the cell sample . 21 injections of HEW lysozyme were used , with 180 seconds spacing between events . The chamber was kept under constant stirring at 350 rpm and all experiments were performed at 25°C . The binding reaction reached saturation during the experiment and control experiments where HEW lysozyme was injected into buffer showed that the heats of dilution were constant across all injections . The constant heat of dilution , as determined by the average of the last 3–5 injections , was subtracted and the data are analyzed using the single site binding model provided in the ITC analysis package . The values for affinity were averaged from three separate runs from two different protein preps . Protein from 6xHis-2xFlag-RsiV59–285 and 6xHis-2xFlag-RsiV59-285 , S169W , P259A , Y261A was purified as described above . Samples were further purified on a BioRad DuoFlow FPLC with a Superdex 75 ( GE ) size exclusion column . Fractions were pooled and the concentration was determined by OD280 with an extinction coefficient calculated for each mutant as described [97] . Samples were diluted to 20 μM and analyzed using a 1mm cuvette in a Jasco J-815 CD spectropolarimeter . Increasing concentrations of purified RsiV ( prepared as described above ) were mixed with lysozyme ( 20 μg/ml ) for 15 minutes . The purified RsiV + lysozyme ( 100 μl ) mixture was combined with M . lysodekticus ( 100 μl , OD450 = 0 . 9 , Sigma ) The OD450 was measured every minute for 30 minutes to monitor M . lysodekticus degradation . Lysozyme activity was calculated based on the equation as described by Sigma [41] . Briefly , units/ml enzyme = ( ΔOD450 ) * ( dilution factor ) / ( 0 . 001*0 . 05 ) , Specific Activity = ( units/ml enzyme ) / ( mg solid/ ml enzyme ) . The minimum inhibitor concentration ( MIC ) was determined by diluting overnight cultures 1:100 in 1:50 LB + 1 mM IPTG . The cells were inoculated into 250 μl of LB + 1 mM IPTG containing serial dilutions of hen egg white lysozyme ranging from 20 μg/ml to 1 . 875 μg/ml in a round bottom 96 well plate . The cells were grown for 16 hours at 37°C . Growth was defined as OD600 of greater than 0 . 05 . All assays were performed using all strains listed in triplicate on the same day . | The exposed cell wall of Gram-positive bacteria renders them particularly susceptible to the innate immune defense enzyme lysozyme . Several Gram-positive bacteria activate lysozyme resistance via a signal transduction system , σV , which is induced by lysozyme . Here we report the co-structure of lysozyme with its bacterial receptor the anti-σ factor RsiV . In the absence of lysozyme , RsiV inhibits activity of σV . In the presence of lysozyme , RsiV is destroyed via proteolytic cascade . We demonstrate that binding of lysozyme to RsiV triggers the proteolytic destruction of the anti-σ factor RsiV and thus activation of σV . In addition , we demonstrate that RsiV also acts as an inhibitor of lysozyme activity . Thus , the anti-σ factor RsiV allows for the cell to sense lysozyme and inhibit its activity as well as inducing additional lysozyme resistance mechanisms . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"sequencing",
"techniques",
"medicine",
"and",
"health",
"sciences",
"crystal",
"structure",
"pathology",
"and",
"laboratory",
"medicine",
"enzymes",
"pathogens",
"bacillus",
"microbiology",
"enzymology",
"vertebrates",
"condensed",
"matter",
"physics",
"animals",
"mammal... | 2016 | The Anti-sigma Factor RsiV Is a Bacterial Receptor for Lysozyme: Co-crystal Structure Determination and Demonstration That Binding of Lysozyme to RsiV Is Required for σV Activation |
Haploid germline nuclei of many filamentous fungi have the capacity to detect homologous nucleotide sequences present on the same or different chromosomes . Once recognized , such sequences can undergo cytosine methylation or cytosine-to-thymine mutation specifically over the extent of shared homology . In Neurospora crassa this process is known as Repeat-Induced Point mutation ( RIP ) . Previously , we showed that RIP did not require MEI-3 , the only RecA homolog in Neurospora , and that it could detect homologous trinucleotides interspersed with a matching periodicity of 11 or 12 base-pairs along participating chromosomal segments . This pattern was consistent with a mechanism of homology recognition that involved direct interactions between co-aligned double-stranded ( ds ) DNA molecules , where sequence-specific dsDNA/dsDNA contacts could be established using no more than one triplet per turn . In the present study we have further explored the DNA sequence requirements for RIP . In our previous work , interspersed homologies were always examined in the context of a relatively long adjoining region of perfect homology . Using a new repeat system lacking this strong interaction , we now show that interspersed homologies with overall sequence identity of only 36% can be efficiently detected by RIP in the absence of any perfect homology . Furthermore , in this new system , where the total amount of homology is near the critical threshold required for RIP , the nucleotide composition of participating DNA molecules is identified as an important factor . Our results specifically pinpoint the triplet 5'-GAC-3' as a particularly efficient unit of homology recognition . Finally , we present experimental evidence that the process of homology sensing can be uncoupled from the downstream mutation . Taken together , our results advance the notion that sequence information can be compared directly between double-stranded DNA molecules during RIP and , potentially , in other processes where homologous pairing of intact DNA molecules is observed .
Many biological systems exhibit specific co-localization ( pairing ) of homologous DNA molecules independent of recombination-based mechanisms . A paradigmatic example is provided by the persistent association of homologous chromosomes in somatic nuclei of the Diptera insects [1] . Somatic pairing of homologous loci has been documented in yeast ( references in [2] ) and in mammals ( references in [3 , 4] ) . In early meiosis , significant pairing of homologous chromosomes occurs prior to or in the absence of recombination in mice , flies , worms , fission yeast and filamentous fungi ( references in [2] ) . The molecular mechanisms by which homologous sequences are recognized in all of these cases remain largely unknown [5] . Roles of direct interactions between DNA molecules or the indirect readout of the base-pair sequence by RNA or proteins have been proposed [6 , 7] . In addition to the physical co-localization , recombination-independent interactions between homologous DNA molecules can trigger their chemical modification . In filamentous fungi , two closely-related processes can detect duplicated DNA sequences irrespectively of their origin and location in the genome [8] . Once detected , DNA duplications can undergo cytosine methylation ( the phenomenon of Methylation Induced Premeiotically , MIP , discovered in Ascobolus immersus [9] ) or cytosine-to-thymine mutation ( the phenomenon of Repeat-Induced Point mutation , RIP , discovered in Neurospora crassa [10] ) . Both RIP and MIP are restricted to the sexual phase of the fungal life-cycle and occur in haploid germline nuclei that prepare to enter meiosis . Both processes are mediated by specialized cytosine methyltransferases of the RID/Masc1 family [11 , 12] . Although RIP coincides with a period of increased intrachromosomal recombination [13] , it does not require MEI-3 , the only RecA homolog in Neurospora [14] . No DNA sequence , with the exception of ribosomal DNA in the nucleolar organizer region , can escape RIP [15] . The general and efficient nature of RIP makes it an especially attractive experimental system for elucidating the general mechanism of recombination-independent recognition of DNA homology . Previously , we developed a sensitive genetic assay for RIP based on the quantitative analysis of individual mutations induced by a pair of closely-positioned DNA repeats [14] . Using this experimental system , we demonstrated that as few as 155 base-pairs of perfect homology could trigger RIP , and that 220 base-pairs of perfect homology could promote substantial RIP . We further found that the number of RIP mutations accurately reflected the length of homology for repeats ranging between 220 and 520 base-pairs . We then explored the ability of RIP to recognize different patterns of imperfect homology by extending the 220-bp region of perfect homology with additional 200 base-pairs of synthetic DNA that provided partial homology in the adjacent region . Here we discovered that RIP could detect weak partial homologies organized as arrays of base-pair triplets interspersed with a matching periodicity of 11 or 12 base-pairs along participating chromosomal segments . These and other observations led us to propose that sequence homology could be sensed by direct interactions between intact , slightly underwound double-stranded DNA molecules [14] . Our previous studies examined recognition of interspersed homologies in the repeat system that always included the 220-bp region of adjacent perfect homology , which was expected to provide a strong , persistent point of interaction . Those studies did not give any indication that RIP might be sensitive to the actual base-pair sequence of homologous units . In the present study , we have focused on the analysis of interspersed homologies spanning 500 base-pairs . We demonstrate that interspersed homologies with overall sequence identity of only 36% can be efficiently detected by RIP in the absence of perfect homology; yet short regions of adjacent perfect homology can have a strong activating effect . We have dissected the sequence requirements for RIP within one such short region of adjacent homology , under conditions where the overall amount of homology in the repeat system remains near the minimum threshold required for RIP . Analysis under these particularly stringent circumstances reveals that the base-pair composition of participating sequences can play a critical role in RIP and identifies one candidate trinucleotide , 5'-GAC-3' , as an especially efficient unit of homology . We also present experimental evidence suggesting that RIP integrates sequence information over hundreds of base-pairs , and that the actual process of homology sensing can be distinguished from ensuing mutation . These and other results further illuminate the properties of DNA homology recognition during Neurospora RIP , paving the way for more mechanistic studies in the future .
We previously developed a sensitive quantitative assay for RIP based on the analysis of individual mutations induced by a pair of closely-positioned DNA repeats [14] . The sequence of the “left” repeat ( so-called "Reference" ) was always held constant , while the sequence of the “right” repeat ( so-called "Test" ) was varied as desired ( Fig 1A ) . To examine the capacity of different interspersed homologies to trigger RIP , we designed test sequences that contained short units of homology of length X separated by units of non-homology of length Y . The periodicity with which homologous units occurred along a particular pair of reference/test sequences is defined by the sum of these two distances ( X+Y ) . Our previous research showed that only certain periodicities ( X+Y = 11 and X+Y = 12 ) could promote RIP in conjunction with an adjacent region of perfect homology [14] . Because an array of homologous units can be positioned in a particular relationship with respect to the reference sequence , an additional parameter Z is now introduced to define the sequence position of the first homologous unit ( Fig 1B ) . Patterns of interspersed homology investigated in this study are thus represented as XH-YN_Z ( Fig 1B ) . Because a particular reference region is selected arbitrarily from the Neurospora genome , the actual nucleotide sequences of homologous units are expected to be different for every combination of X , Y , and Z . However , by keeping X and Y constant and only changing the sequence position parameter Z , the sequences of homologous units can be altered without disrupting basic homology-block structure ( Fig 1C ) . The present study shows that such sequence differences can be important for the efficiency of homology recognition for RIP . In our previous work , we analyzed mutation of the 500-bp interspersed homology 4H-7N_1 created by replacing the cyclosporin-resistant-1 gene ( csr-1 ) with synthetic DNA that was designed to be partially homologous to an arbitrarily selected reference region in a nearby gene ( Fig 2A ) . In that study , we found that 4H-7N_1 alone exhibited barely detectable RIP but that , when a 337-bp region of perfect homology was present at an adjacent position , strong mutation of 4H-7N_1 was observed [14] . The present study was initiated to further define the role of perfect homology in promoting mutation of the interspersed homology 4H-7N_1 . As in our previous analysis , the mean number of mutations ( per spore ) in the entire sequenced region was used as a quantitative measure of RIP . Empirical distributions of mutation counts were compared by the Kolmogorov-Smirnov test . Systematic analysis shows that combining 4H-7N_1 with at least 50 base-pairs of perfect homology significantly increases RIP , and that 100 base-pairs of perfect homology promote very efficient RIP ( Fig 2B–2D ) . A similar effect is observed regardless of whether the fully-homologous region is added to the "left" or the "right" end of the partially homologous region , implying that two different 100-bp sequences , at two different positions in the repeat construct , are equivalently effective ( Fig 2D: compare v and vi ) . We further find that the insertion of 11 base-pairs of random DNA sequence at the junction between 4H-7N_1 and the fully-homologous region leads in a partial reduction in RIP , but that substantial mutation still occurs despite the apparent physical impediment to overall alignment ( Fig 2D: compare v and vii ) . A similar effect is conferred by replacing 22 base-pairs of perfect homology by 22 base-pairs of non-homology in the middle of the 100-bp region , thereby interrupting the 100-bp perfect homology into two segments of 38 and 40 base pairs ( Fig 2D: compare vii and viii ) . We similarly examined the effect of adjacent perfect homology on mutation of another interspersed homology , 5H-6N_1 ( Fig 3 ) . This new homology pattern relates to the same reference sequence , and has the same periodicity ( X+Y ) and sequence position ( Z ) as 4H-7N_1 , but contains homologous units of five base-pairs ( X = 5 ) instead of four . We find that 5H-6N_1 alone can trigger substantial RIP , demonstrating a significant improvement over 4H-7N_1 ( Fig 3 ) . This observation provides a new conclusion: given an appropriate pattern of interspersed homology , perfect homology is dispensable for RIP . We further find that , despite the substantial intrinsic activity of 5H-6N_1 , the addition of only 15 base-pairs of perfect homology ( representing 3% of the total homology length ) increases RIP nearly 2 . 5-fold ( from 3 . 6±0 . 9 to 9 . 0±1 . 5 mutations per spore , Fig 3A ) . This result has two implications . First , the presence of short perfect homology can still be highly advantageous even against the substantial background activity of 5H-6N_1 . Second , such a strong response to 15 base-pairs of perfect homology contrasts with a negligible effect of combining the same 15-bp sequence with the interspersed homology 4H-7N_1 ( Fig 3A and 3B ) . This difference might be explained by a synergistic interaction of the fully- and partially-homologous regions , such that more activity in the partially-homologous region leads to a greater effect of the fully-homologous region and vice versa . However , even from the more robust starting point of 5H-6N_1 , and similarly to 4H-7N_1 , a steep increase in mutation is observed when the region of perfect homology is extended from 75 base-pairs to 100 base-pairs , pointing to a critical threshold at about this length ( Fig 3A and 3B ) . We also note that for all examined combinations of perfect and interspersed homologies , mutations tend to be distributed symmetrically near or around the center of each compound repeat , well inside the partially-homologous region . Taken together , the above findings further imply that the mechanism of homology recognition for RIP integrates sequence information from interspersed and perfect homologies collectively over several hundreds of base pairs . The above results show that recognition of long interspersed homologies can be promoted by relatively short segments of adjacent perfect homology . To obtain more insight into the interplay between these two homology types , we investigated the effects of substituting interspersed homology for perfect homology in a 100-bp region adjacent to 4H-7N_1 ( Fig 2D , construct v ) . We chose to use 4H-7N_1 because it displayed almost no activity in the absence and strong activity in the presence of the adjacent 100-bp region ( Fig 2C ) . We designed new patterns of interspersed homology in this 100-bp region by fixing the homologous unit length to 6 base-pairs ( X = 6 ) and varying the other two parameters: periodicity ( X+Y ) and sequence position ( Z ) ( Fig 4A ) . We have chosen X = 6 because our previous studies [14] suggested that homologous units of this particular length would provide a strong signal which , while not involving continuous homology , might have a significant effect on RIP in combination with 4H-7N_1 . Systematic analysis shows that different patterns of adjacent interspersed homology can elicit different effects . We first varied the periodicity parameter from 8 to 15 while keeping constant sequence position Z = 1 . Corresponding interspersed homologies are shown in Fig 4A ( from 6H-2N_1 to 6H-9N_1 ) . A periodicity of 10 base-pairs confers the strongest effect , reaching nearly 50% the level observed with perfect homology; periodicities of 8 and 9 base-pairs induce lower but significant levels of RIP; and periodicities of 11–15 base-pairs produce little or no detectable effect ( Fig 4B , left ) . These results were quite unexpected , because our previous analyses implied that periodicities of 11 or 12 should be favored [14] , but are explained by further investigation ( below ) . We next explored the effects of varying the sequence-position parameter Z . In the above analysis , the 100-bp interspersed homology 6H-4N with sequence position Z = 1 ( pattern 6H-4N_1 ) conferred the strongest positive effect on RIP . Thus , we have varied the sequence-position parameter of 6H-4N in this same context from Z = 1 to Z = 12 ( Fig 4A ) . We find that sequence positions of 1 and 2 correspond to equally strong levels of RIP , with a progressive decrease for sequence positions 7–10 , followed by an increase again for sequence positions 11 and 12 ( Fig 4B , right ) . These results are intriguing: since the sequence-position parameter Z varies the nucleotide composition of the homologous units without altering basic homology-block structure , the possibility is raised that DNA sequence per se can have an important effect on RIP . In the context of this possibility , particular combinations of base-pairs in the homologous units of 6H-4N_1/11 and 6H-4N_2/12 might be optimal in synergizing with 4H-7N_1 . However , alternatively , the differences in the ability to synergize , observed in this particular situation , could potentially arise from the variation in the distance of separation between homologous units at the junction between the 500-bp and 100-bp regions . To further explore the issues raised in the previous section , we compared mutation of two 500-bp homologies that differed with respect to the sequence-position parameter: 4H-7N_1 ( discussed above ) and the newly designed interspersed homology 4H-7N_7 ( Fig 5A , compare i and ii ) . We first examined these interspersed homologies alone , without any accompanying 100-bp region . Here we have observed a very dramatic difference with respect to their capacities to promote RIP ( Fig 5B and 5C , compare i and ii ) : while RIP is virtually undetectable in the case of 4H-7N_1 ( 0 . 08±0 . 08 mutations per spore ) , it is quite strong in the case of 4H-7N_7 ( 10±1 . 2 mutations per spore ) . This comparison provides a clean demonstration that , in a situation where the overall level of homology is weak , differences in DNA sequence at positions of homology can have a substantial effect . In addition , these results show that periodically-spaced homologous units of four base-pairs , corresponding to overall sequence identity of only 36% , are sufficient to drive homology recognition . We have then further examined the effects of the adjacent 100-bp regions 6H-4N_1 and 6H-4N_7 in combination with the 500-bp regions specifying 4H-7N_1 or 4H-7N_7 ( Fig 5A ) . We find that each 100-bp region confers its characteristic effect , which appears superimposed on the effects conferred by the 500-bp regions: 6H-4N_1 increases RIP in both contexts , whereas 6H-4N_7 has no detectable effect in either context ( Fig 5B and 5C ) . These results show that the relative sequence positions of homologous base-pairs within the 500-bp and 100-bp regions are not relevant—only the absolute sequence positions matter . These findings provide strong additional support for the notion that the sequence composition of homologous units is important for RIP . If DNA sequence plays a critical role in RIP , what is the basis for this effect ? Our previous work has suggested that the triplet of base-pairs represents the basic unit of homology recognition for RIP [14] . Therefore , we wondered whether , among other sequence determinants , “RIP-proficient” homologies might have a higher abundance of certain triplets as compared to “RIP-deficient” homologies . We have shown that the 500-bp homologies 4H-7N_1 and 4H-7N_7 , which differ in the positions of the homologous units and , therefore , have the homologous units comprising different base-pair combinations , exhibit a 100-fold difference in their ability to promote RIP . Thus , we have compared nucleotide sequences of 4H-7N_1 and 4H-7N_7 with respect to the natures of triplets included in their homologous units ( Fig 6A ) . We find that most triplets are present in comparable numbers , but with two exceptions: six GAC and eight TGA triplets are present in 4H-7N_7 ( which promotes substantial RIP ) , while neither of these triplets appears in 4H-7N_1 ( which promotes very little RIP ) . To investigate whether either or both of these differences were significant for RIP , we first explored the role of GAC triplets . We have designed two variants of 4H-7N_7 that lacked either two or all six GAC triplets ( Fig 6B , patterns “Δ2GAC” and “Δ6GAC” ) . The single-nucleotide substitutions that specifically eliminated GAC triplets were chosen to accommodate the fact that each 4-bp unit of homology in 4H-7N_7 contains two overlapping triplets ( Fig 1C ) , making it possible to alter the GAC triplet without affecting the other overlapping triplet . This is accomplished by mutating G to C if the GAC triplet occupies positions 1–3 of the 4-bp unit , or by mutating C to G if the GAC triplet occupies positions 2–4 . Our results show that removing the two GAC triplets has reduced RIP by about 25% , while removing all six GAC triplets nearly eliminated all RIP activity ( Fig 6C and 6D ) . We next analogously examined the possible role of TGA triplets ( Fig 6B , pattern “Δ7TGA” ) . In contrast to the results obtained for GAC triplets , deleting seven TGA triplets using the above approach had no significant effect on RIP ( Fig 6C and 6D ) . Taken together , these results strongly suggest that GAC triplets , but not TGA triplets , may play a privileged role in RIP . As an additional confirmation of this conclusion , we have examined the role of GAC triplets in another context . Our results have shown that the 100-bp region of perfect homology produces a much stronger increase in mutation compared to the 75-bp region ( Fig 3A ) . We have noticed that the 100-bp region , but not the 75-bp region , contains two GAC triplets ( Fig 6E ) . These GAC triplets are also present in the 100-bp interspersed homology 6H-4N_11 that promotes substantial RIP in combination with 4H-7N_1 ( Fig 4B , right ) . Using the same approach as described above , we mutated the two GAC triplets in 6H-4N_11 ( Fig 6E: “ΔGAC” ) . As a control , we introduced two C/G and G/C substitutions in unrelated triplets ( Fig 6E: “Mock” ) . Mutation analysis shows that the two single-nucleotide changes that deleted the GAC triplets have effectively eliminated the ability of 6H-4N_11 to activate RIP , while the two changes that removed non-GAC triplets have had no significant effect ( Fig 6F and 6G ) . These findings further support the idea that GAC triplets may indeed have a special role in RIP . Previous research had shown that repeat units were mutated specifically over the extent of shared homology , with only a few mutations occurring outside the duplicated regions [16] . Such high accuracy implies tight coupling between the processes of homology recognition and ensuing mutation . However , our previous work showed that single-copy regions located between pairs of closely-positioned repeats could be strongly mutated , suggesting that homology recognition and mutation can , in fact , be uncoupled regionally [14] . Here we explored the possibility of uncoupling at the base-pair level . Specifically , we have asked whether the local occurrence of mutations is affected by the positions of homologous units . For this purpose , we have compared mutation patterns produced by interspersed homologies 4H-7N_1 and 4H-7N_7 in combination with appropriate 100-bp regions: ( i ) 4H-7N_1 with 100 base-pairs of perfect homology and ( ii ) 4H-7N_7 with 100 base-pairs of the interspersed homology 6H-4N_1 ( Fig 7A ) . These repeat constructs share the same reference sequence , but , because of the difference in the sequence-position parameter , their 4-bp homologous units do not overlap ( Fig 7B ) . While the overall RIP profiles appear somewhat different between the two repeat constructs , with the apparent paucity of mutations in the first ~130 base-pairs of 4H-7N_1 ( Fig 7A , marked with “*” ) , similar levels of RIP can be observed within the 200-bp portion of the reference sequence ( Fig 7A , outlined ) . Focusing our analysis on this 200-bp region , we find that cytosines at identical positions are mutated similarly and irrespectively of their spatial relationship to the underlying homologous units ( Fig 7B and 7C ) . This result suggests that , at the level of individual base-pairs , homology recognition and mutation are separable events , either functionally and/or temporally .
We previously examined recognition of interspersed homologies using repeat constructs that included a 200-bp region of tested interspersed homology adjoining a 220-bp region of perfect homology [14] , see also ( S1 Fig ) . In this context , the effective patterns of interspersed homology collectively implied that the sequence information was sensed in units of three base-pairs spaced at intervals of 11 or 12 base-pairs , over the total length of several hundred base-pairs . We interpreted this result as evidence that two co-aligned double-stranded DNA molecules were compared by direct contacts . The current study extends these findings in several respects . Taken together , previous and current findings make it clear that a number of different sequence features contribute to DNA homology recognition for RIP . Our previous study [14] identified triplet homology units and 11/12 base-pair periodicities of those units as important features for RIP . The present study uncovers another important factor: the underlying DNA sequence in general and GAC triplets in particular . It is important to appreciate that different features emerged in the two studies because of the built-in differences in the repeat configurations being used . In the previous work , we examined variations in the basic pattern of interspersed homology within a 200-bp region that was positioned adjacent to a 220-bp region of perfect homology which , by itself , triggered substantial RIP . Thus , the effectiveness of different homology patterns was being evaluated in a situation where the basal level of homology ( provided by the 220-bp region ) was already above the critical threshold required for RIP . In this experimental system , a preference for homology in triplets of base-pairs spaced with the 11- or 12-bp periodicity was revealed . The present study , in contrast , examined situations in which the basal level of homology was close to or below the critical threshold for RIP . Starting with the intrinsically weak interspersed homology 4H-7N_1 ( Fig 2 ) , we found that 100 base-pairs of adjacent perfect homology promoted strong RIP , whereas 75 base-pairs of perfect homology were much less effective , implying that the 100 base-pairs were just barely effective . Starting from this suboptimal situation , examining requirements for interspersed homology in the 100-bp region revealed a role of the underlying DNA sequence . The actual patterns of interspersed homology that permitted the strongest effect did not obviously involve 11/12-bp periodicities . This is likely because the 100-bp region is predicted to contain only ~9 duplex/duplex contact points rather than ~18 as in the previous study , thus giving the DNA sequence a more prominent role . Overall , these observations highlight the fact that homology recognition for RIP can collectively integrate and evaluate diverse underlying features over substantial distances . Evidence that RIP involves direct dsDNA/dsDNA homology recognition is still indirect . In principle , homology recognition might involve dsRNA interacting with dsDNA or , even , some other type of sequence-specific interactions not involving two double-helical nucleic acids . We note , however , that RIP can recognize 4-bp units of homology embedded in a completely non-homologous sequence , and that four base-pairs are significantly below the threshold length of 6–7 nucleotides required for stable pairing of single-stranded nucleic acids in the context of Argonaute proteins [17] , arguing against the role of these proteins in homology sensing for RIP . The presented observations provide new grounds to justify consideration of existing models of sequence-specific pairing between intact DNA molecules . Two long-standing models invoke the intriguing principle of self-complementarity of Watson-Crick base-pairs , by which identical base-pairs can form planar quadruplexes via major-groove [18] or minor-groove [19] interactions . The possibility of such interactions was confirmed experimentally by NMR [20 , 21] . Further , in response to our previous work , the principle of base-pair self-complementarity was suggested to be capable of mediating pairing between long intact DNA molecules by the formation of short interspersed dsDNA/dsDNA quadruplexes [22] . This model also predicts that such quadruplex interactions may be sensitive to the underlying DNA sequence [22] . However , pairing of long double-stranded DNAs by this mechanism could involve two duplexes that are either plectonemically intertwined over the "pairing region" ( which would permit many contacts along the region ) or paranemically related ( which would limit direct contacts to one per ~55 base-pairs ) [22] . Neither of these conditions is impossible but also neither is necessarily attractive a priori . Another proposed model includes the intertwining of two negatively supercoiled double-stranded DNA molecules in a form of so-called “PX-DNA” , where homology recognition is mediated by standard Watson-Crick hydrogen bonds [23] . Recent data , however , suggest that the actual structure of the paired complex in this case remains elusive [24] . In a third type of model , non-Watson-Crick hydrogen bonds are proposed to mediate association of specialized sequences , such as G-quartets [25] or triplex structures involving long polypurine/polypyrimidine runs [26] . However , the RIP phenomenon in general , and the newly discovered patterns of sequence recognition in particular , are not compatible with such specialized mechanisms . Finally , interactions between homologous DNA molecules were also proposed to occur in the absence of direct contacts , by a very different type of a mechanism called "electrostatic zipper" [27] . In this model , two double-stranded DNA molecules with identical sequences can align by making multiple electrostatic contacts between negatively charged backbone phosphates on one duplex and positive charges in the grooves of the other duplex . In this model , DNA homology is read out indirectly: pairing occurs specifically between homologous sequences because only such sequences can produce complimentary patterns of negative and positive charges [27] . As pointed out by the authors , this model cannot account energetically for pairing between DNA molecules that are significantly different as is observed for the interspersed homologies that trigger RIP [28] . Consideration of these models , together with our constraining results , suggests that direct DNA/DNA homology recognition , which also occurs in vitro [29] may involve mechanism ( s ) that still remain ( s ) to be described . Furthermore , regardless of the local basis for homology recognition , there also remains the fundamental unsolved issue of how a pair of relatively short chromosomal regions with similar nucleotide sequences can accurately identify one another within the vast space and genomic complexity of the nucleus on a time scale that makes the “genome-by-genome” homology search for RIP feasible . In Neurospora , RIP evolved as a genome defense mechanism to control the expansion of mobile DNA [15] . In most eukaryotic organisms , including mammals , genomes contain vast amounts of repetitive DNA normally silenced in the form of heterochromatin [30] . While the role of RNA intermediates and sequence-specific DNA binding proteins were clearly implicated in many cases involving epigenetic silencing of DNA repeats [31 , 32] , in the other cases there is no clear understanding how the heterochromatic state can be induced over repetitive sequences [33 , 34 , 35] . It is possible that homology-dependent interactions analogous to those that underlie RIP could play a not yet considered role in other phenomena that promote the assembly of heterochromatin on repetitive DNA . We also note that recombination-independent pairing is a prominent feature of inter-chromosomal interactions in both somatic and meiotic systems [2] , and thus the rules that underlie homology recognition for RIP may well underlie a wide variety of homology-dependent phenomena in other biological systems .
Methods for creating interspersed homologies , constructing plasmids and strains , setting up crosses , recovering RIP products and analyzing mutations were previously described [14] . Briefly , interspersed homologies were designed in silico by substituting each designated base with one of the three remaining alternatives chosen with equal probabilities and independently from neighboring bases . The exact algorithm for creating interspersed homologies ( written in Perl ) is provided in S4 File . Synthetic DNA was ordered as “gBlocks” from Integrated DNA Technologies . Repeat cassettes were integrated as a replacement of the csr-1 gene in the mus-52Δ strain FGSC#9720 , which is deficient in the non-homologous end joining pathway and can only be transformed by homologous recombination [36] . 1–2 homokaryotic transformants were typically selected for further analysis . All integration events were validated by sequencing . Plasmids and strains created in this study are listed in S1 Table . Individual plasmid maps ( in GenBank format ) are provided in S1 File ( as tar/gz archive ) . The standard wildtype strain FGSC#4200 was used as a female parent for all the crosses . 1–3 replica crosses were analyzed for each repeat construct; at least 30 random “late” spores were randomly sampled from each cross . The number of replica crosses and the total number of analyzed spores are provided in S1 Table . PCR-amplified repeat cassettes were sequenced directly by the Sanger method . Sequencing reactions were read with an ABI3730xl DNA analyzer at the DNA Resource Core of Dana-Farber/Harvard Cancer Center ( funded in part by NCI Cancer Center support grant 2P30CA006516-48 ) . Individual chromatograms were assembled into contigs with Phred/Phrap . Assembled contigs were inspected manually in Consed . Sequences of all contigs analyzed in this study are provided in S3 File . | Recombination-independent pairing of homologous double-stranded DNA molecules is associated with many important biological processes including the alignment of homologous chromosomes in early meiosis , monoallelic gene expression in mammals , and somatic pairing of homologous chromosomes in Drosophila . The molecular mechanism ( s ) by which homologous sequences are recognized in all of these cases remain ( s ) largely unknown . The phenomenon of Repeat-Induced Point mutation ( RIP ) in the filamentous fungus Neurospora crassa provides an especially advantageous model system for elucidating the general mechanism of recombination-independent recognition of DNA homology . Here we show that imperfect ( interspersed ) homologies with overall sequence identity of only 36% can be efficiently detected by RIP in the absence of nearby perfect homology . Under these particularly stringent conditions , specific DNA sequence motifs are found to play a critical role . Our analysis of one such situation identifies the triplet 5'-GAC-3' as an especially favorable unit of homology . We also present experimental evidence that the process of homology recognition for RIP integrates sequence information over hundreds of base-pairs of chromosomal DNA , and that this process can be uncoupled from the downstream mutation . Taken together , our results advance the notion that sequence information can be compared directly between double-stranded DNA molecules during RIP and , potentially , in other processes where homologous pairing of intact DNA molecules is observed . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"sequencing",
"techniques",
"homologous",
"chromosomes",
"mutation",
"sequence",
"motif",
"analysis",
"molecular",
"biology",
"techniques",
"dna",
"dna",
"recombination",
"homologous",
"recombination",
"research",
"and",
"analysis",
"methods",
"sequence",
"analysis",
"seq... | 2016 | Recombination-Independent Recognition of DNA Homology for Repeat-Induced Point Mutation (RIP) Is Modulated by the Underlying Nucleotide Sequence |
Taenia solium cysticercosis affects millions of impoverished people worldwide and can cause neurocysticercosis , an infection of the central nervous system which is potentially fatal . Children may represent an especially vulnerable population to neurocysticercosis , due to the risk of cognitive impairment during formative school years . While previous epidemiologic studies have suggested high prevalence in rural China , the prevalence in children as well as risk factors and impact of disease in low-resource areas remain poorly characterized . Utilizing school based sampling , we conducted a cross-sectional study , administering a questionnaire and collecting blood for T . solium cysticercosis antibodies in 2867 fifth and sixth grade students across 27 schools in west Sichuan . We used mixed-effects logistic regression models controlling for school-level clustering to study associations between risk factors and to characterize factors influencing the administration of deworming medication . Overall prevalence of cysticercosis antibodies was 6% , but prevalence was significantly higher in three schools which all had prevalences of 15% or higher . Students from households owning pigs ( adjusted odds ratio [OR] 1 . 81 , 95% CI 1 . 08–3 . 03 ) , from households reporting feeding their pigs human feces ( adjusted OR 1 . 49 , 95% CI 1 . 03–2 . 16 ) , and self-reporting worms in their feces ( adjusted OR 1 . 85 , 95% CI 1 . 18–2 . 91 ) were more likely to have cysticercosis IgG antibodies . Students attending high prevalence schools were more likely to come from households allowing pigs to freely forage for food ( OR 2 . 26 , 95% CI 1 . 72–2 . 98 ) and lacking a toilet ( OR 1 . 84 , 95% CI 1 . 38–2 . 46 ) . Children who were boarding at school were less likely to have received treatment for gastrointestinal worms ( adjusted OR 0 . 58 , 95% CI 0 . 42–0 . 80 ) . Our study indicates high prevalences of cysticercosis antibodies in young school aged children in rural China . While further studies to assess potential for school-based transmission are needed , school-based disease control may be an important intervention to ensure the health of vulnerable pediatric populations in T . solium endemic areas .
Infection with the zoonotic tapeworm Taenia solium affects millions of people living in poverty throughout Asia , Africa , and Latin America [1] . Considered a neglected tropical disease , infection is linked to inadequate sanitation and hygiene , presence of free roaming pigs , and poverty [2] . Infection in humans has two manifestations: intestinal taeniasis where humans serve as the definitive hosts for the adult tapeworm which inhabits the gastrointestinal tract , and cysticercosis , a tissue infection where humans are the accidental dead-end host for the cystic larvae ( cysticercus ) . Intestinal infestation with the adult tapeworm develops when humans consume improperly cooked pork containing cysticerci . The consumed cyst is released in the small intestines where the adult worm develops , attaches to the intestinal wall , and liberates thousands of eggs , which are shed , along with gravid proglottids , in human feces . T . solium eggs contaminate the environment and are consumed by pigs that ingest human feces directly or indirectly through contaminated agricultural products . Once consumed , the larval forms encyst in porcine muscle completing the cycle [3] . Human cysticercosis develops when T . solium eggs are consumed by humans through auto-infection , consumption of contaminated food or water , or close contact with a tapeworm carrier [4] . Upon ingestion of mature eggs , the hatched parasite migrates to tissues throughout the body including muscle , sub-cutaneous tissues , orbits , and the central nervous system ( CNS ) [5–7] . Neurocysticercosis ( NCC ) develops when T . solium larva establishes itself in the CNS and may lead to morbidity which can be fatal [5 , 8 , 9] . NCC causes a range of symptoms depending on number , stage of involution , volume , and location of the lesions [10] , including seizures , chronic headaches , focal neurological deficits , psychiatric disturbances , and cognitive impairment [3 , 11–15] . The full prevalence of NCC is difficult to establish , it is estimated to be responsible for 29% of acquired epilepsy in endemic areas [16] , and has been identified as a leading cause of death from foodborne disease resulting in 2 . 8 million disability-adjusted life years lost in 2010 [2] . T . solium has been reported throughout China , with hyperendemic foci mainly in southwest regions [17] . Overall , NCC may affect up to an estimated 7 million people in China [1] . However , the risk factors for infection and impact of disease in rural areas remain poorly characterized [17 , 18] . Children may represent an especially vulnerable population to NCC , with resulting neurological problems and cognitive impairment during formative school years possibly leading to poor academic performance , contributing to high drop-out rates and , eventually , propagating cycles of poverty . Schools may represent centers of transmission combining poor hygienic standards and close contact in a large and vulnerable population . Despite this hypothetical risk , the prevalence of cysticercosis within schools has not been well evaluated . Here we report the prevalence of cysticercosis antibodies and associated risk factors in school-aged children using school-based sampling in western Sichuan , People’s Republic of China .
The school principals , who are the children’s legal guardians while they are boarding at school , provided initial written consent for student participation , and each student provided verbal assent prior to participating . Consent and assent status were documented by field staff , and students who did not assent did not participate in the questionnaire or blood collection . Field staff and school staff described the purpose of the study using pre-written scripts and were available on hand to answer questions that students might have either about the study or questionnaires . Information on the study and a consent form were part of the take-home questionnaire , requesting written consent from adult caretakers in regard to their and their family member’s involvement . All participants were allowed to keep a copy of the consent/assent document . The study and all methodology were approved by the institutional review board of Stanford University ( study ID 35415 ) and the ethical review board of West China School of Public Health , Sichuan University ( K2015031 ) . The data used in this analysis were collected as part of an epidemiological field study conducted in November 2015 investigating T . solium cysticercosis and NCC prevalence , associations between NCC infection and academic performance , and risk factors for infection in school-aged children . Results of the cognitive and academic assessments , prevalences of NCC cases , and frequency of neurologic problems will be presented in another manuscript and are not further discussed here . The study was carried out in rural mountainous areas of western Sichuan . Located at the eastern extremity of the Tibetan Plateau and with an average altitude of 2700 meters , these areas are largely characterized by smallholder farmers who raise pigs and partake in small-scale agriculture . These areas were selected as they had been identified as having high endemicity of Taenia species based on smaller scale studies [18] . In these remote areas , there is no routine mass antihelminthic drug administration for pediatric or adult populations . To collect data which could be generalizable to school-aged children living in farming and pig raising communities in western Sichuan with known T . solium risk factors we used a school-based sampling technique . The study was conducted in three counties , each from one of three prefectures in western Sichuan . In selecting the three study counties ( S1 Fig ) , we considered all counties within Aba ( 13 counties total ) , Ganzi ( 18 counties total ) , and Liangshan ( 16 counties and 1 autonomous county total ) prefectures for the study . All counties in these prefectures are known to have smallholder pig raising activity and risk factors for human cysticercosis including open defecation and free range pigs based on reports from local public health workers to Sichuan Centers for Disease Control and Prevention ( Sichuan CDC ) . Because we wanted to enroll students at risk for T . solium cysticercosis , we only considered counties where previous cases of human cysticercosis and NCC had previously been detected by Sichuan CDC or where there were previous reports to suggest T . solium taeniasis [18] . We further favored counties with the largest populations of fifth and sixth grade students as estimated by demographic data collected previously by colleagues at Sichuan CDC . Because these mountainous communities are often difficult to access due to poor roads and long distances , we focused on counties where all schools could be accessed by field teams and where local public health services were supportive of the field work . All schools enrolling fifth and sixth graders were sampled within the three selected counties and all students in both grades were sampled at each school . We selected fifth and sixth grade students for the cognitive study because standardized tests can be administered to this age group easily and these students are old enough to have repeat and significant exposures . At each school , field teams administered a student questionnaire , aseptically collected approximately 5 ml of blood by venous puncture for T . solium cysticercosis IgG testing by ELISA , and provided a take-home questionnaire to be completed by the child’s parent or guardian . Student questionnaires ( S2 Document ) covered basic demographic data , home environment and family asset ownership , animal ownership , pork consumption , and student toileting behavior . If the family owned pigs , students were asked if they ever saw their family’s pigs eating human feces or if the pigs ever went to the areas where people defecated . While data on specific drugs could not reliably be collected , use of antihelminthics was assessed by asking students if they had taken any medication for “intestinal worms” in the past year . Finally , students were asked about symptoms and perceptions of intestinal worms . To assess if tapeworms or proglottids might be present in the child’s feces , students were asked if they had seen “worms” or “pieces of worms” in their feces . Children were also provided with a take-home adult questionnaire which was completed by the head of household in their home . After completion , adult questionnaire forms were returned to the school by the child . On the adult questionnaire , head of households were asked about pig ownership and pig husbandry methods over the last year and slaughtering and meat preparation practices . Adults were also asked about agricultural practices in the five years preceding the questionnaire and about their knowledge and attitudes towards intestinal worms and the administration of medications for intestinal worms . Serum was tested using an enzyme-linked immunosorbent assay ( ELISA ) based on low-molecular-weight antigens ( LMWAgs ) of T . solium cysticerci collected from pigs in Chinese endemic areas . LMWAgs based assays have been shown to be highly sensitive and specific [19 , 20] , have been used in previous field studies [21] , and are especially attractive—given their low cost , quantifiable result , and simplicity—for use in low resource areas [20] . Detailed assay methodology has been published previously [20] . Antigen for the LMWAgs ELISA was obtained from cyst fluids of T . solium metacestodes collected from infected pigs in endemic areas of China . ELISA plates ( Nunc-ImmunoTM plate , Maxisorp TM Surface , Thermo Fisher Scientific , Denmark ) were coated with 100 μl of diluted LMWAgs at 1 μg/ml in PBS overnight at 4°C . Plates were rinsed twice with PBS containing 0 . 1% Tween 20 ( PBST ) and then blocked with 300 μl of blocking solution ( 10 mM Maleic acid pH 7 . 5 , 150 mM NaCl , 1 . 0% casein , 0 . 1% Tween 20 ) at 37°C for 1 hour . Serum samples were diluted in blocking solution at 1:100 . Plates with 100 μl of diluted sera in duplicate wells were incubated at 37°C for 1 hour . The wells were washed 3 times with PBST , incubated with 100μl of recombinant protein G conjugated with peroxidase ( Invitrogen ) at 1:4000 in blocking solution at 37°C for 1 hour . After washing 3 times with PBST , plates were incubated with 100 μl of substrate ( 0 . 4 mM 2 , 2’- azino-di-[3-ethyl-benzhiazoline sulfonate] in 0 . 1 M citric acid buffer , pH 5 . 3 ) for 30 minutes at 37°C . Color reaction was stopped by application of 1% SDS in each well . The optical density at 405 nm was evaluated with an ELISA reader . The cut-off point was determined as the mean optical density ( OD ) plus 3 times the standard deviation for a panel of serum samples obtained from healthy Chinese donors ( n = 30 ) . We first characterized prevalence of cysticercosis antibodies , student demographics , reported toileting behaviors , and reported pig husbandry and agricultural practices using basic descriptive statistics . We constructed mixed-effects logistic regression models to better characterize differences in T . solium antibody seroprevalence at the school-level , to study associations between T . solium cysticercosis exposure and demographic , environmental , and behavioral factors , and to investigate what factors affected the likelihood of children receiving deworming medications in the year preceding the study . We identified schools with higher student populations with T . solium cysticercosis antibodies by building mixed-effects logistic models with the serologic result as the dependent variable , school as the independent variable , and county as a random effect . A school with a cysticercosis antibody prevalence closest to the mean value in the dataset was used as the model reference . To build models to assess the associations between risk factors and the presence of T . solium cysticercosis IgG antibodies or deworming medication administration , we first framed a causal diagram to identify associations between variables of interest ( Fig 1 ) [22] . In the model assessing associations between risk factors and the presence of T . solium cysticercosis IgG antibodies we used the child’s serologic result for human cysticercosis IgG as the dependent variable . In the model assessing deworming medication administration we used the report of the child receiving medication for intestinal worms in the year preceding the study as the dependent variable . Missing values in independent variables were imputed assuming an ignorable missingness mechanism using multivariate imputation by chained equations [23] . Numeric variables were imputed using predictive mean matching , logical variables using logistic regression , and categorical variables with more than two levels with multinomial logit models [23] . Predictors for multiple imputation included the independent variable in question , all dependent variables outlined in the causal diagram , demographics ( age , sex , ethnicity , and asset score ) , and geographic location ( school and county ) . Fifty imputed datasets were generated . Conditional rules were applied to ensure that imputation did not create impossible combinations ( for example , a non-crop growing household reporting using human waste as crop fertilizer ) . Finally , results obtained using multiple imputation were compared both to available-case and complete-case analyses ( shown in supplemental tables ) . All results reported within the manuscript are from multiple imputed analyses . Odds ratios were calculated by pooling results across all imputed datasets . For each of the two dependent variables , we first constructed mixed-effects models consisting of a single independent variable controlling for school clustering as a random effect . To select variables to include in a best-fit multivariable mixed-effects logistic regression model , we used a two-step approach . Five imputed datasets were used for variable selection , and the final selected models were run on all fifty imputed datasets [24] . We first used information-theoretic model selection to generate all possible combinations of independent variables which had resulted in p-values less than 0 . 1 in our initial analysis and selected the model with the lowest corrected Akaike information criterion ( AIC ) for each of the five imputed dataset [25] . In the second variable selection step , variables selected in at least 50% of the five imputed datasets were assessed by backwards selection and retained if the Wald test resulted in a p value of less than 0 . 05 [24 , 26] . To add a measure of wealth to our models we used principal components analysis ( PCA ) to aggregate asset ownership variables into one standardized asset score [27] . To identify if exposures and reported behaviors differed between students attending schools with the highest seroprevalences of T . solium IgG antibiodies compared to students attending the lowest prevalence schools , we compared the proportion of students reporting selected behaviors and exposures in the highest prevalence schools to students in all remaining lower prevalence schools using Fisher’s Exact Test . Independent continuous variables included in mixed-effects models were confirmed to be linearly related to the log odds . We assessed multicollinearity between independent variables in multivariable models using variance inflation factors . Analysis was conducted in R [28] utilizing the lme4 package [29] for logistic mixed-effects models , the MuMIn package for information-theoretic model selection [30] , and the MICE package to perform multivariate imputation by chained equations [23] .
Overall , 11% ( 283/2606 , 95% confidence interval [CI] 10–12% ) of all fifth and sixth grade children self-reported that they had seen what appeared to be worms or worm segments in their feces ( Table 2 ) . The highest prevalence was in Muli County , where 14% ( 141/1013 , 95% CI 12–16% ) reported worms or worm segments in their feces . The overall prevalence of serum T . solium cysticercosis IgG antibodies in fifth and sixth grade students in the study area was 6% ( 180/2867 , 95% CI 5–7% ) ( Table 2 ) . The county prevalences in all enrolled fifth and sixth grade students ranged from 8% ( 78/1016 , 95% CI 6–10% ) in Muli County to 6% ( 72/1138 , 95% CI 5–8% ) in Yajiang County , and 4% ( 30/713 , 95% CI 3–6% ) in Ruoergai County . Three schools had significantly higher prevalences of students with serum T . solium cysticercosis IgG antibodies when compared to a school with the mean prevalence in the study area ( Fig 2 ) : a school in Muli County with a prevalence of 15% ( 16/105 , odds ratio [OR] 2 . 3 , 95% CI 1 . 2–4 . 6 ) , a school in Yajiang County with a prevalence of 22% ( 19/88 , OR 3 . 6 , 95% CI 1 . 9–6 . 9 ) , and a school in Ruoergai County with a prevalence of 20% ( 22/111 , OR 3 . 2 , 95% CI 1 . 7–5 . 9 ) . Four schools , three in Ruoergai County and one in Muli County , had no students with antibodies , although these results were not found to be statistically different than the mean prevalence in the study area . In mixed-effects logistic models consisting of a single independent variable ( Table 3 , see S1 Table for comparison of available-case , complete-case , and multiple imputed analyses ) and controlling for school-level clustering , children were more likely to have IgG antibodies to T . solium cysticerci if they lived in households that owned pigs ( 4% vs 7% , OR 1 . 81 , 95% CI 1 . 09–3 . 01 ) and if the family reported feeding household human feces to pigs ( 5% vs 11% , OR 1 . 54 , 95% CI 1 . 07–2 . 24 ) . While the odds ratio crossed one , suggesting no clear benefit , there was a trend suggesting treatment of human feces prior to use as a crop fertilizer resulted in less cysticercosis exposure: 6% of children had cysticerosis antibodies in households reporting never treating feces before use compared to 2% if human feces were always treated . There was also a trend towards more children with cysticercosis antibodies in households that did not own toilets ( 5% compared to 8% ) . Children who reported worms or worm segments in their feces were more likely to have serologic evidence of T . solium cysticercosis ( 10% compared to 6% , OR 1 . 60 , 95% CI 1 . 03–2 . 5 ) . The factors most associated with the presence of IgG antibodies to cysticercosis were pig ownership , feeding human feces to pigs , the presence of worms or worm segments in the child’s feces , and the child having been given medication for gastrointestinal worms ( see S2 Table for variable selection results ) . In the multivariate model ( Table 3 ) , children who came from households that owned pigs and reported feeding the household’s human feces to their pigs were more likely to have serologic evidence of cysticercosis antibodies ( adjusted OR 1 . 81 , 95% CI 1 . 08–3 . 03 and adjusted OR 1 . 49 , 95% CI 1 . 03–2 . 16 , respectively ) . In the multivariate model , children who reported worms or worm segments in their feces were more likely to have antibodies ( adjusted OR 1 . 85 , 95% CI 1 . 18–2 . 91 ) , children who reported receiving medication for gastrointestinal worms in the year preceding the study were less likely to have cysticercosis antibodies ( adjusted OR 0 . 52 , 95% CI 0 . 31–0 . 90 ) . Students attending the three schools with the highest seroprevalences of T . solium IgG antibiodies differed from those attending schools with lower prevalences in reported demographics , behaviors , and exposures ( Fig 3 ) . Students attending the highest prevalence schools were more likely to be Tibetan ( OR 7 . 15 , 95% CI 3 . 66–13 . 99 ) , report boarding at school ( OR 3 . 69 , 95% CI 2 . 64–5 . 14 ) , and come from households without toilets ( OR 1 . 84 , 95% CI 1 . 38–2 . 46 ) . Students attending the highest prevalence schools were less likely to have received medication for gastrointestinal worms in the year preceding the study ( OR 0 . 36 , 95% CI 0 . 20–0 . 66 ) ( Fig 3A ) . Students attending the highest prevalence schools were more likely to come from pig owning households ( OR 4 . 75 , 95% CI 3 . 08–7 . 33 ) , report that their household’s human feces were fed to pigs ( OR 1 . 55 , 95% CI 1 . 18–2 . 05 ) , and report that they had seen pigs consuming human feces in the environment ( OR 2 . 66 , 95% CI 1 . 93–3 . 67 ) ( Fig 3B ) . Students attending the highest prevalence schools were more likely to come from households reporting always allowing their pigs to freely forage ( OR 2 . 26 , 95% CI 1 . 72–2 . 98 ) and less likely to come from households not allowing their pigs to forage ( OR 0 . 30 , 95% CI 0 . 20–0 . 44 ) . While the frequency of reported pork consumption ( Fig 3C ) was the same in both student populations , students attending the highest prevalence schools were more likely to report that their pork came from home raised pigs ( OR 2 . 62 , 95% CI 1 . 94–3 . 55 ) . Differences in agricultural practices were less pronounced ( Fig 3D ) , although children attending the highest prevalence schools were slightly more likely to come from households growing crops ( OR 2 . 82 , 95% CI 1 . 59–5 . 00 ) , using human feces as a fertilizer ( OR 1 . 34 , 95% CI 1 . 06–1 . 70 ) , and attempting to treat human feces through fermentation or composting before use as a fertilizer ( OR 1 . 35 , 95% CI 1 . 04–1 . 77 ) . Thirty-two percent of children reported having “intestinal worms” in the year preceding the study ( 818/2579 ) . Most children reported that they had realized they were infected due to abdominal pain ( 55% , 448/818 ) , while a smaller number reported seeing worms or worm pieces in their feces ( 20% , 163/817 ) or had been told that they had intestinal worms by a doctor ( 14% , 114/818 ) . When asked about their impression of gastrointestinal worms in their children , 35% ( 866/2469 ) of adult household respondents felt that intestinal worms had no adverse effects , 30% felt that worms could stunt children’s growth ( 765/2469 ) , and 3% ( 69/2468 ) felt that worms could have a positive effect on children . When asked to identify the best treatments for gastrointestinal worms , 75% of adults ( 1832/2431 ) identified medication provided by a doctor , 10% ( 232/2431 ) suggested a combination of reducing outdoor activity and drinking hot water , and 9% ( 210/2431 ) suggested that spicy food should be consumed . Fourteen percent ( 379/2613 ) of children reported that they had taken a medication for gastrointestinal worms in the year preceding the study . In mixed-effects logistic models consisting of a single independent variable ( Table 4 , see S3 Table for comparison of available-case , complete-case , and multiple imputed analyses ) and controlling for school-level clustering , children who reported their ethnicity as “other” ( OR 2 . 43 , 95% CI 1 . 36–4 . 32 ) and who reported seeing worms or worm segments in their feces in the preceding year ( OR 4 . 41 , 95% CI 3 . 29–5 . 91 ) were more likely to have received treatment . Children who were older ( OR 0 . 89 , 95% CI 0 . 82–0 . 97 ) , were of the Yi ethnicity ( OR 0 . 48 , 95% CI 0 . 23–0 . 99 ) , were boarding at school ( OR 0 . 52 , 95% CI 0 . 38–0 . 70 ) , and were wealthier as classified by household asset score ( wealthiest quartile: OR 0 . 60 , 95% CI 0 . 40–0 . 91 ) were less likely to have received medication for gastrointestinal worms in the year preceding the study . Parental educational level and impressions of gastrointestinal worm influenced if children received treatment . In families where adults reported worms having no adverse effects , fewer children received medication ( OR 0 . 72 , 95% CI 0 . 54–0 . 95 ) . Higher levels of education were associated with increasing medication administration , with parents achieving a junior high education or higher more likely to provide medication ( junior high school: OR 1 . 60 , 95% CI 1 . 08–2 . 38; high school or higher: OR 2 . 37 , 95% CI 1 . 48–3 . 79 ) . The factors most associated with children receiving medication for gastrointestinal worms were age , school boarding status , level of parental education , parental understanding of adverse events caused by gastrointestinal worms , and the child reporting worms or worm segments in their feces in the last year ( see S4 Table for variable selection results ) . In the multivariate model ( Table 4 ) , children of more highly educated parents ( high school or higher: adjusted OR 1 . 81 , 95% CI 1 . 11–2 . 98 ) and with worms or worm segments in their feces ( adjusted OR 4 . 43 , 95% CI 3 . 29–5 . 98 ) were more likely to receive treatment for gastrointestinal worms . Children who were older ( adjusted OR 0 . 91 , 95% CI 0 . 83–0 . 99 ) , boarding at school ( adjusted OR 0 . 58 , 95% CI 0 . 42–0 . 80 ) , and who had parents who felt intestinal worm infestation had no adverse effects ( adjusted OR 0 . 68 , 95% CI 0 . 51–0 . 92 ) were less likely to receive medication .
Our study demonstrates high prevalence of T . solium cysticercosis antibodies in school-aged children in poor , pig-raising areas in western Sichuan . The use of schools as a unit , rather than the typical village , is a unique approach and reveals variation in T . solium antibody seroprevalence across schools in close geographic proximity . We identified three schools with significantly higher prevalences of T . solium cysticercosis antibodies than surrounding schools . Schools with the highest prevalence of T . solium cysticercosis antibodies had differences in reported behaviors and exposures compared to lower prevalence schools , with higher proportions of students in the highest prevalence schools reporting the consumption of home raised pigs , living in households without toilets , and coming from households were the family’s pigs are allowed to freely forage and fed human feces . The seroprevalence of cysticercosis in children varies widely in the literature , from approximately 20% of 10–19 years olds found to be antigen positive in a village based study in the Democratic Republic of Congo [31] , to approximately 12% of 11–20 year olds having antibodies in a hyperendemic area of Peru [32] , and a study in three provinces of Burkina Faso showing T . solium antigen prevalence ranging from 2 . 3% to 0 . 7% in the youngest cohorts [33] . Although it is difficult to compare results given differing laboratory methodology and our school centered approach , the prevalence of T . solium cysticercosis antibodies in the highest prevalence schools in our study seem similar to levels reported in children in high endemic areas . The risk factors most associated with cysticercosis antibodies in children identified in our best-fit multivariate analysis included pig ownership , the child self-identifying worms or worm segments in their feces , and households allowing their pigs to consume human feces . The link between the presence of antibodies and a recent history of young children possibly passing proglottids is consistent with previously published literature [32] . It is unclear from our study how many children are tapeworm carriers and are auto-infecting themselves , although given the poor handwashing practices among young children [34] , there is likely substantial risk for auto-infection . Another potential explanation for this finding is that children who are passing proglottids are , along with their family members , consuming undercooked pork and therefore are likely surrounded by multiple members of their household who are harboring intestinal tapeworms . The consumption of human feces by pigs results in infected pork and likely results in higher proportions of human intestinal infestation with the adult T . solium tapeworm . Children and adults in the study area often report either pigs eating the household’s human feces or report seeing their pigs eating human feces while foraging in the environment . Free range pigs’ access to human feces—made easier by lack of latrines and open defecation—has been frequently identified as a risk factor for cysticercosis [33 , 35–37] . The use and acceptance of human feces as pig feed has been recognized as affecting household practices and preferences , for example respondents in a qualitative study on latrine use in Zambia voiced concern that building latrines would result in less available pig feed [38] . Risk factors identified in the school comparison are also consistent with factors previously identified in the literature . Pig ownership has been identified as a risk factor for cysticercosis and taeniasis in studies in Africa and South America [33 , 39] , and higher risk of seropositivity has been seen in households that consume home raised pigs [39] . While not always meeting criteria for statistical significance , our analysis did suggest trends between agricultural techniques and cysticercosis exposure . Our comparison of the highest with lower prevalence schools suggested that children who attend the highest prevalence schools are more likely to come from households that grow their own crops . The use of human feces as fertilizer for crops has been recognized as a potential risk factor in previous studies [40 , 41] . While some households reported treating human feces prior to use as fertilizer , this practice is not common and the effectiveness of household techniques is unclear , especially given that consistently achieving the required levels of temperature , pH , and dryness to deactivate T . solium eggs may be difficult in household latrines [42] . The role of treating human feces prior to use as fertilizer and other agricultural practices in reducing infection risk deserves further characterization and study . Unlike previous published studies , our study did not show any association between serologic status and pork consumption levels [33 , 39] . Our failure to detect this is perhaps due to the high pork consumption in the area and the fact that children , who do not prepare their own meals at home or at school , are likely poor judges of their pork intake . Additionally , we did not find an association between serologic status and poverty level [33] . This may be related to economic status being similar across the entire study area or to little variation in risk factors with increasing wealth given the agricultural and remote nature of our study communities . Given that cyticercosis cases cluster around tapeworm carriers [32 , 35 , 43] , it is possible that schools are acting as centers for transmission in pediatric populations . Schools represent large congregations of children , and risk for fecal-oral transmission and passage of eggs from tapeworm carriers is likely high . If this is the case , efforts to reduce school fecal-oral transmission may serve as a tool to interrupt disease transmission . Treating human tapeworm carriers with antihelminthic medication eliminates the adult tapeworm , destroying the source of infection and preventing human and porcine cysticercosis [44] . While we did not collect data on specific antihelminthic use , our study , which did include questions regarding administration of medication to treat gastrointestinal worms , suggests that few children receive therapy . More concerning , our analysis showed that children who are boarding at school were less likely to receive medication than students living at home . Treatment is most likely to be administered by more educated parents who are aware of the potential adverse effects of tapeworm infestation . If tapeworm carriers are present in schools , the distribution of antihelminthic medications at schools could decrease possible school-based transmission between students . Our study has some weaknesses that limit our scientific inference . We used a LMWAgs ELISA to detect antibodies to human T . solium cysticercosis . While historically less sensitive and specific than enzyme-linked immunotransfer blot ( EITB ) assays [45] , ELISA performance for serodiagnosis of human cysticercosis has improved with the development of more sophisticated methods for producing antigenic proteins [46] . Use of LMWAgs rather than crude cyst fluid results in improved performance and less cross-reactivity with other pathogens [20] . However , evaluation of LMWAgs based ELISAs continues to suggest some weak cross-reaction with alveolar and cystic echinococcosis [20] . Because echinococcosis is endemic throughout regions of northwestern Sichuan [47] , cross-reactivity may be causing us to over-estimate the prevalence of human T . solium cysticercosis antibodies . In this case , we suspect that misclassification caused by cross-reactivity is minimal given that echinococcosis in northwestern Sichuan is more common in pastoral herding communities than the farming communities which inhabit the regions included in this work [47] . Because the overall larger study was designed to assess the relationship between NCC and cognitive outcomes , we selected counties and schools to maximize disease based in small initial studies suggesting presence of NCC and human cysticercosis in the study areas . This means that our results may not be fully representative over a larger geographical area , as prevalences of disease may be higher in our area of study . Some of the risk factors that failed to achieve significance in the best-fit model are risk factors for gastrointestinal taeniasis and likely failed to achieve significance because our selected laboratory outcome was a serologic test for cysticercosis antibodies . Our measure of taeniasis was based on students self-reporting worms or worm segments in their feces . We were not able to confirm if these reported gastrointestinal infestations were caused by T . solium nor were we able to conduct large scale stool testing to detect cases of taeniasis . T . saginata and T . asiatica are known to be present in the region [18] , so some of these self-reported cases may represent other Taenia species or soil transmitted helminths . Our findings do suggest that students reporting worms or worm segments in their feces are more likely to have T . solium antibodies . Given the likely inclusion of gastrointestinal worms other than T . solium in our data collection , we may be underestimating the risk for cysticercosis associated with T . solium taeniasis . Stool testing and laboratory confirmation will be required to better characterize the prevalence of taeniasis and better clarify the associated risk for cysticercosis in the study area . Because we do not have infection data on pigs raised in the study area , we cannot correlate prevalence with presence and density of infected pigs . Because of the mountainous terrain , long distances , and presumed limited movement of both villagers and pig populations , it is very possible that prevalence of T . solium cysticerci in pigs may vary widely in areas that are geographically proximal but isolated due to terrain features , and this may explain some of the variation in human seroprevalence . Given the complex biology of T . solium , the addition of measures for gastrointestinal taeniasis in humans and prevalence of cysticercosis in pigs would provide a more complete picture of the disease ecology . Finally , because our study is questionnaire based , children failing to answer questions and adults failing to return take home questionnaires may have limited our ability to make school specific characterizations . In this case , however , overall participation in the study across the entire geographic area was high . We have shown a high prevalence of T . solium cysticercosis antibodies in school-aged children with school-based clustering . These findings raise concerns for NCC in school-aged children and possible cognitive deficits caused by CNS infection , which could result in long-term negative health , economic , and social effects . T . solium is an eradicable disease . Combined approaches addressing community education , improvements in hygiene and sanitation , improved pig management and meat handling , treatment of tapeworm carriers with antihelminthics , and porcine treatment through vaccination and chemotherapy have shown success in reducing transmission [48–50] . While further work identifying tapeworm carriers and potential routes of transmission within schools is needed , our work raises the hypothesis that schools may be sites of T . solium cysticercosis transmission and that school based interventions may , therefore , be an important addition to reduce disease among vulnerable pediatric populations in T . solium endemic areas . | The zoonotic tapeworm , Taenia solium , affects millions of impoverished people worldwide and can cause neurocysticercosis ( NCC ) , an infection of the central nervous system which is potentially fatal . Hypothetically , children may be a vulnerable population to infection as neurological problems and cognitive impairment caused by NCC during formative school years may lead to poor academic performance , contributing to drop-out rates and , eventually , propagating cycles of poverty . We carried out a school-based study of T . solium cysticerosis in primary school-aged children in rural western Sichuan . Our results indicate high levels of T . solium exposure in young school-aged children in rural China . While further studies to assess disease transmission within schools are needed , school-based disease control may be an important intervention to ensure the health of pediatric populations at risk for infection . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"children",
"medicine",
"and",
"health",
"sciences",
"enzyme-linked",
"immunoassays",
"education",
"sociology",
"tropical",
"diseases",
"vertebrates",
"social",
"sciences",
"diet",
"parasitic",
"diseases",
"animals",
"mammals",
"pediatrics",
"animal",
"products",
"age",
... | 2018 | Prevalence and risk factors for Taenia solium cysticercosis in school-aged children: A school based study in western Sichuan, People’s Republic of China |
Metabolic control of gene expression coordinates the levels of specific gene products to meet cellular demand for their activities . This control can be exerted by metabolites acting as regulatory signals and/or a class of metabolic enzymes with dual functions as regulators of gene expression . However , little is known about how metabolic signals affect the balance between enzymatic and regulatory roles of these dual functional proteins . We previously described the RNA binding activity of a 63 kDa chloroplast protein from Chlamydomonas reinhardtii , which has been implicated in expression of the psbA mRNA , encoding the D1 protein of photosystem II . Here , we identify this factor as dihydrolipoamide acetyltransferase ( DLA2 ) , a subunit of the chloroplast pyruvate dehydrogenase complex ( cpPDC ) , which is known to provide acetyl-CoA for fatty acid synthesis . Analyses of RNAi lines revealed that DLA2 is involved in the synthesis of both D1 and acetyl-CoA . Gel filtration analyses demonstrated an RNP complex containing DLA2 and the chloroplast psbA mRNA specifically in cells metabolizing acetate . An intrinsic RNA binding activity of DLA2 was confirmed by in vitro RNA binding assays . Results of fluorescence microscopy and subcellular fractionation experiments support a role of DLA2 in acetate-dependent localization of the psbA mRNA to a translation zone within the chloroplast . Reciprocally , the activity of the cpPDC was specifically affected by binding of psbA mRNA . Beyond that , in silico analysis and in vitro RNA binding studies using recombinant proteins support the possibility that RNA binding is an ancient feature of dihydrolipoamide acetyltransferases . Our results suggest a regulatory function of DLA2 in response to growth on reduced carbon energy sources . This raises the intriguing possibility that this regulation functions to coordinate the synthesis of lipids and proteins for the biogenesis of photosynthetic membranes .
Accumulating evidence suggests that metabolism and gene expression are tightly linked . For instance , changes in metabolite levels affect protein modification , for example by acetylation or N-glycosylation , which in turn influences signal transduction and gene expression [1]–[3] . In line with this , several metabolic enzymes functioning in diverse pathways were found to possess unexpected RNA-binding properties by which they are proposed to regulate gene expression and other cellular processes ( reviewed in [4] , [5] ) . Often these proteins represent key enzymes of metabolic pathways , which make them particularly suitable to coordinate distinct biochemical pathways in response to changes in metabolism . In eukaryotic organisms , photosynthesis is performed in endosymbiotically acquired organelles , the chloroplasts . Within chloroplasts , the light-driven reactions of photosynthesis take place in thylakoid membranes , which represent a highly organized system of lipid membranes and embedded multisubunit protein complexes . These complexes include photosystem I ( PSI ) and photosystem II ( PSII ) , the cytochrome b6f complex , and the chloroplastic ATP synthase . The biogenesis of thylakoid membranes requires the synthesis of both lipids and proteins . Major lipids include two glycolipids , monogalactosyl diacylglycerol ( MGDG ) and digalactosyl diacylglycerol ( DGDG ) , the synthesis of which necessitates acetyl-CoA for fatty acid production within the chloroplast ( reviewed in [6] , [7] ) . This acetyl-CoA is mainly generated from pyruvate by the chloroplast pyruvate dehydrogenase complex ( cpPDC ) , which—like its mitochondrial counterpart ( mtPDC ) —is a megadalton complex consisting of multiple copies of three subunits; a pyruvate dehydrogenase ( E1 ) , a dihydrolipoamide acetyltransferase ( DLA , E2 ) , and a dihydrolipoyl dehydrogenase ( E3 ) [8]–[11] . The decarboxylation of pyruvate by cpPDC is a central reaction in chloroplast carbon metabolism and is regulated by light , Mg2+ , and feedback inhibition by acetyl-CoA and NADH ( for reviews , see [9] , [11] ) . The biogenesis of thylakoid membranes also necessitates the synthesis of polypeptides and their assembly into multisubunit complexes . Since their origin as a cyanobacterial endosymbiont , chloroplasts have retained downsized genomes and gene expression systems . However , most chloroplast proteins are encoded by the nuclear genome , synthesized in the cytosol , and then imported into the chloroplast . Therefore , the synthesis of thylakoid membrane protein complexes requires an intracellular coordination , which is mainly mediated via nucleus-encoded factors acting at all levels of chloroplast gene expression ( for a recent overview , see [12] ) . Particularly chloroplast translation initiation has been considered to play a key role in determining the levels of photosynthesis-related proteins ( for recent overviews , see [12]–[14] ) . Moreover , targeting of mRNAs to specific subcellular sites for localized translation contributes a further level of regulation of chloroplast gene expression [15] , [16] . These targeting mechanisms are thought to involve membrane-associated RNA binding proteins ( RBPs ) that tether respective mRNAs to specific membrane regions within the chloroplast [16] , [17] . The unicellular green alga Chlamydomonas reinhardtii is an established model organism for the analysis of biogenesis and gene expression in chloroplasts [18] . Moreover , it has the potential to be an ideal model system to study metabolic regulation of gene expression because it can adjust its metabolism to different energy sources . Unlike plants and animals , which derive energy exclusively from light and reduced carbon , respectively , C . reinhardtii can use both sources . It can exclusively use light energy in photoautotrophic growth , acetate in heterotrophic growth , or a combination of both in mixotrophic growth . The C . reinhardtii psbA mRNA is probably one of the most studied models of translational regulation in chloroplasts . The de novo assembly of PSII requires the synthesis of all subunits of this multimeric complex , including the psbA gene product , the D1 subunit , and its assembly partner D2 . This thylakoid membrane protein synthesis is localized to a specialized membrane region called the T ( translation ) -zone surrounding the pyrenoid , a spherical body and the primary site of CO2 assimilation in the chloroplasts of most algae [19] , [20] . However , under higher intensity light , D1 is damaged . Under these conditions , the so-called damage-repair cycle replaces degraded D1 proteins by newly synthesized ones ( reviewed in [21] ) . Interestingly , repair synthesis of the PSII reaction center protein D1 is not localized to the T-zone but distributed over stroma-exposed thylakoid membranes indicating a strict spatial separation from the PSII de novo synthesis machinery [20] . Using in vitro RNA-binding assays , we previously identified an RBP of 63 kDa , RBP63 , in C . reinhardtii , which is associated with stromal thylakoids and preferentially binds to an A-rich element within the 5′ UTR of the psbA mRNA [17] . This A-rich element is important for psbA translation [22] . Based on these findings , RBP63 was suggested to fulfill a role in membrane targeting and translation of the psbA mRNA [17] . Here , we report the identification and further characterization of RBP63 , which surprisingly turned out to be the dihydrolipoamide acetyltransferase subunit , DLA2 , of the cpPDC . Therefore , DLA2 might represent another example of a metabolic enzyme with an alternate function as an RBP involved in gene expression . Our data support a role of DLA2 in the localized translation of the psbA mRNA and confirm its known enzymatic role in acetyl Co-A synthesis . Moreover , the binding of this subunit to RNA might not be restricted to C . reinhardtii but appears to be a more general feature of dihydrolipoamide acetyltransferases , including the mitochondrial forms . Taken together , these results lead us to a new concept for the coordination of chloroplast translation and carbon metabolism that involves a cross-talk between protein and lipid synthesis in thylakoid membrane biogenesis .
RBP63 was purified from detergent-solubilized thylakoid membranes of C . reinhardtii by two sequential affinity chromatography steps , using first heparin and then poly ( A ) . RBP63 was monitored in the fractions on the basis of its psbA mRNA binding activity with a UV-cross-linking assay ( Figure 1 ) . We observed the highest RNA binding activity in the fraction eluted with 0 . 5 M KCl from poly ( A ) -Sepharose ( Figure 1B , pA E500 ) . SDS-PAGE and subsequent Coomassie Blue staining revealed that this fraction contained predominately the 63 kDa protein as well as minor amounts of proteins with different sizes ( Figure 1B , pA E500 ) . For the identification of the gene encoding RBP63 , the excised 63 kDa protein species was then subjected to proteolytic digestion and subsequent analysis by mass spectrometry . Surprisingly , four different peptides mapped to DLA ( DLA2 ) , the E2 subunit of the cpPDC ( Figure 2 ) . Blast searches of the C . reinhardtii nuclear genome using the DLA2 amino acid sequence identified two homologues: DLA1 and DLA3 . However , the mass spectrometrically identified peptides unambiguously assigned the purified protein to DLA2 ( Figure 2 ) . Moreover , in silico analysis of the N-terminal region of DLA2 by TargetP [23] revealed properties of a 30 aa N-terminal chloroplast transit peptide , while DLA1 and DLA3 were predicted to localize to mitochondria , and thus are likely to represent putative E2 subunits of the mitochondrial PDC . Mitochondrial localization of DLA1 and DLA3 is further supported by their identification during mass spectrometrical determination of the C . reinhardtii mitochondrial proteome [24] , [25] . To establish the gene structure of DLA2 , we sequenced the EST clone MXL069g06 ( BP097085 ) from the Kazusa DNA Research Institute . The clone included the full-length DLA2 cDNA , whose sequence corresponds to the gene model au5 . g10333_t1 ( chromosome 3:1419737–1423311 ) annotated in JGI v4 ( Joint Genome Institute; http://genome . jgi-psf . org/Chlre4/Chlre4 . home . html ) . The 2 , 174 nt DLA2 transcript contains six exons , comprising 1 , 485 nt , flanked by untranslated regions of 38 and 651 nt at its 5′ and 3′ ends , respectively . In silico analysis of the genome sequence and DNA hybridizations confirm that DLA2 is a single copy gene ( Figure S1 ) . The predicted protein has 494 amino acids with a molecular mass of 49 . 8 kDa . The discrepancy between the predicted molecular weight of DLA2 ( 49 . 8 kDa ) and the apparent MW determined from SDS-PAGE ( 63 kDa , Figure S7 ) represents a size anomaly observed for all PDC E2 subunits analyzed to date and has been explained by frequently occurring turn-inducing and charged amino acid residues within the interdomain linker regions of these proteins [26]–[29] . Consistently , the same phenomenon was found with recombinant His-DLA2 protein ( see below ) . In agreement with α-proteobacterial and cyanobacterial origins of mitochondrial and chloroplast E2 subunits , respectively , DLA2 shows an overall amino acid sequence identity of more than 50% to the E2 subunit from the cyanobacterium Synechocystis sp . PCC 6803 and the chloroplast E2 orthologue from the vascular plant Arabidopsis thaliana ( also named LTA2 , At3g25860 ) . In contrast , DLA2 shows only an identity of ∼30% to DLA1 , DLA3 , and mitochondrial orthologues from A . thaliana , S . cerevisiae , and H . sapiens [30] . The DLA2 sequence exhibits conserved regions for lipoamide attachment , E3 subunit binding , and a C-terminal 2-oxoacid dehydrogenase catalytic domain ( Figure 2 ) . The lipoamide attachment site and the catalytic domain show a relatively high conservation between the different organisms , whereas the E3 binding region is more variable . C . reinhardtii DLA2 contains a single predicted lipoamide attachment site , similar to the Synechocystis sp . PCC 6803 protein and the chloroplast E2 subunit from A . thaliana , whereas DLA1 and DLA3 exhibit two lipoyl binding sites similar to the mitochondrial enzymes from A . thaliana and H . sapiens ( Figure 2 ) . To verify the predicted chloroplast localization and exclude an additional targeting of DLA2 to mitochondria , we first carried out subcellular fractionation detecting DLA2 with an antiserum raised against the recombinant protein . The antibody detected a protein with an apparent molecular weight of 63 kDa in whole cell extracts ( Figure S7 ) . Subsequently , the comparisons of the level of DLA2 in subcellular fractions revealed it to be localized to chloroplasts , with approximately similar levels in thylakoid membranes versus the soluble stromal compartment ( Figure 3A ) . As is often seen for C . reinhardtii , the chloroplast and thylakoid fractions were contaminated with mitochondria as judged by following the mitochondrial alternative oxidase ( AOX; Figure 3A [31] , [32] ) . However , the most significant result of this analysis was that no DLA2-specific signal was detected in the mitochondrial fraction , confirming that DLA2 does not form part of the mitochondrial PDC . In addition , this result reveals that a cross-reactivity of the antibody to the putative mitochondrial isoforms , DLA1 and DLA3 , is improbable . As a second independent approach to determine the intracellular localization ( s ) of DLA2 , a DLA2–GFP fusion protein was expressed in C . reinhardtii ( Figure 2 ) . The DLA2–GFP signal was detected in the chloroplast , which was revealed by chlorophyll autofluorescence ( Figure 3B ) . By contrast , when only GFP was expressed from the same expression vector , its signal was detected primarily in the central nuclear-cytosolic region . Because DLA2 ( RBP63 ) was previously characterized as a psbA mRNA binding protein , we tested whether DLA2 forms part of a HMW ribonucleoprotein ( RNP ) complex . Detergent-solubilized thylakoid membranes were prepared from wild-type cells cultured under photoautotrophic ( light , no acetate ) , mixotrophic ( light , with acetate ) , or heterotrophic ( no light , with acetate ) conditions and then subjected to size exclusion chromatography ( SEC , [33] ) . By following the elution pattern of DLA2 , we verified that it forms part of a HMW complex in a size range between 700 kDa and more than 2 , 800 kDa under each of the conditions ( Figure 4 ) . Peak fractions of eluted DLA2 slightly varied between samples from light versus dark grown cells . Under photoautotrophic and mixotrophic growth conditions , the DLA2 complex was more than 2 , 800 kDa , as indicated by a peak of elution in fraction 1 ( Figure 4A , C ) . In cells grown heterotrophically , the peak fractions were moderately shifted toward smaller molecular sizes ( i . e . , fraction 2 ) ( Figure 4B ) . This suggests a light-dependent formation of these different DLA2 HMW complexes . To investigate whether the observed HMW complexes contain RNA , solubilized thylakoid membranes were treated with RNase prior to SEC analysis . Intriguingly , the detected complexes revealed a growth condition-dependent RNase sensitivity: only under mixotrophic conditions did the complex shift toward lower molecular weight fractions upon RNase treatment ( Figure 4C , peak of elution in fraction 4 ) . In contrast , no such size shifts were observed for material from cells grown under photoautotrophic or heterotrophic growth conditions ( Figure 4A , B ) . This strongly suggests that DLA2 forms a HMW–RNP complex in an acetate and light-dependent manner . As localization studies revealed an almost equal distribution of DLA2 in the stroma and the membrane fraction ( Figure 3A ) , we also investigated the size and RNase sensitivity of stromal DLA2 complexes of the wild-type grown under different conditions by SEC ( Figure S2 ) . In contrast to what was seen for the membrane-bound DLA2 complex ( Figure 4A–C ) , we did not observe significant RNase sensitivity of stromal DLA2 complexes under any condition . To obtain an indication of which of the gel filtration fractions contained active PDH complexes and additionally investigate if it is the active PDH complex itself that binds RNA or an alternative RNP complex , we performed cpPDC enzyme tests on the SEC fractions by measuring the reduction of NAD+ spectrophotometrically . These assays were performed on stroma or thylakoid SEC fractions from mixotrophically grown wild-type cells . The results revealed cpPDC activity mainly in the highest molecular weight fraction as expected ( Figure S3 ) . Approximately 10% of the activity measured in fraction 1 was detected in fraction 2 in a size range of ∼1 . 2–1 . 7 MDa , whereas no significant cpPDC activity was detectable in fractions containing smaller complexes . Moreover , no changes in the elution profile of cpPDC activity were observed upon RNase treatment even though significant amounts of DLA2 proteins were detected in lower molecular weight fractions 4 and 5 . Therefore , it is probably not the active cpPDC that binds RNA ( Figure S3A ) . To test whether the RNA bound by the DLA2 complex is the psbA mRNA , solubilized thylakoid membranes from the chloroplast psbA deletion mutant FuD7 ( grown under mixotrophic conditions ) were subjected to SEC ( Figure 4D , [34] ) . With this mutant , the DLA2 complex was mainly detected in fraction 4 , resembling the elution pattern of RNase-treated DLA2 complexes from mixotrophically grown wild-type cells . Reduced amounts of HMW complexes were detected in fractions 1+2 as compared to the WT , indicating that the lack of psbA mRNA in a mutant background reduces the DLA2 complex size . Therefore , this result supports the psbA mRNA as being the RNA component of the DLA2 RNP complex . We cannot exclude DLA2 binding to RNAs other than the psbA message . However , another mixotrophically grown PSII mutant ( nac2–26 ) tested in this analysis , which lacks the psbD mRNA , was not affected in DLA2 complex formation ( Figure 4D [35] , [36] ) . In conclusion , the data revealed a psbA mRNA-dependent DLA2 RNP complex under mixotrophic growth conditions . To further verify a specific interaction of DLA2 and the psbA mRNA in vivo , RNA co-immunoprecipitations ( co-IPs ) using solubilized thylakoids from mixotrophically grown wild-type cells were performed ( Figure S4 ) . The immunoprecipitate obtained with the αDLA2 antiserum contained the psbA mRNA , but not the mRNAs of rbcL or atpB , chloroplast genes encoding the large subunit of Rubisco and the beta subunit of ATP synthase , respectively . This confirms the formation of a specific psbA mRNA/DLA2 complex in vivo ( Figure S4B ) . In agreement with the results of SEC analysis , no psbA mRNA was co-immunoprecipitated from cells grown under heterotrophic conditions ( Figures 4B and S4B ) . It should be noted that this IP was relatively inefficient because the DLA2 antiserum only weakly recognizes the native protein ( Figure S4A ) . As a next step , we further substantiated the RNA binding capacity of DLA2 by testing the hexahistidine-tagged recombinant protein ( His-DLA2 ) for in vitro RNA binding activity ( Figure 5 ) . In UV cross-linking experiments with various 32P-labeled RNAs , His-DLA2 showed a clear intrinsic RNA binding activity for all probes tested ( e . g . , the 5′ UTRs of psbA , psbD—encoding the PSII reaction center protein D2—and rbcL ) ( Figure 5A ) . We next tested the specificity of RNA binding by applying RNA competition assays . Therefore , the UV cross-linking assays were performed in the presence of increasing concentrations of homologous or heterologous unlabeled RNA probes as competitors ( Figure 5B ) . In contrast to what was previously reported for the native DLA2 protein , His-DLA2 exhibited no specificity for the psbA RNA ( [17] , Figure 5B ) . This is probably due to a lack of other components of the DLA2–RNP complex that facilitate recognition of the psbA 5′ UTR by DLA2 in vivo . Such a behavior would resemble other chloroplast RBPs , like RBP40 from C . reinhardtii , which—as an isolated protein—unspecifically recognizes any RNA . Only as part of a complex with its cognate binding partner Nac2 does RBP40 have binding specificity to the psbD mRNA [37] . However , even though the protein appeared to bind nonspecifically to the psbA mRNA in vitro , we determined its equilibrium binding constant ( KD ) for the psbA 5′ UTR to put DLA2 on a comparable basis with other known RBPs ( Figure 5C ) . To this end , we applied an independent filter binding assay according to Ostersetzer et al . [38] , which does not involve a UV cross-linking step . The obtained KD value of ≈51 nM is in line with those determined for other RBPs ( for examples , see [38]–[40] ) . To further elucidate the role of DLA2 in psbA expression and acetyl-CoA production , DLA2–RNAi lines were generated [41] . Of ca . 800 transformants , ca . 100 clones survived the selection procedure . Three of these lines , namely iDLA2-1 , -2 , and -3 , exhibited the highest reductions of DLA2 level , which were determined to be 5% ( ±2 ) , 10% ( ±4 ) , and 6% ( ±2 ) % , respectively , relative to the DLA2 level in the recipient strain transformed with the empty vector . These strains maintained the DLA2 deficiency for over 2 years under selection for the RNAi-induced phenotype . To explore the phenotypic consequences of DLA2 deficiency in these RNAi lines , their growth rates were monitored under photoautotrophic , mixotrophic , and heterotrophic conditions ( Figure 6 ) . Growth of RNAi lines was only slightly affected under photoautotrophic and not at all affected under heterotrophic conditions . However , under mixotrophic conditions , severe growth retardation was observed for the DLA2 knock-down strains as compared to the wild-type . Therefore , DLA2 seems to be required for wild-type growth rate specifically under the condition associated with the formation of an RNP complex ( Figure 4 ) . To determine whether DLA2 is required for psbA expression and accumulation of PSII , D2 protein levels were measured in the DLA2–RNAi lines by immunoblot analyses . It was previously shown in C . reinhardtii that D1 and D2 accumulate in a 1∶1 stoichiometry , whereas unassembled proteins are rapidly degraded [42]–[44] . Therefore , D2 can be used as a proxy measure of D1 accumulation . Interestingly , we observed a light- and acetate-dependent accumulation of PSII in the DLA2–RNAi lines ( Figure 7A ) . In contrast , under photoautotrophic growth conditions , PSII accumulation was higher in the RNAi lines as compared to the wild-type . In contrast , in lines cultured mixotrophically under the same light conditions or heterotrophically in the dark , RNAi-mediated DLA2 deficiency leads to slightly reduced PSII levels . Other protein complexes in the chloroplast were not affected in the RNAi lines , as judged by parallel monitoring of steady-state levels of the large subunit of Rubisco and cytochrome b6 ( Figure 7A ) . To test whether altered psbA mRNA levels or D1 protein synthesis rates are responsible for the observed changes of the levels of D1 protein and , consequently , PSII accumulation , we performed Northern blots as well as 35S protein pulse labeling assays ( Figure S5 , Figure 7B ) . No alterations in psbA transcript levels were observed in iDLA2 lines as compared to the wild-type from cells grown under photoautotrophic , mixotrophic , or heterotrophic growth conditions ( Figure S5 ) . However , whereas only minor effects on D1 protein accumulation were observed , D1 synthesis rates were clearly altered in the iDLA2 lines relative to in the wild-type strain ( Figure 7 ) . For example , higher D1 protein synthesis rates in the iDLA2 lines were detected under phototautotrophic conditions , whereas D1 synthesis rates were reduced under mixo- and heterotrophic growth conditions , respectively ( Figure 7B ) . The less obvious effect of DLA2 deficiency on D1 protein accumulation under mixotrophic and heterotrophic conditions as compared to D1 protein synthesis might be explained by posttranslational stabilization effects counteracting a reduced translation rate . As another means of testing whether DLA2 plays a role in psbA translation , we asked whether it is associated with a “chloroplast translation membrane” ( CTM ) . This membrane subfraction was identified as a privileged location of translation in the C . reinhardtii chloroplast . CTMs are characterized by their higher density as compared to thylakoids and their association with the translation machinery including the translational regulator RBP40 [45] . Accordingly , membranes from wild-type cells from mixotrophic conditions were separated on the basis of density by floatation from a 2 . 5 M sucrose cushion into a 0 . 5–2 . 2 M sucrose gradient during isopycnic ultracentrifugation . Most thylakoid membranes were detected in the intermediate-density fractions 3 and 4 , as revealed by their high chlorophyll concentrations ( Figure 8A , left panel ) . Progressively less thylakoid membrane was present in fractions with increasing density ( fractions 5–8 ) . The CTM marker RBP40 was detected in fractions 3–8 , as reported previously [45] . Immunodetection of DLA2 revealed that its concentrations were highest in lanes 3–6 and progressively decreased in the denser fractions 7–8 . The presence of DLA2 in fractions 3 and 4 was inconclusive because this is consistent with DLA2 association with thylakoid membranes , CTM , or both membrane types . However , DLA2's sustained high levels in lanes 5 and 6 ( i . e . , in fractions in which thylakoid membrane levels progressively decrease ) indicate that DLA2 is associated with a nonthylakoid membrane such as the CTM in these fractions ( Figure 8A , left panel ) . That the RBP40 distribution extends to the bottom of the gradient ( lanes 7 and 8 ) indicates that the DLA2 is not associated with the densest CTMs to detectable extent . To determine whether the DLA2 that cofractionates with CTM ( lanes 5 and 6 ) in membranes from mixotrophically cultured cells ( Figure 8A , left panel ) could be relevant to its role in psbA translation , we carried out the same analysis on membranes from cells cultured photoautotrophically , which lack the DLA2 RNP complex ( Figure 4A ) . In this experiment , DLA2 more closely co-fractionated with the thylakoid membranes ( Figure 8A , right panel , fractions 3–5 ) . The density distribution of CTM was not dramatically altered relative to mixotrophically cultured cells , based on the distribution of RBP40 . Together , these results substantiate the proposed acetate-dependent role of DLA2 in psbA translation . It was proposed previously that DLA2 binds the psbA mRNA to localize it for translation and cotranslational membrane insertion of D1 [17] . To test this possibility , we asked whether DLA2 is required to localize the psbA mRNA to a specific , spatially defined , “translation zone” ( T-zone ) in the C . reinhardtii chloroplast , which is believed to be a privileged subcompartment for protein synthesis and to contain the CTM [16] , [20] . The T-zone was originally defined by results of confocal microscopy , which revealed in the outer perimeter of the pyrenoid by the colocalization of the psbA mRNA , chloroplast ribosome subunits , and the PSII translation factor RBP40 ( RB38 ) [20] . The pyrenoid serves as a cytological landmark for the T-zone because it is large ( 1–2 µm in diameter ) and located in the same position in every chloroplast ( Figure 8B ) . To determine whether DLA2 localizes the psbA mRNA to the T-zone , we first asked whether this localization is altered in the most severe RNAi line: iDLA2-1 . When we visualized the suborganellar distribution the psbA mRNA by fluorescence in situ hybridization ( FISH ) in the recipient strain used for RNAi , WT-NE , we found that the psbA mRNA was localized to the T-zone in 78% of cells ( Figure 8B ) . In the representative WT-NE cell , the green psbA FISH signal is concentrated in distinct regions adjacent to the pyrenoid , which is seen in the accompanying phase contrast image . These cells were from the mixotrophic growth conditions in which DLA2/psbA mRNA complex formation was detected ( Figure 4C ) . In contrast , when iDLA2-1 cells were examined , only 30% showed this psbA mRNA localization pattern , a greater than 2-fold reduction relative to WT-NE . In the representative iDLA2-1 cell , the psbA FISH signal is more dispersed and not concentrated in distinct regions adjacent to the pyrenoid ( Figure 8B ) . Together , these results reveal that DLA2 is involved in psbA mRNA localization to the T-zone and that the sustained psbA localization in 30% of these cells probably reflects the residual DLA2 in this RNAi line ( i . e . , 5% of the wild-type level , Figure 7A ) , the activity of a partially redundant DLA2-independent localization mechanism , or both . As another means of determining whether DLA2 localizes the psbA mRNA , we asked whether they colocalize in the T-zone . When the WT-NE cells were immunofluorescence ( IF ) stained for DLA2 , 65% showed the DLA2 IF signal near the pyrenoid ( Figure 8B ) . However , these analyses were hampered by the dispersed IF signal , reflecting the difficulty of specifically detecting the RNA-binding form of DLA2 amidst the “background” IF signal from DLA2 of cpPDC , which is most probably nonlocalized based on its detection as soluble and membrane-bound forms ( Figures 3A and S3 ) . To reveal a pool of DLA2 involved in psbA mRNA localization , amidst the nonlocalized signal , we used the program Colocalization Finder ( ImageJ ) to display signals of maximal intensity from both DLA2 and the psbA mRNA [20] . Our prediction was that the strongest signals from each should be colocalized for translation . As shown in Figure 8B , clusters of maximal colocalized signals from DLA2 and the psbA mRNA were seen in the T-zone in 60% of the cells analyzed ( and are labeled white in the right-hand most image ) . Moreover , this pattern requires DLA2 because parallel analyses of the most severe knock-down line , iDLA2-1 , revealed that only 8% of these cells showed the colocalization with DLA2 in the T-zone ( Figure 8B ) . Therefore , these in situ results and the biochemical evidence that DLA2 is associated with the CTM ( Figure 8A ) support our hypothesis that DLA2 is required for the localization of the psbA mRNA to the T-zone and , thereby , targets newly synthesized D1 protein to this PSII biogenesis center . cpPDC catalyzes the oxidative decarboxylation of pyruvate to acetyl-CoA for chloroplast fatty acid synthesis with the concomitant generation of NADH [9] , [11] . Thus , to test whether DLA2 is an active subunit of the cpPDC , enzyme activity assays were performed on protein extracts of photoautotrophically grown DLA2–RNAi lines by measuring the reduction of NAD+ . As shown in Figure 9 , cpPDC activity in extracts of the three DLA2–RNAi lines was reduced to approximately 15%–25% of the enzyme activity measured with the wild-type used for RNAi . The level of cpPDC activity thus correlated with the level of DLA2 protein in the RNAi lines ( compare Figure 7A and Figure 9 ) . We therefore conclude that DLA2 represents an active component of the chloroplast PDH complex . However , reduced enzyme activity did not lead to dramatic changes in overall lipid accumulation as assayed by thin layer chromatography ( Figure S6 ) . This might be explained by sufficiently high residual cpPDC levels in the investigated RNAi lines that still mediate efficient production of acetyl-CoA for fatty acid synthesis . In order to explore possible signals for the transition of the enzymatic form of DLA2 to its RNA-binding form for psbA translation , we asked whether cpPDC activity is altered by interaction of DLA2 with the psbA mRNA . This idea is supported by bioinformatical analysis of the DLA2 amino acid sequence . By using the RNABindR software to predict possible RNA-binding residues within the amino acid sequence of DLA2 [46] , we detected a putative RNA-binding region that overlaps the E3 subunit attachment site ( aa 191–216 , compare Figure 2 and Figure S9A ) . Hence , a competitive binding of either E3 or psbA mRNA to this DLA2 site might be involved in regulation . Since binding of E3 is required for cpPDC function , the lack of psbA mRNA should then lead to increased cpPDC activity . We therefore decided to analyze cpPDC activity in the absence of psbA mRNA . Data from higher plants have revealed an activation of the complex by light , due to an increase of stromal pH and Mg2+ concentration ( reviewed in [11] ) . A similar light-dependent activation of cpPDC could be confirmed for C . reinhardtii wild-type as indicated by an increase of activity of ca . 30% of light versus dark grown cells ( Figure 10A ) . This light activation depends on electron transport from PSII to PSI , as indicated by clearly reduced cpPDC activity upon treatment of cells with DCMU , a chemical that specifically blocks the electron flow from PSII ( Figure 10A ) . Next , we tested cpPDC activity in the PSII mutants FuD7 and nac2–26 specifically lacking either the psbA or the psbD mRNA , respectively . Consistent with a requirement for photosynthetic electron flow , both mutants exhibited reduced cpPDC activities as compared to the wild-type . However , in the FuD7 mutant , this reduction ( 66% of wild-type level ) was less pronounced than in nac2-26 ( 47% of wild-type level; Figure 10A ) . This might suggest that the psbA mRNA , which is still present in nac2–26 but lacking in FuD7 , has an additional inhibitory effect on cpPDC activity . To further test this , we measured the cpPDC activity in extracts of light grown wild-type and FuD7 cells after incubation with in vitro transcribed RNAs derived from the 5′ UTRs of the psbA or rbcL mRNAs ( Figure 10B ) . Whereas we detected only minor changes of cpPDC activity in the wild-type upon addition of exogenous psbA or rbcL RNAs , a dose-dependent reduction of enzyme activity was obtained for the FuD7 mutant after incubation with psbA , but not with rbcL RNA . Addition of 450 pmol psbA RNA thereby significantly ( p<0 . 05 ) reduced the activity of FuD7 cpPDC to approximately 48% of the activity measured in the wild-type . Therefore , the activity level measured for nac2–26 strongly resembles that measured for FuD7 in the presence of 450 pmol RNA , suggesting that the observed differences between nac2–26 and FuD7 are due to the absence of psbA mRNA in FuD7 ( Figure 10A , B ) . The marginal reduction of cpPDC activity upon addition of psbA mRNA to wild-type extracts might indicate a saturation of DLA2 with psbA mRNA under these conditions ( Figure 10B ) . Taken together , our data strongly suggest that cpPDC activity is specifically affected by the presence of psbA mRNA , presumably via its binding to the E3 attachment site of DLA2 . The dihydrolipoamide acetyltransferase component of the PDC represents an evolutionary and functionally conserved gene family from prokaryotes to eukaryotes [9] , [30] , [47] . But does this also hold for its ability to bind RNA and fulfill a possible role in gene expression ? To gain first insights , we predicted the possible RNA binding sites within the amino acid sequences of E2 subunits from phylogenetically distant organisms ( i . e . , human , yeast , and a cyanobacterium ) ( Figure S9 ) . Similar to C . reinhardtii DLA2 , all analyzed proteins revealed a putative RNA binding domain within the proposed E3 attachment site . Consequently , we tested recombinant E2 versions from these organisms for their ability to bind RNA in vitro . Therefore , E2 subunits were heterologously expressed in Escherichia coli as hexahistidine-tagged fusion proteins and purified on Ni-NTA Sepharose ( Figure 11A ) . Even though the recombinant proteins were markedly enriched after purification , especially the preparation of the E2 subunit from human ( Hs-E2 ) revealed contaminations with proteins in a size range of 30–45 kDa . These most likely represent Hs-E2 degradation products because they are specifically recognized by an anti-hexahistidine antibody ( unpublished data , Figure 11A ) . As controls for the subsequent RNA binding assay , we also included PratA , which is an unrelated his-tagged protein not described to possess any RNA binding activity [48] . As a positive control , we used the RNA binding protein RBP40 , which was already shown to unspecifically bind to RNA in vitro [49] , [50] . Since no specific RNA targets of the different E2 subunits are known , psbA 5′ UTR RNA was used to detect general binding activity ( Figure 11B ) . Interestingly , all tested E2 proteins showed a binding to the RNA probe applied in the UV cross-linking assay , suggesting that RNA binding is an intrinsic capacity of all E2 subunits , even of those from mitochondria and prokaryotes . However , the binding of the mitochondrial proteins , especially of the human protein , to the psbA RNA appeared to be weaker as compared to the recombinant proteins of the green lineage .
After mass spectrometrical identification of DLA2 , we confirmed its in silico predicted chloroplast localization ( Figure 3 ) . Results of enzyme activity assays of DLA2 knock-down lines demonstrated DLA2 to be a functional subunit of the cpPDC ( Figures 8 and 9 ) . SEC analyses revealed a psbA mRNA-specific DLA2 complex of ca . 700 kDa only in cells cultured mixotrophically ( i . e . , in the presence of exogenous acetate and light ) ( Figure 4 ) . In addition , a large >2 . 8 MDa , RNase-insensitive , DLA2 complex is likely to represent the functional cpPDC ( Figures 4 and S3 ) . This complex accumulated predominantly in cells cultured in photoautotrophic conditions and only to a lower level in mixotrophically or heterotrophically grown cells . Which of these complexes is formed , therefore , seems to be dependent on whether light or acetate is available as an energy source . Under photoautotrophic conditions , C . reinhardtii chloroplasts depend on the production of acetyl-CoA via cpPDC . Therefore , it is likely that most of DLA2 is in complexes of >2 . 8 MDa . In contrast , under mixotrophic conditions acetate can be converted into acetyl-CoA by acetyl-CoA synthetase ( ACS ) and/or by the acetate kinase ( ACK ) /phosphate acetyltransferase ( PAT ) system [10] . Accumulating acetyl-CoA then signals substrate availability for fatty acid synthesis , which might cause a product-inhibition of the cpPDC and/or acetylation of its subunits [2] , [3] . This , we propose , leads to its partial disassembly , thereby stimulating the light-dependent binding of the psbA mRNA to DLA2 and , consequently , the formation of the smaller DLA2 RNP complex ( Figure 12 ) . How this light-regulation might be exerted on DLA2 remains to be shown . The alternate function of DLA2 reported here may be conserved in higher plants . Several hints suggest that cpPDC subunits are present in complexes other than fully assembled cpPDC in higher plants . The analysis of HMW complexes from A . thaliana chloroplasts by SEC revealed the chloroplast E2 subunit ( LTA2 , At3g25860 ) as a component of two chloroplast complexes , one of >5 MDa and another one of 1–2 MDa [52] . The former most likely represents a fully assembled cpPDC , whereas the latter did not cofractionate with the E1 and E3 subunits . Instead , most E2 coeluted with ribosomal proteins and RNA binding proteins . Likewise , the A . thaliana chloroplast E2 subunit has been reported to co-immunoprecipitate with CSP41b , a protein involved in chloroplast RNA metabolism [53]–[55] . Further indications for E2 subunits possessing an additional function besides being a cpPDC subunit are derived from A . thaliana T-DNA insertion lines . Assuming that all three cpPDC enzyme components are necessary for its functionality , it is surprising that T-DNA insertions in E2 genes lead to embryo lethality in the homozygous state , whereas the E3 subunit is dispensable , thereby indicating an essential function of E2 distinct from its role in the cpPDC [56]–[58] . Similarly , an extensive PCR-based search for a DLA2 insertion mutant in an indexed library of C . reinhardtii failed , suggesting that DLA2 also is essential in green algae ( Bohne , Grossman , and Nickelsen , unpublished results ) . Consistent with a functional role of the DLA2–RNP complex , the DLA2–RNAi lines exhibited growth retardation and reduced D1 synthesis rates under mixotrophic conditions ( Figures 6 and 7 ) . Based on results reported here and previously , we propose a role of DLA2 in localizing the psbA mRNA to the T-zone for its translation and the synthesis of D1 in de novo PSII biogenesis . First , the existence of mRNA localization factors in the T-zone was suggested by a previous demonstration that the psbA mRNA localizes there in the absence of the nascent D1 polypeptide and , hence , any localization information in the polypeptide sequence [16] . Second , we show here that DLA2 is associated with membranes that are distinct from thylakoid membranes in density and within the density range of CTM , a biogenic membrane that is probably localized in the T-zone ( Figure 8A , left image , [45] ) . DLA2 is also associated with stromal thylakoid membranes , another proposed location of psbA translation [17] . Third , DLA2 functions in the localization of the psbA mRNA in the T-zone ( Figure 8C ) . Together , these results support a model in which the RNA-binding form of DLA2 tethers the psbA mRNA to CTM in the T-Zone for translation and cotranslational membrane insertion of D1 for its incorporation into assembling PSII complexes ( Figure 12 ) . While the phenotypes of the DLA2–RNAi lines suggest a role in psbA translation under mixotrophic conditions , the effects of DLA2 silencing under photoautotrophic conditions appear more pleiotropic . Although PSII levels increase as compared to the wild-type ( Figure 7 ) , growth rates remained unaltered despite the requirements for efficient photosynthesis under these conditions ( Figure 6 ) . This might be due to counteracting growth limitations caused by reduced cpPDC activity in the absence of acetate . Moreover , this raises the question: What is the molecular basis of the observed differences in the rates of PSII subunit synthesis in the DLA2–RNAi lines versus the wild-type in photoautotrophic and heterotrophic conditions ( i . e . , in the absence of DLA2–RNP complex formation ) ? Evaluation of the protein pulse-labeling data in Figure 7B revealed that D2 synthesis also increased and decreased concomitantly with D1 synthesis in DLA2–RNAi lines under photoautotrophic and heterotrophic growth , again suggesting more pleiotropic effects ( Figure S8 ) . In contrast under mixotrophic conditions when DLA2–RNP complex formation occurs , only D1 synthesis but not D2 synthesis was affected , confirming a psbA-specific function of DLA2 in the presence of light and acetate ( Figure S8 ) . The translational regulation of psbA mRNA is complex and has been shown to involve many factors in addition to DLA2 [59] . Nevertheless , based on the presented data , DLA2 appears to be specifically required for adjusting chloroplast gene expression when high levels of reduced carbon energy sources are available to the cell ( Figure 12 ) . Little is known about the regulation of cpPDC activity in green algae . DLA2 is constitutively expressed in C . reinhardtii under different light conditions , and thus , it is likely that its enzymatic activity is regulated at the posttranslational level ( Figure S7 ) . One mechanism to regulate cpPDC activity has been attributed to product inhibition by NADH and acetyl-CoA [11] . Additionally , cpPDC activity is stimulated in the light by an increase of stromal pH as well as Mg2+ concentration due to enhanced photosynthetic electron flow ( Figure 10A , reviewed in [11] ) . However , the prediction of a putative RNA binding site overlapping the E3 attachment site within the DLA2 primary amino acid sequence raised the possibility of competitive binding of either E3 or the psbA mRNA to this region and , therefore , a further level of cpPDC activity regulation . This idea is supported by the observation that cpPDC activity in the psbA deletion mutant FuD7 is higher than in the PSII mutant nac2–26 ( Figure 10 ) . This is likely due to the lacking inhibitory effect of the psbA mRNA in FuD7 , which can be recovered by the addition of exogenous psbA 5′ UTR RNA ( Figure 10A , B ) . This inhibitory effect of psbA RNA on cpPDC activity in FuD7 might suggest a continuous assembly and disassembly of the PDH complex , as it is difficult to imagine how otherwise psbA should be able to access the DLA2 subunit , which was reported to be located in the inner core of the enzymatic complex [60] . Disassembled DLA2 could bind RNA and might then be inhibited from re-association with other PDC subunits . Alternatively , the structure of the cpPDC might be different from those known from mitochondrial or bacterial PDCs . This is supported by sedimentation analyses that show that cpPDC forms less defined complexes as compared to mitochondrial or bacterial PDCs [61] . It was speculated that this is due to dissociation of one or more of the component enzymes or association with other enzymes [61] . Taken together , the data indicate a reciprocal light and acetate-dependent regulation of D1 and cpPDC activity , which might guarantee a coordination of protein and lipid synthesis in the course of thylakoid membrane biogenesis . A linkage between carbon and protein metabolism is additionally supported by earlier studies that revealed that chloroplast protein synthesis , including that of D1 , can be induced in cells in the dark phase of a diurnal regime after addition of acetate to the medium [62] . Our data demonstrate that recombinant E2 proteins from C . reinhardtii , human , yeast , and a cyanobacterium all have an intrinsic unspecific RNA binding ability in vitro ( Figures 5 and 11 ) . This might point to a very ancient feature of dihydrolipoamide acetyltransferases as RBPs , which is likely to have occurred even before the separation of mitochondrial and chloroplastic homologues . It should be noted , however , that the binding of the human E2 subunit to the provided plant-specific mRNA was very weak as compared to the other tested E2 subunits , and the significance of its RNA binding capacity requires further investigations . Further indications that RNA binding by E2 subunits is conserved are provided by the human mitochondrial E2 subunit of the mtPDC ( Hs-E2 ) . Only recently Hs-E2 has been found to interact with the transcription factor STAT5 ( Signal Transducer and Activator of Transcription 5 ) , which was demonstrated to bind mitochondrial DNA [63] . Furthermore , Hs-E2 and STAT5 are translocated to the nucleus under certain conditions , where Hs-E2 is thought to function as a co-activator in STAT5-dependent nuclear gene expression [64] . This raises the possibility that regulation of gene expression by E2 subunits occurs in diverse phylogenetic contexts . How the RNA binding ability might have been acquired is uncertain , but one might speculate that this occurred before the complex composition of PDC evolved . The ancient E2 enzyme might have had the capacity to bind NAD+ , a role that is now performed by the E3 subunit of PDC to reoxidize E2-bound dihydrolipoamide produced during catalysis . This is indicated by prediction of a typical NAD+/FAD+ binding site known as the Rossmann fold within the E3 binding region of many E2 subunits including C . reinhardtii DLA2 ( Figure S10 , [65] ) . Interestingly , the Rossmann fold , which is typically described as a functional NAD+ binding domain , is highly conserved among dehydrogenases and has been reported to be involved in binding of RNA molecules by metabolic enzymes [5] , [66] . One well-studied example is the human glyceraldehyde-3-phosphate dehydrogenase ( GAPDH ) , which is described to specifically bind to AU-rich elements within the 3′ UTRs of several mRNAs via its Rossmann fold domain [67] , [68] . A wide variety of physiological functions has been attributed to this enzyme–RNA interaction including regulation of mRNA stability , degradation , and/or translation; tRNA and mRNA transport; and as an RNA chaperone ( reviewed in [4] , [5] ) . As reported in this study for DLA2 , Nagy et al . [67] observed a decreased enzymatic activity of GAPDH upon RNA binding in vitro , which also might indicate a reciprocal regulation between metabolic enzyme activity and RNA-binding ( Figure 10B ) . In conclusion , the available data suggest a complex and coordinated interplay between metabolic pathways and gene expression in the chloroplast .
As the wild-type C . reinhardtii strain , we used CC-406 , which has a defective cell wall to allow chloroplast isolation . The chloroplast PSII mutants were FuD7 , in which the psbA genes have been deleted [34] , and a nac2 mutant , which carries nac2–26 [35] and lacks a stability factor for the psbD mRNA . Strains were maintained on 0 . 8% agar-solidified Tris/acetate/phosphate ( TAP ) medium [69] at 25°C under constant light ( 30 µE/m2/s ) if not indicated otherwise . Liquid cultures were grown under agitation at 25°C to a density of ∼2×106 cells/mL in TAP medium containing 1% sorbitol ( TAPS medium ) for mixotrophic and heterotrophic growth . Photoautotrophic growth was in high-salt minimal ( HSM ) medium [69] . Light conditions were as indicated: moderate light ( 30 µE/m−2/s−1 ) , high light ( 200 µE/m2/s ) , or darkness . For GFP import studies and the generation of RNAi lines , the UVM4 expression strain described by Neupert et al . [70] was used . Chloroplasts from cell wall–deficient strains carrying the cw15 mutation were isolated from a discontinuous Percoll gradient ( 45% to 75% ) as described previously [71] . To remove stromal proteins , isolated chloroplasts were osmotically lysed in hypotonic buffer ( 10 mM tricine/KOH , pH 7 . 8 , 10 mM EDTA , and 5 mM 2-mercaptoethanol ) by repeated pipetting . Membrane material was pelleted by ultracentrifugation for 30 min at 100 , 000 g through a 1 M sucrose cushion in hypotonic buffer in a SW40 rotor ( Beckman ) . Pellets were solubilized in equal volumes of 1 . 8 M sucrose in hypotonic buffer and floated using an upper 1 . 3 M sucrose cushion and an additional hypotonic layer ( 180 min , 4°C , 100 , 000 g ) . Thylakoid membranes were taken from the interphase , resuspended in twice the volume of hypotonic buffer , and pelleted again ( 30 min , 4°C , 100 , 000 g ) . The pellet was then lysed in Brij buffer ( 20 mM tricine/KOH , pH 7 . 8 , 0 . 12 mM KCl , 0 . 4 mM EDTA , 10 mM β-mercaptoethanol , 1% Brij-35 ) . After centrifugation ( 60 min , 4°C , 100 , 000 g ) an aliquot of the resulting supernatant containing solubilized proteins was then applied to a 5 mL heparin–Sepharose 4B ( GE Healthcare ) column equilibrated with Buffer I ( 50 mM KCl , 10 mM tricine/KOH , pH 7 . 8 , 10 mM EDTA , and 5 mM 2-mercaptoethanol ) . Bound proteins were eluted using a discontinuous salt gradient ( 150 mM , 500 mM , and 1 M KCl in buffer I ) after washing the sample with 4 CV of buffer I . Proteins eluting at 150 mM KCl were desalted using Amicon Ultra centrifugal filtration devices ( Millipore ) with a 10 kDa molecular mass cutoff according to the manufacturer's instructions . The protein solution ( in Buffer I ) was then applied to a 2 mL poly ( A ) –Sepharose 4B ( GE Healthcare ) column equilibrated with buffer I . The column was washed with 4 volumes of buffer I , and bound proteins were eluted with a discontinuous salt gradient ( 150 mM , 500 mM , and 1 M KCl in buffer I ) . Prior to use in UV cross-linking assays , all protein fractions were dialyzed against RNA binding buffer ( 30 mM Tris-HCl , pH 7 . 0 , 50 mM KCl , 5 mM MgCl2 , and 5 mM 2-mercaptoethanol ) . Protein concentrations were determined using the Bradford assay ( Bio-Rad ) . For mass spectrometric peptide sequencing , RBP63-containing gel pieces were treated with trypsin ( Promega ) , and the resulting peptides were analyzed on a Q-TOF2 mass spectrometer ( Micromass ) as described [72] . A fusion protein containing glutathione-S-transferase ( GST ) and the C-terminal region of the DLA2 protein was used as an antigen for production of a polyclonal rabbit antiserum . For generation of the plasmid expression vector , a DNA fragment encoding the amino acids 391–488 of DLA2 was amplified from genomic DNA by PCR with the primers FWD63BamHI ( 5′-GGATCCGACCTGGTCAAGCGCGCTCG-3′ ) and REV63SalI ( 5′-GTCGACGTTCTCAATCACAGCCTTGA-3′ ) . Attached restriction sites are underlined . The fragment was inserted into the expression vector pGEX4T1 ( GE Healthcare ) via the BamHI and SalI restriction sites . Overexpression and purification of the DLA2–GST fusion protein in the E . coli strain BL21 were performed according to the manufacturer's protocol using glutathione–Sepharose 4B ( GE Healthcare ) . A polyclonal antiserum was produced by immunizing rabbits with this protein fraction ( Biogenes ) . For analysis of HMW complexes , chloroplasts isolated from cw15 strains according to Zerges and Rochaix [71] were lysed in nonreducing hypotonic solution ( 10 mM EDTA , 10 mM tricine-KOH , pH 7 . 8 , and Roche Complete Mini protease inhibitors ) . Crude thylakoid membranes were separated from soluble proteins by centrifugation on a 1 M sucrose cushion ( 100 , 000× g , 30 min , 4°C ) . The supernatant of this centrifugation step was defined as stromal proteins . Pellets were solubilized in equal volumes of lysis buffer ( 120 mM KCl , 0 . 4 mM EDTA , 0 . 1% Triton X-100 , 20 mM tricine , pH 7 . 8 ) , and insoluble particles were removed by an additional sucrose cushion step . If RNase treatment was required , samples were incubated with 400 U RNase One ( Promega ) /mg protein for 60 min at 4°C before application to the gel filtration column . Gel filtration samples were loaded through an online filter onto a Superose 6 10/300 GL column ( GE Healthcare ) , and elution was performed at 4°C with buffer containing 50 mM KCl , 2 . 5 mM EDTA , 5 mM ε-aminocaproic acid , 0 . 1% Triton X-100 , and 20 mM tricine-KOH , pH 7 . 8 , at a flow of 0 . 3 mL/min using an ÄKTApurifier 10 system ( GE Healthcare ) . Aliquots of each elution fraction were subjected to immunoblotting . For protein isolation , cells were placed into 20 mL of liquid TAPS medium and grown under indicated light conditions on a rotary shaker ( 125 rpm ) to mid-log phase ( ∼5×106–1×107 cells/mL ) . Cells were harvested by centrifugation and lysed under pipetting in a buffer containing 20 mM KCl , 20 mM tricine , pH 7 . 8 , 0 . 4 mM EDTA , 5 mM β-mercaptoethanol , and 1% Triton X-100 . For cell fractionation , chloroplasts were isolated as described above . Mitochondria were basically prepared according to [73] . Immunoblot analysis was performed using standard procedures . The procedures for subcellular fractionation experiments shown in Figure 8A were described previously [45] . Protein concentrations were determined using the BCA ( Pierce ) or the Bradford ( C . Roth ) assay following the manufacturer's instructions . A codon-adapted CrGFP has been integrated as NdeI/EcoRI into the PsaD expression vector [74] , [75] . This expression cassette was then inserted into the pBC1 vector as XhoI/XbaI , which contains the APHVIII resistance gene under control of the constitutively active HSP70/RBCS2 promoter regions [76] , [77] to result in pBC1-CrGFP . For DLA2–GFP import studies , the coding sequence of the N-terminal amino acids 1–114 including the predicted transit peptide and the complete lipoyl attachment site was PCR-amplified from genomic DNA with the primer pair RBP63 fw ( 5′-AACATATGCAGGCCACGACCCG-3′ ) /RBP63 rv ( 5′-AACATATGCTCGTTGGCGTTTTCGGCCAC-3′ ) , introducing 5′ and 3′ NdeI sites ( restriction sites underlined ) . The NdeI fragment was then inserted into pBC1–CrGFP to result in pBC1–TP–DLA2–CrGFP . This construct was transformed into UVM4 , and positive transformants were selected on TAP plates supplemented with 10 µg/mL paromomycin . As a control for cytosolic CrGFP expression , the pBC1–CrGFP vector was directly transformed into UVM4 . GFP fluorescence of transformed cells was observed with a confocal laser scanning system ( Zeiss LSM 51 Meta ) . For expression of recombinant DLA2 protein from C . reinhardtii , a 1 , 389 bp fragment encoding the C-terminal amino acids 32–494 was PCR-amplified from EST clone MXL069g06 ( Kazusa DNA Research Institute ) using the primer pair RBP63–pQEfwBamHI ( 5′-aaggatccAACGCGGTCAAGGATG-3′ ) /RBP63-pQErevSalI ( 5′- gtcgacTTAGAACAGCAGCTGGTCGG-3′ ) and inserted into the plasmid pQE30 ( Qiagen ) via the BamHI/SalI restriction sites to yield the plasmid pQE–DLA2 . Expression was accomplished in E . coli M15 cells ( Stratagene ) . Cells were grown to an OD600 of 0 . 5–0 . 6 , and protein expression induced by addition of IPTG to a final concentration of 1 mM followed by growth at 37°C for 3 h . The recombinant protein was purified according to the GE Healthcare protocol for purification of histidine-tagged recombinant proteins under native conditions using Ni Sepharose 6 Fast Flow ( GE Healthcare ) . Primers used for cloning of other His-E2 fusion proteins excluding predicted signal and transit peptides are as follows: Synechocystis sp . 6803 Fw ( 5′-aaggatccATTTACGACATTTTCATGCC-3′ ) /Rev ( 5′-gtcgacGTCAAAGACTGGGCATTC-3′ ) , Saccharomyces cerevisiae Fw ( 5′-ggatccCCAGAGCACACCATTATTG-3′ ) /Rev ( 5′-gtcgacTCACAATAGCATTTCCAAAGG-3′ ) , and Homo sapiens Fw ( 5′-ggatccCCGCATCAGAAGGTTCCATTG-3′ ) /Rev ( 5′-gtcgacAGTGTGACCTGGGAGAGTTTA-3′ ) . For cloning of the expression vector for cyanobacterial PratA ( slr2048 ) , the CDS ( aa 38–383 ) was amplified using the following primers: Ss_PratA_fw01 ( 5′-ctaggatccAATCTTCCTGACGTTACCC-3′ ) /Ss_PratA_rv01 ( 5′-ctactgcagTTAGAGATTATCCAGCTTTTCTTGG-3′ ) and cloned via BamHI/PstI sites into the vector pET28b SUMO-Ser . The full length CDS of RBP40 was amplified with primers BamHI–RBP40 ( 5′-aaggatccATGCTGACCTTGAGACGTGC-3′ ) /RB38DN44revSalI ( 5′-ttgtcgacCTAGTAGCGGGCGCCC-3′ ) and cloned into the vector pQE30 via BamHI/SalI sites . Expression and purification were as described above with some minor changes: Expression for Hs-E2 was performed overnight at 17°C ( 0 . 5 mM IPTG ) , Syn-E2 for 5 h at 25°C , Sc-E2 for 5 h at 18°C , and RBP40 for 5 h at 30°C . PratA was expressed in BL21 DE3 cells overnight at 12°C . Concentrations of recombinant proteins were determined along with a BSA dilution series . In vitro synthesis of RNA and UV cross-linking experiments were basically performed as described by Zerges and Rochaix [71] . DNA templates for the in vitro synthesis of rbcL , psbD , and psbA leader RNA probes were generated by PCR using the following primers: T7rbcL5 ( 5′-gtaatacgactcactatagggTATGCTCGACTGATAAGAC-3′ ) /rbcL3 ( 5′-CTGCTTTAGTTTCTGTTTGTGGAACC-3′ ) ; T7psbD5 ( 5′-gtaatacgactcactatagggCCACAATGATTAAAATTAAA-3′ ) /psbDUTR3 ( 5′-ACCGATCGCAATTGTCAT-3′ ) ; and T7psbA5 ( 5′-gtaatacgactcactatagggTACCATGCTTTTAATAGAAG-3′ ) /2054-psbA ( 5′-GATCCATGG TCATATGTTAATTTTTTTAAAG-3′ ) . Each template contained the promoter of the T7 RNA polymerase ( written in lowercase letters in the fw-primer sequence ) fused to the 5′ end of the described fragments . A total of 0 . 5 µg of the PCR products were transcribed in vitro by T7 RNA Polymerase ( Fermentas ) in a 20 µl reaction in the presence of 20 U RNase inhibitor ( Fermentas ) , 40 µCi of α-32P-UTP ( 3 , 000 Ci/mmol; Hartmann Analytic ) , 30 µM nonradiolabeled UTP , and 0 . 5 mM each of ATP , CTP , and GTP according to the manufacturer's protocol . We added 1 U of RNase-free DNase ( Promega ) , and the reaction was incubated for an additional 15 min at 37°C . Unincorporated nucleotides were removed using a MicroSpin S-200 HR column ( GE Healthcare ) . The reactions were extracted once with phenol-chloroform and ethanol precipitated in the presence of ammonium . Binding reactions ( 20 µl ) were performed at RT for 5 min and contained 20 mM HEPES/KOH , pH 7 . 8 , 5 mM MgCl2 , 60 mM KCl , and 200 ng or 10 ng protein . Each reaction contained 50–100 kcpm of 32P-RNA probe . For competition experiments , protein and RNA probe were used in equimolar amounts or in 5-fold to 200-fold excess of cold RNA . Radiolabeled RNA and nonlabeled competitors were mixed prior to the addition of proteins in competition experiments . Quantification of competitor RNAs was performed by measuring the incorporation of low levels of radioactivity into transcripts . Subsequent exposure to a 254 nm UV irradiation of 1 J/cm2 using a Stratalinker UV cross-linker ( Stratagene ) covalently cross-linked the RNA probe and bound proteins . After irradiation , the nonbound 32P-RNA probes were digested by treatment with 10 U RNase One ( Promega ) for 20 min at 37°C . Samples were fractionated by SDS-PAGE and analyzed by autoradiography or phosphorimaging . The KD was determined as described by Ostersetzer et al . [38] . Increasing amounts of recombinant DLA2 protein were incubated for 15 min at RT with in vitro transcribed 32P-labeled psbA mRNA ( 6 . 7 pM ) in 30 µl reactions in the same binding buffer used for UV cross-linking assays . Subsequently , the reactions were filtered through stacked nitrocellulose ( Reprobe nitrocellulose plus , 0 . 45 µm; Applichem ) and nylon membranes ( Nylon plus , 0 . 45 µm; C . Roth ) using a dot blot apparatus ( Minifold SRC96 , Schleicher & Schuell ) . The membranes were washed once with 100 µl of binding buffer , dried , and subjected to phosphorimaging and quantitation with AlphaEase software ( Alpha Innotech Corporation ) . To create DLA2-deficient mutants of C . reinhardtii , we used the RNAi system previously described by Rohr et al . [41] . For the generation of an inverted repeat construct specific for DLA2 RNA , a 400 bp fragment corresponding to the last exon and part of the 3′ UTR of DLA2 was amplified by PCR using genomic DNA as a template with the primers 5/63 SHE 5′-GTCGACAAGCTTGAATTCCAACTGGGCTCA-3′ and 3/63 400 B 5′-GGATCCGCTAACCCTGCAGCCCACCT-3′ , which add SalI , HindIII , EcoRI , or BamHI restriction sites , respectively ( restriction sites underlined ) . A longer 600 bp fragment containing an additional 200 bp of the 3′ UTR that functioned as a spacer for the inverted repeat was amplified using the primers 5/63 SHE and 3/63 600 B ( BamHI ) 5′-GGATCCGGCATTCAAGCCACCCTGCT-3′ . These two fragments were ligated and cloned as an inverted repeat ( with the central spacer ) into the EcoRI site of the vector NE537 [41] . The UVM4 strain was transformed with the resulting construct , kept for 2 d in liquid culture ( TAP +1 . 5 mM L-tryptophan ) in dim light , and then plated on TAP plates containing 5 µg/mL paromomycin and 1 . 5 mM L-tryptophan [70] . C . reinhardtii cells transformed with the empty NE537 vector served as a control ( WT-NE ) . At intervals of 2 wk , colonies were transferred to TAP plates containing 5 and then 10 µM 5-fluoroindole ( 5-FI ) . Plates were kept in low light ( ∼10 µE/m2/s−1 ) under a yellow foil ( Q-MAX 010 medium yellow , Multi-Lite , Hamburg , Germany ) , which filters out wavelengths of light between 400 and 470 nm to inhibit photodegradation of tryptophan and 5-FI . FISH and IF were performed according to Uniacke et al . [78] . The psbA FISH probes were labeled with Alexa Flour 488 , and the IF staining involved Alexa Fluor 568 conjugated anti-rabbit secondary antibody ( Invitrogen ) . Images were captured on a Leica DMI6000B microscope ( Leica Microsystems ) using a 40×/0 . 75 objective , a Hamamatsu OrcaR2 camera , and Volocity acquisition software ( Perkin Elmer ) . For each condition , ≥20 cells were observed . Chlamydomonas liquid cultures were grown in TAPS or HSM medium to a density of ∼1–2×106 cells/mL , pelleted by centrifugation ( 10 min , 4°C , 1 , 000× g ) , resuspended in the same medium in which all sulfur-containing ingredients were replaced by the respective chloride salts ( TAPS-S/HSM-S ) , and incubated for 16 h at 23°C in the light . Cells were pelleted , washed , and resuspended in TAPS-S/-T or HSM-S/-T ( lacking both sulfur salts and trace elements ) , respectively , and grown under indicated light conditions for 2 h . Cells were then washed again and resuspended in TAPS-S/-T or HSM-S/-T to a concentration of 80 µg chlorophyll per mL . Aliquots ( 225 µl ) of the cell suspension were incubated with cycloheximide ( 10 µg/mL ) for 10 min . Subsequently , 100 µCi H235SO4 ( Hartmann Analytic ) was added to each , followed by incubation for 15 min in the light as before . After centrifugation , sedimented cells were frozen in liquid nitrogen . Cells were resuspended in 10 mM HEPES-KOH , pH 7 . 5 , 10 mM EDTA in the presence of CompleteMini protease inhibitors ( Roche ) and disrupted by sonication ( 30 s , RT ) . The homogenate was then centrifuged at 20 , 000 g for 30 min . The pellet was resuspended in 10 mM HEPES-KOH , pH 7 . 5 , 10 mM EDTA . Samples were fractionated by electrophoresis on 16% SDS-polyacrylamide gels containing 8 M urea . Radioactive protein signals were detected on the dried gel by phosphorimaging . Significance of difference between the mean D1/AtpA/B signal ratio for each RNAi line and WT-NE ( 100% ) was determined with a one-sample t test ( p<0 . 05 ) . Cells were lysed in a buffer containing 25 mM MgCl2 , 100 mM tricine ( pH 8 . 0 ) , and 0 . 1% Triton X-100 at 4°C by sonication . Insoluble material was removed by centrifugation ( 10 min , 4°C , 10 , 000 g ) . cpPDC activity was measured photometrically at 23°C using a Pharmacia Biotech Ultrospec 3000 spectrophotometer . The assay was based on that described by Qi et al . [79] and performed under conditions that favor the activity of the cpPDC over mtPDC ( high Mg2+ concentration , alkaline pH; [61] , [80] ) . The reaction mixture used contained 0 . 1 mM TPP , 5 mM MgCl2 , 2 mM NAD+ , 0 . 1 mM CoA , 3 mM cysteine , 0 . 05% Triton X-100 , 0 . 1 M tricine ( pH 8 . 0 ) , and 200 µg of proteins in a final volume of 0 . 990 mL . Reactions were initiated by the addition of 1 µmol sodium pyruvate in a volume of 10 µl , and the change in absorbance at 340 nm caused by NADH production was followed for 2 min . If the influence of DCMU was tested , it was added to a final concentration of 20 µM 3 h prior to cell harvest . An equivalent volume of ethanol , the solvent for the DCMU stock , was used instead of DCMU for control strains . For pre-incubation of cell lysates with psbA or rbcL mRNAs , 5′ UTRs were in vitro transcribed as described above . Lysates were prepared as described above in the presence of 250 U/mL RNase inhibitor ( Fermentas ) . Protein samples ( 200 µg ) were incubated with 150 or 450 pmol RNA in a final volume of 130 µl for 10 min at RT prior to the PDC activity assay . Unpaired two-sample t tests were used to determine whether the mean activity under each condition is significantly different from that in the absence of RNA ( p<0 . 05 ) . If cpPDC activity tests were performed with size exclusion fractions , SEC elution was performed in 60 mM KCl , 5 mM MgCl2 , and 100 mM tricine-KOH , pH 7 . 8 , and 0 . 05% . We used 300 µl of 1 . 2 ml fractions for the assay as described above . | Metabolic control of gene expression coordinates the levels of specific gene products to meet cellular demand for their activities . This control can be exerted by metabolites acting as regulatory signals on a class of metabolic enzymes with dual functions as regulators of gene expression . However , little is known about how metabolic signals affect the balance between enzymatic and regulatory roles of these proteins . Here , we report an example of a protein with dual functions in gene expression and carbon metabolism . The chloroplast pyruvate dehydrogenase complex is well-known to produce activated di-carbon precursors for fatty acid , which is required for lipid synthesis . Our results show that a subunit of this enzyme forms ribonucleoprotein particles and influences chloroplast mRNA translation . Conversely , RNA binding affects pyruvate dehydrogenase ( metabolic ) activity . These findings offer insight into how intracellular metabolic signaling and gene expression are reciprocally regulated during membrane biogenesis . In addition , our results suggest that these dual roles of the protein might exist in evolutionary distant organisms ranging from cyanobacteria to humans . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biochemistry",
"plant",
"biochemistry",
"plant",
"science",
"model",
"organisms",
"plant",
"biology",
"plant",
"cell",
"biology",
"botany",
"biology",
"molecular",
"cell",
"biology"
] | 2013 | Reciprocal Regulation of Protein Synthesis and Carbon Metabolism for Thylakoid Membrane Biogenesis |
A fundamental step in the evolution of the visual system is the gene duplication of visual opsins and differentiation between the duplicates in absorption spectra and expression pattern in the retina . However , our understanding of the mechanism of expression differentiation is far behind that of spectral tuning of opsins . Zebrafish ( Danio rerio ) have two red-sensitive cone opsin genes , LWS-1 and LWS-2 . These genes are arrayed in a tail-to-head manner , in this order , and are both expressed in the long member of double cones ( LDCs ) in the retina . Expression of the longer-wave sensitive LWS-1 occurs later in development and is thus confined to the peripheral , especially ventral-nasal region of the adult retina , whereas expression of LWS-2 occurs earlier and is confined to the central region of the adult retina , shifted slightly to the dorsal-temporal region . In this study , we employed a transgenic reporter assay using fluorescent proteins and P1-artificial chromosome ( PAC ) clones encompassing the two genes and identified a 0 . 6-kb “LWS-activating region” ( LAR ) upstream of LWS-1 , which regulates expression of both genes . Under the 2 . 6-kb flanking upstream region containing the LAR , the expression pattern of LWS-1 was recapitulated by the fluorescent reporter . On the other hand , when LAR was directly conjugated to the LWS-2 upstream region , the reporter was expressed in the LDCs but also across the entire outer nuclear layer . Deletion of LAR from the PAC clones drastically lowered the reporter expression of the two genes . These results suggest that LAR regulates both LWS-1 and LWS-2 by enhancing their expression and that interaction of LAR with the promoters is competitive between the two genes in a developmentally restricted manner . Sharing a regulatory region between duplicated genes could be a general way to facilitate the expression differentiation in duplicated visual opsins .
Gene duplication is a fundamental step in evolution [1] . Most often , one of the resulting daughter genes simply becomes a pseudogene and may be eventually lost from the genome due to functional redundancy between the duplicates and reduction of selective constraint to maintain its function . However , observation of another fate for duplicated genes , such as acquisition of a new function ( neofunctionalization ) or subdivision of parental gene function between daughter genes ( subfunctionalization ) , implies an evolutionary advantage by the process [2] . Subfunctionalization often involves differentiation of expression pattern between daughter genes and has been a subject of intense scrutiny to understand the regulatory mechanism to achieve the differentiation [3]–[5] . In vertebrates , color vision is enabled by the presence of multiple classes of cone visual cells in the retina , each of which has a different absorption spectrum . The absorption spectrum of a visual cell is mainly determined by the visual pigment it contains . A visual pigment consists of a protein moiety , visual opsin , and a photo-sensing chromophore , either 11-cis retinal or 11-cis 3 , 4-dehydroretinal [6] . The five types of visual opsins found among extant vertebrates are RH1 ( rod opsin or rhodopsin ) and four types of cone opsins: RH2 ( RH1-like , or green ) , SWS1 ( short wavelength-sensitive type 1 , or ultraviolet-blue ) , SWS2 ( short wavelength-sensitive type 2 , or blue ) and M/LWS ( middle to long wavelength-sensitive , or red-green ) [7] . The SWS2 and M/LWS type genes are closely located on the same chromosome [8]–[10] and could represent the most ancient gene duplication in vertebrate visual opsin genes , from which other types could have arisen through whole-genome duplications and subsequent gene losses in early vertebrate evolution [7] , [11]–[13] . Thus , visual opsin genes represent an excellent case of gene duplication to study the mechanism of neofunctionalization ( in absorption spectrum ) and subfunctionalization ( in expression pattern ) . While the spectral tuning mechanism of visual opsins has been intensively studied [14]–[18] , the regulatory mechanism of their expression differentiation , especially that of cone opsins , has been less explored . Among vertebrates , fish are known to possess a rich and varied repertoire of visual opsins , including two or more opsin subtypes within the five types by further gene duplications [19]–[21] , presumably reflecting their evolutionary adaptation to diverse aquatic light environments [22] . In fish , the eyes continue to grow throughout their lifetime by adding new cells to the peripheral zones , such that the peripheral cells are developmentally younger than central cells [23] , [24] . Thus , in the fish retina the timing of gene expression is partly reflected in the region of expression in the retina . All visual opsin genes have been isolated and characterized for zebrafish ( Danio rerio ) [25] , medaka ( Oryzias latipes ) [26] and cichlids ( Family Cichlidae ) [27]–[30] . Among them , the expression pattern of visual opsin genes has been best documented for zebrafish . Zebrafish have nine visual opsin genes consisting of two M/LWS ( red ) , four RH2 ( green ) , and single-copy SWS1 ( UV ) , SWS2 ( blue ) and RH1 ( rod ) opsin genes [25] . The red , green , UV and blue opsin genes are expressed in the long-member of double cones ( LDCs ) , the short-member of double cones ( SDCs ) , the short single cones ( SSCs ) and the long single cones ( LSCs ) , respectively , which are arranged in a regular mosaic pattern in the retina [31] , [32] . The two red opsin genes , LWS-1 and LWS-2 , are arrayed in a tail to head manner , in this order , and encode photopigments with wavelengths of maximal absorption ( λmax ) at 558 and 548 nm , respectively [25] . The four green opsin genes , RH2-1 , RH2-2 , RH2-3 and RH2-4 , are also arrayed in a tail to head manner , in this order , and encode photopigments with λmax at 467 , 476 , 488 , and 505 nm , respectively [25] . In both red and green opsins , expression of longer-wave subtypes occurs later in development and is confined to the peripheral , especially ventral-nasal region of the adult retina , whereas expression of shorter-wave subtypes occurs earlier and is confined to the central region of the adult retina , shifted slightly to the dorsal-temporal region [33] . It remains largely unknown how subtypes of an opsin class are directed to express in different regions of the retina while keeping the cell type identical between them . Thus , the zebrafish visual opsins are an excellent model to study the regulatory mechanism of not only cell-type specific expression of opsin types , but also developmental-stage ( and thus retinal-region ) specific expression of opsin subtypes . With the feasibility to employ transgenic technology , cis-regulatory regions relevant to the cell-type specific expression of opsin types have been elucidated using a living color reporter such as the green fluorescent protein ( GFP ) for zebrafish single-copy opsin genes ( i . e . , rod opsin [34]–[37] , UV opsin [38] , [39] and blue opsin genes [40] ) . A regulatory region relevant to not only cell-type specific but also retinal-region specific expression of opsin subtypes has also been reported for the zebrafish green opsin genes [41] . In the present study , we focus on the zebrafish red opsin genes , LWS-1 and LWS-2 , and report a cis-regulatory region , “LWS-activating region” ( LAR ) , which is relevant to their expression differentiation .
In the two PAC clones we obtained [LWS-PAC ( E ) and LWS-PAC ( H ) ] , the first exons of LWS-1 and LWS-2 were replaced after their initiation codons with DNA segments encoding green and red fluorescent proteins ( GFP and RFP ) , respectively ( Figure 1A ) . The modified clones were designated LWS1/GFP-LWS2/RFP-PAC ( E ) and LWS1/GFP-LWS2/RFP-PAC ( H ) , respectively . One transgenic zebrafish line was established using each construct: Tg ( LWS1/GFP-LWS2/RFP-PAC ( E ) ) #1229 and Tg ( LWS1/GFP-LWS2/RFP-PAC ( H ) ) #430 . Adult fish of both lines expressed RFP in the central-dorsal-temporal region of the retina and GFP in the peripheral-ventral-nasal region of the retina circumscribing the RFP region ( Figure 1B ) . It was also confirmed that the expression was specific to the LDCs , which were immunostained by an antibody against the zebrafish red opsin ( Figure 1C ) . Thus , in both transgenic lines , the expression of GFP and RFP reporter genes recapitulated the expression of LWS-1 and LWS-2 , respectively , demonstrating that both the LWS-PAC ( E ) and LWS-PAC ( H ) clones contain sufficient regulatory region ( s ) for the proper expression of the two red opsin genes . Next , we used only the intergenic region between the stop codon of SWS2 and the initiation codon of LWS-1 , designated LWS1up2 . 6kb , and the region between the stop codon of LWS-1 and the initiation codon of LWS-2 , designated LWS2up1 . 8kb . We created a double-reporter construct consisting of the LWS1up2 . 6kb , GFP reporter , LWS2up1 . 8kb and RFP reporter , in this order ( LWS1up2 . 6kb:GFP-LWS2up1 . 8kb:RFP , Figure 2A ) , and obtained three transgenic lines: Tg ( LWS1up2 . 6kb:GFP-LWS2up1 . 8kb:RFP ) #1464 , 1631 , and 1640 . In two of the three lines , #1631 ( Figure 2B , 2C ) and #1640 , the GFP and the RFP recapitulated the expression patterns of LWS-1 and LWS-2 , respectively . In the third line , #1464 , the expression of RFP was weaker and sparser but still confined to the central region of the retina and the expression of GFP appeared to be identical to the other two lines ( Figure S1 ) . The expression pattern of LWS-1 was also recapitulated when only the LWS1up2 . 6kb was used with GFP ( LWS1up2 . 6kb:GFP ) in all three transgenic lines obtained: Tg ( LWS1up2 . 6kb:GFP ) #1508 , 1509 ( Figure 2A , 2D–2F ) , and 1515 . On the other hand , when only the LWS2up1 . 8kb was used with GFP ( LWS2up1 . 8kb:GFP ) , no GFP signal was observed in the transgenic line obtained: Tg ( LWS2up1 . 8kb:GFP ) #1433 . These results suggest that the LWS1up2 . 6kb contains a regulatory region not only for LWS-1 , but also for LWS-2 . In order to search for the regulatory region from the LWS1up2 . 6kb , we employed the transient transgenic assay in which the regulatory activity of a GFP-reporter construct was evaluated in the fish injected with the construct . This was done by examining the incidence of fish bearing GFP-expressing eyes at a larval stage . As in previous studies [39]–[41] , the expression level of GFP was graded into four categories , +++ , ++ , + , − , at 7 days post-fertilization ( dpf ) ( Figure 3 ) . First , we used a whole PAC clone , LWS-PAC ( E ) , and modified it to LWS1/GFP-PAC ( E ) and LWS2/GFP-PAC ( E ) , in which the first exon of LWS-1 and LWS-2 , respectively , was replaced after its initiation codon with GFP-encoding DNA ( Figure 3A left ) . We confirmed that the GFP expression pattern from the two PAC constructs was consistent with the expression patterns of LWS-1 and LWS-2 , respectively , at the larval stage ( Figure 3A right ) ( i . e . , LWS-2 is expressed predominantly and LWS-1 is expressed only faintly in the retina [33] ) . Next , as shown in Figure 3B left , we isolated from LWS2/GFP-PAC ( E ) a series of DNA regions consisting basically of the LWS2up1 . 8kb-GFP-LWS-2 region and varying ranges of its upstream region . The GFP signal was apparent when the upstream region contained 1 . 3-kb or more upstream of LWS-1 , but was almost undetectable when it contained 0 . 6-kb or less upstream of LWS-1 or when only the LWS2up1 . 8kb-GFP-LWS-2 region was used ( Figure 3B right ) . This implies that the LWS2up1 . 8kb region does not contain a sufficient regulatory region for the expression of LWS-2 , consistent with the absence of a GFP signal in the transgenic line Tg ( LWS2up1 . 8kb:GFP ) described above . This also suggests that the regulatory region is located in the 0 . 6-kb region between 1 . 3-kb and 0 . 6-kb upstream of LWS-1 . To test if the 0 . 6-kb region plays a regulatory role by itself for the expression of LWS-2 , a coinjection protocol was employed using mixed concatamers of separate DNA fragments formed upon integration into the genome [42] . The LWS2up1 . 8kb-GFP-LWS-2 region was injected together with a variety of DNA segments from the LWS1up2 . 6kb region ( Figure 3C left ) . GFP expression was apparent in the retina only when the segment contained the 0 . 6-kb region ( Figure 3C right ) . We thus designated the 0 . 6-kb region as an “LWS-activating region” ( LAR ) . Using a DNA construct consisting of LAR and LWS2up1 . 8kb:GFP , designated LAR:LWS2up1 . 8kb:GFP ( Figure 4A ) , we obtained five transgenic lines: Tg ( LAR:LWS2up1 . 8kb:GFP ) #1481 , 1491 , 1496 , 1499 , and 1501 . In one line , #1499 , a GFP signal was observed specifically in the LDCs but across the entire outer nuclear layer , not confined to the central-temporal-dorsal region ( Figure 4B–4D ) . The absence of retinal region specificity is in sharp contrast to the case in which the double reporter construct , LWS1up2 . 6kb:GFP-LWS2up1 . 8kb:RFP , was used ( Figure 2B , 2C ) . This suggests that the relative position of the LAR to the gene is relevant to the regional specificity of the retina . In another line of Tg ( LAR:LWS2up1 . 8kb:GFP ) #1501 , the GFP signal also appeared throughout the retina , but was sparser ( Figure S2A ) . At a finer level , the signal appeared not only in LDCs but also weakly in some bipolar cells ( Figure S2B ) . In the other three Tg ( LAR:LWS2up1 . 8kb:GFP ) lines , the GFP signal was not detectable . This instability of the reporter signal among the transgenic lines could be attributed not only to the general effect of their insertion sites in the genome , but also to the dependency of LAR to work cooperatively with its adjacent regions in the LWS1up2 . 6kb and LWS2up1 . 8kb . Consistently , as in the transient transgenic assay shown in Figure 3B , GFP expression level was much higher when the entire LWS1up2 . 6kb region was used than when only the proximal 1 . 3 kb region was used . To examine if the LDC-specificity of the GFP expression was attributed to LAR itself , we tested the 564-bp adjacent upstream region of a non-retinal keratin 8 gene [43] , designated krt8up564bp [41] . The krt8up564bp induces gene expression specifically in the epithelial tissues , but not in the retina , and has been used for enhancer trapping as a basal promoter [44] . When krt8up564bp was conjugated to the LAR and GFP reporter ( LAR:krt8up564bp:GFP ) , no GFP expression was observed in the retina of the two transgenic lines obtained: Tg ( LAR:krt8up564bp:GFP ) #1469 and 1477 . This is in sharp contrast to the case in which krt8up564bp was conjugated to the RH2-LCR and GFP expression was observed in the SDCs throughout the zebrafish retina in our previous study [41] . This suggests that LAR itself is not capable of determining the cell-type specificity of gene expression , unlike RH2-LCR , but works as an enhancer which interacts with cell-type determining regions that should reside in both LWS1up2 . 6kb and LWS2up1 . 8kb . Next , we removed the LAR from LWS1/GFP-LWS2/RFP-PAC ( E ) and LWS1/GFP-LWS2/RFP-PAC ( H ) ( designated ΔLAR-LWS1/GFP-LWS2/RFP-PAC ( E ) and ΔLAR-LWS1/GFP-LWS2/RFP-PAC ( H ) , respectively ) ( Figure 5A ) . Two transgenic lines were found for each of the two constructs: Tg ( ΔLAR-LWS1/GFP-LWS2/RFP-PAC ( E ) ) #1143 ( Figure 5C ) and 1166 and Tg ( ΔLAR-LWS1/GFP-LWS2/RFP-PAC ( H ) ) #1107 ( Figure 5D , 5E ) and 1100 . All four of these transgenic lines showed a similar expression pattern of the reporters in the retina with the LAR-bearing Tg ( LWS1/GFP-LWS2/RFP-PAC ( E ) ) or Tg ( LWS1/GFP-LWS2/RFP-PAC ( H ) ) ( Figure 5B ) line . The reporter-expressing cells were confined to LDSs ( Figure 5E ) . The GFP and RFP signals were observed in the ventral and dorsal regions of the retina , respectively ( Figure 5B–5D ) . However , the fluorescent signal in each cell was lowered . The number of the reporter-expressing cells decreased and their spatial distribution was restricted to a narrow range in both of these regions ( Figure 5B–5E ) . These results support the deduced role of LAR as an enhancer but not as the cell-type determining factor from the experiments thus far . This experiment also provided the first direct evidence that LWS-1 expression is affected by LAR .
The present study identified a 0 . 6-kb regulatory region , named LAR , for the expression of the duplicated red opsin genes of zebrafish , LWS-1 and LWS-2 , in the upstream of the gene array . The LAR functions to enhance the LDC-specific expression of both genes but does not determine the cell-type specificity of the gene expression . The regulatory region for the cell-type specificity of the gene expression appears to reside in the 2 . 6-kb and 1 . 8-kb upstream regions of the two genes . The relative position of LAR to a gene is relevant to the retinal region specificity of the expression of the gene . In the primate L/M opsin genes , the locus control region ( LCR ) is located at ∼3 . 5-kb upstream of the gene array and is necessary for the expression of both L and M opsin genes [45] , [46] . Although there is no clear overall similarity between the zebrafish LAR and primate L/M opsin LCR , LAR contains two OTX ( A/GGATTA ) and one OTX-like ( TGATTA ) sequences ( Figure S3 ) which are also present in the primate L/M opsin LCR [47] , [48] . These sequences , or their reverse complement sequences , are the binding sites of the cone-rod homeobox ( Crx ) protein , a member of the Otx family of the paired-like homeodomain proteins and a key trans-acting regulatory factor responsible for the gene expression in the retina and pineal organ [47] , [48] . The mammalian Crx is produced predominantly in both the retinal photoreceptors and pineal cells and regulates expression of retinal photoreceptor-specific genes and of pineal-specific genes [47]–[50] . In zebrafish , Otx5 , a paralog of Crx , is produced in the retina and pineal organ and regulates genes that show circadian expression in the pineal organ [51] . The OTX or OTX-like sequences have also been found in the upstream region of the zebrafish SWS2 [40] and in the RH2-LCR [41] . Thus , the LAR could be orthologous to the primate L/M opsin LCR and also be paralogous to the SWS2 regulatory region and the RH2-LCR ( see ref . [12] for a similar discussion ) . The primate L/M opsin LCR interacts with only the most proximal or the second proximal gene of the array , often L and M opsin genes respectively , through their proximal promoters [46] , [52] , [53] . The choice of the promoters by the LCR is largely a stochastic process [54] , [55] . These characteristics enable the mutually exclusive expression of the L and M opsin genes and nearly a random distribution of the L and M cone photoreceptor cells in the primate retina . In zebrafish , the expression of LWS-1 and LWS-2 is also nearly mutually exclusive in the retina [33] . Unlike the primate L/M opsin system , however , the expression of the two zebrafish red opsin genes is temporally and spatially organized and not random in the retina [33] . Whereas expression of LWS-2 is first observed at 40 hours post-fertilization ( hpf ) and spread throughout the retina by 72 hpf , initial expression of LWS-1 is observed at 3 . 5–5 . 5 days post-fertilization ( dpf ) in the marginal side of the ventral retina [33] . In sexually mature adults , LWS-2 is expressed in the central-dorsal-temporal region of the retina . Expression of LWS-1 is complementary to the LWS-2 observed in the peripheral-ventral-nasal region of the rest of the retina , although cells at the boundary of the two fields appear to express both gene subtypes and LWS-1 is sparsely expressed in the LWS-2 zone [33] . In this study , the spatially restricted patterns of gene expression were recapitulated by fluorescent reporters for both LWS-1 and LWS-2 in the adult retina of Tg ( LWS1up2 . 6kb:GFP-LWS2up1 . 8kb:RFP ) ( Figure 2B , 2C ) . The expression pattern of LWS-1 was also recapitulated in Tg ( LWS1up2 . 6kb:GFP ) ( Figure 2D , 2E ) , whereas that of LWS-2 was not , and GFP was expressed throughout the adult retina in Tg ( LAR:LWS2up1 . 8kb:GFP ) ( Figure 4B , 4C ) . This suggests that the LWS1up2 . 6kb contains a region susceptive to a developmental control that represses gene expression in the early stage or activates it in the later stage in LDCs , while the LWS2up1 . 8kb allows LDC-specific expression throughout development with the aid of LAR . This also suggests that LAR , which is shared by LWS-1 and LWS-2 , interacts with the LWS-2 promoter during the time LWS-1 expression is repressed ( or not activated ) in the early stage and then interacts with the LWS-1 promoter once the LWS-1 expression is enabled . This preference in interaction of LAR for LWS-1 over LWS-2 could be attributed to the closer distance of LAR to LWS-1 , as in the case of the primate L/M opsin LCR [46] , [52] , [53] and the zebrafish RH2-LCR [41] . Sharing a regulatory region among duplicated genes is a common feature among the zebrafish M/LWS ( red ) and RH2 ( green ) and the primate M/LWS ( L and M ) opsin genes . This system should be advantageous in facilitating differential ( i . e . , mutually exclusive ) expression of duplicated opsin genes by using the regulatory region in a competitive manner between the duplicated genes . If the competition is largely stochastic , an intermingled pattern of photoreceptor cells expressing different daughter genes can be expected in the retina , as in the case of primate L/M opsin genes . The trichromatic color vision is enabled by this stochastic-type system in primates . If the competition is developmentally controlled , for example , so that the regulatory region interacts with a proximal gene in an early stage and shifts the interaction target to a distal gene , the proximal gene would be expressed in the central region and the distal gene in the peripheral region of the retina as in the case of the zebrafish green opsin genes . In the case of the zebrafish red opsin genes , the interaction would start with the distal gene and switch to the proximal gene . In fish , such a control is feasible because the retina continues to grow throughout their lifetime by adding new cells to the peripheral zone [24] . Expression of different opsin genes among different retinal regions results in sights with varying wavelength sensitivity as a function of visual angles , which could be advantageous in the aquatic light environment where wavelength composition differs depending on directions [56] . This could explain why many examples of gene duplication have been found in fish visual opsin genes . Further studies of the regulatory mechanism of differential expression of fish visual opsin genes should contribute to our understanding of the adaptive significance of gene duplications in general .
All animal protocols were approved by the University of Tokyo animal care and use committee . Through the screening service of the Resource Center Primary Database ( RZPD , Germany; https://www . rzpd . de ) of a zebrafish PAC library ( no . 706 , originally created by C . Amemiya ) , two clone DNAs ( BUSMP706E19271Q9 and BUSMP706H1397Q9 ) , designated LWS-PAC ( E ) and LWS-PAC ( H ) , were obtained using the LWS-2 cDNA as a probe . Both clones encompass SWS2 , LWS-1 and LWS-2 in their ∼80-kb and ∼110-kb inserts , respectively ( Figure 1A ) . Sequencing both ends of the inserts revealed that the nucleotide sequences of LWS-PAC ( E ) and LWS-PAC ( H ) correspond to the nucleotide positions 25222648–25311454 and 25174505–25295119 of chromosome 11 in the Ensembl zebrafish assembly version 8 , respectively ( http://www . ensembl . org/Danio_rerio/Info/Index ) . The I-SceI meganuclease system [57] was used for efficient transgenesis of the PAC-derived constructs . Two I-SceI recognition sites ( 5′-TAGGGATAACAGGGTAAT-3′ ) were introduced into the vector backbone of the LWS-PAC clones as follows . The ampicillin-resistance ( Ampr ) gene was PCR-amplified from the pUC18 plasmid using primers harboring the I-SceI recognition site at their 5′ ends to create the I-SceI-Ampr-I-SceI segment ( see “I-SceI-Ampr-I-SceI” in Table S1 for primers ) . The I-SceI-Ampr-I-SceI segment was inserted into the EcoRV site of pBluescript II ( SK- ) plasmid vector ( Stratagene , Tokyo ) . The I-SceI-Ampr-I-SceI segment was isolated from the pBluescript clone using primers harboring the flanking sequences of the kanamycin-resistance ( Kmr ) gene site of the LWS-PAC clones to create the I-SceI-Ampr-I-SceI cassette ( see “Kmr<>I-SceI-Ampr-I-SceI” in Table S1 for primers ) . The Kmr of the LWS-PAC clones was replaced with the I-SceI-Ampr-I-SceI cassette by the site-specific homologous recombination system coupled with drug selection using the E . coli strain EL250 [58] as in our previous study [41] . The first exon after the initiation codon of LWS-1 and LWS-2 in the LWS-PAC clones was replaced with the GFP or the RFP gene as follows . The chloramphenicol acetyl transferase ( CAT ) and the Kmr gene fragments were PCR-amplified from pBR328 and pCYPAC6 plasmids , respectively ( see “CAT” and “Kmr” in Table S1 for primers ) . The CAT gene was inserted into the pEGFP-1 plasmid vector ( BD Biosciences Clontech , Tokyo ) at the AflII site , which is located immediately downstream of the SV40 polyadenylation signal ( polyA ) , linked downstream of the GFP coding sequence to create the GFP-polyA-CAT segment . Similarly , the Kmr gene was inserted into the pDsRed-1 or pDsRed-Express-1 plasmid vector ( BD Biosciences Clontech , Tokyo ) at the AflII site to create the RFP-polyA-Kmr segment . For the LWS1/GFP-LWS2/RFP-PAC ( H ) , the pDsRed-1 was used . For the other RFP-containing constructs ( LWS1/GFP-LWS2/RFP-PAC ( E ) , ΔLAR-LWS1/GFP-LWS2/RFP-PAC ( E ) and ΔLAR-LWS1/GFP-LWS2/RFP-PAC ( H ) ) , the pDsRed-Express-1 was used . The GFP-polyA-CAT segment was isolated from the pEGFP-1 construct by PCR using primers harboring the flanking sequences of the exon 1 of LWS-1 or LWS-2 to create the GFP-polyA-CAT cassette ( see “LWS-1<>GFP-polyA-CAT” and “LWS-2<>GFP-polyA-CAT” in Table S1 for primers ) . The RFP-polyA-Kmr segment was isolated from the pDsRed-1 or the pDsRed-Express-1 construct by PCR using primers harboring the flanking sequences of the exon 1 of LWS-2 to create the RFP-polyA-Kmr cassette ( see “LWS-2<>RFP-polyA-Kmr” in Table S1 for primers ) . These cassettes were replaced with the exon 1 of LWS-1 or LWS-2 in the LWS-PAC clones by the site-specific homologous recombination system in EL250 . The LAR was removed from the LWS-PAC clones by the site-specific homologous recombination system and by the flpe-FRT recombination system for excision of a DNA region sandwiched by FRT sequences in EL250 [41] , [58] as follows . The CAT gene was PCR-amplified from the pBR328 using primers harboring the FRT sequences to create the FRT-CAT-FRT segment ( see “FRT-CAT-FRT” in Table S1 for primers ) . The FRT-CAT-FRT segment was inserted into the EcoRV site of pBluescript II ( SK- ) plasmid . Then , the FRT-CAT-FRT segment was isolated by PCR using primers harboring the flanking sequences of the LAR to create the FRT-CAT-FRT cassette ( see “LAR<>FRT-CAT-FRT” in Table S1 for primers ) . The LAR was replaced with the FRT-CAT-FRT cassette in the LWS-PAC clones by the site-specific homologous recombination system in EL250 . The FRT-CAT-FRT cassette was then excised from the modified LWS-PAC clones in EL250 by the flpe-FRT recombination system for excision of a DNA region sandwiched by FRT sequences , leaving one FRT sequence in this region of the clones [41] , [58] . A plasmid construct , pT2AL200R150G [59] , was modified as follows . The pT2AL200R150G contains a GFP-expression cassette between XhoI and BglII sites surrounded respectively by the L200 and R150 minimum recognition sequences of the Tol2 transposase . The Tol2 transposase excises the DNA region between the recognition sequences from the plasmid and integrates it into the host genome as a single copy with the recognition sequences attached as in the plasmid [60] . The GFP-expression cassette contains a promoter sequence of a ubiquitously expressed gene ( the Xenopus elongation factor ( EF ) 1α ) , the rabbit β-globin intron , GFP gene and the SV40 polyA signal . The construct contains two Not I sites , one in the junction between the GFP gene and the SV40 polyA and another upstream of the L200 in the vector backbone . We first removed the Not I site in the vector backbone by eliminating a DNA segment between a Sac I site and L200 encompassing the NotI site . Next , we removed the promoter from the construct by replacing the region from the EF1α promoter to the GFP gene ( from XhoI to NotI sites ) with a DNA segment in the pEGFP-1 vector consisting of a part of the multiple cloning site ( MCS ) and the GFP gene ( from XhoI to NotI sites of pEGFP-1 ) . Finally , we replaced the SV40 polyA signal in the construct ( from NotI to BglII sites ) with the polyA signal sequence derived from the herpes simplex virus thymidine kinase ( HSV-TK ) , which was PCR-isolated from the pEGFP-1 vector using a forward primer harboring NotI site and a reverse primer harboring BglII site ( “HSV-TK-polyA” in Table S2 ) . This modified construct was designated pT2GFP-TKPA . The replacement of the polyA signal from SV40 to HSV-TK was done to facilitate , in a later stage , the insertion of a DNA fragment containing the SV40 polyA in an appropriate orientation into the pT2GFP-TKPA by avoiding a possible interaction between the two SV40 polyA sequences . Using the pT2GFP-TKPA as a transfer vector , we constructed the LWS1up2 . 6kb:GFP and the LWS2up1 . 8kb:GFP in Figure 2A as the followings . The region from LWS1up2 . 6kb to GFP in the LWS1/GFP-PAC ( E ) clone and the region from LWS2up1 . 8kb to GFP in the LWS2/GFP-PAC ( E ) clone were isolated by PCR using forward primers harboring a SalI site and reverse primers harboring a NotI site ( “LWS1up2 . 6kb:GFP” and “LWS2up1 . 8kb:GFP” in Table S2 ) . A DNA segment in the pT2GFP-TKPA from SalI in MCS to NotI in the junction between GFP and HSV-TK polyA was replaced with those segments isolated from the PAC constructs through restriction digestion and ligation at the SalI and NotI sites . In the resulting constructs , the region from LWS1up2 . 6kb to GFP and that from LWS2up1 . 8kb to GFP were connected to the HSV-TK polyA ( LWS1up2 . 6kb:GFP and the LWS2up1 . 8kb:GFP , respectively ) at the NotI site . The LWS1up2 . 6kb:GFP-LWS2up1 . 8kb:RFP ( Figure 2A ) was constructed as follows . The SV40 polyA in the pEGFP-1 vector was isolated together with a NotI site located at its 5′ side by PCR using a forward primer harboring a KpnI site and a reverse primer harboring a SalI site ( “SV40-polyA” in Table S2 ) . The isolated fragment was cloned into the pBluescript II ( SK- ) vector at KpnI and SalI sites . The region from the LWS2up1 . 8kb to the RFP gene including a NotI site located just downstream of the RFP gene in the LWS1/GFP-LWS2/RFP-PAC ( E ) clone was isolated with a SalI site attached to the 5′ end of the LWS2up1 . 8kb . The region was connected to the 3′ side of the SV40 polyA cloned in the pBluescript II ( SK- ) at SalI site . Then , from the pBluescript construct , the region consisting of the SV40 polyA , LWS2up1 . 8kb , and RFP gene ( from the Not I site at 5′ side of the SV40 polyA to the NotI site at 3′ side of the RFP gene ) was inserted into the LWS1up2 . 6kb:GFP construct in the pT2GFP-TKPA at the NotI site located between the GFP gene and the HSV-TK polyA . This results in the LWS1up2 . 6kb-GFP segment connected to 5′ side of the SV40 polyA and the LWS2up1 . 8kb-RFP segment connected to 5′ side of the HSV-TK polyA in the pT2GFP-TKPA ( LWS1up2 . 6kb:GFP-LWS2up1 . 8kb:RFP ) . The LAR:LWS2up1 . 8kb:GFP and the LAR:krt8up564bp:GFP ( see Figure 4A and Results section ) were constructed as follows . The LAR was isolated from the LWS-PAC ( E ) clone by PCR using a forward primer harboring a HindIII site and a reverse primer harboring an EcoRI site ( “LAR” in Table S2 ) and was inserted into the HindIII/EcoRI sites in the MCS of pT2GFP-TKPA . For making the LAR:LWS2up1 . 8kb:GFP , the GFP gene region in the LAR-inserted pT2GFP-TKPA construct ( from the SalI site in MCS to the NotI site at the 3′ side of the GFP gene ) was replaced with the region from the LWS2up1 . 8kb to the GFP gene in LWS2up1 . 8kb:GFP construct in the pT2GFP-TKPA ( from the SalI site at the 5′ side of LWS2up1 . 8kb to the NotI site at the 3′ side of the GFP ) by restriction digestion and ligation at the SalI and NotI sites . Similarly , for making the LAR:krt8up564bp:GFP , the GFP gene region in the LAR-inserted pT2GFP-TKPA was replaced with the region from the krt8up564bp to the GFP gene in LCR:krt8 construct reported in ref . [41] by restriction digestion and ligation at the SalI and NotI sites . A series of the GFP-reporter constructs and DNA fragments for the transient transgenic assay ( Figure 3B , 3C ) were obtained by PCR from LWS-2/GFP-PAC ( E ) using primers listed in Table S3 . These DNA fragments were purified through gel extraction before the microinjection . Zebrafish were maintained at 28 . 5°C in a 14-h light/10-h dark cycle as described by ref . [61] . The LWS-PAC derived constructs bearing the I-SceI recognition sequence were injected into the cytoplasm of embryos at the one-cell stage at 20 ng/µl with I-SceI meganuclease ( 0 . 5 units/µl ) ( New England Biolabs , Beverly , MA ) in a solution of 0 . 5× commercial meganuclease buffer with tetramethyl-rhodamin dextran tracer [57] . The reporter constructs in the pT2GFP-TKPA vector were resuspended at a final concentration of 25 ng/µl in 0 . 1 M KCl and tetramethyl-rhodamin dextran tracer . They were co-injected with mRNA of Tol2 transpsase of 27 ng/µl that was prepared through in vitro transcription from pCS-TP using the mMESSAGE mMACHINE kit ( Ambion , Austin , TX ) [59] , [60] . For generation of transgenic lines , the injected embryos were grown to sexual maturity and crossed with non-injected fish in a pair-wise fashion . Founders and fish of subsequent generations transmitting a reporter transgene were screened by PCR-based genotyping as described in ref . [41] . All the transgenic lines analyzed in this study are listed in Table S4 . The GFP-reporter constructs for the transient transgenic assay ( Figure 3B ) were microinjected with 0 . 1 M KCl and tetramethyl-rhodamin dextran at a final concentration of 25–50 ng/µl . The LWS2up1 . 8kb-GFP-LWS-2 region was injected together with a variety of DNA segments from the LWS1up2 . 6kb region ( Figure 3C ) at a final concentration of approximately 25–50 ng/µl each in 0 . 1 M KCl and tetramethyl-rhodamin dextran tracer . For transient transgenic assay of GFP expression , embryos injected with the GFP-expression constructs were grown in 0 . 003% 1-phenyl–2-thiourea after 12–24 hpf to disrupt pigment formation . One eye per injected embryo was examined at 7 dpf for GFP fluorescence under a dissecting fluorescent microscope . The eyes were scored as “+++” , “++” , “+” , and “−” when GFP was expressed in more than 50 cells , in 11–50 cells , in 1–10 cells , and in no cell per eye , respectively [39]–[41] . Immunostaining was carried out against adult retinal sections following the procedure of ref . [39] . An antibody against the zebrafish red opsin raised in rabbits [32] was used to stain LDCs . The Cy3-conjugated anti-rabbit IgG was used as a secondary antibody . Images of GFP , RFP and Cy3 fluorescence of the sections were captured using a Zeiss 510 laser-scanning confocal microscope ( Zeiss , Thornwood , NY ) . | Among vertebrates , fish may have the most advanced color vision . They have greatly varied repertoires of color sensors called visual opsins , possibly reflecting evolutionary adaptation to their diverse photic environments in water , and are an excellent model to study the evolution of vertebrate color vision . This is achieved by multiplying opsin genes and differentiating their absorption light spectra and expression patterns . However , little is understood regarding how the opsin genes are regulated to achieve the differential expression pattern . In this study , we focused on the duplicated red-sensitive opsin genes of zebrafish to tackle this problem . We discovered an “enhancer” region near the two red opsin genes that plays a crucial role in their differential expression pattern . Our results suggest that the two red opsin genes interact with the enhancer competitively in a developmentally restricted manner . Sharing a regulatory region could be a general way to facilitate the expression differentiation in duplicated visual opsin genes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genetics",
"and",
"genomics/gene",
"expression",
"developmental",
"biology/molecular",
"development"
] | 2010 | A Single Enhancer Regulating the Differential Expression of Duplicated Red-Sensitive Opsin Genes in Zebrafish |
The Cucumber mosaic virus ( CMV ) Y-satellite RNA ( Y-Sat ) has a small non-protein-coding RNA genome that induces yellowing symptoms in infected Nicotiana tabacum ( tobacco ) . How this RNA pathogen induces such symptoms has been a longstanding question . We show that the yellowing symptoms are a result of small interfering RNA ( siRNA ) -directed RNA silencing of the chlorophyll biosynthetic gene , CHLI . The CHLI mRNA contains a 22-nucleotide ( nt ) complementary sequence to the Y-Sat genome , and in Y-Sat-infected plants , CHLI expression is dramatically down-regulated . Small RNA sequencing and 5′ RACE analyses confirmed that this 22-nt sequence was targeted for mRNA cleavage by Y-Sat-derived siRNAs . Transformation of tobacco with a RNA interference ( RNAi ) vector targeting CHLI induced Y-Sat-like symptoms . In addition , the symptoms of Y-Sat infection can be completely prevented by transforming tobacco with a silencing-resistant variant of the CHLI gene . These results suggest that siRNA-directed silencing of CHLI is solely responsible for the Y-Sat-induced symptoms . Furthermore , we demonstrate that two Nicotiana species , which do not develop yellowing symptoms upon Y-Sat infection , contain a single nucleotide polymorphism within the siRNA-targeted CHLI sequence . This suggests that the previously observed species specificity of Y-Sat-induced symptoms is due to natural sequence variation in the CHLI gene , preventing CHLI silencing in species with a mismatch to the Y-Sat siRNA . Taken together , these findings provide the first demonstration of small RNA-mediated viral disease symptom production and offer an explanation of the species specificity of the viral disease .
Plant viruses are often accompanied by small parasitic RNAs termed satellite RNAs . Satellite RNAs range in size from ∼220 to 1400 nucleotides ( nt ) in length and depend on their associated viruses ( known as the helper virus ) for replication , encapsidation , movement and transmission , but share little or no sequence homology to the helper virus itself [1] . Most satellite RNAs do not encode functional proteins , yet can induce disease symptoms which range from chlorosis and necrosis , to total death of the infected plant [1] , [2] . How such non-protein-coding RNA pathogens induce disease symptoms has been a longstanding question . Early studies showed that the pathogenicity of a satellite RNA is determined at the nucleotide level , with one to several nucleotide changes dramatically altering both the virulence and host specificity of disease induction [2]–[6] . Subsequent studies demonstrated that satellite RNA replication is associated with the accumulation of high levels of satellite RNA-derived small interfering RNAs ( siRNA ) [7] . This class of small RNA ( sRNA ) has been shown to direct RNA silencing in plants through sequence-specific mRNA cleavage or DNA methylation [8] , [9] . Taken together , these findings led to the suggestion that pathogenic satellite-derived siRNAs might have sequence complementarity to a physiologically important host gene , and that the observed disease symptoms are in fact due to satellite siRNA-directed silencing of the targeted host gene [10] . However , to date , no such host gene has been identified , leaving the satellite RNA-induced disease mechanism unsolved . In this report we explore the sRNA-mediated disease mechanism using the Y-satellite of Cucumber mosaic virus ( CMV Y-Sat ) . The CMV Y-Sat consists of a 369-nt single-stranded RNA genome and induces distinct yellowing symptoms in a number of Nicotiana species including N . tabacum ( tobacco ) [11] . We show that the Y-Sat-induced yellowing symptoms result from Y-Sat siRNA-directed silencing of the host chlorophyll biosynthetic gene , CHLI . Furthermore , we demonstrate that Y-Sat-induced symptoms can be prevented by transforming tobacco with a silencing-resistant version of CHLI and provide evidence that the observed species specificity of Y-Sat-induced disease symptoms is due to natural sequence variation within the targeted region of the CHLI transcript .
The nucleotide sequence responsible for the yellowing symptoms of the CMV satellite disease has been mapped to a small 24-nt region of the Y-Sat genome ( nt . 177–200 and Figure 1A , referred to hereon as the ‘yellow domain’ ) [11]–[14] . Genetic analysis of progeny plants from a cross between the disease-susceptible species N . bigelovii and the disease-resistant species N . clevelandii , suggested that the yellowing symptoms induced upon CMV Y-Sat infection are associated with a single , incompletely dominant gene in the Nicotiana species [15] . We hypothesized that a siRNA derived from the yellow domain was directing RNA silencing of a tobacco gene , which in turn led to the expression of the observed yellowing symptoms . BLAST searches with the 24-nt yellow domain sequence against tobacco sequences in the NCBI database were performed to identify 21-nt or longer target sequences . No cDNA with perfect 21-nt complementarity was identified , however , accommodating weak G:U base pairings as a match allowed for the identification of a single N . tabacum sequence complementary to 22 nt of the Y-Sat yellow domain ( from nt . 178 to nt . 199; Figure 1A ) . The identified sequence is part of the coding region of the magnesium chelatase subunit CHLI , a 426-amino acid protein essential for chlorophyll biosynthesis ( Figure 1B; see Text S1 for sequences ) . Previous studies have shown that Y-Sat-induced yellowing symptoms are associated with reduced chlorophyll content [16] , making CHLI a strong target candidate for Y-Sat siRNA-directed silencing . Transformation of N . tabacum with a RNA interference ( RNAi ) vector targeting CHLI resulted in a dramatic decrease in CHLI expression and severe chlorosis of the transgenic plants , which ranged from yellowing to complete bleaching ( Figure 1C–D ) . The phenotypes expressed by RNAi plants parallels the appearance of Y-Sat-induced symptoms , and are consistent with CHLI silencing being responsible for the disease phenotype . Northern blot analysis confirmed that CHLI mRNA was dramatically down-regulated upon CMV Y-Sat infection ( Figure 1E ) . CHLI expression was not affected by infection with the CMV helper virus alone ( Figure S1 ) . The level of CHLI down-regulation in CMV Y-Sat-infected plants correlated strongly with the severity of chlorosis , and with the accumulation of yellow domain-specific siRNAs ( Figure 1E ) . As expected , CHLI down-regulation was associated with a dramatic reduction in chlorophyll content ( Figure 1F ) . 5′ rapid amplification of cDNA ends ( 5′ RACE ) on RNA samples extracted from CMV Y-Sat-infected and uninfected tobacco plants showed that the down-regulation of CHLI was due to siRNA-directed RNA cleavage at the predicted 22-nt target site . An expected 310-base pair ( bp ) product was amplified from the CMV Y-Sat-infected plants; however , no such product could be amplified from RNA extracts of uninfected tobacco ( Figure 1G ) . Sequencing of the 5′ RACE product showed that RNA cleavage occurred at three distinct positions within the 22-nt Y-Sat siRNA-targeted CHLI sequence ( Figure 1A ) . siRNA-directed cleavage of a target RNA normally occurs across nucleotides 10 and 11 of the siRNA [17] . The three cleavage sites detected by the 5′ RACE analyses therefore implied that the silencing of CHLI is directed by three individual Y-Sat siRNA species with their 5′-terminal nucleotides corresponding to nt . 178 , 180 and 181 of the Y-Sat RNA genome respectively ( Figure 1A ) . To confirm that these specific siRNAs were present , the total sRNA population of CMV Y-Sat-infected tobacco plants was sequenced using Solexa technology . Approximately 4 million sRNA reads were obtained , of which 1 million were of the 21–22 nt siRNA size class derived from the plus ( + ) strand of the Y-Sat RNA genome . From this ( + ) strand-specific pool , siRNAs corresponding to the yellow domain region form part of a siRNA hot spot along the Y-Sat genome ( Figure S2 ) . Furthermore , the 21–22-nt siRNAs starting at nt . 178 , 180 and 181 accumulated at relatively high frequencies , with 1576 , 2368 and 3352 reads detected respectively ( Figure 2 ) . The above results correlate with the proposal that the yellowing symptoms associated with CMV Y-Sat infection of tobacco are due to Y-Sat yellow domain-specific siRNA-directed silencing of the CHLI gene . To determine if CHLI silencing alone was responsible for the Y-Sat-induced symptoms , we introduced 10-nt silent mutations into the 22-nt complementary sequence of the wild-type CHLI gene ( wtCHLI ) , and transformed tobacco plants with the mutated version of the CHLI gene ( mtCHLI ) . The modified CHLI transgene contains 10 nucleotide changes within the 22-nt region complementary to Y-Sat yellow domain siRNAs , bringing nine mismatches to this region of complementarity , rending it resistant to cleavage by these Y-Sat siRNAs ( Figure 3A–B ) . Strikingly , the Y-Sat-induced symptoms were completely abolished in tobacco plants transformed with the mtCHLI constructs; none of the 44 independent transgenic lines developed yellowing symptoms upon CMV Y-Sat infection ( Figure 3C; Table S1 ) . This is in direct contrast to the population of tobacco plants transformed with the wtCHLI constructs , where 34 of 36 plant lines analysed showed yellowing symptoms upon CMV Y-Sat infection . The absence of symptoms in mtCHLI lines was not due to a lack of Y-Sat accumulation; mtCHLI plants grafted with diseased wtCHLI lines , either as the scion or rootstock , remained symptom free ( Figure 3D ) . Northern blot hybridization analysis confirmed that the Y-Sat RNA accumulated to high levels in both wtCHLI and mtCHLI plants ( Figure 3E ) . The CHLI transcript level was dramatically reduced in infected wtCHLI plants , but remained high in infected mtCHLI plants ( Figure 3E , compare lanes 5–6 with lanes 9–10 ) . These analyses indicated that the modified CHLI transcript was resistant to Y-Sat siRNA-directed silencing , allowing for sufficient accumulation of this modified transcript in infected plants for normal chlorophyll biosynthesis . These results also suggest that sRNA-directed silencing of CHLI alone is sufficient for the induction of the disease symptoms associated with CMV Y-Sat infection . Taken together , our findings strongly indicate that sRNA-mediated viral disease symptoms can be prevented through the introduction of a silencing-resistant version of a sRNA targeted host gene ( s ) . Not all Nicotiana species are susceptible to Y-Sat-induced yellowing symptoms [15] . This suggests that sequence variations might exist in the coding sequence of the CHLI gene in some Nicotiana species , rendering these species ‘resistant’ or free of Y-Sat siRNA-induced CHLI silencing . We sequenced the CHLI transcript from five different Nicotiana species , including three disease-susceptible ( N . tabacum , N . glutinosa and N . benthamiana ) and two resistant ( N . clevelandii and N . debneyi ) species ( Figure 4A ) . The three susceptible species which develop yellowing symptoms upon CMV Y-Sat infection had identical target sequences in the CHLI gene ( Figure 4B ) . In contrast , of the two disease-resistant species which do not develop yellowing symptoms upon infection , N . clevelandii expressed two CHLI mRNA species ( presumably because it is an allotetraploid containing two different chromosome pairs ) , with the predominant species containing an A to G change at the targeted CHLI sequence , converting tobacco's A:U base pairing to a weaker G:U wobble pair near the mRNA cleavage site ( Figure 4B ) . Similarly , N . debneyi expressed two CHLI mRNA species with the predominant form containing a G to U conversion compared with the tobacco sequence , changing a strong G:C pairing to a U-C mismatch . Northern blot hybridization of CMV Y-Sat-infected and uninfected plants of these five Nicotiana species showed that CHLI expression was strongly silenced in infected plants of the susceptible species ( Figure 4C ) . In contrast , CHLI mRNA levels remained high in infected N . clevelandii and N . debneyi plants suggesting that the mRNA variants expressed by these two symptomless species are resistant to Y-Sat siRNA-induced silencing . These results suggest that the previously observed host species specificity of satellite RNA-induced diseases [1] , [2] results from sequence variation within the viral sRNA-targeted host gene ( s ) . In addition , these results further demonstrate that the disease symptoms associated with CMV Y-Sat infection are solely due to silencing of the CHLI gene .
Our results demonstrate that the yellowing symptoms associated with CMV Y-Sat infection in tobacco result from Y-Sat siRNA-directed silencing of the chlorophyll biosynthetic gene CHLI . This finding is consistent with results from previous site-directed mutagenesis studies showing that Y-Sat mutants with nucleotide changes inside the yellow domain either retain or lose the ability to induce yellowing symptoms [13] , [14] . In agreement with siRNA-directed silencing of CHLI solely accounting for the yellowing symptoms associated with CMV Y-Sat infection , Y-Sat variants which are capable of inducing such symptoms have higher degrees of sequence complementarity to the CHLI target sequence than those that can no longer induce yellowing upon CMV Y-Sat infection ( 13 , 14; Figure S3 ) . As observed for the yellowing symptoms associated with CMV Y-Sat infection , several other disease symptoms induced by viral satellite RNAs have also been associated with a short sequence within the respective satellite RNA genomes . For instance , the chlorotic phenotypes induced by the B2 and WL3 satellite RNAs of CMV in tobacco and tomato are determined by a ∼26-nt region ( nt . 141–166 ) of the satellite genome [2] , [5] , [6] . Interestingly , BLAST searching with this sequence identifies a 21-nt match to a tobacco cDNA ( accession # U62485 ) encoding a glycolate oxidase , which is involved in plant photorespiration and , when the expression of the gene is down-regulated , plants develop yellowing [18] . Furthermore , the necrotic symptoms induced by CMV Y-Sat infection in tomato are also associated with a short sequence , from nt . 309 to 334 , of the Y-Sat genome [4] . Single nucleotide changes to these short sequences have been shown to dramatically affect both the severity and host specificity of these satellite RNA-induced disease symptoms [4]–[6] . Taken together , these observations suggest that siRNA-directed host gene silencing is a common mechanism for satellite RNA-induced symptoms . This mechanism could also be extended to viroids , another group of small non-protein-coding RNA pathogens in plants [19] . The pathogenicity of Potato spindle tuber viroid ( PSTVd ) , a 359-nt non-protein-coding RNA pathogen , has been associated with two ∼20-nt partially complementary regions of the PSTVd RNA genome known as virulence-modulating regions [20] . Furthermore , expression of an hairpin RNA transgene derived from PSTVd in tomato resulted in the expression of symptoms similar to PSTVd infection [10] , suggesting that a sRNA-directed RNA silencing mechanism is also responsible for the disease symptoms induced by PSTVd infection . The siRNA-mediated disease mechanism reported here is consistent with the previous observation that disease induction by satellite RNAs involves the interaction of satellite RNAs , their helper viruses , and the host plant [1] , [2] . Diseases caused by satellite-derived siRNA-mediated host gene silencing would require i ) the existence of sufficient sequence complementarity between the satellite RNA genome and a host gene mRNA; ii ) a helper virus that supports high levels of replication of the satellite; and iii ) host RNA silencing machineries for efficient siRNA biogenesis and siRNA-directed mRNA degradation . One of the key helper virus-encoded factors for satellite-induced disease symptoms would be a RNA silencing suppressor protein . Most plant viruses encode silencing suppressor proteins , which inhibit sRNA-mediated silencing in their host [21] . Silencing suppressors from helper viruses could affect satellite siRNA-mediated diseases in two ways . The suppressor protein could inhibit host antiviral silencing to enhance the accumulation of satellite RNAs , or , it could act to inhibit satellite siRNA-induced silencing of complementary host gene sequences to minimize the disease symptoms . Consistent with the latter possibility , transgenic expression of the potyvirus silencing suppressor P1/HC-Pro in tobacco almost completely abolished the yellowing symptoms induced by CMV Y-Sat infection [10] . Also , the chlorotic symptoms induced by the B2 and WL3 satellites of CMV in tobacco are associated specifically with RNA2 of subgroup II CMV: the symptoms are diminished when the satellites are replicated by subgroup I CMV [22] . A recent study demonstrated that subgroup I CMV encodes a stronger silencing suppressor protein ( 2b , encoded by RNA 2 ) than subgroup II CMV [23] , raising the possibility that the lack of strong B2 and WL3 satellite-induced chlorosis in the presence of subgroup I CMV is due to CMV 2b-mediated suppression of satellite siRNA-induced host gene silencing . Thus , viral-encoded silencing suppressor proteins may have a dual function , helping the virus or subviral agent to evade sRNA-mediated antiviral defence by preventing the silencing of viral RNAs , and minimizing symptom severity by inhibiting the silencing of host genes . RNA silencing has been suggested to be a driving force for the evolution of subviral RNAs including viral satellite RNAs [1] , [10] . These RNA species tend to form stable secondary structures that are resistant to siRNA-mediated degradation [10] , [24] . Also , satellite RNAs have little or no sequence homology with their helper virus genome [1] , and this is presumably to avoid the silencing of the helper viruses by satellite-derived siRNAs . Our findings have further implications for the involvement of RNA silencing in the evolution of viral satellite RNAs in plants . Satellite RNAs with extensive sequence homologies to host genes would result in silencing of the targeted gene ( s ) , affecting the viability of the host plant . This in turn , would affect the survival of both the helper virus and its satellite RNA . Consistent with this scenario , there has been no report showing extensive sequence homology between satellite RNAs and their host genome . As discussed for satellites and viroids , infection of plants with both RNA and DNA viruses is also associated with the accumulation of virus-derived siRNAs [21] . Furthermore , more than 100 miRNAs have been identified from animal viruses [25] . Whether sRNA-directed host gene silencing plays a wider role in plant and animal viral disease remains to be investigated . Nevertheless , numerous plant and animal viral sRNAs have been shown to match host gene sequences and therefore have the potential to down-regulate their expression [26]–[30] . This raises the possibility that at least some virus-induced diseases are due to viral sRNA-directed silencing of host genes . Recent advances in high-throughput DNA and RNA sequencing technologies are expected to result in complete genomic sequences , not only for the infecting viruses , but also for their respective hosts , providing new opportunities to identify host genes that are potentially targeted by virus-derived sRNAs . We have demonstrated here that transformation of tobacco with a silencing-resistant version of the CHLI gene completely prevented the yellowing symptoms associated with Y-Sat infection . This finding offers a potential new strategy for preventing viral siRNA-mediated diseases in plants and animals . However , viral replication has relatively high error rates and viruses often exist as quasispecies ( mixtures of minor sequence variants ) [31] . Thus , a modified host target gene protected against siRNAs of the original virus could potentially be silenced by siRNAs from a variant virus with an altered sequence . Introducing multiple nucleotide changes into the target sequence of the host gene , as demonstrated for the modified CHLI sequence harbouring 10 single nucleotide modifications , could minimize the recurrence of host gene silencing by siRNAs of variant viruses .
All Nicotiana species were grown in a 25°C glasshouse with natural light . Infection of tobacco plants with Cucumber mosaic virus plus Y-Sat , total RNA extraction and northern blot hybridization were performed as previously described [32] . Solexa sequencing of small RNAs from Y-Sat-infected tobacco was carried out at the Allan Wilson Centre Genome Service ( New Zealand ) . BLAST searching to identify Y-Sat-targeted tobacco genes was performed as follows: 21-nt segments of the Y-Sat yellow domain sequence ( e . g . nt . 177–197 , nt . 178–198 , nt . 179–199 ) were used to BLAST search common tobacco sequences ( taxid:4097 ) of the NCBI database “nucleotide collection ( nr/nt ) ” . This search identified the N . tobacum CHLI cDNA ( accessions: U67064 and AF014053 ) as the “best” match with the Y-Sat sequence . Full-length genomic and cDNA sequences of the CHLI gene were amplified from total DNA and RNA using the NEB Long Amp Taq kit and Qiagen One Step RT-PCR kit respectively according to the manufacturer's instructions . The CHLI forward primer ( 5′ ATCTGGTACCAAAATGGCTTCACTACTAGGAACTTCC 3′ ) and reverse primer ( 5′ TCTAGTCTAGAAGCTTAAAACAGCTTAGGCGAAAACCTC 3′ ) were used for both the PCR and RT-PCR reactions . PCR products were purified using a QIAquick PCR purification kit ( Qiagen ) , cloned into the pGem-T Easy cloning vector ( Promega ) and sequenced . The full-length genomic and cDNA sequences ( see Text S1 ) were digested with KpnI/XbaI and cloned into the intermediate vector pBC ( Strategene ) . To create the modified CHLI sequence ( mtCHLI and gmtCHLI ) , a 630 bp sequence containing the Y-Sat targeted 22-nt sequence was amplified as two halves; i ) the 5′ half was amplified with forward ( WT-F1 , 5′ TGGCACAATCGACATTGAGAAAGC 3′ ) and reverse ( MT-R1 , 5′ ACGTAATTCGGCGTCACGTACGGTCCCCACTTGGGGATGC 3′ ) primers that spanned the 22-nt CHLI target site and allowed for the introduction of modified nucleotides , and ii ) the 3′ segment was amplified with a forward primer ( MT-F2 , 5′ GTACGTGACGCCGAATTACGTGTGAAGATAGTTGAGGAAAGAG 3′ ) that also contained modified nucleotides spanning the 22-nt CHLI target sequence overlapping with MT-R1 and a reverse primer ( WT-R2 , 5′ AGCAGTTGGGAATGACAGTGGC3′ ) . The two amplified products were joined together using overlapping PCR with Pfu polymerase ( Promega ) to generate a 630 bp fragment with a modified Y-satellite target sequence . A 223 bp fragment , containing the modified target sequence , was released by PstI and EcoRI digestion and used to replace the PstI-EcoRI fragment of the wild type sequence in the CHLI cDNA , giving rise to the modified sequence mtCHLI . Digestion of the mtCHLI sequence with PstI and BamHI released the modified region which was used to replace the corresponding region in the wild-type CHLI genomic sequence , giving rise to the modified genomic clone gmtCHLI . The wild-type and mutated cDNA or genomic sequences were digested with KpnI and XbaI and inserted between the 35S promoter and the Ocs terminator in pART7 [33] , and the resulting expression cassettes were cloned into the binary vector pART27 [33] as a NotI fragment . Similarly , the wild-type and mutant CHLI cDNA clones were cloned into a binary vector with the rubisco small sub unit promoter . To prepare the RNAi construct , CHLI cDNA was digested with PstI and XbaI releasing a 571 bp fragment spanning the Y-Sat target site . This fragment was cloned into the Gateway-based hairpin RNA gene silencing vector Hellsgate 12 which incorporates a spliceable intron for improved silencing efficiency [34] , [35] . All plant expression vectors were introduced into Agrobacterium tumefaciens strain LBA4404 by triparental mating in the presence of the helper plasmid pRK2013 . Agrobacterium-mediated transformation of tobacco was carried out as described previously [36] , using 50 mg/L kanamycin as the selective agent . Total RNA ( 2 µg ) was ligated to a 24-nt RNA adaptor ( 5′ AACAGACGCGUGGUUACAGUCUUG 3′ ) using T4 RNA ligase ( Promega ) at room temperature for 2 hours in 50 mM HEPES pH 7 . 5 , 0 . 1 mg/mL BSA , 8% glycerol , 2 units/µL RNasein RNase inhibitor ( Promega ) and 0 . 5 unit/µL T4 RNA ligase ( Promega ) . The ligation was purified by phenol-chloroform extraction and ethanol precipitation . The purified product was reverse-transcribed using a CHLI-specific reverse primer ( 5′ AGCAGTTGGGAATGACAGTGGC 3′ ) . Primary PCR was then performed using a forward primer matching the RNA adaptor ( 5′ AACAGACGCGTGGTTACAGTC 3′ ) and the CHLI reverse primer ( as above ) . The RT-PCR product was then amplified using the same forward primer with a nested CHLI reverse primer ( 5′ ATATCTTCCGGAGTTACCTTATC 3′ ) . The nested PCR product was separated on a 2% agarose gel , purified with the Ultra Clean-15 DNA purification kit ( Mo Bio Laboratories ) , and ligated into the pGEM-T Easy vector ( Promega ) for sequencing . Approximately 0 . 5 grams fresh weight of plant material was ground into fine powder in liquid nitrogen , mixed with 15 mL methanol and filtered through filter paper . Chlorophyll a and chlorophyll b were measured in a spectrophotometer ( Biochrom WPA light wave II ) at a wavelength of 663 nm and 645 nm respectively . Chlorophyll concentration was measured as nmol per gram of fresh weight . | Viral infections result in a variety of disease symptoms that vary in character and severity depending on the type of viral infection and individual host factors . Despite extensive research , the molecular basis of viral disease development has remained poorly understood . Both plant and animal viruses express 20–25 nucleotide siRNAs or microRNAs ( miRNAs ) in their host . These virus-specific small RNAs ( sRNAs ) direct RNA silencing of homologous viral genes to restrict virus replication forming part of the host's antiviral defense mechanism . Using a plant viral satellite RNA as a model system , we demonstrate here a new function for virus-derived sRNAs: induction of disease symptoms by silencing of a physiologically important host gene . Furthermore , we demonstrate that such viral-derived sRNA-induced disease symptoms can be prevented by the expression of a either naturally evolved , or artificially introduced , silencing-resistant sequence variant of the viral siRNA-targeted gene . These findings not only provide the first demonstration of a sRNA-mediated viral disease mechanism , but also offer an alternate strategy to prevent the onset of such viral diseases . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"plant",
"science",
"plant",
"biology",
"plant",
"pathogens",
"plant",
"pathology",
"biology"
] | 2011 | Viral Small Interfering RNAs Target Host Genes to Mediate Disease
Symptoms in Plants |
Informed consent is one of the principal ethical requirements of conducting clinical research , regardless of the study setting . Breaches in the quality of the informed consent process are frequently described in reference to clinical trials conducted in developing countries , due to low levels of formal education , a lack of familiarity with biomedical research , and limited access to health services in these countries . However , few studies have directly compared the quality of the informed consent process in developed and developing countries using the same tool and in similar clinical trials . This study was conducted to compare the quality of the informed consent process of a series of clinical trials of an investigational hookworm vaccine that were performed in Brazil and the United States . A standardized questionnaire was used to assess the ethical quality of the informed consent process in a series of Phase 1 clinical trials of the Na-GST-1/Alhydrogel hookworm vaccine that were conducted in healthy adults in Brazil and the United States . In Brazil , the trial was conducted at two sites , one in the hookworm non-endemic urban area of Belo Horizonte , Minas , and one in the rural , resource-limited town of Americaninhas , both in the state of Minas Gerais; the American trial was conducted in Washington , DC . A 32-question survey was administered after the informed consent document was signed at each of the three trial sites; it assessed participants’ understanding of information about the study presented in the document as well as the voluntariness of their decision to participate . 105 participants completed the questionnaire: 63 in Americaninhas , 18 in Belo Horizonte , and 24 in Washington , DC . Overall knowledge about the trial was suboptimal: the mean number of correct answers to questions about study objectives , methods , duration , rights , and potential risks and benefits , was 45 . 6% in Americaninhas , 65 . 2% in Belo Horizonte , and 59 . 1% in Washington , DC . Although there was no difference in the rate of correct answers between participants in Belo Horizonte and Washington , DC , there was a significant gap between participants at these two locations compared to Americaninhas ( p = 0 . 0002 and p = 0 . 0001 , respectively ) , which had a lower percentage of correct answers . Attitudes towards participating in the clinical trial also differed by site: while approximately 40% had doubts about participating in Washington , DC and Belo Horizonte , only 1 . 5% had concerns in Americaninhas . Finally , in Belo Horizonte and Washington , high percentages cited a desire to help others as motivation for participating , whereas in Americaninhas , the most common reason for participating was personal interest ( p = 0 . 001 ) . Understanding of information about a Phase 1 clinical trial of an experimental hookworm vaccine following informed consent was suboptimal , regardless of study site . Although overall there were no differences in knowledge between Brazil and the US , a lower level of understanding about the trial was seen in participants at the rural , resource-limited Brazilian site . These findings demonstrate the need for educational interventions directed at potential clinical trial participants , both in developing and developed countries , in order to improve understanding of the informed consent document .
The principle of informed consent is internationally recognized as one of the essential elements of the ethical conduct of research involving human subjects [1 , 2] . Within its ethical and legal foundations , obtaining informed consent has two specific objectives: to respect and promote the autonomy of research participants , and to protect the research subjects from possible harm or exploitation [3] . The informed consent process depends upon five criteria: the willingness to participate , the capacity to make a decision , disclosure of information , comprehension , and the decision to participate [4] . The quality of the informed consent process is determined primarily by the level of the study volunteers’ understanding and by the absence of coercion from the decision-making process [5 , 6] . The findings of research indicate that , in general , there are gaps in individuals’ knowledge of various aspects of the clinical trials for which they are being consented , which may potentially impact their decision to participate [7–12] . In this sense , critics fear that for many clinical trials , the informed consent process may not be fully meeting its intended objectives [13] . Breaches in the informed consent process are frequently described in reference to clinical trials conducted in developing countries [14] . Low levels of formal education , a lack of familiarity with biomedical research , and limited access to health services in these countries have been associated with an inadequate informed consent [15–20] . A meta-analysis of the subject , however , demonstrates that this problem is not limited to developing countries , as several aspects of the informed consent process are poorly understood by participants in clinical trials both in developing and in developed countries [3] . In fact , one of the few studies dedicated to an empirical comparison of consent obtained in developed and developing countries revealed that there were no substantial differences in the participants’ knowledge between the two settings [21] . Furthermore , Mandava et al observed that the understanding of information about studies varies within both groups of volunteers and that to assume that clinical trials conducted in developing countries are less ethical than those conducted in developed countries is an oversimplification of an undoubtedly complex situation [11] . Most of the studies published to date on the informed consent process , however , have acknowledged limitations in their methodologies . Critically , the use of different measurement instruments has hindered comparison of results between different clinical trials . In this sense , these investigations have provided limited contributions to the discussion on the comparative quality of the consent process in developed and developing countries . Given the methodological limitations of studies reported in the literature , we conducted an investigation to compare the quality of the informed consent process of a series of clinical trials performed in Brazil and the United States , using a standardized questionnaire . Our research sought to answer the following question: can the ethical quality of the informed consent of participants in clinical trials carried out in developed countries be considered superior to that obtained in developing countries ? The authors hypothesize that there is no substantial difference in the quality of informed consent of research subjects living in developed and developing countries . The justification for this study resides in the need to assess whether there is cause for concern regarding the protection of research participants in countries in which clinical trials are being conducted . The identification of differences and similarities between informed consent processes in developed and developing countries may aid in the implementation of specific strategies to protect participants in each research setting . Among these strategies , we can cite the need to support and inform institutional ethical review committees in their evaluations of the informed consent process of proposed clinical research .
A descriptive quantitative study with a cross-sectional design was conducted to compare the quality of the informed consent process of two Phase 1 clinical trials performed in Brazil and the United States , respectively , of the Na-GST-1/Alhydrogel hookworm vaccine that is being developed by the Sabin Vaccine Development Partnership [22] . The trials were carried out in the cities of Belo Horizonte and Americaninhas ( Brazil ) and in Washington , D . C . ( United States of America ) . In Brazil , a Phase 1 clinical trial was conducted between 2011 and 2014 of the safety and immunogenicity of Na-GST-1/Alhydrogel administered with or without the GLA-AF immunostimulant in healthy adults ( protocol SVI-10-01 , NCT01261130 ) . The principal objective of this trial was to estimate the frequency of adverse events to the candidate hookworm vaccine . Vaccinations were conducted first in the hookworm non-endemic site of Belo Horizonte and then in the hookworm-endemic area of Americaninhas , to establish the vaccine’s safety in a hookworm-unexposed population before testing it in endemic areas . A Phase 1 clinical trial of similar design was conducted of Na-GST-1/Alhydrogel administered with or without a different immunostimulant ( a CPG oligodeoxynucleotide ) in Washington , DC , starting in 2014 ( protocol SVI-GST-03 , NCT02143518 ) . In both clinical trials , healthy adults ( aged 18–45 years in Brazil and 18–50 years in the USA ) were enrolled and vaccinated by intramuscular injection according to a 0 , 2 , and 4-month schedule . In Brazil , the SVI-10-01 clinical trial was carried out in two separate centers: in Belo Horizonte and in Americaninhas , 556 km from Belo Horizonte . Americaninhas is a town of approximately 1500 residents located in the mostly rural municipality of Novo Oriente de Minas Gerais , in the Mucuri Valley , in the northeast part of the state of Minas Gerais . Belo Horizonte is the capital of Minas Gerais , with a population of 2 , 479 , 175 inhabitants and a human development index ( HDI ) of 0 . 81 , which is considered very high [23] . On the other hand , Americaninhas is a region with low social indicators: it has an HDI of 0 . 60 , the 6th-worst amongst Minas Gerais municipalities [24] . Low levels of formal education are a concern: 57 . 1% of its inhabitants are illiterate [24] . In the United States , the SVI-GST-03 clinical trial was conducted at the George Washington ( GW ) Medical Faculty Associates , a high-volume outpatient clinic affiliated with the GW hospital in the urban center of Washington , District of Columbia . For both clinical trials , all participants underwent the informed consent interview and , if they decided to participate , signed an informed consent form ( ICF ) that had been approved by the ethical review committees of the Centro de Pesquisas René Rachou and the Brazilian federal Ministry of Health ( for SVI-10-01 ) , as well as the George Washington University ( for both trials ) . The approved ICFs that were used for the trial in Brazil in Belo Horizonte ( S1 ICF ) and Americaninhas ( S2 ICF ) , as well as the approved ICF that was used for the trial in Washington , DC ( S3 ICF ) are included as supporting information . The ICFs for the two trials differed primarily in the description of the different adjuvants that were used in the vaccine formulations ( GLA-AF in Brazil vs . CPG in the United States ) and in country-specific requirements such as the inclusion of language related to the Health Insurance Portability and Accountability Act in the United States . However , the ICFs were the same regarding the description of the study rationale and the nature of the hookworm vaccine , the risks and benefits of the Na-GST-1/Alhydrogel hookworm vaccine , the number of vaccinations to be administered , the duration of the study ( per participant ) , the type of procedures to be conducted , the fact that participation was voluntary , and that consent could be withdrawn at any time with no negative consequences to participants . Completion of the informed consent questionnaire was optional and was not required for participation in the rest of the respective study . Data were collected through the use of a semi-structured questionnaire consisting of 32 questions that assessed the participants’ understanding of the information about the study presented in the informed consent document for the respective clinical trial as well as the voluntariness of their decision to participate . The questions sought to evaluate their knowledge of the purpose of the clinical trial , the study methods , the duration of the trial , the participants’ rights , and the potential risks and benefits of participation . The participants’ socio-demographic and economic information were also collected . Given the lack of appropriate existing questionnaires for this type of clinical trial , especially in two languages , questions , although not pre-tested , were based on published questionnaires used for clinical studies for other disease areas [25–27]; the International Ethical Guidelines for Biomedical Research [1]; and , on the experience the researchers have gained from working in this field for over 13 years . In order to improve understanding of the questions , the authors followed the recommendations of Vieira , which included using plain and easily understandable language ( assessed by the Flesch reading-ease score ) ; using general language rather than technical terminology; and , avoiding negative phrases and words with double meaning [28] . The questionnaire was originally formulated in Brazilian Portuguese and later translated into English by experts on the research subject . With the agreement of the researchers involved , this process favored an interpretation of concepts rather than a literal translation of terms . Questions were also tailored to the specific goals , risks and benefits of each clinical trial . Given the need for reliability of the measuring instrument , the researchers opted for using open-ended questions since pre-determined answers to close-ended questions might influence the participants’ responses [28] . The preference for this type of question arises from a study by Lindegger et al , which revealed that the participants’ understanding of the informed consent information was overestimated when evaluated by instruments using close-ended questions compared with those using open-ended questions [26] . Data were collected after all volunteers participated in the informed consent process and signed the informed consent form . In Brazil , the questionnaire was administered by interviewers who had received training in how to standardize data collection and improve reliability , in order to minimize the risk of information bias . The interviewers were undergraduate and graduate students in nursing , education , psychology and medicine , had no relationship with the clinical research staff , and were specially trained to comprehensively transcribe the participants' responses . In the USA , the questionnaires were self-administered at the study site clinic . Different methods of applying the questionnaire in the USA and Brazil were chosen due to differences in the level of education of the volunteers , as had been observed in previous studies carried out in the same areas . The administration of the questionnaire lasted on average 10 minutes and it was carried out at the same setting in which the informed consent process was conducted . Most of the questionnaires were completed on the day of first vaccination or , in a small number of cases at all sites , immediately after signing the informed consent form . After collection , data were coded and entered into an SPSS database ( version 14 . 0 ) and Microsoft Excel . In order to ensure reliability , data were independently entered twice . In cases of discrepancy between the two entries , the lead researchers referred to the original questionnaire and determined the actual response by consensus . Analysis of the open questions followed the categorization of the responses , based on the criterion of appearance frequency . To avoid bias in the process of categorization , this step was performed independently by two different professionals . The end results of this stage were compared; in cases of disagreement between the categorizations , the professionals debated , each justifying their choice . After an agreement had been reached by consensus , the participant’s response was classified into the appropriate category . Data were initially analyzed using descriptive statistics including frequency calculations ( simple and relative ) , as well as mean and standard deviation . Subsequent analyses compared the percentages of correct answers ( categorical variables ) using the chi-square test . A Knowledge Index ( KI ) was created to measure participants' knowledge on all issues evaluated . This index consists of the sum of the participants’ correct responses divided by the total number of questions ( 11 ) and is expressed as a percentage ranging from 0% ( the participant answered all questions incorrectly ) to 100% ( the participant answered all questions correctly ) . The analysis variable was analysed by calculating the mean , median and interquartile ranges . The KI was compared between study sites using the one-way ANOVA and Tukey-HSD tests . A significance level ( p value ) of 0 . 05 was used for all analyses . The normality of continuous variables was assessed by the Kolmogorov-Smirnov test .
A total of 105 study participants completed the informed consent questionnaire and were included in the analysis: 63 ( 60% ) from Americaninhas , 18 ( 17% ) from Belo Horizonte and 24 ( 23% ) from Washington , DC . In Americaninhas , 3 of 66 ( 4 . 5% ) participants enrolled in the clinical trial declined to complete the questionnaire , whereas 12 of 36 ( 33% ) and 0 of 24 ( 0% ) declined in Belo Horizonte and Washington , DC , respectively . Significantly more participants in Belo Horizonte declined to complete the questionnaire than in either Americaninhas or Washington , DC ( p = 0 . 016 ) , although the reasons for refusal were not recorded . The average age of those completing the questionnaire was 29 . 3 years ( SD 8 . 9 , range 18 to 50 ) , which varied significantly by study site ( Belo Horizonte , 23 . 7 years , Americaninhas , 29 . 6 years , Washington DC , 32 . 8 years; p = 0 . 021 , Kruskal-Wallis test ) . The proportion of study participants who were female ( 46 . 7% ) , on the other hand , did not vary significantly between the study sites . Regarding the maximum level of education achieved by participants , 37 ( 35 . 2% ) had primary education , 33 ( 31 . 4% ) had secondary education , 25 ( 23 . 8% ) had post-secondary education and 6 ( 5 . 7% ) had post-graduate education; 4 ( 3 . 8% ) participants were deemed illiterate ( all in Americaninhas ) . Levels of education were not uniform across the study sites: the chi-square test revealed a statistically significant difference between the site of the clinical trial and the participants’ maximum level of education , with the lowest levels of education observed in Americaninhas ( p<0 . 001 ) . The distribution of the participants according to their level of education is shown in Table 1 . Most participants did not have a health insurance plan ( n = 77; 73 . 3% ) , had never participated in a clinical trial ( n = 78; 74 . 3% ) , and had no formal employment contract ( n = 72 , 68 . 6% ) . The chi-square and Kruskal-Wallis tests revealed statistically significant differences between the location of the clinical trial and having a health insurance plan ( p<0 . 001 ) , or having previously participated in a clinical trial ( p = 0 . 030 ) . Regarding health insurance , it appears that only three participants had formal insurance in Americaninhas ( 4 . 0% ) , with higher proportions being found in Belo Horizonte ( 16 . 6% ) and Washington , DC ( 79 . 1% ) . In terms of formal employment , most participants in Americaninhas ( 71 . 3% ) and in Washington , DC ( 79 . 1% ) had jobs , with a lower proportion in Belo Horizonte ( 44 . 4% ) . While 55 . 5% of participants in Washington , DC had previously participated in a clinical trial , much lower rates were seen among the participants in either Belo Horizonte ( 0 . 5% ) or in Americaninhas ( 25 . 5% ) . Table 2 shows the absolute and relative frequencies of correct answers to questions that evaluated participants’ knowledge regarding information about the clinical trial that was contained in the informed consent form . A majority of participants knew the correct answers to at least seven out of the eleven questions . The research subjects from Belo Horizonte had the highest percentage of correct answers , with an average of five questions ( 45% ) answered correctly , followed by four in the United States ( 36% ) , and one in Americaninhas ( 19% ) . The analysis of each question demonstrated that the majority of participants understood that deciding not to participate in the clinical trial for which they were being asked to volunteer would not result in any negative consequences . However , in Americaninhas only 51% recognized that declining to participate was not associated with of any negative consequences of not participating , compared to 83% and 87% in USA and Belo Horizonte , respectively . In addition , high proportions were aware of what they should do in case of illness during the trial , the possibility that they might experience anticipated or unanticipated adverse effects after being vaccinated , and that they could contact members of the study team if they had any doubts or questions about the trial . As shown in Table 2 , there were statistically significant differences between the study sites in the comprehension of the following items: the study objectives ( p = 0 . 001 ) ; the consequences of choosing not to participate in the study ( p = 0 . 002 ) ; the potential risks of the investigational vaccine ( p = 0 . 037 ) ; and the possibility of unanticipated adverse effects ( p = 0 . 001 ) . Table 3 summarizes the participants’ knowledge about information contained in the informed consent form by calculating a “knowledge index” ( KI ) consisting of the mean number of correct answers to these questions on the questionnaire , by study site . In Americaninhas , participants had an average of 45 . 9% of correct answers; in Belo Horizonte , 65 . 2%; while in Washington , DC , the percentage was 59 . 1% . Table 4 provides comparisons between the average number of correct answers in the KI according to the one-way ANOVA and Tukey-HSD tests . The overall association between the study site and the average knowledge about the clinical trial was statistically significant ( one-way ANOVA: F = 13 . 931 , p = 0 . 0001 ) . Although there was no difference in the rate of correct answers between participants in Belo Horizonte and those in Washington , DC ( p = 0 . 437 ) , there was a significant gap between the KI of participants at these two locations compared to Americaninhas ( p = 0 . 0002 and p = 0 . 0001 , respectively ) , where a lower percentage of correct answers was recorded . Table 5 provides the absolute and relative frequencies of responses concerning the participants’ attitudes towards clinical research and the voluntariness of their decision to participate in the clinical trial . In all three locations , the majority of participants reported not being afraid of participating in the research and trusting the investigators responsible for the trial . However , in Belo Horizonte , 72 . 2% of participants declared having enrolled in the study only for the benefits , a situation that was not observed at the two other sites . Regarding the participants’ doubts about participating in the clinical trial , whereas in Washington , DC , and Belo Horizonte approximately 40% reported having doubts , in Americaninhas only 1 . 5% admitted having some concerns about participating . Virtually all participants ( 90 . 5% ) resident at that site believed that participation in the clinical trial could lead to improvements in their health . In Washington , DC , only 4 . 2% of participants admitted to being afraid to participate , while the comparable values in Americaninhas and in Belo Horizonte were higher , at 38 . 1% and 27 . 7% , respectively ( p = 0 . 01 ) . It appears that at the Brazilian study sites , participation in informational meetings about the study with study team members was a significant factor that influenced their decision to participate ( 76 . 2% and 77 . 8% in Americaninhas and Belo Horizonte , respectively ) , while in Washington only 41 . 7% cited this influence ( p = 0 . 001 ) ( Table 5 ) . Table 6 details participants’ responses regarding their attitudes to the clinical trial using the Likert scale . In both Brazil and the United States , most ( 90 . 4% and 100% , respectively ) respondents agreed or strongly agreed when asked if they agreed or disagreed with the statement , “You want to participate in the study . ” When asked if they only “tolerated” participation in the clinical trial , most subjects in the United States and in Belo Horizonte disagreed or strongly disagreed . On the other hand , in Americaninhas , most participants agreed or strongly agreed with the same question ( p = 0 . 02 ) . Table 7 shows the absolute and relative frequencies of the study subjects’ motivations for participating in the clinical trial . When asked about their main motivations for participating , those in Belo Horizonte and Americaninhas reported personal or social advantages that would benefit them . In contrast , of the study participants in Washington , DC , 17 . 4% said that participating in the trial would bring more benefits mainly to society through development of a new vaccine for hookworm . In Americaninhas , the most common reason for participating was motivated by personal interest; in Belo Horizonte , half of participants reported that their main reason for participating was a desire to help others . In the United States , high percentages cited the possibility of helping others and receiving monetary compensation as reasons for participating . The reasons for participating in the clinical trial varied significantly by study site ( p = 0 . 001 , chi-square test ) .
The results of the study reported herein indicate that there were no substantial differences between the overall quality of the informed consent obtained from participants in similar clinical trials conducted in the United States , a developed country , and in Brazil , a developing one . Such a conclusion is supported by the absence of any statistically significant differences between participants in Belo Horizonte and the United States in their knowledge of information about the clinical trial contained in the informed consent form . However , our research nevertheless showed a significant association between the particular site where the trial was conducted and the quality of the informed consent process: statistically significant differences were observed in the study participants’ knowledge about the trial between Americaninhas , Belo Horizonte and Washington , DC , with residents of Americaninhas having the lowest percentage of correct answers on the informed consent questionnaire . The inequality of living conditions within the Brazilian population is a widespread reality . In many regions of Brazil , substantial disparities exist between urban and rural areas with regard to household income , basic infrastructure , access to healthcare , and quality of education . For example , the municipality of Padre Paraíso , the seat of the Americaninhas district where one of the clinical trials of this study took place , has an HDI of 0 . 60 that is classified as low , whereas in Belo Horizonte , the capital of the state of Minas Gerais , the HDI is considered “very high” at 0 . 81 [24] . Socio-demographic and economic characteristics such as advanced age , low level of education , female gender , and low socioeconomic status have been associated with a reduced quality of the informed consent process [29–33] . Therefore , it is essential that the characteristics of the potential research participants being recruited into a clinical trial be adequately analyzed in order to identify factors that may negatively influence the quality of the informed consent obtained from them . These characteristics must also be analyzed to support researchers in developing strategies to encourage the dissemination and understanding of information about clinical research , such as the use of appropriate language in the informed consent form and the development of more relevant educational interventions that match the context of study participants . There are suggestions in the literature that efforts to establish greater links between researchers and participants , such as creating an atmosphere of openness to dialogue and giving opportunities for asking questions , can facilitate the consent process [34] . Regarding the lower percentage of success on the informed consent questionnaire observed in Americaninhas , the relative lack of understanding at this site about the scientific purpose of the study suggests that the study subjects in Americaninhas did not consider themselves to be participants in clinical research . Instead , they may have conflated the scientific purpose of the clinical trial with the provision of medical care , a phenomenon that has been termed the “therapeutic misconception” [33] . Although the study volunteers were participating in a Phase 1 trial that by definition may not provide any direct benefit to participants , this phenomenon may have resulted from the fact that they either received the hepatitis B vaccine as the comparator vaccine or were offered it at the end of the study and , if necessary , received treatment for hookworm and anemia , as well as being referred for further investigations or management in cases of other illnesses or medical conditions . Commonly observed amongst clinical trial participants who are socially and economically disadvantaged , the therapeutic misconception results from confusion between routine medical care provided to study participants in the context of a trial and the objectives of the study in which an experimental product is being tested . Instead of understanding that they are participating in an experiment , these individuals believe that the research protocols are tailor-made for them and their health-related issues [35 , 36] . This phenomenon was observed in a previous study conducted in Americaninhas , in which many participants in a clinical trial of the Na-ASP-2 vaccine against hookworm believed that the purpose of the investigation was the medical treatment of its participants rather than the testing of an experimental preventative vaccine [7] . The manifestation of the therapeutic misconception is in conflict with the doctrine of informed consent since the participant experiencing such a misconception may not adequately weigh the risks and benefits of their participation in the clinical trial and instead base their decision to participate on incorrect criteria and false expectations [37] . The limited knowledge of the consequences of not participating in a clinical trial that was observed in Americaninhas is similar to the findings of Tam et al [3] , which demonstrated that participants in clinical trials conducted in developing countries are less acquainted with issues related to participation or refusal of participation in a study . Many studies have affirmed that information about the right to refuse participation and to withdraw consent at any time , without affecting their rights or access to medical care is one of the most important items to be conveyed to prospective research participants , particularly to those in developing countries [38 , 39] . Despite the above differences , it should be noted that most participants—at all study sites—had an incomplete knowledge of the information contained in the informed consent form . The highest average knowledge as assessed by a structured questionnaire amongst the three locations was in Belo Horizonte ( 65 . 2% ) , which is lower than the average reported in other studies of the informed consent process in developing countries [40–42] , in developed countries [43] , and in both types of countries [21] . Regardless of the setting in which the research is conducted , a lack of understanding of the information about the research impairs the quality of the individuals’ decision to participate [44] . The inability to describe the risks and possible adverse effects of participation in a particular clinical trial , observed mostly in Americaninhas and in Washington , DC , has been seen in other studies of the informed consent process , such as in a breast cancer clinical trial conducted in Japan [45] . In that study , the scores achieved by participants were acceptable in terms of a broad understanding of the informed consent document , but were low for particular items such as the experimental nature of the study , potential risks , benefits , and compensation . The limited understanding of information contained in the informed consent form may influence the voluntariness of an individual’s decision to participate [8] . In this sense , the results of the questionnaire related to knowledge about the clinical trial discussed above suggest that the informed consent obtained in Americaninhas may have been relatively compromised in relation to the voluntary decision to participate , in comparison to other places . Other aspects related to the willingness of volunteers from Brazil to participate in the study consist of potential indirect benefits associated with participation in the clinical research , such as learning about the disease and receiving a medical examination as part of screening . While such care is provided free of charge by the universal Brazilian public system of health in Americaninhas , access to medical care is hampered due to a shortage of medical professionals in the region and to excess demand , conditions that may have had a direct influence on willingness to participate in the study . Aspects related to the participants’ attitudes also influence the voluntariness and the quality of the informed consent , such as fear of participation , trust in the study team , the expression of doubt , and the underlying intentions and motivations for participating . Regarding the voluntariness of the decision to participate in the clinical trials that were the subject of the current research , most participants at all three sites reported not being afraid of participating and having trust in the investigators who were responsible for the clinical trials , an important finding that contributed to the quality of the informed consent in both countries . Jenkins and Fallowfield remark that fear and dissatisfaction with certain research procedures were cited as reasons for people declining participation in cancer research studies [46] . Despite this , a study aimed at analyzing attitudes toward participation in clinical research in a developing country demonstrated that the majority of respondents reported that they would not like to entrust decisions about their health to physicians . Only a small proportion of participants , particularly those without any formal education , would leave the process of making health decisions in the hands of physicians [47] . The study participants from Americaninhas , unlike those from Belo Horizonte or Washington , DC , had very few doubts or concerns about the clinical research in which they were participating . This might be explained by a reduced capacity to ask questions during the informed consent process due to lower levels of education [48] . In addition , we found that the popular image of the physician overlapped with the image of the researcher in this community , something previously observed in this region [49] . This situation might intimidate study volunteers , making them less likely to ask critical questions of the researchers [40] . It has previously been shown that clinical trial participants with higher levels of education are more inclined to discuss their potential participation in the study with the research team [35] . In contrast , in Americaninhas the belief that participation in a clinical trial may improve their health status , may lead them to be less critical of the research aspects of the study [50] . Regarding the desire to enroll in the clinical trial , we found that most participants actively wished to participate; however , paradoxically , in Americaninhas , half merely “tolerated” participation . This contradiction might be attributed to two factors: a difficulty in understanding the question , despite the care taken during questionnaire preparation and implementation; or , a limited voluntariness in the decision-making process , resulting from a lack of understanding of the information about the trial , third-party influences , or the level of trust in the researchers [8] . Another aspect that may be associated with limited voluntariness and that may help to explain this contradiction is the notion that participation in the clinical trial might improve one’s health . Especially in developing countries , the level of trust in the investigators conducting a clinical trial might influence a reluctant volunteer during the consent process due to their degree of authority in such an environment; a potential research participant might feel that participation cannot be refused to such an individual [51] . For example , a review by Mandava et al revealed that participants from developing countries are less likely to refuse participation in research and are more likely to worry about the consequences of their refusal to participate [11] . Regarding the motivation for participating in the clinical trial conducted in Americaninhas , most individuals attributed it to a personal decision taken of their own free will . Such motivation may be due to the endemic level of hookworm infection in the study area; that is , proximity to the disease being studied might lead the volunteer to associate participation in the clinical trial with the prospect of improving their health . In contrast , in both Belo Horizonte and in Washington , DC , the participants reported that their participation was motivated by the possibility of helping other people , perhaps driven by their reduced exposure to the disease for which the vaccine is being developed . Specifically in the case of the United States , volunteers named the financial benefit deriving from their participation as a major factor in enrolling in the research . Receiving a financial incentive for participation in clinical trials was reported as a reason for enrolling in another healthy volunteer study conducted in New Haven , Connecticut , in which 58% of the respondents reported it as the primary motivation for participating [52] . The same study identified a positive correlation between financial interest and a greater understanding of the informed consent document , an aspect that could not be evaluated in the current study given its dissimilar objectives and methodology and the fact that monetary compensation for clinical trial participation is not permitted in Brazil . The findings of our study are supported by the fact that we utilized a set of strategies employed to ensure method reliability , internal data validity , and minimization of bias . First , the questionnaires used open-ended questions to assess the participants’ knowledge of information contained in the informed consent document . This type of tool is able to measure more accurately the actual knowledge of a topic , and avoids overestimation of responses or influencing responses by presenting pre-determined options [26 , 28] . We also sought to ensure that knowledge of the information conveyed in the informed consent form derived merely from the reading of the document and was not influenced by the experience of participating in the trial . This was achieved by administering the questionnaires after the consent procedure but before the first study vaccination . In most cases the questionnaire was administered on the first day of vaccination although in a handful of cases it was administered on the same day as consent was obtained; it is possible that differences in time between signing the consent form and completing the questionnaire may have affected the observed results , however the number of participants who completed the questionnaire immediately was too small to make valid comparisons . This is a potential limitation of our research , as is the fact that the informed consent forms were not identical between the studies conducted in Brazil and in the United States due to slight differences in study design , primarily the use of different adjuvants in the two trials . However , this is unlikely to have significantly impacted our results since the major differences observed were between participants in Americaninhas and those in Belo Horizonte and Washington , DC , rather than between the Brazilian and American studies . The validity of the data was also ensured by standardizing the questions and how the responses were recorded , thereby increasing reliability and permitting more robust statistical analysis of the data [53] . Standardization was essential to compare the participants’ knowledge , especially since the questionnaires were administered in three diverse study locations . In order to ensure standardization of the process , interviewers received training prior to the application of the questionnaire . The questionnaire used in this research addressed all of the necessary aspects of an ethically sound informed consent , in contrast to the study by Ellis et al in which only seven themes were evaluated [21] . This same study pioneered the comparison between levels of quality of informed consent in developed ( USA ) and developing ( Mali ) countries . However , the investigators used a questionnaire with a primarily educational purpose that did not take into consideration the rigor of scientific research . In order to build upon the findings of this study , the questionnaire design used in our research was based on other tools validated for this purpose and on the researchers’ experience developing and administering similar surveys [7–8 , 27 , 54] . Despite the fact that international ethical guidelines and literature on the subject of informed consent call for additional measures to protect the rights of research participants in developing countries , the findings of this study suggest that the characteristics of participants at each specific study site need to be considered , regardless of the country in which they are located [1 , 26] . This observation was emphasized by Lobato [55] , who demonstrated that certain characteristics of study participants may be negatively associated with the quality of the informed consent that is obtained , particularly those associated with extrinsic or intrinsic vulnerability [16] .
The study described herein provides the first empirical comparative analysis of the quality of the informed consent process of participants in a clinical trial in a developed country with participants from Brazil . In addition , it compared the quality of the informed consent process in different Brazilian contexts . Regarding the methodology , this is the first study to investigate the theme using clinical trials of similar design and testing the same investigational products , and that measured results through a standardized questionnaire designed specifically for that purpose . From this perspective , it may contribute especially to building a body of knowledge about the quality of informed consent worldwide . As described above , potential limitations of this study include the lack of validation of the questionnaire prior to its use , as well as the different methods of its implementation in both countries . Furthermore , it is understood that a larger sample of participants could give a different result is that the same sample used could produce other results in different population groups and ages or in other countries , or in different locations within Brazil and the US given local differences . Despite this , based on our results , we conclude that the use of the terms “developed” and “developing” to describe countries is a reductionist exercise to define participants as vulnerable , whereas a rigorous consideration of the specific characteristics of each group of individuals recruited as participants in a clinical trial is necessary . These findings demonstrate also the need for educational interventions directed at clinical trial participants , both in developing and developed countries , in order to improve understanding of the informed consent document . | Informed consent is an essential element of the ethical conduct of clinical trials of new vaccines , regardless of the study setting . However , the quality of informed consent is often suboptimal . Some research has suggested that the quality of the informed consent process may be reduced in resource-limited areas compared to developed country settings . To test this , we conducted a study of the quality of the informed consent process in two similar Phase 1 clinical trials of the Na-GST-1/Alhydrogel hookworm vaccine that were conducted in healthy adult volunteers in Brazil and in the United States . In Brazil , the trial was conducted at two sites , one a large urban area ( Belo Horizonte ) , and the other a rural , resource-limited region of the state of Minas Gerais; in the United States , the trial was conducted in Washington , DC . A structured questionnaire was administered after the informed consent document was signed at each of the three clinical trial sites , which tested understanding about the information contained in the document and attitudes toward the volunteers’ participation in the clinical trial . The results indicate that there were no substantial differences between the overall quality of the informed consent obtained from participants in the United States and in Brazil . However , a significant association was found between the particular site where the trial was conducted and the quality of the informed consent process , with residents of the site in rural Brazil having the lowest percentage of correct answers on the informed consent questionnaire . The informed consent process should therefore take into account the specific characteristics of the population in which the trial is being conducted . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion",
"Conclusions"
] | [
"united",
"states",
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"health",
"services",
"research",
"helminths",
"immunology",
"geographical",
"locations",
"hookworms",
"animals",
"health",
"care",
"vaccines",
"preventive",
"medicine",
"research",
"design",
... | 2017 | A Comparison of the Quality of Informed Consent for Clinical Trials of an Experimental Hookworm Vaccine Conducted in Developed and Developing Countries |
Eukaryotic cells often use proteins localized to the ciliary membrane to monitor the extracellular environment . The mechanism by which proteins are sorted , specifically to this subdomain of the plasma membrane , is almost completely unknown . Previously , we showed that the IFT20 subunit of the intraflagellar transport particle is localized to the Golgi complex , in addition to the cilium and centrosome , and hypothesized that the Golgi pool of IFT20 plays a role in sorting proteins to the ciliary membrane . Here , we show that IFT20 is anchored to the Golgi complex by the golgin protein GMAP210/Trip11 . Mice lacking GMAP210 die at birth with a pleiotropic phenotype that includes growth restriction , ventricular septal defects of the heart , omphalocele , and lung hypoplasia . Cells lacking GMAP210 have normal Golgi structure , but IFT20 is no longer localized to this organelle . GMAP210 is not absolutely required for ciliary assembly , but cilia on GMAP210 mutant cells are shorter than normal and have reduced amounts of the membrane protein polycystin-2 localized to them . This work suggests that GMAP210 and IFT20 function together at the Golgi in the sorting or transport of proteins destined for the ciliary membrane .
Most vertebrate cells have a non-motile primary cilium projecting from their surface [1] , [2] . Defects in these organelles lead to a wide range of developmental disorders and diseases ranging from embryonic lethality in severe cases to polycystic kidney disease and retinal degeneration with less extreme alleles . These non-motile primary cilia are thought to be sensors of the extracellular environment . A number of receptors and channels have been localized to the ciliary membrane including the opsin photoreceptors of the vertebrate retina , the odorant receptors of the olfactory system , the SSTR3 isoform of the somatostatin receptor [3] , smoothened and patched , transmembrane receptors in the hedgehog signaling pathway [4] , [5] , the PDGFRα isoform of the platelet derived growth factor receptor [6] , and the polycystins and fibrocystin , products of the human polycystic kidney disease genes [7]–[9] . Little is known about how the ciliary membrane is assembled and maintained despite the fact that this membrane is vitally important for the sensory functions of cilia . While the ciliary membrane is continuous with the plasma membrane of the cell it is a separate domain with a unique complement of proteins localized to it [10] . The mechanism separating the ciliary membrane domain from the rest of the apical plasma membrane is likely to involve a membrane-cytoskeletal complex called the ciliary necklace [11] . The proteins that make up these complexes are as yet unknown , but probably help form the diffusional barrier separating the two zones . There is also a zone of condensed lipid at the base of the cilium that may contribute to the barrier [12] . Membranous vesicles containing ciliary membrane proteins appear to dock on the plasma membrane just outside of the cilium [13] , [14] . Recent studies are beginning to identify the protein machinery required for trafficking to the ciliary membrane . In C . elegans , progress has been made in identifying proteins required for transport of membrane proteins into the dendrite , which is a prerequisite step for ciliary membrane targeting in this organism , but proteins required specifically at the cilium are still unknown [15] . In vertebrates , Rab8 appears to regulate the transport of membrane proteins to the cilium as expression of dominant negative Rab8 causes opsin-containing vesicles to accumulate at the base of the cilium [16] and also prevents the formation of cilia in cultured cells [17] . Defects in proteins required for polarization of mammalian cells such as FAPP2 [12] , Crumbs3-CLPI [18] , annexin-13 , and syntaxin-3 [19] also perturb ciliogenesis , but whether these are acting directly on transport of ciliary proteins or indirectly in the formation of the apical domain is not known [12] . Smoothened transport in mammalian cells requires beta-arrestin [20] although this is not required for transport of polycystin-2 in C . elegans [15] . Intraflagellar transport ( IFT ) is responsible for assembling the non-membrane portions of the cilium ( reviewed in [21] , [22] ) but its role in movement of membrane proteins is not clear . During IFT , large complexes , composed of ∼20 proteins are transported along the ciliary microtubules under the membrane [23] , [24] . The complexes are thought to carry proteins from their site of synthesis in the cell body to sites of assembly in the cilium . The IFT particles traffic along the microtubule axoneme just under the flagellar membrane and probably interact with the membrane [25] , [26] . The nature of the connection between the ciliary membrane and the particle is not obvious as none of the known IFT particle proteins have any predicted transmembrane domains [27] . In C . elegans , membrane channels move in cilia at rates that are comparable to those of IFT , suggesting that IFT moves proteins within the ciliary membrane [28] and in Chlamydomonas , movement of a membrane associated kinase into the cilium requires IFT [29] . Levels of the transmembrane protein , polycystin-2 , are elevated in cilia when the IFT88 subunit is mutated in C . elegans [30] , mouse [7] , and Chlamydomonas [31] suggesting that IFT88 may be more important for removing polycystin-2 from the cilium than inserting it into the cilium . We previously showed that one of the IFT particle proteins , IFT20 , is localized to the Golgi complex as well as to the cilium and the peri-basal body pool . We hypothesized that IFT20 plays a role in the sorting or transport of membrane proteins processed through the Golgi complex and destined for the ciliary membrane . This idea was based on the observation that IFT20 moved between the Golgi and ciliary compartments and the demonstration that partial reduction of IFT20 by RNAi reduced the level of the membrane protein polycystin-2 in cilia [32] . In this work we sought to further our understanding of the function of the Golgi-associated pool of IFT20 by identifying proteins that interact with IFT20 at the Golgi complex . To do this , we immunoprecipitated an IFT20-containing complex from mouse kidney cells and used mass spectrometry ( MS ) to identify one of the subunits as a golgin known as GMAP210 or TRIP11 . This peripheral membrane protein was previously shown to be localized to the Golgi complex by a number of groups [33] . Beyond localization to the Golgi complex , there is little agreement in the literature about the function of this protein in mammals and it has been proposed to play roles ranging from regulating gene expression , controlling Golgi structure , and polarized secretion . To understand the in vivo function of GMAP210 , we generated a GMAP210 mutant mouse . The mutant mice are viable until birth , when they die from a pleiotrophic phenotype that includes growth retardation and lung and heart defects . Cells derived from these animals do not have structural defects in their Golgi complexes indicating that this protein is not required for Golgi organization . However , IFT20 is displaced from the Golgi complex in mutant cells indicating that GMAP210 anchors IFT20 to the Golgi membrane . In addition , the mutant cells have slightly shorter cilia and have significantly less polycystin-2 in these cilia . This suggests that GMAP210 functions with IFT20 in the trafficking of proteins to the ciliary membrane .
IFT20 is the only IFT particle protein known to be associated with the Golgi complex [32] . The identification of proteins that interact with IFT20 at the Golgi membrane is likely to yield new information about the function of IFT20 . To this end , we generated stable mouse kidney cell lines expressing FLAG-tagged IFT20 and as controls , FLAG-tagged IFT25 and FLAG-tagged GFP ( Figure 1A ) . IFT25 is a small IFT complex B subunit that is not Golgi associated ( Follit et al . , in preparation ) . FLAG-IFT20 localizes predominantly to the Golgi complex , whereas FLAG-IFT25 localizes to the cilium and basal body region as well as the cell body . FLAG-GFP is found in the cell body and is not enriched at either the cilium or Golgi complex . To identify candidate proteins that potentially interact with FLAG-tagged proteins , FLAG-tagged proteins were immunoprecipitated ( IPed ) from cell lysates using FLAG antibody coupled to agarose , fractionated by SDS-PAGE and the gels silver stained ( Figure 1B ) . Proteins found in all three lanes are background proteins that non-specifically bound to the resin whereas proteins found in the IFT20 and IFT25 extracts are likely to be IFT complex B proteins . This appears to be the case since the ∼200 kD band found in both the IFT20 and IFT25 lanes was identified by mass spectrometry ( MS ) as IFT172 . Two bands were identified in the IFT20 extract but not in either of the controls suggesting that these are IFT20 interacting proteins that are not part of complex B . We were not able to identify the larger band ( indicated by an asterisk ) , but MS identified the smaller band as a golgin protein known in mammals as Thyroid Hormone Receptor Interacting Protein 11 ( TRIP11 ) [34] , Golgi Microtubule Associated Protein 210 ( GMAP210 ) [35] , and Clonal Evolution Related Protein ( CEV14 ) [36] . The yeast orthologue is known as RUD3p [37] . To verify the interaction between IFT20 and GMAP210 , we used a monoclonal antibody against GMAP210 ( Clone 15 , BD Transduction Laboratories ) to perform inverse IPs . This antibody recognizes a single protein in extracts made from human cells ( Figure 1C , starting material ) but does not recognize the mouse orthologue . Extracts of human retinal pigmented epithelial ( RPE ) cells were IPed using the GMAP210 monoclonal Ab and a GFP monoclonal Ab ( JL-8 , Clontech ) as a negative control . The GMAP210 Ab but not the GFP Ab precipitated IFT20 ( Figure 1C ) . The IP extracts also were probed with our collection of antibodies directed against mouse IFT proteins that also recognize the human orthologues . Even though all of these proteins were present in the extract , only IFT20 was precipitated by the GMAP210 antibody ( Figure 1C ) . This corroborates the identification of GMAP210 as an IFT20-binding protein and indicates that IFT20 and GMAP210 interact independently of IFT complex B . Furthermore , IFT20 and GMAP210 extensively co-localize at the Golgi complex as would be expected for interacting proteins ( Figure 1D ) . To map the IFT20 binding site on GMAP210 , we tested whether IPing FLAG-tagged fragments of GMAP210 also brought down IFT20 and whether these FLAG-GMAP210 fragments could displace IFT20 from the Golgi apparatus by competing with native GMAP210 for binding to IFT20 . The identity of the Golgi-targeting sequence within GMAP210 is controversial , with the targeting sequence variably being located to the N- or C-terminal ends of the protein [33] so we also examined the cellular distribution of our FLAG-tagged GMAP210 constructs . Data are graphically displayed in Figure 2A , key examples of IF and IP that document the IFT20 binding site are shown in Figures 2B and 2C while IF data supporting the Golgi localization are shown in Figure S1A . Initially , we expressed GMAP210 as two fragments split at the junction between the coiled-coil and Grab domains ( JAF157 and JAF172 , Figure 2A ) . Both fragments localized to the Golgi-complex , although the N-terminal fragment ( JAF172 ) also was found in the cytoplasm ( Figure S1A ) . The C-terminal fragment ( JAF157 ) did not affect the localization of native IFT20 or bring down IFT20 in an IP . The N-terminal fragment ( JAF172 ) precipitated IFT20 and partially displaced IFT20 from the Golgi complex indicating that it contains an IFT20 binding site . We then split the JAF172 fragment into two smaller fragments . The N-terminal JAF175 fragment partially localized to the Golgi complex indicating that there are Golgi-targeting domains at both the N- and C-terminal ends of the protein ( Figure S1A ) . Thus our results explain the apparent discrepancy in the literature [33] , which can be ascribed to a non-systematic analysis of the protein in previous studies [34] , [37] , [38] . We did not precisely map the Golgi-binding domain at the N-terminus , but it is likely to involve the ALPS domain that has recently been shown to bind curved membranes [39] . The JAF174 fragment displaced IFT20 from the Golgi and was able to IP IFT20 indicating that it contained the IFT20 binding site . Expression of these GMAP210 fragments did not alter Golgi structure ( Figure S1B ) . We progressively split the JAF174 fragment into smaller pieces and tested their ability to IP IFT20 and displace IFT20 from the Golgi complex . The smallest fragment of GMAP210 that IPed IFT20 and displaced IFT20 from the Golgi contained residues 1180 to 1319 ( JAF203 ) . However , this peptide was not as effective as the slightly larger 1157 to 1319 fragment ( JAF192 ) . In all cases , the ability to IP IFT20 correlated with the ability to displace IFT20 from the Golgi complex . In contrast , the ability of the GMAP210 fragment to localize to the Golgi complex was not correlated with the presence of the IFT20 binding site . This suggests that GMAP210 localization to the Golgi complex is not dependent on IFT20 . This appears to be the case as cells lacking IFT20 still localize GMAP210 to the Golgi complex ( data not shown ) . The amino acid sequence of the IFT20 binding domain in GMAP210 is 95% identical between humans and mice while overall the two proteins are 80% identical , suggesting that there is selective pressure maintaining the IFT20-binding sequence . The IFT20 binding site is not found in the Caenorhabditis or Drosophila GMAP210 homologues . To begin to understand the in vivo function of GMAP210 , we obtained mouse gene trap ES cell line AJ0490 from the Sanger Institute [40] and used these cells to generate a mutant mouse . Cell line AJ0490 contains a splice acceptor site and a β-galactosidase-neomycin resistance gene fusion inserted into intron 4 of GMAP210 . There also is an insertion of 531 bp derived by duplication from chromosome 16 at the junction between the vector and intron 4 ( Figure 3A ) . In spite of this duplication , the rest of the gene appears intact as measured by PCR of exons and selected other regions of genomic DNA ( Figure 3A ) . Sequencing of cDNA made from the AJ0490 allele indicates that the first four GMAP210 exons are spliced to the 5′ end of the β-galactosidase message , potentially producing a fusion protein containing the first 197 residues of GMAP210 fused to the N-terminus of β-galactosidase . Real time RT-PCR of mRNA from e18 . 5 lungs indicates that the message derived from exons upstream of the insertion is found at about the same level as controls but significantly less message is made from the exons downstream of the insertion ( Figure 3C ) . Since the commercially available GMAP210 clone 15 Ab did not detect mouse GMAP210 , we generated a rabbit polyclonal directed against the C-terminal tail of the mouse protein . In extracts made from wild type and heterozygous mouse cells , this antibody recognizes a band of ∼200 kD that is likely to be GMAP210 and a cross reacting band of ∼60 kD . The observation that the 200 kD band is missing in the homozygous mutants , without the presence of any new smaller bands , suggests that the downstream exons in the mutant allele are not translated significantly ( Figure 3D ) . In addition , immunofluorescence analysis of MEFs from mutant animals did not show any staining with this antibody whereas GMAP210 was readily detected at the Golgi complex in the control MEFs ( Figure 3E ) . This data suggests that the AJ0490 allele is either null or a strong hypomorph . The GMAP210 gene trap allele causes a neonate lethal phenotype as all homozygous mutant animals were found either dead or close to death on the morning of their birth and none survived past postnatal day 0 ( p0 ) ( 25 +/+ , 56 +/− , 14 −/− from 18 litters . +/+ and +/− were genotyped at various ages between p0 and p21 , all −/− were genotyped at p0 ) . Mutants on p0 never achieve the healthy pink color of normal littermates but rather appear cyanotic or pale bluish pink ( Figure 3F ) . Mutants that were found alive were inactive but occasionally made a convulsive or a gasping like movement . Less than expected numbers of homozygous mutants were found but this is likely due to cannibalism of dead pups , since roughly Mendelian numbers of mutant embryos ( 44 +/+ , 50 +/− , 33 −/− from 18 litters ) were found at embryonic day 18 . 5 ( e18 . 5 ) , one day prior to birth . Mutant embryos at e18 . 5 were smaller than normal , weighing on average 70±9% ( n = 6 litters ) of what +/− and +/+ embryos weigh . In addition , the mutants usually had their mouths open with protruding tongues , suggesting craniofacial anomalies , and some exhibited an omphalocele or abdominal wall hernia , indicating a body wall closure defect ( Figure 4A ) . The omphalocele has also been observed by D . Beier in an independently identified allele ( David Beier , personal communication ) . We observed no evidence of polydactyly , left-right patterning defects or hydrocephaly , which are common phenotypes also associated with cilia defects . To understand the pathology causing neonatal lethality in the mutant animals , we fixed embryos at e18 . 5 , the day prior to birth , and examined them histologically . The abdominal organs did not appear to be greatly affected by the absence of GMAP210 and we did not detect any abnormalities in the kidney or pancreas . However in the thoracic cavity , both the heart and lungs were affected . To characterize the heart defect , 5 mutant and control animals were fixed in formalin and the hearts analyzed using episcopic fluorescence image capture ( EFIC ) [41] . With EFIC imaging , we generated serial 2D image stacks and 3D reconstructions that allowed detailed examination of the cardiac anatomy in multiple imaging planes ( see Video S1 , S2 , S3 ) . All five mutant hearts showed ventricular septal defects ( VSD ) . Normally , at e18 . 5 ventricular septation is complete , allowing for separate pulmonary vs . systemic circulation from the right vs . left ventricles ( Figure 4Ba , b +/+ ) . However in the mutants , muscular VSDs are found at the anterior and posterior walls of the heart , which would cause inappropriate mixing of blood ( Figure 4Bc–f −/−; see Video S1 , S2 , S3 ) . In one mutant , a VSD was observed in conjunction with an overriding aorta , which is an abnormal positioning of the aorta between the right and left ventricle . This was accompanied by a narrowing of the pulmonary outflow ( pulmonary stenosis ) and thickening of the ventricular chamber walls ( Figure 4Bf ) . Together this constellation of defects is consistent with Tetralogy of Fallot ( Figure 4Bf ) , which in humans , is a relatively common but serious congenital heart condition . The lungs showed the normal four right and one left lobe structure suggesting that the early stages of development had occurred normally . However , at e18 . 5 the mutant animals had notably smaller saccules with thicker inter-saccule mesenchyme ( Figure 4C ) . Mutants had about one third as much saccule space as littermate controls ( Wild type = 33 . 6±9% , Mutant = 12 . 9±5% , n = 5 animals for each genotype ) . At e18 . 5 mouse lungs are normally in the terminal sac stage of development . During this stage , which lasts from e17 . 5 to p5 , the lung mesenchyme thins to bring the capillaries next to the prospective alveoli and the alveolar type I and II cells differentiate [42] . During their differentiation , the Type I cells flatten to reduce the distance between the capillaries and the air exchange surface of the saccule and the Type II cells produce surfactant for secretion into the saccules . Staining control and mutant lungs with markers for the Type I and II epithelial indicates that both types of cells are present . However in the mutant lungs , the Type II surfactant secreting cells are not as clearly interdigitated between the Type I cells and the SP-C staining is more punctate and distributed throughout the cell rather than being located at the apical end as it is in the control lungs ( Figure S2B ) . Staining with PECAM1 , to mark the endothelial cells of the capillaries , indicates that capillaries are forming in the mutant lungs like the wild type but in the mutant lungs the capillaries are less associated with the saccules ( Figure S2A ) . EM analysis indicates that the type I cells are less flattened in the mutant lungs as compared to the controls ( Figure S2C ) . Quantitative PCR was used to examine the expression levels of a number of lung development genes . Genes examined included sonic hedgehog ( SHH ) , which is critical for branching morphogenesis , VEGF-A , which is a regulator of vascular development , Hif1a and its binding partner ARNT , which regulate transcription of VEGF-A and other genes , SP-C , which encodes a surfactant molecule critical for lung function at birth , and the selenium binding protein , SelenBP1 , which is up regulated just before birth [43] . No differences were seen between the mutant and control animals ( data not shown ) . To begin to understand GMAP210s function in cells , we generated embryonic fibroblasts ( MEFs ) and kidney ( MEKs ) cells from e18 . 5 animals . All three genotypes ( +/+ , +/− , −/− ) grew at similar rates and outwardly appeared indistinguishable . Since GMAP210 was identified as an IFT20 binding partner , we sought to understand how the lack of GMAP210 affected IFT20 and cilia formation . Wild-type MEKs and MEFs ( not shown ) localize IFT20 to the Golgi apparatus ( Figure 5A , Wild Type ) as we described earlier for other cell types [32] . However , IFT20 is completely dispersed from the Golgi complex in cells lacking GMAP210 ( Figure 5A Mutant ) . This is not simply an indirect result caused by dispersal of the Golgi as the cis-medial and trans compartments of the Golgi complex appear normal in the GMAP210 mutant cells by Helix pomatia agglutinin ( HPA ) , golgin97 , wheat germ agglutinin ( WGA ) , giantin , and GM130 staining ( Figure 5B ) . The dispersal of IFT20 from the Golgi complex is caused by the lack of GMAP210 because re-expression of FLAG-tagged GMAP210 restores IFT20 to the Golgi complex ( Figure 5A Rescue ) . IFT20 protein levels are slightly reduced in the mutant cells ( data not shown ) suggesting that some IFT20 is degraded when GMAP210 is absent with the rest being distributed throughout the cell . In addition to being localized to the Golgi complex , IFT20 is also found at the centrosome [32] . We were unable to detect GMAP210 at the centrosome in either the mutant or wild-type cells ( Figure 5C top row ) and consistent with this observation , the centrosomal pool of IFT20 is not affected in the GMAP210 mutant cells ( Figure 5C bottom row ) . This suggests that IFT20 is not required to be trafficked through the Golgi complex in order to be assembled into an IFT particle . We previously showed that IFT20 is required for ciliary assembly [32] . GMAP210 in contrast , is not absolutely required for cilia assembly ( Figure 3E , 6A , Figure S2D ) . Quantification showed that GMAP210 mutant cells are ciliated nearly as well as control cells , and the level of ciliation did not increase upon re-expression of FLAG-tagged GMAP210 ( Figure 6B , left panel ) . However we did note that the cilia on the GMAP210 mutant cells were often shorter than those on control cells . Measurement of cilia length on MEK cells indicated that the cilia are only about 2/3 as long as cilia on control cells . The length difference can be restored by expression of FLAG-tagged GMAP210 indicating that this result is specifically caused by the lack of GMAP210 ( Figure 6B , right panel ) . This suggests that GMAP210 is involved in ciliary assembly . One of the proposed roles of GMAP210 is in ER to Golgi transport [44] and there is clear evidence in yeast for the involvement of the homologue Rud3p and Rud3p-interacting proteins in membrane protein transport [45] , [46] . Partial knockdown of IFT20 by RNAi reduced the amount of the membrane protein polycystin-2 on cilia suggesting that the Golgi pool of IFT20 was important for transport or retention of polycystin-2 in cilia [32] . To test the involvement of GMAP210 in ciliary transport , we measured the ciliary levels of polycystin-2 in wild-type and mutant MEKs ( Figure 6C , D ) . The level of polycystin-2 in the mutant cilia was reduced to about one quarter the amount seen in the control line . The results are also displayed as polycystin-2 per unit of ciliary length to show that this is not an indirect effect of having shorter cilia on the mutant cells ( Figure 6D right panel ) . To test if the decrease in ciliary polycystin-2 was specifically due to the lack of GMAP210 , we transfected in FLAG-tagged GMAP210 and measured the levels of ciliary polycystin-2 in the rescued cells . Rescue with FLAG-tagged GMAP210 was able to restore ciliary polycystin-2 levels to wild type levels ( Figure 6C , D ) . These results indicate that GMAP210 is important for efficient targeting of polycystin-2 to the cilium .
IFT20 is unique among the known IFT particle proteins in that it is the only one shown to localize to the Golgi complex in addition to the basal body and cilium , where the other IFT particle proteins are found . As such , it is in a unique position to coordinate the sorting or transport of ciliary membrane proteins . In prior work , we showed that an RNAi-mediated reduction of IFT20 that depleted the Golgi pool but did not greatly affect the basal body pool was sufficient to block ciliary assembly suggesting that the Golgi pool of IFT20 played an important role in ciliary assembly . We also showed that cells with a moderate reduction of IFT20 could still assemble cilia , but these cilia had less polycystin-2 in them supporting a role for IFT20 in the sorting or transport of ciliary membrane proteins [32] . In the present work , we sought to further our understanding of the function of the Golgi-associated pool of IFT20 by identifying proteins that interact with IFT20 at the Golgi complex . This analysis identified a protein called GMAP210 that binds to IFT20 . Cells lacking GMAP210 fail to localize IFT20 to the Golgi complex , indicating that this protein is the linker that holds IFT20 to the Golgi . These cells can still form cilia , but they are shorter than normal and have reduced amounts of polycystin-2 . Mice lacking GMAP210 die at birth likely from heart and lung defects . As discussed below , previously published in vitro studies have implicated GMAP210 in a wide variety of processes ranging from Golgi structure to regulating gene expression , so the ability of GMAP210 mutant embryos to progress through embryonic development was unexpected . When examined just prior to birth , the major organs , with the exception of the lungs and heart , appear fairly normal and do not show signs of cystic disease . Our finding that ciliary levels of polycystin-2 are reduced in cells derived from the GMAP210 mutant animals would suggest that these animals should develop kidney cysts . However , in other work , we found that mice lacking cilia due to a mutation in IFT20 do not show signs of cystic disease until five to ten days after birth . When the IFT20 mutant kidneys are examined at e18 . 5 , cilia are absent but there is no sign of cysts or even dilation of the kidney tubules [57] . Thus , it is likely that if the GMAP210 mutant animals were to live for a few weeks longer they would develop cystic kidney disease , but instead , animals die at birth before cysts have time to develop within the kidneys . GMAP210 mutants exhibit serious congenital heart defects ( VSD and Tetralogy of Fallot ) that are a likely cause of the neonatal lethality observed in these animals . These disorders are common in humans , where it is estimated that as many as 1% of live births have congenital heart defects with VSDs being the most common form [47] . Tetralogy of Fallot accounts for 10% of human congenital heart disease and is the leading cause of cyanotic congenital heart disease in newborns [48] . VSDs and malalignment of the great arteries also are observed in mice with mutations in Vangl2 , Dvl2 , and Scrib . These genes are members of the planar cell polarity pathway ( PCP ) , which regulates cell polarity and polarized cell movement via non canonical Wnt signaling [49]–[51]; for review see [52] . It is thought that formation of the ventricular septum is mediated by compaction of the trabeculae , with growth of the muscular septum generated by addition of sheets of trabeculae [53] . Cre mediated cell lineage tracing indicates the ventricular septum is derived from cells originating from the ventral aspect of the primitive ventricle , with closure of the ventricular foramen mediated by dorsal migration of this precursor cell population [54] . These cell migration events could be regulated by PCP signaling and thus VSDs might arise in animals with defects in PCP components . Similarly , the outflow tract alignment defects in the Vangl2 mutant hearts may involve inhibition of polarized cell migration associated with myocardialization of the outflow tract . In the Scrib mutants , PCP defects are suggested to cause abnormalities of cardiomyocyte organization , which may result in abnormal trabeculation and ventricular noncompaction [51] . Cilia mutants often show defects in PCP [55]–[57] . For example , the deletion of IFT20 in the kidney collecting duct disrupts PCP by randomizing the orientation of the cell division plane [57] . These observations suggest cardiac defects in the GMAP210 mutants could arise from dysregulated PCP signaling due to defects in the cilia . Cilia are present in the developing mouse heart from e9 . 5 onward and defects in ciliary assembly cause severe heart development defects including malformation of the trabeculae that normally contribute to the formation of the ventricular septum [58] . In the GMAP210 mouse , the cilia are not absent but are likely missing key sensory receptors , analogous to the reduction of polycystin-2 in the kidney cilia ( Figure 6 ) . The lung phenotype of the GMAP210 mutant mouse is similar to several mouse models of infantile respiratory distress syndrome caused by defects in signaling between cells of the developing lungs such as the Wnt5a [59] and the nitric oxide synthase ( eNOS ) mutant [60] . The Wnt5a mutant mouse is neonatal lethal and shares a number of features with the GMAP210 mutant mouse , including the thickened mesenchyme , reduced saccule space and a failure to organize its capillaries under the saccule epithelium . Wnt5a is a secreted ligand thought to be produced by both the mesenchyme and epithelial cells of the lung to regulate lung development [59] . It is of significance to note that Wnt5a mediates non canonical Wnt signaling in the PCP pathway . Thus the observed disorganization of the alveoli may reflect a disturbance of PCP signaling related to ciliary dysfunction . Similarly , the eNOS mouse is neonatal lethal and has reduced saccule space with thickened mesenchyme . In this case , it is thought that signaling between the endothelial and epithelial cells via eNOS influences development of the lung [60] . Recent studies suggest NO production in endothelial cells is regulated by shear stress transduced through the cilia , with polycystin-1 cleavage associated with loss of responsiveness to high shear stress [61] . Thus it is also possible that abnormal regulation of NO production due to ciliary abnormalities in the GMAP210 mutant may play a role in the lung defects . Based on the reduced amount of polycystin-2 in the kidney cilia , we speculate that the cilia present in the developing lung may be deficient in membrane-localized receptors and hence unable to respond to cues from the environment . GMAP210 is a member of the golgin family of proteins . Golgins are thought to function in the formation of the Golgi matrix , which organizes the Golgi membranes and regulates membrane trafficking . Members of this family typically have large coiled-coil regions and GRIP and GRAB domains that bind to small GTPases in the ARF and ARL subfamilies [62] . In addition , GMAP210 contains an ALPS domain , which is an amphipathic helix that binds preferentially to curved membranes . In GMAP210 the ALPS domain is at the N-terminus and is separated from the GRIP and GRAB domains at the C-terminus by a long stretch of coiled-coil suggesting that it may be able to hold small vesicles on the end of a long tether [39] . Clear homologues of GMAP210 are found throughout the vertebrates and in organisms as distantly related as Drosophila and Caenorhabditis . The yeast orthologue is reported to be Rud3p as this protein shares the same domain structure being largely coiled-coil with Grab and Grab-associated domains [37] . However sequence identity between the yeast and mouse proteins is low ( 20–24% , BLAST E = 1e-9 ) . BLAST analysis does not identify a Chlamydomonas homologue , but XP_001702340 is a coiled-coil protein containing Grab and Grab-associated domains and thus may be the Chlamydomonas orthologue . The IFT20 binding domain localizes within a 163 amino acid region of the GMAP210 coiled-coil domain . This sequence is well conserved throughout the vertebrate kingdom but is not present in the yeast , Drosophila or Caenorhabditis proteins nor is it found in the putative Chlamydomonas GMAP210 homologue . At this point , it is not known if IFT20 associates with the Golgi complex in Drosophila , Caenorhabditis or Chlamydomonas but if it does , it is likely to use a different mechanism . It is possible that the sorting mechanism of ciliary membrane proteins is fundamentally different in vertebrate cells as compared to Caenorhabditis or Drosophila since the cilia assembled by IFT in these invertebrates are found on dendrites and so ciliary trafficking requires sorting to dendrites before sorting to cilia . In vertebrates , this arrangement is found in olfactory sensory neurons but the majority of cells assemble their cilia directly from the cell body and do not require sorting to dendrites first . Dendritic sorting shares features with sorting to the basal-lateral domain [63] whereas most vertebrate cilia project from the apical surface if the cell is polarized . The proposed functions of GMAP210 in the literature fall into disparate categories of organizing the microtubule cytoskeleton , organizing the Golgi complex , regulating gene expression , and playing roles in vesicular transport . Many of these studies have either not been repeated independently or are controversial . For example , Barr and Egerer called into question the role of GMAP210 as a microtubule associated protein involved in Golgi organization [33] and our data indicating that cells lacking GMAP210 still form normal Golgi structures further brings this result into question . The strongest data on the role of GMAP210 suggests that it plays roles in vesicular trafficking within the endomembrane system . In yeast , Rud3p , the GMAP210 homologue ( with the caveats described above ) , was identified as a suppressor of mutations causing defective ER to Golgi transport [45] . Deletion of Rud3p causes glycosylation defects but the gene is not essential for viability [64] . Erv14p is required to localize Rud3p to the Golgi membrane [37] . Erv14 in yeast and its orthologue Cnih in Drosophila appear to play critical roles in polarized secretion . In yeast , Erv14 mutants retain the transmembrane protein Axl2p in the ER rather than inserting at the bud site [46] . In Drosophila , Cnih mutants retain the membrane protein Gurkin in the endoplasmic reticulum instead of secreting it at the anterodorsal corner of the oocyte [65] . Mammals have four Erv14/Cnih homologues but very little is known about their function . It will be interesting to learn if any of the Erv14/Cnih homologues are required for localization of mouse GMAP210 and IFT20 to the Golgi complex . In mammalian cells , overexpression of GMAP210 blocked the secretion of alkaline phosphatase into the medium and inhibited the retrograde transport of a KDEL-containing substrate from the Golgi to the ER [44] . The proposed function of GMAP210 in polarized secretion of proteins is interesting in the context of GMAP210 anchoring IFT20 to the Golgi complex and in being required for localization of polycystin-2 to cilia . Polarized secretion at the bud site in yeast and at the base of the cilium in other eukaryotes may be evolutionarily related and share common components . It has been proposed that the entire IFT process evolved from the coated vesicle transport system [66] . Whether this is true remains to be determined . However , it is likely that transport of membrane proteins to the ciliary membrane evolved as a specialized form of transport to the apical plasma membrane . We proposed earlier that IFT20 may function to mark vesicles that are destined for the ciliary membrane [32] . The unique ability of GMAP210 to bind IFT20 and anchor it to the Golgi membrane in addition to its ability to bind curved membranes [39] puts GMAP210 in a position to play a key role in sorting proteins to the ciliary membrane .
IMCD3 ( ATCC ) and hTERT-RPE cells ( Clontech , Palo Alto , CA ) were grown in 47 . 5% DMEM ( high glucose for IMCD3 , low glucose for hTERT-RPE ) , 47 . 5% F12 , 5% fetal bovine serum , with penicillin and streptomycin at 37° C in 5% CO2 . Cells were transfected by electroporation ( Biorad , Hercules CA ) . Stable cell lines were selected by supplementing the medium with 400 µg/ml of G418 ( Sigma , St . Louis , MO ) . Clonal lines were selected by dilution cloning after drug selection . Primary mouse embryonic fibroblasts ( MEF ) were generated by dispersing e18 . 5 embryos in trypsin then plating in 45% DMEM ( high glucose ) , 45% F12 , 10% fetal bovine serum , with penicillin and streptomycin . Mouse embryonic kidney ( MEK ) cells were made by trypsin , collagenase , and DNAse dispersion [67] of e18 . 5 kidneys and grown in the same medium as the MEFs . 24 hrs after the MEKs were initially plated , the medium was supplemented with 150 mM NaCl and 150 mM urea to select against fibroblasts and maintained until the fibroblasts were gone . Cells for immunofluorescence microscopy were grown , fixed , and stained as described [32] except that the paraformaldehyde fixation time was reduced to 15 min . For embryonic lung immunofluorescence , lungs from e18 . 5 embryos were fixed overnight at 4°C with 4% paraformaldehyde in PBS and embedded in paraffin . Sections were treated with the antibodies after antigen retrieval . Labeling of the GMAP210 and PECAM-1 antibodies was enhanced with a biotin-streptavidin layer . For electron microscopy , the lungs were fixed in 4% paraformaldehyde and 2% glutaraldehyde . Primary antibodies used included anti-tubulins ( 611β1 , GTU-88 , Sigma , St . Louis MO ) , anti-FLAG ( Sigma ) , anti-MmIFT20 , anti-MmIFT52 , anti-MmIFT57 , anti-MmIFT88 [68] , anti-MmPKD2 [7] , anti-human GMAP210 ( clone 15 , BD Transduction Laboratories ) , anti-T1α ( 8 . 1 . 1 , DSHB , Univ . Iowa ) , anti-PECAM1 ( M-20 , Santa Cruz Biotechnology ) , anti-SP-C ( FL-197 , Santa Cruz Biotechnology ) , anti-golgin97 ( CDF4 , Molecular Probes ) . Anti-giantin , anti-GM130 ( gifts from M . Fritzler , Univ . of Calgary ) , Anti-MmGMAP210 was made by expressing the C-terminal end of GMAP210 in bacteria ( residues 1761–1976 , same fragment as in JAF157 , Figure 2 ) as a maltose binding protein fusion and injecting into rabbits . Antibodies were affinity purified against the same fragment expressed as a glutathione S-transferase fusion . Alexa 488 conjugated Helix pomatia agglutinin and wheat germ agglutinin was from Molecular Probes ( Eugene , OR ) . Widefield images were acquired by an Orca ER camera ( Hamamatsu , Bridgewater , NJ ) on a Zeiss Axiovert 200 M microscope equipped with a Zeiss 100× plan-Apochromat 1 . 4 NA objective . Images were captured by Openlab ( Improvision , Waltham , MA ) and adjusted for contrast in Adobe Photoshop . If comparisons are to be made between images , the photos were taken with identical conditions and manipulated equally . For the quantification of polycystin-2 in the cilia , the length , area , and average fluorescence intensity of the cilia was measured using the measurement tools of Openlab . To determine significance of differences , data were logarithmically transformed to normalize variance , subjected to one-way analysis of variance , followed by post-hoc analysis with a Tukey-Kramer test ( SuperANOVA , Abacus Concepts , Berkeley CA ) . Confocal images were acquired by a Nikon TE-2000E2 inverted microscope equipped with a Solamere Technology modified Yokogawa CSU10 spinning disk confocal scan head . Z-stacks were acquired at 0 . 5 micron intervals and converted to single planes by maximum projection with MetaMorph software . Bright field images were acquired using a Zeiss Axioskop 2 Plus equipped with an Axiocam HRC color digital camera and Axiovision 4 . 0 acquisition software . FLAG-tagged IFT20 , IFT25 , GMAP210 , and GFP were constructed by PCR amplifying the open reading frames and inserting them into p3XFLAG-myc-CMV-26 ( Sigma , St . Louis , MO ) . FLAG IPs were carried out on stable cell lines expressing FLAG-Tagged IFT20 ( JAF134 ) , IFT25 ( JAF143 ) , GFP ( JAF146 ) or GMAP210 ( full length = JAF205 , shorter fragments are listed in Figure 2 ) . Cells were rinsed once with cold PBS and lysed in Cell Lytic M+0 . 1% NP40 ( Sigma ) , 0 . 1% CHAPSO ( BioRad ) , plus Complete Protease Inhibitor ( Roche ) at 4° Celsius . Lysates were centrifuged at 18 , 000 g for 10 minutes and clarified lysates were incubated with Agarose beads coupled with FLAG M2 antibody ( Sigma ) for one hour . FLAG beads were washed 3 times with Wash Buffer ( 50 mM Tris , 150 mM NaCl , pH 7 . 4 ) plus 1% NP40 . Bound FLAG proteins were eluted with 200 µg/ml 3× FLAG peptide ( Sigma ) . ES cell line AJ0490 was obtained from the Sanger Center and injected into C57Bl6J blastocysts to generate chimeric mice . Chimeric mice were backcrossed to the C57Bl6J inbred line and the animals used in this study were a mix of 129 and C57Bl6 backgrounds . Embryonic ages were determined by timed mating with the day of the plug being embryonic day 0 . 5 . Genotyping was carried out with the following primer pairs: GMAPwt3 AAACAGGAGCATTTCCGAGA+GMAPwt4 AAGACATGCGCCACTATGC ( product size = 295 bp in wild type ) and GMAPmt1 GGGCATCCACTTCTGTGTTT+GMAPmt2 TGTCCTCCAGTCTCCTCCAC ( product size = 168 bp in mutant ) ( Figure 3B ) . Mouse work was approved by the UMMS IACUC . Pregnant mice were euthanized by isoflurane overdose , their uteri were removed and submerged in ice cold PBS . While remaining submerged in cold PBS , the embryos were dissected from the uteri and their chests opened . The lungs were then fixed , paraffin embedded , sectioned , stained with H&E , and photographed at 4× magnification . The percent of open space ( excluding bronchioles and vasculature ) was calculated using the measure particle function of ImageJ . Individual lungs were dissected and frozen at −80°C in RNAlater ( Qiagen Inc , Valencia , CA ) until RNA was isolated with RNeasy kits ( Qiagen ) , including on-column DNA digestion . First strand cDNA was synthesized from 1 µg of total lung RNA per mouse , using a SuperScript II First-Strand Synthesis System ( Invitrogen , Carlsbad , CA ) and random hexameric primers . PCR primers were designed to produce amplicons between 100–150 nucleotides in length , using the online primer3 web PCR primer tool ( http://fokker . wi . mit . edu/primer3/input . htm ) and the IDT Primer Express software tool ( http://www . idtdna . com/Scitools/Applications/Primerquest/ ) . PCR primers were synthesized by Integrated DNA Technologies Inc ( Coralville , IA ) and are listed in Table S1 . Real-time qRT-PCR analysis was performed using the ABI Prism 7500 sequence detection system ( Applied Biosystems , Foster City , CA ) . Each reaction contained 2 . 5 ng first strand cDNA , 0 . 1 µM each specific forward and reverse primers and 1× Power SYBR Green ( Applied Biosystems , Foster City , CA ) in a 15 µl volume . Mouse IFT20 = NM_018854 , Mouse GMAP210/TRIP11 = XM_001001171 . | The primary cilium is a sensory organelle used by cells to monitor the extracellular environment . In mouse , severe defects in primary cilia lead to embryonic lethality while less severe defects cause a pleiotrophic phenotype that includes cystic kidney disease , retinal degeneration , obesity , and hydrocephaly , among others . The sensory functions of cilia rely on proteins localized to the ciliary membrane , which is continuous with the plasma membrane of the cell . Cells have the ability to specifically localize proteins to the ciliary membrane to the exclusion of the rest of the plasma membrane . Little is known about how this is accomplished . In prior work , we showed that the ciliary assembly protein IFT20 is localized to the Golgi complex , in addition to the cilium , and we proposed that it is involved in sorting or transport of membrane proteins to the cilium . In this work , we show that IFT20 is anchored to the Golgi complex by the golgin GMAP210 . Mice defective in GMAP210 die at birth with lung and heart defects . Cells from these animals have ciliary defects , suggesting that IFT20 and GMAP210 function together at the Golgi complex in the trafficking of ciliary membrane proteins . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genetics",
"and",
"genomics/disease",
"models",
"cell",
"biology",
"genetics",
"and",
"genomics"
] | 2008 | The Golgin GMAP210/TRIP11 Anchors IFT20 to the Golgi Complex |
Genes are regulated because their expression involves a fitness cost to the organism . The production of proteins by transcription and translation is a well-known cost factor , but the enzymatic activity of the proteins produced can also reduce fitness , depending on the internal state and the environment of the cell . Here , we map the fitness costs of a key metabolic network , the lactose utilization pathway in Escherichia coli . We measure the growth of several regulatory lac operon mutants in different environments inducing expression of the lac genes . We find a strikingly nonlinear fitness landscape , which depends on the production rate and on the activity rate of the lac proteins . A simple fitness model of the lac pathway , based on elementary biophysical processes , predicts the growth rate of all observed strains . The nonlinearity of fitness is explained by a feedback loop: production and activity of the lac proteins reduce growth , but growth also affects the density of these molecules . This nonlinearity has important consequences for molecular function and evolution . It generates a cliff in the fitness landscape , beyond which populations cannot maintain growth . In viable populations , there is an expression barrier of the lac genes , which cannot be exceeded in any stationary growth process . Furthermore , the nonlinearity determines how the fitness of operon mutants depends on the inducer environment . We argue that fitness nonlinearities , expression barriers , and gene–environment interactions are generic features of fitness landscapes for metabolic pathways , and we discuss their implications for the evolution of regulation .
Gene regulation is a major factor of molecular evolution , and changes in gene expression contribute to phenotypic differences between species [1] . Expression levels are under natural selection , which results from a balance between costs and benefits for the organism . For single-cell organisms , fitness benefits include the ability to digest nutrients in different environments . The cost of gene expression , on the other hand , depends on the biophysics of protein production and of protein activity . The cost of protein production has been studied extensively [2]–[6] . However , enzymatic activities of proteins can also reduce fitness due to energy consumption or toxic effects of the reaction products . What are the relative contributions of these two effects ? How do they interact ? To address these questions , we have to understand the fitness effects of an entire metabolic pathway , in which protein production is coupled to function and growth . This is the subject of the present paper . For our analysis , we use the lactose utilization pathway in Escherichia coli , which is one of the best characterized molecular pathways [7] . It is coded in a set of genes referred to as the lac operon . Several studies have addressed fitness effects associated with expression of the lac genes . In particular , production of the lac proteins in the absence of lactose has been shown to involve a fitness cost , that is , to reduce the growth rate of a cell population [2]–[4] , [8] , [9] . This cost has been ascribed to transcription and translation of the lac genes [10] , because toxic effects of the gene products have not been observed . Growth is also reduced by the presence of inducers in the medium , even after the maximum of expression is reached [8] . This fitness cost is likely to arise from inducer transport through the cell membrane [11] , [12] . Furthermore , the lac operon has been used to study the interplay of cost and benefit in the evolution of gene expression [9] . Taken together , these observations make the lac operon an ideal system to study the coupled fitness costs of protein production and activity . Here we determine a fitness landscape of the lac pathway by a combined experimental and theoretical approach . We measure the fitness of different regulatory mutant strains in the presence and absence of the lac inducer IPTG and of the natural sugar lactose . LacY proteins act as transporters ( so-called permeases ) for IPTG and lactose ( i . e . , these molecules are substrates of LacY ) . We develop a quantitative biophysical growth model to disentangle the fitness contributions of protein production ( i . e . , transcription and translation ) and of protein activity ( i . e . , intra-cellular transport ) . The model explains the growth rate of all observed mutants in different inducer environments . Its key element is a feedback loop between the lac pathway and fitness: at constant rate of protein production , faster cell growth leads to stronger dilution of proteins and lowers the cost of protein activity . In addition , the rate of lac gene expression itself can depend on growth [13] , [14] . Similar growth feedback mechanisms have been argued to play an important role in bacterial drug resistance [15] , [16] , and to generate diversity in an isogenic population [13] , [15] , [17] . Our analysis suggests that growth feedback is a pervasive feature of the activity-dependent fitness of metabolic pathways . This feature has important evolutionary consequences . In particular , our model predicts a fitness cliff , beyond which populations cannot maintain viable growth , and an expression barrier , that is , an upper bound for protein production and activity in viable populations . As a consequence , gene regulation in metabolic pathways is likely to be under stronger selection than the mere cost of protein production would suggest .
There are two generic sources of fitness cost for a molecular pathway: the cost of protein production and the cost of enzymatic activity [2]–[6] . In the case of the lac pathway , fitness depends strongly on the presence of substrates of the lac proteins , even when these substrates cannot be used as a carbon source [8] , [11] , [12] . One such substrate is IPTG ( isopropyl-1-thio--D-galactoside ) , which is transported by LacY and induces lac expression ( see Figure 1 ) . Hence , there are two potential phenotypes affecting the fitness of the lac operon in an IPTG environment: the rate of lac protein production and the rate of IPTG transport into the cell . We measure the fitness effects of lac protein production and activity in thirteen regulatory mutants in the lac operon of Escherichia coli . Twelve mutant strains have substitutions in the lacO1 operator region , which affect expression of the lac genes , and one strain has a deletion in the gene of the repressor lacI ( see Figure 1 and Text S1 for details ) . We determine the lac protein concentration and the fitness of these mutants both with and without substrates of the lac permease LacY . Specifically , we compete each mutant strain against a reference strain with deleted lac genes . This assay defines the fitness cost of the lac pathway as the difference in growth rate , or difference in Malthusian fitness , between reference strain and mutant , ( see Materials and Methods and Text S1 for details ) . Figure 2 summarizes the results of these experiments . They show that fitness always decreases with increasing concentration of lac proteins inside the cell , but the form of this dependence depends on presence or absence of the substrate . Without substrate , the fitness cost can be fitted to a linear form , which we associate with lac protein production ( blue line ) . When substrate is added , the magnitude of the fitness cost strongly increases and its dependence on concentration becomes nonlinear ( purple line ) . The additional , nonlinear fitness cost in the presence of IPTG can be associated with the transport activity of the LacY proteins . This is shown by a control mutant with deleted lacY gene , for which we only observe the linear cost of protein production ( red dot ) . Deviations of individual data points from the fit curves can be caused by different sources of noise . Competition assays involve experimental errors , in particular for large fitness differences between strains . For example , there can be slight day-to-day differences in medium composition . Furthermore , some of the strains might have acquired mutations with a fitness effect outside the lac operator sequence , although we have controlled for random mutations elsewhere the genome ( see Text S1 ) . As a further experimental step , we test whether these results extend to lactose , which is a natural nutrient of E . coli . The sugar used to support cell growth in the above experiments is glycerol , which is a poor carbon source . Lactose supports faster growth and is known to give an advantage to cells which are able to metabolize it . With 1 mM of lactose , the wild type has a fitness benefit over the reference strain , which amounts to ( mean of 4 replicates standard error ) . To assess whether lactose metabolism also involves a cost , we construct a mutant with deleted lacZ and lacI genes . This mutant cannot use lactose and expresses lacY constitutively . In the presence of lactose , it has a fitness cost ( mean of 12 replicates standard error ) against the reference strain , which indicates that lactose and IPTG cause a similar decrease in fitness in the presence of the lac permease ( see Materials and Methods for details ) . We conclude that both the rate of protein production and the rate of protein activity ( intra-cellular transport by LacY ) are phenotypes that affect the fitness cost of the lac pathway . But what is the cause of the fitness nonlinearity in the presence of substrates , and what are its biological consequences ? To address these questions , we now describe our experiments in terms of a simple biophysical model . We use a minimal model of gene expression and inducer transport to disentangle the fitness effects of protein production and activity in a quantitative way . The underlying intra-cellular processes involve transcription and translation , uptake of substrate by active transport , and dilution by cell division . Given the complexity of these processes and their effects on cell growth , our model does not aim at a complete description . However , the model does account for a large part of the fitness variation between strains and between cellular growth conditions . At the same time , it contains only few phenotypes and few parameters , which can be inferred from our fitness measurements . Within the model , the cost of lac protein production is proportional the production rate , and we infer this rate from our measurements of fitness and LacZ concentration ( see Figure 1 and Materials and Methods ) . The cost of LacY activity has two different potential contributions: the energy consumption of the transport process ( direct transport costs ) and growth effects of the molecules transported inside the cell ( toxicity costs ) . Direct transport costs can arise from futile transport cycles: LacY transports one proton with every IPTG molecule , and ATP is consumed to pump the excess protons out of the cell . These costs are proportional to the total LacY pumping rate inside the cell , . Toxicity costs are likely to arise from an excess concentration of the transported protons , i . e . , a reduction of the intra-cellular pH value [11] , [12] . The toxicity of IPTG itself appears to be negligible ( see Text S1 and [12] ) . Toxicity costs are proportional to the steady-state concentration of the toxic molecules , which depends on their uptake rate , the rate of dilution by cell divisions , and the cell volume [18] . The excess concentration of protons is , thus , proportional to . Furthermore , the steady-state cell volume itself depends on the growth rate , [13] . The combined fitness cost of protein production and activity in the lac pathway takes the form ( 1 ) in terms of the pathway phenotypes and . Here , denotes the fitness of the reference strain with deleted lac genes ( for which ) . Our model contains a feedback loop: fitness depends on the rates and , which in turn depend on fitness . This feedback between pathway phenotypes and fitness is illustrated in Figure 3 . It has an important consequence: although the cost contributions in Equation 1 are taken to be additive at any given value of , the resulting dependence of fitness on the pathway phenotypes , , becomes nonlinear . By calibrating this model to our experimental data , we can infer the amplitudes , , and of the different cost factors . Bayesian analysis shows that there are significant fitness contributions of protein production and steady-state concentration ( with maximum-likelihood parameter values , , ) , but the data are also compatible with a larger direct cost of transport ( ) ( see Materials and Methods and Text S1 ) . As shown in Figure 2 , the maximum-likelihood model provides a good fit to the data: the fitness feedback loop quantitatively explains the cost nonlinearity observed in our experiments . We use Equation 1 to derive two representations of a fitness landscape for the lac pathway , which highlight different biological implications of its form . First , we solve this equation to display the dependence of fitness on the pathway phenotypes , , as shown in Figure 4 . Second , we display the dependence of fitness on the external IPTG concentration , , and on two genotype summary variables , which depend only on lac O1 sequence . As genotype variables , we use the maximal rate of lac protein production at a fixed growth rate of one cell division per hour , , and the ratio of repressed to unrepressed protein production rates , [19] . The resulting function , which is shown in Figure 5 , can be called a genotype-environment-fitness map . We note that the change from the phenotype variables to the genotype-environment variables depends itself on fitness . This dependence has two reasons: ( i ) The LacY pumping rate depends on the production rate , the pumping rate per LacY molecule , and fitness , , because LacY molecules are diluted by cell divisions just like the transported molecules . This generic dependence reinforces the basic growth feedback loop by dilution , which also enters Equation 1 . ( ii ) For fixed genotype and environment , the production rate itself can depend on fitness , . This growth effect on gene expression generates an additional feedback between the lac pathway and fitness , which is expected under several growth conditions [13] , [14] . Including this feedback in our model significantly improves the agreement between data and theory ( see Materials and Methods and Text S1 for details ) . The fitness landscapes of Figure 4 and Figure 5 are obtained from our model using maximum-likelihood parameters , but their shape depends only on the presence of a fitness nonlinearity ( ) . We now discuss their form and their biological implications in more detail . The phenotype-fitness landscape of the lac pathway resulting from our model is shown in Figure 4 , together with fitness measurements of different lac O1 operator mutants in different inducer environments . The experimental data are plotted as a function of the pathway phenotypes and inferred from our model; for each mutant , the dependence of these phenotypes on the IPTG concentration is indicated by a red line . Data and model consistently show that protein production and activity of the lac pathway affect fitness in a highly nonlinear way . Our model explains the nonlinearity in terms of the growth feedback mechanism contained in Equation 1 . This form of the phenotype-fitness landscape has two important aspects . First , the nonlinearity of fitness translates into epistatic interactions between the pathway phenotypes: the effect of a change in the production rate , which is proportional to the slope , depends on the transport rate , and vice versa . Second , the fitness landscape is not univalued: for some values of and , there are two possible fitness values , for others , there is none . Phenotype values in the no-solution regime cannot be attained by a cell population in steady growth . This regime is bounded by a dotted line in the plane , which marks an expression barrier for the lac genes . The barrier occurs at a finite growth rate ( in contrast to the model of ref . [9] ) . Double-valued fitness solutions and the existence of an expression barrier for given phenotype values are a direct consequence of the growth feedback loop in Equation 1 . The stability analysis described below shows that only the full-shaded part of the landscape describes viable cell populations in stationary growth , whereas the striped part is unstable . Hence , for parameter values between the dotted and the solid lines in the plane , populations can reach two different steady-state growth rates with the same lac pathway phenotypes . We now turn to the dependence of fitness on the lac O1 operator sequence and on the external inducer concentration , which are the quantities we manipulate in our experiments . To display the sequence-dependence , we use the genotype summary variables ( maximal rate of lac protein production at a fixed growth rate of one cell division per hour ) and ( ratio of repressed to unrepressed protein production rates ) . These variables reflect the double role of the operator sequence: it acts as a binding site for the repressor LacI , but it also affects other processes that lead to changes in protein production [20] . Figure 5 shows the fitness cost as a function of the maximal production rate and the IPTG concentration , . The ratio is kept fixed to its wild type value; the figure shows fitness data for the corresponding subset of strains ( see Figure S1 and Figure S2 for the full dependence of on , and ) . Again , cell populations in stationary growth cannot exist for some genotype-environment parameters; this regime is bounded by a blue line in the plane . The fitness of different mutants as a function of the inducer concentration is again shown as a family of lines . The nonlinearity in the landscape indicates that the inducer environment affects the selective effect of regulatory mutations: higher IPTG concentrations lead to increased fitness differences between mutants . This interaction between genotype and the environment is due to an increase in the pumping rate with increasing IPTG , to the coupling of uptake rate and production rate in the term , and to the growth feedback through dilution . Figure S2 further illustrates this interaction . Genotype-environment interactions in the lac operon have been observed previously [21] . Our model shows how such interactions emerge from the basic architecture of metabolic pathways . The fitness landscapes of Figure 4 and Figure 5 have a common feature: over a wide range of parameters , there are two possible fitness values . This double-valued fitness landscape is partitioned into a stable part ( full-shaded ) and an unstable part ( striped ) ; see Text S1 , Figure S3 , and Figure S4 . The stable part of the landscape describes stationary growth of viable populations; i . e . , cells with growth rates close to a point on this surface reach a steady state given by a point on the surface . A large part of the lower surface is unstable , i . e . , cells with fitness cost are unable to dilute their proteins and transported molecules fast enough to maintain stable growth . These cells will further decline in fitness , whereas cells with will increase fitness up to the stable value . The stable and the unstable part of the fitness landscape are separated by a fitness cliff , which is shown as a blue line in Figure 4 and 5 . The cliff marks an extinction threshold: If a cell population is driven beyond this cliff by mutations or environment changes , it suffers a sudden drop in fitness and cannot maintain a finite growth rate . Existence and position of the fitness cliff depend on the amount of inducer present ( see Figure 2 , Figure 5 , and Figure S2 ) . For IPTG concentrations used in our and other experiments , the cliff is far from the wild type ( Figure S2B ) . We note , however , that lactose is often used in higher concentrations , and lack of growth due to the presence of lactose ( lactose killing ) has indeed been observed [22] .
We have shown that in the presence of an inducer , the fitness cost of the lac pathway arises not only from protein production , but also from transport activity of the permease LacY . The cost is governed by a feedback loop , which is the result of two repressive interactions: protein activity results in reduced growth , and growth dilutes proteins as well as transported molecules ( see Figure 3 ) . We note that our feedback mechanism does not rely on a limitation of cellular resources to generate a nonlinear relation between lac gene expression and growth ( in contrast to the model of ref . [9] ) . This feedback produces a strongly nonlinear dependence of fitness on pathway phenotypes or on genotype and environment , as shown in the fitness landscapes of Figure 4 and Figure 5 . Both landscapes contain a fitness cliff , which is an extinction threshold for cell populations . The nonlinearity of fitness is likely to persist for any substrate of the permease LacY and sets an upper bound for its rates of expression and activity . Thus , changes in lac permease activity or expression can have strong impacts on fitness . This is consistent with the observation that lacY is under particularly strong selection [23] , as reflected notably by its low number of synonymous single-nucleotide polymorphisms [24] . The nonlinearity of fitness and its consequences are expected to hold in the presence of lactose . If the benefit conferred by lactose ( or other sugars ) also depends on its internal concentration , we expect an effect of diminishing return: the faster a cell grows , the more it will dilute lactose , which leads to a sublinear increase of fitness with lactose concentration . Hence , combining costs and benefits of the lac proteins will lead to more complex fitness landscapes; their detailed dependence on pathway phenotypes will be addressed in a future study . Importantly , the full landscapes are expected to have a fitness cliff similar to the cost landscapes derived in this paper . This might explain why induced cells grown in a chemostat die after exposure to high concentrations of lactose , a phenomenon known as lactose killing [22] . Moreover , many other metabolic pathways in microorganisms contain a membrane pump or transporter accumulating substrates inside the cell , which often uses the proton motive force as an energy source . Our results are expected to apply to these pathways as well . In particular , we note the similarity of our fitness landscapes and those of the glucose utilization pathway in yeast [25] ( see Figure 5 and Figure S2 ) . Other protein activities such as hydrolysis of substrates can produce the same type of feedback , because they also depend on internal concentrations of molecules . The shape of the fitness landscape described here has various implications for the genomic evolution of the lac pathway . Our fitness model of protein production and activity contains two types of epistasis on the operator lac O1 . Within the operator , the fitness reduction caused by two mutations that increase expression is larger than the sum of the fitness costs of either one ( see Figure 5 ) . Furthermore , the selection pressure on expression depends on the protein activity rate and , hence , on the sequence of the downstream gene lac Y . The total pumping rate of the cell also depends on the concentration of LacY substrates in the environment , which generates fitness interactions between the operator genotype and the inducer environment . In a broader context , the costs of gene expression due to protein activity and due to protein production affect the evolution of regulatory systems in a different way . Taking into account only protein synthesis , we expect the length of genes to be the main determinant of the fitness cost of gene expression . Including protein activity , however , the selective pressure against expression of a gene can depend primarily on the coding sequence of functional domains and on the environment . For the lac pathway , the cost contributions of protein production and of protein activity are of similar magnitude , and both effects contribute to selection on regulatory sequences . Generalizing the results of this study , we expect the full landscape of a metabolic network to be filled with cliffs and valleys , whose importance depends on which pathways are more active in a given environment . In addition , a metabolic pathway with growth feedback generates ubiquitous epistasis . For example , any mutation under selection has fitness interactions with mutations in the lac operon: In the presence of IPTG , deleterious ( beneficial ) mutations outside the lac pathway affect the protein production rate and the transport rate , and hence increase ( reduce ) the fitness cost of lac activity-enhancing mutations . Thus , higher-dimensional fitness landscapes including more and more metabolic phenotypes are expected to be increasingly rugged . Previous experiments have produced fitness landscapes as a function of genotype ( see for example [26] , [27] ) . This kind of fitness landscapes omits the intermediate level of phenotypes , which describes how genotype changes affect biophysical functions . Here , we record fitness as a function of well-defined phenotypes of a metabolic pathway . These can be connected to a biophysical model , which describes the dependence of fitness on the operator sequence and on the inducer concentration . Phenotype- and model-based fitness landscapes are predictive: Once the model constants are fixed by one set of measurements , the model predicts the outcome of further experiments with different input parameters . In this study , the most striking model prediction is the extinction of populations beyond a fitness cliff . Our fitness landscape also differs from previous phenotype-fitness maps , perhaps the most popular of which is Fisher's geometric model [28] . Fitting this model to fitness data is a method to infer distributions of fitness effects of mutations and of epistasic effects between mutations [29] , [30] . Fisher's geometric model contains an a priori arbitrary number of unknown molecular phenotypes . In contrast , our model contains a small number of known phenotypes associated to a specific pathway , which are shown to capture salient features of fitness variation between populations ( clearly , this does not rule out further phenotypes of this pathway affecting fitness ) . In the classical geometric model , the fitness landscape is assumed to be smooth , and different phenotypes to contribute additively to fitness . Our fitness landscape contradicts both of these assumptions: there is strong epistasis and ruggedness . These features have been extensively analyzed for genotype-fitness maps ( a well-known example is the NK model [31] ) , but the dependence of fitness on quantitative phenotypes is generally assumed to be smoother . Our study shows that strong epistasis and ruggedness can persist in phenotype-fitness landscapes . It calls for new statistical models of such landscapes , which address their broad consequences for speed and constraints of molecular evolution . An interesting example is a recent extension of the geometric model , which contains epistasis and a fitness cliff [32] . In summary , our measurements and modeling show that the lac pathway of E . coli is governed by a strongly nonlinear fitness landscape depending on phenotypes of protein production and activity . These phenotypes , in turn , depend on the lac operon genotype and on environmental parameters in a coupled way . Fitness nonlinearities and genotype-environment interactions are not specific to the system studied here , but are likely to be general features of metabolic pathways . Thus , the fitness landscape of a metabolic network is much more than a simple superposition of the cost of protein production and the benefit of protein activity . It describes the entire network as a unit of natural selection . Such system-level fitness landscapes emerge already at simplest level of cell growth and metabolism .
The background of all strains used in this study is Escherichia coli BW30270 . The lacO1 mutant strains ( summarized in Table S1 and Table S2 ) are constructed as described in [33] . First , the complete lac promoter is deleted and replaced with the chloramphenicol resistance cassette from plasmid pKD3 ( see Table S3 for a list of plasmids used in this study ) . This yields strain S4146 which is , and . The full lac promoter and 5′UTR of wild-type Escherichia coli are amplified and cloned ( see Table S4 for a list of oligonucleotides used in this study ) . Specific lac O1 mutations are inserted using PCR mediated mutagenesis [34] , and the mutant sequences are cloned in a high-copy-number plasmid ( derived from pUC12 ) . The same gene replacement method [33] is then used to replace the chloramphenicol resistance cassette in strain S4146 with the chosen lac promoter and O1 operator . The strains produced in this way are all , and . We noticed that these strains have a general lower fitness than strain BW30270 that cannot be explained by the inserted mutations ( see Figure S5 ) so we use T4GT7 mediated transduction [35] to transfer the lac mutations back to the parent background ( BW30270 ) . First , the resistance cassette from strain S4146 is transduced to BW30270 , producing strain T218 . Then , the mutated lac operon is transduced from each lacO1 mutant to T218 . The lac promoter and O1 operator are then sequenced to confirm the correct insertion of the lac operator allele . As a control for the transduction , a wild type construct is obtained in the same way ( T273 ) . It has the same fitness as BW30270 . The reference strain for the competition , the lac permease and the lac repressor mutants ( and ) are constructed as described in [33] . Strain is constructed by first deleting lac Z following [36] and then deleting lac I following [33] . Unless stated otherwise , all measurements are made in M9 minimal medium with glycerol ( 0 . 1% v/v ) as carbon source . To distinguish strains in competition , tetrazolium lactose ( TL ) medium ( 1% bacto-tryptone , 0 . 1% yeast extract , 0 . 5% NaCl , 1% lactose , 0 . 005% tetrazolium chloride and 1 . 5% agar ) is used . colonies are white and colonies are red in TL plates [37] . is also white on TL plates . To distinguish this strain from and , LB-XGal-IPTG plates are used ( 1% bacto-tryptone , 0 . 5% yeast extract , 0 . 5% NaCl , 1 mM isopropyl-1-thio--D-galactoside ( IPTG ) , 5-Bromo-4-chloro-3-indolyl--D-galactoside ( X-Gal ) and 1 . 5% agar ) . Protein concentration is estimated using a -galactosidase ( LacZ ) activity assay [38] . Since all our mutants have the same coding sequence for this protein , changes in activity reflect changes in protein concentration inside the cell . The LacZ assays are performed as described in [38] . Overnight cultures are diluted in fresh medium to an optical density at 600 nm ( OD600 ) of 0 . 05 and harvested after growth in the indicated media at to an OD600 of 0 . 3 . IPTG was added to the overnight culture and to the test cultures in the concentrations mentioned in the text . The enzyme activities are determined from at least three independent cultures . Figure S6 shows the measured LacZ levels for all strains used in this study , in absence and presence of IPTG . Fitness is measured in head to head competition as described in [39] . Briefly , frozen cultures ( stored at ) are streaked on a Luria broth agar plate and grown over night at . An isolated colony is randomly selected and grown overnight in 3 ml of the same medium used in the competition , in particular with the same amount of IPTG . Both the reference strain ( , unless stated otherwise ) and assay strains are treated in this way separately . The strains are then mixed and diluted in saline solution ( 10 mM and 0 . 85% NaCl ) , such that about 50 , 000 colony forming units ( CFUs ) of each strain are used to start the competition . The mixed dilutions are also used to count the starting titer . Cultures are grown for 24 h on 96 deep-well plates in 1 ml of medium , shaken at 150 RPM , reaching saturation ( CFUs ) . They are then diluted and plated on TL or LB-XGal-IPTG medium . We measure the Malthusian fitness , i . e . , the growth rate , of each strain in units of the growth rate of the reference strain ( such that ) . The fitness value of a mutant is inferred from a competition experiment with the reference strain , where , are the final and initial number of mutant CFUs after and before the competition , and are the corresponding numbers for the reference strain . The growth rate of the reference strain is not affected by IPTG ( see Figure S7 ) . Thus , the doubling time of the reference strain is a fixed time unit and fitness measurements across environments are directly comparable . We report the fitness cost of a mutant compared to the reference strain , ( which is proportional to its selection coefficient measured in units of doubling time , [40] ) . The strain has the same phenotype as the reference strain ( both are red on TL plates and white on LB-XGal-IPTG plates ) , so the two cannot be competed directly . Instead , we measure the fitness of this strain by competing it with . has the same fitness as the reference strain in competition in glycerol minimal medium with 1 mM lactose ( ) . As explained in the Results section , the protein production rate and the transport rate can be expressed in terms of genotypic and environmental parameters , and fitness . This map relates the fitness landscapes of Figure 4 and Figure 5 and can be obtained as follows . The first phenotype of the lac pathway , the protein production rate , has two main components: one is independent of the lac repressor ( LacI ) and the other depends on the probability of the repressor to bind the operator . The independent component is given by the direct effect of the operator sequence ( quantified by the first genetic component ) and by the growth rate ( through a function specified below ) . The LacI-dependent component of depends on the affinity of the operator sequence ( measured by the second genetic component ) and on the concentration of inducer in the environment . The dependence on has the form of a Hill function with parameters ( the half saturation constant , taken to be ) and ( the Hill coefficient , taken to be ) [18] . The protein production rate is then . We now derive the form of , and estimate and . As mentioned before , is the dependence of the production rate on the growth rate . Following [13] , [14] , is expressed relative to the fitness of a strain growing at the rate of 1 doubling/hour , , such that . Note that the reference strain has a growth rate of , so . The parameter reflects the following observation: When the growth rate changes due to nutrient quality , there is a linear inverse correlation between protein concentration ( ) and growth rate [14] , ( see Figure S8A ) . This relationship can be extended to the protein production rate , because at steady state . We choose a linear dependence of the cell volume on the fitness , ; see Text S1 and Figure S8B . Using the dependences inferred above and assuming to be independent of , the dependence of protein production on growth rate can be estimated: . We have verified that including significantly improves the agreement between model and data ( see Text S1 ) , although it is not obvious a priori that a correlation between and growth rate is relevant in the context of our experiments . The two genetic components , ( the maximal protein production rate at fixed growth rate ) and ( the ratio of repressed to unrepressed protein production rates ) , depend only on the genotype and were calculated for each strain separately . can be derived from the protein concentration and fitness measured at a concentration IPTG , where the LacI proteins cannot bind DNA ( ) . As explained above , the cell volume , the growth rate , and the effects of growth on expression affect , such that , where is the measured growth rate at 1 mM of IPTG . Similarly , can be estimated using and measured at 0 mM of IPTG: , where is fitness in the absence of IPTG . Both and are independent of the model in Equation 1 and of the growth-dependence of the volume . Inferred values of and are shown in Figure S9 and Table S1 . The parameter is related to the “repression level” defined by Müller-Hill and co-workers as the ratio of LacZ activity between strains differing only by the presence/absence of the lac repressor , [19] . Neglecting the growth difference between both strains , these quantities are inversely related , . The second phenotype of the lac pathway , the total transport rate , is the product of the number of LacY molecules in the cell and the transport rate per LacY molecule , . The number is equal to , with the protein production rate and the growth rate , because LacY molecules are diluted by cell divisions . Note that is measured for LacZ , but all proteins of the operon are produced proportionally . The ratio of LacY molecules per LacZ molecule , which is close to 3 [10] , and other numerical constants are absorbed in the coefficients , and . The transport rate depends on the external IPTG concentration , , and on the half-saturation constant for inducer uptake , [41] . An expression for can be derived from the known functioning of the permease [42] , with efflux neglected ( see Text S1 ) . We obtain , normalizing to of IPTG . The uncertainties on and are obtained by standard error propagation , assuming independent experimental errors on and ( see Text S1 ) . A possible error in the IPTG concentration is not considered , because it is expected to be small . The coefficients , and in Equation 1 are obtained by likelihood analysis of our model and the experimental data . This analysis is based on the dependence , where and are inferred for each mutant as described above . The fitting procedure and score-based model comparisons are detailed in Text S1 ( see also Figure S10 and Figure S11 ) . | The levels of protein produced by an organism are likely to change its fitness , potentially driving the evolution of genetic regulation . Importantly , protein expression generates costs as well as benefits . Here , we use a model genetic system , the lac operon of Escherichia coli , to investigate different sources of fitness costs . We find that fitness depends not only on the production rate of proteins but also on their enzymatic activity . A simple quantitative model , which is based on the biophysics of protein production and activity , accurately reproduces the experimental results and provides testable predictions . The model describes a feedback cycle between a molecular pathway and the growth rate of cells: pathway activity impedes growth , but growth itself affects the pathway . This feedback can generate dramatic effects , such as gene expression barriers , fitness cliffs , and population extinctions , which can be triggered by small environmental or genetic changes . Our results disentangle the complex interplay of protein production and activity , and they show how these processes shape the evolution of simple organisms . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"bacteriology",
"microbial",
"mutation",
"population",
"genetics",
"evolutionary",
"selection",
"microbiology",
"mutation",
"microbial",
"evolution",
"biology",
"evolutionary",
"theory",
"evolutionary",
"genetics",
"systems",
"biology",
"natural",
"selection",
"evolutionary",... | 2011 | Nonlinear Fitness Landscape of a Molecular Pathway |
Inner ear mechanosensory hair cells transduce sound and balance information . Auditory hair cells emerge from a Sox2-positive sensory patch in the inner ear epithelium , which is progressively restricted during development . This restriction depends on the action of signaling molecules . Fibroblast growth factor ( FGF ) signalling is important during sensory specification: attenuation of Fgfr1 disrupts cochlear hair cell formation; however , the underlying mechanisms remain unknown . Here we report that in the absence of FGFR1 signaling , the expression of Sox2 within the sensory patch is not maintained . Despite the down-regulation of the prosensory domain markers , p27Kip1 , Hey2 , and Hes5 , progenitors can still exit the cell cycle to form the zone of non-proliferating cells ( ZNPC ) , however the number of cells that form sensory cells is reduced . Analysis of a mutant Fgfr1 allele , unable to bind to the adaptor protein , Frs2/3 , indicates that Sox2 maintenance can be regulated by MAP kinase . We suggest that FGF signaling , through the activation of MAP kinase , is necessary for the maintenance of sensory progenitors and commits precursors to sensory cell differentiation in the mammalian cochlea .
The mammalian cochlea transduces sound using a dedicated sensory organ , the organ of Corti , which comprises of a highly ordered array of mechanosensory hair cells ( HCs ) and their associated support cells ( SCs ) . The arrangement of cochlear HCs , 3 rows of outer hair cells ( OHCs ) and one row of inner hair cells ( IHCs ) , together with SCs results from a balance between specification , progenitor expansion and differentiation [1] . The first step in HC specification is the induction of a Sox2-positive territory known as the sensory patch . Sox2 is critical for neurosensory precursor formation in the inner ear [2]–[4] and is induced by Notch signalling through its ligand Jagged ( Jag ) 1 [5]–[9] . BMP signalling [10] then specifies the prosensory domain , the immediate precursors of the HCs and SCs , from within this Sox2-positive sensory patch . At specification , the prosensory domain exits the cell cycle , expressing the cell cycle inhibitor p27Kip1 as well as other prosensory domain markers . Importantly , the prosensory domain first becomes post-mitotic at the apical end of the cochlea from E12 . 5 , spreading basally until E14 . 5 [11] , [12] . HCs and SCs are picked out from within the prosensory domain through Notch signalling from putative SCs , acting on Delta1 or Jag2 in potential HCs [5] , [13]–[15] . This lateral inhibition ensures that only some of the cells of the prosensory domain retain the transcription factor Atoh1 [16] , [17] . Atoh1 is both necessary and sufficient for HC differentiation [18] . In contrast to the apical to basal wave of cell cycle exit of the prosensory domain , the wave of differentiation occurs basally at E14 . 5 extending apically at E17 . 5 [19] . In addition to the above , fibroblast growth factor ( FGF ) signalling has also been shown to be important in the development of the cochlear HC . Conditional deletion of Fgf receptor ( Fgfr ) 1 , results in the loss of HCs [20] . This phenotype is observed to a lesser extent , when the proposed ligand for FGFR1 , Fgf20 , is deleted [21] . Ex vivo explant studies suggest that FGF signalling enhances Notch-Jag signalling after sensory patch induction [22] . However the in vivo significance of these observations and how they lead to the Fgfr1 deletion phenotype are not clear . Fgf ligand binding causes the dimerization and activation of the canonical receptor tyrosine kinase [23] . Activation , generally by phosphorylation of particular tyrosine residues in the intracellular domain of the Fgf receptor , results in recruitment of adaptor proteins that are essential for the intracellular response to the extracellular signal . Each group of phosphorylated residues mediate distinct functions , for example phosphorylation of tyrosine 766 in FGFR1 serves as a potential binding site for phospholipase C-γ ( PLCγ ) [24] . Other adaptor proteins include Fgf Receptor Substrate ( Frs ) 2 or 3 ( collectively termed Frs2/3 ) [25] , [26] . Frs2/3 recruitment and activation leads to the stimulation of multiple FGFR-dependent signaling pathways such as Ras/MAP kinase pathway , and the phosphatidylinositol-3-kinase ( PI3K ) pathway [27] . Studies into a mouse allele in which the Frs2/3 interaction motif has been deleted , reveal that Frs2/3 recruitment mediates aspects of FGFR1 signalling [28] . However , the necessity of these pathways in inner ear development had not been investigated . In this study , we found that FGFR1 signalling through Frs2/3 is necessary for prosensory formation . Even in the absence of FGFR1-Frs2/3 signalling , the prosensory domain becomes post-mitotic , however the expression of prosensory markers is impaired . This results in fewer sensory precursors , giving rise to a reduction in HC numbers . We also found that the expression of Sox2 is transient , suggesting that the strength and duration of Sox2 expression , under the direct or indirect control of FGF-mediated MAP kinase activation , commits progenitors to sensory cell differentiation .
To determine gross morphology , the inner ear from Six1enh21-Cre::Fgfr1flox/flox and Fgfr1ΔFrs/ΔFrs at E14 . 5 were examined first by paint-filling [29] . The cochlear duct of the conditional mutant ( Six1enh21-Cre::Fgfr1flox/flox ) was shorter than control ( Figure 1A and B ) . Fgfr1ΔFrs/ΔFrs also exhibited a truncated cochlear duct although the phenotype was milder than that of the conditional mutant ( Figure 1C ) . No significant difference in the formation of vestibular components was observed . A requirement for FGFR1 function in cochlear HC development had been previously shown [20] , however the mechanisms used remained unknown . We asked when FGFR1 signalling was acting during HC development , by exploiting the difference in the timing activation of of two different Cre driver lines . To first verify Cre activity , we crossed these lines with a Rosa26-flox-STOP-flox-EYFP reporter , in which the expression of EYFP is initiated after the Cre-mediated excision of the STOP , transcription terminator sequence . Six1enh21-Cre activity can be detected as early as E9 . 5 specifically in whole otic epithelium ( Figure 2A–C and G ) . In contrast , Emx2-Cre activity cannot be detected at E9 . 5 , but is active at E12 . 5 , with EYFP labelled in the almost all putative sensory organs except three semicircular ampullae ( Figure 2D–G and data not shown ) . Quantitative PCR for the deleted portion of Fgfr1 confirmed the temporal activity of the two Cre lines ( Figure 2H ) . Fgfr1 levels in the cochlear rudiment were reduced to approximately 20% of normal in Six1enh21-Cre::Fgfr1flox/flox from E10 . 5 . In contrast , Fgfr1 levels in Emx2-Cre::Fgfr1flox/flox cochleae were close to wild-type levels at E10 . 5 , falling to 60% at E12 . 5 and 20% by E14 . 5 . We thus used these lines to examine the cochlear phenotypes when Fgfr1 deletion occurred at around E9 . 5 to 10 . 5 ( using Six1enh21-Cre ) or at around E12 . 5 ( using Emx2-Cre ) . To investigate HC phenotype , whole-mount cochlear samples from E18 . 5 mice were dissected and immunostained for Myo7a . Control , wild-type , cochleae showed the typical arrangement of three rows of OHCs and one row of IHCs along the entire length of the cochlea ( Figure 3A , B ) . In Six1enh21-Cre::Fgfr1flox/flox the arrangement of HCs was altered , with those in the apical third of the cochlea more severly affected ( Figure 3C–E ) . Here the rows of HCs were discontinuous , and arranged in islands . Typically , OHCs were missing , although isolated OHCs could be found basally . The cochlear phenotype of Emx2-Cre::Fgfr1flox/flox inner ears was milder ( Figure 3F–H ) . Basally , OHC loss was less pronounced with the outer-most row most severely affected ( Figure 3F ) . Further apically , the HC row became discontinuous , and islands that were present were made up of IHCs and OHCs , with occasional additional IHCs observed ( Figure 3G ) . HCs were more sparsely distributed in the apical-most part of the cochlea ( Figure 3H ) . We next addressed the downstream pathway employed by FGFR1 during cochlear HC formation using two alleles of Fgfr1 , Fgfr1ΔFrs/ΔFrs and Fgfr1Y766F/Y766F . Y766F carries a point mutation converting a tyrosine at position 766 to a phenylalanine , rendering it resistant to phosphorylation . This has been postulated to result in a failure of PLCγ phosphorylation and thus its activation [30] . The cochlear HC phenotype of Fgfr1ΔFrs/ΔFrs inner ears closely resembled that of Six1enh21-Cre::Fgfr1flox/flox , showing the severe OHC loss apically and the islands of HCs ( Figure 3I–K ) . In contrast , surface preparations from the inner ear of Fgfr1Y766F/Y766F showed that cochlear HCs were normal ( Figure 3L ) . The correspondance of the HC phenotypes was confirmed after quantifying the number of cochlear HCs , and also compared to the previously published Foxg1-Cre::Fgfr1flox/flox [20] . The total number of HCs per cochlea averaged 2494±160 ( n = 4 ) in wild-type controls . There were 201±26 ( n = 4 ) HCs in Six1enh21-Cre::Fgfr1flox/flox inner ears , 218±44 in Foxg1-Cre::Fgfr1flox/flox cochlea ( n = 4 ) , 728±274 ( n = 6 ) in Emx2-Cre::Fgfr1flox/flox , 420±60 ( n = 5 ) in Fgfr1ΔFrs/ΔFrs , and 2532±23 ( n = 6 ) in Fgfr1Y766F/Y766F ( mean ± SD , respectively ) ( Figure 3M ) . The significant difference was also determined when comparing Six1enh21-Cre::Fgfr1flox/flox and Emx2-Cre::Fgfr1flox/flox cochleae ( p<0 . 05 ) . Given the differences in the timing of the two Cre drivers ( Figure 2 ) , these results suggest that FGFR1 signalling commences prior to E12 . 5 . Next we counted the number of IHCs and OHCs ( Figure 3M ) . By comparisons with control cochlea ( 1876±160 ) , OHC loss were evident in Six1enh21-Cre::Fgfr1flox/flox ( 22±14 , decreased by 99% ) , Emx2-Cre::Fgfr1flox/flox ( 379±144 , decreased by 80% ) , and Fgfr1ΔFrs/ΔFrs ( 93±77 , decreased by 95% ) , but not in Fgfr1Y766F/Y766F ( 1894±24 ) . With the exception of Fgfr1Y766F/Y766F mutants ( 626±19 ) , the number of IHCs were also reduced in FGFR1 signaling mutants; Six1enh21-Cre::Fgfr1flox/flox ( 179±25 , decreased by 72% ) , Emx2-Cre::Fgfr1flox/flox ( 349±130 , decreased by 44% ) , and Fgfr1ΔFrs/ΔFrs ( 259±61 , decreased by 58% ) , compared with wild type control ( 618±31 ) ( p<0 . 05 ) . In addition , cochlear length was decreased by 41% in Six1enh21-Cre:: Fgfr1flox/flox , by 49% in Emx2-Cre::Fgfr1flox/flox , and by 37% in Fgfr1ΔFrs/ΔFrs mutants , respectively ( Figure 3N ) . To exclude the influence of cochlear length on total HC number , we counted the number of each HC type normalized to 100 µm length ( Figure 3O ) . IHCs were decreased ( p<0 . 05 ) by 33% in Six1enh21-Cre::Fgfr1flox/flox ( 6 . 8±2 . 1 ) , by 17% in Fgfr1ΔFrs/ΔFrs ( 8±0 . 1 ) of wild type levels ( 10 . 3±0 . 1 ) . However , normalized number of IHCs was statistically the same in Emx2-Cre::Fgfr1flox/flox ( 13 . 4±4 . 1 ) and wild type ( 10 . 3±0 . 1 ) . In contrast , OHC number per 100 µm was decreased by 98% in Six1enh21-Cre::Fgfr1flox/flox ( 0 . 7±1 . 2 ) , by 56% in Emx2-Cre::Fgfr1flox/flox ( 12 . 8±11 . 1 ) , and by 87% in Fgfr1ΔFrs/ΔFrs ( 2 . 0±2 . 2 ) when compared to wild type levels ( 29 . 1±0 . 4 ) . These findings suggested that FGFR1-Frs2/3 activity was required for OHC development from E10 . 5 , whereas FGFR1-Frs2/3 activity was only required for IHC development prior to E12 . 5 . Taken together , these results demonstrate that signalling via Frs2/3 recruitment is necessary for FGFR1 activity during the formation of the cochlear HCs . In addition to the cochlear HC phenotype , we analyzed the number of HCs in one of the vestibular sense organs , the utricle dissected from E16 . 5 mice . While utricilar HCs number was comparable between Six1enh21-Cre:: Fgfr1flox/+ ( 577±21 , n = 3 ) and Six1enh21-Cre:: Fgfr1flox/flox ( 550±10 , n = 4 ) ( Figure 4A , B ) , HC number was significantly decreased ( p<0 . 05 ) in Fgfr1ΔFrs/ΔFrs mutants ( 473±57 , n = 6 ) , by comparison with Fgfr1ΔFrs/+ control ( 718±81 , n = 4 ) ( Figure 4C , D ) . As this mutant is non-conditional , it may suggest that FGFR1 signalling outside of the inner ear epithelium plays a role in vestibular HC formation . The decision by sensory precursors to generate either HCs or SCs is controlled by Notch-Delta cell-cell signalling [5] , [13]–[15] . Therefore , one possible mechanism of FGFR1 action is in modifying the action of Notch and Delta in this choice . We thus investigated whether SC formation was disrupted in the absence of Frs2/3-mediated FGFR1 signalling . We crossed Fgfr1ΔFrs/ΔFrs onto an Atoh1-GFP reporter line to reveal HCs . At E18 . 5 , Prox1 is strongly expressed in the Deiter's cells and in the pillar cells [31] . In the control , Fgfr1ΔFrs/+ cochlea , Prox1-labeled 5 rows of cells ( Figure 5A ) . In mutant Fgfr1ΔFrs/ΔFrs cochlea , only two to three rows of Prox1-labelled cells were detected and were confined within sensory islands ( Figure 5B ) . In whole mount view of Fgfr1ΔFrs/+ cochlea , p75 expression was apparent in the inner pillar cells that are found along the length of the cochlear duct ( Figure 5C ) . In Fgfr1ΔFrs/ΔFrs cochlea , p75 staining was only found in the sensory cell islands highlighted by Atoh1-GFP and not found in the intervening spaces ( Figure 5D ) . Within severely affected region , the row of p75-positive cells was mostly present lateral to the one row of HCs , suggesting that these islands were exclusively IHCs . The other SC marker at this stage , Sox2 , was also only found within the sensory islands ( Figure 5E and F ) . Section analysis revealed that Sox2 is expressed in SCs , in both control organ of Corti ( Figure 5G ) and in sections taken through the level of the islands in Fgfr1ΔFrs/ΔFrs cochlea ( Figure 5H ) . In sections taken through the gaps in between the islands , we could only detect weak Sox2 expression in the Kölliker's organ , a region medial to lateral compartment ( Figure 5I ) . Combined , these results suggest that the FGFR1-Frs2/3 signalling axis also affects the formation of SCs , and is thus acting upstream of HC/SC decision mediated by Notch-Delta signalling . Precursors of auditory HCs and SCs form from a domain known as the prosensory domain [1] . This region emerges from within the Sox2-positive sensory patch between E12 . 5 and E14 . 5 , depending on the exact position within the cochlea . It is initially characterised by the cessation of mitosis , forming the zone of non-proliferating cells ( ZNPC ) , as well as the expression of a cell cycle inhibitor , p27Kip1 . Subsequently , the ZNPC expresses specific markers of the prosensory domain such as Hey2 and Hes5 . It had been previously shown that a conditional deletion of Fgfr1 regulates proliferation in the cochlea [20] . We thus asked if cell cycle regulation within the cochlea was mediated by FGFR1-mediated Frs2/3 activity . We first asked if Six1enh21-Cre:: Fgfr1flox/flox mutants used in this study recapitulated the reported cell cycle deficit shown previously in FoxG1-Cre:: Fgfr1flox/flox [20] . Prosensory domain progenitors become post-mitotic commencing at the apex at E12 . 5 and ending at the base at E14 . 5 . Cell cycle exit correlates with the expression of p27Kip1 , as was observed in whole mount preparations of control heterozygous cochlea stained for p27Kip1 and BrdU ( Figure 6A and B ) . Consistent with previous observations , no cell cycle defect was detected in Six1enh21-Cre:: Fgfr1flox/flox mutant at E10 . 5 ( data not shown ) [20] . However , a reduction in cell proliferation within the epithelial cells of the cochlea was detected in Six1enh21-Cre:: Fgfr1flox/flox mice at E12 . 5 ( Figure 6C ) . This phenotype was more prominent in Kölliker's organ at E13 . 5 and E14 . 5 . Surprisingly , and despite the proper formation of the ZNPC , p27Kip1 was down-regulated in Six1enh21-Cre:: Fgfr1flox/flox cochleae at E13 . 5 and E14 . 5 when compared to controls ( Figure 6D ) . Quantification of BrdU-labelled cells showed far fewer proliferating cells in E12 . 5 nascent cochlear duct of Six1enh21-Cre:: Fgfr1flox/flox ( 14±8: n = 5 compared with 41±8: n = 4 in wild type controls ) and E14 . 5 Kölliker's organ ( 2±2: n = 5 compared with 21±4: n = 5 in wild type controls ) ( p<0 . 05 ) ( Figure 6G ) . We next investigated proliferation in Fgfr1ΔFrs/ΔFrs cochlea . Fgfr1ΔFrs/ΔFrs mutant cochleae still exhibited down-regulation of p27Kip1 throughout cochlear duct ( Figure 6F ) . However in contrast to Six1enh21-Cre:: Fgfr1flox/flox mutant cochlea , BrdU-positive cells were observed in Kölliker's organ of Fgfr1ΔFrs/ΔFrs mutant ( Figure 6E ) . Comparable number of BrdU-positive cells were detected in Fgfr1ΔFrs/ΔFrs at both stages ( 36±8: n = 4 at E12 . 5 , and 21±5: n = 5 at E14 . 5 ) ( p<0 . 05 ) ( Figure 6G ) . We also quantified the number of BrdU-positive cells in Emx2-Cre::Fgfr1flox/flox cohleae . Reduced proliferation was only detected at E14 . 5 and was milder than that observed for Six1enh21-Cre:: Fgfr1flox/flox ( 39±1: n = 4 at E12 . 5 , and 9±1: n = 4 at E14 . 5 ) . These results indicate that Frs2/3 recruitment does not mediate FGFR1-induced cell proliferation in Kölliker's organ during inner ear development . Furthermore , these results showed that FGFR1-Frs2/3 signaling is not necessary for the formation of the ZNPC , but is required for p27Kip1 expression . The down-regulation of p27Kip1 expression in the prosensory domain indicated that even though prosensory precursors had become post-mitotic , a marker of the prosensory domain was not correctly expressed . Section analysis revealed that as well as p27Kip1 ( Figure 7A–C ) , the prosensory domain marker Hey2 was also reduced in cochlea from both Six1enh21-Cre::Fgfr1flox/flox and Fgfr1ΔFrs/ΔFrs mutants ( Figure 7D–F ) . We confirmed the down-regulation of p27Kip1 and Hey2 , as well as two other prosensory markers , Hes5 and Atoh1 , by quantitative PCR ( Figure 7P ) . The down-regulation of prosensory domain markers was significantly milder in Emx2-Cre::Fgfr1flox/flox cochleae than in either Six1enh21-Cre::Fgfr1flox/flox or Fgfr1ΔFrs/ΔFrs mutants ( Figure 7P ) . As well as the prosensory domain , the Sox2-positive sensory patch also forms Kölliker's organ and the outer sulcus . We thus asked if Fgfr1 mutation also affected these structures . Cells in Kölliker's organ normally express Fgf10 and Jag1 . In both Six1enh21-Cre::Fgfr1flox/flox and Fgfr1ΔFrs/ΔFrs mutants , the spatial expression of Jag1 ( Figure 7G–I ) and Fgf10 ( Figure 7J–L ) was unchanged . However , quantitative PCR revealed a down-regulation of Fgf10 expression , although Jag1 did not show any significant difference ( Figure 7P ) . The spatial pattern of Bmp4 , a marker for the outer sulcus located lateral to prosensory domain , was also unchanged in Six1enh21-Cre::Fgfr1flox/flox and Fgfr1ΔFrs/ΔFrs mutants ( Figure 7M–O ) . Quantitation revealed up-regulation of Bmp4 only in Six1enh21-Cre::Fgfr1flox/flox mutant but not in Fgfr1ΔFrs/ΔFrs ( Figure 7P ) . These results indicate that although cell cycle exit , an aspect of prosensory domain induction , occured normally , the induction of genes marking the prosensory domain is impaired in the absence of Frs2/3-mediated FGFR1 signalling . This signalling also contributes to the up-regulation of Fgf10 in Kölliker's organ . However , FGFR1 signalling independently of Frs2/3 recruitment , may negatively regulate Bmp4 expression in the outer sulcus . The expression of Sox2 in the sensory patch is known to be critical in the formation of prosensory domain and subsequent HC formation; mutation or reduction in Sox2 expression affects their development in a dose-dependant fashion [2] . Furthermore , FGF signalling has been shown to be sufficient for Sox2 expression [22] . We thus hypothesised that the HC phenotype observed in Fgfr1 mutants were , in part , due to alterations in Sox2 expression . Initially , Sox2 is expressed in the neuronal and sensory precursors in the otocyst at E10 . 5 . Between E12 . 5 to E14 . 5 , Sox2 expression in the cochlear duct is detected in the thickened epithelial cells that mark the site of the prosensory domain [32] . By E18 . 5 , Sox2 is confined to the SCs of the organ of Corti [4] . Sox2 was initially expressed at comparable levels between control , heterozygous , inner ears and Six1enh21-Cre::Fgfr1flox/flox mutants at E10 . 5 ( Figure 8A and B ) . By E11 . 5 , expression in Six1enh21-Cre::Fgfr1flox/flox inner ears was decreased ( Figure 8D ) , although in Fgfr1ΔFrs/ΔFrs expression levels were equivalent to those in control inner ears ( Figure 8C and E ) . By E12 . 5 , decreased expression of Sox2 in the cochlea of both Fgfr1 mutant lines was apparent ( Figure 8F–H ) , although Sox2 expression in the saccule was unchanged . To quantify this decrease , we measured Sox2 protein levels in E12 . 5 mouse cochlea . Levels were reduced by approximately 78% in Six1enh21-Cre::Fgfr1flox/flox to the levels found in Six1enh21-Cre::Fgfr1flox/+ , while a 55% decrease was observed in Fgfr1ΔFrs/ΔFrs when compared to heterozygous controls ( Figure 9A ) . This down-regulation was confirmed by immunostaining whole cochleae with Sox2 antibody ( Figure 9B and C ) . To exclude the possibility that the early Sox2 down-regulation occurred due to accelerated prosensory domain development , we used BrdU uptake to indicate its formation . At E12 . 5 , even though Sox2 is down-regulated in the cochlear rudiment of Six1enh21-Cre::Fgfr1flox/flox , BrdU-positive cells can still be detected ( Figure 9D and E ) , indicating that Sox2 down-regulation occured prior to prosensory domain formation . Furthermore , the down-regulation is not a result of cell survival: No difference in cell death was observed between controls and Six1enh21-Cre::Fgfr1flox/flox cochleae using an antibody against activated caspase-3 to detect apoptotic cells ( data not shown ) . At E14 . 5 , the onset of sensory cell differentiation , Sox2 is expressed robustly in the prosensory domain ( Figure 10A ) . Six1enh21-Cre::Fgfr1flox/flox ( Figure 10B ) and Fgfr1ΔFrs/ΔFrs cochleae ( Figure 10C ) showed weak Sox2 expression in prosensory domain . When compared to E14 . 5 heterozygous controls , Sox2 expression was decreased by approximately 66% in Six1enh21-Cre::Fgfr1flox/flox mutant cochlea , and by 49% in Fgfr1ΔFrs/ΔFrs . Only a 12% decrease of Sox2 expression levels was observed in Emx2-Cre::Fgfr1flox/flox mutants ( Figure 10D ) . To exclude the possibility that reduced Sox2 expression was as a result of reduced cell numbers , Sox2-positive cells in the prosensory domain were counted ( Figure 10E ) . No significant difference between controls ( 19 . 8±1 . 5: n = 5 ) and both Six1enh21-Cre::Fgfr1flox/flox ( 18±1 . 0: n = 4 ) and Fgfr1ΔFrs/ΔFrs ( 21 . 6±1 . 5: n = 4 ) cochleae was detected . These results indicate that reduced expression of Sox2 is independent of cell number . In addition , reduced Sox2 expression was also detected in SCs of E18 . 5 Fgfr1ΔFrs/ΔFrs cochlea ( Figure 5G–I ) . Sox2 expression in the sensory patch is induced by activation of the Notch receptor by its ligand Jag1 [6] , [14] . Expression analysis of Jag1 in Six1enh21-Cre::Fgfr1flox/flox mutant revealed that its expression pattern is unchanged ( Figure 11A and B ) , suggesting that FGFR1 signalling affects Sox2 expression independent of any affect on Jag1 regulation . Taken together , we suggest that FGFR1-Frs2/3 signalling is required for Sox2 maintenance in sensory progenitors . Frs2/3-mediated FGFR1 signalling is transduced by a number of downstream pathways . We investigated which were activated during Sox2 maintenance in the sensory patch . The MAP kinase pathway is one of the key mediators of receptor tyrosine kinase signalling , and is activated through Frs2/3 recruitment to FGFR1 [25] . To determine if this pathway was activated in the inner ear , we used antibodies specific for the di-phosphorylated form of Erk1 and Erk2 ( dpERK ) , an indicator of MAPK activity [33] , to investigate the spatiotemporal activation of this pathway in the inner ear . Our data thus far suggested that FGFR1 activity commencing prior to E12 . 5 and was necessary for Sox2 maintenance . In agreement with this timing , we detected ventral localization of dpErk in the otocyst of E10 . 5 Fgfr1ΔFrs/+ heterozygous embryos ( Figure 12A ) . In contrast , otocyst expression could not be detected in homozygous Fgfr1ΔFrs/ΔFrs embryos ( Figure 12B ) . At E11 . 5 , sections revealed ventromedial dpErk localization in the otocyst of Six1enh21-Cre::Fgfr1flox/+ heterozygous control ( Figure 12C ) but is down-regulated in both homozygous Six1enh21-Cre::Fgfr1flox/flox otocyst as well as Fgfr1ΔFrs/ΔFrs homozygote embryo ( Figure 12D and E ) . Frs2/3-mediated FGFR1 signalling also activates PI3K , which results in the phosphorylation of Akt [26] . We thus asked if this pathway was also affected in FGFR1 signalling mutants . At E12 . 5 we found no difference in the levels of phospho-Akt between Six1enh21-Cre::Fgfr1flox/+ heterozygous and Six1enh21-Cre::Fgfr1flox/flox otocysts ( Figure 12F ) . At later stages of sensory cell development , FGF8 signalling mediated through FGFR3 is thought to play a role in the specification of pillar and Deiter's cells [34] , [35] . To verify the specificity of the FGFR1 signalling mutants , we asked if ERK phosphorylation was affected at these later stages . We found no obvious difference in dpErk localization to the cells of E14 . 5 Six1enh21-Cre::Fgfr1flox/+ heterozygous and Six1enh21-Cre::Fgfr1flox/flox homozygous inner ears ( Figure 12G and H ) , where nascent pillar cells IHCs are present . Thus , inhibition of signalling by FGFR1 specifically affects early ERK phosphorylation at E10 . 5 and E11 . 5 , but does not affect later activation at E14 . 5 by other FGF receptors .
The regulation of Sox2 by FGF signaling has been characterized in a number of other systems , for example during foregut development [37] , retinal pigmented epithelia [38] , the lens placode [39] and in the differentiation of osteoblasts [40] . We show that in the cochlear precursor , FGF signalling maintains Sox2 expression . The reduction of Sox2 is not as a consequence of reduced proliferation ( and hence reduced numbers ) of Sox2-positive cells . While the number of proliferating cells in Six1enh21-Cre::Fgfr1flox/flox cochlea is reduced , the numbers in the Fgfr1ΔFrs/ΔFrs allele are not . Despite this difference , Sox2 levels are reduced in both mutants at E12 . 5 and E14 . 5 , suggesting that during cochlear HC formation one role for FGFR1 signaling is in the maintenance of Sox2 expression . Further support for the regulation of Sox2 by FGFR1 signaling comes from the correspondence of HC loss seen in Six1enh21-Cre::Fgfr1flox/flox and Fgfr1ΔFrs/ΔFrs cochlea with other mutants . Sensory cell loss is more prominent apically in the cochlea , with the phenotype becoming milder basally . Such phenotypes are similar to knockouts or hypomorphic alleles of Jag1 and Sox2 [2] , [6] , suggesting their involvement in a gene network with Fgfr1 . Indeed further support for this molecular network comes from explant studies that show that exogenous application of FGF20 can overcome Notch-Jagged-mediated inhibition of Sox2 [22] . One caveat is that it is unclear whether the regulation of Sox2 maintenance by FGFR1 signalling is direct or indirect , through the upstream regulation of other factors important in Sox2 maintenance . It is clear that further studies are necessary to determine the exact mechanism by which FGFR1 signalling regulates Sox2 . At least two roles for Sox2 have been described during the formation of the cochlear sensory cells . The above-mentioned network , apparent from E10 . 5 to E12 . 5 , maintains the competence of precursor cells to form sensory progenitors . This is supported by the analysis of the cochlear phenotype of mutant mice with little or no Sox2 . These mutants show reduced , or absent HCs in the cochlea [2] . A later role for Sox2 , from around E15 , has been proposed . Here , Sox2 maintains SC fate , and preventing ectopic HC formation through the repression of Atoh1 [4] . This is suggested by hypomorphic alleles where the reduction of Sox2 is not as severe . Here , HC number is increased [2] , [4] . Our results suggest that these two activities are separable , with FGFR1 signalling maintaining sensory commitment , partly through Sox2 regulation . The question remains , how does decreased Sox2 as a result of reduced FGFR1 signalling translate into reduced sensory cells in the cochlea ? Sox2 expression as well as other prosensory markers expressed in prosensory domain were down-regulated in both Six1enh21-Cre::Fgfr1flox/flox and Fgfr1ΔFrs/ΔFrs mutants , whereas only Six1enh21-Cre::Fgfr1flox/flox mutants showed defect in cell proliferation . Moreover , both mutants showed similar effects on the formation of HCs . We thus conclude that early cell cycle exit provides , at most , a minor contribution to the disruption of prosensory formation , and hence cochlear HC development in FGFR1 signalling mutants . Instead , it is possible that the level or duration of Sox2 expression determines the commitment or competence to form HCs . A number of studies have described the quantitative requirement for Sox2 in other systems such as in the retinal progenitors [41] , anterior foregut [37] and in taste buds [42] . Indeed , over-expression studies have suggested this is also the case in HC [4] . One possible mechanism , through which the duration of Sox2 expression in progenitors and precursors may be translated into effects on commitment and differentiation , is suggested from work on the effects of Sox2 binding to target gene enhancers in other systems [43] , [44] . Here silenced genes , important for cell type differentiation , are pre-bound with Sox2 . Pre-binding is thought to be associated with the generation of local epigenetic changes [44] or is required for successive binding of co-operative factors [43] , important in gene activation , priming the genes for activation . Consistent with this is data showing Sox2 binding sites in the Atoh1 , a gene that is responsible for sensory cell differentiation in the inner ear [45] . Similarly , we suggest that one function of maintained early Sox2 expression , controlled by FGFR1 signalling , is to prime prosensory genes , such as Atoh1 , for subsequent activation and thus control the differentiation of the sensory cells . The disruption of the transition from Sox2-positive sensory progenitors to prosensory precursors also provides an explanation for the discontinuous “island” phenotype of HCs in the cochlea of FGFR1 signalling mutants . Convergent extension movements that partially drive cochlear extension normally distribute sensory precursors over the length of the organ of Corti [46] , . However the fewer numbers of precursors in FGFR1 signalling mutants cannot be evenly dispersed . The apical to basal difference in the distribution of the sensory cells in these mutants may suggest directionality for these rearrangements . Several studies have proposed FGF20 as the FGFR1 ligand during mouse cochlear development [21] , [48] . Indeed there is good correlation of the phenotype between Fgf20−/− mutants and Emx2-Cre::Fgfr1flox/flox described in this study; both have moderate reduction in the number of OHC , and IHC remains unaffected . In addition , their prosensory domain formation is largely unaffected . In contrast , there are important differences between Fgf20 nulls and both Six1enh21-Cre::Fgfr1flox/flox and Fgfr1ΔFrs/ΔFrs mutants . In these more severe Fgfr1 mutants , HC number is more severely reduced and IHC are also affected . Analysis of Fgf20 nulls revealed a function for Fgf20 in HC differentiation since undifferentiated Sox2-positive cells between sensory islands have been reported [21] . In Fgfr1ΔFrs/ΔFrs mutant cochleae , however , there are no Sox2-positive cells detected in the lateral compartment among the HC islands . Furthermore , and in contrast to Fgf20−/− mutant cochlea , Sox2 is down-regulated in both Six1enh21-Cre::Fgfr1flox/flox and Fgfr1ΔFrs/ΔFrs mutants from E12 . 5 to at least E14 . 5 , and prosensory domain formation is disrupted . Our use of the two Cre drivers suggests a reason for this discrepancy . We propose that the FGFR1 has at least two distinct functions in auditory HC development . An early role , prior to E13 . 5 , is in the maintenance of prosensory function , in part through the regulation of Sox2 , and in the development of IHC . A later role , in OHC development , is demonstrated by the use of Emx2-Cre , which only reaches the same level of driver activity as Six1enh21-Cre at E14 . 5 . Here , Sox2 expression in prosensory domain is not severely affected despite significant reduction in OHC numbers . This suggests that a second Fgf ligand , operating either earlier or in combination with Fgf20 , is required for the maintenance of Sox2 . Although Fgf20 is expressed in the sensory patches from E10 . 5 to E14 . 5 [21] , [22] , it is likely that prosensory development , but not OHC development , could be compensated by the second ligand in Fgf20−/− mutant cochlea . A number of Fgf ligands are expressed in the inner ear at these stages of development . Fgf3 , -4 , -5 , -9 , -10 , -16 , as well as Fgf20 are all detected in the mammalian inner ear at early stages [21] , [48]–[54] . Receptor specificity can be used to narrow down the likely early ligand for FGFR1 . It is known that mutation of the Fgfr1-IIIb isoform does not affect inner ear development , thus it is likely that the Fgfr1-IIIc isoform is operating in the sensory epithelium [20] . Of these 7 ligands , FGF4 , -5 , -9 , -16 , and FGF20 can bind and signal through FGFR1-IIIc [55] , [56] , suggesting that one or more of these FGF molecules may act with FGF20 to maintain early Sox2 expression . FGF signalling triggers a downstream response , transducing external cues into an internal response . We find that in the absence of Fgfr1 , or Frs2/3-mediated FGFR1 signalling , MAP kinase phosphorylation is attenuated , suggesting that this pathway is necessary for sensory progenitor maintenance . The similarity of the Six1enh21-Cre::Fgfr1flox/flox phenotype with that of Fgfr1ΔFrs/ΔFrs suggests that adaptor proteins Frs2/3 transduce the FGF signal during sensory progenitor maintenance . However there is an important difference between the two mutants . The defect in proliferation seen in Six1enh21-Cre::Fgfr1flox/flox ( and previously in Foxg1-Cre::Fgfr1flox/flox [20] ) is rescued in Fgfr1ΔFrs/ΔFrs . This suggests the involvement of another downstream pathway in control proliferation in the cochlea . Indeed , the recovery of cell cycle impairment in Fgfr1ΔFrs/ΔFrs is consistent with previous findings that cell lines obtained from Fgfr1ΔFrs/ΔFrs are still capable of proliferating [28] . It is likely that other binding partners of FGFR1 , such as Grb14 , Crk , and Shc , which are known to regulate FGFR1-dependent cell proliferation may respond to mitogenic stimulation in the developing cochlea [57]–[59] . In contrast to the Fgfr1ΔFrs/ΔFrs , which lacks the Frs2/3 interaction motif on FGFR1 , mice carrying a point mutation in tyrosine at position 766 , Fgfr1Y766F/Y766F mice , showed no defect in inner ear development . Previous reports have suggested that Y766 phosphorylation may act to negatively regulate FGFR1 activity [30] . It is likely that FGFR1 activity is up-regulated in the inner ear of Fgfr1Y766F/Y766F mutants . Given that previous studies have suggested that exogenous FGF ligands do not result in an obvious phenotype in the normal mouse cochlea [21] , our observation of a normal cochlea in Fgfr1Y766F/Y766F mice is not unreasonable . Our analysis of a mutant of Frs2 in which its subsequent binding to Shp2 is impaired ( Frs2α/2F ) revealed a very early defect in inner ear development , with the inner ear arrested at the otocyst stage ( unpublished observations ) . This phenotype is more reminiscent of the Fgfr2 ( IIIb ) mutant , which is thought to mediate signalling from Fgf3 and Fgf10 during inner ear induction [60] . This indicates that Frs2/3-mediated FGFR signalling , like FGF signalling itself , is re-iteratively employed during inner ear formation .
Experiments were conducted and mice were housed , in accordance with local ( RIKEN CDB ) and national guidelines for animal experiments . Full details of the construction of Six1enh21-Cre mice will be presented elsewhere ( S . S and K . K . , in preparation ) . Briefly , a transgene was constructed in which the otic/epibranchial progenitor domain ( OEPD ) enhancer of the Six1 homeobox gene ( Six1enh21 ) [61] was placed upstream of Cre recombinase . Transgenic males were crossed with Rosa26-flox-STOP-flox-eYFP reporter females [62] and embryos were collected at stages E8 . 5 to E11 . 5 , LacZ expression was found in the otic/epibranchial progenitor domain ( OEPD ) as early as E8 . 5 . At subsequent stages ( E9 . 5 to E11 . 5 ) , LacZ expression was detected in the otic vesicle and epibranchial placodes/ganglia , scattered cells in the epibranchial ectoderm , the pharyngeal pouch endoderm as well as the olfactory placode/epithelium . The transgenic mouse line , mSix1-21-NLSCre ( Acc . No . CDB0466T: http://www . cdb . riken . jp/arg/TG%20mutant%20mice%20list . html ) , and is available from the RIKEN BioResource Center ( BRC ) . Mice were housed in accordance with local and national guidelines for animal experiments . The Fgfr1flox mutant mice have been described previously [20] . Fgfr1Y766F mice were generated by crossing Fgfr1n15YF with the ubiquitously expressed Cre from EIIa-Cre [30] . Fgfr1ΔFrs mice have been previously described [28] . The Rosa26-flox-STOP-flox-eYFP was obtained from Jackson Laboratory ( Bar Harbor , ME ) . The Atoh1-GFP line was provided by Dr . Jane Johnson [63] . Emx2-Cre mice were provided by Dr . Shinichi Aizawa [64] . FoxG1-Cre mice were provided by Jean Herbert , via Carina Hanashima [65] . Frs2α2F/2F were as described previously [66] . Staged mouse heads were fixed in 4% paraformaldehyde for 1–4 . 5 hours , depending on stage , and then prepared and mounted for cryo-sectioning . Immunofluorescence was performed as has previously been described [67] . The following antibodies were used: anti-p27Kip1 ( #RB-006-P , Thermo Scientific , Fremont , CA ) , anti-Sox2 ( #AB5603 , Millipore , Temecula , CA ) , anti-pErk1/2 ( #4370 , Cell Signalling , Beverly , MA ) , anti-Hey2 ( gifted by Neil Segil , House Ear Institute , Los Angeles , USA ) , anti-Prox1 ( #AB5475 , Millipore , Temecula , CA ) , anti-p75 ( #AB1554 , Millipore , Temecula , CA ) , anti-BrdU ( #555627 , BD Pharmingen , Franklin Lakes , NJ ) , anti- Jag1 ( #sc-6011 , Santa Cruz Biotechnology , Santa Cruz , CA ) , anti-GFP ( #04406-26 . Nacalai Tesque ) , anti-Caspase-3 ( #G748A , Promega , Madison , WI ) , and anti-Myosin7a ( #25-6790 , Proteus , Ramona , CA ) . For BrdU staining , the specimens were pre-treated in 2N HCl for 20 min at 37°C , and neutralized with 0 . 01M PBS ( pH 8 . 5 ) for 10 min at room temperature . For whole-embryo dpERK staining , fixed embryos were dehydrated in a graded methanol series and then treated with 5% H2O2 for 1 hr . Rehydrated embryos were processed as previously described [68] . Signal was detected using DAB substrate kit for peroxidase ( Vector ) . Alexa-488 , Alexa-594 , or anti-rabbit-HRP ( Dako ) conjugated secondary antibodies were used to detect primary antibodies . F-actin was detected using phalloidin conjugated to Alexa-488 ( Molecular Probes ) . For cochlear and vestibular HC counting , E16 . 5–E18 . 5 samples stained with Myo7a or expressing Atoh1-GFP were used since most Fgfr1 mutants die before birth . Inner and outer HC were distinguished by location and morphology as described previously [21] . Group of single row of HCs was regarded as IHCs since they were located medial side of p75-expressing pillar cells . Relative cochlear length was measured using ImageJ software . For evaluation of HC number per length , we counted more than 300 µm regions of the base , middle , and apex of the cochlea and normalized counts to 100 µm ( more than n = 4 in each HC type ) as described [21] . For Sox2-positive cell counting , cross sections from E14 . 5 samples were made and middle turn of cochlear duct stained with anti-Sox2 antibody was chosen . RNA in situ hybridization on cryo-sections was performed as previously described [69] . The gross anatomy of bony labyrinths at E14 . 5 was visualized by paint-filling as previously described [29] . Briefly , decapitated heads were fixed in Bodian's fixative over night . Specimens were subsequently immersed in a graded ethanol series to dehydrate , and cleared in a 2∶1 mixture of benzyl benzoate and benzoic acid ( BABB ) . The inner ears were visualized by injection of 1% white paint in BABB into the common crus . BrdU ( 100 mg/g body weight ) was injected into pregnant mice intra-peritoneally at E10 . 5–E14 . 5 . BrdU injected mice were sacrificed 2 hours after injection , and then fixed in 4% PFA . BrdU-positive cells were counted in at least four cross sections of the cochlear apical ( at E12 . 5 ) or middle ( at E14 . 5 ) turn . For E14 . 5 samples , only BrdU-labelled cells in Kölliker's organ were counted . Data shown are mean ± SD . P-values were calculated using unpaired t-test , to determine the significance of the difference between experimental and control samples . Whole otocyst or cochlear epithelial cells were dissected from embryos at E10 . 5–E14 . 5 ( at least n = 2 in each sample ) . Enzymatic treatment was conducted to remove mesenchyme [70] . Total RNA from pure otic epithelial cells was extracted using the RNAqueous-Micro kit ( #AM1931 , Ambion , Austin , TX ) and then reverse-transcribed using First Strand cDNA Synthesis Kit for RT-PCR ( #11483188001 , Roche , Indianapolis , IN ) . Synthesized cDNA and primer sets were mixed with Power SYBR Green PCR Master Mix ( #4367669 , Applied Biosystems , Warrington , UK ) , and real-time quantitative PCR was performed using an ABI Prism 7900 Sequence Detection System ( Applied Biosystems ) . All reactions were carried out in duplicate . The relative amount of mRNA was calculated by standard curve method , and normalized to that of 36B4 mRNA [71] . P-values were calculated using unpaired t-test , to determine the significance of the difference between experimental and control samples . E12 . 5 or E14 . 5 cochlear epithelial cells , purified from underlying mesenchymal cells were lysed in a buffered solution , consisting of SDS , salt , phosphatase inhibitor , and proteinase inhibitor . A mixture of lysate , sample buffer , and 2-mercaptoethanol , were boiled at 98°C for 2 min and separated on a SuperSep Ace gel ( Wako ) , and subsequently transferred into PVDF membrane ( GE Healthcare ) . The following antibodies were used: rabbit anti-Akt antibody ( 1∶ 1000 ) ( #9272 , Cell Signalling Technology ) , anti-Sox2 ( 1∶ 1000 ) ( #AB5603 , Millipore , Temecula , CA ) , rabbit anti-Phospho Akt antibody ( 1∶ 1000 ) ( #9271 , Cell Signalling Technology ) , and anti-Actin antibody ( 1∶ 10000 ) ( MBL ) . Horseradish peroxidase-linked anti-rabbit IgGs were used as secondary antibodies ( 1∶10 , 000 ) ( GE Healthcare ) and revealed using Amersham ECL Prime Western Blotting Detection Reagent ( GE Healthcare ) according to the manufacturer's instruction . ImageJ software was used to compare the relative Sox2 protein amount between control and Fgfr1 mutants . | The ability of our brain to perceive sound depends on its conversion into electrical impulses within the cochlea of the inner ear . The cochlea has dedicated specialized cells , called inner ear hair cells , which register sound energy . Environmental effects , genetic disorders or just the passage of time can damage these cells , and the damage impairs our ability to hear . If we could understand how these cells develop , we might be able to exploit this knowledge to generate new hair cells . In this study we address an old problem: how do signals from the fibroblast growth factor ( FGF ) family control hair cell number ? We used mice in which one of the receptors for FGF ( Fgfr1 ) is mutated and found that the expression of a stem cell protein , Sox2 is not maintained . Sox2 generally acts to keep precursors in the cochlea in a pre-hair cell state . However , in mutant mice Sox2 expression is transient , diminishing the ability of precursors to commit to a hair cell fate . These findings suggest that it may be possible to amplify the number of hair cell progenitors in culture by tuning FGF activity , providing a route to replace damaged inner ear hair cells . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"morphogens",
"medicine",
"developmental",
"biology",
"cell",
"fate",
"determination",
"developmental",
"neuroscience",
"neurogenesis",
"otology",
"signaling",
"molecular",
"development",
"hearing",
"disorders",
"biology",
"otorhinolaryngology",
"neuroscience",
"cell",
"diffe... | 2014 | FGFR1-Frs2/3 Signalling Maintains Sensory Progenitors during Inner Ear Hair Cell Formation |
At many promoters , transcription is regulated by simultaneous binding of a protein to multiple sites on DNA , but the structures and dynamics of such transcription factor-mediated DNA loops are poorly understood . We directly examined in vitro loop formation mediated by Escherichia coli lactose repressor using single-molecule structural and kinetics methods . Small ( ∼150 bp ) loops form quickly and stably , even with out-of-phase operator spacings . Unexpectedly , repeated spontaneous transitions between two distinct loop structures were observed in individual protein–DNA complexes . The results imply a dynamic equilibrium between a novel loop structure with the repressor in its crystallographic “V” conformation and a second structure with a more extended linear repressor conformation that substantially lessens the DNA bending strain . The ability to switch between different loop structures may help to explain how robust transcription regulation is maintained even though the mechanical work required to form a loop may change substantially with metabolic conditions .
DNA looping , in which a protein or protein complex interacts simultaneously with two separated sites on a DNA molecule , is a recurring theme in transcription regulation [1] . A prototypical example is transcription initiation at the E . coli lacZYA promoter , which is modulated through DNA looping by the lactose repressor . The promoter vicinity includes three operator sites: a primary operator ( O1 ) located 11 bp downstream from the lacZ transcription start site , and two auxiliary operators ( O2 and O3 ) with lower affinities for the repressor located 401 bp downstream and 92 bp upstream of O1 , respectively ( see review [2] ) . Repressor binding to O1 blocks transcription from the lacZYA promoter . Nevertheless , the presence of O2 and O3 is indispensable for complete transcriptional repression in wild-type bacterial strains because the repressor loops DNA by binding simultaneously to O1 and O2 or O3 [3–6] , and such looping enhances repression by increasing the occupancy of O1 by repressor [5 , 7] . In many transcription factors that function at least in part by DNA looping ( for example , the lambda , ara , and gal repressors [8–10] ) , the protein complex interacts with two binding sites displayed on the same face of the double helix . Both in vitro and in vivo , these systems display a characteristic dependence of repression on interoperator spacing , with strong repression when operators are separated by an integer number of helical repeats ( “in phase” ) , and repression weak or absent when an additional half turn of the helix is added ( “out of phase” ) [9 , 11–14] . The reduced repression with out-of-phase operators is consistent with simple models of DNA elasticity , which predict a substantial energy cost to twist a short interoperator DNA segment by a half turn . In contrast , the effects of operator phasing on DNA looping by Lac repressor are in general weaker than those seen with other well-characterized bacterial repressors . Also , there is strong evidence from studies in vitro ( [15 , 16] and references therein ) and in vivo [17] that stable looped repressor–DNA complexes can form with operator spacings as small as or smaller than the 92-bp O1–O3 spacing . Even spacing the operators so that they are positioned on opposite sides of the double helix only 14 . 5 and 15 . 5 turns apart , so that substantial DNA twisting and bending may be required to close the loop , allows formation of putatively looped species , apparently with only a modest reduction in stability relative to similarly sized in-phase loops [15] . Out-of-phase operator spacings of similar size also give levels of repression in vivo consistent with looping [18] . No direct determinations of the structures of small Lac repressor–DNA looped complexes are available . The availability of crystallographic structures for the repressor alone and in complex with two DNA oligonucleotides [19 , 20] , together with studies of the thermodynamic and kinetic stabilities of Lac repressor–DNA looped complexes in vitro [7 , 21–25] , have led to the proposal of a variety of different structural models for looped protein–DNA complexes [1 , 19 , 20 , 24 , 26–28] . Most of these models are based on the crystallographic repressor–oligonucleotide model and a smoothly bent interoperator DNA segment . However , the tightly bent or strongly twisted interoperator DNA in these models is predicted to be highly energetically unfavorable based on simple worm-like-chain ( WLC ) models of DNA mechanics [29] . One proposal is that the energy of bent or twisted loops is reduced by introducing a kink into the DNA [30–33] , but this idea is controversial [34] . It has also been suggested that large-scale alterations in repressor structure might help to accommodate small loops [19 , 20 , 26 , 28 , 35–39] . Nevertheless , neither the number of looped species that can form with small operator spacings nor their structure ( s ) nor their dynamics is known with any certainty . To further characterize the reactions and structural properties of small Lac repressor–DNA loops , we employed single-molecule kinetic and imaging techniques to examine loop formation from two different di-operator DNA constructs , O-158-O and O-153-O , that have interoperator separation of 15 and 14 . 5 helical turns , respectively ( Figure 1A ) . These two constructs are identical in sequence , except for the 5-bp insertion in O-158-O . To characterize the conformation of the Lac repressor–DNA complexes formed with these DNAs , we used atomic force microscopy ( AFM ) to directly visualize individual complexes . To follow loop formation and breakdown in single DNA molecules , we used tethered particle motion ( TPM ) single-molecule light microscopy [40 , 41] , which can directly observe DNA looping events mediated by Lac repressor [23 , 42 , 43] and other proteins [44–49] by monitoring the extent of Brownian motion of a microscopic bead attached to a single surface-immobilized DNA molecule . AFM images of individual surface-immobilized looped complexes revealed that the DNA is wrapped around Lac repressor , and analysis of loop formation and breakdown by TPM demonstrated the presence of multiple looped structures in equilibrium with one another ( in the case of O-158-O ) , defined the kinetic mechanisms of looping , and provided additional structural information . The data imply novel structures for some of the loops and suggest that the Lac repressor–DNA system is capable of adopting multiple polymorphic structures that help to lessen the mechanical strains inherent in forming small loops .
AFM is an excellent method for examining protein-induced conformational changes in DNA because it allows direct visualization of individual protein–DNA complexes . It has been used to assess DNA looping by other proteins [50–56] . We used AFM to examine the complexes of Lac repressor with either of the two di-operator DNAs . A representative image of O-153-O DNA deposited in the presence of Lac repressor is shown in Figure 1B . Both free DNA and DNA with Lac repressor bound can be seen . Because we know the positions of the operator binding sites in the DNA , we can determine whether Lac repressor is bound to a single operator site , to two operator sites simultaneously ( looping the DNA ) , or to nonspecific sites by measuring the contour length of the DNA and the distance of Lac repressor from the end of the DNA . The contour length ( see Materials and Methods ) measured for images of DNA molecules without repressor molecules bound is 169 ± 8 nm ( mean ± standard deviation [S . D . ]; n = 105 ) , close to the expected contour lengths of the O-153-O and O-158-O DNAs ( 183 and 185 nm , respectively ) . This confirms that the image analysis method reliably reports DNA contour length . For DNA with protein bound , the contour length of each DNA arm was measured from the center of the Lac repressor protein . The two operator sites are each roughly the same distance from the DNA ends . Thus , if repressor is bound to either site alone , it will have a long arm and a short arm , with the sum of the arm lengths equal to the free DNA length . Alternatively , if repressor is simultaneously bound to both sites , both arms should be of approximately equal length and the sum of their lengths should be approximately 150 bp shorter than that of free DNA . Figure 1C shows a plot of the contour length of the longer arm versus the shorter arm for all DNAs with a single Lac repressor bound . For a little more than half ( 205 of 364 ) of the complexes , the sum of the lengths of the two arms falls within two S . D . of the length of free DNA ( the region delineated by black dashed lines in Figure 1C ) , consistent with Lac repressor being bound to linear unlooped DNA . Of these 205 complexes , most ( 136 ) fall into a distinct , small cluster with the combination of longer- and shorter-arm contour lengths predicted for a linear repressor–operator complexes ( Figure 1C , green squares ) . The clustering of the points at this particular position shows that many of the 205 complexes have repressor specifically bound at the operator sequence , rather than associated nonspecifically with the DNA or surface . The remainder of the complexes with the sum of arm lengths equal to 169 ± 16 nm have a more uniformly distributed combination of arm lengths , as would be expected for complexes in which the repressor is bound to the linear DNA in a sequence-independent manner . In addition to these linear repressor–DNA complexes , we observed another cluster ( Figure 1C , red square ) of complexes with approximately equal arm lengths but shorter total DNA length ( ∼120 nm ) . Both the individual arm lengths and their sum agree with those expected for specific looped complexes in which a repressor binds simultaneously to both operator sites ( see Materials and Methods ) . These results demonstrate that Lac repressor can form looped complexes on both in-phase and out-of-phase di-operator molecules , even with small loop sizes . Single-operator DNA bound to Lac repressor is not appreciably bent [20] . Consistent with the expected structures , we found the angles at which the DNA arms exit from the protein to be significantly more acute for the looped than the unlooped complexes ( see Materials and Methods ) . Although Lac repressor–DNA looped complexes with short operator spacings have been examined by electron microscopy [15] , they have not been imaged with sufficient resolution to determine the position of the DNA relative to the protein in the three-dimensional structure of the complexes . Two kinds of models have been proposed: one in which the bulk of the repressor is positioned external to the looped segment of DNA ( Figure 1D ) ( for example , [26] ) and one in which the repressor is positioned in the center of the DNA loop , possibly making stabilizing contacts with the looped segment ( Figure 1E ) ( for example , [24] ) . To determine whether the AFM images are capable of differentiating between these two types of proposed structures , we modeled repressor and DNA as simple geometrical solids ( with overall dimensions corresponding to those determined crystallographically ) positioned flat against a surface , with the repressor within ( Figure 1E ) or outside of ( Figure 1D ) the DNA loop . The corresponding AFM images were then computed using a numerical simulation that accounts for the image distortion caused by the shape and finite width of the AFM tip [57] . The simulated images clearly show that these two conformations should be differentiated by AFM . Although the images of looped repressor–DNA complexes have a variety of shapes ( Figure 1F ) , we do not observe any complexes ( out of 102 analyzed ) that are consistent with the simulated image ( Figure 1D ) in which the repressor protein lies outside of the DNA loop . In contrast , many images similar to that simulated for the repressor lying within the loop are seen ( e . g . , Figure 1F ) . Assuming that at least some complexes adhere to the surface in the orientation modeled , the AFM images exclude the structure of Figure 1D and strongly favor a model in which the repressor is positioned within the DNA loop . To characterize the dynamics of Lac repressor–induced DNA looping , we used TPM to monitor changes in effective DNA length in real time . Before examining Lac repressor–induced DNA looping with constructs containing two operator sites ( Figure 2A ) , we performed control experiments using two DNA fragments , each containing a single operator site ( O-539 and O-349; Figure 1A ) . In the presence of repressor , both DNAs exhibited effective lengths that are indistinguishable from the tether lengths of the DNA in the absence of repressor ( Figure 2B and 2E , and unpublished data ) . Thus , repressor binding to a single operator site or to nonspecific sites does not change the effective length of the DNA tether . This observation confirms that Lac repressor does not induce appreciable bending in single-operator DNA [45 , 58] . In contrast , for beads tethered by two-operator O-153-O DNA in the presence of Lac repressor ( Figure 2A ) , time records of bead Brownian motion ( expressed as effective DNA tether length; Figure 2C ) alternate stochastically between a long-tether-length ( unlooped ) state and a short-tether-length ( looped ) state . This parallels previous observations using di-operator DNAs with larger interoperator separations [23 , 42 , 43 , 59] . Tether length distribution histograms from individual data records ( e . g . , Figure 2F ) typically have two peaks . In the example shown , the peak centered at approximately 525 bp corresponds to unlooped O-153-O ( length 539 bp ) and a second peak at approximately 310 bp corresponds to the looped DNA . Measured TPM peak positions can vary slightly from molecule to molecule because of small differences in positioning of the molecules with respect to the microscope focal plane [41]; it is therefore more precise to measure peak spacings than to measure absolute positions . In a set of 16 records , the mean spacing between the two histogram peaks was 204 ± 37 ( S . D . ) bp . The observations of two TPM peaks with a well-defined spacing further support the AFM data demonstrating that Lac repressor can form specific looped complexes even when the operators are out of phase . Unexpectedly , when the spacing between the operators was increased by 5 bp to 158 bp , three ( not two ) discrete tether lengths are observed in the presence of repressor ( Figure 2D ) in 49 out of 70 records . In the other 21 records ( most of which are of comparatively short duration ) , only two of the three histogram peaks were clearly discernable . In the example shown , the peaks are at approximately 500 , 360 , and 250 bp ( Figure 2D and 2G ) . The observed tether lengths of approximately 500 bp are assumed to arise from the nonlooped states of the 544-bp-long DNA . Consistent with this assumption , only tether lengths of approximately 500 bp were observed with both the O-153-O and O-158-O constructs in control experiments in which 0 . 2 mM isopropyl β-D-thiogalactoside ( IPTG ) was added to block repressor–operator binding or in which repressor was omitted ( unpublished data ) . In the 36 tether-length records in which the positions of the three peaks could be most reliably measured , the positions of two shorter tether-length peaks were 114 ± 29 ( S . D . ) bp and 228 ± 40 ( S . D . ) bp less than that of the approximately 500-bp unlooped peak , with the 114-bp shorter peak ( ∼386 bp ) being the predominant species . These results indicate that Lac repressor–di-operator complexes can exist in two different looped conformations when bound to O-158-O . If the two structures interconvert directly , they likely involve two distinct conformations of the repressor . In contrast , if the two species can interconvert only by going through the unlooped state , there is the additional possibility that the two species are topological isomers that differ in the configuration of the DNA loop but in which repressor conformation does not change significantly [19 , 26 , 28 , 60] . Interconversion of such topoisomers requires transient dissociation of one of the two operator sites from the repressor . To distinguish between these possibilities , we further analyzed TPM records of O-158-O ( such as those in Figure 2D ) and counted the relative number of times the short-tether loop state was immediately followed by the long-tether loop state rather than the unlooped state . This partition ratio , P′′short-tether→long-tether/short-tether→unlooped = 3 . 0 ± 1 . 1 ( see Materials and Methods ) is significantly greater than one , demonstrating that conversion to the long-tether loop is the favored pathway of exit from the short-tether loop . Similarly , the long-tether loop preferentially converts to the short-tether loop relative to converting to the unlooped species , with P′′long-tether→short-tether/long-tether→unlooped = 2 . 3 ± 0 . 5 . These observations show that the two looped states can interconvert directly without passing through the unlooped state . Thus , they contain distinct conformations of the protein; they are not simply topological isomers that differ ( for example ) in operator binding orientation [28 , 60] . The single looped species seen with O-153-O has a tether length significantly different ( p < 0 . 025; t-test with 50 d . f . ) from those of either O-158-O looped species . Since the lengths of the two DNAs differ by an amount ( 5 bp ) too small to detect by TPM , the observed differences in the tether lengths of the three distinct looped species are likely due to differences in the preferred angle between the DNA arms as they exit the loop . This in turn strongly suggests that the looped segment of the DNA can have different three-dimensional geometries in Lac repressor looped complexes depending on operator phasing . Taken together , the O-153-O and O-158-O data indicate that Lac repressor can induce substantially different loop structures depending on operator spacing and that a single di-operator-repressor complex can exist in multiple ( at least two ) different looped conformations . In crystal structures of intact tetrameric Lac repressor , the two DNA binding domains have roughly similar orientations [20] . They are therefore well situated to form looped complexes with DNAs such as O-158-O , in which the two operator sites are separated by an integral number of helical turns putting them on the same face of the double helix . In contrast , the operator sites are positioned on opposite faces of the DNA in O-153-O . If looped complexes formed by O-158-O and O-153-O are identical in three-dimensional structure except that the DNA has no torsional strain in the former and one-half twist between the operators in the latter , the equilibrium constant for loop formation with O-153-O should be approximately 104-fold lower ( see Materials and Methods ) than that for O-158-O due to the significant energetic cost predicted by standard models of DNA mechanics of introducing a half twist into the 153-bp DNA segment . The TPM looping experiments permit thermodynamically rigorous measurement of the equilibrium constant for loop formation as the ratio of the total time spent in a particular looped state to the total time spent in the unlooped state . Under the conditions of the experiment , these equilibrium constants are 0 . 5 ± 0 . 1 , 2 . 9 ± 0 . 5 , and 0 . 39 ± 0 . 08 for the O-153-O loop , O-158-O long-tether loop , and O-158-O short-tether loop , respectively . The long-tether loop is approximately 7-fold more populated than short-tether loop and therefore is the preferred looped conformation for O-158-O . Although the O-153-O looped complex is less stable ( relative to the unlooped state ) than the preferred looped complex of O-158-O , the equilibrium constants differ by only 5-fold , not the approximately 104-fold expected from DNA twisting alone . This analysis suggests that the O-153-O and O-158-O looped complexes may incorporate significantly different repressor conformations that accommodate the different operator spacing without significantly twisting the interoperator DNA . To more fully characterize the mechanism by which repressor interacts with O-153-O , the complete set of O-153-O TPM records ( e . g . , Figure 2C ) was analyzed to determine the lifetime distributions of the unlooped ( Figure 3A ) and looped ( Figure 3B ) states . The loop lifetime histogram is well fit by a simple exponential function , consistent with the interpretation that the O-153-O looped state is a single chemical species , not an unresolved mixture of two states . In contrast , the unlooped state lifetime histogram requires a distribution function that is the sum of at least two exponential terms to produce an acceptable fit . A multiexponential distribution is expected [23] because we know a priori that an unlooped di-operator DNA can exist in a minimum of four different states , an “O2R2” state that has two bound repressor molecules , two equivalent “O2R linear” states in which a single repressor molecule interacts with one operator , and an “O2” state with no bound repressor . These four unlooped species , together with the single looped species , comprise the minimal kinetic scheme ( Figure 3F ) for the interaction of repressor with O-153-O DNA . The scheme has only four independent rate constants; because the TPM experiments provide the shapes of the lifetime distributions ( for both the looped and aggregate unlooped states ) and the equilibrium constant ( between the looped and aggregate unlooped states ) , they allow determination of well-constrained values for all four ( Figure 3F; see Materials and Methods ) . The shapes of the lifetime distributions predicted by this scheme reproduce the empirical data within experimental uncertainty ( Figure 3A and 3B ) ; similarly , the value of the equilibrium constant predicted by this scheme and that determined by experiment are also in good agreement ( 0 . 48 and 0 . 53 ± 0 . 11 , respectively ) . In the analysis of O-158-O state lifetimes , both the long-tether and short-tether looped state distributions were fit well by single exponential functions ( Figure 3D and 3E ) , consistent with mechanisms in which each looped state consists of a single chemical species . O-158-O is expected to have the same four unlooped states as O-153-O , leading to the kinetic scheme of Figure 3G . In addition to the three lifetime distributions , the TPM experiments also directly measure the two equilibrium constants and three partition ratios . These data taken together with the value of the repressor–operator dissociation rate constant , kd , measured in the O-153-O experiment permitted precise determination ( see Materials and Methods ) of the forward and reverse rate constants for reaction pathways in the kinetic scheme ( Figure 3G ) . As with O-153-O , the kinetic scheme with the deduced rate constants reproduces the lifetime distributions ( Figure 3C–3E ) and other experimental data . The values calculated for k3 and k4 are significantly greater than zero , confirming that all features of the TPM data ( not merely the partition ratios ) are consistent with the ability of long- and short-tether loops to interconvert directly without going through the O2R linear species . In contrast , when we performed a separate kinetic analysis using the scheme of Figure 3G but with fixed k3 = k4 = 0 ( unpublished ) , the mechanism predicted values of equilibrium constants and partition ratios substantially different from those measured , confirming that direct interconversion of the two looped states is necessary to explain the experimental observations through the mechanism of Figure 3G . Finally , the rates of interconversion are consistent with observation that the two species are not resolved in gel mobility shift experiments ( unpublished data; see also [15] ) , because they interconvert on a time scale ( <1 min; Figure 3G ) that is much shorter than the time required for electrophoresis .
As previously discussed , a likely cause of the existence of two loop structures that are in equilibrium with one another for the in-phase O-158-O DNA is a conformational change in the repressor protein itself . However , the effective tether lengths of the two looped species differ by 114 bp , corresponding to a DNA contour length difference of 38 nm . As this distance is larger than the overall dimension of the repressor protein itself , only a reconfiguration of the repressor structure that substantially changes the angle between the two DNA arms as they exit from the loop is likely to account for the detected difference between the two types of loops . We previously showed that a single acute DNA bend could shorten the observed TPM tether length by 284 bp [45] . Thus , a large-scale repressor conformational change that changes the relative orientation of the operator-binding domains could be sufficient to explain the difference in tether length measured for the two O-158-O loops . Prior studies raised the possibility that Lac repressor may in fact be capable of such a large-scale conformational change . Crystal structures of intact tetrameric Lac repressor [20] , either alone or in complex with operator DNA , show an asymmetrical “V” structure with two operator-binding domains located at the tips of the arms of the V ( Figure 4A ) . The major contacts between the two halves of the tetramer are restricted to the 4-helix bundle at the tip of the V . Friedman et al . [19] and Lewis et al . [20] both noted that spreading open of the V into a more extended conformation would result in only a small increase ( ∼300 Å2 [20] ) in water-accessible surface area and thus is predicted to be only moderately unfavorable in free energy . Indeed , Friedman et al . [19] speculated that such an extended conformation might facilitate looping of DNA with closely spaced operators . Early solution X-ray scattering , powder diffraction , and electron microscopy results are consistent with an extended conformation [61 , 62] . A more recent electron microscopy study suggested that both a compact and an extended conformation of repressor may coexist in solution [63] . Finally , hyperstable looped complexes formed from di-operator DNAs that contain static bends between the operators may have at least two alternative loop shapes , depending on the phasing of the bend and operators [26 , 35] . Fluorescence resonance energy transfer measurements are consistent with a “V” repressor structure in one of these loops , and a more extended structure has been proposed for the other [35 , 36] . Our results lend further credence to the idea that Lac repressor can exist in a V-shaped and an extended conformation and in addition show that the two structures can exist in dynamic equilibrium in the same protein–DNA complex . To assess the possible structures that could explain our TPM and AFM data , we made a comprehensive delineation ( Figure 5 , left column ) of the types of loop geometries that can be made with the repressor in its crystallographic conformation and a small interoperator spacing DNA , together with the loop configurations predicted to arise from these structures if the repressor is distorted into an extended conformation by pivoting the two core dimers about the axis of the 4-helix bundle ( Figure 4A ) . The analysis assumes a simple rigid-body mechanical model in which this pivoting is the only permitted structural change in the repressor . As noted previously , we consider “wrap away” models such as that of Figure 1D ( Figure 5B and 5D , right ) to be unlikely because they are inconsistent with the colocalization of DNA loop and repressor protein we observe by AFM . More significantly , the left-column structures of Figure 5B–5E produce strained , energetically implausible configurations if the repressor is opened to the extended conformation without allowing one of the DNA-binding domains to temporarily dissociate from the operator to relieve the strain . Thus , the structures in Figure 5A are the only loop geometries that adequately explain the observed ability of the long- and short-tether states to interconvert directly without passing through an unlooped configuration . Although the loop structures in Figure 5B–5D have been previously proposed , they were proposed for DNAs that differed materially from the ones used here because the DNAs had permanent bends [35] , were bent by the catabolite activator protein [20] , or had a considerably shorter interoperator spacing [20 , 24 , 35] . As previously noted [28 , 35 , 64] , it is quite possible that the lowest energy loop structure would be different in such different circumstances . Loop configurations other than those shown in Figure 5 are possible if the repressor is capable of other structural changes ( e . g . , headpiece reorientation; see [1 , 28] ) in addition to the simple hinge motion considered here; however , there is little experimental evidence for such changes . Steered molecular dynamics calculations [1] that allowed the operator axis to rotate to a position nearly perpendicular to the plane of the Figure 4A illustration produced wrap-away loop structures that appear inconsistent with our AFM data . However , these calculations used a smaller interoperator spacing which could direct formation of different loop geometries than those obtained with the approximately 150-bp spacings used here . To determine whether the two loop structures in Figure 5A can account for the two tether lengths observed with O-158-O , we made more detailed models based on crystallographic and electron microscopy data [19 , 20 , 63] . In one model ( Figure 4B ) , the repressor adopts the V-shaped conformation determined by X-ray crystallography and the DNA loop wraps around the repressor in the fashion analogous to the “wrapping toward” loop model of Friedman et al . [19] . The interoperator segment has near-zero twist . In the other model ( Figure 4C ) , we propose that the repressor is in an almost fully extended conformation with a shape similar to that inferred from electron microscopy [63] . In this model , the repressor is still positioned in the interior of the curved DNA segment and the interoperator segment is still not twisted . A rough calculation ( see Materials and Methods ) of the effective tether lengths predicted by these loop models yields 354 and 296 bp for the V-shaped and extended repressor conformation , respectively . Given the approximate nature of the calculation , these lengths are in reasonable agreement with the values ( 421 ± 143 [S . D . ] and 309 ± 132 bp ) measured by TPM and suggest that the long-tether-length conformation is the one in which the Lac repressor is in the V-shaped conformation ( Figure 4B ) seen in the crystal structures [19 , 20] . The measured difference in free energy between the two in-phase loops is small , amounting to little more than twice the energy of thermal agitation ( Figure 6 , species 3 and 6 ) . The small energy difference is consistent with the energies estimated for the proposed loop structures ( see Materials and Methods ) , demonstrating that the proposed structures are good models for the two looped species seen with O-158-O . The closure of the loop is predicted to be energetically downhill for both structures ( Figure 6 , species 4 → 3 and species 5 → 6 ) because the formation of the highly favorable repressor–operator interaction more than compensates for the energy required to bend the DNA . Although opening the repressor “V” structure ( species 4 → 5 ) is expected to be energetically costly because of the elimination of some favorable subunit–subunit interactions , this energy cost is almost fully offset by the fact that the DNA bending strain required to form the extended repressor loop structure ( species 5 → 6 ) is less than that required for the “V” repressor loop structure ( species 4 → 3 ) . This difference in DNA bending strain also explains the measured kinetics of loop closure . In the transition state for a loop closure ( Figure 6 , species 4 → 3 or species 5 → 6 ) , the DNA is bent into a shape very close to that of the fully formed loop and the favorable repressor–operator contacts are just beginning to form . The energy input required to surmount the transition state barrier in going from species 5 → 6 is lower than that in going from 4 → 3; this is readily explained by the smaller DNA bending required for the former . In summary , the proposed conformations in Figure 4B and 4C are consistent with both the kinetics and thermodynamics measured for formation of the two kinds of in-phase loops . In addition , conformations appearing consistent with both of the proposed structures can be seen in the AFM images ( Figure 1F ) . Nevertheless , our structural conclusions are based on a simple rigid-body model of the Lac repressor protein in which pivoting around the four helix bundle is the only allowed internal motion . The conclusions may require revision if future experiments reveal that additional regions of the repressor structure can hinge or deform during loop formation . The WLC model of DNA mechanics predicts a substantial energy cost to twist an approximately 150-bp segment of DNA by a half turn . Although this cost might be reduced substantially by local denaturation or kinking of the DNA , the rate at which such discontinuities form spontaneously in DNA is likely too low to explain the rates of loop formation observed here [30–33] . Based on the WLC model , the out-of-phase looped structure would be expected to be greatly destabilized relative to the in-phase structure if the repressor protein has the same conformation in both cases . Our data show that even with an out-of-phase interoperator spacing as small as 153 bp , Lac repressor can form highly stable loops . The rough features of these out-of-phase looped complexes are remarkably similar to those of the two different O-158-O in-phase loops: the free energies are similar ( Figure 6 , species 2 , 3 , and 6 ) ; the tether length of the O-153-O loop is between those measured for the O-158-O loops; and all loops appear in AFM images to have the same wrap-around configuration with only subtle differences in the observed DNA exit angles ( Figure S1 ) that might be attributed to differences in conformation of the looped species . Most importantly , the energy of the transition state for closure of the out-of-phase loop ( Figure 6 , species 1 → 2 ) is almost identical to that for the closure of the long-tether in-phase loop ( Figure 6 species 4 → 3 ) , even though the strain in the out-of-phase DNA construct would be expected a priori to be the same or larger . If formation of the out-of-phase loop requires the prior formation of the energetically unfavorable extended form of the repressor protein , which is not necessary for the formation of the long-tether in-phase loop ( Figure 6 , species 3 ) but is necessary for the short-tether in-phase loop ( Figure 6 , species 6 ) , the loop closure transition state energy for 1 → 2 would be expected to be higher than that for 4 → 5 → 6 , which is inconsistent with the kinetics we measure ( Figures 3 and 6 ) . Therefore , the simplest hypothesis is that the out-of-phase loop has the same repressor configuration as the long-tether in-phase loop , for which we propose a structure with the “V” repressor conformation . Nevertheless , the detailed three-dimensional structures of species 2 and 3 must be different to accommodate the different operator phasing; this difference is also necessary to account for the somewhat different tether lengths and gel mobility shifts [15] observed for 153 and 158 loops . Species 2 is unlikely to contain a highly strained structure of the repressor because its free energy is similar to that of species 3 . Instead , we propose that closure of the out-of-phase loop captures a dynamic conformation of an already flexible region of the repressor . One possible hypothesis is that the DNA binding headpieces move apart ( by rotation away from the plane of the illustration in Figure 4A ) , thereby reducing or eliminating the twist in the loop . Other modes of conformational flexibility in the repressor [1 , 28] could also account for the stability of the out-of-phase loop . Although we observe only a single tether length for the out-of-phase loop ( and only two tether lengths of in-phase loops ) , it is clearly possible that each DNA can form additional loop structures ( possibly including those in Figure 5B–5E ) that are not detected in our experiments because they are not sufficiently thermodynamically or kinetically stable to be seen . The idea that two or more different looped species can form on DNAs with small operator spacings is consistent with previous observations that 153- and 158-bp spaced operators produce putatively looped species with substantially different gel mobilities [15] . Swigon et al . [28] showed that various alternative loop structures are energetically accessible and also demonstrated that earlier DNase I footprinting data on loops with small interoperator spacings [15] are consistent with a structural model incorporating an extended-repressor loop . Kahn and coworkers [26 , 35 , 36] used a variety of techniques to study the hyperstable loops formed from DNAs with strong intrinsic bends in the interoperator segment . These studies demonstrate that the mechanical constraints imposed by bends at different positions relative to operators induce formation of alternative loop structures . However , they do not determine whether the alternative structures have substantially different repressor conformations or are merely loop topoisomers . While this manuscript was under review , Normanno et al . [59] reported that twisted DNA with much larger operator spacings can form two alternative types of loops . The three-dimensional structures of these loops and the extent to which they correspond to the alternative structures of small loops that we observe is unclear , particularly since multiple topoisomeric species have similar energies in the limit of large operator spacing [28] . Our studies demonstrate two looped states in DNA without static bends and further show that these must arise from a large-scale structural reorganization of the repressor itself . Furthermore , the experiments directly demonstrate that different kinds of loops can form at equilibrium from the same DNA . By going beyond analysis of equilibrium structure to examine the dynamics of looped species formation , breakdown , and interconversion , we also show that these processes occur on a timescale ( tens to hundreds of seconds ) similar to that of rapid transcriptional responses to changes in environmental conditions . The dynamic data also provide key clues about loop configuration that lead to our proposal of novel loop geometries with decreased DNA bending or twisting strain energies relative to previously proposed structures . Recent studies support the notion that alternative loop structures like those proposed here are necessary to quantitatively account for the extent of looping-mediated repression in measured in living cells [38 , 39] . On the basis of DNA mechanics alone , the DNA loops formed by Lac repressor in vivo would be expected to have enormous differences in stability . Loops between operator pairs separated by distances both much larger than and much smaller than the persistence length of DNA ( and hence , with predicted DNA bending energies both much larger and much smaller than the thermal energy ½RT ) are important for repression . Also , changes in the extent of supercoiling , the binding of DNA bending proteins ( e . g . , catabolite activator protein ) , and association with proteins that alter DNA flexibility [17 , 43 , 65] could cause dramatic differences in the mechanics of loop closure and thus greatly perturb loop stability . The ability of Lac repressor to dynamically alter its conformation ( and thus , the geometries of and mechanical strain in the resulting loops ) that is observed here suggests the repressor may have evolved the ability to produce polymorphic loop structures as a way of stably maintaining regulation of transcription under conditions of widely varying mechanical difficulty of loop formation . This is a testable hypothesis—it predicts that mutant repressor proteins designed to have decreased conformational flexibility should be less able to achieve stable repression both in vivo and in vitro over a range of conditions thought to alter DNA mechanics .
Lac repressor was a kind gift of Kathleen Matthews ( Rice University ) . All repressor concentrations are expressed as the concentration of tetramer . Avidin-conjugated beads ( 0 . 098-μm diameter ) were prepared as described [23] . Di-operator and mono-operator parent plasmids pH108 ( contains two symmetric operators separated by 114 bp ) and pH109 ( identical to pH108 except that three point mutations in one of the operator sites abolish recognition of that site by repressor ) were gifts of Sankar Adyha [66] . Two-operator plasmids pOKW153 and pOKW158 were constructed from pH108 by inserting 39-bp ( 5′-GTTACCTTAGGTACCACTAG-TCTAGAATGCATTCCGCGG-3′ ) or 44-bp ( 5′-GTTACCTTAGGTACCACTAG-TCTAGACCGCGGAGATCTCAATTG-3′ ) linkers , respectively , into the unique BstEII site . One-operator plasmid pOKW153C , which is identical to pOKW153 except for the three point mutations , was constructed by inserting the 39-bp linker into the BstEII site of pH109 . 5′ Digoxigenin-labeled oligonucleotide P31-dig ( digoxigenin-5′-TCGATAGCG-TGATCGTGC-3′ ) and 5′ biotin-labeled oligonucleotides P32-bio ( biotin-5′-CGTATCA-CGAGGCCCTTT-3′ ) and P86-bio ( biotin-5′-CAATAATTTATTCCATGTCAC-3′ ) were synthesized from the corresponding 5′ amino-labeled oligonucleotides by reacting with either digoxigenin-3-O-methylcarbonyl-ε-aminocaproic acid-NHS ester ( Roche Applied Sciences ) or biotin-XX-NHS ester ( Glen Research ) . Labeled oligonucleotides were subsequently purified by anion exchange high-performance liquid chromatography ( HPLC ) . Mono- and di-operator DNA fragments O-153-O , O-158-O , and O-539 were generated by polymerase chain reaction ( PCR ) using primers P31-dig and P32-bio with pOKW153 , pOKW158 , and pOKW153C as templates . O-349 was made from pOKW158 using P31-dig and P86-bio . PCR products were purified by extraction with buffered phenol and with water-saturated 1-butanol , followed by four cycles of greater than 10-fold dilution and reconcentration in a Centricon-100 concentrator [67] . In constructs O-153-O and O-158-O , the center-to-center separation between the two operators are 153 bp ( 14 . 5 helical turns , assuming 10 . 5 bp/turn ) and 158 bp ( 15 turns ) respectively . Mono-operator construct O-539 is identical to O-153-O , except that only the biotin-proximal operator is functional in repressor binding ( Figure 1A ) . In O-349 , only the digoxigenin-proximal operator is present ( Figure 1A ) . Formation and breakdown of repressor–DNA looped complexes formed from O-153-O or O-158-O were monitored using previously described single-molecule light microscopy techniques [23] with the following modifications: DNA–bead complexes were preformed by incubating 35 pM DNA with 0 . 56 nM avidin-conjugated beads in PTC buffer ( 20 mM Tris-acetate [pH 8 . 0] , 130 mM KCl , 4 mM MgCl2 , 0 . 1 mM EDTA , 0 . 1 mM DTT , 20 μg/ml acetylated BSA ) for >60 min . Under this condition , the probability of having multiple DNA molecules attached to the same avidin-conjugated bead is less than 0 . 10 . After attaching the DNA-bead complexes to the anti-digoxigenin–coated surface and washing the microscope flow cell with PTC buffer supplemented with 6 mg/ml casein , a solution of 5 . 4 nM repressor in LRB ( 10 mM Tris-HCl [pH 7 . 4] , 200 mM KCl , 0 . 1 mM EDTA , 0 . 2 mM DTT , 5% DMSO , 0 . 6 mg/ml casein ) was introduced . The cell was then monitored by video-enhanced differential interference contrast light microscopy at approximately 22 °C to observe the processes of looped complex formation and breakdown . Time sequences of digitized images of DNA-tethered beads were collected using Glimpse ( http://www . brandeis . edu/projects/gelleslab/glimpse/glimpse . html ) ; each recorded image was the average of 64 consecutive video frames ( 2 . 1 s ) . Digitized images , together with the times at which each was acquired , were stored in the computer for subsequent off-line analysis . During the experiments , correct microscope focus was maintained automatically every 20 s by a stepper motor that moved the stage to the position giving the highest contrast in the image of a bead rigidly attached to the coverslip surface . Data acquisition was temporarily halted during the focusing process so that out-of-focus images were not included in the bead Brownian motion data . The beads in TPM experiments experience a polymer confinement force directed away from the surface [68] , but in these experiments , this force is expected to be insignificant because of the small ( 98 nm ) bead diameter . Brownian motion of DNA-tethered beads in each image was calculated as described [23] , except that data were not filtered . Brownian motion records were converted to tether length measurements using a proportionality constant of 0 . 047 nm/bp obtained as described [41] . For O-153-O tethered beads , two Brownian motion states ( looped and unlooped ) were observed and their lifetimes were analyzed using the one-threshold discrimination algorithm [23] . Time records of O-158-O tethered bead Brownian motion with histograms that showed three discrete Brownian motion states were analyzed using an analogous two-threshold algorithm , with the thresholds positioned at the troughs between the histogram peaks representing the three states . The minority of records in which three states could not be clearly distinguished ( see Results ) were excluded from the analysis . Under the image acquisition conditions used for both O-153-O and O-158-O , states with lifetime < tmin = 10 s could not be reliably detected; therefore , such data were excluded from further analysis . Measured lifetimes were binned and plotted as scaled lifetime probability density histograms where n ( t ) is the number of events in the histogram bin centered at time t , N is the total number of observed events , W is the bin width , and F is the estimated fraction of missed events calculated as described below . Time constants for the theoretical lifetime distributions predicted by the kinetic schemes were calculated by the method of Colquhoun and Hawkes [69] . Under the image acquisition conditions used in the TPM experiments , occasions in which a DNA molecule existed in a particular looped or unlooped state for a time < tmin = 10 s could not be reliably detected . To estimate the number of such missed events , we first fit the raw lifetime histograms ( using the Levenberg-Marquardt algorithm ) for each looped state to the exponential distribution function [70] to obtain the time constant , τ , for each state . For the aggregated unlooped states of O-153-O and O-158-O , the corresponding bi-exponential function [71] , was instead used to obtain time constants τ1 and τ2 and amplitude A for each DNA . All fits were excellent with randomly distributed residuals . For each looped state i , the fraction Fi of events that were not detected ( i . e . , events with lifetime < tmin ) was calculated as the mean lifetime di was taken to be the lifetime distribution fit parameter τ , and its standard error was computed as For the unlooped states , the corresponding equations were and For any state , the total time Di in each state adjusted for the missed events , and its standard error σDi , were calculated as and Results of these calculations are reported in Table S1 . The equilibrium constant for loop formation , Ki , j , is the ratio of the total time spent in state i to that spent in state j: and the standard error of the equilibrium constant was calculated by error propagation as To determine the partition ratios for the interconversion between the unlooped state and the two looped states of O-158-O , we first measured Ca→b , the number of observed instances in the O-158-O TPM records in which state a was immediately followed by state b . This measurement was independently made for each pairwise combination of the unlooped , long-tether , and short-tether states . Only transitions in which both the beginning and ending states had lifetimes equal to or greater than 10 s were counted . Each Ca→b was then used to calculate C′a→b and C′′a→b , values corrected for missed state b events and for both missed state a and state b events , respectively , as where Na is the number of state a events with lifetimes equal to or greater than 10 s and Fa is the fraction of missed state a events as defined earlier . The partition ratio P′′a→b/a→c and its standard error S′′a→b/a→c were then calculated as In all cases , the corrected partition ratios P′′a→b/a→c differ from the uncorrected values Pa→b/a→c by less than 18% . Molecules in the unlooped state can in principal interconvert between four different chemical species ( O2R2 , O2 , and two equivalent O2R ) before looping . The analytical expressions relating the O-153-O rate constants kloop , kd , and ka* ( Figure 3F ) to the unlooped state lifetime distribution are highly complex and therefore were not used . Instead , values and standard errors for the three rate constants ( Figure 3F ) were determined by numerical optimization to the observed set of unlooped state lifetimes using the MIL program [72 , 73] as implemented in the QuB software suite [74] . The analogous O-158-O rate constants ( Figure 3G ) were determined the same way , except that kd was held fixed at the value determined for O-153-O . That constraint was imposed because the values of kd , the single-operator repressor dissociation rate constant , are expected to be identical for the two DNA constructs because they have identical operator sequences . The apparent first-order rate constant for repressor–operator association , ka* , was allowed to vary in order to accommodate small unintended differences in the concentration of free repressor between the two experiments; however , the difference between the resulting values was roughly that expected merely from the calculated uncertainties . Preliminary fits of the O-158-O data in which both kd and ka* were allowed to vary did not adequately constrain the rate constant values because of the comparatively small difference between the two principal time constants for the O-158-O unlooped state lifetime distribution . Rate constants related to looped state lifetimes ( k−loop in Figure 3F for O-153-O; k1 through k6 in Figure 3G for O-158-O ) were determined by global optimization of the equations ( for O-153-O ) , or ( for O-158-O ) to minimize chi-square with respect to the values of the empirical quantities given on the left sides of the equations . Fits were weighted using the calculated standard errors of the empirical quantities; in cases in which the calculated error was less than 10% , a 10% error was assumed to allow for small systematic errors in the measurements . The calculated rate constants reproduced the experimental data almost exactly; all fit residuals were less than 15% . Error estimates for the rate constants were calculated by propagating errors from the empirical quantities using a Monte Carlo simulation [75] . All fitting and error propagation calculations were performed using custom computer software ( available from the authors by request ) implemented in MATLAB . Di-operator DNA ( 40–60 nM O-158-O or 50 nM O-153-O ) was incubated with 2-fold molar excess ( over DNA ) of repressor in LB buffer ( 35 mM Tris-HCl [pH 7 . 4] , 140–180 mM KCl , and 0 . 3 mM EDTA ) for 10 min at approximately 22 °C . An aliquot of the solution was then mixed with an equal volume of 1 . 7% glutaraldehyde in 10 mM Tris-HCl ( pH 7 . 4 ) and incubated for 2 min at approximately 22 °C to allow protein–DNA cross-linking to occur . ( This step helps to preserve the repressor–DNA looped complexes during the subsequent deposition process . ) A volume of 1 μl of the cross-linked sample was then diluted with approximately 9 μl of DB buffer ( 10 mM Tris-HCl [pH 7 . 4] , 10 mM MgCl2 ) and deposited onto a disk of freshly cleaved ruby mica ( Asheville-Schoonmaker Mica Co . ) . After 1–2 min , the mica disk was rinsed with water and dried with a stream of nitrogen . AFM images were obtained in air with a Nanoscope IIIa microscope ( Digital Instruments ) operating in the tapping mode using high-frequency silicon tapping-mode cantilevers ( f0 ∼ 330 kHz; Nanosensors ) . Images ( 512 × 512 pixels ) were collected with a scan area of either 1 . 5 × 1 . 5 μm or 2 × 2 μm at a scan rate of two to four scan lines/second . To distinguish Lac repressor–DNA looped complexes ( in which protein is bound to both operator sites ) from RO complexes ( in which protein is bound to only one operator site ) in the AFM images , we measured DNA arm length as the image contour length from the center of the protein to the end of each DNA arm using Nanoscope III ( Digital Instruments ) image analysis software . DNA contours were traced with short segments of straight line , and DNA arm length was obtained by adding the length of these line segments . The total image contour length of each complex was then calculated by summing the two DNA arm lengths . Measurements of the image contour lengths of 539- and 544-bp DNA molecules without bound repressor systematically underestimated the contour length by an average of 7 . 6% ( 14 nm ) ; this underestimation is expected because the DNA path was approximated with line segments [55] . Since the DNA arms for both di-operator constructs are 190 or 196 bp ( measured from the center of each of the operators to its closest 5′ end ) , the 7 . 6% systematic underestimation should lead to measured arm lengths of approximately 61 nm in the looped complexes . Therefore , all protein-bound molecules with both arm lengths in the range 61 ± 16 nm ( 16 nm is two standard deviations of the random error in contour length measurements as determined by measurements on free DNA as described above ) were classified as looped complexes . RO complexes are defined as molecules that have one DNA arm length within 61 ± 16 nm and the other arm in the range of 108 ± 16 nm ( 108 nm is predicted based on the assumption that the sum of two DNA arm lengths should be approximately 169 nm , similar to the average image contour length measured for the 539- and 544-bp DNA molecules ) . To permit objective statistical tests for differences in the population-averaged geometries of repressor–DNA complexes , the angle between the two DNA arms at their exit from the repressor , defined as the acute angle θ ( Figure S1A ) between two lines each tangent to the DNA arms at the two exit points , was measured by two independent observers . On average , images classified as non-looped repressor–operator complexes based on arm length criteria ( Figure 1C , green squares ) have a nearly colinear arm geometry ( Figure S1A–S1C ) , with a mean arm exit angle ( Figure S1A ) of approximately 136° . This result is consistent with previous demonstrations that binding of repressor to a single operator site does not severely bend the DNA [76 , 77] . In contrast , complexes scored as looped ( Figure 1C , red square ) have a near orthogonal arm geometry on average , with mean exit angles of approximately 114° for looped complexes of O-158-O and approximately 100° for O-153-O ( Figure S1D–S1M ) . Although from TPM experiments two looped species are observed in O-158-O , two peaks are not clearly resolved in the exit angle histogram ( Figure S1H ) . The failure to resolve the peaks is perhaps not surprising; the equilibrium constants calculated from the TPM data predict that the equilibrium concentration of the short-tether loop is only approximately 13% that of the long-tether loop . Nevertheless , the population-average exit angle , measured by two independent observers , of O-158-O is significantly larger than that of O-153-O ( p < 0 . 02 and p < 0 . 12 for observers A and B , respectively; unpublished data ) . Thus , AFM data support the conclusion from the TPM experiments that the population of looped complexes formed with O-158-O is structurally different from that formed with O-153-O . The simulated images were generated using a program that allows the user to model objects on a surface using sphere swept lines ( SSLs ) , which are cylinders capped on each end with a hemisphere [57] . The program then simulates an AFM image of the model by modeling the tip as a sphere swept cone , which is a cone capped with an 8-nm diameter hemisphere ( comparable to the size of tip used in the AFM experiments ) at the end , and convoluting the shape of the tip and modeled object using dilation and erosion methods . Lac repressor is modeled as an SSL with a diameter of 9 nm and a length of 11 nm ( estimated from the crystal structures [20] ) , and the DNA is modeled as a chain made up of connected SSLs with flexible links , with each SSL being 4-nm long and 2 nm in diameter . The various conformation of the repressor–DNA complexes were modeled by manually wrapping or looping the flexible chain ( DNA ) around the 9 nm × 11 nm SSL ( repressor ) . In all simulations , both the chain and the SSL lie on the surface . The length of the chain in the loop is approximately 50 nm , which is the distance between the two lac operator sites in the DNA used in the AFM experiments . The difference in standard free energy ( ) for O-158-O long-tether and short-tether looped complexes can be estimated from where is the difference in the energy required to bend the DNA into a long-tether versus a short-tether loop and is the difference in conformation energy required to disrupt the protein-protein interactions in the V-shaped repressor upon loop formation . The value for is calculated from [78] , where ρ is the bending persistence length of double-stranded DNA ( ∼154 bp , [79] ) , θlong = 1 . 74π ( 2π − angle of the V arms of the repressor measured from the crystal structure ) is the bend angle for the long-tether loop in Figure 4B , θshort = π is the bend angle for the short-tether loop ( Figure 4C ) and L = 158 bp is the length of the DNA loop . Therefore , is 9 . 8 RT , where R is the gas constant and T is the absolute temperature , taken to be 298 °K in these calculations . Difference in conformation energy is calculated from: where A = 300 Å2 is the difference in solvent-exposed surface area [20] assuming short-tether loop formation requires opening up the V-shaped repressor to a fully extended structure and E = 66 . 9 J mol−1Å−2 [80] . Thus = ∼ −8 . 1 RT and = 1 . 7 RT . The values of the latter determined from the measured equilibrium constant and from the global fit to the kinetic data are both −2 . 0 RT , in reasonable agreement with the estimate . The small discrepancy between the calculation and measurement may be attributable to either of two factors: First , the model does not include favorable sequence nonspecific interactions between the repressor core domain and the looped DNA segment [24] that may exist in the long-tether loop . Second , the two operator-binding sites are not precisely coplanar in the crystal structure , nor are they expected to be in the open structure; the DNA strain energies will thus be somewhat different from those in the estimate , which assumes a planar structure . The change in free energy in introducing half a twist to DNA of length L is calculated from a simple elasticity model as: [78] where p is the torsional persistence length of double-stranded DNA ( ∼286 bp; [81] ) and θt is the torsional angle ( π for half a twist ) and L = 153 bp . Therefore , the energy needed for twisting 153-bp DNA half helical turn is = 9 . 2 RT , which is equivalent to an equilibrium constant of approximately 104 . To estimate the long-tether loop tether length ( Figure 4B ) , the unknown length of DNA that makes up each of the two overlapping arcs between the two DNA binding domains was approximated by the distance ( corresponding to the length of approximately 32 bp of duplex DNA; see Figure 4B ) between the center of one operator and the distal end of the other operator measured from the repressor–operator co-crystal structure [20] . Thus , the tether length for long-tether loop was estimated to be 354 bp , the sum of the two DNA arms ( 190 bp and 196 bp ) minus 32 bp . For the proposed short-tether loop structure ( Figure 4C ) , the arms do not overlap and the repressor is extended ( both of which lengthen the tether ) , but the overall tether length is shorter . This is because the angle between the DNA arms as they exit the loop is roughly zero as opposed to the approximately 180° angle proposed in Figure 4B . The Figure 4C tether length is estimated to be the sum of the length of the extended repressor structure and the tether length contributed by the two DNA arms . Since the arm length is approximately the persistence length of the DNA , the arms will on average be bent by θ = ∼90° . Making the crude assumption that the effective tether length of the arm length L = 190 bp can be modeled as equivalent to a semicircular arc with radius r = L / θ = 121 bp , each DNA arm will contribute approximately 121 bp to the tether length of the short-tether loop . The length of the fully extended repressor conformation is approximately 54 bp , approximately twice the length of a repressor dimer estimated from the crystal structures . Therefore , the predicted tether length for the short-tether loop is 296 bp . This calculation is only an approximation; it assumes an arbitrary value for the arm exit angle and does not take into account sequence effects or protein or DNA dynamics . | Some proteins that regulate DNA transcription do so by binding simultaneously to two separated sites on the DNA molecule , forming a DNA loop . Although such loops are common , many of their features are poorly characterized . Of particular interest is the question of how some proteins accommodate the formation of loops of different sizes , particularly when the loops are small and thus require strong bending ( and , in some cases , twisting ) of the DNA to form . We observed the shape and behavior of individual DNA molecules bent into tight loops by Lac repressor , a transcription-regulating protein from the bacterium Escherichia coli . Loops were formed in DNA molecules with repressor-binding sites on opposite faces of the DNA double helix almost as readily as in those with sites on the same side , suggesting that the repressor is highly flexible . The DNA can switch back and forth between a tighter and a looser loop structure “on the fly” during the lifetime of a single loop , further evidence that Lac repressor is capable of adopting different shapes that may serve to minimize DNA bending or twisting in loops . The ability of the repressor to readily switch between different loop shapes may allow it to maintain effective control of transcription across situations in which the difficulty of bending or twisting DNA changes substantially . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biochemistry",
"biophysics",
"molecular",
"biology"
] | 2008 | Interconvertible Lac Repressor–DNA Loops Revealed by Single-Molecule Experiments |
Intrinsically disordered proteins play an important role in cellular signalling , mediated by their interactions with other biomolecules . A key question concerns the nature of their binding mechanism , and whether the bound structure is induced only by proximity to the binding partner . This is difficult to answer through experiment alone because of the very heterogeneous nature of the unbound ensemble , and the probable rapid interconversion of the various unbound structures . Here we report the most extensive set of simulations on NCBD to date: we use large-scale replica exchange molecular dynamics to explore the unbound state . An important feature of the study is the use of an atomistic force field that has been parametrised against experimental data for weakly structured peptides , together with an accurate explicit water model . Neither the force field nor the starting conformations are biased towards a particular structure . The regions of NCBD that have high helical propensity in the simulations correspond closely to helices in the ‘core’ unbound conformation determined by NMR , although no single member of the simulated unbound ensemble closely resembles the core conformation , or either of the two known bound conformations . We have validated the results against NMR spectroscopy and SAXS measurements , obtaining reasonable agreement . The two helices which most stabilise the binding of NCBD with ACTR are formed readily; the third helix , which is less important for binding but is involved in most of the intraprotein contacts of NCBD in the bound conformation , is formed more rarely , and tends not to coexist with the other helices . These results support a mechanism by which NCBD gains the advantages of disorder , while forming binding-competent structures in the unbound state . We obtain support for this mechanism from coarse-grained simulations of NCBD with , and without , its binding partner .
We must consider all the conformations that a protein populates , if we want to understand completely its function and behaviour . Although many proteins fold promptly to a native conformation—thereby allowing us to use a simplified picture of conformational space—many others do not [1] . Intrinsically disordered proteins ( IDPs ) [2] , [3] , [4] , [5] do not form stable structures in isolation under physiological conditions . Instead , they sample multiple conformations , often while retaining some residual structure [6] . An IDP may perform a function while partly disordered [7] , [8] , or it may fold to one or more ordered conformations as it binds ligands [9] in a process known as coupled folding and binding [10]; indeed IDPs are often signalling molecules , for example in transcription regulation [11] . Protein disorder has a number of possible advantages: for example , a larger effective binding surface may increase binding rates ( via ‘fly-casting’ [12] or ‘non-native steering’ [13] ) , while greater conformational flexibility might help a protein to bind multiple ligands [2] , [14] , [15] . It has been proposed that disorder can aid allosteric coupling [16] , enable ultrasensitivity [17] or , in contrast , help to stabilise a system against perturbations in its environment [18] , low temperatures [19] or desiccation [20] . The occurrence of coupled folding and binding challenges us to consider the mechanism that causes it [10]: is it ‘induced fit’ ( the binding partner induces the disordered IDP to adopt the bound conformation ) or ‘conformational selection’ ( the binding partner selects a binding-competent conformation from an ensemble of conformations visited in the unbound state ) ? Although these two descriptive mechanisms may not be mutually exclusive , and most likely represent extremes of a continuum [21] , they emphasise the importance of the unbound ensemble [22] . However , characterising all the conformations belonging to the unbound ensemble of the IDP is challenging , because experimental techniques [23] , [24] , [25] generally provide only ensemble-averaged properties [5] . Since the unbound state samples a very heterogeneous set of conformations , and is likely to interconvert rapidly between them , experimental observations can be hard to interpret . Molecular simulations can therefore play an important role in the understanding of IDP behaviour , as they provide a detailed picture of the dynamics in unfolded proteins . Phenomenological “coarse-grained” models using a simplified representation of the system are a powerful tool for investigating general features of coupled folding-binding due to their low computational cost [26] , [13] , [27] . However , the most detailed and accurate computational models for proteins are all-atom simulations in which all of the protein atoms and solvent molecules are included explicitly [28] . Such models should capture sequence-specific and solvent-mediated effects in a more predictive fashion . There are two main limitations of this type of simulation approach: the first is that it becomes very computationally demanding even for relatively small proteins . Enhanced sampling techniques [29] , [30] such as the replica-exchange molecular dynamics ( REMD ) [31] , [32] method used here , can help overcome this difficulty . Recent studies have used all-atom simulations to investigate the conformational dynamics of histone tails [33] and the binding of a small IDP [34] . The second limitation is that atomistic simulations represent a system as a collection of classical particles interacting via an empirical energy function or “force field” [35] . This picture is only an approximation of nature , and so the force field itself can be only approximately correct . For example , most force fields only include pairwise interactions , and neglect effects such as polarisation and charge transfer . This leads to the consequence ( familiar in all coarse-graining problems ) that the force field may not be fully transferable: a force field that has been parametrised with data from small molecules may suffer from flaws if it is applied to a large molecule [36] , and a force field that is optimised for folded proteins may be less appropriate for IDPs . One of the most extensively characterised IDPs is NCBD ( also known as IBiD or SID [37] , [38] ) , the nuclear coactivator binding domain of the transcriptional coactivator CBP ( CREB-binding protein ) , which is a regulator of gene expression in animals . The domain is disordered in isolation—it forms a molten globule with some helical structure—but it undergoes synergistic folding and binding with another intrinsically disordered protein , CABD ( the coactivator binding domain of the p160 transcriptional coactivator ACTR ) [39] , [40] , [41] . NCBD also forms a complex with the transcription factor IRF-3 [42] , [43] , [44] , which is not an IDP [45] . Each of the bound conformations of NCBD contains three helices , but the tertiary structures differ . A study of unbound NCBD [46] has found that the ensemble of the molten globule includes a ‘core’ conformer , which resembles the conformation of NCBD bound to ACTR . This experimental evidence can therefore be compared with simulation results . In the present work , we investigate the unbound ensemble of NCBD using extensive atomistic REMD simulations in explicit solvent . Unlike recent computational studies of NCBD [47] , [27] , our approach is intended to be predictive: the force field is not biased towards a particular tertiary structure , and the simulation does not start from the native structure . Other recent atomistic simulations of IDPs have investigated smaller fragments using replica-exchange molecular dynamics [33] and multicanonical molecular dynamics [34] to enhance sampling . These earlier studies provide contrasting examples of the degree of secondary structure formation in the unbound state . We use an atomistic force field [48] that has been parametrised with experimental data for weakly structured peptides , which should make it appropriate for this system . By looking at the unfolded ensemble , we can reach some conclusions about how unfolded NCBD may reflect the demands of binding with ACTR . Ultimately , however , a definitive description of the binding mechanism will require explicit consideration of the binding partner . Our simulations reveal a large amount of residual secondary structure in the unbound ensemble , in agreement with experimental evidence . The regions of NCBD that have strong -helical propensity correspond to the three helices in the experimental ‘core’ conformer of the unbound state . However , no member of the simulated ensemble has an overall structure closely resembling the core conformer , or either of the bound conformations . We find that the two ( end ) helices that most stabilise NCBD-ACTR binding are formed most readily . The middle helix , which is less important for binding ACTR , but which is implicated in most of the intraprotein contacts in the ACTR-bound conformation of NCBD , is formed more rarely , and tends not to coexist with the end helices . We argue that this may indicate a ‘binding interface preference’ mechanism by which NCBD retains the advantages of being disordered , while forming binding-competent structures in the unbound state . As such NCBD exhibits features of both conformational selection ( with respect to the binding interface ) and induced fit ( with respect to contacts involved in folding ) .
The structures which NCBD has been found experimentally to adopt in complex with ACTR ( PDB ID: 1KBH ) [40] and IRF-3 ( PDB ID: 1ZOQ ) [43] are largely -helical: both contain approximately the same three helical regions . The main difference between the two structures is in the tertiary arrangement of the helices . An experimental study [46] has found that a ‘core’ conformer ( PDB ID: 2KKJ ) , with a structure similar to the ACTR-bound conformation 1KBH , is highly populated in the unbound ensemble . The three structures are shown in Figure 1 , together with the sequence of NCBD . In order to compare the secondary structure of our simulated unbound ensemble , we have analysed the -helix propensity in the 304 K replica from the REMD simulations ( this is the replica whose temperature is closest to that at which the 2KKJ structure was determined [46] ) . Recent atomistic studies of smaller IDPs [33] , [34] have shown evidence of secondary structure preference . Figure 2 shows the proportion of time spent by each residue of NCBD in a helical state; the helices in the PDB structures are also indicated . The simulations produced three clear regions of high helical propensity , corresponding exactly to the three helices in the core unbound conformer 2KKJ: residues 7–18 ( hereafter referred to as region I ) , 23–35 ( region II ) and 37–46 ( region III ) . These results agree with the experimental observation that NCBD is intrinsically unfolded , but has residual helical structure . The helical content appears to be qualitatively consistent with CD and NMR data ( discussed further below ) , which show a very significant helical content in the unbound state [40] , [53] . The simulations also correctly predict the location of the residual helical structure within the sequence , corresponding to the regions occupied by helices in the core unbound conformer [46] . These boundaries of the helices are also in agreement with chemical shift data [53] . The main ways in which the core conformer differs from 1KBH are: ( i ) helix II is lengthened to cover residues 23–27 , and ( ii ) helix III is shortened by six residues at the C-terminal end . This successful prediction provides support for the accuracy of the ff03w force field . Our results may also show traces of the differences between the three PDB structures: the last two residues of region I ( 17 and 18 ) , which are helical in 2KKJ and 1KBH but not in 1ZOQ , have lower helix propensity than the rest of helix I , while the simulations indicate a detached slightly helical region ( residues 49–51 ) that seems to suggest the longer helix III in 1KBH , compared to the other structures; this C-terminal extension of helix III is presumably stabilised by specific interactions with the binding partner in that structure . A recent simulation study [54] of NCBD included an REMD simulation of the unbound state , using a 47 residue fragment ( corresponding to residues 8–54 of the present study ) with an implicit solvent model . The authors argued that the results were limited by convergence problems; nevertheless , a simulation initiated from a fully unfolded state developed some helical propensity in regions II and III ( and in region I , to a small extent ) . When the simulation was initiated from the folded state 1KBH , high helical propensity was found in all regions—including the part of region II ( residues 23–27 ) that is not helical in 1KBH . This is consistent with our own findings for this portion of the sequence . Structure formation has also been seen in previous all-atom simulations of “disordered” proteins . For example , in simulations of fragments of histone tails , flickering elements of both -helix and -hairpin were observed [33]; similarly , a recent computational study of neural restrictive silencer factor ( NRSF ) , which is helical when bound to its partner mSin3 , was found to populate significantly both and structure [34] . However , the population of helical structures is particularly high in NCBD , in comparison with these other studies . This may be related to the fact that it needs to bind another IDP , ACTR , in contrast with some other IDPs like NRSF that bind to folded domains . In addition to individual residues , we can look at larger sections of the protein , and compute how far they deviate from some ordered reference structure , in order to evaluate the extent to which the ordered structure is present in the unbound ensemble . It makes most sense to look at regions I , II and III; since they match the helices in the 2KKJ structure , we will use that structure as the reference . The top left panel of Figure 2 shows the distribution of the root-mean-squared deviations of the individual regions at 304 K , calculated by separate least-squares fitting of each region to the 2KKJ reference structure . In this case a low RMSD is already quite a strict definition of helix formation , as it will not include short partial helices . Since the observed structures are predominantly helical , this seems like a reasonable choice of order parameters . Helix III is predicted to be the most likely to form completely ( 36 . 8% within 0 . 3 nm ) . Helix I is a little less likely to be fully formed ( 26 . 6% ) , probably because it is slightly longer . Helix II is significantly less stable than the others ( 6 . 15% ) , in agreement with the helical propensity results in Figure 2 . The other three panels of Figure 3 show two-dimensional distributions of the rmsds of pairs of regions , which allow us to see the extent to which helices coexist . The darker areas , corresponding to high probability density , can be viewed as clusters of heavily populated states . The results indicate that the most common pair of coexisting helices is I and III . The joint probability that both regions are within 0 . 3 nm of the 2KKJ structure , at any given time , is 10 . 9% . This is close to the product of separate probabilities for the two regions ( 9 . 77% ) , which is the value that would be expected for the joint probability if the two regions were uncorrelated . Helix II is found much more rarely with either of the others . In particular , the joint probability for regions II and III is 0 . 140% , much lower than the product of separate probabilities ( 2 . 27% ) . The equivalent figures for helices I and II are 0 . 825% ( joint ) and 1 . 63% ( separate ) . One question which arises regarding any molecular simulation is how representative is the sampling . We need to be sure that the sub-ensembles which are shown in Figure 3 arise from genuine attractors in the free energy landscape of the protein , and are not a consequence of an inadequate exploration of conformational space . Evidence for adequate sampling is that free energy surfaces and helix populations calculated from the first and second halves of the trajectory ( excluding the first 50 ns ) produced comparable results ( Figure S1 ) . To be sure that this does not arise from configurations merely being “stuck” in local traps for the duration of the run , in Figure 4 we show eight conformations observed during the simulation , and we also mark their positions , and the continuous trajectory of which each forms a part , on a shadow of Figure 3 . Only the 304 K parts of the trajectories are plotted , so each trajectory is divided into regions , joined by lines which bypass the states visited at higher temperatures . It is clear that each trajectory was able to explore a number of different regions , and also that a given region could be visited by more than one trajectory . The trajectories in Figure 4 must not be interpreted as folding paths , since REMD aims to reproduce the canonical distribution of a system , rather than its dynamics . However , we need to inspect the trajectories , to check that they encompass changes in secondary structure . Trajectory 13 ( shown in blue ) forms helices I and II between conformations A ( which includes a hairpin in place of helix III ) and B , while helix III remains unformed . Helix III is formed in trajectory 37 ( cyan ) between E and F , the other two helices remaining unformed . Helix I unfolds in trajectory 42 ( orange ) between G and H; helix III is present , and helix II absent , throughout . Trajectory 35 ( C and D ) , in green , illustrates multiple transitions between clusters , and the folding and unfolding of helix II . To see more of what the simulation results can tell us about NCBD as an intrinsically disordered protein , and NCBD-ACTR binding , we look at the contacts that are present in the bound structure 1KBH . Figure 5 shows the intraprotein ( within NCBD ) and interprotein ( between NCBD and ACTR ) contacts . Of the 25 long-range ( greater than 5 residues apart in sequence ) intra-protein contacts , 17 involve helix II , while only five link helix III to helix I or to the N-terminal region of the protein . Turning to the inter-protein contacts , we see that only eight of the 76 contacts involve helix II ( there are two more that involve nonhelical residues close to helix II ) . Most of the inter-protein contacts involve helix I ( 14 contacts ) or III ( 23 contacts , and an additional 18 in the region 47–53 , which is helical in 1KBH but not in 2KKJ ) . A protein that exploits different conformations to bind different partners , as NCBD does , may face a difficulty . On the one hand , binding might be helped by the presence , in the unbound ensemble , of nativelike structure in regions of the protein that are involved in binding [55] ( a recent simulation study of fast-folding proteins found that formation of secondary structure tends to precede formation of contacts [56] ) . This is likely to be particularly important when the ligand is itself an IDP ( as ACTR is ) : we might expect the binding of two IDPs to be more difficult , if each spends too much of its time in conformations that are incompatible with the other protein . On the other hand , if there is too much nativelike structure , it might actually slow the binding [57] , or get in the way of binding another ligand such as IRF-3 . Our results suggest that the unbound ensemble of NCBD may show adaptations that ameliorate this difficulty by means of a ‘binding interface preference’ mechanism . The regions around helices I and III contain most of the inter-protein contacts in the NCBD-ACTR bound structure , so binding might be easier if one ( or both ) of these regions has nativelike structure in unbound NCBD . Accordingly , the simulation results show a high degree of helicity in these regions ( Figure 2; top left panel of Figure 3 ) , and helices I and III readily occur together ( bottom right panel of Figure 3 ) . On the other hand , most of the intraprotein contacts involve helix II: a lack of nativelike structure in this region , and a tendency to avoid conformations containing both helix II and another helix , might help to make NCBD an IDP rather than a folded protein , and protect its ability to bind IRF-3 . We do indeed find that helical structure is scarcer in this region ( Figure 2; top left panel of Figure 3 ) , while helix II coexists rarely with helix I , and even more rarely with helix III , which is the helix formed most commonly ( top right and bottom left panels of Figure 3 ) . ‘Binding interface preference’ is a hypothesis about the unbound ensemble , suggesting how it might be distributed in order to facilitate binding without compromising disorder in the unbound state . It is a statistical rather than a mechanistic hypothesis , and does not propose a specific binding path—indeed , it might be misleading to infer a binding path from information about the unbound ensemble alone . Unbound NCBD retains some of the structure of the ACTR-bound conformer [46]; this supports the idea that , on average , preformed nativelike structure is likely to help binding , especially when the binding partner is also an IDP . However , individual binding events might follow many different paths , not all of which need benefit from preformed nativelike structure . Experimental studies of ExsE [58] and of the preS1 surface antigen of hepatitis B virus [59] suggest that , in the unbound state of some IDPs , the binding interface is more strongly structured than other parts of the protein . We can establish some further theoretical support for binding interface preference , using a coarse-grained native-centric ( G ) model for the NCBD-ACTR complex . Figure 6 shows the probability distribution for a nativeness parameter , for NCBD alone and in the presence of ACTR . The model for NCBD is identical in the two cases . Because of its additional ( intermolecular ) interactions , the native state of the complex is more energetically favourable than that of isolated NCBD; this pushes to higher values . Results are shown for a baseline version of the model , together with three variations in which subsets of the native interactions are weakened . When the interactions between helix II and the other two helices are weakened , the distribution for solitary NCBD shifts to lower values of ( red line ) ; weakening the intrahelical interactions has a smaller effect ( blue and green lines ) . This supports the idea that the disordered character of isolated NCBD is enhanced by a tendency to avoid the simultaneous formation of helix II and another helix . In the presence of ACTR , the distribution is perturbed most readily by weakening of the intrahelical interactions of helices I and III , which form the binding interface . This result ties in with the idea that helical structure in these regions is particularly important for binding . Each of the bound structures of NCBD contains three helices packed together . To visualise the difference between the two tertiary structures ( Figure 1 ) , imagine a plane formed by helices II and III . Helix I can take up either of two positions: above the plane ( 1KBH ) or below it ( 1ZOQ ) . Whether or not the helices are actually present , we can describe the tertiary structure in the coordinate system defined in parts ( a ) and ( b ) of Figure 7 , which considers the orientation of region I relative to the plane formed by regions II and III . This is calculated from the relative orientations of the regions , rather than their relative positions . With only two coordinates , it cannot be a complete description of the relative orientations: it neglects the angle between regions II and III . Therefore , it is only one of many possible descriptions of the tertiary structure . However , the coordinate system is particularly appropriate for the tertiary structure of NCBD , as it captures the essential difference between 1KBH and 1ZOQ as a difference between positive and negative values of . The main panel of Figure 7 shows the probability distribution of the simulation results at 304 K , as a function of and . There is a favoured region on this graph , in which the three PDB structures also lie . As expected , we find that the simulated system spent more time close to the core conformer 2KKJ and the ACTR-bound structure 1KBH than in the vicinity of the IRF3-bound structure 1ZOQ . Notably , however , we observe a significant population of conformations whose topology is similar to that of 1ZOQ . The peaks in the probability do not coincide precisely with the PDB structures: the simulated ensemble does not match the secondary structures of the PDB conformations exactly ( Figures 3 and 1 ) , and therefore it lacks the specific packing interactions that would be needed to stabilise their tertiary structures; moreover , the presence of the binding partners would be expected to have an effect on the energy landscape . The azimuth of the favoured region is closer to than to , which means that the orientation of region I tends to be approximately opposite to that of region II . This orientation maximises packing interactions between helix I and each of the other helices: since all the helical regions are approximately the same length , an azimuth of ( I and II pointing in the same direction ) would make interaction between helices I and III difficult . Interestingly , both in the simulations and in the experimental structures , the azimuth shows a bias towards angles below , rather than above , , while the elevation shows a bias towards negative angles . Since there is no geometrical reason for such preferences , they can only be due to specific tertiary packing interactions . To assess how close our computed ensemble is to that observed experimentally , we have back-calculated several sets of experimental data from our simulations . First , we compare our results with NMR data reported for unbound NCBD [46] . The distance restraints derived from NOE data in the experimental study were largely satisfied or near-satisfied by the 2KKJ conformation [46] , which led to the proposal that 2KKJ is a ‘core’ conformer in the unbound ensemble . We compare our simulation results against these NMR restraints . The left hand panel of Figure 8 shows the proportion of the restraints that are satisfied , as a function of the tolerance , in the simulations at 304 K and in the set of 48 randomised conformations from which the simulations were initialised . Distances were computed by averaging over the ensemble [60] . The simulation does much better than the random set , both for medium-range constraints ( between atoms separated by 2–4 residues ) , and for long-range constraints ( separated by more than four residues ) . The simulation results satisfied 90 . 8% of the medium-range constraints within a tolerance of 1 Å ( that is , allowing the simulation predictions to fall outside the range of the NMR results by 1 Å ) , but did less well against the long-range constraints , satisfying 50 . 0% within 1 Å . The remaining discrepancies are probably due to the limited sampling that was possible within the finite duration of our simulations , and to inaccuracies in the force field . Of course , the unfolded ensemble of an IDP is expected , by definition , to explore diverse conformations rather than remaining in a ‘core’ conformer or bound structure . The right hand side of Figure 8 shows chemical shifts calculated from the simulation results using SPARTA+ , and compares them with experimental values and with values similarly calculated from the 2KKJ structure . The simulation correctly reproduces the locations and extensions of the helices , as Figure 2 suggests , but underestimates the overall helical content . In earlier work , we had found that we were able to optimise the helix propensity of a simpler alanine-based sequence ( the 15-residue helical peptide ) , to obtain a good match to experimental chemical shift deviations [48] . The present results clearly show larger deviations from experiment , and present some measure of the transferability of the force field to more complex sequences . Some insight into this issue can be obtained from considering a more recent assessment of the helix propensities of all 20 amino acids in the ff03w force field [61] . In this work , it was found that although ff03w resulted in residue-specific propensities broadly consistent with experiment , there were significant deviations for some residues . It was also shown that optimisation of side-chain torsion potentials and the use of a common backbone charge model are promising directions for improving the helix propensity of other residues to be comparable to that of alanine [61]; thus , there is clearly further work that needs to be done to make force fields fully transferable to all residue types . We should note , however , that close to the transition midpoint , substantial population shifts can be caused by relatively small changes in free energy - e . g . for a simple two-state system a change of can shift the population from % folded to folded . Given that there are certainly appreciable residual errors in current force fields , and the challenge of sampling an equilibrium distribution for this size of system , the discrepancy from experiment in this case is in fact quite reasonable . Small-angle X-ray scattering ( SAXS ) data contain information about the gross features of the structural ensemble . We have computed SAXS curves , using CRYSOL [62] , from structures sampled randomly from the 304 K replica in our simulations . The ensemble-averaged SAXS profile is compared with the experimental data in Figure 9 . The data agree reasonably well in the low- region which is most sensitive to large-lengthscale features of the ensemble . However , in this region the second derivative of the simulated intensity is less negative than that of the experimental intensity . According to the Guinier approximation [63] , this indicates that the simulated ensemble for the unbound state is slightly compacted relative to the experimental system . This is consistent with the average radius of gyration of estimated from the simulations using CRYSOL . The comparable figure estimated from the unbound 2KKJ structure is 15 . 8 Å , while the average value reported from SAXS experiments under nativelike conditions is [46] . Overcompaction of unfolded proteins is a common problem with current force fields . Although previous studies with the TIP4P/2005 water model [48] or TIP4P-Ew ( a water model similar to TIP4P/2005 ) [64] gave more realistic dimensions for the unfolded state than the commonly used TIP3P water , it is clear that there is room for further improvement . The simulation results successfully reproduce the essential characteristics of unbound NCBD: it is found to be a molten globule with residual -helical structure . In respect of secondary structure , we find very good agreement between the simulations and the experimentally observed core unfolded conformer 2KKJ . This agreement can be seen in the proportion of medium-range NMR restraints satisfied , and in the residual helical propensity . Importantly , the helical propensity matches the core conformer very closely indeed , whereas it matches the bound structures of NCBD only to the extent that they resemble the core conformer . The simulations are less successful at reproducing the tertiary structures of the core conformer or of the bound structures: no single simulated conformation is a close match for any of these . However , if the tertiary structure is plotted according to a two-dimensional coordinate system that is based on the difference between the two bound structures 1KBH and 1ZOQ , the simulated ensemble is found to favour regions in the vicinity of the PDB structures . Helices I and III of NCBD contain the bulk of the residues that are in close proximity to ACTR in the NCBD-ACTR bound structure . Therefore , it is likely that these regions are where preformed structure will be helpful for binding , particularly as ACTR itself is intrinsically disordered . In contrast , helix II is involved in most of the intraprotein contacts of NCBD . This suggests that preformed structure in the region of helix II , especially in concert with preformed structure elsewhere , might encourage the formation of the ACTR-bound structure of NCBD , even in the absence of ACTR , thereby eliminating the benefits of disorder . In the simulations , helices I and III form more readily than does helix II , while coexistence between helix II and the other helices seems to be disfavoured . These facts point to the hypothesis that a ‘binding interface preference’ mechanism is helping to maintain disorder in the unbound state of NCBD—by disfavouring structure that might encourage independent folding—while favouring structure that facilitates binding with ACTR . On the continuum between conformational selection and induced fit , regions of NCBD that form contacts with ACTR are perhaps closer to conformational selection , while regions that form intraprotein ( folding ) contacts in the ACTR-bound structure are closer to induced fit . It would be interesting to see if any other IDPs , particularly ones with intrinsically disordered binding partners , use an analogous mechanism .
The simulated system contained a single molecule of 59-residue NCBD in a truncated octahedral water box of size 70 Å between nearest faces , with periodic boundary conditions . NCBD comprised 944 atoms , and the box contained 8334 water molecules , for a total of atoms , including a low ( 0 . 15 M ) concentration of sodium chloride . The water model used was TIP4P/2005 [65] ( a highly optimised version of TIP4P ) while the Amber ff03w force field [48] was used for the protein ( this force field has been adapted for use with the TIP4P/2005 water model ) . Long-range electrostatics were calculated using the particle mesh Ewald ( PME ) method , with a 9 Å cutoff and a 1 . 2 Å grid spacing . Simulations were performed with GROMACS 4 . 5 . 3 [66] , using Phase 2b of HECToR together with in-house computing resources . We used replica-exchange molecular dynamics [32] , with a total of 48 replicas . Replicas differed only in temperature: the lowest temperature was 304 . 000 K , the highest 424 . 458 K . Temperature gaps between replicas were selected to ensure an exchange acceptance probability of slightly above 0 . 2 between neighbouring replicas , throughout the temperature range . This led to a choice of 2 . 000 K for the temperature gap at the low temperature end , increasing to 3 . 221 K at the high temperature end . The above system , with the protein initiated from the 2KKJ structure , was simulated for 8 ns at 304 K under conditions of constant pressure . From the resulting trajectory , a conformation was selected which had a volume close to the trajectory average . This was used as the starting point for the subsequent preparation and simulations , which were performed under conditions of constant volume . To produce 48 starting conformations for the REMD , the system was simulated at constant volume at 600 K . Conformations were selected from different points on the trajectory between 10 and 100 ns . Each conformation was then equilibrated at its appropriate starting temperature ( in the range 304 . 000–424 . 458 K ) over a cooling time of 50 ps . A danger of using a simulation at 600K is that the groups around peptide bonds adjacent to proline residues may isomerise , and be trapped in the incorrect isomer on cooling to the simulation temperature . Therefore , care was taken to select conformations from points on the trajectory where only the correct isomers were present . After this preparation , the replica-exchange molecular dynamics was run for a total of 250 ns . A different random seed was used for the Langevin dynamics of each replica . The simulation time step was 2 fs , and replica exchanges were attempted every 10 ps ( which corresponds to attempting half of the possible exchanges every 5 ps ) . We monitored structural properties , and on this basis the first 50 ns was discarded , leaving us with 200 ns of usable data . A practical problem with explicit solvent simulations of unfolded proteins concerns the choice of box size . Under periodic boundary conditions , it is possible for the protein molecule to interact with itself across the boundary , and possibly form a complex , which is clearly an artefact . Unfortunately , the length of time that can be simulated diminishes rapidly as the box size is increased: for a 59-residue protein such as NCBD , it is unrealistic to use a box so large that self-interaction is impossible . We can only choose a box large enough to make self-interaction a rare event . However , the closer a simulation gets to exploring a representative sample of the whole ensemble ( which this study aimed to move towards ) , the more likely it becomes that such unfortunate rare events will happen . Artefactual complex formation occurred in two of the 48 continuous trajectories , and we disregarded all the data from these two trajectories . None of the other trajectories showed any close approaches between the molecule and its image . The minimum separation between the molecule and its image was never less than 6 Å , and was less than 10 Å only 0 . 2% of the time . The separation was between 20 Å and 40 Å about 98% of the time . Since close approach was such a rare event , it is unlikely that the finite box size had much influence on the ensemble ( Figure S2 shows the probability density for this minimum separation ) . For the G model simulations ( Figure 6 ) , models were constructed from the native state 1KBH according to the prescription of Karanicolas and Brooks [67] , except that the pseudoangles between three consecutive residues were subject to a statistical potential [68] , [69] . Langevin dynamics simulations were run using CHARMM , with a time step of 0 . 01 ps and a friction coefficient of . Each variation of the model was simulated for between 8 . 5 and , and the first 500 ns was discarded . In variants where a subset of native contacts is weakened ( by a reduction of 20% in thair pair energies ) , the pair energies of the remaining contacts are increased so as to preserve the energy of the native state . The nativeness parameter is defined , as a function of residue separations , by ( 1 ) where and . The sum is computed across all native contacts , and is the residue separation in the native state . | While many proteins have a specific ‘native’ conformation , so-called intrinsically disordered proteins ( IDPs ) adopt many different conformations in rapid succession—a characteristic that may be advantageous for rapid binding and promiscuous association . However , this characteristic also makes it very hard to make experimental measurements over times that are short enough to see changes of conformation . In this work , we use the results of a large-scale molecular simulation to explore conformations of NCBD , which is an IDP that adopts specific conformations when it binds either of two other proteins ( ACTR and IRF-3 ) . Our results point to the following hypothesis: to help NCBD bind ACTR , those parts of NCBD that make contact with it in the bound conformation are biased towards that conformation , even in the absence of ACTR . Other parts of NCBD tend to avoid the ACTR-bound conformation , to help ensure that NCBD is disordered when unbound . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Methods"
] | [
"biomacromolecule-ligand",
"interactions",
"protein",
"folding",
"biophysics",
"theory",
"biology",
"biophysics",
"simulations",
"biophysics"
] | 2012 | A Preformed Binding Interface in the Unbound Ensemble of an Intrinsically Disordered Protein: Evidence from Molecular Simulations |
Helminths express various carbohydrate-containing glycoconjugates on their surface , and they release glycan-rich excretion/secretion products that can be very important in their life cycles , infection and pathology . Recent evidence suggests that parasite glycoconjugates could play a role in the evasion of the immune response , leading to a modified Th2-polarized immune response that favors parasite survival in the host . Nevertheless , there is limited information about the nature or function of glycans produced by the trematode Fasciola hepatica , the causative agent of fasciolosis . In this paper , we investigate whether glycosylated molecules from F . hepatica participate in the modulation of host immunity . We also focus on dendritic cells , since they are an important target of immune-modulation by helminths , affecting their activity or function . Our results indicate that glycans from F . hepatica promote the production of IL-4 and IL-10 , suppressing IFNγ production . During infection , this parasite is able to induce a semi-mature phenotype of DCs expressing low levels of MHCII and secrete IL-10 . Furthermore , we show that parasite glycoconjugates mediate the modulation of LPS-induced maturation of DCs since their oxidation restores the capacity of LPS-treated DCs to secrete high levels of the pro-inflammatory cytokines IL-6 and IL-12/23p40 and low levels of the anti-inflammatory cytokine IL-10 . Inhibition assays using carbohydrates suggest that the immune-modulation is mediated , at least in part , by the recognition of a mannose specific-CLR that signals by recruiting the phosphatase Php2 . The results presented here contribute to the understanding of the role of parasite glycosylated molecules in the modulation of the host immunity and might be useful in the design of vaccines against fasciolosis .
Fasciolosis is a major parasitic disease of livestock that causes significant economic losses worldwide [1–2] . Currently , fasciolosis is also considered an emerging zoonosis with an increasing number of human infections globally [1] . In temperate regions this disease is caused by the liver fluke Fasciola hepatica . During infection , this pathogen can modulate the host immune response by different cellular and molecular mechanisms that include the production of immune-suppressive cytokines by the host [3] , the increase of regulatory T cells [4] , the alternative activation of macrophages [3] or the modulation of maturation and function of dendritic cells ( DCs ) [5–7] . Helminths express various carbohydrate-containing glycoconjugates on their surface and they release glycan-rich excretion/secretion products that can be very important in their life cycles and pathology , since they can participate in immune escape [8] . Carbohydrate-signatures from parasites are decoded by the immune system through the interaction of several immune receptors . In particular , receptors of innate immunity that recognize glycan motifs consist of soluble or membrane-associated lectins , siglecs and scavenger receptors , among others . Notably , C-type lectin receptors ( CLRs ) have been described to mediate internalization of parasite glycosylated molecules as well as cell-surface signaling , modulating the host immune response [9] . For instance , Schistosoma mansoni , through a glycosylated RNAse , impairs protein synthesis of IL-12 . The glycans on this enzyme are essential to allow its uptake by DCs where it degrades both ribosomal and messenger RNA , leading to a Th2-polorized T-cell response [10] . On the other hand , glycans from the nematode Brugia malayi were reported to participate in the induction of the specific Th2 immune response , since sodium periodate-treated soluble extracts from this parasite induced lower levels of IL-4 by specific lymph node cells [11] . Evidence demonstrating that helminths can mediate the modulation of the activity or function of DCs has also been reported [5–7] . DCs are potent antigen presenting cells that possess the ability to stimulate naive T cells . In response to infectious agents DCs undergo a maturation process during which they migrate to secondary lymphoid organs where they present captured antigens to naive T cells , for the triggering of specific immunity . This process is associated to an up-regulation of the expression of MHC molecules , adhesion molecules and co-stimulatory molecules ( CD40 , CD80 or CD86 ) as well as a down-regulation of their endocytic capacity [12] . However in the presence of helminth antigens mature DCs express reduced levels of co-stimulatory markers and MHC class II molecules , as compared to DCs matured with Toll-like receptor ( TLR ) ligands such as lipopolysaccharide ( LPS ) [13] . Also , these DCs are not capable of producing high levels of pro-inflammatory cytokines ( IL-12 , IL-6 or TNFα ) [13] . In this sense , independent in vitro studies have reported that different F . hepatica components modulate TLR-initiated DC maturation and their stimulatory function [5–7] . Although under investigation , the identity of the molecular components from helminths that mediate DC immune-modulation is limited . Nevertheless , growing evidence suggests that parasite glycoconjugates could play a role in the modulation of DC-maturation [14] . Indeed , a recent report described that glycosylated components from the whipworm Trichuris suis mediate the suppression of TNFα production by DCs stimulated with LPS , through the recognition of mannose ( Man ) residues or terminal N-acetyl-Galactosamine ( GalNAc ) by specific CLRs [15] . Interestingly , it has been recently reported that fucose-carrying helminth components can trigger a DC-SIGN specific signaling pathway on DCs that directs differentiation of T cells into follicular helper T cells [16] . Little is known about the glycans produced by F . hepatica , with only two recent reports describing lectin reactivity in the miracidial surface [17] or in the gut of adult flukes [18–19] , that suggest the presence of Man and glucose ( Glc ) residues . Another independent work from Wuhrer and collaborators described Galβ1-6Gal-terminating glycolipids by using mass spectrometry [20] . Finally , our group has previously described the expression of the GalNAc-O-Ser/Thr structure ( known as Tn antigen ) [21] . As for their structure , the immune-modulatory roles of F . hepatica glycans have also barely been investigated . Their role in alternative activation of macrophages has been reported by treating glycans with periodate [3] or by inhibiting macrophage binding and function using antibodies specific for CLRs [22–23] . Nevertheless , the evidence available about the function of F . hepatica carbohydrate in the regulation of parasite immunity or DC function is still poor . In this work , we show that glycoconjugates from F . hepatica are involved in the modulation of host immunity , promoting the production of IL-4 and IL-10 , and suppressing IFNγ production . During infection this parasite is able to induce a semi-mature phenotype of DCs which express low levels of MHCII and secrete IL-10 . Furthermore , we show that parasite glycosylated molecules mediate the modulation of LPS-induced maturation of DCs since their oxidation restores the capacity of LPS-treated DCs to secrete high levels of the pro-inflammatory cytokines IL-6 and IL-12/23p40 and low levels of the anti-inflammatory cytokine IL-10 . Inhibition assays using carbohydrates suggest that the immune-modulation is mediated by the recognition of a Man specific-CLR that signals by recruiting the phosphatase Php2 . The results presented here contribute to the understanding of the role of parasite glycoconjugates in the modulation of the host immunity and might be useful in the design of vaccines against fasciolosis .
Mouse experiments were carried out in accordance with strict guidelines from the National Committee on Animal Research ( Comisión Nacional de Experimentación Animal , CNEA , National Law 18 . 611 , Uruguay ) . Adult worms were collected during the routine work of a local abattoir ( Frigorífico Carrasco ) in Montevideo ( Uruguay ) . All procedures involving animals were approved by the Universidad de la República's Committee on Animal Research ( Comisión Honoraria de Experimentación Animal , CHEA Protocol Numbers: 071140-001822-11 and 071140-000143-12 ) . Six- to 8-week-old female BALB/c mice were obtained from DILAVE Laboratories ( Uruguay ) . Animals were kept in the animal house ( URBE , Facultad de Medicina , UdelaR , Uruguay ) with water and food supplied ad libitum , and handled in accordance with institutional guidelines for animal welfare by the Committee on Animal Research ( CHEA , Uruguay ) . Live adult worms of F . hepatica were obtained from the bile ducts of bovine livers , washed in phosphate buffered saline ( PBS ) pH 7 . 4 , then mechanically disrupted and sonicated . After centrifugation at 40 , 000 × g for 60 min supernatants were collected and dialyzed against PBS . The obtained lysate ( FhTE ) was resuspended on PBS containing a cocktail of protein inhibitors ( Sigma-Aldrich , St . Louis , MO ) and dialyzed against PBS for 24 h . Carbohydrate glycol groups present in FhTE were oxidized with sodium periodate ( 10 mM ) . The oxidation was performed at room temperature for 45 min in the dark , followed by the reduction with sodium borohydride ( 50 mM ) of the reactive aldehyde groups . The resulting oxidized lysate is referred as FhmPox . In order to perform control experiments , the following control extracts were prepared: FhCB , consisted of FhTE subjected to the whole treatment excepting for the incubation with sodium periodate; and CmPox , consisting of PBS subjected to the entire treatment . Lysates were dialyzed against PBS and their protein concentration was measured using the bicinchoninic acid assay ( Sigma-Aldrich , St . Louis , MO ) . To remove endotoxin contamination , the lysates were applied to a column containing endotoxin-removing gel ( detoxi-gel , Pierce Biotechnology ) . The endotoxin levels were determined by using the Limulus Amebocyte Lysate kit Pyrochrome ( Associates of Cape Cod ) . Protein preparations showed very low levels of endotoxins and were not able to induce DC maturation ( as IL-12 read out ) on their own . The concentration of all F . hepatica extracts described here and used in culture experiments did not modify cell viability evaluated by MTT ( 2-[4 , 5-dimethyl-2-thiazolyl]-3 , 5-diphenyl-2H-tetrazolium bromide ) assay . The lysates were analyzed by electrophoresis and western blotting using the anti-Tn mAb 83D4 ( kindly provided by E . Osinaga , Uruguay ) and a polyclonal antibody specific for the Cathepsin-L1 from F . hepatica ( FhCL1 , kindly provided by P . Berasain , Uruguay ) . Proteins were separated in a 15% SDS-PAGE and transferred to nitrocellulose sheets ( Amersham , Saclay , France ) at 45 V overnight in 20 mM Tris–HCl , pH 8 . 3 , 192 mM glycine and 10% ethanol . Residual protein-binding sites were blocked by incubation with 1% bovine serum albumin ( BSA ) in PBS at 37°C for 1 h . The nitrocellulose was then incubated for 2 h at room temperature with either the anti-Tn mAb 83D4 or the anti-FhCL . After three washes with PBS containing 0 . 1% Tween-20 , the membrane was incubated for 1 h at room temperature with an anti-mouse or anti rabbit immunoglobulins , conjugated to peroxidase ( Dako , CA , USA ) diluted in PBS containing 0 . 1% Tween-20 and 0 , 5% BSA and reactions were developed with enhanced chemiluminiscence ( ECL ) ( Amersham , Saclay , France ) . The same procedure was performed omitting the primary antibodies as a negative control . BALB/c mice of 8 weeks old ( 5 per group ) were orally infected with 10 uncapped F . hepatica metacercariae ( Baldwin Aquatics , USA ) per animal . After 1 , 2 or 3 weeks of infection spleens , hepatic draining lymph nodes ( HLN ) and peritoneal exudates cells ( PECs ) were removed . PECs were harvested by washing the peritoneal cavity with 10 mL of cold PBS . Splenocytes , HLN , PECs ( 0 . 5–1 × 106 cells/mL ) or purified CD4+ T cells ( 0 . 2 × 106 cells/mL ) were cultured in complete medium consisting of RPMI-1640 with glutamine ( PAA Laboratories , Austria ) supplemented with 10% heat-inactivated fetal bovine serum ( FBS ) , 50 μM 2-mercaptoethanol , 100 U/ml penicillin and 100 mg/ml streptomycin ( Sigma-Aldrich , St . Louis , MO ) , in the presence or absence of FhTE ( 75 μg/ml ) , Concavalin-A ( ConA ) ( 5 μg/ml ) , FhmPox ( 75 μg/ml ) or the controls FhCB and CmPox , for 72 h at 37°C and 5% CO2 . IFNγ , IL-4 , IL-5 and IL-10 levels were evaluated by specific ELISAs or quantitative RT-PCR ( qRT-PCR ) . Uninfected naive animals were used as a control group . Proliferation with the control CmPox always provided background levels , such as with medium alone . Infections were also carried with 5 , 10 or 15 metacercariae/mouse and mice were sacrificed at 3 weeks after the infection . Alanine transaminase activity was measured in sera from infected and non-infected animals using a commercial kit ( Spinreact , Spain ) , according to the manufacturer’s instructions . The presence of flukes in livers from infected animals was analyzed so as to calculate fluke burden . Livers from infected mice presented macroscopic damage and/or necrosis . They were also histologically analyzed and found to present inflammatory infiltration and flukes , in some cases . IFNγ , IL-4 , IL-5 , IL-6 , IL-10 and IL-12p40/23 levels on culture supernatants were quantified by interleukin-specific sandwich ELISA assays ( BD Bioscience , NJ , USA ) . MIP-1α and MIP-2 were also detected in the culture media using sandwich ELISA assays , according to the instructions of the manufacturer ( RayBiotech , Inc . , GA , USA ) . In some cases , cytokines were detected by qRT-PCR using a Corbett Rotor Gene 6000 Real-Time PCR Machine and the SYBR Green 1 dye ( Applied Biosystem ) . Standard amplification conditions were 3 min at 95°C and 40 cycles of 10 s at 95°C , 30 s at 60°C , and 30 s at 72°C . For detection of cytokines the following primers were used: IFNγ: F: 5′-GGAGGAACTGGCAAAAGGATGGTGA-3′ and R: 5′-GCGCTGGACCT-GTGGGTTGT-3′; IL-4: F: 5′-AGGTCACAGGAGAAGGGACGCC-3′ and R: 5′-TGC-GAAGCACCTTGGAAGCCC-3′; IL-10: F: 5′-TTCCCAGTCGGCCAGAGCCA and R: 5′-GGGGAGAAATCGATGACAGCGCC-3′ . Results were expressed as the ratio between each evaluated cytokine and GAPDH expression . For GAPDH detection , sense and antisense primers were 5′-TCGGAGTCAACGGATTG-3′ and 5′-CCTGGAAGAT-GGTGATGG-3′ , respectively . The reactivity of polyclonal antibodies from infected or non-infected animals was evaluated by ELISA . Briefly , ninety-six-well microtiter plates ( Nunc , Roskilde , Denmark ) were coated overnight at 4°C with 1 μg/well of FhTE in 50 mM carbonate buffer ( pH 9 . 6 ) . After blocking with 1% BSA in PBS , three washes with PBS containing 0 . 1% Tween-20 were performed . For the oxidation of the glycan moieties of FhTE , wells were treated with 10 mM of sodium meta-periodate in 50 mM sodium acetate buffer pH 4 . 5 for 30 min at room temperature in the dark , washed with 50 mM sodium acetate buffer and subsequently incubated for 1 h with glycine 1% at room temperature . As controls ( Fhmock ) , wells were subjected to the same treatment except for the incubation with sodium meta-periodate . Serially diluted sera in buffer ( PBS containing 0 . 1% Tween-20 and 0 . 5% BSA ) were added to the wells for 1 h at 37°C . Following three washes , wells were treated 1 h at 37°C using goat anti-mouse polyvalent peroxidase-conjugate ( Sigma-Aldrich , St . Louis , MO ) and o-phenylenediamine-H2O2 was then added as substrate . Plates were read photometrically at 492 nm in an ELISA auto-reader ( Labsystems Multiskan MS , Finland ) . The lectin-reactivity on FhTE , FhmPox or FhCB lysates was evaluated by an ELISA-type assay . Briefly , Nunc microtiter plates were coated with 2 . 5 μg/well of parasite lysates and blocked with 1% BSA in PBS for 1 h at 37°C . Then , different concentrations of biotin coupled lectins were added and incubated for 1 h at 37°C . For inhibition assays , the lectins were pre-incubated for 30 min at 37°C with 50 mM of the indicated monosaccharide . After three washes , streptavidin conjugated to DyLight 800 was added to each well for 30 min at 37°C . Plates were then washed and analyzed with an Odyssey Infrared Imaging System ( LI-COR Biosciences , NE , USA ) . Lectins from Vicia villosa ( VV: GalNAc , Tn antigen ) , Triticum vulgaris ( WGA: ( GlcNAc ) 2 ) , Canavalia ensiformis ( ConA: αMan>αGlc ) , Arachis hpogaea ( PNA: βGal ( 1 , 3 ) GalNAc ) , Ulex europaeus ( UEA: Fucα ( 1 , 2 ) Gal ) , Erythrina cristagalli ( ECA: βGal ( 1–4 ) GlcNAc ) , Sambucus nigra ( SNA: αNeuAc ( 2 , 6 ) Gal ) and Helix pomatia ( HPM: GalNAc ) were used in this study . Splenocytes or PECs from infected and non-infected mice were washed twice with PBS containing 2% FBS and 0 . 1% sodium azide . Cells were then stained with different antibody mixes to identify DCs or macrophages . First , CD3+ cells were excluded from the gatings . DCs were defined as CD11chi F4/80- CD3- cells . Macrophages were identified as F4/80+ CD11c- CD3- cells . The following antibodies were used in these experiments: anti-CD3 ( 17A2 ) , CD11c ( N418 ) , CD40 ( HM40-3 ) , I-A/I-E ( 2G9 ) , F4/80 ( BM8 ) , CD80 ( 16-10A1 ) , CD86 ( GL1 ) . Cells were then washed twice with PBS containing 2% FBS and 0 . 1% sodium azide and fixed with 1% formaldehyde . Cell populations were analyzed using a CyAn ADP Analyzer ( Beckman Coulter ) . Antibodies were obtained from Affymetrix ( CA , USA ) or from BD-Biosciences ( CA , USA ) . IL-10 and IL-12/IL23p40 in vivo production by DCs or macrophages was analyzed by intracellular staining . Splenocytes and PECs from infected and non-infected mice were cultured for 6 h with GolgiPlug ( BD Biosciences ) , washed , stained with CD11c , F4/80 and CD3 , and then fixed and permeabilized using the Cytofix/Cytoperm kit ( BD Biosciences ) and subsequently stained with Abs specific for IL-12/23p40 or IL-10 ( Biolegend , CA , USA ) . Bone Marrow-derived Dendritic Cells ( BMDCs ) were generated from bone marrow precursors from BALB/c mice . Briefly , bone marrow precursor cells were harvested and plated at a density of 2–5 × 105 cells/ml in complete culture medium supplemented with GM-CSF-containing supernatant . After 3 days of culture at 37°C , the medium was replaced . Cells were recovered on day 8 and analyzed for the expression of CD11c by flow cytometry . To analyze DC-maturation , BMDCs ( 2 . 5 × 105/well ) were incubated at 37°C and 5% CO2 in 96-well plates with FhTE , FhPox , FhCB ( 75 μg/ml ) or medium alone in the presence or absence of LPS ( Escherichia coli 0111:B4 , 0 . 5–1 μg/ml ) overnight at 37°C . Alternatively , cells were pre incubated for 45 min . at 37°C with 10 mM of monosaccharides ( Man , GalNAc or arabinose ) or 10 μM of specific signaling inhibitors ( PHPS1; GW5074; and ER27319 ) . Cells were then centrifuged at 1 , 500 rpm for 5 min at 4°C and supernatants were then collected . Cytokine ( IL-12/23p40 , IL-10 and IL-6 ) levels were tested on culture supernatants by interleukin specific sandwich ELISA assays ( BD Bioscience , NJ , USA ) . The in vitro internalization and binding of the lysates were analyzed by flow cytometry . BMDCs were incubated ( 2 . 5 x 105/well ) with Alexa 647-labeled Ag for 1 h at 37°C in complete medium ( to assess uptake ) , or at 4°C in complete medium ( to assess binding ) . Cells were then washed twice and analyzed by FACS . For inhibition assays , cells were preincubated with 5 mM EDTA or 50 mM of different carbohydrates for 30 min at 37°C . The Student t test was used for statistical comparisons; p values <0 . 01 or <0 . 05 were considered to be statistically significant , depending on the experiment .
One of the objectives of this work was to evaluate the role of F . hepatica glycoconjugates structures in the induction of the host immune response . Indeed , the parasite induces a Th2 immune response with a regulatory component [3] . However , the parasite molecules involved in this immune-regulation are still unknown . Thus , we first evaluated different parameters of the immunity against F . hepatica in our experimental model and correlated them with the course of the infection and the level of liver damage . To this end , BALB/c mice were infected with 10 metacercariae and the production of IL-4 , IL-5 IL-10 and IFNγ was evaluated on splenocytes from infected animals at 1 , 2 and 3 weeks post-infection ( wpi ) . Splenocytes removed from infected animals and stimulated in vitro with a total parasite extract ( FhTE ) produced high levels of IL-4 , IL-5 and IL-10 , with significant higher levels of IL-4 and IL-10 as soon as the first wpi , reaching a plateau at week 2 after infection . On the other hand , IFNγ was slightly increased , although at very low levels ( ≤200 pg/ml ) , compared to the 30-fold increase of IL-10 ( around 6 ng/ml ) ( Fig 1A ) . The increase of IL-4 and IL-5 as well as the regulatory cytokine IL-10 coincided with the detection of liver damage evaluated by the alanine aminotransferase ( ALT ) activity in serum , a common marker to detect hepatic dysfunction [24] . Indeed , the ALT activity augmented 5-fold at 2 wpi , while it increased more than 20 fold at 3 wpi compared to levels detected in naïve/control animals ( Fig 1B ) . ALT activity also augmented with parasite dose of infection , confirming its usefulness for detecting liver damage and monitoring F . hepatica infection ( S1A Fig ) . The strong production of IL-4 and IL-10 at 3 wpi was confirmed by qRT-PCR , revealing around a 100-fold increase of IL-10 and only a 3-fold increase of IFNγ expression , with respect to uninfected animals ( Fig 1C ) . The strong modified Th2 polarization observed was also evidenced with a polyclonal stimulus , such as ConA , on splenocytes from infected animals . They produced higher levels of IL-4 , IL-5 and IL-10 , while their capacity to produce IFNγ was significantly diminished ( a 3-fold decrease ) , comparing to splenocytes from naïve animals stimulated in the same conditions ( Fig 1D ) . These results are in agreement with previous work describing the Th2 polarization induced by F . hepatica during infection [3] . Carbohydrate structures produced by parasites participate in critical processes such as infection or invasion [25–26] . Although much advance in the area of glycomics has been gained in recent years , the knowledge about the structure and function of F . hepatica glycans is still poor . In order to identify the carbohydrates present in the FhTE used in this study we carried out lectin-reactivity assays with a panel of different vegetal lectins . Glycoconjugates from FhTE strongly reacted with lectins from Vicia Villosa ( VV ) , Triticum vulgaris ( WGA ) , Canavalia ensiformis ( ConA ) , Arachis hpogaea ( PNA ) and Ulex europaeus ( UEA ) ( Fig 2A ) , revealing the presence of N-acetil-galactosamine-Ser/Thr ( GalNAc-Ser/Thr ) , N-acetyl-glucosamine ( GlcNAc ) 2 , mannose ( Man ) or glucose ( Glc ) , galactose ( Gal ) in ( βGal ( 1–3 ) GalNAc and fucose ( Fuc ) in Fucα ( 1–2 ) Gal , respectively . We also performed inhibition assays with specific monocarbohydrates and with non specific carbohydrates as negative control . More than the 70% of the lectin reactivity was lost when incubating with GalNAc , GlcNAc or Man and VV , WGA or ConA , respectively , confirming the carbohydrate specificity by these lectins ( Fig 2B ) . Finally , we carried out the assays on oxidized FhTE ( FhmPox ) . Mild periodate oxidation of glycans is usually used to evaluate the functional roles of glycoconjugates [11 , 15] . During this process the glycol groups in carbohydrates are oxidized to reactive aldehyde groups , which are in turn reduced with sodium borohydride . Thus , the structure of carbohydrates is lost , as well as the possible biological activity that they can mediate . The recognition of the specific carbohydrates on FhTE by most reactive lectins was completely abrogated with meta-periodate oxidation ( FhmPox ) ( Fig 2B ) . Importantly , the integrity of oxidized parasite glycoconjugates ( FhmPox ) remained unchanged since the electrophoretic mobility of FhTE molecular components was similar to that of FhmPox ( Fig 2C ) . Furthermore , oxidation with meta-periodate totally abolished the recognition of a monoclonal antibody specific for the GalNAc-O-Ser/Thr carbohydrate structure that is reactive to FhTE [27] , while it did not modify the recognition of cathepsin-L1 by a specific polyclonal antibody ( Fig 2D ) . Thus , the chemical oxidation of terminal carbohydrates abrogates recognition of glycans by carbohydrate binding proteins , resulting in an adequate strategy to study their recognition by proteins or receptors and could be useful to study their biological functions . To determine whether F . hepatica glycosylated molecules participate in the induction of high levels of the Th2-type cytokines and the suppression of IFNγ we cultured splenocytes from infected animals with oxidized parasite components . Thus , splenocytes removed from infected animals at 3 wpi were in vitro stimulated with FhTE , and the production of cytokines in the culture supernatant was determined and compared to those incubated with oxidized parasite components ( FhmPox ) . Splenocytes stimulated with FhmPox produced lower levels of IL-4 and IL-10 than cells incubated with FhTE . Surprisingly , in these conditions , spleen cells produced significant higher levels of IFNγ ( Fig 3A ) . On the other hand , the levels of IL-5 remained unchanged . Splenocytes incubated with the control FhCB , consisting in FhTE subjected to the whole treatment excepting for the incubation with sodium periodate , behaved essentially as cells in presence of FhTE , as expected . The cytokine production was specific of parasite components since stimulated spleen cells from non-infected animals did not produce any of the evaluated cytokines . The oxidation-dependent decrease of IL-4 and IL-10 by splenocytes from infected animals was also confirmed by qRT-PCR , although no significant difference was found between the production levels of IFNγ by FhmPox-stimulated splenocytes ( Fig 3B ) . Cells from the hepatic draining lymph nodes of infected animals also produced decreased levels of IL-10 when stimulated with oxidized parasite components , with no changes in the production of IL-5 ( Fig 3C ) . Finally , in order to establish a possible role of glyans in the recognition of parasite molecules by the humoral immune response , we evaluated the recognition of FhmPox by sera from infected animals . IgG antibodies from infected animals recognized in a similar manner both FhTE and FhmPox , suggesting that oxidation of terminal glycans does not modify the recognition of parasite components by antibodies induced during infection ( Fig 3D ) . Evidence demonstrating that F . hepatica components can modulate DC-maturation and function in vitro has been previously reported [5–6 , 10 , 28] . Also , the phenotype of DCs has been evaluated suggesting that the parasite immune-modulates DCs upon infection [13] . Nevertheless , an exhaustive study of DC immune-modulation by the parasite has not been carried out . Thus , in order to deeply evaluate their role in the host immune-modulation by F . hepatica , we sought to evaluate DCs in vivo both in the spleen and in the peritoneum of infected animals . Upon infection , we observed a marked recruitment of cells in the spleen and in the peritoneal cavity of infected animals , which increased with time of infection ( Fig 4A and 4B , respectively ) . Among these cells , DCs ( defined as CD11chi F4/80- cells ) were recruited both at the spleen and the peritoneum since the first wpi , and their number augmented with the course of infection ( Fig 4A and 4B ) . In spite of the fact that all DCs were MHC class II positive , they presented remarkable decreased levels of MHCII expression ( Fig 4C–4E ) . Indeed , splenic DCs presented around 50% reduction of MHCII expression since the first wpi together with lower levels of CD40 on their surface ( S1B Fig ) . MHC class II- or CD40-decreased expression on DC surface was not modified when infecting animals with higher parasite dose ( S1B Fig ) . Moreover , from the second wpi , an increase of IL-10 secreting splenic DCs was evidenced ( Figs 4D and S2 ) . Strikingly , IL-12+ DCs also augmented in the spleen ( Fig 4D ) . DCs recruited into the peritoneum also reduced the expression of MHC class II ( Fig 4E ) while the expression of CD80 and CD86 was increased ( S1C Fig ) . There was also an increase of IL-10 secreting DCs in the peritoneal cavity , while IL-12 secreting DCs remained very low in the peritoneum ( Figs 4E and S2 ) . Macrophages , defined as F4/80+ CD11c- cells were also recruited to the spleen and peritoneum upon infection ( S3 Fig ) . In spleen , the recruitment of IL-10+ macrophages was favored while IL-12 secreting macrophages considerably decreased ( S2A and S3C Figs ) . On the other hand , macrophages from the peritoneal cavity increased the expression of surface MHC class II from the first wpi . IL-10+ or IL-12+ macrophages were also increased in the peritoneum ( S2 and S3D Figs ) . In order to determine whether parasite glycan structures can also modulate the function of DCs , we evaluated the in vitro activation of splenocytes ( Fig 5B ) and purified CD4+ T cells ( Fig 5C ) from infected animals by DCs . To this end , bone marrow derived DCs ( BMDCs ) were loaded with parasite-derived components ( FhTE ) or with oxidized total lysate ( FhmPox ) . A control consisting on FhTE subjected to the chemical process in absence of meta-periodate was also included ( FhCB ) . Importantly , the viability of loaded-BMDC was not affected by the treatment with different parasite lysates ( Fig 5A ) . Then , loaded BMDCs were washed and subsequently incubated with splenocytes or purified CD4+ T cells from infected animals . As seen in Fig 5B and 5C , when incubated with FhmPox-loaded BMDCs , splenocytes and purified CD4+ T cells produced lower levels of IL-4 and IL-10 than cells stimulated with FhTE-loaded BMDCs , while IL-5 and IFNγ production remained unchanged . No cytokine production was detected when loaded BMDCs were incubated with splenocytes or purified CD4+ T cells from uninfected animals ( naïve ) ( Fig 5B and 5C ) . These results strongly suggest that glycoconjugates modulate DC-stimulatory capacity by increasing the production of IL-4 and IL-10 by CD4+ T cells during infection . Glycans can modulate DC function through a variety of mechanisms . For instance , they can interact with lectin receptors expressed on the surface of DCs that can endocytose the glycan components and/or signal through kinase-dependent cascades [29] . Thus , we evaluated whether F . hepatica glycoconjugates could interact with the DC surface , or be internalized by DCs . To this end , Atto-647-labeled FhTE was incubated with DCs both at 4°C ( to evaluate binding ) or at 37°C ( to test internalization ) and the fluorescence intensity was determined by flow cytometry . As shown in Fig 6A and 6B , FhTE both interacted with and was internalized by DCs . To determine whether this binding or uptake was dependent on C-type lectin receptors , requiring Ca2+ for binding , we incubated DCs with FhTE in presence of EDTA , a chelating agent . We observed that around 70% of the FhTE internalization was abrogated with EDTA incubation , indicating that the internalization process was mediated by Ca2+-dependent lectin receptors ( Fig 6A ) , On the contrary , only 25% of the binding was inhibited in presence of EDTA ( Fig 6A ) , suggesting that CLR-dependent recognition of glycosylated molecules is less relevant in this process . In order to confirm the participation of glycoconjugates in the internalization or binding of parasite components to DCs , FhCB and FhmPox were also stained with Atto-647 and further incubated with DCs in the same conditions as FhTE . Oxidation of parasite glycans resulted in a decrease of around 50% of FhTE internalization , while only 25% of reduction in the binding was observed ( Fig 6B ) , indicating that they partially mediate binding and internalization of parasite components by DCs . As expected , no significant difference between FhTE and FhCB ( control ) binding or internalization was observed ( Fig 6B ) . Finally , we carried out inhibition assays with several carbohydrates , as well with laminarin , a ligand of the CLR Dectin-1 [30] . Laminarin was included in these assays since Dectin-1 seems to mediate the binding of F . hepatica glycoconjugates by macrophages [22] . As shown in Fig 6C , the binding of FhTE was inhibited by both Man and GalNAc , while only Man was capable of inhibiting FhTE uptake by DCs . Incubation with laminarin did not significantly modify FhTE binding or internalization by DCs , indicating that Dectin-1 is not involved in this process . In order to evaluate whether parasite glycosylated molecules are able to influence DC-maturation we incubated BMDCs with parasite components or oxidized-FhTE in absence or presence of a maturation stimulus ( LPS ) . Furthermore , the control FhCB ( consisting in FhTE subjected to the whole treatment excepting for the incubation with sodium periodate ) was also included . Then , we evaluated the production of several cytokines by DCs . BMDCs incubated with FhTE in presence of LPS produced higher levels of IL-10 than DCs incubated only with LPS . Interestingly , when oxidized-parasite glycans ( FhmPox ) where incubated with BMDCs , they produced similar levels of IL-10 than those produced by cells incubated with LPS alone , indicating that oxidation of glycans abrogates the immunomodulatory activity of FhTE on DCs ( Fig 7A ) . On the other hand , the opposite situation was observed with the pro-inflammatory cytokines IL-6 and IL-12/23p40 . Indeed , FhTE incubated with BMDCs in presence of LPS decreased the production of IL-6 and IL-12/23p40 induced by LPS , while FhTE oxidation restored the levels of both cytokines ( Fig 7A ) . As expected , the control FhCB/LPS behaved essentially in the same way as FhTE/LPS , while the CmPox/LPS condition induced the same levels of cytokine production than cells incubated only with LPS . Altogether , these results indicate that glycoconjugates from F . hepatica modulate LPS-induced maturation of DCs by augmenting anti-inflammatory cytokines and decreasing pro-inflammatory cytokines . Next , we carried out inhibition assays of the immunomodulatory capacity of FhTE with the carbohydrates Man and GalNAc since they were capable of inhibiting binding or internalization of FhTE . An irrelevant carbohydrate , arabinose ( Ara ) was used . As shown in Fig 7B , only Man could inhibit the immune-modulation of FhTE on DC-maturation . Indeed , incubation with Man restored the levels of IL-10 and IL-12/23p40 production induced by LPS alone on DCs . Nevertheless , the levels of IL-6 were not restored by inhibition assays with Man , suggesting that the production of this cytokine is triggered by a Man-independent signaling process . Incubation of DCs with FhTE in presence of GalNAc or Ara did not modify the levels of the evaluated cytokines ( Fig 7B ) . The production of the inflammatory chemokines MIP-1α and MIP-2 by DCs was also investigated . These chemokines regulate the influx of inflammatory cells , and can be produced by DCs under pathogenic conditions . MIP-1α is a ligand for CCR5 , a chemokine receptor expressed on Th1 cells , while MIP-2 selectively chemoattracts Th2 cells [31] . DCs stimulated with LPS produced both chemokines . When incubated with FhTE the production of MIP-1α significantly decreased , while MIP-2 increased ( Fig 7C ) . However , inhibition assays with Man did not modify the production of either chemokines induced by LPS/FhTE , suggesting that the production of these chemokines does not depend on Man-specific receptors on DCs . Finally , in order to provide evidence about the possible CLR implicated in the recognition of Man-containing glycans from F . hepatica , we performed cell cultures in presence of chemical inhibitors of different signaling pathways: GW5074 , ER27319 and PHPS1 that inhibit pathways mediated by Raf-1 , Syk and Shp2 , respectively . This inhibitors were chosen since a group of CLRs that recognize Man residues from pathogens , such as Dectin-1 , Man Receptor , SIGNR or DCIR , are expressed on BMDCs and signal through these molecular mediators [29] . Thus , after incubation of these molecules , we evaluated the production of IL-12/23p40 . Either GW5074 or ER27319 did not modify the decrease of IL-12/23p40 production induced by FhTE in presence of LPS . Nevertheless , incubation of DCs with PHPS1 in presence of FhTE/LPS completely abrogated the immunomodulatory effect of FhTE on DCs , leading to the production of similar levels of IL-12/23p40 as those obtained with LPS ( Fig 7D ) .
Carbohydrate structures can exert different biological functions , ranging from cell growth or development to tumor growth or metastasis . Moreover , glycans participate in diverse processes such as coagulation , induction of immunity , cell-cell communication or microbial pathogenesis [32–33] . In this context , accumulating evidence demonstrates that glycoconjugates produced by helminths favor parasite survival by influencing the host immune response [34–35] . In this work we show that F . hepatica glycoconjugates are involved in the induction of high levels of IL-10 and IL-4 as well as in the reduction of IFNγ production by splenocytes during infection . In this sense , different previous reports used meta-periodate treatment of glycans to identify the role of glycoconjugates in the regulation of host immunity . Sodium meta-periodate treatment at low concentration does not remove glycans or compromise the integrity of glycoproteins , but instead opens up the “chair” structure altering molecular conformation of the glycans . Thus , when F . hepatica components were treated with meta-periodate and used to stimulate spleen cells from infected animals , they reduced their capacity to produce the Th2 cytokine IL-4 as well as the regulatory cytokine IL-10 , suggesting a role of glycoconjugates in the induction of Th2/regulatory T cell immune response . Interestingly , stimulation of splenocytes from infected mice with oxidized parasite components also increased IFNγ production compared to the IFNγ levels obtained with F . hepatica total lysate containing non-modified glycans , indicating that they could suppress specific Th1 responses . Glycoconjugates from other helminths such as Schistosoma mansoni [36] and Brugia malayi [11] have also been reported to regulate the host immune shift toward a Th2 response , or inducing regulatory responses via induction of IL-10 [11 , 37–38] . Indeed , S . mansoni produces the glycan structure Lewisx ( Galβ1-4 ( Fucα1–3 ) GlcNAc-terminal structure ) that possesses immune-regulatory properties and accounts for induction of host IL-10 induced by the parasite [36] . Fasciola hepatica , however , does not express this glycan structure [39] . Thus , the immune-regulatory glycan structures from this parasite remain unknown . Although carbohydrate moieties from parasites such as Echinococcus [40–42] , S . mansoni [20 , 43–45] , among others [35 , 46] , have been well determined , glycoconjugates produced by F . hepatica still remained poorly characterized . Previous works have identified the presence of Gal ( β1–6 ) Gal and GlcNAc ( α1-HPO3-6 ) Gal terminating glycolipids by mass spectrometry [20 , 47] , as well as glycans carrying Man , Glc , GlcNAc or GalNAc by lectin reactivity [17–19 , 48] both in tegument and tissues throughout the parasite . Here , we identified the presence of glycan structures containing Man/Glc , GalNAc or GlcNAc in the parasite lysate used in this study , confirming the strong binding of Man/Glc , GalNAc and GlcNAc-reactive lectins ( ConA , WGA and VV , respectively ) . It is worth noting that , in our case , the lectin reactivity was abrogated when incubating with specific sugars , confirming the specificity for their respective carbohydrates . Our results also showed a slight recognition of parasite glycans by the Ulex europeus agluttinin ( UEA-1 ) specific for terminal Fuc residues . However , it should be noted that the reactivity was lower than that observed with the other lectins , and that inhibition of lectin binding with Fucose did not completely abolish lectin recognition . Nevertheless , several helminths , including F . hepatica , are reactive to the Lotus tetragonolobus agglutinin , which binds Fucα1-3GlcNAc , confirming the expression of α1 , 3-fucosylated glycans [39] . In all cases , lectin reactivity was also abolished with meta-periodate oxidation of parasite glycans , demonstrating that this procedure suppressed carbohydrate specific recognition by lectins and is suitable for studying the biological role of glycans . To evaluate possible mechanisms that could explain the modulation of host immunity by F . hepatica glycoconjugates we focused on DCs , the most effective antigen presenting cells that possess the ability to stimulate naive T cells , inducing a specific Th polarization [12] . In order to guarantee the identity of selected DCs , we excluded both macrophages and CD3+ cells and then selected CD11chi cells by flow cytometry analyses . Thus , DCs ( CD3- F4/80- CD11chi cells ) from infected animals presented a different phenotype than DCs from naïve animals , characterized by a profound decrease of MHC class II expression on DC-surface , which was not found in macrophages ( CD3- F4/80+ CD11c- cells ) , probably due to a different interaction between these cells and parasite molecules . DCs recruited to the peritoneum also showed an up-regulation of the co-stimulatory molecules CD80 and CD86 , suggesting that these DCs acquire a semi-mature phenotype upon parasite infection . The expression of MHC class II-peptide complexes on the surface of DCs is essential for their ability to activate CD4+ T cells efficiently . Apparently , F . hepatica would restrict the capacity of DCs to present antigens to CD4+ T cells as well as to prime specific CD4+ T cells , an effect that increases with persistence of infection . It has been previously shown that ubiquitination of MHCII-peptide complexes regulates their surface expression , retention and degradation in DCs [49–51] , and that certain pathogens , such as Salmonella typhimurium , induce polyubiquitination of HLA-DR , resulting in a reduced surface expression of all MHC class II isotypes [52] . On the other hand , there are evidences reporting that Mycobacterium tuberculosis diminishes MHC-II synthesis by macrophages [53] in a process dependent on TLR2 ligation [54] , limiting antigen presentation . It would be interesting to evaluate whether any of these molecular mechanisms underlie the reduced expression of MHC class II on the surface of DCs from F . hepatica-infected animals . There are also pieces of evidence highlighting the role of IL-10 in inducing immune anergy by reducing expression of MHC class II on the surface of antigen presenting cells [55] . Thus , the possibility that IL-10 secreted by CD4+ T cells during infection is responsible of MHC class II decrease on DCs cannot be ruled out . To evaluate the influence of glycoconjugates parasites in the specific stimulatory capacity of DCs conditioned with F . hepatica components , we co-cultured FhTE-pulsed BMDCs with CD4+ T cells from infected and non-infected animals . Cultures from infected animals produced high levels of IL-4 , IL-5 and IL-10 with low levels of IFNγ . However , when parasite oxidized components were used for BMDC loading , CD4+ T cells from infected animals reduced their capacity to produce IL-4 and IL-10 , indicating that parasite glycoconjugates are involved in the DC-triggered production of IL-4 and IL-10 by T cells . Thus , our work demonstrates for the first time the role of F . hepatica glycoconjugates in immunomodulating the host immune response during infection and , in particular , by modulating T-cell stimulatory capacity of DCs . Regulatory DCs can exert their functions through different mechanisms , such as by decreasing their capacity of antigen presentation or by inducing regulatory T cells able to suppress inflammatory Th1/Th17 responses [56] . In this context , it is interesting to remark that antigen presenting cells from the peritoneal cavity of infected animals , but not CD11c+ spleen cells , have limited capacity to induce effector T cell response [4] , although their ability to directly induce IL-10 secreting regulatory T cells has not been evaluated . In an attempt to gain more insight in the process of glycan-mediated immune-modulation of DCs , we focused on the study of the effects of glycoconjugates on DC-maturation in vitro . Importantly , previous reports demonstrate the capacity of F . hepatica components to modulate in vitro TLR-induced maturation of DCs and/or their stimulatory function [5–7 , 16] . However , the role of parasite glycans in this immune-modulation had not been previously addressed . When BMDCs were matured in the presence of parasite components they secreted higher levels of IL-10 and lower levels of IL-6 and IL-12/23p40 than cells stimulated only with LPS . On the other hand , when DCs were matured with LPS in presence of oxidized parasite components the production levels of IL-6 , IL-10 and IL12/23p40 were restored , indicating that glycoconjugates from F . hepatica mediate modulation of TLR-induced maturation of DCs . One possible mechanism that can account for this immune-modulation is by triggering carbohydrate specific receptors , such as CLRs , that can cross-talk with TLR-stimulation . Indeed , we found that both binding and uptake of parasite components by DCs were inhibited with EDTA or with Man , suggesting a CLR mediated process of recognition and uptake of parasite glycoconjugates through Man-containing glycans . On the other hand , GalNAc inhibited binding but not uptake by DCs , indicating that GalNAc-residues on parasite components can interact with receptors on the surface of DCs , but do not mediate antigen internalization . Strikingly , incubation with laminarin , a ligand of Dectin-1 , did not inhibit either the binding or the uptake of parasite components present on FhTE by BMDCs . It has been recently published that Dectin-1 is involved in the induction of CD4+ T cell anergy by macrophages [23] . However , it should be noted that in this case excretory-secretory products from F . hepatica were used . Since the immunomodulatory properties of these parasite products on macrophages were found to depend on Dectin-1 signaling , it is likely that the different carbohydrate composition and/or nature of molecules present on FhTE ( total lysate ) and excretory-secretory parasite products determine the different CLR-mediated signaling triggered by F . hepatica components . In this sense , in previous elegant studies using Helicobacter pylori variants which differ in their expression of Lewis glycan antigens , only the variants expressing the Lewis antigen were capable to bind to DC-SIGN and block the induction of a specific Th1 response , while Lewis-negative H . pylori variants were not [57] . Eventual studies on the elucidation of the carbohydrate moieties present on F . hepatica total lysate and excretory-secretory products will be decisive to explain their different immunomodulatory properties . By carrying out DC-maturation assays in presence of Man , we provide evidence that a Man-specific CLR expressed on DC surface mediates the immunomodulatory effects of F . hepatica components . Indeed , Man incubation restored the production levels of IL-10 and IL-12/23p40 , but not those of IL-6 , MIP-1α or MIP-2 . Several CLRs have been reported to cross-talk with TLR-signaling on DCs as well as on other myeloid cells , inducing an increase of IL-10 and a decrease of pro-inflammatory cytokines [29 , 58] . Furthermore , glycans from helminths can interact with CLRs on DCs and regulate their maturation . For instance , S . mansoni , glycans are recognized by the Man receptor resulting in a Th2-polarized cell response [10] . Also , the glycosylated molecules of the whipworm T . suis interact with the Man receptor , DC-SIGN and MGL , which recognize Man and terminal GalNAc residues , respectively , and suppress TNFα production by DCs stimulated with LPS [15] . According to our experimental results obtained with different inhibitors of molecules that participate in Man-specific CLR signaling , the phosphatase SHP2 would participate in the Man-triggered signaling . This indicates a possible role of SHP2-dependent CLR signaling in the recognition of F . hepatica glycans and cross-talk with TLR4-tiggering . One possible candidate is DCIR , a CLR identified on mouse BMDCs [59] and other myeloid cells [29] , that can modulate TLR-induced gene expression at the transcriptional or post-transcriptional level . However , DCIR does not induce gene expression in the absence of other PRR signaling [29] . DCIR-triggering induces the phosphorylation of its ITIM ( Immunoreceptor Tyrosine-based Inhibitory Motif ) , which recruits the phosphatases SHP1 or SHP2 to its cytoplasmic domain [60–61] , and results in the inhibition of TLR8-mediated IL-12 and TNFα production or TLR9-induced IFNα and TNFα production by DCs [62–63] . Experiments are on their way to determine the role of DCIR in the immune-modulation induced by F . hepatica components , by specifically silencing DCIR expression on BMDCs . The possible role of DCIR in the recognition of Man residues from F . hepatica and in mediating the cross-talk with TLR4 signaling could also explain the fact that both IL-10+ DCs and IL-12+ DCs were identified in the spleen from infected animals . Indeed , splenic DC subsets differentially express membrane molecules , some of which are CLRs [64–65] . In mouse spleen two different DC-subsets can be identified , CD11b+ DCs and CD8α+ CD11b- DCs , which differ in antigen presentation and T-cell stimulatory capacity [64] . In fact , CD8α+ splenic DCs are the major subset responsible for cross-presenting cellular antigens [66] . On the other hand , CD8α− DCs preferentially present antigen to CD4+ T cells [67] . These DC subsets also express different sets of surface molecules , including distinct CLRs . Interestingly splenic CD8α- DCs have been reported to express DCIR , while CD8α+ DCs do not [67] . Thus , taking these findings into account , one could speculate that CD8α- DCs expressing DCIR are the main target of immune regulation between different DCs found in mouse spleen by F . hepatica Man-containing glycans . Furthermore , a possible DCIR signaling could explain the immunomodulatory effects of F . hepatica glycoconjugates on splenocytes , since splenic B cells also express DCIR [60] . Additional experiments are needed to support this hypothesis . In spite of the hypothesized role of DCIR in mediating immune-modulation by F . hepatica Man-glycans , inhibition culture assays with Man did not restore IL-6 , MIP-1α or MIP-2 production . Thus , it is likely that another receptor or glycan structure is participating in the immune-modulation process that could not be detected with our assays . It has recently been reported that F . hepatica glycans interact with the DC-SIGN receptor on human DCs . Interestingly when DC-SIGN interacts with Man-rich glycans , modulates the production of pro-inflammatory TLR-induced cytokines by a pathway that depends on the activation of Raf-1 [68] . In fact , the interaction of F . hepatica glycans and DC-SIGN on human DCs , in presence of LPS , induces follicular T helper cell differentiation via IL-27 [16] . However , although the authors assume that this process is Fuc-dependent , the identity of the immunomodulatory glycoconjugates from F . hepatica was not investigated . In mice , however , there are eight DC-SIGN homologs that do not recognize exactly the same glycans as DC-SIGN [69] . According to their glycan specificity , SIGNR1 and SIGNR3 are the closest candidates to fulfill DC-SIGN function in mice [69] . Interestingly , SIGNR3 has been shown to modulate M . tuberculosis immune responses , by a signaling dependent on the tyrosine kinase Syk [70] . Indeed , resistance to M . tuberculosis is impaired in SIGNR3-deficient animals [70] . The fact that our results demonstrate that modulation of TLR-induced maturation of DCs by F . hepatica components does not depend on either Raf-1- or Syk-mediated signaling would exclude the participation of these receptors in this process . Nevertheless , additional studies are needed to elucidate a possible role of other SIGNR receptors in the immune modulation by F . hepatica on mice DCs . In conclusion , the results reported here demonstrate that glycoconjugates from F . hepatica are involved in the induction of high levels of IL-10 and IL-4 by the parasite and are in agreement with the increasing evidence supporting a role for helminth glycans in regulation of the host immune shift toward a Th2/regulatory response via induction of IL-10 . Furthermore , we show that F . hepatica glycoconjugates interact with DCs and modulate DC-function and -maturation by a SHP2-dependent CLR that is inhibited by Man residues . We are currently working on the identification of these glycans and the C-type lectin receptors on DCs that participate in their recognition . These results contribute to the understanding of the role of parasite glycans in the modulation of the host immunity and might be useful in the design of vaccines against fasciolosis . | Fasciola hepatica is a helminth that infects mainly ruminants , causing great economic losses worldwide . Importantly , fasciolosis is also considered an emerging zoonosis with an increasing number of human infections globally . As other helminths , F . hepatica is able to regulate the host immune response favoring parasite survival in the host . In this work we investigated whether glycoconjugates produced by this parasite play a role in the host immune-regulation . Glycans , composed by carbohydrate chains , participate in important biological processes , but their role during Fasciola infection has not been previously addressed . We found that glycoconjugates are involved in the production of the regulatory cytokine IL-10 and in the production of the Th2-like cytokines IL-4 . Furthermore , we found that they are also involved in the modulation of dendritic cell maturation , the most efficient antigen presenting cells . Indeed , the parasite is able to inhibit the maturation of dendritic cells in a process that is glycan-mediated and dependent on a mannose-specific receptor . In conclusion , our results highlight the importance of parasite glycoconjugates in the modulation of host immunity and might be applied in the design of vaccine strategies to prevent infection . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [] | 2015 | Glycans from Fasciola hepatica Modulate the Host Immune Response and TLR-Induced Maturation of Dendritic Cells |
Toxoplasma gondii possesses sets of dense granule proteins ( GRAs ) that either assemble at , or cross the parasitophorous vacuole membrane ( PVM ) and exhibit motifs resembling the HT/PEXEL previously identified in a repertoire of exported Plasmodium proteins . Within Plasmodium spp . , cleavage of the HT/PEXEL motif by the endoplasmic reticulum-resident protease Plasmepsin V precedes trafficking to and export across the PVM of proteins involved in pathogenicity and host cell remodelling . Here , we have functionally characterized the T . gondii aspartyl protease 5 ( ASP5 ) , a Golgi-resident protease that is phylogenetically related to Plasmepsin V . We show that deletion of ASP5 causes a significant loss in parasite fitness in vitro and an altered virulence in vivo . Furthermore , we reveal that ASP5 is necessary for the cleavage of GRA16 , GRA19 and GRA20 at the PEXEL-like motif . In the absence of ASP5 , the intravacuolar nanotubular network disappears and several GRAs fail to localize to the PVM , while GRA16 and GRA24 , both known to be targeted to the host cell nucleus , are retained within the vacuolar space . Additionally , hypermigration of dendritic cells and bradyzoite cyst wall formation are impaired , critically impacting on parasite dissemination and persistence . Overall , the absence of ASP5 dramatically compromises the parasite’s ability to modulate host signalling pathways and immune responses .
The phylum Apicomplexa groups obligate protozoan parasites that are the causative agents of severe diseases in humans and animals such as malaria , toxoplasmosis , babesiosis and coccidiosis . The key process of invasion and subsequent multiplication within their host cells is maintained via secretion from three distinct phylum-specific organelles termed the micronemes , rhoptries and dense granules [1–3] . Plasmodium falciparum is the most notorious member of the Apicomplexa in terms of its impact upon human health [4] . During its intraerythrocytic stage development , P . falciparum modulates the infected red blood cell by exporting a large repertoire of proteins to impact notably on nutrient acquisition , rosetting and cytoadherence [5] . This host cell modulation is governed by the export of various effector proteins , many of which contain a plasmodium export element RxLxE/Q/D ( PEXEL ) , whilst a smaller repertoire of exported proteins lack this motif and are thus termed PEXEL-negative exported proteins [6] . The protease underpinning this cleavage event , Plasmepsin V , is an integral membrane protein localized to the endoplasmic reticulum ( ER ) in an orientation such that the catalytic aspartyl protease domain faces the ER lumen [7] . Plasmepsin V cleaves the PEXEL motif after the leucine residue , ensuring secretion of the effectors into the host erythrocyte and subsequent parasite survival [8–10] . Toxoplasma gondii is amongst the most widely distributed parasites with nearly half of the human population chronically infected . Infection during pregnancy can result in severe neurological birth defects , whilst fatal cerebral toxoplasmosis can occur in association with immunosuppressive diseases and treatments . T . gondii follows a complex life cycle involving a haploid replicative stage , followed by chronic encystation in a broad range of intermediate hosts , and meiosis in the intestine of the definitive felid host [11] . Intermediate hosts are infected either by ingestion of either oocysts shed by felids , or by bradyzoites in tissue cysts within infected meat . The fast-replicating tachyzoites are responsible for the acute stage of infection and dissemination into all tissues , whereas the slow growing bradyzoite stage establishes chronic infection with resultant cysts predominantly found in the brain and striated muscle . Following host cell invasion , T . gondii tachyzoites and bradyzoites are surrounded by a PVM that resists endo/lysosomal fusion [12] and secludes them from the host cell cytosol . Unlike rhoptry effector proteins that are secreted at the onset of invasion , dense granule proteins ( GRAs ) are secreted once the parasite resides within the PV . GRAs have been implicated in a variety of processes linked to the establishment of parasitism , including the formation of the membranous nanotubular network ( MNN ) produced within the PV at the posterior side of invading parasites [13 , 14] . More recently , several GRAs have been demonstrated to cross the PVM and subvert host cellular functions . Specifically , GRA15 modulates host cell signalling pathways by NF-kB nuclear translocation and NF-kB–mediated transcription of cytokines and other effector molecules [15] . GRA16 reaches the host cell nucleus , where it positively modulates genes involved in cell cycle progression and the p53 tumour suppressor pathway [16] . GRA24 modulates the early immune response to infection by promoting host p38 MAPK activation [17] . The activity of these proteins within the host cell is suggestive of the presence of an export pathway similar to that of Plasmodium spp . In this context , the T . gondii genome was reported to contain genes exhibiting a signal peptide in combination with proximal sequences reminiscent of the PEXEL motif ( PEXEL-like motif ) [18] . Some of these genes encode novel dense granule proteins; GRA19 , GRA20 and GRA21 , that are cleaved at the PEXEL-like motif , however do not cross the PVM and are instead incorporated into the PV and PVM . Furthermore , the previously reported GRA3 , GRA5 , and GRA15 were also found to contain the exact PEXEL consensus motif RxLxD/E , whilst most of the other identified GRAs possess an N-terminal motif resembling the pattern RxLxD/E within approximately 140 residues of their predicted start methionine [18] . The presence of PEXEL-like motifs on several GRAs spoke for the existence of an aspartyl protease involved in cleavage and export of these effector proteins in T . gondii . Of the seven aspartic protease paralogues encoded by the T . gondii genome ( ASPs ) , ASP5 and ASP7 are members of the evolutionarily distinguished group ( D ) of apicomplexan aspartic proteases comprising Plasmepsin V [19] . Since ASP7 is not expressed in tachyzoites and bradyzoites , ASP5 emerges as the most promising candidate to cleave T . gondii PEXEL-like motif containing proteins . Here we report the functional characterization of ASP5 in T . gondii type I and type II strains . We demonstrate that ASP5 is responsible for the cleavage of GRAs containing PEXEL-like motifs and is necessary for the export of GRAs beyond the PVM into the host cells . We also report the broader phenotypic consequences of the absence of ASP5: severely compromised parasite fitness , a block in the formation of the MNN , an inability to enhance dendritic cell ( DC ) migration , a dramatic remodelling of the host immune response , and an impairment in cyst wall formation without any impact on tachyzoite to bradyzoite conversion .
ASP5 has previously been described as a Golgi-resident protein when expressed as an epitope-tagged second copy [19] . To determine its role and importance in T . gondii , we first confirmed the localization of ASP5 by inserting a 3Ty-epitope tag at the carboxyl-terminus of the endogenous ASP5 locus in both type I ( RHΔku80 ) and type II ( Prugniaud , PRUΔku80 ) strains . Endogenous ASP5-3Ty co-localizes with the Golgi marker GRASP ( Fig 1A ) and shows two forms by western blot analyses that have not been previously reported . The 100 kDa band is in agreement with the predicted full length protein size ( 108 kDa ) , whereas a smaller form migrates with an apparent molecular weight of 55 kDa ( short-ASP5 ) . Markedly , short-ASP5 is not detectable when a cDNA copy of the gene is expressed in the parasites ( Fig 1B and 1C ) . To determine if the short-ASP5 form identified by western blot is the result of a processing event , a pulse-chase experiment followed by co-immunoprecipitation ( IP ) was performed using anti-Ty antibodies on 35S-methionine metabolically labelled ASP5-3Ty expressing parasites . Even during the short pulse , short-ASP5 is readily detectable suggesting that it originates either from an alternative transcriptional initiation , splicing or translational start , but not from a processing maturation event ( S1A Fig ) . To functionally characterize ASP5 , knockout mutants were generated in type I and II strains . In RHΔku80 , two loxP sites were inserted on either side of the ASP5 coding sequence and the gene was excised by transient expression of Cre recombinase , followed by FACS sorting and cloning . The upstream loxP was inserted along with a KillerRed expression cassette while the downstream loxP was directly fused to a GFP without a promoter as described in S2A Fig . The excised parasites ( RHΔku80Δasp5 ) were confirmed by genomic PCR and immunofluorescence analyses ( Figs 1A and S2B ) . In parallel , the CRISPR/Cas9 approach was used to generate frame-shift knockout parasites in type I ( RHΔasp5 ) and type II strains ( PRUΔku80Δasp5 and ME49Δasp5 , with insertion of a selection marker , HXGPRT and DHFR , respectively ) ( S3A–S3D Fig ) . In the parental lines RHΔku80 and PRUΔku80 , ASP5 was C-terminally epitope-tagged at the endogenous locus prior to disruption of the gene ( S3A–S3D Fig ) . The frame-shift in ASP5 induced by CRISPR/Cas9 editing was confirmed by genomic PCR and sequencing as indicated in S3B–S3D Fig . Plaque assays were performed to assess the importance of ASP5 for parasite fitness over multiple lytic cycles ( Fig 1E ) . RHΔku80Δasp5 , RHΔasp5 , and PRUΔku80Δasp5 parasites formed strikingly smaller plaques compared to the wild-type ( wt ) parasites and their respective parental non-excised lines , indicating a defect in one or more steps of the lytic cycle ( Fig 1F ) . Morphologically , all of the organelles ( inner membrane complex , mitochondrion , apicoplast , rhoptries and micronemes ) appeared normal in the absence of ASP5 ( S1E Fig ) . Both type I RHΔku80Δasp5 and RHΔasp5 parasites were functionally complemented with either a genomic or cDNA version of wt ASP5 , or the cDNA coding for the mutated ASP5D/A where the aspartic residue in the first catalytic site ( DTG ) was converted to alanine ( ASP5D/A ) . Complemented parasites expressing either ASP5 cDNA or gDNA were readily obtained in the absence of positive selection , whereas no transgenic parasites expressing ASP5D/A were obtained unless selection was applied . Although the level of ASP5 expression between endogenously tagged ASP5-3Ty ( 3 tags ) and the complemented strain ASP5g-Ty ( 1 tag ) cannot be directly compared , the full complementation of phenotype by plaque assay with a low level of ASP5g-Ty suggests that the protease is produced in excess in wild type parasites . Interestingly , complementation with ASP5 gDNA resulted in a complete reversion of the Δasp5 phenotype in plaque assay , whereas ASP5 cDNA led to only partial reversion , implying that short-ASP5 contributes to ASP5 functioning ( Fig 1F ) . In contrast , ASP5D/A failed to complement the RHΔasp5 phenotype ( S1B Fig ) . Table 1 recapitulates all the ASP5 modified parasites lines generated in this study . The phenotypic consequences of Δasp5 were investigated for each step of the lytic cycle . Unexpectedly , intracellular growth assays revealed that all strains examined replicated at a similar rate suggesting that ASP5 does not impact on parasite growth ( Fig 1D ) . To exclude that the rich culture media used here is actually masking a phenotype , we performed an intracellular growth assay in glucose depleted medium . While the control BCKDH mutant parasites were severely slowed , both RH and RHΔasp5 were not impacted under glucose starvation conditions ( S1C Fig ) . Spontaneous egress is a none-synchronous event which cannot be assessed quantitatively , whereas the calcium ionophore A23187 is a strong inducer that tends to mask modest impairments in egress . We were only able to observe a significant defect in egress in both RHΔasp5 and PruΔku80Δasp5 parasites when using BIPPO ( Fig 1G ) , a recently described potent inhibitor of phosphodiesterases that triggers egress in apicomplexan parasites [20] . Moreover , during the process of spontaneous egress , a significant fraction of Δasp5 parasites remained enclosed within a membranous structure , either PVM or host plasma membrane , which possibly delayed infection of new host cells ( Figs 1H and S1D , S1 Movie ) . To assess the role of ASP5 in trafficking of the GRAs to the PV and/or the PVM , we first examined the localization of the subset of GRAs for which specific antibodies were available ( GRA1 , 2 , 3 and 7 ) . GRA1 is expressed and secreted into the vacuolar space as a soluble protein that subsequently becomes peripherally associated with the MNN [21] . As shown in Fig 2A , the localization of GRA1 , which possesses a putative PEXEL-like motif ( RALNK ) , is not affected by the absence of ASP5 . In contrast , GRA2 and GRA3 that are associated with the MNN in wt parasites [13] , showed an altered staining pattern in Δasp5 parasites ( Fig 2A ) . Upon strong fixation conditions adapted to visualize proteins accumulated in the vacuolar space , GRA2 is not aggregated and displays instead a punctate staining for around 80% of the PVs observed . Similarly , GRA7 and also GRA3 localization at the PVM was modified in the absence of ASP5 , with no PVM staining observed in more than 70% of the vacuoles ( Fig 2A ) . Given that several GRAs involved in MNN formation appeared perturbed in RHΔasp5 parasites , the morphology of the PV was examined by electron microscopy . Whilst the MNN in wt parasites is comprised of elongated nanotubules , a dramatic change of vacuolar space architecture was observed in the absence of ASP5 . In contrast to RH parasites , the PV of RHΔasp5 parasites did not exhibit a typical MNN which is usually constituted of many long and intricate tubules . Instead , RHΔasp5 parasites contained vesicles and small tubules sparsely distrubuted throughout the vacuolar space ( Figs 2B and S4 ) . This indicates that parasites lacking ASP5 are unable to assemble an elaborated MNN . The PV lumen of RHΔasp5 parasites also appears different to that of parasites depleted in both GRA2 and GRA6 [13] . By electron microscopy , the PVM appeared intact and the host mitochondria and ER still appeared to be recruited at the periphery of the vacuole ( Fig 2C ) . The outer membrane of the host mitochondria shows a close apposition to the PV membrane of RHΔasp5 parasites , with a mean distance of 12 ± 3 nm , as similarly documented for RH parasites ( Fig 2C ) . Morphometric analyses were undertaken to quantify the extent of host mitochondria-PV membrane association in host cells 24 h p . i . : 26 and 18% of the PV membrane was associated with host mitochondria in RH and RHΔasp5 parasite-infected cells , respectively . This suggests that the mutant has the ability to recruit host mitochondria to its PV but to a lesser extent than wt parasites . The recently described GRA19 and GRA20 were investigated here via expression of C-terminally HA-tagged second copies as previously described [18] . Both proteins are known to be processed within their PEXEL-like motif by an unidentified protease [18] . Transiently expressed GRA19-HA and GRA20-HA were modestly processed as previously observed in parental parasites , however this cleavage was abolished in the absence of ASP5 ( Fig 3A ) . An R/A point mutant in the GRA19 PEXEL motif prevented processing as previously reported , and served here as a control . The absence of ASP5 did not alter the localization of either GRA19 or GRA20 in an obvious manner as documented by IFA ( Fig 3B ) . Given that both processing and localization of several GRAs is affected by the absence of ASP5 , we examined whether the overall secretion by dense granules was impaired . We developed a secretion assay whereby released GRAs were collected from the supernatant of extracellular parasites and referred to here as excretory secretory antigens ( ESA ) upon western blot analyses . Secretion of the microneme protein MIC2 was used here as a control for parasite viability and fitness . These assays revealed that secretion of GRA1 , 2 , 3 and 7 were comparable in RH and RHΔasp5 strain parasites ( Fig 3C ) . Interestingly , GRA7 which was previously reported to be phosphorylated by an unidentified host cell kinase [22 , 23] gave rise to a ladder of bands which appears to be extensively reduced or even abolished in the absence of ASP5 ( Fig 3C ) . This suggests that GRA7 may be subtly miss-targeted in the absence of ASP5 and hence no longer accessible to the host kinase . Taken together , these results indicate that ASP5 is responsible for the cleavage of some PVM-enclosed GRAs and in its absence , these proteins are normally secreted by the dense granules yet are impacted in their final destination . This is likely to lead to defects in post-translational modifications ( e . g . phosphorylation of GRA7 ) and altered protein activity given that the MNN is no longer formed . Following parasite internalization and the concomitant PVM formation , two GRAs are known to cross the PVM and be exported into the host cell nucleus [16 , 17] . To assess the fate of GRA16 and GRA24 , second copies of GRA24 driven by a tubulin promoter and GRA16 driven by its endogenous promoter and fused to 3 Myc tags were expressed in type I parasites ( Fig 4A and 4C ) . As previously reported , GRA16 and GRA24 show a dual localization in the PV as well as the host cell nucleus in RH parasites . In sharp contrast , in RHΔasp5 parasites both GRAs accumulate in the PV but fail to reach the host nucleus , even at a high MOI . Whereas ASP5 cDNA or gDNA complementation restored the host cell nuclear localization , the catalytically inactive ASP5D/A is not sufficient to promote GRA16 export ( Fig 4A ) . In wild type parasites GRA16-3Myc showed two forms that presumably correspond to unprocessed and processed forms given the fact that the protein possesses a PEXEL-like motif starting at the arginine residue in position 63 , corresponding after cleavage to a drop of ~5 kDa . Contrastingly , in RHΔasp5 parasites the unprocessed form of GRA16 ( which migrated slightly slower ) strongly accumulates while a residual level of the processed form was still detectable ( Fig 4B ) . This might result from the action of a different protease , or may represent a degradation product . The small shift in the unprocessed band observed between RH and RHΔasp5 parasites cannot be explained and will require further investigation . Scrutiny of the GRA24 sequence did not uncover the presence of such a motif , nor did the protein appear to undergo any detectable processing event by western blot analyses ( Fig 4D ) . These findings demonstrate that ASP5 contributes critically to the export of parasite effector proteins both with and without a PEXEL-like motif suggesting that ASP5 might additionally alter the function of protein ( s ) implicated in the translocation of effectors across the PVM . Type I parasites lacking GRA16 and GRA24 exhibit no decrease in virulence in mice , however the deletion of these genes in type II strain parasites has been reported to show reduced virulence [16 , 17] . In light of this , type I and type II Δasp5 parasites and their parental lines were assessed for virulence upon intraperitoneal ( i . p . ) inoculation into groups of susceptible female C57Bl/6 mice . Mice had to be sacrificed 7 days after infection with 5 . 101 RH parasites , whereas mice receiving the same inoculum of RHΔasp5 parasites survived for 13 days ( Fig 5A ) . Despite this , upon inoculation of a larger number of parasites ( 5 . 103 ) no difference was observed between the two type I parasite lines ( Fig 5A ) . This suggests that the delay observed during low dose infection could be explained by the reduced fitness observed in tissue culture . Mice infected with 106 type II parasites from both ME49 and ME49Δasp5 led to death of the animals over the acute phase of the infection ( Fig 5B ) . In contrast , 80% of the mice infected with 105 ME49Δasp5 survived the infection at day 40 whereas the control ME49 parasite line succumbed to infection within 7 to 13 days ( Fig 5B ) . The type II Δasp5 in vivo phenotype therefore correlates with previous observations made with GRA16 and GRA24 deficient parasites [16 , 17] . Seroconversion was assessed for the four surviving mice , which all display a positive serological profile ( S5 Fig ) . Cyst biogenesis in vivo remains unassessed and will be further investigated . One of the key events in establishing a protective Th1 immune response against T . gondii is the ability of host immune cells to produce the pro-inflammatory cytokine interleukin 12 ( IL-12 ) , which in turn stimulates the production of interferon gamma ( IFNγ ) by natural killer ( NK ) cells , CD4+ and CD8+ T cells [15 , 24] . IFNγ is the major pro-inflammatory cytokine driving multiple cellular defense mechanisms during both the acute and chronic phases of infection [25] . Importantly , the immunity related GTPases ( IRG proteins ) constitute a large family of interferon-inducible proteins that mediate early resistance to T . gondii infection in mice . Several studies have shown that IRGs , in particular Irga6 and Irgb6 are recruited to the nascent PVM , where they cause disruption of the vacuole and parasite death . While the ROP18 complex of type I parasites is able to phosphorylate IRG proteins , thereby preventing their oligomerization and loading onto the PVM , type II parasites are unable to block the action of IRG proteins due to the polymorphic nature of ROP5 , which forms part of the ROP18 complex [26] . Recent studies have associated GRA7 to the ROP18 complex by acting as regulator for ROP18-specific inactivation of Irga6 [23 , 27] . To determine whether vacuoles containing RHΔasp5 parasites failed to block IRG recruitment to the PVM , an IRG recruitment assay was performed for Irgb6 . Our results indicate that RHΔasp5 parasites behave like RH parasites and remain non-susceptible to Irgb6 and Irga6 loading ( Fig 6A , left panel ) . Conversely , Irgb6 was recruited to the PVM of PRUΔku80Δasp5 parasites as previously reported for PRUΔku80 ( Fig 6A , right panel ) [26] . Given these data and in spite of the impact of ASP5 on GRA7 phosphorylation ( which forms part of the ROP18 complex ) , we propose that ASP5 activity is not essential for the activity of the ROP18 complex [23] . In light of the blockage in the export of GRA effectors , we interrogated the impact of Δasp5 on the macrophage response to infection . Differentiated bone marrow derived macrophages ( BMDMs ) were infected with type I and II strain parasites lacking ASP5 , and IL-12p40 levels were measured 40 h pi ( at the peak level of IL-12p40 secretion ) by ELISA . As previously observed , PRU parasites induce significantly higher levels of IL-12p40 synthesis when compared to RH strain parasites ( Fig 6B ) . Both type I and type II strain parasites showed a reproducible and significantly lower level of IL-12p40 produced by macrophages infected with Δasp5 parasites . Expectedly , RHΔasp5/asp5g complemented parasites rescued the IL-12p40 phenotype observed with RHΔasp5 parasites . This assay has also been carried out with RAW264 . 7 cells and with peritoneal exudate cells ( PECs ) , and both assays produced results comparable to those obtained with BMDMs . Based on the current knowledge , both type II GRA15 and GRA24 promote IL-12 secretion in vitro and in vivo and it is therefore plausible that export and function of GRA15 is also blocked in the absence of ASP5 [15] . Chemokines are soluble mediators that are essential to contain parasite spreading and to control the infection . Previous studies have shown that T . gondii induces chemokine up-regulation in several cell types and specifically GRA6 [28] , GRA24 [17] and GRA25 [29] are known to shape the immune response by regulating the expression of CXCL1 , CXCL2 , CCL2 , CCL5 and CXCL10 [30] . Here , monolayers of pMEF cells were infected with type I and type II strain parasites and pelleted after 20 h . cDNA was synthesized from total RNA and the mRNA expression levels for each of these chemokines was measured by qPCR . These data reproducibly demonstrated a pronounced decrease in CXCL1 , CXCL2 , CCL2 , CCL5 and CXCL10 expression reproducibly measured from cells infected with RHΔasp5 parasites when compared with RH and RHΔasp5/asp5g parasites ( Fig 6C ) . The results obtained with PRU and PRUΔasp5 showed no difference for CXCL1 , CXCL2 and CCL2 , while a significant decrease in CCL5 and CXCL10 expression was observed . To obtain a more global picture of the modification caused by the absence of ASP5 , we performed a genome-wide expression profiling by RNA-sequencing of mouse BMDMs infected with parental or Δasp5 parasites from both type I and type II parasites ( Fig 6D ) . We focused our analysis on genes that were modulated with more than twofold change when comparing each Δasp5 mutant with their respective parental strains . A broader , analysis of the highly modulated KEGG pathways in a type I context of infection revealed that a substantial number of both pro- ( e . g . , IL-1a and TNF ) and anti-inflammatory ( e . g . , IL-10 ) cytokines , chemokines and their relative receptors ( e . g . , CXCL1 , CXCL10 ) were significantly differentially-regulated upon Δasp5 parasite infection compared to parental lines ( Fig 6D , left panel ) . Strikingly , in type II Δasp5 parasites , the parasite-induced transcriptional response is strongly reduced to a level resembling un-infected cells ( Figs 6D , right panel , S6A and S6B ) . This result highlights that ASP5 might broadly impact type II-specific virulence factors and host cell effectors . T . gondii tachyzoites can cross biological barriers [31 , 32] , however , once inside the host the precise mechanisms leading to systemic dissemination of the parasites remain unknown . T . gondii can exploit the migratory properties of dendritic cells ( DC ) to spread throughout the organism using the “Trojan horse” strategy . Specifically , upon infection by tachyzoites , DCs exhibit a hypermigratory phenotype [33 , 34] . GRA5 has been described as one of the parasite effector molecules capable of increasing the migratory properties of DCs via CCR7 expression without DC activation [35] . To determine the potential impact of ASP5 in this process we first assessed the hypermotility phenotype and observed no significant differences between Δasp5 and the corresponding parental lines either in RH or PRU parasites ( Fig 7A ) . We then conducted a transmigration assay , whereby we measured the ability of infected DCs to migrate in response to CCL19 , a CCR7 ligand [34] . In this assay both RHΔasp5 and PRUΔku80Δasp5 parasites showed a considerable reduction in transmigration of infected DCs ( Fig 7B ) . Importantly , GRAs are anticipated to play key roles in other stages of the parasite and notably during cyst wall formation , a process that is central for parasite persistence and transmission [36] . To assess the role of ASP5 in cyst wall formation , we used the fluorescent Dolichos biflorus Agglutinin ( DBA ) lectin to detect the glycosylated protein CST1 ( one of the few available markers of the T . gondii cyst wall ) [37] , following pH induced differentiation of PRUΔku80Δasp5 strain parasites in vitro . Stage conversion from tachyzoites to bradyzoites was measured by expression of SAG4 or BAG1 . This conversion took place normally in PRUΔku80Δasp5 parasites and CST1 was also produced , glycosylated and targeted to the PV However , upon closer inspection of the IFA , it became obvious that cyst wall formation was already impaired just one week after induction of differentiation , and even more strikingly impaired two weeks later ( Fig 8A and 8B ) .
Whilst intracellular , apicomplexan parasites reside within a specialised membranous niche ( PVM ) , across which the parasite transports a plethora of effector molecules necessary for subversion and remodelling of host cell functions . Studies in P . falciparum have revealed that this process relies upon a PVM-resident translocation machinery ( PTEX ) that serves to facilitate export of parasite proteins across this membrane into the erythrocyte cytosol [38 , 39] . Intimately associated with this process is Plasmepsin V , a protease known to cleave a specialised motif ( PEXEL ) in a wide repertoire of known exported proteins [40–42] . Similarly , T . gondii possesses components related to the PTEX translocon that are proposed to act as a molecular sieve at the PVM , allowing diffusion of small molecules across this membrane [43] . Here we report the characterisation of the Golgi-resident protease ASP5 , which is responsible for the cleavage of PEXEL-like motif-containing proteins in T . gondii . Whilst Plasmepsin V appears to be primarily dedicated to cleavage of proteins destined to be exported beyond the PVM , deletion of ASP5 causes considerable pleiotropic effects by effecting both exported and PV/PVM-resident proteins ( Fig 9 ) . Accordingly and in contrast to P . falciparum , numerous PEXEL-like containing proteins remain within the PV and are not further exported . Deletion of ASP5 , without affecting dense granule secretion , caused significant morphological aberrations of the PV , most notably being the defect in MNN formation . The role of this elaborated structure is still mysterious , although it is presumed to participate in parasite access to host cell nutrients . In this context , and rather unexpectedly , depletion of ASP5 does not appear to impose any restriction on intracellular parasite replication even in glucose depleted media . The molecular connection between ASP5 activity and MNN formation is not known , however such a phenotype was previously described when individual GRAs were knocked out [13] . Alternatively , the MNN could participate in the process of egress , which is unexpectedly affected in parasites lacking ASP5 . It is known that some GRAs form high molecular weight complexes within the dense granules [44] and also exist as heteromeric complexes in the PV [44] . It is therefore conceivable that deletion of ASP5 could affect formation of these complexes and thus ASP5 will not only affect the activity of its direct substrates but also their interacting partners . Through western blotting analyses we demonstrate that the vacuolar GRA19 and GRA20 , together with the exported GRA16 , are cleaved by ASP5 at the PEXEL-like motif . These proteins share a consensus motif “RRL” in their PEXEL-like sequence , which is likely part of a larger motif recognized and cleaved by ASP5 . This slightly differs from the substrate preference of Plasmepsin V [45] , and may suggest that the motif serves differing functions in T . gondii . In wt parasites , multiple GRAs such as GRA16 and GRA24 are exported to the host cell nucleus . Upon ASP5 deletion however , there is a striking block in the export of GRA16 and GRA24 and they are absent in the host nucleus but still accumulated within the PV . Intriguingly , while GRA16 harbours a PEXEL-like motif and thus accumulates as an un-cleaved product in the absence of ASP5 , GRA24 is apparently devoid of such a signal . Assuming that both GRA16 and GRA24 share the same translocation pathway to cross the PVM , the defect in GRA24 export suggests that a component of the export machinery is defective and hence possibly active only upon cleavage by ASP5 . A recent study has demonstrated that ASP5 is directly responsible for the processing of GRA16 in a PEXEL-dependent manner at the site ( RRLAE ) [46] . In accordance with our findings , this study established that the mapped processing of GRA16 is actually not necessary for export of the protein to the host nucleus . Further dissection of GRA16 processing and identification of components of the translocation machinery and the potential implication of ASP5 in this process still await further investigation . The aforementioned GRAs exported to the host cell nucleus play a significant role in reprogramming host cell gene expression and thus contribute to immunomodulation of the host . This response is significantly perturbed in parasites depleted of ASP5 as reflected by the targeted analysis of IL-12 and the chemokine levels , as well as by the global transcriptome analysis of BMDMs infected with either type I or type II parasites . Previous studies in type II strain parasites reported specific host-transcriptome alterations when selected GRAs [15–17 , 29] were knocked-out individually . While reproducing most of these findings our data significantly broaden the network of putative parasite-dependent host-regulated pathways . Importantly , the broader impact observed upon ASP5 deletion reflects the multiple GRAs affected simultaneously in this mutant . T . gondii also acts to subvert host cell functions through increasing the motility and migration of infected DCs . This parasite-induced hypermigration of infected DCs ensures rapid dissemination of the parasites from the intestinal site of entry to the rest of the body . This step is critical for the establishment of infection and persistence as it gives the parasite time to reach sanctuary organs prior to the onset of the immune response [47] . The mediators of DCs hypermigration are not fully characterized; however GRA5 has been reported to be associated with this phenomenon [35] . Parasites lacking ASP5 show a considerable loss of this capacity , which correlates with the down-regulation of CCR7 observed in the RNA seq data upon Δasp5 type I and type II parasite infection . The establishment of chronic infection is underpinned by cyst formation . Not only do cysts ensure transmission from intermediate to definitive hosts , they are also the source of reactivation in situations of immunosuppression and thus a very important stage from a pathology view-point . Our data indicate that whilst ASP5 does not impact upon the capacity of the parasite to differentiate into bradyzoites , depletion of ASP5 severely compromises the parasites ability to build the cyst wall . The cyst wall provides a protective ‘shell’ within which the parasite is sheltered from the host cell environment , however exchanges across this barrier with the host cell are still anticipated to occur . In conclusion , the role of the T . gondii PEXEL-like motif appears to be significantly broader than that reported to date for Plasmodium spp . , wherein PEXEL cleavage gives rise to a newly exposed N-terminal sequence proposed to serve as a trafficking signal for proteins destined to be exported beyond the PVM . Given the substantial differences in host cell repertoire and accordingly , the host cell subversion requirements of these two apicomplexan parasites , it is not surprising that cleavage of this motif in T . gondii is likely implicated in a variety of functions . For example , the ASP5 PEXEL-like motif cleavage could be needed to elicit conformational changes needed for enzymatic activity , or alternatively , for sequential interaction/recognition by/with a host/parasite partner . In the context of protein targeting , given that T . gondii PEXEL-like motif cleavage occurs within the Golgi prior to trafficking to the DGs , it might specifically target a population of proteins to distinct secretory organelles . Concordantly , not all exported GRAs completely co-localize with canonical GRA-markers within the parasite [16 , 17] . Whilst the data presented here have substantially contributed to the knowledge base surrounding not only ASP5 but also GRA functioning , many questions pertaining to the reasons behind this cleavage remain to be answered . Furthermore , we have not estimated the range of ASP5 substrates . Notably , the option that ASP5 mediated cleavage could expand to proteins from secretory organelles other than dense granules or following the default pathways for secretion , has not been comprehensively investigated . In spite of this , this study will serve as a solid platform upon which further investigations into the essential process of apicomplexan protein export can be completed .
E . coli XL-10 Gold chemo-competent bacteria were used for all recombinant DNA experiments . T . gondii tachyzoites parental and derivative strains were grown in confluent human foreskin fibroblasts ( HFFs ) maintained in Dulbecco’s Modified Eagle’s Medium ( DMEM , Gibco ) supplemented with 5% fetal calf serum ( FCS ) , 2mM glutamine and 25 mg/ml gentamicin . Genomic DNA ( gDNA ) from RH parasite was isolated with the Wizard SV genomic DNA purification system ( Promega ) . Total cDNA was generated by RT-PCR using the Superscript II reverse transcriptase ( Invitrogen ) . TgASP5 ToxoDB accession number: TGME49_242720 . The C-terminal ( Ct ) of ASP5 was amplified with primers 4203–4204 on gDNA and cloned in pT8-TgMIC13-3Ty-HXGPRT [48] between ApaI and NsiI sites to give Ct-ASP5-3Ty-HXGPRT . This vector was then digested ApaI/PacI and cloned in pTub8-loxP-KillerRed-loxP-YFP-HXGPRT [49] to give Ct-ASP-LoxP-YFP-HXGPRT . The 5’ region of ASP5 was amplified with primers 4615–4616 and cloned in pTub8-loxP-KillerRed-loxP-YFP . This vector was digested PacI/SacII and the bleomycin selection cassette from pTub8-ARO-GFP-Ty-Ble [50] , digested with the same sites and was inserted to give 5’ASP5-pTub8-loxP-KillerRed-Ble . Ct-ASP5-3Ty-HXGPRT was digested ApaI/NotI and ligated into p2854-DHFR-TS [51] to create Ct-ASP5-3Ty-DHFR . To create pTub8-ASP5c-Ty , cDNA was amplified with primers 1624–1592 , digested MfeI/NsiI and cloned in pTub8-Ty [52] digested EcoRI/NsiI . To create pTub8-ASP5g-Ty , cosmid PSBL804 ( D . Sibley , Toxodb . org ) was digested BglII/EcoRI , the band corresponding to the genomic DNA of ASP5 was isolated and cloned at the same sites in pTub8-ASP5c-Ty . Importantly , the last intron is not present and the two last exons are fused . ASP5 catalytic residue D431 was mutated to A to create pTub8-ASP5c-D/A-Ty . The Q5 site directed mutagenesis kit ( NEB ) instructions was followed using primers 4795–4796 with the pTub8-ASP5c-Ty as template . TgGRA24 was amplified from gDNA using primers 4814–4815 , and cloned in pTub8TgARO-Myc-Ble [50] between EcoRI and NsiI sites . This vector was digested with NotI and ligated with a PCR product using primer 4943–2642 with pTub5-CAT-Sag1 [52] as template to create the plasmid pTub8-GRA24-Myc-Ble-CAT . To construct the vector pLIC-PGRA16-GRA16-3-Myc , the promoter region and the coding sequence were amplified on gDNA using primers 5419–5421 and cloned into the pLIC-3Myc-dhfr vector using the LIC cloning method as described [53] . gRNA for CRISPR/Cas9 was generated with primers 4883–4969 on the pSAG1-CAS9gfp-U6gRNA [54] following the Q5 site directed mutagenesis kit ( NEB ) instructions . The HXGPRT and DHFR cassettes used to generate the CRISPR/Cas9 mediated KOs of ASP5 were amplified by KOD DNA polymerase ( Novagen ) using primers 5240–5241 and 5142–5143 respectively . PCR products were precipitated in sodium acetate and re-suspended in water prior to transfection . T . gondii tachyzoites were transfected by electroporation as previously described [55] . Selection of transgenic parasites were performed with either mycophenolic acid and xanthine for HXGPRT selection [56] , pyrimethamine for DHFR selection [51] or phleomycin for ble selection [21] . All stable expressing strains were cloned by limited dilution in 96-well plates and analyzed for the expression of the transgenes by IFA and for the genomic integration by PCR . The RHΔku80-DiCre ( abbreviated RHΔku80 ) strain [49] was transfected with 40 μg of the plasmid Ct-ASP-LoxP-YFP-HXGPRT linearized AvrII . The resulting strain , RHΔku80asp5-3Ty , was transfected with 5’ASP5-pTub8-loxP-KillerRed-Ble linearized XhoI to create the RHΔku80loxPasp5-3Ty strain . 40 μg of pTub5-Cre [57] was transfected in this strain to obtain RHΔku80Δasp5 ( S1 Fig ) . To generate RHΔasp5 , 30 μg of pSAG1-CAS9gfp-U6gASP5 was transfected into RH parasites ( S2 Fig ) . Transfected parasites where cloned by GFP+ FACS sorting 48 hr post-transfection . To generate PRUΔku80Δasp5 and ME49Δasp5 , 30 μg of pSAG1-CAS9gfp-U6gASP5 and respectively 15 μg of KOD-amplified HXGPRT or DHFR selection cassette flanked by 25 nt homology regions were transfected into PRUΔku80asp5-3Ty or ME49Δasp5 ( S2 Fig ) . Clones were obtained by limiting dilution after appropriate selection . 60 μg of pTub8-ASP5c-Ty , pTub8-ASP5g-Ty , were transfected in RHΔku80Δasp5 and in RHΔasp5 . Parasites were passaged for several weeks to allow the Ty positive population to gradually increase . Parasites were cloned and Ty positive clones were selected . 60 μg of pTub8-ASP5c-D/A-Ty was transfected in RHΔasp5 and HXGPRT-selected . In absence of selection , no Ty positive population was observed even after 6 weeks post-transfection . Transient transfection of GRASP-GFP [58] , pTub-GRA19-HA , pTub-GRA19-HA R124A , pTub-GRA20-HA [18] and pLIC-PGRA16-GRA16-3Myc was performed by using 40 μg of each plasmid as previously described [55] . A confluent monolayer of HFFs was infected with around 50 freshly egressed parasites for 7 to 8 days before the cells were fixed with PFA/GA . Plaques were visualized by staining with Crystal Violet ( 0 . 1% ) as previously described [59] . RH , RHΔasp5 , PRUΔku80 and PRUΔku80Δasp5 were allowed to grow on HFF for 24 hr prior to fixation with PFA/GA . IFA using α-GAP45 antibodies was performed and the number of parasites per vacuole was scored . For each condition , 200 vacuoles were counted . Data are mean value ± s . d . of three independent experiments . Medium depleted in glucose is DMEM 11966 supplemented with up to 6 mM glutamine and 25 μg/ml gentamicin [60] . Freshly egressed tachyzoites were added to a new monolayer of HFF , washed after 30 min and grown for 30 hr . The infected HFF were then incubated for 5 min at 37°C with DMEM containing either 3 μM of the Ca2+ ionophore A23187 ( from Streptomyces chartreusensis , Calbiochem ) , 50 μM of BIPPO [20] or DMSO as control . Host cells were fixed with PFA/GA , and IFA using α-GAP45 antibodies was performed . 200 vacuoles were counted per strain and per condition , and the number of lysed vacuoles was scored . Data are mean value ± s . d . of three independent experiments . Dense granule secretion assay was performed with freshly egressed parasite washed twice in intracellular buffer ( IC; 5 mM NaCl , 142 mM KCl , 2 mM EGTA , 1 mM MgCl2 , 5 . 6 mM glucose , 25 mM HEPES-KOH , pH 7 . 2 ) containing protease inhibitor cocktail . Parasite constitutive secretion was performed in similar IC buffer . After 1 hr at 37°C , a fraction was collected , which represent the total lysate , and the excreted secreted antigens ( ESA ) were collected upon sequential centrifugation . Following an initial centrifugation 5 min/4°C/1000 g , supernatants were transferred to a new tube and spun again 5 min/4°C/2000 g . The final supernatant was collected and analyzed by immunoblotting . Processing of the micronemal protein 2 ( MIC2 ) upon secretion of the micronemes was used as control . Antibodies described here were used for IFA and western blot analysis . The mAbs α-Ty tag BB2 , α-Myc tag 9E10 , α-HA ( Covance Inc ) , α-ROP2-4 T3-4A7 [61] , α-MIC2 , α-GRA1 , α-GRA2 , α-GRA3 ( J . F . Dubremetz ) , α-ACT1 [62] as well as the polyclonal Abs α-GAP45 [59] , α-CAT [63] , α-GRA7 ( kindly provided by Prof . D . J . P . Ferguson ) α-Hsp70 [64] , α-Cpn60 [65] were used . Infected-HFF monolayers on coverslips were fixed with 4% paraformaldehyde ( PFA ) /0 . 05% glutaraldehyde ( GA ) for 10 min or for 30 min for GRAs PV and PVM localization prior to quenching in 0 . 1M glycine/PBS . Cells were then permeabilized with 0 . 2% Triton X-100/PBS ( PBS/Triton ) and blocked in the same buffer supplemented with 2% BSA ( PBS/Triton/BSA ) . Cells were incubated with primary antibodies ( Abs ) diluted in PBS/Triton X-100/BSA for 1 hr followed by PBS/Triton washes ( 3 x 5 min ) . Cells were incubated with secondary Abs ( Alexa488- or Alexa594-conjugated goat anti-mouse or goat anti-rabbit IgGs ) in PBS/Triton X-100/BSA . Where appropriate , parasite and HFF nuclei staining was performed by incubating cells in DAPI ( 4’ , 6-diamidino-2-phenylindole; 50μg/ml in PBS ) prior to final washing ( 3 x 5 min ) . Coverslips were mounted in Fluoromount G ( Southern Biotech ) on glass slides and stored at 4°C in the dark . Confocal images were collected with a Zeiss microscope ( LSM700 , objective apochromat 63x /1 . 4 oil ) at the Bioimaging core facility of the Faculty of Medicine , University of Geneva . Stacks of sections were processed with ImageJ and projected using the maximum projection tool . Crude extracts of T . gondii tachyzoites were subjected to SDS-PAGE Western blot analysis carried out using polyacrylamide gels under reducing conditions . Proteins were transferred to hybond ECL nitrocellulose . Primary and secondary antibodies ( HRP conjugated , SIGMA ) are diluted in PBS , 0 . 05% Tween20 , 5% skimmed milk . Bound antibodies were visualized using the ECL system ( Amersham ) . Metabolic labelling of the tachyzoites was done with 50 mCi [35S]-labeled methionine/cysteine ( Hartmann analytic GmbH ) per ml for 4 h at 37°C followed by co-IP in RIPA buffer using α-Ty antibodies . A pulse of 7 min was followed by two chases of 15 and 60 min . Freshly egressed parasites were allowed to invade HFF monolayers for 24 hr prior to fixation . Infected host cells were washed with 0 . 1M phosphate buffer pH 7 . 2 and were fixed with 2 . 5% glutaraldehyde in 0 . 1M phosphate buffer pH 7 . 2 , post-fixed in osmium tetroxide , dehydrated in ethanol and treated with propylene oxide to embedding in Spurr’s epoxy resin . Thin sections were stained with uranyl actetate and lead citrate prior to examination using a Technai 20 electron microscope ( FEI Company ) . Two independent sample preparation and multiple thin sections for each sample were examined . HFF infected with wt or mutant parasites for 24 hr were fixed in 2 . 5% glutaraldehyde in 0 . 1 M sodium cacodylate buffer ( pH 7 . 4 ) for 1 hs at room temperature , and processed as described [66] before examination with a Philips CM120 Electron Microscope ( Eindhoven , the Netherlands ) under 80 kV . Morphometric analysis to quantify the extent of association of host mitochondria with PV was performed as described [67] . Total RNA was extracted from infected primary mouse embryonic fibroblasts ( pMEFs ) using TRIzol reagent . cDNA was synthesized using Verso Reverse transcription ( Thermo Fisher Scientific ) . Real-time PCR was performed using the Go-Taq real-time PCR system ( Promega ) and the CFX connect real-time PCR system ( Biorad ) . The values were normalized to the amount of actin in each sample . The primer sets used are listed in [28] . Chemokine production was analyzed from three independent experiments . Bone marrow derived macrophages ( BMDMs ) were obtained by flushing marrow from the hind tibias and femurs of C57Bl/6 mice . The cell suspension was passed through a nylon mesh and cultured in RPMI1640 medium supplemented with 10% FCS , 100 U/ml penicillin , 0 . 1 mg/ml streptomycin , and 15% L-cell conditioned medium at 37°C degrees in humidified 5% CO2 . Non-adherent cells were passed the next day to 10-cm bacteriological petri dishes ( 4 . 106 cells/dish ) and harvested for experiments six days later using a cell scraper . BMDMs were seeded ( 5 . 105 per well ) in 24 well plates containing coverslips and activated with 10 ng/ml of murine recombinant IFNγ for 24 h at 37°C , 5% CO2 . The next day , activated cells were infected with freshly egressed and filtered T . gondii parasites ( MOI = 1 ) for 1 h , then fixed in 4% PFA in PBS for 10 min , semi-permeabilized in 0 . 002% digitonin in PBS for 7 min at 4°C and blocked in 2% BSA/PBS for 30 min . IFA was carried out using the primary antibodies ( Abs ) mouse α-Irga6 ( kindly provided by Prof . J . C . Howard ) , goat α-Irgb6 ( Santa Cruz ) , mouse α-GRA1 and mouse α-GRA2 . GRA1 and GRA2-containing positive vacuoles were analyzed for the presence of Irgb6 at the vacuole by counting 10 fields at a magnification of 100X . Data was analyzed and images were taken using a confocal laser microscope ( FVI1200 IX-83; Olympus ) and the software FLUOVIEW ( Olympus ) and are representative of three independent experiments . BMDMs were seeded ( 105 per well ) in 96-well plates and left to adhere over-night at 37°C , 5% CO2 . Cells were then infected with freshly egressed and filtered T . gondii parasites ( 3 . 105 parasites per well ) and culture supernatants were collected 40 h later and frozen at -20°C degrees . IL-12p40 levels were measured by ELISA according to manufacturer’s instructions in three independent experiments . The assay was also carried out using PECs and RAW264 . 7 cells . BMDMs were plated to 80% confluence in RPMI with 10% FCS , 100 U/ml penicillin , 0 . 1 mg/ml streptomycin and infected with the different T . gondii strains at a MOI of 3 . 18–20 hr post-infection , cells were rinsed with cold PBS , detached with trypsin and pelleted . Total RNA from samples in triplicates was extracted using a hybrid RNA extraction protocol with TRIzol ( Life Technologies ) and QIAGEN RNeasy Mini Kit . Sample pellets were lysed in TRIzol followed by the addition of chloroform to separate the aqueous layer and the organic layer . RNA from the upper aqueous phase was precipitated with 70% ethanol and isolated using the RNeasy column according to the manufacturer's instructions . Isolated RNA was subjected to single read 100 bp on a Illumina HiSeq 2500 at the Genomics platform at the University of Geneva , iGE3 ( the Institute of Genetics and Genomics ) . The RNA samples were multiplexed across 3 sequencing lanes of the flow cell . RNA Sequencing was performed on the Illumina HiSeq 2500 at the iGE3 genomics plateform of the University of Geneva ( http://www . ige3 . unige . ch/genomics-platform . php ) . The adapter sequences from the raw reads obtained from the RNA-Seq were trimmed using FASTX-Toolkit ( http://hannonlab . cshl . edu/fastx_toolkit/ ) ( phred<20 ) . The resulting reads , after quality control , were aligned to the latest mouse reference genome ( GRCm38 ) using TopHat/Bowtie2 aligner and HTSeq-count was used to get the read counts of the genes [68–71] . Differential expression analysis was carried out using edgeR , a Bioconductor package in R ( http://www . R-project . org ) . Normalized expression values from the count data were obtained from the normalization factors calculated by the TMM ( trimmed mean of the M values ) method . The heatmaps for the genes of interest were also generated in R using heatmap . 2 in the gplots package . KEGG ( Kyoto Encyclopedia of Genes and Genomes ) pathway analysis was performed using KOBAS 2 . 0 ( KEGG Orthology Based Annotation System ) [72 , 73] . All the computations were performed at University of Geneva on the Baobab cluster . BM-DCs were obtained as previously described [74] . Briefly , DCs were generated by culturing BM cells for 7–10 days in the presence of 20 ng/ml GMCSF in RPMI supplemented with 10% heat-incativated FCS , 50 mM 2-mercaptoethanol , 100 mM sodium pyruvate and 100 μM penicillin/streptomycin at 37°C in humidified 5% CO2 . DC ( 3–5 . 104 ) were challenged with freshly egressed tachyzoites at a MOI of 3 , treated with LPS ( 100 ng/ml ) or maintained in complete medium ( non-infected ) . Cells were settled on gelatin-coated glass slides for 6–8 hr at 37°C . The cells were imaged every min for 45–60 min ( Zeiss Cell Observer . Z1 ) . Motility patterns were compiled using ImageJ ( image stabilizer software and manual tracking plugins ) . DCs were plated at a density of 1 . 106 cells/well and incubated with freshly egressed T . gondii tachyzoïtes ( MOI 3 ) for 4 hr at 37°C and 5% CO2 . DCs were then transferred into transwell filters ( 8 μm pore size; Corning ) and incubated for 16 hr at 37°C and 5% CO2 . Migrated DC were quantified in a hematocytometer . In vitro tachyzoite to bradyzoite conversion was induced by exposing parasite cultures to pH 8 . 2 as described previously [75] . Briefly , 5 . 104 tachyzoites were allowed to infect HFF grown on glass coverslips inside 24-well plates . 24 hr post infection , bradyzoite differentiation was induced by replacing normal media with RPMI 1640 buffered with 50 mM HEPES to pH 8 . 2 and supplemented with 3% fetal bovine serum . Parasites were allowed to grow at 37°C in absence of CO2 for 4 days and alkaline media was changed daily . After 4 days of conversion , infected HFF were fixed with 3 . 7% formaldehyde , permeabilized with 0 . 5% Triton X-100 in phosphate buffered saline ( PBS ) for 20 min . After 1 hr incubation with 10% foetal calf serum ( FCS ) as blocking agent , the cells were stained for 1 hr with DBA conjugated with Alexa 594 ( used at 10 μg/mL; Vector ) and with α-BAG1 mAb ( kindly provided by Prof . V . B . Carruthers ) followed by Alexa Fluor 594 goat anti-rabbit IgG antibody 200 vacuoles were counted from 20 fields for each experiment to determine the positive/negative rate of DBA and BAG1 staining . Mice were infected by intraperitoneal injection . The health of the mice was monitored daily until they presented severe symptoms of acute toxoplasmosis ( bristled hair and complete prostration with incapacity to drink or eat ) and were sacrificed on that day . All animal experiments were conducted with the authorization Number ( 1026/3604/2 , GE30/l3 ) according to the guidelines and regulations issues by the Swiss Federal Veterinary Office . No human samples were used in these experiments . Human foreskin fibroblasts ( HFF ) were obtained from ATCC . | The opportunistic pathogen Toxoplasma gondii infects a large range of nucleated cells where it replicates intracellularly within a parasitophorous vacuole ( PV ) surrounded by a membrane ( PVM ) . Parasites constitutively secrete dense-granule proteins ( GRAs ) both into and beyond the PV which participate in remodelling of the PVM , recruitment of host organelles , neutralization of the host cellular defences , and subversion of host cell functioning . In addition , the GRAs critically contribute to cyst wall formation , a process that critically ensures parasite persistence and transmission . To act as effector molecules , some of the GRAs must be translocated across the PVM . Within the related apicomplexan parasite P . falciparum , a repertoire of proteins exported beyond the PVM contain a motif cleaved by a specific protease , Plasmepsin V . Examination of the repertoire of GRAs in T . gondii revealed that some proteins exhibit such export-like motifs suggestive of protease involvement . In this study , we have functionally characterized the related aspartyl protease 5 ( TgASP5 ) in both virulent and persistent T . gondii strains , and have investigated the phenotypic consequences of its deletion in the context of overall parasite biology , its intracellular niche , the infected host cells and the murine model . Our findings revealed fundamental roles of TgASP5 at the host-parasite interface . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Fundamental Roles of the Golgi-Associated Toxoplasma Aspartyl Protease, ASP5, at the Host-Parasite Interface |
The poxvirus life cycle , although physically autonomous from the host nucleus , is nevertheless dependent upon cellular functions . A requirement for de novo fatty acid biosynthesis was implied by our previous demonstration that cerulenin , a fatty acid synthase inhibitor , impaired vaccinia virus production . Here we show that additional inhibitors of this pathway , TOFA and C75 , reduce viral yield significantly , with partial rescue provided by exogenous palmitate , the pathway's end-product . Palmitate's major role during infection is not for phospholipid synthesis or protein palmitoylation . Instead , the mitochondrial import and β-oxidation of palmitate are essential , as shown by the impact of etomoxir and trimetazidine , which target these two processes respectively . Moreover , the impact of these inhibitors is exacerbated in the absence of exogenous glucose , which is otherwise dispensable for infection . In contrast to glucose , glutamine is essential for productive viral infection , providing intermediates that sustain the TCA cycle ( anaplerosis ) . Cumulatively , these data suggest that productive infection requires the mitochondrial β-oxidation of palmitate which drives the TCA cycle and energy production . Additionally , infection causes a significant rise in the cellular oxygen consumption rate ( ATP synthesis ) that is ablated by etomoxir . The biochemical progression of the vaccinia life cycle is not impaired in the presence of TOFA , C75 , or etomoxir , although the levels of viral DNA and proteins synthesized are somewhat diminished . However , by reversibly arresting infections at the onset of morphogenesis , and then monitoring virus production after release of the block , we determined that virion assembly is highly sensitive to TOFA and C75 . Electron microscopic analysis of cells released into C75 revealed fragmented aggregates of viroplasm which failed to be enclosed by developing virion membranes . Taken together , these data indicate that vaccinia infection , and in particular virion assembly , relies on the synthesis and mitochondrial import of fatty acids , where their β-oxidation drives robust ATP production .
Global cellular metabolism is an intricate and tightly regulated network of pathways that generate building blocks and energy to sustain the life of the cell . Viral infection often leads to shifts in substrate utilization , dysregulation of metabolic pathways and changes in cellular energetics to facilitate maximal viral replication . This is exemplified by studies conducted on human cytomegalovirus ( HCMV ) , in which infected cells display augmented cellular metabolism [1] , [2] and an anaplerotic shift in which glucose is converted into precursors for nucleotide and lipid synthesis and glutamine is utilized to replenish the tricarboxylic acid ( TCA ) cycle [3] . In addition to virally induced changes in overall metabolism , specific intermediate pathways can be exploited during infection . De novo fatty acid biosynthesis has been shown to be manipulated by several diverse viruses such as HCMV , dengue virus ( DV ) and Hepatitis C virus ( HCV ) . The de novo fatty acid biosynthesis pathway generates the long-chain fatty acid palmitate and is schematically represented in Figure 1A . The first committed step in this pathway is achieved by the conversion of acetyl-CoA to malonyl-CoA by acetyl-CoA carboxylase ( ACC ) . Subsequently , successive condensation reactions of malonyl-CoA with acetyl-CoA are catalyzed by fatty acid synthase ( FASN ) , ultimately generating the 16-carbon fatty acid palmitate . Palmitate contributes to several key biological functions such as protein palmitoylation , phospholipid synthesis and energy production ( Fig . 1A ) [4] . HCMV increases the activity of ACC as a means of upregulating phospholipid synthesis , thereby providing additional membranes for viral assembly [5] . Similarly , the DV protein NS3 has been shown to actively relocalize FASN to sites of viral replication and increase rates of fatty acid synthesis [6] . HCV has been shown to upregulate expression of the two key enzymes in the de novo fatty acid biosynthesis pathway , ACC and FASN , in a sterol regulatory element binding protein-c ( SREBP-c ) dependent manner [7]–[10] , as well as to relocalize FASN to sites of viral replication [11] . Both of these RNA viruses are known to replicate their genome on a scaffold of intracellular membranes . In the case of these three viruses , increased de novo fatty acid biosynthesis is utilized for increased phospholipid synthesis . Vaccinia virus , the prototypic poxvirus , was used as the vaccine for the successful eradication of smallpox . The large size of the viral genome , which encodes approximately 200 genes , enables the virus to replicate solely in the cytoplasm of infected cells . The array of virally encoded proteins include those that mediate viral entry , three temporally regulated phases of gene expression , genome replication and maturation , virion morphogenesis and egress , and a range of interactions with the intrinsic , innate and adaptive defenses of the host . Despite the virus' physical and genetic autonomy from the host nucleus , infection still requires macromolecule precursors such as nucleotides , phospholipids , and amino acids as well as intracellular membranes and the host translational machinery . A comprehensive metabolomic analysis of vaccinia-infected cells is lacking; indeed very little is known about how vaccinia virus manipulates or interfaces with the host metabolome or bioenergetic machinery . Recently , the Smith lab reported that vaccinia induces a ‘pseudo hypoxic’ state early in infection [12] . We have also previously shown that the FASN inhibitor , cerulenin , diminishes viral yield [13] , suggesting that de novo fatty acid biosynthesis is important for viral infection . Here , we report our expanded analysis of the contribution of de novo fatty acid biosynthesis to vaccinia virus replication . Inhibition of this pathway impairs the production of infectious virus in a manner that can be partially rescued by the addition of exogenous palmitate . The mitochondrial import and β-oxidation of palmitate appear to be very important during infection , whereas exogenous glucose is dispensable , suggesting that the conversion of palmitate to acetyl-CoA , rather than the glycolytic processing of glucose , drives the TCA cycle and aerobic energy production . Additionally , glutamine is essential for vaccinia replication , serving to anaplerotically fill the TCA cycle . Examination of how the impairment of de novo fatty acid biosynthesis affected infection revealed the unimpeded progression through the biochemical stages of infection , a modest reduction in DNA and protein synthesis , and a more dramatic block to virion assembly . Cumulatively , our data are consistent with the conclusion that de novo fatty acid biosynthesis is essential for energy production , rather than phospholipid synthesis , during poxvirus infection .
As mentioned above , several viruses manipulate the de novo fatty acid biosynthesis pathway to augment the synthesis of phospholipids and increase the availability of intracellular membranes needed for efficient replication . We have previously shown that the fatty acid synthase ( FASN ) inhibitor , cerulenin , inhibits viral yield [13] , suggesting that this pathway may also be important for vaccinia virus infection . To further assess the contribution of this pathway to vaccinia infection , the impact of two additional pharmacological inhibitors was assessed: 5- ( Tetradecyloxy ) -2-furoic acid ( TOFA ) and C75 , which inhibit the cellular enzymes ACC and FASN , respectively ( Fig . 1A ) . BSC40 cells were infected with WT vaccinia virus and treated with vehicle control ( DMSO ) , TOFA or C75 for 16 h at 37°C . Treatment with TOFA or C75 significantly inhibited viral yield by 95- and 20-fold , respectively ( Fig . 1B , black bars ) . To validate the specificity of these inhibitors , we tested the ability of the end-product of the de novo fatty acid biosynthetic pathway , palmitate , to reverse their impact . Palmitic acid was coupled to fatty-acid free bovine serum albumin [14] and 50 µM exogenous palmitate was added to infected cells treated with DMSO , TOFA , and C75 as described above . Addition of palmitate had no impact on cells treated with DMSO , but rescued viral yield by 8-fold in cells treated with either TOFA or C75 ( Fig . 1B , gray bars ) . Immunoblot analysis of lysates prepared from cells infected under these conditions showed that the accumulation of the viral early protein I3 was unaffected by the inhibitors tested . However , there was a decrease in the accumulation of the viral late protein F18 and a decrease in the cleavage of the viral protein L4 ( a hallmark of virion assembly ) ; both of these inhibitors' effects were ameliorated by the addition of exogenous palmitate ( data not shown ) . Taken together , these data show that the de novo fatty acid biosynthetic pathway , and palmitate specifically , are important for maximum viral production . Palmitate has several biological functions within the cell as shown in Figure 1A: protein palmitoylation , phospholipid synthesis or energy production via ß-oxidation . To determine if palmitate is utilized for palmitoylation , BSC40 cells were infected in the presence of vehicle ( DMSO ) or various concentrations of the palmitate analogue 2-bromopalmitate , which impairs palmitoylation by inhibiting palmitoyl acyl transferases [15] . Inclusion of 25–75 µM 2-bromopalmitate had no impact on the 16 h viral yield ( Fig . 2A ) , whereas coronavirus replication is inhibited by ≤10 µM [16] . To validate the efficacy of this inhibitor in our hands , BSC40 cells were incubated with [3H]-palmitate for 4 h in the presence of DMSO or 50 µM 2-bromopalmitate , and whole cell lysates were resolved and analyzed by fluorography [17] . As expected , the levels of [3H]-palmitoylated proteins was greatly decreased in the 2-bromopalmitate-treated sample ( Fig S1A ) . We next determined whether the utilization of palmitate for phospholipid synthesis is important for viral replication by assessing the impact of the pharmacological inhibitor triacsin C . Triacsin C is a potent competitive inhibitor of long chain acyl-CoA synthetase , and thus prevents the acylation of long chain fatty acids and hence the initiation of phospholipid synthesis [18] . Treatment with 12 . 5 or 25 µM triacsin C had no impact on the 16 h viral yield ( Fig . 2B ) . To validate the efficacy of this inhibitor , BSC40 cells were incubated with 400 µM oleic acid in the presence of DMSO or 6 . 25 µM triacsin C , and stained with BODIPY493/503 to monitor the inclusion of oleic acid into lipid droplets , a process that requires acylation of oleic acid by acyl-CoA synthetase [19]–[22] . Treatment with 6 . 25 µM triacsin C abrogated the formation of lipid droplets ( Fig S1B ) . Of note , concentrations between 1 and 10 µM triacsin C have previously been shown to inhibit rotavirus replication [19] , [23] . Taken together , these data indicate that the importance of palmitate synthesis ( and fatty acid biosynthesis in general ) during vaccinia infection is independent of its contributions to protein modification and phospholipid synthesis . The data presented above suggest that the key contribution of palmitate during vaccinia infection might be its potential to undergo β-oxidation and promote energy production . As a first approach to understanding the bioenergetics pathways that contribute to productive viral infection , we tested the importance of glucose , which can be metabolized to pyruvate and ultimately acetyl-CoA , thereby driving the TCA cycle . BSC40 cells were therefore infected with WT vaccinia virus in glucose-free media or media supplemented with 4 mM glucose ( Fig . 3A , gray and black , respectively ) . Surprisingly , the absence of glucose had no impact on the 16 h viral yield , suggesting that infected cells might be relying on ß-oxidation to drive the TCA cycle and oxidative phosphorylation , rather than glycolysis , to provide energy . To directly test the hypothesis that infected cells rely on the import and utilization of palmitate within mitochondria , cells were infected in the presence of vehicle ( solid bars ) or the pharmacological inhibitor etomoxir ( striped bars ) . Etomoxir is an irreversible inhibitor of carnitine palmitoyltransferase I ( CPT1 ) and prevents the import of long-chain fatty acids into mitochondria . Treatment with etomoxir inhibited the 16 h viral yield 14-fold compared to treatment with DMSO ( Fig . 3B , left bars ) . As expected based on the mode of action of etomoxir , the addition of 50 µM exogenous palmitate was unable to rescue viral yield ( data not shown ) . Intimate feedback between the various facets of cellular metabolism is illustrated by the observation that inhibition of β-oxidation causes a concomitant increase in the glycolytic pathway [24] . In light of this interrelatedness , the impact of etomoxir treatment on viral replication in cells maintained in the absence of glucose was also examined . When glucose was omitted from the medium , treatment with etomoxir caused a much greater decrease ( 125-fold ) in viral yield than had been seen in glucose-containing media ( Fig . 3B , right bars ) . These data show that import of long-chain fatty acids into mitochondria is important for viral infection . To formally test the hypothesis that virus infection requires ß-oxidation of long-chain fatty acids within mitochondria for energy production , cells were infected in the presence of various concentrations of the pharmacological inhibitor trimetazidine ( TMZ ) . TMZ is a competitive inhibitor of 3-ketoacyl coenzyme A thiolase , a key enzyme in ß-oxidation [25] . Inclusion of 2–8 mM TMZ led to a dose-dependent , but modest , decrease in the 16 h viral yield when cells were maintained in the presence of glucose [5-fold decrease at 8 mM ( Fig . 3C , black diamonds ) ] . As was seen with etomoxir ( Fig . 3B ) , treatment with TMZ in the absence of glucose caused a dose-dependent and more severe inhibition in viral yield [36-fold decrease at 4 mM ( Fig . 3C , gray circles ) ] . Taken together , these data show that the import of palmitate into mitochondria and its subsequent β-oxidation , are important for viral infection . In addition , the presence of glucose is not important for viral replication unless β-oxidation of fatty acids is inhibited , suggesting that the TCA cycle within infected cells is normally driven by acetyl-CoA generated by the ß-oxidation of palmitate , rather than from glycolysis . As an alternative to glycolysis , the process of glutaminolysis can provide TCA cycle intermediates through the deamination of glutamine to glutamate , which is imported into mitochondria and converted into α-ketoglutarate . To evaluate the importance of glutamine for viral infection , BSC40 cells were infected with WT vaccinia virus in glutamine-free media or media supplemented with various concentrations of glutamine from 31 . 2–2000 µM ( Fig . 4A ) . In the absence of glutamine , viral yield was reduced ∼1300-fold . The addition of 250 µM or 500 µM glutamine rescued viral yield by 20-fold and 384-fold , respectively; maximum virus production was restored at ≥1 mM glutamine . To determine whether the TCA cycle was being sustained by the anaplerotic utilization of glutamine , cells were infected with WT vaccinia virus in glutamine-free medium or media supplemented with the TCA cycle intermediates oxaloacetate ( diagonally striped bars ) or α-ketoglutarate ( horizontally striped bars ) ( Fig . 4B ) . In the absence of glutamine , the addition of 15 mM oxaloacetate was unable to rescue viral yield , and 5 mM dimethyl-α-ketoglutarate only rescued viral yield 4-fold . In the presence of 200 µM glutamine ( which itself rescued viral yield 14-fold ) , the addition of 15 mM oxaloacetate or 5 mM dimethyl-α-ketoglutarate led to additional increases in virus production of 12- and 22-fold , respectively . Taken together , these data demonstrate that glutamine plays critical roles during infection , among which is the anaplerotic support of the TCA cycle , which can be sustained by an exogenous supply of oxaloacetate or α-ketoglutarate . Since it appears that the key role for palmitate during viral infection is to drive the TCA cycle and energy production , ATP synthesis was monitored during the course of infection . We utilized a Seahorse Bioscience XF-96 extracellular flux analyzer , which measures oxygen consumption rates as a direct surrogate of ATP synthesis [26] . This assay allows for repeated monitoring of ATP synthesis of a single sample over time . BSC40 cells were plated at a density of 40 , 000 cells per well ( 96-well dish ) and were either mock infected or infected with WT vaccinia virus . Oxygen consumption rates were measured every 8 min from 90 min to 12 h post-infection ( hpi ) . At 90 min post infection , mock treated cells had an oxygen consumption rate ( OCR ) of 40 pmol/min , which rose to 75 pmol/min by 12 hpi as nutrients were consumed ( Fig . 5A , gray line ) . Strikingly , the initial OCR ( 90 min post-infection ) of WT infected cells was almost twice that of mock infected cells at 75 pmol/min; this rate remained high throughout the infection , rising to 127 pmol/min at 12 hpi ( Fig . 5A , black line ) . Statistical analysis indicated that the OCR was significantly higher in infected cells than mock-infected cells at all time points , and linear regression analysis indicated that the rate of increase in the OCR over the 10 . 5 hours of measurement was also significantly higher for the infected cells than that measured for mock-infected cells . Because the extracellular acidification rate , a readout of lactate production , remains largely unchanged over time ( data not shown ) , this increase in OCR can be ascribed to mitochondrial rather than glycolytic activity . To confirm that the elevated OCR seen in infected cells was due to increased ß-oxidation of palmitate and TCA cycle-driven ATP production , cells were infected with WT virus in the presence of vehicle or etomoxir [27] and subjected to Seahorse analysis . In DMSO-treated cells , the OCR time course had a similar slope as seen above ( Fig . 5B , black line ) , while the OCR measurements in etomoxir-treated cells were depressed at all times and decreased over the course of infection ( Fig . 5B , gray line ) . Taken together , these data show that even by 90 min post infection , oxygen consumption rates , and hence ATP synthesis , have increased almost 2-fold over mock treated cells and that this increase is maintained throughout infection . The loss of oxygen consumption seen in the presence of etomoxir confirms that the mitochondrial import of long-chain fatty acids is key for sustained and elevated ATP production within infected cells . Previously , FASN has been reported to have a diffuse localization within cells as assessed by immunofluorescence microscopy [11] , and FASN has been shown to relocalize to sites of viral replication in cells infected DV or HCV [6] , [11] . We utilized a variety of commercially available antibodies to monitor the localization of FASN in uninfected and vaccinia-infected BSC40 cells ( 7 hpi ) by immunofluorescence microscopy . The most reproducible and distinct pattern was seen using a polyclonal antibody raised against a peptide from the C′-terminus of FASN . In mock-infected cells ( Fig . 6 , left column ) , a tubular pattern of staining that radiated throughout the cytoplasm was seen . This pattern was reminiscent of mitochondrial staining , and indeed the anti-FASN staining pattern largely colocalized with the Mitotracker signal [merge ( fourth row ) and zoom ( bottom row ) ] ( Pearson's coefficient of 0 . 80 ) . Validation of the specificity of the staining seen with the anti-FASN antibody was obtained by blocking cells with the peptide used as the immunogen; in the presence of this blocking peptide , no anti-FASN staining was seen ( Fig . 6 , middle column ) , although the Mitotracker ( mitochondria ) and DAPI ( nuclei ) staining remained unchanged . At 7 hpi with vaccinia virus , the pattern of anti-FASN staining was quite similar and remained coincident with the Mitotracker profile ( Pearson's coefficient of 0 . 95 ) , although both FASN and mitochondria were more concentrated in the perinuclear region of the cell ( Fig . 6 , right column ) . These data show that at least a portion of FASN , that portion most accessible to this anti-FASN antibody , localizes at or near the mitochondria , suggesting that palmitate may be generated by FASN in close proximity to mitochondria , facilitating its import by CPT1 . Cumulatively , the data shown above indicate that the production , mitochondrial import , and β-oxidation of palmitate are important for productive vaccinia infection . As a first step in determining which stages of the viral life cycle were most affected by these inhibitors , we analyzed viral DNA replication on a per-cell basis ( Fig . 7A ) . BSC40 cells were infected with WT vaccinia virus in the presence of BrdU ( which is incorporated into newly synthesized DNA ) and treated with DMSO , TOFA , or etomoxir for 7 h . Cells were then stained with antisera against BrdU [to mark sites of nascent viral DNA synthesis ( Fig . 7A , second row ) ] , I3 [the viral single-stranded DNA binding protein ( Fig . 7A , top row ) ] and DAPI [to stain all DNA ( Fig . 7A , middle row ) ] . In cells treated with vehicle ( DMSO , left column ) , large sites of viral DNA replication accumulate in the cytoplasm as shown by the colocalization of I3 and BrdU [merge ( fourth row ) and zoom ( bottom row ) ] . Similar replication foci were observed in cells treated with either TOFA ( middle column ) or etomoxir ( right column ) , indicating that genome replication was proceeding in the presence of these inhibitors . Of note , the replication foci in the drug-treated cells appeared to be somewhat smaller and more numerous . To examine the integrity of the viral DNA accumulated in the presence of the various drugs , pulsed-field gel electrophoresis was employed ( Fig . 7B ) . BSC40 cells were infected with WT virus in the presence of DMSO , TOFA , C75 or etomoxir for 10 h . The DNA was resolved and visualized with ethidium bromide . Full-length viral genomes ( ∼200 kbp; arrowhead ) were predominant under all treatments , although the levels of DNA accumulated were diminished to 50 , 70 and 83% of WT levels in the presence of TOFA , C75 and etomoxir , respectively . Similar results were seen when viral DNA replication was assayed by Southern dot blot analysis ( data not shown ) . Cumulatively , these data show that viral DNA replication is only modestly impaired when palmitate synthesis and mitochondrial import are impaired . Since DNA replication is accomplished by early viral proteins , and since the IF analysis ( Fig . 7A ) ( and immunoblot assays , not shown ) indicated that the I3 protein accumulates to high levels in the presence of these drugs , we can conclude that early protein synthesis is also largely unaffected by these inhibitors . Vaccinia virus infection is characterized by a temporally regulated cascade of gene expression , in which each phase of transcription ( early , intermediate , late ) is mediated by proteins expressed in the prior phase . Moreover , as infection progresses , host transcription is quelled , host mRNAs decrease in abundance , and translation is skewed towards viral mRNAs . We therefore assessed whether the progression and robustness of viral protein synthesis was impaired when the cycle of palmitate synthesis and mitochondrial import was disrupted . Cells were mock-infected or infected with WT vaccinia virus in the presence of vehicle ( DMSO ) , TOFA , C75 or etomoxir and pulse labeled with [35S]-methionine for 30 min at the indicated times post-infection . As illustrated by the autoradiograph shown in Fig . 8 , protein synthesis in mock-infected cells was largely unaffected by the inclusion of these inhibitors ( lanes 1–4; 4 . 5 h post-treatment or lanes 9 , 11 , 12; 7 . 5 h post-treatment ) , with the exception of TOFA-treated cells , where protein synthesis was reduced 5-fold at 7 . 5 h post-treatment ( lane 10 ) . In infected cells , we saw the expected reduction in host protein synthesis at both 4 . 5 and 7 . 5 hpi ( lanes 5–8 and 13–16 ) . Intermediate viral proteins were readily seen at 4 . 5 hpi ( circles in lanes 5–8 ) under all treatments , although synthesis levels were somewhat decreased in cells treated with TOFA and etomoxir ( to 60 and 37% of control levels , respectively ) . The synthesis of late viral proteins ( stars in lanes 13–16 ) occurred under all treatments , although a reduction in the level of expression was seen in all drug-treated cells . Overall protein synthesis was diminished to 35% of control levels in cells treated with TOFA and 65% in cells treated with either C75 or etomoxir . These data show that the inhibition of palmitate synthesis and mitochondrial import leads to a modest reduction in the rate of protein synthesis , but does not impair host shutoff or the progression through the characteristic phases of vaccinia gene expression [28] . We next analyzed the process of viral morphogenesis to determine if the inhibition of de novo fatty acid biosynthesis affected the formation of mature virions . To focus on morphogenesis itself , without the complications of the minor impact that the inhibitors exert on protein synthesis and genome accumulation , we took two experimental approaches . First , we utilized a temperature-sensitive mutant with a lesion in the F10 protein kinase ( Cts28 ) ; at the non-permissive temperature ( 39 . 7°C ) the virus has a well-characterized defect in virus assembly [29] , [30] . Viral genome replication and the full cycle of gene expression proceed normally , but the infection arrests prior to membrane biogenesis . This arrest is reversed when cells are shifted to permissive conditions ( 31 . 5°C ) [29] . BSC40 cells were infected with Cts28 for 12 h under non-permissive conditions , and then released to permissive conditions to allow the resumption of assembly . Release was performed for 8 h in the presence of vehicle ( DMSO ) , the inhibitor rifampicin ( rif ) , which allows membrane biogenesis to move forward but blocks the formation of immature and mature virions , TOFA , or C75 . The viral yield from cells shifted to permissive conditions in the presence of DMSO was 40-fold greater than the yield from cells shifted under conditions in which virus production was inhibited ( rif ) ( Fig . 9A ) . Interestingly , the inclusion of TOFA or C75 after the temperature shift was as deleterious to virus production as rifampicin . These data strongly suggest that de novo fatty acid biosynthesis , and by extrapolation ATP production , is particularly important for virion assembly . To further examine the importance of fatty acid biosynthesis to viral assembly , we performed infections in the presence of rifampicin; the reversible inhibitor rifampicin prevents the interaction of the viral proteins D13 and A17 , thereby preventing the assembly of immature virions [31]–[34] . In the presence of rif , DNA replication and the full profile of protein expression proceed normally , but morphogenesis arrests at an early stage [35] . BSC40 cells were infected with WT vaccinia virus in the presence of rif for 12 h , washed with fresh media and incubated for another 8 h in the presence of rif ( negative control , to inhibit resumption of assembly ) , DMSO ( positive control , to allow full recovery ) , TOFA , or C75 ( Fig . 9B ) . Cells released from the rif block into DMSO produced 345-fold more virus than those maintained in rif . In cells treated with TOFA or C75 , an impaired recovery was seen , with viral production being only 41- and 5-fold greater than observed in cultures maintained in rif ( black bars ) . The addition of palmitate gave a significant but incomplete ( 7-fold ) rescue of virus production in cells released from rif in the presence of C75 ( gray bar ) . Having shown that C75 ( and TOFA ) impaired virion assembly even when added after DNA replication and late gene expression had been accomplished , electron microscopic analysis was performed on cells infected in the presence of rif for 12 h and then released into rif , DMSO , or C75 as described above ( Fig . 9B ) . The phenotype seen for cells maintained in rif was as expected: flaccid membranes ( arrowheads ) accumulated around electron-dense virosomes ( V ) ( Fig . 9C , upper left ) . These membranes lack the exterior D13 scaffold and hence do not show the spiked appearance of normal virion membranes , and no progression to typical crescents , immature or mature virions was found . When cells were released into DMSO ( upper right panel ) , the full range of assembly intermediates , including crescents ( C ) , immature virions ( IV ) , and mature virions ( not shown ) was seen . When cells were released into C75 ( lower panels ) , a novel block in assembly was seen . Crescent membranes , IV membranes , and “peanut” shaped membranes ( open circle ) resembling two fused IV were seen . However , the smooth , dense viroplasm observed in rif synchronization was dispersed into splotches or aggregates ( star ) , some of which were being enclosed in IV membranes ( open diamond ) . Many of the IVs , however , appeared devoid of interior contents . No mature virions were seen . Consistent with the data shown in Fig . 9B , the addition of exogenous palmitate to C75 treated cells partially rescued the block observed in cells treated with C75 alone ( data not shown ) . Taken together , these data show that de novo fatty acid biosynthesis is important for virion assembly , even after the constituent components are present . The most dramatic impairment observed was the fragmentation of the virosomal contents into splotchy aggregates , and the inefficient incorporation of this material into nascent virion membranes .
The data presented here support a model that is depicted schematically in Figure 10 . The de novo fatty acid biosynthetic pathway , comprising the two enzymes ACC and FASN , generates palmitate within the cytoplasm . Palmitate is then imported into the mitochondria by CPT1 where it undergoes β-oxidation to generate acetyl-CoA . The generation of acetyl-CoA from palmitate , rather than glucose , drives the TCA cycle to generate sufficient amounts of ATP for maximal viral production . Inhibition of any part in this pathway causes a significant diminution in viral yield . ATP generated by this pathway contributes to viral DNA replication and protein synthesis , but is most important for the morphogenesis of infectious virions . Glucose is largely dispensable for infection , but glutamine is essential . A major role for glutamine is to enable the anaplerotic support of the TCA cycle via deamination to glutamate within the cytoplasm , and conversion to α-ketoglutarate within the mitochondria , by the successive activities of glutaminase and glutamate dehydrogenase . The robust action of the TCA cycle may allow for the shuttling of citrate into the cytoplasm to generate pools of acetyl-CoA for the synthesis of palmitate . Having previously shown that the FASN inhibitor cerulenin impaired vaccinia virus morphogenesis [13] , we began the current study with the expectation that de novo fatty acid biosynthesis would be needed to augment the intracellular membranes available to support viral assembly . However , this expectation was not supported by further experimentation . For example , triacsin C , which inhibits long chain acyl-CoA synthetase and drastically impairs phospholipid synthesis , has no impact on vaccinia virus infection ( Fig . 2B ) . Because cerulenin , TOFA and C75 all inhibited vaccinia virus infection , we considered what other roles the end-product of their activity , palmitate , might play . We were able to discard an essential role for protein palmitoylation in the progression of the life cycle by demonstrating that infection was resistant to 2-bromopalmitate ( Fig . 2A ) . There are several viral proteins that do undergo palmitoylation , including A33 , B5 and F13 [36] , [37] , but these proteins have been shown to participate in the maturation of a small subset of mature virions ( MV ) into extracellular virions ( EV ) ; our studies are focused on the production of MV , which constitute the vast majority of virions and are sufficient for infection in tissue culture . The third key role for palmitate is to undergo β-oxidation within mitochondria and generate acetyl-CoA , a key component of the TCA cycle , which in turn drives oxidative phosphorylation and ATP generation . It seems intuitive that vaccinia infection would require high levels of ATP , given the robust synthesis of viral RNA , DNA and protein and the assembly of large numbers of complex virions . We found that glucose was dispensable for infection ( Fig . 3A ) . Moreover , we also found that oxygen consumption , a surrogate measure of mitochondrial ATP synthesis , was elevated significantly in infected cells ( Fig . 5A ) . This observation was consistent with a previous report demonstrating an increase in ATP levels in vaccinia-infected HeLa cells [38] . Our data indicate that optimal infection requires the synthesis , mitochondrial import and β-oxidation of palmitate ( Figs . 1 and 3 ) . It should be noted that complete oxidation of one molecule of palmitate yields more ATP than can be generated from one molecule of glucose ( glucose→2 acetyl-CoA+36 ATP; palmitate→8 acetyl-CoA+129 ATP ) . Even considering the fact that generating palmitate requires 57 ATP ( 15 ATP to generate and activate acetyl-CoA and 42 ATP to complete 7 condensation reactions ) , palmitate yields twice the amount of ATP compared to glucose . Interestingly , this value correlates with the ∼2-fold increase in OCR we observe upon infection with vaccinia virus . Thus , at least in BSC40 cells , vaccinia virus is more dependent upon β-oxidation of fatty acids within the mitochondria than on glycolysis for maximal ATP production . This altered metabolic profile is an early event during infection , since OCR levels were already high by 1 . 5 hpi ( Fig . 5 ) . Determining how mitochondrial activity is stimulated early in infection is an area of interest for future study . Given the central role of FASN in the generation of palmitate , we considered how vaccinia might exploit or maximize its activity . HCV has been shown to upregulate expression of the enzymes involved in fatty acid synthesis [7]–[10]; however , the levels of FASN are not upregulated during vaccinia infection ( data not shown ) . It will be of interest in the future to determine whether the specific activity of ACC or FASN is upregulated upon vaccinia infection , since both enzymes are known to be modulated by phosphorylation , ubiquitination and by specific interactions with stimulatory and inhibitory binding partners [39]–[46] . DV and HCV induce a change in the localization of FASN to the sites of RNA replication [6] , [11] , but the localization of FASN does not appear to change during vaccinia infection , and we have not been successful in detecting any stable and/or obvious interactions between FASN and vaccinia-encoded proteins ( not shown ) . However , we were surprised to see that , in BSC40 cells , a significant portion of FASN co-localizes with mitochondria ( Fig . 6 ) ; this co-localization has not been reported before . It is reasonable to conclude that this proximity would facilitate the mitochondrial import of palmitate immediately after synthesis . Electron microscopic analysis of vaccinia-infected cells has revealed that mitochondria are numerous at the periphery of viral factories , which may facilitate the delivery of ATP to viral machinery . The de novo fatty acid biosynthetic pathway driven by ACC and FASN requires ample concentrations of acetyl-CoA , which can be generated via two primary pathways . The first mechanism involves the catabolism of pyruvate by pyruvate dehydrogenase . This is unlikely , however , since it has recently been reported that a ‘pseudo-hypoxic’ state is induced early after vaccinia infection [12] , which leads to an up-regulation of HIF1α-dependent genes causing an increase in the flux of glucose to lactate and concomitant decrease in accumulation of acetyl-CoA in the mitochondria [47] . The second mechanism involves the catabolism of citrate to acetyl-CoA and oxaloacetate by ATP citrate lyase ( ACLY ) . This pathway is utilized by highly proliferative cells that utilize citrate as an intermediate for lipogenic pathways . In order for the TCA cycle not to slow down as a result of this diversion of citrate for fatty acid synthesis , replenishment must occur . In HCMV-infected cells , replenishment of the TCA cycle has been shown to occur through glutaminolysis and anaplerosis [48] . Glutaminolysis refers to the use of glutamine as a source for the ultimate synthesis of α-ketoglutarate , which can subsequently be utilized in the synthesis of citrate . Indeed , glutamine is essential for vaccinia virus infection: depletion of glutamine reduces viral yield by ∼1300-fold ( Fig . 4A ) . Oxaloacetate and α-ketoglutarate fill the TCA cycle and are able to rescue viral yield significantly , albeit only in the presence of a low concentration of glutamine ( Fig . 4B ) . This suggests that citrate may indeed provide the acetyl-CoA that is needed in the cytoplasm of vaccinia-infected cells for palmitate synthesis . Further metabolomic studies aimed at determining changes in the levels of acetyl-CoA , malonyl-CoA , palmitate and TCA cycle intermediates during vaccinia infection will be highly informative . It is perhaps not surprising that productive viral infection requires a low level of glutamine that cannot be rescued by TCA cycle intermediates , because glutamine is known to play roles in purine and pyrimidine synthesis as well as protein synthesis . Our studies indicated that the progression of the viral life cycle through three temporally regulated stages of gene expression , and through the process of viral DNA synthesis and maturation , was not impaired in the presence of C75 , TOFA , or etomoxir . We observed a mild decrease in the levels of genomes that accumulated ( Fig . 7 ) , and a modest decrease in the levels of proteins being synthesized ( Fig . 8 ) . Even when C75 or TOFA were added at 12 hpi , when these biosynthetic processes are largely complete , they had a significant impact on virus production . The process of virion morphogenesis appears to be acutely sensitive to the impact of these drugs and the attendant decrease in ATP synthesis , as demonstrated by our use of CtsF10 and rifampicin to arrest cells at the onset of morphogenesis and then monitor virus production when these blocks were released ( Fig . 9A and B ) . When cells were released into C75 or TOFA , the expected burst of virus production was diminished by >10-fold . We utilized electron microscopy to visualize the assembly process after synchronization with rifampicin for 12 h and release into vehicle alone or C75 for an additional 8 h ( Fig . 9C ) . When cells were released into vehicle alone , we saw the full spectrum of assembly intermediates including crescents ( C ) , immature virions ( IV ) , immature virions with nucleoids ( IVN ) and mature virions ( MV ) . The images seen upon release into C75 were strikingly different . The flaccid membranes that accumulate in the presence of rifampicin were indeed “chased” into normal crescent membranes by their association with the D13 scaffold protein . Many of these crescents were enlarged and resembled IV , although a significant number had a “peanut” shape reminiscent of two IVs that had fused together . However , most of these IVs and crescents were “empty” and not associated with smooth virosomal material . The large virosomes that accumulate in the presence of rifampicin were dispersed into fragmented aggregates , as if the solubility of these virosomal proteins was compromised . This phenotype was quite distinct from what has been observed before , and suggests that the solubility of virosomal proteins and their inclusion in nascent virions is highly dependent upon sufficient ATP levels . In sum , our data support the conclusion that the de novo fatty acid biosynthetic pathway plays a key role in viral infection by generating sufficient palmitate for import into mitochondria . The subsequent ß-oxidation of palmitate can drive the TCA cycle and augment the production of ATP; data generated using Seahorse technology confirms that ATP production is highly elevated within infected cells in a manner that depends upon mitochondrial fatty acid import . Thus , the vaccinia infectious cycle is highly dependent upon cellular bioenergetics; moreover , infection seems to shift cellular metabolism towards a more oxidative and less glycolytic state . The de novo fatty acid biosynthesis pathway offers a novel target for developing therapeutics to treat poxvirus infections . Interestingly , this biochemical pathway is often upregulated in cancer [49] , and inhibitors that target this pathway ( such as TOFA , C75 , and cerulenin ) inhibit cancer cell proliferation in vitro . New compounds are being developed that retain efficacy but show increased solubility and reduced toxicity , such as the FASN inhibitor , G28UCM [50]–[52] . It will be interesting to determine if such compounds would have anti-poxviral efficacy in tissue culture and animal models .
[35S]-methionine and [3H]-palmitate were purchased from Perkin Elmer Life Sciences ( Boston , MA ) . [14C]-labeled protein molecular weight markers and Mitotracker Red CMXRos were purchased from Invitrogen ( Carlsbad , CA ) . Protran nitrocellulose membranes were obtained from GE Healthcare Life Sciences ( Buckinghamshire , UK ) . 5- ( Tetradecyloxy ) -2-furoic acid ( TOFA ) and triacsin C were obtained from Enzo Life Sciences . C75 , 2-bromopalmitate , trimetazidine , rifampicin , fatty-acid free bovine serum albumin , sodium palmitate , oleic acid , glucose , glutamine , oxaloacetate and dimethyl-α-ketoglutaric acid were obtained from Sigma ( Saint Louis , MO ) . Etomoxir was obtained from Tocris Bioscience ( Bristol , UK ) . Monolayer cultures of African green monkey BSC40 cells were maintained at 37°C in Dulbecco modified eagle medium ( DMEM; Invitrogen ) containing 5% fetal calf serum ( FCS ) unless otherwise specified . Confluent 35 mm dishes of BSC40 cells were infected with WT vaccinia virus ( MOI of 5 ) in the presence of various pharmacological inhibitors [TOFA ( 154 µM ) and C75 ( 39 µM ) , various concentrations of 2-bromopalmitate , various concentrations of triacsin C , etomoxir ( 360 µM ) , or various concentrations of trimetazidine] as well as exogenous palmitic acid ( 50 µM ) as indicated , incubated at 37°C for 16 h , harvested and analyzed for viral yield and protein expression . Cell viability was monitored visually following all treatments . At the doses reported herein , no significant cell death was observed as determined by cell morphology and adherence . Viral yield was determined by performing plaque assays on BSC40 cells . Viral yield was plotted as the average of six experiments with error bars representing standard error of the mean . A student two-tailed t-test was employed to determine significance . BSC40 cells were seeded at 4×104 cells per well in a 96-well V-3 PET tissue culture plate ( Seahorse Bioscience ) and incubated for 24 h at 37°C . Cells were then either mock infected or infected with WT vaccinia virus ( MOI of 5 ) . Additionally , cells infected with WT vaccinia virus were treated with etomoxir ( 360 µM ) or vehicle control for the duration of the experiment . Prior to the experiment , the cells were incubated for 30 min in High Glucose DMEM ( Gibco ) lacking sodium bicarbonate and supplemented with 10 mM HEPES ( Gibco ) and drugs as appropriate . Oxygen consumption rates were measured every 8 . 2 minutes using the Seahorse Bioscience XF-96 extracellular flux analyzer ( Seahorse Bioscience ) from 1 . 5 to 12 hpi at 37°C . Data were assayed in quadruplicate on three separate days and graphed in Microsoft Excel . Error bars represent standard error of the mean . Linear regression analysis was employed to show a statistical difference between sample groups . Confluent 4-well chamber slides of BSC40 cells were either mock infected or infected with WT vaccinia virus ( MOI of 2 ) for 7 h at 37°C and analyzed for FASN localization and replication foci as described below . Confluent monolayers of BSC40 cells were pretreated with DMSO or 50 µM 2-bromopalmitate for 1 . 5 h . Cells were washed twice with 1 ml wash medium ( DMEM; 5% dialyzed FBS; 3 . 6 mg/ml fatty acid-free BSA; 5 mM sodium pyruvate ) and then incubated with the same medium supplemented with 0 . 5 mCi/ml [3H]-palmitate for 4 h at 37°C . Cells were harvested on ice , washed with cold PBS and lysed in RIPA buffer ( 20 mM Tris , pH 7 . 5; 137 mM NaCl; 2 mM EDTA , 0 . 5% sodium deoxycholate; 0 . 1% SDS; 1% TritonX-100; 10% glycerol and protease inhibitors ) for 20 min on ice . Whole cell lysates were resolved by SDS-PAGE and visualized by fluorography . Confluent 35 mm dishes of BSC40 cells were infected with WT vaccinia virus ( MOI of 5 ) for 10 h in the presence of vehicle control , TOFA ( 154 µM ) , C75 ( 39 µM ) or etomoxir ( 360 µM ) . Cell pellets were processed as described earlier [54] . Briefly , cell pellets were embedded in 0 . 5% low-melting-point agarose and digested for 48 h at 50°C with mild agitation in 500 µl of ESP buffer ( 1% sarcosyl , 0 . 5 M EDTA [pH 9 . 0] , 0 . 5 mg/ml proteinase K ) . The plugs were equilibrated in 0 . 5× TBE buffer prior to insertion into premolded wells of a 1% SeaKem Gold agarose gel cast and resolved in the same buffer . The DNA was resolved on a CHEF Mapper XA apparatus ( Bio-Rad ) at 6 V/cm for 5 hours at 14°C , using a switching time gradient of 0 . 05 to 17 seconds , a 2 . 84 ramping factor ( non-linear ) , and a 120° angle . The DNA was visualized by ethidium bromide staining and the image was captured and quantified using AlphaView software ( ProteinSimple ) . Confluent 35 mm dishes of BSC40 cells were infected with WT virus ( MOI of 5 ) or mock infected and treated with TOFA ( 154 µM ) , C75 ( 39 µM ) , etomoxir ( 360 µM ) or vehicle control . Cells were then rinsed with methionine-free DMEM and pulsed with 100 µCi/ml [35S]-methionine for 30 min at either 4 hpi or 7 hpi . Cells were harvested , rinsed with PBS containing protease inhibitors , resuspended in 250 µl PBS containing 1× PSB and protease inhibitors and resolved on a SDS-10–17% acrylamide gel . The gel was then stained with coomassie brilliant blue and subjected to autoradiography . Effects on protein synthesis were quantified by measuring the radioactivity present in each lane using Typhoon FLA 7000 imager and Image Quant TL imaging software ( GE Healthcare Bio-Sciences , Pittsburgh , PA ) . The signals in the experimental lanes were normalized to the DMSO control sample for each time point . Confluent 60 mm dishes of BSC40 cells were infected with WT virus in the presence of rifampicin and released as described above . Eight hours post release , samples were fixed in situ with 1% glutaraldehyde in 0 . 1 M Sorensen's phosphate buffer ( pH 7 . 4 ) and processed for epoxy embedding and conventional transmission electron microscopy . Sections were examined on a JEOL JEM 2100 electron microscope and digital images were acquired using a GATAN Ultrascan 1000 camera . Original data were scanned on an Epson Perfection scanner ( Long Beach , CA ) and were adjusted with Adobe Photoshop software ( Adobe Systems , Inc . , San Jose , CA ) . Statistical analysis and graph preparation were performed using SigmaPlot software ( Systat Software , Chicago , IL ) or GraphPad Prism ( La Jolla , CA ) . Final figures were assembled and labeled with Canvas software ( Deneba Systems , Miami , FL ) . | Vaccinia virus , the prototypic poxvirus , is closely related to variola virus , the etiological agent of smallpox . A full understanding of the poxviral life cycle is imperative for the development of novel antiviral therapies , the design of new vaccines , and the effective and safe use of these viruses as oncolytic agents . Metabolomic studies have shed light on the novel mechanisms used by viruses to replicate efficiently within their hosts . de novo fatty acid biosynthesis has been shown to be of relevance for numerous viral infections as well as for the development of cancer . Here we describe an important role for de novo fatty acid biosynthesis during vaccinia infection . Ongoing synthesis of palmitate is needed to fuel the production of energy within mitochondria . The biochemical events of viral DNA replication and protein synthesis are minimally affected by inhibition of this pathway , but viral assembly is disrupted more dramatically . Further exploration of this pathway will provide additional insight into the infectious cycle and inform new therapeutic strategies for this important class of pathogen . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"cellular",
"structures",
"subcellular",
"organelles",
"protein",
"metabolism",
"enzymes",
"microbiology",
"oxygen",
"metabolism",
"protein",
"synthesis",
"dna",
"replication",
"dna",
"bioenergetics",
"lipid",
"metabolism",
"proteins",
"metabolic",
"pathways",
"biology",
... | 2014 | De novo Fatty Acid Biosynthesis Contributes Significantly to Establishment of a Bioenergetically Favorable Environment for Vaccinia Virus Infection |
Polypeptides exiting the ribosome must fold and assemble in the crowded environment of the cell . Chaperones and other protein homeostasis factors interact with newly translated polypeptides to facilitate their folding and correct localization . Despite the extensive efforts , little is known about the specificity of the chaperones and other factors that bind nascent polypeptides . To address this question we present an approach that systematically identifies cotranslational chaperone substrates through the mRNAs associated with ribosome-nascent chain-chaperone complexes . We here focused on two Saccharomyces cerevisiae chaperones: the Signal Recognition Particle ( SRP ) , which acts cotranslationally to target proteins to the ER , and the Nascent chain Associated Complex ( NAC ) , whose function has been elusive . Our results provide new insights into SRP selectivity and reveal that NAC is a general cotranslational chaperone . We found surprising differential substrate specificity for the three subunits of NAC , which appear to recognize distinct features within nascent chains . Our results also revealed a partial overlap between the sets of nascent polypeptides that interact with NAC and SRP , respectively , and showed that NAC modulates SRP specificity and fidelity in vivo . These findings give us new insight into the dynamic interplay of chaperones acting on nascent chains . The strategy we used should be generally applicable to mapping the specificity , interplay , and dynamics of the cotranslational protein homeostasis network .
Ribosomes translate the linear genetic code into polypeptide chains that must fold into a specific three-dimensional structure and often assemble with other polypeptides to be born as functional proteins . During this process , as nascent proteins emerge from the ribosome , they lack information to complete their folding and are susceptible to misfolding and aggregation . A diverse set of molecular chaperones act as midwives to stabilize and facilitate the folding of newly translated polypeptides into functional proteins . Among these , Chaperones Linked to Protein Synthesis ( CLIPS ) [1] interact physically with ribosomes and associate cotranslationally with nascent polypeptides . In addition to folding within the cytosol , many polypeptides must be directed to various membrane-bound organelles , such as the ER and mitochondria . A number of specific targeting factors recognize nascent polypeptides before they have a chance to fold in the cytosol and deliver them to specific cellular membranes . One of the best understood mechanisms involves the cotranslational recognition of characteristic hydrophobic nascent chain segments by the Signal Recognition Particle ( SRP ) , which facilitates proper delivery of the entire ribosome-nascent chain complex ( RNC ) to the ER membrane for cotranslational translocation . The multiplicity of fates and possible interactions available to a polypeptide as it emerges from the ribosome in the eukaryotic cytosol raises a number of intriguing questions . Do all nascent chains interact with chaperones ? Is there any specificity in the recognition of nascent chains by chaperones ? How do cytosolic chaperones and targeting factors such as SRP discriminate among their respective substrates , and how is the fidelity of this process achieved ? These questions are fundamental to understanding the mechanisms governing polypeptide fate as it emerges from the ribosome . Much of our understanding of nascent chain interactions with chaperones or other targeting factors comes from the study of model proteins , chosen for a convenient enzymatic or structural assay for folding or translocation . As a result , the overall logic and organization of the system that mediates the critical events in delivery and birth of a nascent polypeptide as a functional protein is still a mystery . To begin to fill this gap , we developed a systematic approach to define the principles underlying the specificity of cotranslational chaperones . In the present work , we apply it to study the specificity and interplay of two important nascent-chain interacting factors: SRP and the Nascent Chain Associated Complex ( NAC ) . Eukaryotic SRP was initially identified as a factor for targeting proteins to the ER . SRP is a ribonucleoprotein complex comprising six proteins ( in yeast Srp72 , Srp68 , Srp54 , Sec65 , Srp21 , Srp14 ) and a non-coding RNA ( scR1 ) [2] . SRP binds weakly to all ribosomes , even those that do not translate ER-destined proteins , by virtue of its contacts with multiple ribosomal sites . One of the contact sites , the ribosomal protein Rpl25 , is also a proposed binding site for NAC [3] , [4] . SRP recognizes characteristic hydrophobic sequences such as the N-terminal signal sequence ( SS ) and transmembrane domains ( TM ) in nascent polypeptides as they emerge from the ribosome . The dual recognition of ribosome and nascent chain by SRP ensures high affinity for cognate RNCs . SRP-bound RNCs are targeted to the membrane through interaction with the membrane bound SRP receptor ( SR ) , where nascent chains bearing SS or TM domains are translocated across the ER membrane by a protein complex called the Sec61 translocon . Interestingly , the Sec61 translocon itself can also interact directly with ribosomes [5] , preferentially recognizing RNCs bearing hydrophobic SS or TM regions; this might provide an SRP independent route to the ER or a proofreading mechanism for ER import . SRP-independent co- and post-translational ER targeting pathways also exist , including post-translational chaperone-assisted translocation [6] and direct ER targeting of mRNA through RNA-binding proteins ( reviewed in [7] ) . The respective contributions of the various targeting pathways to ER import in vivo and the determinants that channel an ER-bound protein through either SRP-dependent or SRP-independent pathways are not entirely understood . Very little is known about the function of the abundant and ubiquitous NAC complex . NAC is phylogenetically conserved across eukaryotes and archaea but is absent from prokaryotes [8] . Structural characterization of archaeal NAC indicates that its subunits must assemble in tightly folded dimers [9] . Most NAC complexes are heterodimers of two subunits , α and β , but homodimers have also been reported [10] . Yeast contains a single alpha subunit gene , EGD2 , and two β subunit genes , EGD1 and BTT1 . NAC contacts Rpl25 [3] and Rpl31 [11] in close proximity to the ribosomal exit site and can crosslink to very short nascent chains [12] , suggesting an early role in the birth of nascent proteins . NAC deletion causes embryonic lethality in mice , flies , and nematodes [13]–[15] but only minor growth defects in yeast [16] . Despite its abundance and conservation , the specificity and function of NAC are obscure and controversial . NAC does not associate with proteins after release from the ribosome and has no apparent chaperone activity . From in vitro experiments , NAC was initially proposed to be essential for faithful SRP-targeting of proteins to the ER [12] and preventing inappropriate association of RNCs lacking SS or TM with the translocon Sec61 [17] . This hypothesis was not supported , however , by subsequent in vitro and in vivo studies , which did not reveal aberrant translocation phenotypes in NAC-deleted strains [16] , [18] . A regulatory role for NAC in mitochondrial protein import , suggested by in vitro experiments [19] , [20] , was not corroborated by in vivo studies [16] . Given the robustness of protein homeostasis pathways , loss of NAC could be compensated by other systems . Indeed , NAC deletions exacerbate the effect of deleting the yeast Hsp70 homolog SSB , leading to higher levels of ribosomal protein aggregation [21] . A number of experimental challenges have hindered progress towards understanding the robust network of chaperones and cofactors acting cotranslationally on nascent chains . Because nascent chains comprise a small , transient , and heterogeneous cellular pool of chaperone substrates , proteomic analyses are currently impractical . The high degree of redundancy within the cellular chaperone network often masks obvious loss-of-function phenotypes . Our understanding of the specificity and mechanism of cotranslationally acting chaperones comes from in vitro translation experiments using individual model proteins , and thus the generality of such experiments is hard to ascertain . To circumvent these difficulties , we developed a sensitive , systematic method for defining the substrate specificity and interplay of cotranslationally acting chaperones and other nascent chain binding and modifying factors ( e . g . , acetylation enzymes ) in vivo . Here we used this approach to characterize the specificity of the interactions of SRP and NAC with nascent polypeptides and how the interplay between these two factors serves to modulate that specificity .
Cotranslationally acting chaperones recognize substrates as they emerge from ribosomes; the identity of the polypeptide substrate is determined by the mRNA programming its translation . We reasoned that we could leverage the specificity , sensitivity , and comprehensiveness of RNA identification to systematically identify the substrates of factors that associate cotranslationally with nascent polypeptides . Our basic experimental strategy was to isolate specific chaperone-bound ribosome-nascent chain complexes ( RNCs ) from cells and identify the polypeptide substrates through their encoding mRNAs ( Figure 1A ) . The isolation exploits Tandem-Affinity-Purification ( TAP ) tagged chaperones expressed from their endogenous chromosomal locations to ensure their expression at physiological levels . Following isolation of a specific tagged chaperone and associated RNCs complexes , we can identify the mRNAs encoding the polypeptide substrates selectively bound by that chaperone , by DNA microarray hybridization ( Figure 1A ) . In this study , we applied this approach to define the substrate specificity of two ribosome-associated factors from the yeast S . cerevisiae: SRP and NAC . The multicomponent SRP complex was isolated using SRP54-TAP . To purify NAC we used TAP-tagged variants of each of the NAC subunits: Egd1/β , Egd2/α , and Btt1/β' . A similar strategy relying on C-terminal TAP-tags of two different solvent exposed ribosomal proteins , Rpl16 and Rpl17 , was used to purify ribosomes directly . Sucrose gradient fractionation confirmed that the TAP-tagged Srp54 , all three NAC subunits , and Rpl16 and Rpl17 all associated with polysomes ( Figure S1A and Figure S6B ) . We initially examined the TAP-purified complexes by SDS-PAGE followed by silver staining ( Figure 1Bi ) . Whereas the untagged purification control ( Figure 1Bi , lane 1 ) revealed only background bands corresponding to the TEV protease preparation , all other lanes showed characteristic associated proteins ( Figure 1Bi , lanes 2–7 ) . Shared among all purifications were a set of low molecular weight proteins identified as ribosomal proteins by mass spectrometry ( Figure 1Bi; unpublished data ) . The presence of the 40S ribosomal subunits in the TAP immunopurifications ( IPs ) was confirmed by RT-PCR detection of the 18S rRNA ( Figure 1Bii ) while immunoblot analysis for ribosomal protein Rpl3 confirmed the presence of the 60S subunit ( Figure 1Biii ) . Importantly , neither 18S rRNA nor Rpl3 were detected in purifications carried out from untagged control cells ( Figure 1Bi , lane 1 ) . These results show that the TAP-tag does not disrupt ribosomal binding of either SRP or NAC and that our isolation procedure efficiently recovers their ribosome-associated complexes . We subsequently employed the TAP-tag isolation approach to systematically identify all mRNAs associated with SRP and the three subunits of NAC , as well as those engaged with translating ribosomes in actively growing cells . The TAP-tags in ribosomal proteins Rpl16 or Rpl17 were used to purify all translating ribosomes irrespective of their association with chaperones ( Figure 1Bi , lanes 2 , 3 ) and the associated mRNA was analyzed using DNA microarrays with the total mRNA from the same cells serving as a comparative standard ( Figure 1C ) . The experiments were carried out as three independent biological replicates for each ribosomal protein . As shown in the clustering analysis in Figure 1C , the results of these experiments were highly reproducible ( Rpl16 , r = 0 . 96; between Rpl16 and Rpl17 , r = 0 . 93 ) . In principle , the relative occupancy of each mRNA with Rpl16 and Rpl17 provides a measure of that mRNA's association with translating ribosomes . At a stringent 1% false discovery rate ( FDR ) [22] , we identified that 1 , 673 mRNAs are highly enriched in both Rpl16 and Rpl17 datasets . As expected , a disproportionate number of these mRNAs encode ribosomal proteins ( GO “ribosome” , n = 212 genes , p<1×10−56 ) , metabolic enzymes ( GO “carboxylic acid metabolic process” , n = 199 genes , p = 1×10−10 ) , and mitochondria ( GO “mitochondrion” , n = 450 , p = 1×10−4 ) , which correspond to the mRNAs with the highest translation rates in actively growing cells . In contrast , the least enriched mRNAs encoded proteins likely not translated at appreciable rates in mid-log phase , including meiosis and transposition . Similar conclusions were obtained when translation was assessed in the same yeast cells by isolation of actively translated mRNAs from the polysome fractions of sucrose gradients ( Figure S1 ) ; our results are also consistent with previous findings [23] , [24] . To identify the cellular substrates of SRP in vivo , we used immunoaffinity isolation of Srp54-TAP along with its cotranslational associated RNC-mRNAs complexes to isolate mRNAs encoding nascent proteins specifically recognized by SRP ( Figure 1B , lane 7; Figure 2A ) . Using DNA microarrays we identified approximately 924 mRNAs reproducibly enriched at a stringent statistical threshold in Srp54 IPs ( Figure 2A , note high reproducibility of three independent Srp54 replicates ) . Disrupting the translating 80S ribosomes with EDTA , which releases the translated mRNAs , prevented the recovery of mRNAs but not the SRP RNA scR1 in the SRP isolations ( unpublished data ) . This indicates that the association of mRNAs with SRP was mediated through translating ribosomes , supporting our premise that analysis of the mRNAs associated with RNC-SRP complexes provides information on the specificity of SRP interaction with the translating polypeptides . Hierarchical clustering based on quantitative enrichment of mRNAs in association with Rpl16/17 and Srp54 respectively indicated clear selectivity of SRP-associated complexes for a distinct subset of translated mRNAs ( Figure 2B ) , consistent with the SRP specificity for a distinct subset of nascent polypeptides ( Figure 2B , yellow highlight ) . Secretory pathway proteins ( Figure 2C , yellow bars ) were disproportionately represented among SRP-associated mRNAs , whereas the mRNAs encoding cytosolic and mitochondrial proteins ( Figure 2C , cyan bars ) were significantly under-represented among SRP-associated mRNAs . The consistency of these results with the known function of SRP suggests that this procedure can selectively identify the mRNAs encoding nascent polypeptides that are in vivo substrates of specific cotranslational chaperones . The recognition code for SRP derived from in vitro studies provided a basis for several algorithms that predict SS and TM domains from sequence information; these are used to identify putative secretory pathway proteins ( reviewed in [25] ) . The systematic identification of SRP substrates provides an unprecedented opportunity to benchmark these predictive algorithms against the experimentally determined SRP substrates from yeast . We used published algorithms ( SignalIP , RPSP , TMHMM , HMMT , and the curated Uniprot database ) to identify putative SS or TM regions encoded by mRNAs that were associated with SRP with high confidence ( 1% FDR ) ( Figure 2D and Figure S2A ) as well as in the mRNAs least enriched in our SRP IPs ( herein the “non-SRP interactors” ) ( Figure S2Bi and S2Bii ) . All these programs predicted , with good agreement , TM domains in ∼60%–75% of the SRP interactors ( Figures 2D and Figure S2Aii , hairline denotes consensus among programs ) and an SS in 15%–35% of the SRP interactors ( Figure S2Aii ) . Notably , however , the algorithms found no SS or TM domains in about a quarter of the proteins encoded by SRP-associated mRNAs ( Figure 2D , 102 targets when using SignalIP and TMHMM ) . These could represent bona fide SRP substrates that are recognized by novel , yet-to-be-determined features . Indeed , 12% of these proteins are annotated as membrane or secretory pathways ( Figure 2E; Table 1 ) . For instance , Sed4 , a known integral ER membrane protein , and Sec20 , a v-SNARE membrane glycoprotein involved in Golgi to the ER retrograde transport , are both encoded by mRNAs that we found to be enriched in association with SRP , though both lack predicted TM or SS regions . Despite the overall consistency of our results , some of the apparent interactions might be stochastic or spurious: For instance , 48% of proteins encoded by SRP-associated mRNAs that lack predicted TM or SS domains localize to the nucleus or mitochondrion ( Table 1; Figure 2E ) . Future studies on the mechanistic and physiological significance of these potential non-canonical SRP interactors may reveal novel aspects of SRP function . As expected , applying the same algorithms to the proteins encoded by mRNAs not associated with SRP yielded few proteins with predicted SS or TM regions ( Figure S2B; “non-SRP interactome” ) . Approximately 6% of these proteins had a predicted SS and ∼6% had a predicted TM domain ( Figure S2B , note slight variations among algorithms ) . Interestingly , some of these proteins are annotated as localizing to the plasma membrane ( Fus1p ) or ER ( Ost4p ) and might therefore represent weakly SRP-bound or SRP-independent secretory proteins . Others correspond to mitochondrial proteins , which are generally not recognized by SRP; although dual targeting of some polypeptides to the mitochondria and the ER has been reported [26] . Still others , such as ribosomal protein Rpl45 , contain a predicted SS yet are clearly cytoplasmic proteins . Messenger RNAs encoding proteins with predicted TM regions were generally more highly enriched by our SRP affinity isolation procedure than proteins with predicted SS ( Figure 2F ) , suggesting that the interaction of the correspondent nascent polypeptide with SRP was stronger or more sustained . Since TM regions are generally more hydrophobic than SS , this is consistent with previous biochemical experiments indicating that proteins with more hydrophobic sequences have a higher dependency on SRP for efficient ER translocation [27] . SRP-binding substrates lacking predicted SS or TM domains were typically less enriched than those containing either predicted TM or SS domains , suggesting that their SRP-binding sequences may be weaker and thus not recognized by algorithms designed to find sequences that bind strongly to SRP . While the hydrophobicity of the SS or TM regions is clearly important for SRP interaction , we only found a very weak correlation between this parameter alone and SRP enrichment ( Figure S3 and unpublished data ) . The presence of canonical SRP-binding , ER-targeting sequences in proteins that did not appear cotranslationally associated with SRP and the apparent enrichment of nascent proteins with no SS or TM regions in association with SRP suggest that our understanding of SRP specificity in vivo is still incomplete and that SRP-binding might be influenced by additional cis- and transacting factors . A number of important questions surround the mechanisms and functions of mRNA association with membranes ( reviewed in [28] ) . SRP inactivation is not lethal to yeast [29]–[31] , indicating that SRP is not the only route to membrane association . mRNA binding proteins known to localize to cellular membranes could provide additional mechanisms for targeting selected mRNAs to the ER [32] . Experimental evidence that many mRNAs encoding cytosolic proteins associate with membranes has led to a suggestion that a substantial fraction of all translation in eukaryotic cells occurs in association with membranes [33] . To examine the contribution of the SRP-mediated route to overall mRNA targeting to membranes , we empirically defined the global complement of mRNAs associated with yeast membranes . We used a previously established differential centrifugation procedure [34] , [35] to isolate membrane-associated mRNAs as well as the cytosolic , membrane-free mRNAs ( Figure 3A ) . At a stringent statistical threshold ( 1% FDR ) , we identified 1 , 168 membrane-associated mRNAs ( ∼45% of the translatome , Figure 3B ) . Hierarchical clustering of SRP-bound and membrane-associated mRNAs demonstrated extensive overlap , as expected ( Figure 3B; r = 0 . 6 ) . A large fraction of membrane-bound mRNAs encoded proteins localized to ER , Golgi , or plasma membrane ( Figure 3C i versus ii; red , pink , and orange , respectively ) , consistent with previous findings [35] . SRP-associated ( Figure 3Ci ) and membrane-associated ( Figure 3Cii ) fractions showed comparable enrichment for mRNAs encoding ER , Golgi , and Plasma membrane proteins . For instance , 60% of all mRNAs annotated as corresponding to ER proteins were enriched in the SRP-associated dataset ( log2 ratio greater than 0 ) ( Figure 3Ci , red line ) and 70% were enriched in the membrane-bound dataset ( log2 ratio greater than 0 ) ( Figure 3Cii , red line ) . In contrast more than 90% of mRNAs encoding cytosolic proteins were included in neither the SRP-associated nor membrane-associated fractions ( Figure 3C green ) . This result suggests that cellular membranes are not the major site of cytosolic protein synthesis , at least in yeast . The enrichment for mRNAs encoding mitochondrial proteins was clearly higher in the membrane-associated than in the SRP-associated fractions ( Figure 3C ii , blue line; Figure S4A and Table 2 ) . This likely reflects the presence of mitochondria in our membrane preparation and supports the idea that a fraction of mitochondrial proteins are imported cotranslationally into mitochondria ( reviewed in [36] ) . Joint analysis of the quantitative enrichment of each mRNA in association with SRP and membrane respectively gave further insight into modes of mRNA localization ( Figure 3D–F and Figure S4 ) . Comparison of both SRP and membrane-associated RNCs ( significantly enriched targets at 1% FDR ) reveals that most mRNAs that were both SRP-associated and membrane-associated ( SRP+/Mem+ ) encoded proteins annotated as belonging to the secretory pathway ( Figure 3D , E for ER; Figure S4B–D for Plasma membrane and Golgi ) . Interestingly , 24% of the SRP-associated RNCs in which the nascent polypeptide lacks either predicted SS or TM regions were also membrane-associated ( Table S1 ) ; thus , these nascent chains are likely bona fide SRP targets despite their lack of a canonical SRP binding site . Virtually no transcripts encoding cytosolic proteins ( Figure 3D , green ) and few encoding mitochondrial proteins ( Figure S3A; Figure S4C ) were SRP+/Mem+ . As expected , these mRNAs were overwhelmingly SRP−/Mem− . Notably , a number of mRNAs encoding secretory pathway proteins also fell into this class . We reasoned that these might represent proteins imported into the ER post-translationally . Indeed , known substrates of post-translational translocation pathways were SRP−/Mem− ( Figure 3E ) . These include tail-anchored proteins ( Figure 3E , TA , highlighted in black ) , which use the post-translational GET pathway [37] and pre-pro-alpha-factor ( Mfa1 , Figure 3E ) , which uses cytosolic chaperones to reach the ER membrane [38] , [39] . Further analysis of this dataset may reveal additional substrates of these pathways . Of particular interest were secretory pathway proteins whose mRNAs were membrane-associated but not SRP-associated ( SRP−/Mem+; Figure 3F and Table 2 ) , such as the chaperone EPS1 , the plasma membrane protein IST2 , and Golgi protease KEX1 . These may represent translocation substrates whose mRNAs are targeted to membranes in an SRP-independent mechanism . Interestingly , IST2 mRNA is known to localize to the bud tip by an actomyosin-driven process and is associated with cortical ER via an SRP-independent mechanism [40] . One possible mechanism for this process could be direct localization through specific membrane-associated RNA-binding proteins ( RBPs ) [32] , [41] . However , we could not detect significant overlap between the SRP−/Mem+ mRNAs and the mRNA targets of previously described membrane-associated RBPs ( unpublished data ) . Thus , novel yet-to-be-determined pathways and factors may function to localize these mRNAs to membranes . To gain insight into the cotranslational specificity of NAC , we systematically identified mRNAs cotranslationally associated with NAC complexes , using DNA microarrays to profile the mRNAs associated with each of the three NAC subunits , Egd2 , Egd1 , and Btt1 ( Figure 4A ) . Importantly , dissociation of the ribosome-mRNA-nascent chain complexes with EDTA abrogated the association of NAC with mRNAs ( unpublished data ) , suggesting that the mRNAs identified by this assay in association with individual NAC subunits reflect the cotranslational specificity of NAC for the nascent polypeptide . Each of the TAP-tagged NAC subunits was ribosome associated ( Figure 4B , lower panel; Figure 1B ) . The extent of α/β heterodimer formation for each β subunit was assessed by immunoblot analysis of α/Egd2 enriched by immunoaffinity purification of each of the two β isoforms ( Figure 4B , middle panel , lanes 2 and 3 ) . As expected , the EGD1-encoded β subunit was strongly associated with the α subunit , Egd2 , consistent with previous reports [16] , [42] . On the other hand , little of the Egd2/α subunit copurified with the BTT1 encoded β' subunit ( Figure 4B , compare lanes 2 and 3 ) . This is consistent with evidence that Btt1 elutes predominantly at a homodimer molecular weight during size exclusion chromatography of yeast cell extracts [16] . Hierarchical clustering of the mRNAs based on their patterns of enrichment in association with each NAC subunit reveals two striking properties of NAC: First , there were clear differences between NAC subunits , suggesting that the different NAC subunits recognize different subsets of mRNA-RNC complexes . Second , NAC targets include virtually every mRNA associated with Rpl16/17 , suggesting that at least one NAC isoform can interact with virtually every nascent polypeptide in the cell . This result is consistent with the estimated stoichiometry of NAC to ribosomes ( 1 . 25∶1 ) together with evidence that most of NAC in the cell are ribosome-bound [43] . Importantly , no mRNA was recovered by NAC complexes from non-ribosome-associated fractions ( unpublished data ) , suggesting that the mRNA association and specificity are mediated through translating ribosomes ( unpublished data ) . Similarly , omission of cycloheximide during cell extract preparation and analysis , which leads to polysome dissociation , dramatically reduced the number of mRNAs associated with Egd2 ( Figure S5 ) . Because association of Egd2 with mRNAs is critically dependent on the presence of intact polysomes , we conclude that Egd2 does not interact with mRNAs directly , but rather , through its association with translating nascent chains . What determines the substrate specificity of different NAC subunits ? The nascent proteins associated with different NAC subunits exhibited significant differences in a number of physicochemical properties , most notably length , hydrophobicity and intrinsic disorder , as well as inferred translation rate ( Figure 4D–G ) . Btt1 associates with mRNAs encoding proteins of higher intrinsic disorder and lower hydrophobicity , whereas Egd2 associated with mRNAs encoding proteins with low intrinsic disorder and high hydrophobicity ( Figure 4D , E ) . The length distribution of predicted protein products , which correlates inversely with the overall rate of folding ( Figure 4F ) [44] , as well as translation rate of the mRNAs ( Figure 4G ) were also significantly different among sets of mRNA respectively associated with different NAC subunits . These differences suggest that each NAC subunit participates in recognizing specific features of the nascent polypeptide; Egd2 may have higher affinity for longer , more slowly folding polypeptides exposing hydrophobic determinants , whereas Btt1 may preferentially recognize more polar , disordered chains . There were also differences in the distribution of functional roles of nascent chains associated with the different NAC ( Figure 5A ) subunits . Egd1 and Egd2 targets were enriched for mRNAs encoding metabolic enzymes , whereas the targets of Btt1 were enriched in mitochondrial and ribosomal proteins ( Figure 5A ) . RNCs translating membrane and secretory pathway proteins were also associated with NAC α/Egd2 . Preferential NAC association with nascent proteins sharing specific physicochemical properties may have resulted indirectly in the enrichment for specific functional categories . For instance , the preferential interaction with nascent ribosomal proteins with Btt1 may reflect its preference for short , highly disordered polypeptide chains with high translation rates . Alternatively , some features differentially associated with both overall physicochemical properties and functional roles of the translated proteins may underlie the observed differential specificity of NAC subunits . Our GO analysis also revealed overlaps in specificity among pairs of subunits , most notably for Egd1 and Egd2 . mRNA association patterns of these two subunits were similar to each other ( r = 0 . 74 average of three replicates for every subunit ) and more distinct from the alternative β subunit , Btt1 ( Figure 4C ) . This is consistent with , and likely reflects , the predominance of the Egd1/Egd2 heterodimer in vivo [16] . NAC subunits appear to exist as a combination of homo- and hetero-dimers in the cell [10] , [42] , and each complex may have a different set of specificities . To explore this possibility , we extracted those substrate sets shared by a α/β pair: that is , likely Egd1/Egd2 or Btt1/Egd2 substrates , and those associating solely with individual NAC subunits , that is , likely substrates of a NAC homodimer . We thus examined whether specific functional themes were significantly enriched in each category ( Figure 5B ) . Few nascent polypeptides associated with Egd1 alone , suggesting that Egd1 primarily functions in a complex with Egd2 . Egd1/Egd2 preferentially associated with RNCs translating proteins that function in carbohydrate metabolism , while the Btt1 and Btt1/Egd2 preferentially associated with RNCs translating mitochondrial and ribosomal proteins ( Figure 5B ) . Some protein classes , including redox and nucleotide metabolism , interacted with all NAC subunits , whereas Egd2 only and to a lesser extent Egd1/Egd2 also associated with RNCs translating secretory proteins; notably , this subset of nascent polypeptides also associated with SRP . We next examined how this analysis reflected on the physicochemical properties of substrates ( Figure 5C–F ) . Incorporating into our analysis the idea that NAC exists as heterodimers and homodimers exacerbated the differences in the intrinsic properties observed for each subunit set . The binding specificities of Egd2/Egd1 and Egd2/Btt1 appeared to reflect the combined specificity of the subunits in the dimer ( Figure 5C–F; green , Egd1/Egd2; orange , Egd2/Btt1; blue , Egd2/Egd2; purple , Btt1/Btt1 ) . In contrast , the nascent polypeptides associated exclusively with Btt1 ( Figure 5C , D , purple ) comprised proteins with the highest intrinsic disorder and lowest hydrophobicity , whereas the RNCs associated with Egd2 translated the most hydrophobic proteins ( Figure 5C , D blue ) . Importantly , the fact that the interaction specificity of each subunit correlated so strongly with the predicted physical properties of the translated polypeptide is strong evidence that each NAC subunit recognizes determinants in the nascent chain itself . Furthermore , both components of each NAC dimer appear to contribute to nascent chain recognition , expanding both the specificity and number of RNCs recognized by NAC . Although different NAC homo- or heterodimers differentially associated with ribosomes translating different sets of mRNAs , the specificity of the ensemble of NAC complexes appears to encompass virtually every translated polypeptide . The role of NAC in SRP specificity and substrate selection has been a matter of debate [17] , [45] , [46] . NAC and SRP both contact the ribosomal protein Rpl25 [3] . NAC was originally proposed to compete with SRP for ribosome binding [17] . However , our global analysis revealed that many nascent secretory pathway proteins can interact with both SRP and the NAC subunits Egd2 and Egd1 . We tested whether the specificity overlap might reflect joint binding at the ribosome . We isolated SRP-containing RNCs and tested for the presence of NAC ( Figure S6A ) . Indeed , immunopurification via either Srp54p , Srp68p , or Srp72p revealed the presence of Egd2p in SRP-associated polysomes ( Figure S6A ) , suggesting that NAC and SRP might bind simultaneously to the same RNCs , though this experiment does not exclude that these factors might bind to different ribosomes engaged in translation of the same mRNA . However , NAC does not detectably affect the extent of SRP association with ribosomes , as shown by the similarity of SRP cofractionation with polysomes in WT and Δegd1Δegd2 cells ( , see also below Figure 6B ) . To further explore the functional interplay between SRP and NAC , we examined whether the absence of NAC affects SRP recognition of nascent proteins , as reflected by its pattern of association with mRNA ( Figure 6 ) . To this end , we compared the ribosome-nascent polypeptide interaction specificity of SRP in wild type cells ( herein NAC+ ) with that in cells lacking NAC subunits Egd1 and Egd2 ( Δegd1Δegd2 , herein ΔNAC ) ( Figure 6A ) . Clustering analysis highlights the striking differences between the patterns of SRP-bound mRNAs in NAC+ and ΔNAC cells ( Figure 6A; Figure S6A ) . SRP association with a core set of RNCs encoding secretory proteins was relatively independent of NAC , as these were enriched by SRP affinity isolation in both NAC+ and ΔNAC cells ( Figures 6A and S7A , purple ) . In contrast , the SRP-association with another set of secretory proteins appeared to be NAC-dependent , as it was lost in ΔNAC cells ( Figures 6A and S7A , blue ) . Two SRP-dependent proteins , DPAPp and Kar2p [27] , fell into this category ( Figure S7A ) . Strikingly , a third set of mRNA-RNC complexes only associated with SRP in ΔNAC cells; most of these mRNAs encoded cytosolic proteins ( Figures 6A and S7A , green; Off-target interactors ) . The relative depletion from the SRP-associated RNCs of transcripts encoding “bona fide” SRP substrates , that is , secretory pathway proteins and the corresponding increase in mRNAs encoding cytosolic proteins from the SRP-associated RNCs , was also reflected in the GO analysis of the proteins interacting cotranslationally with SRP ( Figure S7B ) . Using a 1% FDR to analyze the SRP interactomes , we find that 70% of SRP-associated RNCs in NAC+ cells contained nascent polypeptides with predicted SS or TM regions ( Figure 6B ) whereas in ΔNAC cells , only 40% of the SRP-associated mRNAs encoded proteins with SS or TM domains . Of note , the depletion in “bona fide” SRP substrates in ΔNAC cells was independent of the statistical stringency of the analyses ( Figure S8 ) . Interestingly , deletion of the second NAC isoform ( Δegd2/Δbtt1 ) ( ΔNAC' ) had a similar effect on SRP specificity ( Figure S8 ) . As observed for ΔNAC cells , the SRP interactome in ΔNAC' cells was also depleted in proteins containing predicted SS or TM regions ( Figure S8B , Figure S8C ) and mRNAs encoding secretory proteins ( Figure S8D ) . In these cells , SRP also associated with more cytosolic and mitochondrial proteins ( Off target , Figure S8A , Figure S8D ) . Thus , NAC significantly influences the in vivo specificity of SRP interactions . How does NAC affect SRP specificity ? The extent of SRP interaction with ribosomes and the salt-sensitivity of this interaction were not affected by the absence of NAC ( Figure 6C ) . The fact that NAC appears to enhance the association of SRP with some nascent polypeptides ( i . e . , NAC-dependent ) and prevent SRP interactions with others , leaving still others unaffected , suggests a complex mode of regulation . We first hypothesized that less hydrophobic SRP-binding nascent polypeptide sequences might be more easily displaced in the absence of NAC . Our analysis did not support this hypothesis; NAC-dependent and NAC-independent SRP interacting proteins were indistinguishable based on the length and hydrophobicity of their predicted SS or TM domains ( Figures 6D , S7C and unpublished data ) . mRNA abundance and translation rate provided the strongest identifiable differences between NAC-dependent and NAC-independent SRP interactions ( Figure 6E , note log scale in 6Ei ) . NAC-independent SRP substrates were relatively highly translated , abundant proteins; NAC-dependent SRP substrates tended to be much less abundant membrane and secreted proteins ( Figure 6E ) . Because abundance and translation rate appeared key to the NAC-modulation of SRP specificity , we compared the relative enrichment of each class of SRP-associated mRNAs in the presence or absence of NAC ( Figure 6F ) . The abundant , NAC-independent SRP substrates were the most highly enriched SRP interactors even in wild type cells; their level of enrichment was only modestly affected by the absence of NAC . In contrast , the NAC-dependent SRP substrates were less highly enriched in association with SRP , even in wild type cells . Their interaction with SRP was completely undetectable in the absence of NAC . The nascent cytosolic nascent proteins whose latent ability to interact with SRP was apparently blocked by NAC were generally highly abundant cytosolic proteins with high translation rates ( Figure 6E , Off-target ) . These cytosolic proteins do not bind appreciably to SRP in wild type cells , but displayed an enrichment level comparable to bona fide SRP substrates in the absence of NAC , despite their lack of canonical SRP binding sequences . It is known that SRP can recognize hydrophobic sequences with broad specificity [47] . One of the most striking aspects of SRP function observed here is that , in vivo , in wild type cells , SRP displays exquisite specificity for its cognate substrates independent of their concentration in the cell . In the absence of NAC , however , SRP also interacts with very abundant nascent polypeptides that lack high affinity SS or TM binding sites . These may compete for SRP , effectively lowering its availability to sample less abundant cognate sequences . NAC thus effectively acts as both a positive and negative regulator of SRP interactions with potential binding targets , tuning out the noise and enhancing the specific interactions with low abundance cognate substrates ( Figure 6G ) . Deletion of NAC has few phenotypic consequences for the cell ( unpublished data ) [16] , [21] , while loss of SRP function severely compromises growth . We reasoned that if SRP were to bind inappropriately to nascent cytosolic proteins in the absence of NAC and directs the paused ribosome to the ER , the associated mRNA should inappropriately localize at the ER membrane . We thus investigated whether loss of NAC would also affect global mRNA targeting to the ER ( Figure 7A ) . Strikingly , the distribution of mRNAs between the membrane-associated and soluble fractions was indistinguishable between NAC+ and ΔNAC yeast cells ( Figure 7B; r = 0 . 96 ) . For instance , cytosolic proteins that inappropriately interacted with SRP in ΔNAC cells ( Figure 7C , “Off-target” , green ) were nevertheless largely absent from the membrane fraction in both wild type and ΔNAC cells ( Figure 7C ) . Conversely , the “NAC-dependent” nascent polypeptides whose association with SRP was impaired in ΔNAC cells still efficiently associated with membranes in ΔNAC cells , despite their diminished association with SRP ( Figure 7D , blue , compare with Figure S7B ) . We conclude that despite the loss in SRP specificity under these conditions , loss of NAC activity has little or no effect on the fidelity of mRNA targeting to membranes , despite the loss in SRP specificity . We chose two NAC-dependent SRP substrates , Kar2 and DPAP , whose association with SRP but not with membranes was impaired in ΔNAC cells , for further biochemical analysis ( Figure 7E ) . The efficiency of their ER translocation in wild type or ΔNAC cells was evaluated by determining the ratio of processed versus unprocessed protein following a short pulse with 35S-methionine . Defective translocation results in accumulation of precursors of these two proteins , that is , uncleaved Kar2 ( pre-Kar2 ) and non-glycosylated DPAP . The presence or absence of NAC did not affect the speed or efficiency of translocation for either Kar2 or DPAP ( Figure 7E ) . In contrast , impairing SRP function using the temperature-sensitive mutation sec65-2 did reduce translocation of both proteins . Thus , the apparent decrease in SRP binding of these SRP substrates in the absence of NAC did not appreciably impair their translocation to the ER . These experiments highlight the robustness and fidelity of membrane targeting pathways . To better understand how the cell compensates for the loss of NAC function , we examined the transcriptional response to joint deletion of either Egd1 and Egd2 or Egd2 and Btt1 ( Figure 7F ) . A different complement of genes was induced in response to these two perturbations , but major features of the responses were shared . Transcripts encoding ribosomal proteins , ribosome biogenesis and mitochondrial biogenesis machines , and chaperones and stress response genes were induced in response to both defects ( Figure 7F and Tables 3 and 4 ) . Notably , loss of NAC activity did not lead to transcriptional induction of an unfolded protein response ( UPR ) , consistent with the lack of a translocation defect in these cells ( Figure 7F and unpublished data ) . The chaperones induced in response to NAC deletion included stress-inducible chaperones like SSA3 and small HSPs , as well as CLIPS , most notably SSB1/2 . The synthetic genetic interaction between SSB1/2 and the NAC complex suggests that induction of SSB1/2 may contribute to functional compensation for the loss of NAC [21] . The induction of ribosomal proteins and ribosome biogenesis genes is in accord with the observation that NAC has a role in ribosome biogenesis ( [21]; Figures 4 and 5 ) . Loss of Egd2/Btt1 led to induction of numerous mitochondrial biogenesis factors , including AFG3 , SED1 , and MIA4 , suggesting a role for this NAC complex in mitochondrial biogenesis .
Affinity isolation of cotranslationally acting chaperones from cells under conditions that preserve their interaction with the nascent polypeptide and associated ribosomes and quantitative profiling of the associated mRNAs open a window on the specificity and interplay of chaperones and targeting factors responsible for cotranslational protein homeostasis . This approach should enable us to probe the structure of the CLIPS network and the interplay between different chaperones and targeting systems . Unlike previous studies defining chaperone interactors by proteomic analysis , our approach focuses on cotranslational interactors as potential chaperone substrates . The approach presented here opens a window to understand the pathways and principles of cotranslational chaperone action . The full complement of nascent chains that interact with SRP in vivo has never been defined . Studies of SRP recognition using model substrates and peptides have shown that SRP recognizes highly hydrophobic signal sequences and transmembrane regions [27] but can also recognize hydrophobic stretches found in cytoplasmic proteins [39] , [48] . We find that in vivo SRP displays considerable specificity for previously defined recognition sequences; approximately 80% of the in vivo substrates we identified contained a predicted SS or TM domain ( Figure 2D ) . Our analysis also indicates that additional factors contribute to SRP specificity and affinity in vivo: approximately 20% of SRP interactors lack a discernible SS or TM domain . Since several SRP-associated mRNAs encode secreted or membrane-associated proteins that appear to lack canonical SS or TM domains , these interactions may be functionally relevant . Conversely , a number of proteins with clear SS or TM domains were translated in these cells but were not enriched in association with SRP , suggesting that SRP recognition might be regulated by features or mechanisms beyond its intrinsic affinity for SS or TM regions . Interestingly , it has been reported that , in bacteria , basic residues promote binding of SRP to a subset of signal peptides whose hydrophobicity falls slightly below a critical level [49] . Recent studies also suggest that a hydrophobic stretch can recruit SRP to the ribosome before it emerges from the ribosomal exit tunnel , presumably by changing the conformation of the ribosome [50] , [51] . SRP-substrates without canonical SS or TMs may contain sequence elements that similarly enhance binding to SRP by this indirect mechanism or even by recruiting bridging factors . The SRP-interactome also yielded some surprising observations . Although the glucose metabolism enzyme glyceraldehyde-3-phosphate dehydrogenase ( GAPDH , Tdh1-3 in yeast ) is reportedly mostly cytoplasmic [52] , the mRNAs encoding all Tdh isoforms ( i . e . , Tdh1-3 ) were both SRP-associated and enriched in the membrane fraction ( see Table 2 ) . Notably , all three isoforms of this enzyme in yeast have predicted signal sequences at the N-terminus and have been detected on the outer surface of the cell wall [53] . Our findings suggest a mechanism by which Tdh can reach the outer cell wall and may also explain the observation that other primarily cytosolic proteins , including glycolytic enzymes , are secreted by yeast spheroplasts and found as integral components of the cell wall [54] , . Interestingly , GAPDH mRNA has also been found to associate with membranes in mammalian cells [56] . The variable efficiency of the SRP-export pathways for different mRNAs has been recognized both as a regulatory mechanism and as a source of misfolded proteins [57] . Even a small fraction of untranslocated precursors may represent a substantial burden for the cytosolic protein quality control machinery . Our analyses show quantitative variation in binding of SRP to its targets , which may be related to previous observations that some secretory and membrane proteins are more efficiently translocated than others [58] , [59] . Differential translocation efficiency is proposed to underlie disease pathologies , such as prion formation , and to regulate protein flux into the secretory pathway [57] , [60] . We find that mRNAs encoding proteins with predicted TM regions are more enriched in association with SRP than either polypeptides with SS domains or those with no detectable SS or TM domains ( Figure 2F ) . Notably , in bacteria , SRP is a main targeting factor for membrane proteins , while secretory proteins follow a different , SecB-dependent , pathway ( reviewed in [61] ) . Although differences among signal sequences have been shown to modulate translocation in yeast for a small number of substrates ( reviewed in [57] , [62] ) , we could not identify a clear correlation between SRP enrichment and any defined physicochemical property in the SS or TM domains themselves , such as length , overall or maximal hydrophobicity , or amino acid composition . In bacteria , the codon adaptation index of the SS and the efficiency in translation initiation have been proposed to influence targeting efficiency ( reviewed in [63] ) ; we found no significant correlation between SRP enrichment and these parameters ( unpublished data ) . Our data indicate that , in vivo , the features that control SRP recognition of a given nascent polypeptide are more complex than expected . In principle , additional features promoting a pre-recruitment of SRP to translating ribosomes could increase the efficiency of SRP-dependent translocation and enhance physiological robustness . For example , the translation properties of a given mRNA might influence the efficiency with which a potential SS or TM domain or other hydrophobic stretches are recognized . Additional ribosome-associated factors could also modulate the SRP association with nascent chains , as shown here for NAC . What determines the variable efficiency we observe in SRP association with SS and TM containing nascent polypeptides remains an important unanswered question . The comparative analysis of SRP- and membrane-bound mRNAs provides an overview of overall partitioning of co- and posttranslational events in membrane targeting . Our data indicate that most cytosolic mRNAs are not membrane-associated , suggesting the existence of mechanisms that separate cytosol-bound from membrane-bound mRNAs . SRP appears to be involved in cotranslational targeting of most membrane and secretory proteins to the ER; ∼80% of membrane-associated mRNAs encoding these proteins were also SRP-associated , at a 1% FDR threshold . However , we also found evidence for SRP-independent translocation pathways . A significant minority of mRNAs , roughly 20% , appears to associate with membranes through SRP-independent pathways . We estimate that 25% of the secretory pathway proteins that do not associate cotranslationally with membranes are translocated posttranslationally to the ER; these include tail-anchored proteins , which use the GET pathway [37] , [64] , and Mfa1 , whose translocation is assisted by cytosolic chaperones including Hsp70 and TRiC/CCT [38] , [39] . SRP is not essential in yeast [65] and downregulation of SRP in mammalian cells has a mild effect on growth [33] , [66] , indicating the existence of SRP-independent mechanisms for translocation [27] . The membrane-associated mRNAs that encode membrane or secreted proteins but do not bind SRP ( SRP−/Mem+ ) are candidates for substrates of SRP-independent cotranslational translocation pathways . Little is known about these pathways . They may involve direct recruitment of RNCs to the Sec61 translocon [67] , as well as additional factors [31] , [68] . Direct , translation-independent targeting of mRNAs to membranes could also involve RNA-binding proteins ( RBPs ) such as Puf1 , Puf2 , Pub1 , Scp160 , Ypl184c , Khd1 [41] , and Whi3 [69] , all of which bind distinct sets of mRNAs encoding membrane or secreted proteins . Interestingly , while few of these RBPs bind mRNAs in the SRP−/Mem+ set , there is also considerable overlap between their targets and the mRNAs we found enriched in association with SRP ( unpublished data ) , suggesting these RBPs may provide redundancy or another level of control to cotranslational SRP targeting to membranes . Our experiments will provide an opportunity to refine our understanding of the signals and features that direct secretory proteins along these alternative non-SRP pathways . The functions and localization patterns of mRNAs that were membrane-associated but not SRP-bound suggest several additional roles for SRP-independent membrane sorting of mRNAs . Many of these ( 150 out of 541 ) encoded mitochondrial protein precursors , which may be imported cotranslationally into mitochondria [70] . Among the remaining non-mitochondrial mRNAs , there was a paucity of mRNAs encoding ER-localized proteins ( Lro1 ) and proteins with transmembrane domains , but the set included many mRNAs encoding proteins involved in other membrane systems in the cell . Two other She2 targets , Mtl1 and Lsb1 , were included in this set and may also be associated with the cortical ER during trafficking to the bud [71] . Most strikingly , there were a number of mRNAs encoding proteins involved in endocytosis and actin patch assembly ( Vps35 , Aly2 , Swh1 , Lsb6 , and Pan1 ) , clathrin-mediated vesicle transport ( Apl4 , Apl3 , Laa1 , and Sec16 ) , bud formation ( Sbe2 , Ypk1 , Lrg1 , Prm10 , and Bem3 ) , and vacuole function and assembly ( Sch9 , Vps13 , Vac8 , Tre2 , Tre1 , Fab1 , and YIR014W ) . Potentially , these mRNAs are localized to specific membrane compartments to preferentially translate the proteins near their site of action . Alternatively , the nascent polypeptides could associate cotranslationally with membrane-associated interacting partners . The set of membrane-associated mRNAs included an abundance of regulatory factors , including transcription factors ( Stp2 , Ino2 , and Ppr1 ) , RBPs ( Puf2 and Puf3 ) , and signaling molecules ( Tor1 , Bem3 , Fab1 , and Sch9 ) . Localization of these mRNAs to appropriate membrane structures may facilitate co-translational association of their products with localized signaling partners or enable locally controlled regulation of their translation by signaling systems linked to these membranes . For instance , Stp2 promotes expression of permease genes and is synthesized as an inactive precursor that associates with the plasma membrane and is cleaved upon sensing of external amino acids [72]; Tor1 , a component of the TOR complex , is a peripheral membrane protein that regulates cell growth in response to nutrient availability and stress [73] , and Bem3 is a Rho GTPase activating protein specific to Cdc42 , which controls establishment and maintenance of cell polarity , including bud-site assembly [74] . The abundant , ubiquitous , and evolutionarily conserved Nascent Chain Associated Complex ( NAC ) binds ribosomes in close proximity to the nascent chain exit site [75] . Despite its conservation , little is known about its function . Our analysis of the association of the three NAC subunits with nascent polypeptides revealed a surprising and unanticipated division of labor . Considered as a group , the three NAC subunits have translation-dependent interactions with almost every mRNA . Each subunit , however , exhibits distinct specificity for RNCs engaged in translation of mRNAs with different functional themes . Based on the crystal structure of the archaeal NAC domain , NAC complexes are obligate dimers , where two subunits must complete the folded beta-sheet NAC-domain . Our analysis supports the idea that NAC subunits can function as either homodimers or heterodimers [10] , [16] . We found a large overlap between the sets of transcripts associated with Egd1 and Egd2 , consistent with the idea that the Egd1/Egd2 complex is the most abundant form . This dimer associated preferentially with nascent metabolic enzymes , including those in carbohydrate metabolism , such as glycolysis . Egd2 , either as a homodimer or in a complex with Egd1 , was also cotranslationally associated with a large fraction of mRNAs encoding membrane or secreted proteins , many of which also associate with SRP . Btt1 , either as a homodimer or in a seemingly minor Btt1/Egd2 complex , associated primarily with RNCs translating ribosomal proteins and nuclear-encoded mitochondrial proteins . In yeast , the three NAC subunits can be deleted with minimal impact on growth . Deletion of all three NAC subunits leads to enhanced ribosomal protein aggregation in cells also lacking the Hsp70 homologs Ssb1 and Ssb2 [21] . This would suggest that the putative function of NAC is masked by the redundancy of the CLIPS protein homeostasis network . Our analysis of the transcriptional response to NAC deletion ( Figure 7F–G ) provides insight into how the cellular circuitry compensates for the loss of NAC: A set of chaperones including both stress-inducible chaperones ( e . g . , Ssa2/4 , Hsp42 , and Hsp104 ) and CLIPS ( e . g . , Ssb1 and Ssb2 ) , as well as several ribosomal proteins and assembly factors ( Tables 3 and 4 ) , were induced . This multifaceted response suggests that loss of NAC impairs protein folding and ribosome assembly . NAC has been proposed to have a role in mitochondrial targeting , as shown by a synthetic growth defect by deletion of cells lacking both Egd2 and the mitochondrial targeting factor Mft1 [19] . Our analysis revealed that mRNAs encoding mitochondrial proteins are enriched in association with both Btt1 and Egd2 ( Figure 5A , B ) . Moreover we found that several proteins involved in mitochondrial assembly were induced in cells lacking NAC . Thus , a possible auxiliary role for NAC in cotranslational targeting polypeptides to the mitochondria deserves further investigation . Is NAC a chaperone ? Purified NAC does not prevent protein aggregation and NAC cannot bind directly to nascent chains unless they are ribosome associated [12] , [39] . While this is unexpected for a traditional chaperone , NAC may be akin to trigger factor in bacteria , which interacts primarily with nascent chains in the context of the ribosome [76] . The distinct physicochemical properties of the nascent polypeptides associated with different NAC subunits may reflect the direct binding specificity of each individual subunit . A more detailed understanding of NAC substrate specificity must await better structural and biochemical understanding of this complex . The results of our global analysis will open the way for these experiments . The interplay between SRP and NAC has been controversial [12] , [16] , [17] , [45] , [77] , [78] . In vitro experiments suggested that SRP can bind to cytosolic non-cognate nascent chains and that NAC and SRP can compete for RNC binding . On the other hand , in vivo analyses did not support the idea that NAC is required for proper SRP function and translocation [16] . Our experiments reconcile these observations and provide an integrated view of the regulation of SRP specificity by NAC . NAC modulates the interaction of SRP with nascent chains in vivo , favoring SRP binding to cognate substrates and disfavoring interactions with non-cognate targets ( Figure 6F ) . Some ER-bound nascent proteins appear to depend on the presence of NAC in the cell for their interaction with SRP ( NAC-dependent ) while others do not ( NAC-independent ) . Surprisingly , mRNA abundance and translation rate , rather than direct determinants of SRP affinity such as SS or TM hydrophobicity , are the major distinguishing features of the NAC-dependent versus the NAC-independent SRP interactions . This raises the idea that , in vivo , the specificity of the factors that interact with nascent proteins is governed not only by properties of the nascent polypeptide sequence , such as the intrinsic affinity of a given nascent chain for SRP , but also by the competition among cognate and non-cognate nascent polypeptides for these factors and by interactions between factors , exemplified by NAC and SRP . Our analysis provides insight into the question of how signals such as SS or TM , which are recognized in a variable manner depending on affinity and concentration , can be read in the cell to determine a binary fate such as translocation , that is . proteins do or do not get translocated . Our data show that in wild type cells , SRP does bind with exquisite specificity to cognate substrates spanning a very wide range of cellular mRNA abundances , while disregarding very abundant cytosolic substrates that contain hydrophobic stretches with potential SRP-binding affinity . In ΔNAC cells , however , this specificity is relaxed , so that highly abundant non-cognate substrates bind to SRP and low abundance cognate substrates are lost from SRP . Thus NAC provides an additional level of specificity that fine-tunes SRP interactions to “sharpen” the response . Our data can be explained in light of previous biochemical and biophysical measurements . NAC and SRP both contact the same ribosomal protein , L25 , but have additional non-overlapping binding sites on the ribosome [11] . We observed that SRP-associated polysomes also contain associated NAC ( Figure S6A ) . The interplay between these factors appears to be relevant for SRP specificity . SRP samples most translating ribosomes to encounter RNCs translating cognate polypeptides . Affinity measurements indicate that all translating ribosomes can bind SRP [58] . RNCs translating cytosolic polypeptides have significant affinity for SRP ( ca . 8 nM ) [58] , however this interaction is salt sensitive and likely has a higher dissociation rate [79] . In contrast , SRP binds with extraordinarily high , subnanomolar affinity to RNCs bearing cognate substrates; this interaction is also salt-resistant , perhaps related to its low dissociation rate in vivo [80] . Of note , NAC was shown to reduce association of SRP to non-cognate RNCs . Accordingly , loss of NAC would result in a higher residence time for SRP on ribosomes translating highly abundant non-cognate mRNAs and a lower availability of SRP to bind low abundance cognate mRNAs . Our data suggesting that SRP and NAC overlap in binding to RNCs , much as proposed for trigger factor and SRP in bacteria , open the possibility for them acting in concert on a translating nascent chain . Because it appears that the conformational state of the ribosome contributes to SRP recruitment [43] , [51] , a more speculative possibility is that NAC exerts its regulatory activity through modulation of the ribosomal cycle . Despite the relaxed specificity of SRP binding to nascent chains in ΔNAC cells , there was no detectable difference in mRNA targeting to membranes in these cells , and no significant induction of a UPR response ( Figure 7 ) , supporting previous findings that NAC has no impact on translocation or the interaction of RNCs with membranes [77] , [78] . This is likely the combined result of the redundancy of mRNA targeting pathways , which ensure that secretory proteins reach the membrane , together with proofreading mechanisms that prevent non-cognate SRP-RNC complexes from associating with membranes . For instance , the SRP targeting pathway contains an inbuilt proofreading mechanism at the SRP receptor ( SR ) level whereby the SRP-SR interaction is enhanced when SRP is bound to a signal sequence [81] , [82] . Furthermore , the Sec61 translocon can stringently recognize signal sequence RNCs [5] , [47] . These different mechanisms may together provide a robust system that ensures the fidelity of translocation even when the specificity of SRP interactions is impaired .
Strains carrying chromosomally integrated Rpl16-TAP , Rpl17-TAP , Egd1-TAP , Egd2-TAP , Btt1-TAP , Srp68-TAP , and Srp72-TAP were obtained from Open Biosystems , Srp54-TAP from Euroscarf . Δegd1 and Δegd2 yeast strains from the Saccharomyces Genome Deletion Project [83] were used to obtain Δegd1/Δegd2 by mating , sporulation , and tetrad dissection . Sec65-1 strain was a kind gift of Peter Walter . Immunoaffinity purification of specific ribosome-associated factors together with ribosomes and associated RNAs was carried out exploiting the C-terminal TAP-tagged derivative of each selected protein [84] . Briefly , 1 liter cultures were grown to OD 0 . 7–0 . 8 in YPD . Following addition of cycloheximide ( CHX ) ( 0 . 1 mg/ml ) to stabilize ribosome-nascent chain complexes , cells were harvested by centrifugation , washed twice in buffer A ( 50 mM Hepes-KOH [pH 7 . 5] , 140 mM KCl , 10 mM MgCl2 , 0 . 1% NP-40 , 0 . 1 mg/ml CHX ) , resuspended in 2 ml of buffer B ( buffer A plus 0 . 5 mM DTT , 1 mM PMSF , 20 µg/ml pepstatin A , 15 µg/ml leupeptin , 1 mM benzamidine , 10 µg/ml aprotinin , 0 . 2 mM AEBSF ( Sigma ) , 0 . 2 mg/ml heparin , 50 U/ml Superasin ( Ambion ) , and 50 U/ml RNAseOUT ( Invitrogen ) ) , and dripped into a conical 50 ml Falcon tube filled with and immersed in liquid nitrogen . Frozen cells were pulverized for 1 min at 30 Hz on a Retsch MM301 mixer mill . Pulverized cells were thawed and resuspended in 5 ml of buffer B; cell debris was removed by two sequential centrifugation at 8 , 000 g for 5 min at 4°C . A 100 µl aliquot ( 5% ) of the supernatant was removed for reference RNA isolation . The remaining lysate was incubated with 6 . 7×106 beads/µl of IgG-coupled magnetic beads ( Dynabeads , Invitrogen ) at 4°C for 2 h . Beads were washed once in 5 ml of buffer B for 2 min and 5 times in 1 ml buffer C ( 50 mM Hepes-KOH [pH 7 . 5] , 140 mM KCl , 10 mM MgCl2 , 0 . 01% NP-40 , 10% glycerol , 0 . 5 mM DTT , 10 U/ml superasin , 10 U/ml RNAseOUT , 0 . 1 mg/ml CHX ) for 1 min , resuspended in 100 µl of buffer C , and incubated for 2 h in 0 . 3 U/µl TEV protease ( Invitrogen ) at 16°C . Supernatant was recovered as final pulldown for protein and RNA isolation . Reference RNA was isolated using RNeasy mini kit ( Quiagen ) , while RNA from the eluate was isolated by sequential extraction with Acid Phenol:Chloroform 125∶24∶1 ( Ambion ) , Phenol/Chloroform/Isoamyl Alcohol 25∶24∶1 ( Invitrogen ) , and chloroform followed by isopropanol precipitation with 15 µg of Glycoblue ( Ambion ) as carrier . A total of 20 OD254 nm were loaded on a 7%–47% sucrose gradient in buffer B without NP-40 . The samples were centrifuged on a Beckman SW-41 rotor for 90 min at 42 , 000 rpm at 4°C . Gradients were continuously fractionated on an ISCO collector with a flow cell UV detector recording the absorbance at 254 nm . For protein detection by western blotting , fractions were precipitated with trichloroacetic acid , separated by SDS-PAGE , and analyzed by immunoblotting using the indicated antibodies . For RNA isolation and microarrays analysis , fractions corresponding to 60S , 80S , and polysomes were pooled to isolate polysome-associated RNA and the supernatant and low sucrose fractions were pooled to isolate free RNA . RNA was purified with RNeasy Mini Columns kit ( Qiagen ) . Free cytosolic polysomes and membrane-bound polysomes were fractionated by sedimentation velocity exactly as described [34] , [85] starting with 250 ml of exponential growth cultures ( of WT or Δegd1/egd2 ) in YPD . Total RNA from free and membrane-associated polysomes was purified with RNeasy Mini Columns Kit ( Qiagen ) . 50 ml cultures of WT , Δegd1/egd2 , or sec65-1 cells were grown in YPD at 30°C or followed by 1 h at 37°C ( sec65-1 ) . Cells were starved in SD-Met media for 30 min and labeled with 35S-methionine for 7 min . Endogenously expressed Kar2 and DPAP-B were immunoprecipitated with specific antibodies ( a kind gift of Peter Walter and Mark Rose , respectively ) and analyzed by SDS-PAGE in 8% acrylamide gels . Translocation defects were measured by comparing the ratio of non-processed precursor versus processed mature protein , namely non-signal sequence-cleaved versus cleaved Kar2 and an unglycosylated precursor versus glycosylated protein for DPAP-B . 3 µg of reference RNA and 50% or up to 3 µg of TAP-tag affinity purified RNA were reverse transcribed with Superscript II ( Invitrogen ) in the presence of 5- ( 3-aminoallyl ) -dUTP ( Ambion ) and dNTPs ( Invitrogen ) with a 1∶1 mixture of N9 and dT20V primers . The resulting cDNA was covalently linked to Cy3 ( reference RNA ) and Cy5 ( purified RNA ) NHS-monoesters ( GE HealthSciences ) . Dye-labeled DNA was diluted into 20–40 µl solution containing 3× SSC , 25 mM Hepes-NaOH , pH 7 . 0 , 20 µg poly ( A ) RNA ( Sigma ) , and 0 . 3% SDS . The sample was incubated at 95°C for 2 min , spun at 14 , 000 rpm for 10 min in a microcentrifuge , and hybridized at 65°C for 12–16 h in the MAUI hybridization system ( BioMicro ) . Following hybridization , microarrays were washed in 400 ml of four subsequent wash buffers made of 2×SSC with 0 . 05% SDS , 2×SSC , 1×SSC , and 0 . 2×SSC . The first wash was performed at 65°C for 5 min and the following washes for 2 min each at room temperature . Slides were briefly immersed in 95% ethanol and dried by centrifugation in a low-ozone environment to prevent Cy5/3 dyes destruction . Once dry , the microarrays were kept in a low-ozone environment during storage and scanning . For fractionation experiments , 10 µg of free polysomes-RNA ( Cy3 ) and 3 µg of rER polysomes-RNA were used for reverse transcription . For analysis of transcriptional levels on mutant strains , 3 µg of reference RNA ( wild type strain ) ( Cy3 ) and 3 µg of experimental RNA ( mutant strain ) ( Cy5 ) were used for reverse transcription . Microarrays were scanned with an Axon Instrument Scanner 4000B ( Molecular Devices ) . PMP levels were adjusted to achieve 0 . 05% pixel saturation for IP experiments and 0% saturation for analysis of transcriptional levels . Data were collected with the GENEPIX 5 . 1 ( Molecular Devices ) , and spots with abnormal morphology were excluded from further analysis . Arrays were computer normalized by the Stanford Microarray Database ( SMD ) [86] . Log2 median ratios were filtered for a regression correlation greater than 0 . 6 and a signal over background greater than 2 . 5 to remove low-confidence measurements . Hierachical clustering was performed with Cluster 3 . 0 [87] , and results were visualized with Java TreeView [88] . At least three , usually four , independent biological replicates were employed for each condition . After removing features missing two or more values , we generated a representative dataset by running a one-class t test with 800 ( SAM ) [22] . Ultimately , substrates were defined as those encoded by mRNAs that were differentially expressed with a false discovery rate ( FDR ) ( q-value ) of 1 . Enriched GO terms among the identified targets were retrieved with GO Term Finder [89] , which uses the hypergeometric density distribution function to calculate p values and the programs Genetrail [90] and FuncAssociate [91] . The GO database [89] was used to collect a list of GO categories . In these classifications , gene products can be affiliated with one or more GO category assignments . Lists of proteins with predicted signal peptides and transmembrane regions were downloaded from the Saccharomyces Genome Database ( SGD ) , which uses the prediction programs SignalIP [92] and TMHMM 2 . 0 [93] , respectively . Intrinsic disorder was predicted from the protein sequences with the Disopred2 software [94] after filtering out coiled-coil and transmembrane regions with the program pfilt ( http://bioinf . cs . ucl . ac . uk/downloads/pfilt ) . Reported is the fraction of the protein sequence that is predicted to be unstructured . Sequence hydrophobicity was approximated by the average of the hydrophobicity profile , computed from the Kyte and Doolittle scale [95] with averaging over sliding windows of size 7 . Hydrophobic stretches were defined as 5 or more consecutive amino acids that surpassed a threshold of 1 in the hydrophobicity profiles . Data on translation rate , ribosomal density , and mRNA expression were retrieved from [23] and [96] . Statistical data analysis was performed in R ( www . r-project . org ) . Box plots indicate the data distribution through median , 25% , and 75% quartiles ( filled box ) , as well as the range of non-outlier extremes ( dashed lines ) . | In every cell , ribosomes translate the genetic instructions carried by messenger RNAs into the proteins they encode . Molecular midwives called chaperones often bind to nascent protein chains as they emerge from the ribosome to help them fold . Very little is known about this process . Do all proteins need chaperone assistance as they exit the ribosome ? Do different chaperones recognize different polypeptide chains and , if so , how ? Answering these questions has been hard because most studies have examined only a handful of model proteins and their interactions with a specific chaperone . Here , we used a systematic approach to investigate the challenging question of chaperone specificity in living cells . We isolated specific chaperones that interact with nascent proteins during translation along with the ribosomes and associated mRNAs encoding the emerging proteins . We then used DNA microarrays to identify the full suite of mRNAs and thus the encoded proteins that interact cotranslationally with each of these factors . We learned from these studies that individual chaperones interact with a specific set of nascent proteins . Furthermore , overlapping specificity enables one chaperone to modulate the specificity and fidelity of another . The picture that emerges suggests that these molecular midwives are an important part of the intricate systems that maintain specificity , precision , and efficiency in expressing the genome's instructions . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"systems",
"biology",
"biochemistry",
"biology",
"molecular",
"cell",
"biology",
"genetics",
"and",
"genomics"
] | 2011 | Defining the Specificity of Cotranslationally Acting Chaperones by Systematic Analysis of mRNAs Associated with Ribosome-Nascent Chain Complexes |
White blood cell ( WBC ) count is a common clinical measure from complete blood count assays , and it varies widely among healthy individuals . Total WBC count and its constituent subtypes have been shown to be moderately heritable , with the heritability estimates varying across cell types . We studied 19 , 509 subjects from seven cohorts in a discovery analysis , and 11 , 823 subjects from ten cohorts for replication analyses , to determine genetic factors influencing variability within the normal hematological range for total WBC count and five WBC subtype measures . Cohort specific data was supplied by the CHARGE , HeamGen , and INGI consortia , as well as independent collaborative studies . We identified and replicated ten associations with total WBC count and five WBC subtypes at seven different genomic loci ( total WBC count—6p21 in the HLA region , 17q21 near ORMDL3 , and CSF3; neutrophil count—17q21; basophil count- 3p21 near RPN1 and C3orf27; lymphocyte count—6p21 , 19p13 at EPS15L1; monocyte count—2q31 at ITGA4 , 3q21 , 8q24 an intergenic region , 9q31 near EDG2 ) , including three previously reported associations and seven novel associations . To investigate functional relationships among variants contributing to variability in the six WBC traits , we utilized gene expression- and pathways-based analyses . We implemented gene-clustering algorithms to evaluate functional connectivity among implicated loci and showed functional relationships across cell types . Gene expression data from whole blood was utilized to show that significant biological consequences can be extracted from our genome-wide analyses , with effect estimates for significant loci from the meta-analyses being highly corellated with the proximal gene expression . In addition , collaborative efforts between the groups contributing to this study and related studies conducted by the COGENT and RIKEN groups allowed for the examination of effect homogeneity for genome-wide significant associations across populations of diverse ancestral backgrounds .
The WBC count , a classic marker of immune or inflammatory response , varies substantially among healthy individuals . The counts of constituent cell subtypes comprising the WBC count measure are assayed as part of a standard clinical WBC differential test . While the WBC count and WBC differential count are often obtained to assess for evidence of infection or underlying inflammation , prospective epidemiologic studies have consistently linked higher WBC counts , within the clinically designated normal range , along with other inflammatory markers , to increased risk of coronary artery disease , cancer , and total mortality [1] , [2] , [3] , [4] , [5] , [6] , [7] , [8] , [9] , [10] . Studies are often not consistent on the specific WBC subpopulations involved , but granulocytes , in general , and neutrophils , in particular , are most often implicated in these observations [11] , [12] . In the general population , the total WBC count is also directly associated with many cardiovascular disease risk factors , such as higher blood pressure , cigarette smoking , adiposity , lower socioeconomic status , and higher levels of plasma inflammatory markers [13] . WBC counts are also moderately heritable [14] , with heritability estimates varing from approximately 0 . 14 to 0 . 4 across total leukocyte count and cell subtypes , as assessed in a Sardinian population , with the highest heritability estimates for monocyte counts [14] . In addition , the substantially lower neutrophil count and total WBC count in African Americans compared to European-ancestry individuals seems to be at least partially explained by a regulatory variant in the Duffy Antigen Receptor for Chemokine ( DARC ) gene , which accounts for ∼20% of total variation in the measures [15] , [16] . Recent studies have sought to investigate the common genetic variants associated with several blood count traits in European-ancestry and Japanese individuals , but have not focused specifically on the multiple cell types comprising the total WBC count measurement [17] , [18] , [19] , [20] . In the current study , we sought to identify and replicate common genetic variants that influence normal variation in six WBC phenotypes commonly measured in the clinical setting and in population studies that comprise the CHARGE Consortium [21] , the HaemGen Consortium [18] and independent collaborative studies . We utilized genome-wide association study ( GWAS ) data and meta-analytic techniques to identify 10 genome-wide significant loci in a study of over 31 , 000 individuals ( Table 1 ) , examining variants possibly associated with total WBC count , three granulocyte phenotypes ( neutrophil , basophil and eosinophil counts ) , and two non-granulocyte phenotypes ( lymphocytes and monocytes ) . We also investigated the shared functional connectivity of identified loci across phenotypes with regards to both known biological pathways and nearby gene expression effects . As previous research has shown strong selective pressures at the locus identified to affect WBC counts in African American populaftions , we examined the possibility of recent selective effects at significantly associated loci in European ancestry individuals [15] , [16] . Additional efforts were made in collaboration with RIKEN and COGENT investigators to identify homogenous associations across populations of diverse ancestral backgrounds .
In the discovery meta-analysis of 19 , 509 subjects from seven cohorts , we identified 11 genome-wide significant associations with six white cell phenotypes ( total WBC , neutrophil , basophil , eosinophil , lymphocyte and monocyte counts , see Table 1 , Table S1 , Figure 1 ) . Further , we found strong evidence for replication of 10 of the 11 trait-locus associations in 11 , 823 independent samples from 10 GWAS cohorts who contributed summary statistics for SNPs of interest . The discovery analysis results ( Figure 1 and Figures S1 , S2 , S3 , S4 , S5 , S6 , S7 , S8 , S9 , S10 for details ) , and the results of replication testing for the 10 replicated trait-loci are summarized in Table 2 . These results are presented in greater detail for all genome-wide significant SNPs in Table S2 . Total WBC count was associated with two independent loci in the discovery phase of analyses; the first locus was on chromosome 6p21 encompassing a region from 31 , 131 , 127 bp to 31 , 161 , 846 bp near HLA and PSORS1 gene families , and the second locus on chromosome 17q21 from 35 , 345 , 186 bp to 35 , 470 , 048 bp near candidate genes ORMDL3 and CSF3 . Both loci showed independent replication ( Table 2 ) . Neutrophil count was associated with a 196 , 381 bp region on chromosme 17q21 containing 46 genome-wide significant SNPs from the discovery phase . This region overlaps the locus on chromosome 17q21 identified for the total WBC count phenotype and showed positive association in replication testing at all but 2 SNPs . Basophil count was associated with one SNP , rs4328821 on chromosome 3q21 near RPN1 and C3orf27 . This SNP showed a significant positive association between basophil count and minor allele dosage ( minor allele frequency 0 . 110 , p-value in discovery phase 2 . 58E-08 , p-value in replication phase 8 . 40E-06 ) . No regions showed genome-wide significance for association with eosinophil count . Lymphocyte count was associated with two loci , with one locus on chromosome 6p21 overlapping with the chromosome 6p21 total WBC count locus . The second locus associated with lymphocyte counts is on chromosome 19p13 , from 16 , 32 , 871 bp to 16 , 429 , 197 bp at EPS15L1 and was successfully replicated . Monocyte count was associated with the largest number of independent hits for any of the traits , with five loci identified in the discovery analysis , four of which showed significant associations in replication testing . These four loci include two intergenic regions ( >100 kb to nearest known genes ) on chromosome 8q24 ( from 130 , 672 , 817 to 130 , 693 , 287 ) and chromosome 9q31 ( 112 , 917 , 232 to 113 , 073 , 157 ) . We also identified a novel association on chromosome 2q31 ( 182 , 027 , 546 to 182 , 036 , 459 ) near ITGA4 , and a single genome-wide significant SNP on chromosome 3q21 , which is located 18 , 866 bp from the SNP significantly associated with basophil count ( r2 = 0 . 076 , D′ = 0 . 841 ) . The monocyte- associated locus on chromosome 1q22 contained only one genome-wide significant SNP which failed to replicate and is not included in Table 2 ( denoted by rs17131683 , which exhibited a replication p-value of 0 . 770 but a consistent negative effect associated with the A allele ) . Many of the loci described in this report contain genome-wide significant SNPs spread throughout much of each locus suggestive of either extensive LD or multiple association signals at each locus . For loci described in Table S2 ( except those containing less than 3 genome-wide significant SNPs ) , conditional analyses were conducted using the allele dosage of the most significant SNP per locus as a covariate in a subset of discovery cohorts ( AGES , ARIC , BLSA , Health ABC and InChianti ) . Statistical models were identical to those used in the discovery analyses except for the additional SNP covariates . No SNPs analyzed remained significant after correcting for 147 tests , showing that only one signal of association exists per locus . The complete results for these analyses are evident in Table S3 . As each locus accounts for only one unique signal per trait , adjusted r2 estimates were calculated for each trait across loci within this subset of cohorts , and may be found in Table 2 . To assess the independence of associated SNPs from the total WBC count measurement , all genome-wide significant SNP associations for white cell subtypes ( Table S2 ) were re-analyzed as per the discovery phase analysis methods adjusting for total WBC count as an additional covariate in a subset of discovery cohorts ( AGES , ARIC , BLSA , Health ABC and InChianti ) . After Bonferroni correction for 97 independent tests at least one SNP per locus remained significant , suggesting some independence from the total WBC measure in the SNP associations . The extended results of this analysis and a table of r2 values for the phenotypes of interest based on the same subset of discovery cohorts may be found in Table S4 and Table S5 . These results demonstrate a high degree of relatedness across traits , with individual loci affecting multiple WBC traits , that may be pleiotropic to some degree or due to the biological relatedness of the traits . Neutrophils are the most abundant WBC subtype , and the locus on chromosome 17q21 associated with both total WBC count and neutrophil count independently based on conditional analyses described above . In this region , 38 SNPs were common to both traits in the discovery analysis , and these SNPs showed identical directions of effects across both traits . In the chromosome 3q21 between C3orf27 and RPN1 , rs9880192 was associated with monocyte count and rs4328821 was associated with basophil count , suggesting pleiotropy in this region . The region on chromosome 6p21 near the PSORS1 family of genes as well as HLA-C and HLA-B contains associated SNPs with both total WBC count and lymphocyte count , although , interestingly , spatially overlapping SNPs failed to replicate in the other trait , further suggesting independence of effects seen in conditional analyses of all loci . To further examine the functional connectivity of these loci across white blood cell phenotypes , the Gene Relationships Among Implicated Loci package ( GRAIL , http://www . broad . mit . edu/mpg/grail/ ) was utilized to mine PubMed archives using textual analyses of known functional associations to identify concurrent effects and relationships across phenotypes [22] . In brief , GRAIL incorporates functional annotations from text mining related to specific genomic loci , usually genome-wide association study results , to assess the functional inter-relatedness of genes in linkage disequilibrium with the regions of interest and construct networks of related genes sharing biological function . In our analyses , we utilized GRAIL to survey the functional relatedness of all regions containing significant results passing Bonferroni correction in both the discovery and replication phases . We identified four clusters of related genes with false-discovery rate adjusted p-values<0 . 05 out of the 49 gene clusters generated by the GRAIL package , which are described in Figure 2 and Table S6 . All four clusters show significant interconnectivity between genes proximal to loci on chromosome 17q21 associated with total WBC and neutrophil counts and the chromosome 19p13 locus associated with lymphocyte count . The two most significant clusters also show relationships between genes proximal to the previously mentioned loci and candidate genes at the chromosome 8q24 region associated with monocyte counts . Genes at the chromosome 17q21 locus associated with both total WBC and neutrophil counts appeared in all significant pathways identified in the GRAIL analyses , suggesting biological connectivity across both granulocyte and non-granulocyte cell lineages . Candidate genes from the gasdermin ( GSDML ) and mediator complex subunit ( MED ) families were highly enriched in the significant gene clusters and comprised 37 . 5% of genes in these clusters . These results suggest shared biological pathways between these genes and cell proliferation among WBC subtypes associated with these genomic regions , although as emphasized in conditional analyses , these effects at these loci remain to some degree independent of the total WBC measure . We also examined possible functional consequences of individual SNP associations by analyzing whole blood genome-wide gene expression data from the InChianti study to identify associations between SNPs found to be significant in the meta-analysis and cis changes in gene expression . All SNPs significant in both the discovery and replication phases for all phenotypes were used in the expression analysis . Each SNP in this dataset which was within 500 kb of an expression probe was treated as a possible expression quantitative trait locus ( eQTL ) . For 85 SNPs in our subset of significant meta-analysis results , we tested at total 741 candidate eQTL associations using multivariate linear regression . This analysis tested each SNP in the subset for an association with each expression probe within 500 kb . After Bonferroni correction for the 741 tests conducted , 36 SNPs in the chromosome 17q21 locus associated with total WBC count and neutrophil count in the meta-analysis were also significantly associated with cis-expression levels . In total , these 36 SNPs were associated with three expression probes , 2 probes tagging transcripts in the GSDML gene ( probes ILMN_2347193 and ILMN_1666206 ) and another probe tagging a single probe in ORMD3L ( probe ILMN_1662174 ) , for a total of 103 signficicant eQTLs ( Figure 3 ) . Both probes in GSDML were highly correlated ( r2 = 0 . 582 ) , although neither GSDML probe was strongly corellated with the probe of interest in the ORMD3L gene ( r2<0 . 200 ) . With each SNP's minor allele as the point of reference , all effect directions for significant meta-analysis and eQTL associations were concordant , showing a strong correlation between the effect sizes in the meta-analysis and eQTL analysis , suggesting that the effect of the identified SNPs on the corresponding WBC trait may be transcriptionally mediated . For example , the correlation of effect estimates between the eQTL and meta-analysis for neutrophil associated SNPs also associated with the ILMN_1666206 probe highly significant , with an r2 of 0 . 898 . For details of all eQTLs tested , please refer to Table S7 and the Methods and Materials section . Previous studies of WBC genetic variation in African American populations have shown evidence of WBC associated loci in a region of high selective pressures , with the putative functional SNP rs2814778 being almost fixed in frequency in sub-Saharan African populations where malaria is endemic [15] . We therefore undertook an investigation of recent selective pressures acting upon SNPs identified associated with WBC phenotypes in European-ancestry populations . Evidence for selection for all 10 significant trait-loci from the meta-analysis was evaluated by mining HapMap2 CEU data . Integrated Haplotype Scores ( iHS ) were used to quantify selection at each locus based on homozygosity decay in extended haplotypes and are availible for download from the Haplotter website ( http://hg-wen . uchicago . edu/selection ) [23] . Selection was quantified by an absolute value of iHS>2 indicating strong selective pressures , and absolute iHS values ≤2 and ≥1 . 635 were interpreted as indicating recent-moderate selection , positive or negative iHS values indicated the direction of selection [24] . All replicated SNPs were evaluated for signatures of selection . One locus showed evidence of selection , and this was the lymphocyte associated locus on chromosome 19p13 , from 16 , 336 , 388–16 , 429 , 197 bp . This locus on chromosome 19p13 showed evidence of selection for all SNPs analyzed in the replication phase . All SNPs in this locus appear to be under some degree of negative selection with 2 of the 11 SNPs being under strong negative selection and the rest under recent moderate selection . However , when testing the correlation between effect estimates and iHS at this locus using the ancestral allele as a reference for effect direction , no clear correlation between iHS and effect size was detected ( r2 = 0 . 226 , p-value = 0 . 140 ) . Full iHS annotation for replicated SNPs are shown in in Table S8 . As part of collaborative efforts with the COGENT and RIKEN groups , a coordinated exchange of summary statistics for SNP identified in Table 2 was organized from the parallel studies conducted by these groups . Random-effects meta-analysis techniques were utilized to identify effects that were consistent across studies of diverese ethnic backgrounds . While all joint effect estimates remained in consistent directions with those described in Table 2 , heterogeneity of effects across the 3 ancestral populatons severely attenuated the strength of the associations for all but 2 of the genome-wide significant associations identified in this report . The associations at rs4794822 ( WBC count ) and rs11878602 ( lymphocyte count ) remained genome-wide significant . Consistent robust effects across ancestral populations at rs11878602 may lend some support to the recent-moderate negative selection at this SNP ( iHS = −1 . 924 ) being associated with an increase in frequency of the derived A allele associated with decreasing monocyte counts . These results suggest that these SNPs may be very close to the functional variants associated with these effects , as well as exhibiting relatively consistent effects across multiple population ancestries with differing LD structures . The results of these analyses are detailed in Table 3 .
This meta-analysis has identified ten genome-wide significant trait-locus associations with WBC phenotypes spread across seven genomic loci . Of these trait-locus associations , seven associations represent novel findings , and three associations , across two genomic loci , confirm previously identified associations . The association of chromosome 17q21 near ORMDL3 and CSF3 associated in our study with total WBC count and neutrophil count , and the 9q31 locus associated with monocyte count have been previously demonstrated in European-ancestry and Japanese populations [25] , [26] . The chromosome 3q21 locus near RPN1 and C3orf27 has been previously shown to be associated with eosinophil count , and is instead associated with related granulocyte cell measures of monocyte and basophil counts in this study , although we do identify suggestive p-values at ∼1 . 00E-4 to 1 . 00E-8 and the same direction of effect for the additional loci identified in Gubjartsson et al . , 2010 [19] , [27] . In addition , a number of previous GWAS have implicated the monocyte count associated locus at chromosome 8q24 . 21 as affecting height , renal function , serum protein levels , multiple sclerosis , glioma , leukemia and a number of other cancers [28]–[59] . Through conditional analyses and an analysis of functional relatedness , we have shown correlation among related traits and possible pleiotropic connectivity of these loci across phenotypes that represent measures of biologically related cellular lineages , as well as identified loci showing a direct association between allelic gene expression differences and variation in phenotypic measures . The associations identified in this study are robust , and several have been previously identified in GWAS studies of immunologically relevant phenotypes , such as the association of celiac disease with the chromosome 2q31 . 1 locus containing ITGA4 . ITGA4 encodes an alpha integrin subunit present on monocytes , lymphocytes , endothelial cells and erythrocytes that serves as an adhesion molecular receptor for VCAM-1 , fibronectin . VCAM-1 is expressed at high levels on the vasculature of the bone marrow , and therefore alpha4integrin receptors may play a role in homing and recruitment of certain cell types during hematopoiesis [60] , [61] . The overlapping loci on chromosome 17q21 associated with both total WBC and neutrophil counts constitute a single plausible candidate locus contributing to general variability in total WBC count and neutrophil count via hematopoietic mechanisms . These measures are highly correlated , and the estimated SNP effects are also likely correlated for this reason . From a functional perspective , the role of G-CSF , the CSF3 gene product , has been well-described in the biology of myeloid progenitor production and differentiation . This locus previously reached genome-wide significance in a joint analysis of discovery and replication cohorts of total WBC count in individuals of European ancestry [18] . The same locus , containing the genes PSMD3 and CSF3 , was associated with neutrophil count in a cohort of Japanese participants [27] . This study of Japanese subjects also reported a significant genome-wide association with neutrophil count for three SNPs in a locus at chromosome 20 , containing the gene PLCB4 . Due to the lower minor allele frequency in European ancestry individuals of these three correlated SNPs ( minor allele frequency = 0 . 076 in HapMap CEU samples versus 0 . 289 in HapMap JPT samples for rs2072910 ) and the marginal effect size detected in the original report , statistical power to detect this association was limited in our analysis [27] . Our data suggest that the locus on chromosome 17q21 has functional connectivity across white cell subtypes . Multiple genes at this locus appeared in all significant pathways identified in the GRAIL analyses showing a functional connectivity across both granulocyte and non-granulocyte cell lineages . Gene clusters detailed in the GRAIL analyses show significant functional clusters including genomic regions that are separately associated with granulocyte and non-granulocyte traits within the same cluster . However , the results of the GRAIL analyses may be influenced by both funding avenues and publication bias , as the classifications are based only on PubMed searchable published results . Cis-eQTL effects on transcripts in the ORMDL3 and GSDML were shown to be highly correlated with allelic effects associated with neutrophil and total WBC count measures quantified in the meta-analysis . This suggests that variation in total WBC count and neutrophil counts are at least in part due to polymorphism-based regulation of gene expression . This regulation of gene expression by SNPs in this region may be related to some form of systemic immunological function , as variants in this region were also shown to be associated with expression of ORMDL3 in childhood asthma [62] . Interestingly , Okada et al . also showed significant eQTL associations between SNPs and transcripts at this locus , although their associations implicate regulation of PSMD3 as affecting neutrophil variation , rather than the exact transcript associations identified in this report [27] . This slight difference may be attributable to the larger sample size in our study , differences in population ancestry of the two studies , and the use of expression data derived from lymphoblastoid cell lines instead of whole blood . The CSF3 gene at this locus is a possible candidate contributing to effects across multiple cell subtypes at this locus , as functional studies have demonstrated that this gene creates a protein integral to the differentiation and functionality of granulocyte cells [63] , [64] , [65] , [66] . One important caveat to our eQTL analysis is the lack of representation of the CSF3 gene on the expression array used . The loci on chromosome 6p21 associated with total WBC and lymphocyte counts appear to be independent of each other as the lymphocyte association persists after adjustment for total WBC . While the closest replicated SNP associations are roughly 200 kb from each other , and a shared functional connectivity between these regions was not elucidated in the GRAIL analysis . At this locus , rs2524079 associated with lymphocyte count is in moderate LD with a number of SNPs in the periphery of the total WBC count locus , including rs2844619 , a SNP significant in the total WBC count discovery phase analysis ( D′ = 0 . 762 , r2 = 0 . 305 , from HapMap Phase II CEU samples ) . The finding of multiple independent effects at a single locus has occurred in prior studies and include examples such as the finding of two independent signals within the PLAG1 locus associated with human height , suggesting a locus specific effect , in the current example affecting leucopoiesis or leukocyte homeostasis [48] . Our chromosome 6p21 WBC and lymphocyte loci , both harbor candidate genes that have been previously implicated as associated with phenotypes closely related to immunological function . The locus associated with total WBC count on chromosome 6p21 contains genes associated with follicular lymphoma ( CHCG22 ) , progression of HIV-1 infection ( CDSN , PSORS1C1 and PSORS1C2 ) and psoriasis ( CCHCR1 , PSORS1C1 and PSORS1C2 ) [67] , [68] , [69] . This gene rich region includes HLA family genes , particularly HLA-B and HLA-C that are candidates within the lymphocyte associated locus , and actually overlap with the psoriasis candidate locus identified previously via linkage mapping studies and includes PSORS1C1 and PSORS1C2 , showing the relatedness of these two loci on chromosome 6p21 [69] , although conditional analyses adjusting for total white blood cell count validate these as primarily independent effects . The lymphocyte associated regions containing HLA-B and HLA-C , harbors two genes that have been implicated in multiple GWAS as modifiers of immunological responses , associated with IL-18 levels , HIV-1 control , vitiligo , multiple clerosis , and psoriasis [49] , [68] , [70] , [71] , [72] , [73] , [74] , [75] , [76] . The region associated with basophil and monocyte counts on chromosome 3q21 proximal to the GATA2 gene was previously described as associated with variation in eosinophil count in European and Asian ancestry populations [19] . This association with eosinophil count was not identified as genome-wide significant in our analyses , with multiple SNPs in the GATA2 region ( GATA2+/−250 kb ) approaching genome-wide significance , exhibiting p-values in the discovery meta-analysis including a regional minimum of 6 . 73E-07 ( rs4328821 ) . This is likely due to our decreased sample size for analyses of eosinophil count compared to the original report . However , our data show a significant association between this genomic region and basophil count , and basophils and eosinophils are both granulocytic cells with common lineage in WBC differentiation . GATA2 is a well-known transcription factor involved in maintenance of early hematopoietic cell pools and proximal hematopoietic pathways . In addition , this region proximal to GATA2 is also significantly associated with monocyte counts , showing overlapping associations across both granulocyte and non-granulocyte cell lineages and supporting the previously described functional role of GATA2 more proximally in the WBC differentiation process [77] , [78] . The independence across granulocyte and non-granulocyte lineages is evident as both associations showed independent signals of association after adjustment for total WBC . Our analyses have identified genomic loci associated with total WBC and constituent white cell subtypes in European ancestry cohorts . Our findings differ from the results of similar studies of African American and Asian ancestry populations . Population variation at previously identified total WBC count associated loci of DARC in African American cohorts and PLCB4 in Asian ancestry samples motivated our investigation of possible selection at the loci identified in this report , as allele frequencies of SNPs in DARC and PLCB4 differ across populations . This may be suggestive of recent selection . Our analyses of iHS statistics for genome-wide significant SNPs yielded only one locus under selection , with all SNPs investigated in this region being under negative selection in European ancestry populations . These SNPs on chromosome 19p13 associated with lymphocyte counts are proximal to candidate genes such as CHREP , which function in calcium homeostasis in lymphocytes , and mutations in the coding region of CALR3 are associated with familial hypertrophic cardiomyopathy [79] , [80] . A thorough search of literature did not reveal any known selective factors associated with this locus . In addition , the fact that this locus remained genome-wide significant in random-effects modeling across diverse ancestral populations suggests a highly generalizable effect at this locus that may or may not be related to selective factors . In conclusion , we have identified and replicated a set of 10 independent trait-locus associations influencing multiple related WBC traits , of which seven are novel associations . Integrative analyses of our association data and gene expression analyses support pleiotropic effects that will require further functional testing to clarify .
All participating studies conducted their research in accordance with their respective institutional scientific and ethical review boards . All human participants provided informed consent and all clinical investigation was conducted in accordance with the Declaration of Helsinki . WBC counts were measured in 19 , 509 subjects in 7 discovery cohorts ( The Rotterdam Study ( RS ) , Framingham Heart Study ( FHS ) , the NHLBI's Atherosclerosis Risk in Communities ( ARIC ) Study , the Age , Gene/Environment Susceptibility – Reykjavik Study ( AGES ) Study , Health Aging and Body Composition Study ( Health ABC ) , the Baltimore Longitudinal Study of Aging ( BLSA ) , and the Invecchiare in Chianti Study ( InChianti ) ) and 11 , 823 subjects in 10 replication cohorts ( the Sorbs , the Twins UK cohort ( TwinsUK ) , Kooperative Gesundheitsforschung in der Region Augsburg ( KORAF3 & KORAF4 ) and UK Blood Services Donor Panel 1 ( UKBS1 ) studies , three of the Italian Network on Genetic Isolates ( INGI ) studies ( Carlantino , Val Borbera and Friuli Venezia Giulia ) , the Rotterdam Study II ( RSII ) and the Heart and Vascular Health Study ( HVH ) ) . In order to study genetic factors affecting variation of these traits within normal ranges , each study excluded all participants with any WBC measure ( total WBC or one of the 5 cell subtypes ) outside of +/−2 standard deviations from the mean value for that trait . WBC phenotypes were derived from data provided by fluorescence activated cell sorting technologies commonly employed in clinical and epidemiological studies to interrogate common hematological elements found in peripheral blood . Total WBC count was reported in thousands of cells per ml , and sub-type specific cell counts were calculated by multiplying the proportion of the WBC count comprised by each cell type by the total WBC measure . Any subject with a trait value greater than 2 SD from the corresponding mean of that trait in each cohort or missing data for any assayed phenotype were excluded from all analyses . Shapiro-Wilk tests of normality were implemented in the smallest discovery cohorts as the data was available at the time of study design ( the InChianti study and the Baltimore Longitudinal Study of Aging , BLSA ) to evaluate normality of the phenotypes for analyses . Raw values , natural log transformed and square-root transformed values for each phenotype of interest in these two studies were compared with regard to deviations from normality based assessment of the Shapiro-Wilk test statistic in the two studies . Based upon these reviews , a uniform analysis plan was established for conducting each study-level analyses , analysis , using either log transformation ( total WBC count , neutrophil count , and monocyte count ) or square-root transformation ( basophil count , eosinophil count and lymphocyte count ) in order to normalize the distributions of the phenotypic data . At the study level , GWAS analyses were conducted on unrelated participants ( except for FHS ) of confirmed European ancestry based on either multi-dimensional scaling or principal components analyses , concordance between genotypic and self-reported gender , and successfully genotyped at >95% of attempted SNPs . SNPs were filtered based on criteria of minor allele frequency >0 . 01 ( MAF ) , missingness per SNP<5% and Hardy-Weinberg equilibrium p-value>1 . 00E-7 ( HWE , used to exclude poorly clustered genotypes ) . Participants and SNPs passing basic quality control were imputed to >2 . 4 million SNPs based on HapMapII haplotype data . All studies utilized multivariate linear regression to generate study level summary statistics for each phenotype , with allelic dosages at each SNP used as the independent variable and primary covariates of age at hematology assay , current smoking status and sex . Detailed descriptions of participating studies , their quality control practices and study-level analyses which may differ slightly from those described above are provided in Text S1 . To conduct meta-analyses , all studies submitted summary statistics from the study-level linear regression analyses for each phenotype . Meta-analyses were conducted using inverse-variance weighted fixed-effects models to combine beta coefficients and standard errors from study level regression results for each SNP to derive a combined p-value . Prior to discovery meta-analyses , SNPs were excluded if imputation quality metrics ( equivalent to the squared correlation between proximal imputed and genotyped SNPs ) were less than 0 . 30 . Study level results were also corrected for genomic inflation factors ( λGC ) by incorporating study specific λGC estimates into the scaling of the standard errors ( SE ) of the regression coefficients by multiplying the SE by the square-root of the genomic inflation factor ( see Table S1 for study and phenotype specific genomic inflation factors ) [81] . Study specific genomic inflation factor estimates for all discovery cohorts were all <1 . 05 except for 1 . 12 in the Health ABC analysis of basophil count and 1 . 09 in the analysis of total white blood cell count in AGES ( Table S1 ) . No definitive cause of this inflation could be identified , and of particular interest , the genomic inflation factors for related traits in these two studies were within the normally accepted range . Meta-analyses were implemented using METAL and independently re-analyzed using R to confirm results [82] . We chose a priori to carry over all results from discovery meta-analyses at p-values<5 . 00E-08 to replication meta-analyses , excluding any SNPs with Cochran's Q test of heterogeneity p-values<0 . 01 or missing in more than 2 studies . These conservative exclusion criteria caused the exclusion of 6 of 167 genome-wide significant SNPs from replication analyses , and these SNPs did not constitute any new loci of interest . The final number of SNPs for replication analyses was then reduced to 161 candidate SNPs across all phenotypes . For replication meta-analyses of individual SNPs , each phenotype was analyzed separately using similar inverse-variance weighted meta-analyses as in the discovery stage analyses , although no genomic control was used . P-values for significant associations in the replication stage were corrected for the number of SNPs tested per phenotype using the standard Bonferroni correction for multiple testing ( total WBC count corrected for 63 SNPs , with a significance threshold of p-value≤7 . 94E-4; neutrophil count corrected for 46 SNPs , with a significance threshold of p-value≤1 . 09E-3; basophil count corrected for 1 SNP , with a significance threshold of p-value≤0 . 05; lymphocyte count corrected for 14 SNPs , with a significance threshold of p-value≤3 . 57E-3; and monocyte count corrected for 37 SNPs , with a significance threshold of p-value≤1 . 35E-3 ) . Of the 161 candidate SNPs included in the replication phase , 152 SNPs passed the trait-specific replication p-value thresholds . Ony one genome-wide significant locus failed to replicate , and this locus on chromosome 1q22 contained only one genome-wide significant SNP associated with monocyte count in the discovery phase . Of the 152 successfully replicated associations , 109 SNPs were unique , since some SNPs were significant across multiple phenotypes . These replicated SNPs were analyzed in GRAIL to infer a possible biological connection between significant meta-analysis loci . GRAIL was used to mine textual data based on PubMed keywords to examine functional relatedness across phenotypes based on inferred biological interconectivity between genes proximal to meta-analysis results . SNP ( rs ) identifiers for these associated SNPs were input into the GRAIL webserver as a means of specifying genomic regions of interest in constructing query and seed regions to be analyzed . Genes for text mining of the functional datasource were identified using the LD structure of HapMap2 Release 22 CEU samples , gathering gene identifiers to search indexed abstracts from PubMed last curated on May 2010 . Genes in regions of interest were clustered based on keyword similarity . These genes and clusters were then scored based on ranked similarity , adjusting for gene size , to generate p-values evaluating the strength of the functional interconnectivity of genes in the regions of interest . P-values for these functional clusters were then false-discovery rate adjusted ( FDR ) to correct for multiple testing , with the FDR adjusted p-value of 0 . 05 considered the threshold for significance . For the eQTL analysis , 501 participants with complete genotyping data from the InChianti study were also successfully assayed on Illumina HT12v . 3 genome-wide expression arrays using RNA isolated from whole blood . Quality control of the genome-wide expression data included the exclusion of probes with detection p-values>0 . 01 with missing data for greater than 5% of participants . Samples must have been assayed with at least 95% of the filtered probe sets passing quality control in order to be included in analyses . 5094 probes passed quality control and were subsequently cubic-spline normalized prior to analysis . In the investigation of possible cis-eQTL associations at regions of interest identified in the meta-analysis , all probes within 500 kb of successfully replicated SNPs from the meta-analysis were identified based on annotations from ReMOAT ( http://www . compbio . group . cam . ac . uk/Resources/Annotation/ ) [83] . Thus , we tested 741 possible cis-eQTLs . Multivariate linear regression was implemented using PLINKv . 1 . 07 [84] , testing the dosage of minor alleles as a predictor of gene expression level for each probe . These linear regression models were adjusted for hybridization batch , amplification batches , sex , smoking , study site and age at baseline of study . The p-values generated by each analysis was corrected for the number of tests , with a minimum threshold of significance at p-value≤6 . 75E-05 . | WBC traits are highly variable , moderately heritable , and commonly assayed as part of clinical complete blood count ( CBC ) examinations . The counts of constituent cell subtypes comprising the WBC count measure are assayed as part of a standard clinical WBC differential test . In this study we employed meta-analytic techniques and identified ten associations with WBC measures at seven genomic loci in a large sample set of over 31 , 000 participants . Cohort specific data was supplied by the CHARGE , HeamGen , and INGI consortia , as well as independent collaborative studies . We confirm previous associations of WBC traits with three loci and identified seven novel loci . We also utilize a number of additional analytic methods to infer the functional relatedness of independently implicated loci across WBC phenotypes , as well as investigate direct functional consequences of these loci through analyses of genomic variation affecting the expression of proximal genes in samples of whole blood . In addition , subsequent collaborative efforts with studies of WBC traits in African-American and Japanese cohorts allowed for the investigation of the effects of these genomic variants across populations of diverse continental ancestries . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biotechnology",
"medicine",
"public",
"health",
"and",
"epidemiology",
"statistics",
"applied",
"mathematics",
"immunology",
"text",
"mining",
"mathematics",
"population",
"biology",
"information",
"technology",
"epidemiology",
"biology",
"hematology",
"computer",
"science... | 2011 | Multiple Loci Are Associated with White Blood Cell Phenotypes |
We propose a biologically plausible architecture for unsupervised ensemble learning in a population of spiking neural network classifiers . A mixture of experts type organisation is shown to be effective , with the individual classifier outputs combined via a gating network whose operation is driven by input timing dependent plasticity ( ITDP ) . The ITDP gating mechanism is based on recent experimental findings . An abstract , analytically tractable model of the ITDP driven ensemble architecture is derived from a logical model based on the probabilities of neural firing events . A detailed analysis of this model provides insights that allow it to be extended into a full , biologically plausible , computational implementation of the architecture which is demonstrated on a visual classification task . The extended model makes use of a style of spiking network , first introduced as a model of cortical microcircuits , that is capable of Bayesian inference , effectively performing expectation maximization . The unsupervised ensemble learning mechanism , based around such spiking expectation maximization ( SEM ) networks whose combined outputs are mediated by ITDP , is shown to perform the visual classification task well and to generalize to unseen data . The combined ensemble performance is significantly better than that of the individual classifiers , validating the ensemble architecture and learning mechanisms . The properties of the full model are analysed in the light of extensive experiments with the classification task , including an investigation into the influence of different input feature selection schemes and a comparison with a hierarchical STDP based ensemble architecture .
The standard MoE architecture [24 , 32] used in machine learning is shown in Fig 1 . The outputs of an ensemble of N classifiers feed into a final decision unit whose output is the combined classification . A separate gating network , with N outputs , weights the individual classifier outputs , typically by multiplying them by the corresponding gating output ( Fig 1 ) . The final decision unit uses a simple rule ( often some variation of the highest weighted classification from the ensemble classifiers ) to generate the final classification . The classifiers and the gating network are typically feedforward nets which are trained by a gradient descent algorithm in a supervised manner . In the standard setup the classifiers in the ensemble and the gating network all receive the same input data . The classifiers and the combining mechanism , via the gating network , adapt together , with the gating mechanism helping to ‘guide’ learning . This often leads to some degree of specialization among the ensemble with different classifiers performing better in different areas of the input space . Extensions can include more explicit variation among the classifiers by providing them with different inputs ( e . g . different sub samples , or features , of some overall input vector ) . Techniques such as this can encourage diversity among the classifiers which is generally a good thing in terms of performance [18] . In general , ensemble methods , such as MoE , have been shown to outperform single classifier methods in many circumstances . The combined performance of an ensemble of relatively simple , cheap classifiers is often much better than that of the individual classifiers themselves [16 , 18 , 20] . Our model of ensemble learning in biologically plausible spiking neural networks does not attempt to slavishly follow the methods and structure of the standard MoE architecture , but instead adapts some of the basic underlying principles to produce a MoE like system which can operate according to biologically plausible mechanisms which are based on empirical findings . The term input timing dependent plasticity ( ITDP ) was first coined in [26] where it was empirically demonstrated in the hippocampus . It is a form of heterosynaptic plasticity—where the activity of a particular neuron leads to changes in the strength of synaptic connections between another pair of neurons , rather than its own connections . Classical Hebbian plasticity involves correlations between pre- and post- synaptic activity , specifically activity in the presynaptic cell is causally related to activity in the postsynaptic cell [33] . By contrast , ITDP involves synaptic plasticity which is induced by correlations between two presynaptic pathways . Dudman et al . ( 2007 ) [26] observed that stimulation of distal perforant path ( PP ) inputs to hippocampal CA1 pyramidal neurons induced long-term potentiation at the CA1 proximal Schaffer collateral ( SC ) synapses when the two inputs were paired at a precise interval . The neural system is illustrated in Fig 2 left . Plasticity at the synapse ( SC ) between neurons CA3 and CA1 is induced when there is a precise interval between stimulations from CA3 and from the distal ( PP ) perforant pathway from neuron EC in the entorhinal cortex ( see timing curve , Fig 2 left ) . More recently , Basu et al . ( 2016 ) [29] have extended these findings by investigating the role of additional long-range inhibitory projections ( LRIPs ) from EC to CA1 , the function of which were largely unknown . They showed that the LRIPs have a powerful gating role , by disinhibiting intrahippocampal information flow . This enables the induction of plasticity when cortical and hippocampal inputs arrive at CA1 pyramidal neurons with a precise 20ms interval . Humeau et al . ( 2003 ) [25] observed a very similar form of heterosynaptic plasticity in the mammalian lateral amygdala . Specifically , simultaneous activation of converging cortical and thalamic afferents induced plasticity . More recently ITDP has been demonstrated in the cortex [27] and in the cortico-amygdala regions [28] . Another study [34] predicted the function of the vestibule-occular reflex gain adaptation by modeling heterosynaptic spike-timing dependent depression from the interaction between vestibular and floccular inputs converging on the medial vestibular nucleus in the cerebellum . Dong et al . ( 2008 ) [35] also reported a related kind of heterosynaptic plasticity operating in the hippocampus , but on different pathways from those studied by Dudman et al ( 2007 ) [26] and Basu et al . ( 2016 ) [29] . Thus this , as yet little studied , form of plasticity appears to exist in many of the main brain regions associated with learning and the coordination of information from multiple sensory/internal pathways . In the above example of ITDP acting in the hippocampus ( Fig 2 ) , the role of neuron EC in enabling ITDP driven plasticity at synapse SC is somewhat reminiscent of the action of the gating neurons in the MoE architecture outlined in the previous section , especially when we take into account the new findings that the EC to CA1 inhibitory projections do indeed enable a gating mechanism [29] . Moreover , distal projection from the entorhinal cortex to the CA1 region are topographic [36 , 37] and the enhancement of excitatory postsynaptic potentials ( EPSP ) is specific to the paired pathway [26] , indicating that only the ITDP synapse which is paired with the distal signal is potentiated . These facts suggest the possibility of specific targeted pathways enabling ‘instructor’ signals . In addition , the EPSP from the distal input is attenuated [26] , meaning that the ‘instructor’ signal would not directly influence any final network output , rather it indirectly influences through ‘instructions’ that enable plasticity . These properties are exactly those needed to operate a biologically plausible spiking MoE type architecture . This led us to the development of such an architecture using an ensemble of spiking networks with ITDP-activating distal connections playing a kind of gating role which allows coordinated learning in the ensemble ( these connections are a slight abstraction of the PP and LRIP connections rolled into one , to provide a temporally precise mechanism ) . This system is described over the following sections and embodies our biologically founded hypothesis of a potential role for ITDP in coordinating ensemble learning . First a tractable analytic model of the biologically plausible ITDP driven spiking ensemble architecture and its attendant MoE type mechanisms is developed . Derived from a logical model based on the probabilities of neural firing events , this gives insights into the system’s performance and stability . With this knowledge in hand , the analytic model is extended into a full , biologically plausible , computational implementation of the architecture which is demonstrated on a visual classification task ( identifying hand written characters ) . The unsupervised ensemble learning mechanism is shown to perform the task well , with the combined ensemble performance being significantly better than that of the individual classifiers . The properties of the full model are analysed in the light of extensive experiments with the classification task , including an investigation into the influence of different input feature selection schemes and a comparison with a hierarchical STDP-only based ensemble architecture .
This section describes the analytic formulation of ITDP driven spiking ensemble learning using probability metrics . The development of such an analytic/logical model serves two purposes: to demonstrate and better understand the mechanisms of spike-based ensemble learning , particularly the coordination of classifier outputs through ITDP , and as the basis of a fast , simplified model which can be used to provide unsupervised learning in an ensemble of arbitrary base classifiers . Later in the paper we extend the proposed model to a more biologically plausible spiking neural network ensemble learning architecture . Next we built an extended logical model for learning the weighted combination of a population ( ensemble ) of spiking neuronal voters ( classifiers ) using the simplified ITDP model described earlier . A voter was assumed to have a set of output neurons ( one for each class ) each of which fires an event ( spike ) according to its firing probability distribution . The voter follows the mechanism of stochastic winner-takes-all ( sWTA ) , where only a single neuron can fire for any presented input data . The firing probabilities of the neurons in a voter sum up to unity and these probabilities are determined by the input presented to the voter . Therefore , a voter generates a stochastic decision ( casts a vote representing the classification ) by firing a spike from one of its output neurons whenever an input pattern is presented to the voter . The input pattern shown to the voter can be any neurally coded information ( such as an image , sound , or tactile information ) which is to be classified by the voter . A pattern given to the voter is regarded as being labeled as belonging to a certain class ( c ) , where the number of existing classes is assumed to be initially known . However , it is unnecessary to relate the absolute value of the class label to the specific neuron index , since any voter neuron can represent an arbitrary data class by firing dominantly . In this abstract model , which was primarily motivated as a vehicle to test the efficacy of ITDP driven coordination of ensemble member outputs , the individual ensemble classifiers were assumed to be fully trained in advance using an arbitrary set of input data . Their tables of firing probabilities ( as in Fig 3 ) effectively represent the posterior probabilities of each class for a given input vector . Using the simplified voter model , we can build an analytically tractable voter ensemble network capable of learning the spike-based weighted combination of the individual voters . In other words , learn to combine the individual votes by weighting them appropriately so as to give a better overall classification . The ensemble system consists of three subsystems similar to those in the MoE architecture: an ensemble of voters , a final voter which receives the decisions from the ensemble and combines them to give the final classification output , and a gating voter which guides ITDP between the ensemble and the final voter ( Fig 3 right ) . The neurons of all voters in the ensemble project connections to all the neurons in the final voter ( c . f . proximal projections from CA3 in the hippocampal case ) , whereas the gating voter projects topographic ( one to one ) distal connections to the final voter ( Fig 3 right , c . f . distal topographic projections from EC in the hippocampal case ) . Every ensemble voter and the gating voter take their own representation vectors derived either from the same input pattern or from different patterns from distinct input subsets ( e . g . different regions of an image ) . The spikes from the gating voter passing through the topographic distal connection are assumed to have no significant contribution to the final voter output ( except indirectly through guiding ITDP ) . This is because , following the biological data , in our model long range EPSP propagation from distal synapses to the soma is significantly attenuated and therefore has little influence on evoking postsynaptic action potentials [26] . The gating voter guides ITDP via its topographic projections , which selectively enhance the connection strengths from the ensemble voter neurons representing the same class to one of the final voter neurons ( the gating voter’s topographic counterpart ) regardless of the ensemble neuron indices . Therefore , the system produces the ‘unsupervised’ weighted combination of ensemble outputs by learning the ITDP weights to reflect the long term co-firing statistics of the ensemble and the gating voter so that the most coherent neuronal paths for a specific class are converged to one of the final voter neurons . We derived the following analytic solution ( Eq 6 ) for the values of the weights of the ITDP synapses projecting from the voter ensemble to the final voter ( Fig 3 ) under equilibrium ( i . e . when they have converged after learning ) . See Methods for details of the derivation . w k i j = log ( a ) - log ∑ l = 1 M { p ( m i j | x l ) + p ( g k | x l ) } ∑ l = 1 M p ( m i j | x l ) p ( g k | x l ) - 1 ( 6 ) Where p ( m i j | x l ) is the firing probability of the ith neuron of the jth ensemble voter for input sample xl , w k i j is the weight from m i j to the kth neuron ( fk ) of the final voter , and p ( gk|xl ) is the firing probability of the corresponding gating voter neuron which projects to fk . We also derived an analytic solution for the expected firing probability of a final voter neuron under the presentation of the samples belonging to a particular class as given in Eq 7 ( see Methods for derivation ) . E [ p ( f k | c ) ] = 1 M c ∑ l = 1 M c ∑ q = 1 N S p ( s q | x l ) · e u k ( q ) ∑ r = 1 N C e u r ( q ) , ( 7 ) where p ( fk|c ) is the firing probability of a final voter neuron at qth ensemble state sq under presentation of the samples from class c , uk ( q ) is the weighted sum of spikes from the ensemble in state sq arriving at the postsynaptic neuron k , and NC is the number of classes ( see Methods for full explanation of all terms ) . This gives the analytic solution of the final voter firing probabilities as a function of joint probabilities of ensemble voter firings under each class presentation . The addition of these expression now gives us a complete analytic spiking ensemble model . The logical voter ensemble model described in the previous section showed that the computational characteristics of ITDP provide a novel functionality which can be used to coordinate multiple neural classifiers such that they perform spike based online ensemble learning . This form of ensemble learning simultaneously solves both the weighted vote and combining problems of arbitrarily ordered decisions from individual classifiers in an unsupervised manner . After this validation of the overall ensemble scheme , we next investigated an extended neural architecture for combined learning in an ensemble of biologically plausible spiking neural network classifiers using ITDP . The overall scheme is based on the initial simplified model , but the components are now significantly extended . Instead of assuming the individual classifiers are pre-trained , they are fully implemented as spiking networks with their own inherent plasticity . Individual classifier and overall ensemble learning dynamics occur simultaneously . The individual classifiers in the ensemble are implemented as Spiking Expectation Maximisation ( SEM ) neural network which have been shown to perform spike based Bayesian inference [30] , an ability that is often cited as an important mechanism for perception [38–40] in which hidden causes ( e . g . the categories of objects ) underlying noisy and potentially ambiguous sensory inputs have to be inferred . A body of experimental data proposes that the brain can be viewed as using principles of Bayesian inference for processing sensory information in order to solve cognitive tasks such as reasoning and for producing adequate sensorimotor responses [41 , 42] . Learning using Bayesian inference updates the probability estimate for a hypothesis ( a posterior probability distribution for hidden causes ) as additional evidence is acquired . Recently , a spike-based neuronal implementation of Bayesian processing has been proposed by Nessler et al . [30 , 43 , 44] as a model of common cortical microcircuits . Their feedforward network architecture implements Bayesian computations using population-coded input neurons and a soft winner takes all ( WTA ) output layer , in which internal generative models are represented implicitly through the synaptic weights to be learnt , and the inference for the probability of hidden causes is carried out by integrating such weighted inputs and competing for firing in a WTA circuit . The synaptic learning uses a spike-timing dependent plasticity ( STDP ) rule which has been shown to effectively implement Maximum Likelihood Estimation ( MLE ) allowing the network to emulate the Expectation Maximization ( EM ) algorithm . The behaviour of such networks was validated by a rigorous mathematical formulation which explains its relation to the EM algorithm [30] . Our reimplementation and extension of Nessler’s [30] model forms the basis of our classifiers and is well-suited for integration into our spike-based ensemble system . Viewing the SEM model as a unit cortical microcircuit for solving classification tasks , we can naturally build an integrated ITDP-based ensemble architecture as an extension of the logical ITDP ensemble model described earlier . Fig 4 shows the two layer feedforward neural architecture for the SEM-ITDP ensemble system . The first layer consists of an ensemble of SEM networks and a gating SEM , which share the presynaptic input neurons encoding the input data . Reflecting the often non-uniform , and specifically targeted , convergent receptive fields of cortical neurons involved in perceptual processing [45] , each WTA circuit receives a projection from a subset of input neurons ( representing e . g . a specific retinal area ) , which enables learning for different ‘feature’ subsets of the input data . All synapses in the ensemble layer are subjected to STDP learning . Following Nessler et al . ( 2013 ) [30] and others , in order to demonstrate and test the operation of the system , binarized MNIST handwritten digit images [46] were used as input data for classification , where the ON/OFF state of each pixel is encoded by two input neurons . The MNIST dataset is a large database of handwritten digits covering a wide range of writing styles , making it a challenging problem . The output from the ensemble layer is fed to the final WTA circuit via ITDP synapses which are driven by the more biologically plausible ITDP curve shown in Fig 2 . The following sections will describe in detail the model SEM circuit and the ITDP dynamics , followed by an investigation into how the SEM-ITDP ensemble system applied to image classification performed simultaneous realtime learning of both the individual classifier networks and the ITDP layer in parallel . Although the starting points for the ITDP based ensemble architecture proposed in this paper were the earlier hypotheses about MoE type architectures operating in the brain [6] , and the realization that the circuits involved in ITDP studied in [26 , 29] had exactly the properties required for an ITDP driven gating mechanism that could control ensemble learning , an alternative hypotheses involves a hierarchical STDP only architecture . A multi-layered STDP system where the final layer learns to coordinate the decisions of an earlier layer of classifiers might also provide a mechanisms for effective ensemble learning . The SEM neural network classifiers realize expectation maximization by learning the co-firing statistics of pre and postsynaptic neurons via STDP . The neurons of the input layer represents discrete-valued multidimensional data ( ex . digital pixel image ) using a spike-coded vector , where the value of each dimension is expressed by a group of exclusively firing neurons representing its corresponding states . Since the spike output of a WTA ensemble similarly can be regarded as the binary-coded multidimensional input data for the final layer ( ex . NE dimensional data where the value of each dimension has NC states ) , this naturally leads to the possibility that the latent variable ( hypothesis ) of a given ensemble state can be inferred by the final WTA network using STDP learning instead of ITDP . One difference between the ensemble WTA layer and the input layer during the presentation of input data is that the firing probabilities of WTA neurons are not exclusive for a given input sample ( more than one neuron can have a non-zero firing probability ) , while the population code used in the input layer neurons always have all-or-none firing rates , which means that the state of the given input data is represented stochastically in the WTA layer . Although , as a form of interference , this might inherently affect the behavior of a SEM network , previous work [30] indicates that it should still be able to deal with incomplete or missing data . Possible applications of multi-layered SEM microcircuit were suggested in [30] , and a further study [61] has shown the power of recurrent networks of multiple SEM circuits when used as a neural reservoir for performing classification , memorization , and recall of spatiotemporal patterns . These insights suggest an STDP only implementation of the MoE type architecture presented earlier might be viable . Hence we conducted a preliminary investigation of using STDP to learn the second layer connection weights ( i . e . connections between the ensemble and final layer , Fig 4 ) , making a comparison of the use of STDP and ITDP in that part of the ensemble classifier system . The learning of the second layer of weights by STDP was done straightforwardly by applying the same learning rule as in the first layer connections ( between the input and ensemble layers ) . All other settings and parameters were exactly the same as the original system . In order to avoid the influence of the inevitable trial-to-trial variance of the presynaptic SEM ensemble when the two learning rules are tested separately , the original ensemble network architecture was expanded by having two final ( parallel ) WTA circuits which both receive connections from the same ensemble WTAs , but are subject to different synaptic learning rules ( one for STDP and the other for ITDP ) . This setup , where the learning rules are tested in parallel , ensures that both final layer WTAs receive exactly the same inputs , so that any differences in their final performances depend only on the different synaptic learning rules . For the repeated simulations with the normal Gaussian feature selection scheme , the same initial mean positions were used without the random mean placement ‘jittering’ . This is because the purpose of the current experiment is to compare the two plasticity methods under as identical conditions as possible , and we know from the earlier experiments with the ITDP ensemble that the performance and trends of the fixed normal Gaussians was very close to the average of the randomly jittered placements . These procedures enables a well-defined , unbiased comparison between the two learning rules . The connections from the gating WTA to the ITDP final layer operate exactly as in the experiments described earlier ( i . e . as genuine gating connections involved in the heterosynaptic plasticity process ) . For comparability , the STDP final layer also receives projections from the gating WTA , but they of course operate very differently—they are just like the connections to the final network from any of the ensemble networks . Therefore in the STDP case the gating WTA does not have an actual gating function but effectively operates as an extra ensemble member . The corresponding synaptic weights are learnt by STDP in just the same way as for all other ensemble WTA projections to the STDP final layer neurons . This use of an additional ensemble member is potentially advantageous for the STDP final network in terms of the amount of information used . The results of multiple runs of the expanded comparative architecture on the MNIST handwritten digits recognition task with random feature selection are illustrated in Fig 13 . It is clear from these initial tests that the STDP version compares favourably with the ITDP version , although is generally not as good . The performance of the STDP final WTA over repeated trials shows that on many runs it outperforms most of the ensemble WTAs ( i . e . ensemble learning is successful in this version of the architecture ) . Although the STDP net is capable of bringing improved classification from the SEM ensemble , its performance variance over repeated trials is higher than the ITDP net , indicating less robustness against the various ensemble conditions . However , while the ITDP net is dependent on the gating WTA performance ( as we know from earlier experiments—Fig 7 ) , no single presynaptic WTA circuit strongly influences the STDP net performance . The result of repeated runs sorted by the gating WTA performance ( Fig 13B ) indeed shows this dependency of the ITDP net , and the STDP net outperforms the ITDP net in the region where the gating WTA performances are the worst . However , as was shown with earlier experiments , it is relatively easy to find good initial gating network settings , and it might not be unreasonable to assume these would be partially or fully hardwired in by evolution in an ITDP ensemble . The dependence of ITDP on ( a reasonable ) gating signal may be disadvantageous in terms of the performance consistency in this type of neural system in isolation , and without any biases in initial settings , but on the other hand , the gating mechanism ( which after all is the very essence of the ITDP system ) can act as an effective and compact interface for providing higher control when connected to other neural modules . For example , the supervising signal could be directly provided via a gating network from the higher nervous system , or the gating signal could be continuously updated by reward-based learning performed by an external neural system such as the basal ganglia . Also it is possible that multiple ITDP ensemble modules could be connected such that the final output of one module is fed to the gating signal of other modules ( similar to the multilayered STDP SEM networks ) , achieving successive improvements of system performance as information is passed through modules . Fig 13C and 13D show the performances using a high performing ensemble WTA as the gating WTA which is automatically selected during the early simulation period . The gating WTA was continuously updated during the first round of dataset presentation ( 0 < t < 224 ) by assigning one of the ensemble WTAs as the gating WTA whenever the current gating WTA is outranked by it . This procedure was used , rather than assigning previously found good ( ITDP ) gating network settings , in an attempt not to potentially bias proceeding against STDP by using a network known to be good for ITDP . When the gating WTA is replaced by the selected ensemble WTA , the indices of its neurons representing corresponding classes also changes . Thus the entire set of ITDP weights are automatically re-learnt to new values , which causes the transition in the NCE value of the final WTA until re-stabilization ( the hills in the red line in Fig 13C ) . Indeed , we can see from the results of the more detailed set of comparative experiments shown in Fig 14 that given a qualified gating signal of the kind describe above ( i . e . from a gating network that performs classification reasonably well ) , the ITDP final net consistently and significantly outperformed the STDP final net over a wide range of conditions ( feature selection scheme , ensemble size ) in both training and testing . This was the case even though the STDP net uses one more presynaptic WTA circuit ensemble member , which can be seen to confer an advantage ( first two columns in Fig 14 ) . Clearly , if the gating network was used only in the ITDP case , and the main ensemble was the same size under both conditions , then the ITDP version’s margin of superiority would be increased further . It is interesting to note that the overall trends of the final performances of both methods are similar to each other over the repeated trials in the region of good performance gating WTAs ( the ups and downs of the red and green plots over the trials in Fig 13B and 13D follow each other quite closely ) . There is also a similar dependency of the average performances on the ensemble sizes ( Fig 14 ) , which suggests that there might be some shared underlying mechanisms in both combining methods . In the STDP ensemble , the synapses carrying the presynaptic spikes onto the postsynaptic neurons get enhanced after a few milliseconds of neural firing . Since all the WTA neurons fire highly synchronous bursts of spikes during every input presentation ( the behavior is similar to the clocked output of the abstract voter ensemble model ) , in most cases the last spike of the final WTA burst triggering STDP follows right after the end of the presynaptic bursts . This leads to the synaptic potentiation by STDP reflecting all the presynaptic bursts . Considering the plasticity curves of STDP and ITDP in our model are of a similar type with a few tens of milliseconds of time shift , both plasticities can be generally thought as enhancing the synaptic weight if two neuron co-fire around the peak of the curve , and depressing it otherwise . This insight leads to the hypothesis that the final WTA in the STDP network acts functionally quite similarly to the gating WTA in the ITDP network . Among the presynaptic ensemble WTA neurons , the better performing neurons ( those which fire only under the presentation of a specific class ) will fire more spikes than the worse performing neurons . This is because the neurons of each ensemble WTA typically fire highly regular burst of 3-4 spikes in total . The best performing neurons in the ensemble layer fires all 4 spikes for its corresponding class and remains silent for the other classes . In the poorer performing WTA neurons , more than two neurons will fire 1-2 spikes each , resulting in the dispersion of spikes . Thus , over the course of STDP weight updates , the weights from the better performing presynaptic WTA neurons will get more potentiation ( by summing EPSPs from all 4 presynaptic spikes ) than the connection weight from the more poorly performing neurons ( which typically carry only 1-2 spikes ) . This leads us to infer that the best performing presynaptic WTA neurons under each class presentation generally influence the final WTA most as learning proceeds ( through the Hebbian STDP process ) . This autonomously drives the final WTA towards better performance through increased correlated activity with the ensemble , effectively making it a good ‘gating’ WTA ( or at least ‘guiding’ WTA ) . This ‘guiding’ results in better performance of the combined ensemble output in an analogous way to the explicit gating signals in the ITDP ensemble mechanism . Of course the STDP version requires correlated pre- and post-synaptic firing from the start in order to gain traction , whereas the more direct ITDP version does not require post-synaptic firing . Although this STDP ‘gating signal’ may result in positive feedback of the final WTA behavior , inputs from the other presynaptic neurons always interfere with it , preventing an indefinite increase of system performance . The effect of supervised gating signals shown in Fig 14 indeed shows the difference between the two mechanisms: the STDP final net has increased performances driven by the supervised signal from one of the presynaptic WTAs during the training phase , but its performance drop is much larger than for the ITDP final net in the test phase after the supervised signal is removed . In particular , the odd dependence of the STDP net on ensemble size in the stretched Gaussian selection case ( where performance decreases with ensemble size in the training phase , instead of increasing as in all other cases , and the discrepancy with the test phase is particularly marked: Fig 14 bottom of 3rd column ) indicates the possibility of a negative effect of the supervised signal when the ensemble size is small , where the training result can be deceptive because of the large influence of the supervising signal on the final WTA relative to the inputs from the rest of the presynaptic WTAs . By contrast the explicit gating signal in the ITDP system is more stable and less prone to such effects , providing better overall performance .
The main aim of this paper was to explore a hypothesized role for ITDP in the coordination of ensemble learning , and in so doing present a biologically plausible architecture , with attendant mechanisms , capable of producing unsupervised ensemble learning in a population of spiking neural networks . We believe this has been achieved through the development of an MoE type architecture built around SEM networks whose outputs are combined via an ITDP based mechanism . While the architecture was successfully demonstrated on a visual classification task , and performed well , our central concern in this paper was not to try and absolutely maximize its performance ( although of course we have striven for good performance ) . There are various methods and tricks from the machine learning literature on ensemble learning that could be employed in order to increase performance a little , but a detailed investigation of such extensions is outside the scope of the current paper , making it far too long , and some would involve data manipulation that would move the system away from the realms of biological plausibility , which would detract from our main aims . However , one interesting direction for future work related to this involves using different input data subsets for each ensemble member . This can increase diversity in the ensemble which has been shown to boost performance in many circumstances [18 , 49] , a finding that seems to carry over to our spiking ensemble system according to the observations on diversity described in the previous section . Preliminary experiments were carried out in which each SEM classifier was fed its own distinct and separate dataset ( all classifiers were fed in parallel , with an expanded , separate set of input neurons for each classifier , rather than them all using the same ones as in the setup described earlier in this paper ) . A significant increase in the overall ensemble performance after training was observed as shown in Fig 15 . Further work needs to be done to investigate the generalization of these results and to analyse differences in learning dynamics for the ensemble system with single ( one set for all classifier ) and multiple ( different sets for each classifier ) input data sets . The issue of how such multiple input data sets might impinge on biological plausibility must also be examined . A related area of further study is in applying the architecture to multiple sensory modes , with data from different sensory modes feeding into separate ensemble networks . Some of the biological evidence for ensemble learning , as discussed in the Introduction section , appears to involve the combination of multiple modes . Although we have tested the architecture using a single sensory mode , there is no reason why it cannot be extended to handle multiple modes . While our SEM ensemble model mimics the general MoE architecture , the overall process is not identical to that used in the classic MoE system [18 , 24] . A key difference is that the operation of the SEM gating WTA on the ensemble outputs is not based on immediate training input but is accumulated by slow additive synaptic plasticity over a relatively long time scale , whereas the standard MoE gating network instantaneously changes the strength of ensemble efferents for each input vector . Therefore our spiking system is not as adept at directly and quickly filtering out the wrong output from the ensemble WTAs when an output neuron in the ensemble fires for multiple classes . In this case the false spikes are also passed to the final layer through the enhanced connections . However , because such a neuron has a higher probability of firing for multiple classes , it dissipates its synaptic resource over multiple efferent connections , resulting in lower synaptic weights than in the case of a neuron which fires predominantly for one class . Hence the neuron that fires for multiple classes has less chance of winning at the final output layer WTA . Similarly , false spikes from the gating WTA will result in less chance of enhancing the corresponding target set of ITDP weights because of timing mismatch . In this way our spiking ensemble system can effectively filter out these false classifications , but using different learning dynamics from the classical system . However , if a large number of ensemble WTAs fire equally wrongly for the same class , the final output develops a higher chance of generating the wrong output . The standard architecture can of course suffer from the same problem [18 , 49] . This can happen , for instance , when two input images are hard to discriminate ( such as the digits 3 and 8 ) , even if the input subfeatures are randomly selected . Therefore the system is not entirely free from the feature selection problem as experienced in other ensemble learning methods . This limitation meant that in such circumstances simulations using high ensemble sizes did not significantly improve the overall performance ( Fig 11 ) , indicating a lack of ensemble diversity . Preliminary experiments indicated that by using an evolutionary search algorithm to evolve individual feature selection schemes for each ensemble member , diversity is increased , alleviating this problem greatly and significantly increasing performance . This is reminiscent of individually evolved receptive fields/input ‘features’ for spatially separated networks in the cortex and other areas . Future work will explore this issue more thoroughly . An interesting extension is the possibility of a form of evolutionary search being neuronally integrated into the current architecture [62] so that feature selection is performed in parallel with the other plastic processes , becoming part of the overall adaptation . The empirical work on which we base our ITDP model [26 , 29] was conducted in vitro . While this was of course because of the technical difficulty of conducting such research in vivo , it should be noted that work by Dong et al . ( 2008 ) [35] suggests that in some circumstances there can be activity dependent differences in the dynamics of heterosynaptic plasticity operating in vivo . While Dong et al . were looking at heterosynaptic plasticity in the hippocampus , they were not studying ITDP as defined in [26] and they were observing quite different neural pathways from Dudman et al . ( specifically , Dong’s system involved Schaffer and commissural pathways , crucially without the different proximal and distal projections onto CA1 found in Dudman’s system , from EC and CA3 respectively—instead the two CA1 inputs are both from CA3 ) . However , Dong et al . ( 2008 ) [35] made the interesting finding that in the system they were studying , in vivo , coincident activity of converging afferent pathways tended to switch the pathways to be LTP only or LTP/LTD depending on the activity states of the hippocampus [35] . If such findings extended to the system we have based our learning rule on , then of course our hypothesis would have to be revised . We are working under the assumption that the behaviour is stable in vivo . Recently Basu et al . ( 2016 ) [29] have provided some indirect evidence that the ITDP behaviour of the particular circuits we are basing our functional model on does hold in vivo . They cite studies of the temporal relation of oscillatory activity in the entorhinal cortex and the hippocampus in vivo that suggest that the disinhibitory gating mechanism enabled by the LRIPs may indeed be engaged during spatial behavior [63 , 64] and associational learning [65] . For example , during running and memory tasks , fast gamma oscillations ( 100Hz ) arising from EC are observed in CA1 and precede the slow gamma oscillations ( 50Hz ) in CA1 , which are thought to reflect the CA3 pyramidal neuron input [63] . Crucially , EC-CA1 gamma activity and CA3-CA1 gamma activity display a 90° phase offset during theta frequency oscillations ( 8 to 9Hz ) [63] which is consistent with a 20-25ms time delay . However , since any ensemble learning of the kind we have presented here would be part of a wider ‘cognitive architecture’ , it is interesting to speculate that some activity dependent influence on the dynamics of such learning might occur in the bigger picture ( e . g . moderating or switching on/off ensemble learning in certain conditions ) . For reasons discussed earlier in this paper , ITDP seems a very good candidate for involvement in a biological mechanism ideal for combining ensemble member outputs , but it was naturally interesting to also attempt an all STDP implementation . Although we had imagined interference effects would compromise its learning ability , this version of the architecture performed surprisingly well . When the gating network performed relatively poorly , the STDP version compared very favourably with the ITDP version . However , with good ( or at least reasonably ) performing gating networks the ITDP version was significantly better over all conditions . This highlighted the dependence of the ITDP architecture on a gating network that achieves reasonable performance in agreement with the similar findings from the initial more abstract ( voter ) model . This shows that there is a small price to pay for the advantage the ITDP process confers , namely that it strengthens connections without a need for the corresponding final output neuron to be firing , thus providing a strong guiding function . The various methods for reducing this reliance ( or at least ensuring the gating performance is always reasonable ) that were outlined in the previous section will be the subject of future work . Preliminary analysis , as discussed in the previous section , suggests that there are some very interesting parallels between the ways the successful ITDP and STDP architectures operated , notably that the best performing ensemble WTA neurons in the STDP version had a guiding role functionally similar to that of the gating network in the ITDP version . While the differences and commonalities between ITDP and STDP dynamics in combining ensemble classifiers were briefly discussed in relation to the preliminary experiments , a more thorough comparative analysis of the effects of various conditions on both learning schemes will be addressed in the future work . Certainly the ITDP vs STDP work undertaken so far suggests that STDP-only architectures are another plausible hypothesis for ensemble learning in populations of biological spiking networks . Lateral inhibition in the SEM networks—which provides the competition mechanism in the WTA circuits—is modeled as a common global signal that depends on the activity of the neurons in the network [30] . This effectively models a form of strong uniform local lateral inhibition as widely experimentally observed in the cortex [66 , 67] . This inhibition mechanism is a core part of the SEM network dynamics and reflects the fact that they are small locally organised networks . We assume multiple such networks act as the ensemble members in our architecture . However , it might be possible to model the ensemble layer by a bigger single group of neurons which inhibit each other according to a ‘Mexican hat’ type function . Since with this form of inhibition ( which is also commonly assumed [68] ) the effect drops off with distance , with strong interaction among nearby neurons , a set of overlapping networks could emerge that function similarly to a smoothed version of multiple WTA circuits . Dealing with arbitrary ( unknown ) numbers of classes with our ITDP ensemble architecture in a general unsupervised manner is a challenging future direction , although an individual SEM network with a sufficient number of output neurons has been shown to perform unsupervised clustering of a given dataset to some extent [30] . It might be possible to employ a higher control to vary the number of classes in a supervised way as shown in [72] . More preferably , the smoothed version of a lateral inhibition mechanism using the Mexican hat topology may be capable of dealing with unknown numbers of classes in a more biologically plausible way by incorporating more sophisticated synaptic and neuromodulatory mechanisms . The novel architecture presented here demonstrates for the first time how unsupervised ( or indeed any form of ) ensemble learning can be neurally implemented with populations of spiking networks . Our results show that the ensemble performs better than individual networks . The lack of diversity within the population , which sometimes becomes apparent , will be tackled in the next phase of work as outlined above . It is also possible that the relative strength of the ensemble method , in terms of efficiency of learning , might change when reducing the time spent on learning in the SEM networks ( i . e . there may be an interesting resource/performance trade-off ) . This issue will also be investigated .
The detailed methods for the full SEM-ITDP ensemble architecture are as follows . | Ensemble effects appear to be common in the nervous system . That is , there are many examples of where groups of neurons , or groups of neural circuits , act together to give better performance than is possible from a single neuron or single neural circuit . For instance , there is evidence that ensembles of spatially distinct neural circuits are involved in some classification tasks . Several authors have suggested that architectures for ensemble learning similar to those developed in machine learning and artificial intelligence might be active in the brain , coordinating the activity of populations of classifier circuits . However , to date it has not been clear what kinds of biologically plausible mechanism might underpin such a scheme . Our model shows how such an architecture can be successfully constructed though the use of the rather understudied mechanism of input timing dependent plasticity ( ITDP ) as a way of coordinating and guiding the activity of a population of model cortical microcircuits . The model is successfully demonstrated on a visual classification task ( recognizing hand written integers ) . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"learning",
"action",
"potentials",
"medicine",
"and",
"health",
"sciences",
"neural",
"networks",
"nervous",
"system",
"membrane",
"potential",
"social",
"sciences",
"electrophysiology",
"neuroscience",
"learning",
"and",
"memory",
"synaptic",
"plasticity",
"cognitive",
... | 2016 | Unsupervised Learning in an Ensemble of Spiking Neural Networks Mediated by ITDP |
The nematode Caenorhabditis elegans is a popular model system in genetics , not least because a majority of human disease genes are conserved in C . elegans . To generate a comprehensive inventory of its expressed proteome , we performed extensive shotgun proteomics and identified more than half of all predicted C . elegans proteins . This allowed us to confirm and extend genome annotations , characterize the role of operons in C . elegans , and semiquantitatively infer abundance levels for thousands of proteins . Furthermore , for the first time to our knowledge , we were able to compare two animal proteomes ( C . elegans and Drosophila melanogaster ) . We found that the abundances of orthologous proteins in metazoans correlate remarkably well , better than protein abundance versus transcript abundance within each organism or transcript abundances across organisms; this suggests that changes in transcript abundance may have been partially offset during evolution by opposing changes in protein abundance .
The rapid lifecycle , small size , reproducible development , and ease of cultivation in the laboratory have made Caenorhabditis elegans an important experimental system for biological studies . Numerous human disease-related genes ( e . g . , related to cancer or neurological diseases ) have orthologs in the worm [1] . Sequencing and annotation of its genome has revealed more than 19 , 000 genes [2] coding for more than 22 , 000 proteins , including splice variants . Extensive systematic studies of gene function have been performed . However , to completely understand complex biological processes such as development , aging , or disease , the analysis of the proteome—i . e . , the entire set of the expressed proteins—is becoming increasingly important . Knowledge of the complete sequence of a genome is a necessary prerequisite for proteomics , but the DNA sequence itself does not reveal which proteins are actually expressed when , where , and to what level . Furthermore , in contrast to the genome , the proteome is changing under different biological conditions . Although for many years , transcriptome data ( i . e . , the collection of transcribed mRNAs ) has been used to approximate the proteome , a number of studies have demonstrated that the correlation between mRNA and protein abundance is surprisingly low [3–5] because of posttranscriptional regulation and variable protein half-lives . The analysis of the proteome is therefore a key method to provide systems-level information about protein function in time and space , and to obtain a concise view of biological processes . In the case of C . elegans , previous analyses of the proteome were either limited in scope and coverage [6 , 7] , or largely focused on improving genome annotation [8] , with the biggest C . elegans proteome dataset published so far encompassing 6 , 779 proteins [8] . To generate a comprehensive , deeply sampled C . elegans proteome database that can be used for quantitative proteome analysis , we applied subcellular and biochemical fractionation methods to the worm proteins , performed tryptic digests , separated the resulting peptides using a variety of techniques , and identified the peptides by mass spectrometry ( MS ) . This resulted in a unique global view on the expression status of the C . elegans proteome . We identified a number of protein features and functions that are underrepresented in the expressed proteome , likely representing specialized functional systems expressed only in a small subset of cells and/or developmental stages . We demonstrate the importance of proteomics data towards improved genome annotation . Finally , we compared the proteome data with similar data from the fruit fly Drosophila melanogaster . The latter comparison provided—for the first time to our knowledge—an overview of the expressed “core animal proteome , ” which should arguably become the initial focus for monitoring the basic metazoan cellular machinery in the future .
To identify C . elegans proteins , we collected worms at various developmental stages and homogenized whole animals and eggs to isolate the proteins . Their tryptic peptides were separated using strong cation exchange chromatography ( SCX ) , in several cases after labeling them with isotope-coded affinity tags ( ICAT ) [9] to reduce sample complexity , or by isoelectric focusing ( applying free-flow electrophoresis and immobilized pH gradient strips ) . The peptides were finally identified using microcapillary liquid chromatography–electrospray ionization–tandem MS ( μLC-ESI-MS/MS ) . With this extensive shotgun proteomics approach , we identified 10 , 977 different proteins , including splice variants , via 84 , 962 nonredundant peptide identifications ( Table S1; 759 , 320 peptide identifications were obtained in total ) . We identified 10 , 631 gene loci , corresponding to 54% of the gene loci in WormBase ( WS140: 19 , 735 loci ) . Of these , 7 , 476 loci ( 38% ) were detected via several distinct peptides , 580 ( 3% ) were detected via the same peptide more than once , and 2 , 575 ( 13% ) were detected only via a single peptide identification ( Figure 1 ) . When considering individual annotated exons ( irrespective of their various splicing contexts ) , our peptide data covered 28 . 2% of the 129 , 047 exons contained in WormBase . Protein identification from MS peptide spectra is prone to false-positive assignments , and we employed strict search cutoffs using PeptideProphet ( see Materials and Methods ) . To independently estimate our false discovery rate ( FDR ) , in particular for identifications based on a single peptide spectrum ( “single hits” ) , we first took advantage of one of our experiments that used isoelectric focusing to fractionate peptides . In each peptide fraction , true-positive identifications should scatter around a narrow range of isoelectric points ( pIs ) , whereas false-positive identifications should follow the background distribution in the database . This analysis , using computational predictions of pIs to check all peptides , yielded an estimated FDR of 35% for single hits in this particular experiment . Independently , a newly developed model based on a robust decoy search strategy yielded an upper limit for the FDR of single-hit identifications at around 63% for all combined experiments ( L . Reiter , M . Claassen , S . P . Schrimpf , J . M . Buhmann , M . O . Hengartner , et al . , unpublished data ) . By the latter method , multi-hit identifications were found to be much more reliable , resulting in an FDR of 7% in our study . Since almost half of all single-hit identifications do represent bona fide protein identifications , we chose to include single-hit identifications in our subsequent analyses . A separate analysis focusing on just these proteins alone showed that they often belonged to groups that were underrepresented in the complete dataset and are therefore presumably of low abundance in C . elegans ( short , uncharacterized proteins and in particular those with seven transmembrane domains; also see below ) . This means that they do represent valuable information about which proteins are expressed at low levels in C . elegans . It should also be stressed that all conclusions reported below remained valid when single-hit identifications were excluded . To assess whether proteins from sources other than C . elegans were present in our preparations , we focused on the bacteria on which the worms were feeding ( Escherichia coli ) . We tested a single , representative experiment , encompassing 67 MS/MS analyses by searching the spectra against a combined C . elegans and E . coli database . A total of 1 . 3% of the protein identifications mapped to E . coli , among them 14 hits mapping to both organisms . However , for each of these 14 proteins , there was at least one additional C . elegans peptide identified , confirming that these overlapping detections did not influence the C . elegans results . In order to characterize C . elegans proteins that were not detected , and that are therefore most likely expressed at particularly low levels , or in specialized cells or developmental stages only , we classified the entire predicted C . elegans proteome with respect to several aspects ( length , pI , hydrophobicity , transmembrane topology , and functional annotation ) . This should reveal the nature of underrepresented proteins ( with potentially more peripheral , or even worm-specific functions ) , and separate them from abundant proteins involved in basic cellular processes such as growth , metabolism , and information processing . It should also reveal potential technical limitations ( proteins/peptides difficult to detect using our procedure ) , which is important to assess for future systematic uses of MS . Our bias analyses revealed an underrepresentation of proteins shorter than 400 amino acids ( Figure 2A ) and of proteins with basic pIs ( Figure 2B ) . A similar bias has previously been observed for D . melanogaster [10] . The underrepresentation of basic proteins was partly to be expected , due to our isoelectric focusing experiments , which centered on the pH range 3–7 . The underrepresentation of short ( low molecular weight ) proteins might be caused by a generally higher prevalence of spurious gene predictions among short genes , and also by a lower probability of detecting one of the few tryptic peptides generated by short proteins . We observed a bimodal distribution of hydrophobicity values within the annotated set of all C . elegans proteins , and a strong underrepresentation of proteins in the second , high hydrophobicity peak in our dataset ( Figure 2C ) . This second peak consists mostly of multipass transmembrane proteins ( ∼64% of these proteins have seven or more predicted transmembrane domains ) . To better understand how membrane association relates to protein abundance and detectability , we globally characterized WormBase proteins with respect to their content of transmembrane segments , using Phobius [11] . Overall , we found a notable underrepresentation of transmembrane proteins in our proteomics data , and decided to subdivide these proteins further according to the number of transmembrane sections and annotated functions as shown for other species [12 , 13] ( Figure 2D and 2E ) . Remarkably , we found that the strongest underrepresentation is observed for proteins with seven transmembrane regions , in particular those annotated with the function “receptor activity . ” This may point to a biological ( rather than technical ) explanation for the relative paucity of transmembrane proteins in our data: Seven-transmembrane chemosensory receptors are widespread in the C . elegans genome , but many of these are known to be expressed only in a small number of neurons each [14–16] . Because we assessed whole animals , those proteins might be too rare to be successfully detected . This general underrepresentation in our proteome data suggests similar sensory functions for other transmembrane proteins of hitherto unknown function that we also found to be of too low abundance to be detected . Finally , we globally analyzed the functional classifications of all the detected proteins . We observed a clear bias towards proteins with known functions . The same bias was also observed for the D . melanogaster proteome [10] . A possible explanation could be that some of the undetected proteins with unknown functions are actually erroneous gene predictions or pseudogenes . It could also be a testament to the biases of previous studies: abundant proteins are easier to work with biochemically , and may therefore have obtained a functional annotation more easily . In total , our proteomics approach identified proteins belonging to 125 out of the 127 Gene Ontology ( GO ) slim categories defined for WormBase . The global GO slim analyses confirmed the underrepresentation of proteins with receptor activity mentioned above , and of “membrane” or “integral to membrane” proteins in general ( Figure 2F ) . Large-scale proteome analyses ( such as ours ) represent an important cornerstone for an improved genome annotation . In WormBase ( WS160 ) , 4 , 987 gene loci were still listed with the gene status “predicted” only , i . e . , without any supporting transcript data ( expressed sequence tag [EST] , mRNA ) . We experimentally confirmed the protein expression of 1 , 062 of these predicted genes ( among them , more than 40% via multiple peptide detections ) . As was the case for the whole proteome , this subset was enriched for proteins with GO slim annotations ( 45% in our dataset , as compared to 38% expected for this subset in WormBase; p-value: 4 . 65E−08 ) . Apart from these gene confirmations , our C . elegans proteomics dataset contains numerous spectra originating from nonannotated regions in the worm genome . In computationally intensive analyses , we are identifying these by searching our data against six-frame translations of the genome , and filtering the results for high confidence spectra that map to nonannotated regions . For example , from one particular experiment , we identified 78 likely novel peptides . Two of these are illustrated in Figure 3 ( the corresponding MS/MS spectra are provided as Figure S1 ) . These data suggest an alternative translational start site for the protein SYN-4 ( T01B11 . 3; Figure 3A ) : the observed peptide is located upstream of the annotated translational start site , and only partially overlaps with the currently annotated protein sequence . The second example demonstrates a novel splice variant for the gene F47B7 . 7 ( Figure 3B ) . In this case , we identified a peptide that extends an existing annotated exon downstream , in the correct frame . These and similar analyses , suggesting altered or new gene models , are computationally very intensive and were not yet completed at the time of submission . Furthermore , due to the increased search space when searching proteomics against the genome , extra scrutiny is needed when interpreting each reannotation instance , and additional experimental data should probably be taken into account before fully accepting these gene annotation changes . C . elegans and its relatives are unique among characterized metazoans in that a large number of their genes are organized into operons ( multicistronic transcription units , containing up to eight genes that are strictly coexpressed [17 , 18] ) . Following transcription , the primary transcript is split up through a unique trans-splicing mechanism , and the individual open reading frames are subsequently translated separately into distinct mature proteins . In order to assess the potential influence of operon structure on the regulation and abundance of proteins , we studied the expression status of genes in operons , compared to individually transcribed genes . Although an absolute quantification of protein levels is not possible with our shotgun approach , we performed a semiquantitative analysis based on spectral counting [19–23] . Surprisingly , we observed that proteins encoded by operons are expressed far more strongly than those encoded by individually transcribed genes: we observed 84% of the former , with a median relative abundance of 20 ppm ( parts per million of total protein molecules ) , but only 47% of the latter with a median relative abundance of 5 ppm ( Figure 4A ) . The same tendency was found when analyzing publicly available transcript-abundance data ( Figure 4B ) . This striking observation confirms that operons are preferentially made up of genes that are strongly transcribed , and we now establish that this is reflected also at the protein level: operon proteins , on average , are more than 3-fold more abundant than proteins from single-gene transcripts . Apart from grouping strongly expressed proteins , operons are also expected to facilitate coordinated regulation of their constituent genes . We assessed whether this is the case by searching for operons that were either fully expressed ( i . e . , all encoded proteins detectable ) or silenced ( none or very few of the encoded proteins detectable ) . Indeed , we found significantly more operons of both types than expected by chance , as illustrated for operons of lengths 4 to 6 ( Figure S2 ) . In principle , our observations could be stemming from a limited selection of tissues only , for example from the hermaphrodite germline , where operons are thought to be strongly expressed during oogenesis [24] . However , we observed that operon proteins are more abundant even in dauer and L1-stage larvae , which both should have very little germline material ( Figure S3 ) . We further checked whether our observation could be explained by systematic differences in length or transmembrane segments of operon proteins . Although we did observe slight differences in length and transmembrane content—operon proteins are on average 11% longer , and transmembrane proteins are 40% less frequent—these differences were not sufficient to explain the increased abundance of operon proteins ( unpublished data ) . Together , our observations indicate , for the first time , that operons in C . elegans ensure the coordinated regulation of highly expressed proteins . In this study , we had the unique opportunity to compare large-scale proteome datasets from two different animal species , owing to the recent publication of the D . melanogaster proteome [10] ( http://www . mop . uzh . ch/peptideatlas/; previous work in D . melanogaster had mainly focused on protein–protein interactions or subproteomes only [25 , 26] ) . We performed spectral counting for both organisms to obtain semiquantitative measurements of protein abundance , and compared these to published mRNA expression data derived from Affymetrix [27 , 28] and serial analysis of gene expression ( SAGE ) platforms [27 , 29] . In the C . elegans and D . melanogaster proteomes , 2 , 695 pairs of orthologs were identified for which all three types of data were available . Surprisingly , we observed that orthologs showed a strong correlation in protein abundances across the two organisms , despite more than 600 million years of separate evolution ( Spearman rank correlation RS = 0 . 79; Figure 5A ) . Notably , this biological correlation at the protein level between the two species is even higher than the within-species correlation between protein and transcript abundances ( within C . elegans: RS = 0 . 59 and 0 . 44 for protein-Affymetrix and protein-SAGE , respectively; within D . melanogaster: RS = 0 . 66 and 0 . 36 , respectively ) . In contrast to the protein-level correlation , the abundance correlations at the transcript level between the two species were also rather low ( Figure 5B ) . Interestingly , the overall protein-abundance correlations are not equally tight across functional categories: the highest correlation was observed for the functional category “translation” ( RS = 0 . 93 ) and the lowest for the category “regulation of biological process” ( RS = 0 . 65 ) . Despite the fact that it is difficult to compare tissues and developmental stages across organisms , our analysis provides a first insight into the evolutionary behavior of animal proteins over long time scales . It is important to point out that for all six data points , several developmental stages and/or tissues had been mixed , but that these were not , of course , always directly equivalent and comparable between the two organisms . However , many of the ancient animal orthologs that we studied here can be expected to be expressed similarly across many cell types and stages , and we thus attempted to capture an organism-wide “average” proteome for both animals . That notwithstanding , we also repeated the analysis for one set of samples that is arguably more directly comparable: mixed staged embryos sampled in both D . melanogaster and C . elegans at the proteome and at the transcript levels ( Figure S4 ) . Here again , we saw that protein abundances correlated far better ( RS = 0 . 70 ) across organisms than transcript abundances ( RS = 0 . 50 ) . Another potential complication for our analysis lies in the technique of spectral counting . Individual tryptic peptides are known to ionize and be detected with widely differing efficiencies in mass spectrometers . Although protein conservation between C . elegans and D . melanogaster is low ( ∼40% sequence identity ) , a higher-than-expected abundance correlation might still result if equivalent peptides in both organisms were correlated in their suitability for MS . We assessed the extent of this effect by making the spectral counts independent: For any given section in the alignment of two orthologs , only one of the proteins was allowed to generate peptide counts; these sections were alternated across the length of the alignment , effectively reducing the data by half . As expected , this lowered the abundance correlation , but not by much ( RS = 0 . 68 ) . Importantly , the resulting correlation is still much higher than the correlation of transcript abundances across organisms ( Figure S5 ) . Since one of our original interests was to characterize the “core animal proteome , ” we also analyzed lower-coverage datasets from two additional organisms: Saccharomyces cerevisiae [30] and Mus musculus ( [21] ) ( for the latter , we additionally included plasma data from PeptideAtlas; http://www . peptideatlas . org/ ) . Comparative proteomics using multiple organisms has recently become popular , for example in bacteria [31] , but has not yet been possible for animals . We searched for groups of orthologs that were detected in all four organisms; these would constitute the universally detectable eukaryotic proteome core . We found 847 such proteins , mostly from information-processing and metabolism genes . Conversely , we found 1 , 287 proteins to be detectable in all three animals , but not in yeast . This latter set might be considered the specific core of multicellular animal proteomes . However , it is clear that neither of these sets is complete , as of yet , mostly due to low coverage in mouse . Our protein-abundance estimates from two organisms also allowed us to study in more detail the fate of duplicated genes . Here , of particular interest , are cases in which a gene family has duplicated in one lineage ( fly or worm ) , but not in the other . It is known that long-term retention of duplicated gene copies requires neo- or subfunctionalization [32–34] , but it is unclear what consequences this has for overall protein-abundance levels . We found that when averaging over all cases of lineage-specific gene duplications ( Figure S6 ) , the abundance of duplicated genes is significantly lower than that of their nonduplicated counterparts in the other lineage . Strikingly , however , when all the duplicated genes of a given gene family are pooled , they tend to add up again to the original abundance of the nonduplicated counterpart ( Figure S6 ) .
We describe here a comprehensive inventory of C . elegans proteins , the functional characterization of this inventory , and the first-ever comparison of two such inventories between two model animals ( “comparative proteomics” ) . Although some subsets of the proteome are more difficult to analyze ( e . g . , the membrane compartment ) , we achieved a relatively thorough representation of the genome , where the major exceptions can be explained biologically . For example , the systematic underrepresentation of seven-transmembrane proteins appears to be caused mainly by G protein–coupled receptors . The putative chemoreceptor gene families in C . elegans encompass about 7% of its total genome [35] , and many are thought to be expressed only in a few neurons each [14–16] . Despite their generally low abundance , we did identify 172 seven-transmembrane receptor proteins , showing that they are , in principle , amenable to high-throughput MS analysis ( this is relevant , for example , for screens of putative therapeutic targets ) . We also demonstrated that a whole-proteome analysis of a model organism can contribute to an improved genome annotation . First , we experimentally confirmed the expression of 1 , 062 predicted genes for which no transcript data were available , but for which our proteome data allowed the extraction of a first rough expression pattern . Second , we identified novel peptides from spectra that could not be matched to annotated gene models , suggesting a way to more precisely map open reading frames and splice isoforms to the genome . With respect to genome organization , we found that , in C . elegans , genes in operons are far more consistently and more strongly expressed than individually transcribed genes . In principle , this observation could be an artifact of genome annotation—if a disproportionally large number of annotated nonoperon genes were misannotations that are biologically meaningless . This is highly unlikely , however , since more than 6 , 000 such misannotations would be needed to reconcile the observed differences . Instead , it is likely that operons in C . elegans indeed serve to group strongly expressed genes into coregulated transcription units . Another question that arises is whether these genes were highly expressed even before they were grouped into operons , which would hint at a possible selective advantage for the grouping ( e . g . , to enable more efficient , more reliable , or more uniform transcription of genes whose products are in high demand ) . This is difficult to address conclusively , but our comparison to D . melanogaster provides some information: we observe that orthologs of operon genes are more strongly expressed even in the fly ( Figure S7 ) , where they are not arranged in operons nor are even neighbors on the genome . If one assumes that the operons in C . elegans are the derived state , then the corresponding genes were indeed already strongly expressed before they formed operons . The comparison of our data to the D . melanogaster proteome also sheds some light on an important evolutionary puzzle , namely the surprisingly low correlation between mRNA expression levels of orthologous genes across animal species [36 , 37] , despite evidence for strong stabilizing selection against expression changes in experimental evolution [38] . We found that the abundances of orthologous proteins from worm and fly correlate well ( RS = 0 . 79 ) , far better than the corresponding abundances of mRNA transcripts ( RS < 0 . 50; Figure 5B ) . There are several possible explanations for this finding: First , sweeping changes within the transcriptional machineries in one or both organisms could have resulted in global differences in transcript abundance , whereas selection would have kept protein abundances at least partially stable . One candidate for such a mechanistic change could be , for example , the unique trans-splicing mechanism of nematodes . A second possible explanation might be that posttranslational regulation may have changed systematically , for example due to differences in developmental strategies , physiology , or life styles of the two animals . Here , possibly relevant changes include the fixed cell lineage of nematodes , differences in reproductive strategies , increased endurance in nematodes ( dauer stage ) , or the constraints imposed on D . melanogaster because of its need for metamorphosis and its higher motility ( flight ) . However , in our view , the most parsimonious explanation might be that many changes in the transcriptome might be neutral , or at least nearly neutral [36] . Ultimately , it is the protein levels that are under selection . Protein levels are not only determined by mRNA abundance , but are equally affected by translation efficiencies , protein half-lives , and other factors . Genetic mutations resulting in small changes on any of these levels might persist for some time in a population , as long as their fitness effects are small ( around 1/[2Ne] or less ) . This might be sufficient time to allow for compensatory mutations either in the same gene or elsewhere in the genome , which would reconstitute optimal protein abundance through action on the same or another factor that influences protein abundance . Thus , changes in mRNA expression could be offset by opposite changes in translation rate or protein half-life , and vice versa . Over evolutionary time scales , such small changes may accumulate , resulting in appreciable changes of mRNA abundance , whereas protein abundance would remain roughly constant . This model is a generalization of the concept of compensatory mutations that explains the rapid divergence of some cis-regulatory nucleotide sequences despite the maintenance of stable transcript levels [39] , or the conserved expression of assembled protein complexes despite variable expression patterns of their individual components [40] . The presence of several interacting levels of protein-abundance regulation also would explain another two of our observations: a wide variance of the number of mature proteins per transcript , and a correspondingly low correlation between protein and transcript abundance within an organism ( interestingly , the latter correlation is quite similar between our C . elegans data and data published in yeast [41] [RS = 0 . 57] ) . Our data , in principle , provides an opportunity to study transcript features that would directly influence the ratio of proteins per transcript ( and thereby potentially uncover novel mechanisms of translational regulation ) . However , when checking the influence of transcript length , GC content , or UTR length , we failed to detect correlations with protein/transcript ratios ( unpublished data ) . We did observe a weak , but significant , positive correlation of our protein/transcript ratios and experimental protein half-life measurements of orthologous proteins in yeast [42] ( unpublished data ) , suggesting that protein stability is indeed one of the factors determining the steady-state protein/transcript ratio . We note that the most abundant proteins ( often found in central pathways like energy metabolism or protein synthesis ) also tend to be the ones that show the best abundance correlation between species . This may simply be the case because of a greater relative measurement accuracy for abundant proteins . However , highly expressed genes are also more likely to be housekeeping genes [43] , and may thus be more likely to be under the same evolutionary pressures in different organisms . Strong and constant stabilizing selection is also consistent with our observation that amino acid sequences of more highly expressed proteins evolve more slowly ( Figure 5C ) , mirroring the analogous observation for mRNA expression data [44] . When we stratify proteins by functional categories , we find that those involved in translation and in core metabolism are those with the most highly correlated abundances across species . These functional groups are also those where the coexpression between pairs of transcripts is most highly conserved across species [45] . Furthermore , the same categories also tend to show the best correlation within each organism , with respect to rank-correlation between transcripts and proteins ( Table S2 ) . We also find that the correlation between transcript and protein levels is particularly poor for genes that are presumably heavily regulated ( the categories “signal transduction” or “transcriptional regulation” ) , arguing for abundant posttranscriptional regulation in these functional classes . Proteins differ not only in their mean abundance , but also in the variance of this abundance among individuals ( “noise” ) [46] . Interestingly , whereas yeast proteins involved in translation also show low levels of noise [47] , other groups of proteins found here to be conserved in their abundance between species ( e . g . , protein metabolism ) are characterized by high protein expression noise [47] . Thus , it appears that abundance fluctuations on short time scales ( within populations ) are partially decoupled from fluctuations on long time scales ( between species ) . However , as natural variation is the substrate of evolutionary change , we expect that changes in mRNA levels via compensatory mutations may occur faster in proteins that exhibit higher levels of noise; this remains to be tested in future studies . Our comparative analysis underlines clearly the necessity and usefulness of quantitative proteome analyses , since these better reflect the abundance of the actual effectors of biological processes . Most likely , the actual conservation of protein levels is even higher than what we report here , due to the shortcomings of a simple spectral-counting procedure . In fact , comparisons across organisms might generally provide a good test scenario to improve spectral-counting algorithms or other proteomics algorithms: the higher the abundance correlation , the more precise the measurements ( due to the high number of data points , and due to the quickly changing positions of tryptic cleavages , this is difficult to “over-train” by choosing biased parameters ) . With respect to the transcriptomics datasets that we used , the above test argued for a better quality of the Affymetrix data , as compared to SAGE , because the latter were seen to correlate less well across organisms . This is intriguing , and it may point to additional biases in the SAGE procedure ( for example , due to the added molecular biology steps of cleavage and ligation ) [48] . For those instances where orthologs were not found to be of similar abundance , one can speculate that this difference reflects differing roles ( or even molecular functions ) of the orthologs . Thus , these proteins are of particular interest when studying the evolutionary differences between species . Alternatively , differences in technical aspects for particular proteins might occur , such as shifted or absent trypsin cleavage sites or differences in protein solubility . Interestingly , we did not lose the observed interspecies correlation even for quite low-abundance proteins such as those involved in signal transduction ( our measurements have a dynamic range of more than three orders of magnitude ) . This means that low-abundance measurements are still quantitative , at least to some degree . In our analysis of gene families with lineage-specific duplications , we found that duplicated proteins generally have lower abundance than their nonduplicated counterparts , whereas the summed abundances per gene family remained roughly constant . This finding might be most parsimoniously explained by a prevalence of subfunctionalization among duplicated genes , although it is also consistent with other scenarios ( e . g . , complementarity of tissue expression domains , functional fine-tuning , or subfunctionalization followed by neofunctionalization [49] ) . Of course , protein abundances alone cannot directly inform us about any changes in the functions of duplicated genes . However , our finding does suggest that cases where an increased demand for protein product would provide the sole driving force behind gene copy retention are probably rare . With our dataset , we established an inventory of where and how proteins of interest can be specifically accessed using MS . It enables the generation of a proteotypic peptide library ( i . e . , peptides in a protein sequence that are most likely to be consistently and confidently observed by current MS-based proteomics methods ) . This library in turn can be used for targeted analyses and comparative studies of expressed proteins [10 , 50–52] by spiking the samples to be analyzed with chemically synthesized proteotypic peptides , or by selected reaction monitoring ( SRM ) MS . Our C . elegans proteome dataset will be made publicly available within WormBase and will thus be useful for the entire C . elegans research community . In general , proteomics data like ours is closer to the biologically active players than transcriptomics data . It should therefore be increasingly used to investigate biological phenomena and mechanisms underlying disease pathogenesis such as neuronal degeneration and cancer development , and for the identification of conserved therapeutic target proteins .
C . elegans wild-type strain N2 ( Bristol ) was grown on 9-cm nematode growth medium ( NGM ) agar plates seeded with a lawn of the E . coli strain OP50 or in 100 ml of liquid cultures in S-basal buffer in beveled flasks . Worms were harvested from plates or liquid culture , and separated from the bacteria by washing with water or sucrose flotation . For the collection of embryos , the worms were synchronized , and eggs were removed from agar plates or obtained from the hermaphrodites by bleaching . Worm and egg pellets were homogenized with glass beads ( diameter of 212–300 μm; Sigma-Aldrich ) in the ratio of 1:1:2 ( worms:beads:buffer ) in a cell disrupter ( FastPrep FP120 , Thermo Savant; Qbiogene ) at 4 °C three times for 45 s at level 6 . The buffer used was 50 mM Tris/HCl ( pH 8 . 3 ) , 5 mM EDTA , 8 M urea . After glass bead treatment , 0 . 125% SDS was added , and the homogenate was incubated for 1 h at room temperature ( RT ) to solubilize proteins . For other experiments , the worms or eggs were homogenized with glass beads in 50 mM Tris/HCl ( pH 8 . 3 ) , 5 mM EDTA , then 0 . 75% or 1% Rapigest ( Waters ) was added , the homogenate was heated at 95 °C for 5 min , and incubated at RT for 30–60 min with gentle agitation . Cell debris was removed by centrifugation , and the protein concentration was determined using the Bradford reagent ( Sigma-Aldrich ) . The peptides were subjected to reversed-phase capillary chromatography using a 75-μm × 8-cm self-packed C18 column ( Magic C18; Michrom ) at a flow rate of 250 nl/min . Peptides were eluted with a gradient between solvent A ( 5% ACN , 0 . 2% formic acid ) and solvent B ( 80% ACN , 0 . 2% formic acid ) . The gradient was from 5% up to 45% solvent B within 69 min . The peptides were identified by CID ( collision induced dissociation ) on a Thermo-Finnigan ion trap mass spectrometer “LTQ” . Six dependent scans followed each survey scan . Raw data were converted into mzXML files and searched against a C . elegans database derived from the Wormpep database ( http://www . wormbase . org , release WS140 ) using the Sequest program [53] . The search parameters used were two missed cleavage sites , two tryptic termini , a mass tolerance of 3 Da for the parent ion and 0 . 95 Da for the fragment ion , optional oxidized methionine , and depending on the experiment , modified cysteine . Peptide assignments were statistically validated at peptide level using PeptideProphet [54] , and peptides with a probability score of 0 . 9 or higher and the proteins they belong to were selected . For the qualitative analysis of the proteome ( Figure 2 ) , peptides matching to more than one protein ( such as duplicated tubulins or histones ) , or matching to several splice variants of a protein , were counted only once ( for the first entry of the search results ) . For the quantitative analysis , however , such peptides were assigned fractionally ( see below ) . From a total of 18 different experiments ( Table S3 ) , we identified 10 , 977 proteins from 10 , 631 gene loci ( Table S1 ) . The comparative analysis of the different protein parameters was also based on WS140 . For technical reasons , all the information for the other functional analyses was extracted from release WS160 using WormMart ( http://www . wormbase . org/biomart/martview ) . The FDR for single hits was estimated first based on an experiment in which isoelectric focusing of peptides was performed on an immobilized pH gradient strip ( pH range 3–5 . 6 ) , followed by subsequent analysis of computationally predicted pIs for each peptide identification , and second by a new model based on a decoy search strategy ( L . Reiter , M . Claassen , S . P . Schrimpf , J . M . Buhmann , M . O . Hengartner , et al . , unpublished data ) . To evaluate potential bacterial contamination in our dataset , one experiment was searched against a combined C . elegans ( WormBase WS140 ) and E . coli ( SPproteomes at the European Bioinformatics Institute [EBI] , release 2005-03-19 , 4 , 338 entries ) database using the same search parameters as for the searches against the C . elegans database . After redundancy analysis , 22 , 269 distinct proteins ( including splice variants , WormBase WS140 ) and 10 , 977 proteins in our dataset were compared for the bias analysis with respect to different protein parameters . Tools from the ExPASy Web site ( http://www . expasy . ch ) were used to calculate the pIs of proteins ( protein parameter tool “protparams” ) and their hydrophobicity ( gravy computation “grand average hydrophobicity” ) . The statistical analysis shown in Figure S8 was carried out as described before [10]; the p-values for all parameter analyses were 1E−10 or better . The number and orientation of transmembrane domains of the proteins in WormBase ( WS160 ) and in our dataset were predicted using Phobius [11] . Only gene loci—not splice variants—were processed . Whenever transmembrane predictions differed for splice variants , the predictions for the longest splice variants were used . For the GO slim analysis , the GO terms listed in WormBase ( WS160 ) were mapped onto higher-level terms using the GO slim guide ( http://www . geneontology . org ) , with two exceptions: the terms “membrane” and “integral to membrane” were not mapped to the higher category term “cell , ” but instead were retained . In Figures 2E and S9 , we assigned the GO slim terms of the category “molecular function” to the predicted transmembrane proteins . For 412 proteins , there was more than one entry for molecular function . For the statistical analysis of the GO slim categories in Figure 2 , we applied the Fisher exact test and included the Bonferroni correction for multiple testing . We plotted the log ratio of observed versus expected , using the proportions in WormBase as the expectation . The GO slim categories with a p-value better than 0 . 05 are shown ( Figure 2E and 2F ) . We mined our dataset for nonannotated translated regions by preparing a whole-genome open reading frame database that was searched using the Sequest algorithm [53] . To do this , WormBase release WS160 was used to translate each chromosome into all six reading frames . Open reading frames longer than 20 amino acids were assembled into a database with headers containing the coordinates of the sequences on the genome . The resulting database contains 3 , 136 , 258 open reading frames and 132 , 018 , 220 amino acids . A subset of our data ( experiment 15 ) obtained by isoelectric focusing , comprising approximately 304 , 000 MS/MS spectra , was searched at the Functional Genomics Center Zurich . We allowed fully tryptic peptides with up to two missed cleavages , and specified oxidized methionine as variable modification and carbamylated cysteine as static modification . The results were further analyzed with PeptideProphet [54] , and 27 , 940 search hits with a PeptideProphet score greater than or equal to 0 . 95 were selected . From these , we removed 26 , 952 scans that also generated a hit against the normal Wormpep140 protein database with a score greater than 0 . 8 . Of the remaining 988 spectra , 789 were further observed to exist in Wormpep178 or an E . coli database and were therefore omitted , resulting in a final set of 199 spectra belonging to 173 different peptides . For the resulting peptides , a theoretical pI value was calculated and compared to the mean pI of all peptides in the corresponding fraction . Only peptides with a delta pI value smaller than or equal to 0 . 5 were selected . This resulted in 78 distinct peptides . WormMart ( http://www . wormbase . org/biomart/martview ) was used to extract operon architectures from WormBase release WS160 . To test whether the coregulation of genes in operons would be detectable also at the level of translated proteins , operons were first divided into length classes ( here , length is defined as the number of cotranscribed genes in each operon ) . For each length class , the fraction of operon genes was then determined for which at least one peptide was detected in at least one proteomics experiment . This fraction determines how many proteins should , on average , be detectable from a single operon if expression of the operon genes were truly independent ( when assuming independence , the number of detections per operon should follow a binomial distribution , shown as grey lines in Figure S2 ) . Applying the two-sided Kolmogorov-Smirnov test yielded p-values better than 1E−10 . For the study of operons in specific stages ( Figure S3 ) , only the proteome data was analyzed , limited to the experiments done in these stages ( with concomitantly reduced spectral counts ) . For the semiquantitative comparison between C . elegans and D . melanogaster proteomes , we used the STRING database and the Smith-Waterman similarity relations stored therein to compute orthologous groups [55] . This analysis retrieved 4 , 184 loci in C . elegans , and 4 , 302 in D . melanogaster . When working with orthology sets , each pair of orthologs was aligned with “muscle” [56] , available from http://www . drive5 . com/muscle/ . The protein sequences used were extracted from WormBase WS160 and from FlyBase release 5 . 1 ( http://flybase . org ) . Due to lineage-specific gene duplications , some proteins had several orthologs . For the interspecies abundance correlation comparison , we summed up the abundances in these cases . We independently tested another source of orthology information , InParanoid [57] , which resulted in slightly more orthologs but also in a somewhat lower interspecies abundance correlation ( RS = 0 . 76 versus 0 . 79 ) . Conversely , we also tested a stricter set of orthologs , to test for and exclude artifacts caused by potentially undetected paralogy . To conclusively separate orthologs from paralogs can be difficult , and this is the subject of intense study [58–60] . Therefore , we constructed a very strict set of orthologs by searching for reciprocal best matches between worm and fly , with the additional constraint that any extra homologs within these genomes had to exhibit no more than half the alignment score than the score between these organism ( plus , the score between the organisms had to be 60 bits or higher ) . This strict set contained only 2 , 001 pairs of orthologs , and resulted in an interspecies abundance correlation of 0 . 80 . This shows that our high correlation is not caused , or affected , by the presence of paralogs in the comparison . We calculated the relative abundance of a protein by counting how often any of its amino acids had been identified in any peptide , divided by the total number of amino acids of the protein sequence . A length restriction to peptides with ≥7 and ≤40 amino acids ( modified from [22] ) was applied . where a = protein abundance , p = identified peptides , q = tryptic peptides ( in silico digest ) , and f ( q ) = peptide length correction factor . The peptide-length correction factor takes into account the technical bias of the MS instrument , which resulted in certain peptide lengths being observed more often than others . This was learned from the data by comparing the observed peptide-length spectrum with the expected , and was corrected accordingly ( similar to [22] ) . In our hands , peptide length proved to be the most important determinant of peptide observability , since using the original APEX implementation ( “absolution protein expression profiling” ) [22] or a retrained version of the same classifier , did not further improve the observed cross-organism abundance correlation between C . elegans and D . melanogaster ( RS = 0 . 78 ) . A relative protein abundance of 1 means that the total number of amino acids in the identified peptides equals the number of amino acids in the protein . Whenever a peptide could be assigned to several proteins ( because of identical predicted tryptic peptides ) , the amino acids were assigned fractionally . Peptides specific for any of the splice isoforms originating from a given locus were pooled . This approach means that the unit of interest in our comparisons is the gene locus—not individual splice isoforms—consistent with the observed lack of conservation of alternative splicing at very large evolutionary distances [61] . Finally , protein abundances were normalized to total amount of protein detected . To plot the data , orthologs were binned into eight groups of equal size ( sorting for binning was x + y ) , and the means , as well as first and third quartiles , for each group were calculated . For the comparison of gene and protein expression , SAGE data for C . elegans [27] were downloaded from http://tock . bcgsc . bc . ca/cgi-bin/sage160 . In order to best reflect the developmental stages analyzed in our proteome data , we chose the stages SWN21 , SWL12 , SWL21 , SWL32 , SWL41 , SWYA1 , MIXED , SW022 , and DAUER . Only entries with “source = coding_RNA” were considered , and the average of the nine columns was calculated . SAGE data for D . melanogaster [29] were obtained from Professor San Ming Wang ( Northwestern University , Evanston , Illinois ) . D . melanogaster SAGE tags were mapped to all transcripts from FlyBase release 5 . 3 . The C . elegans Affymetrix GeneChip data were obtained from the Genome British Columbia C . elegans Gene Expression Consortium at http://elegans . bcgsc . bc . ca . The D . melanogaster Affymetrix GeneChip data [28] were obtained from http://www . flyatlas . org . For 2 , 695 pairs of orthologs protein abundance , SAGE and Affymetrix data were compared ( in case of several paralogs , only one of them had to have data from all three measurements ) . For the comparisons of different abundances , Spearman rank correlation coefficients were computed to avoid assumptions about the underlying distributions . Probabilities for the correlation coefficients were calculated as implemented in R; all corresponding p-values were better than 2 . 2E−16 . Further supporting the validity of spectral counting as a semiquantitative measure is a comparison of C . elegans protein abundance data against protein abundance data in yeast [41] . Importantly , the latter is not based on MS , but on immunodetection of tagged open reading frames . Orthologs correlate linearly in their abundance over two orders of magnitude ( RS = 0 . 54; Figure S10 ) . The correlation for sequence conservation ( aligned to D . melanogaster ) and protein abundance was calculated for 4 , 013 C . elegans proteins . Orthologs were binned into eight groups of equal size ( Figure 5C ) . | Proteins are the active players that execute the genetic program of a cell , and their levels and interactions are precisely controlled . Routinely monitoring thousands of proteins is difficult , as they can be present at vastly different abundances , come with various sizes , shapes , and charge , and have a more complex alphabet of twenty “letters , ” in contrast to the four letters of the genome itself . Here , we used mass spectrometry to extensively characterize the proteins of a popular model organism , the nematode Caenorhabditis elegans . Together with previous data from the fruit fly Drosophila melanogaster , this allows us to compare the protein levels of two animals on a global scale . Surprisingly , we find that individual protein abundance is highly conserved between the two species . So , although worms and flies look very different , they need similar amounts of each conserved , orthologous protein . Because many C . elegans and D . melanogaster proteins also have counterparts in humans , our results suggest that similar rules may apply to our own proteins . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biotechnology",
"biochemistry",
"evolutionary",
"biology"
] | 2009 | Comparative Functional Analysis of the Caenorhabditis elegans and Drosophila melanogaster Proteomes |
Sexually dimorphic phenotypes are a universal phenomenon in animals . In the model animal fruit fly Drosophila , males and females exhibit long- and short-sleep phenotypes , respectively . However , the mechanism is still a mystery . In this study , we showed that juvenile hormone ( JH ) is involved in regulation of sexually dimorphic sleep in Drosophila , in which gain of JH function enlarges differences of the dimorphic sleep phenotype with higher sleep in males and lower sleep in females , while loss of JH function blurs these differences and results in feminization of male sleep and masculinization of female sleep . Further studies indicate that germ cell-expressed ( GCE ) , one of the JH receptors , mediates the response in the JH pathway because the sexually dimorphic sleep phenotypes cannot be rescued by JH hormone in a gce deletion mutant . The JH-GCE regulated sleep dimorphism is generated through the sex differentiation-related genes -fruitless ( fru ) and doublesex ( dsx ) in males and sex-lethal ( sxl ) , transformer ( tra ) and doublesex ( dsx ) in females . These are the “switch” genes that separately control the sleep pattern in males and females . Moreover , analysis of sleep deprivation and circadian behaviors showed that the sexually dimorphic sleep induced by JH signals is a change of sleep drive and independent of the circadian clock . Furthermore , we found that JH seems to also play an unanticipated role in antagonism of an aging-induced sleep decrease in male flies . Taken together , these results indicate that the JH signal pathway is critical for maintenance of sexually dimorphic sleep by regulating sex-relevant genes .
Sleep is a general phenomenon identified in many species and is usually accompanied with various sex-specific properties . For instance , in human beings , women have higher sleep spindle number and density than men[1] . Also there are sex differences in responses of male and female rats to sleep deprivation[2] . In recent years , the Drosophila genetic model system for sleep research has been shown to share all the key characteristics of mammalian sleep[3 , 4 , 5] . Similar to a variety of organisms , sex-specific properties in sleep pattern are present in fruit flies . Previous studies reported that there is a large sex difference in the total sleep amounts during the daylight hours , in which female sleep is only 40% of male sleep in D . melanogaster[6] . To address that phenomenon , specific neurons[7 , 8] and non-neural factors [8] relevant to sexually dimorphic sleep have been discussed . Other types of sexual dimorphism ( e . g . , stress resistance , feeding , and physical characteristics ) mediated by insulin , dopamine and Juvenile Hormone ( JH ) may participate in the relevant regulatory circuits[9 , 10] , but their specific interactions and the molecular mechanism are still unclear . Here , we examine the role of JH in the control of a sexually dimorphic sleep phenotype . Juvenile hormones are a group of acyclic sesquiterpenoids that regulate insect physiological processes , such as development , reproduction and diapause[11] . JHs are synthesized primarily in the corpora allata ( CA ) , a pair of endocrine glands with neural connections to the brain[12] . The pars intercerebralis ( PI ) has been recently identified as a structure of neuroendocrine regulation for sleep and wakefulness in Drosophila[13] . It has been suggested that at least part of the CA is derived from the PIa/c ( PIa: located around the medial lobes of the mushroom body; PIc: located dorsal to the central complex ) [14] . Furthermore , JH could serve as a molecular link between PI neurons and the corpus cardiacum—copora allata ( CC-CA ) gland[15] . Previous studies mapped the PI area of the brain which feminizes male locomotor activity through up-regulated expression of the sex-determination gene transformer ( tra ) [16] . In D . melanogaster , the presence of XX ( or XY ) sex chromosome constitution promotes female ( or male ) somatic development through a sex determination hierarchy[17] . The binary switch gene sex-lethal ( sxl ) regulated by the X chromosome counting system controls sexual identity in Drosophila . When SXL is on , it imposes female development , whereas male development proceeds when it is off[18] . Transformer ( tra ) , fruitless ( fru ) and doublesex ( dsx ) act downstream of sxl to control sexual development in Drosophila . TRA regulates female-specific alternative splicing of dsx to encode DSXF protein . Without TRA , male-specific alternative splicing of dsx pre-mRNA occurs , and this transcript encodes DSXM protein[19] . Fru , a pleiotropic gene which lies at the bottom of the sex-determination hierarchy , controls male sexual behavior and is essential for viability in Drosophila[20] . Indeed , these genes products control sexual differentiation of the organism body and also contribute to relevant behavioral and neural development[21] . Biosynthetic activity in the CA is considered to be a major factor in the regulation of JH titer[22 , 23] . The JH biosynthetic pathway mainly includes two steps: the generation of farnesyl diphosphate ( FPP ) via the mevalonate pathway in early steps , and then conversion of farnesoic acid ( FA ) generated from FPP to active JH in later steps[11] , in which many enzymes have been found to be responsible for the biosynthesis of JH , such as HMG-CoA synthase ( HMGS ) , HMG-CoA reductase ( HMGCR ) , phosphomevalonate kinase ( MevPK ) , and JH acid O-methyltransferase ( JHAMT ) [24 , 25] . Among these , the enzyme JHAMT expressed in the CA activates the JH biosynthesis pathway in the last step and is considered the key regulator in JH biosynthesis . In D . melanogaster , the JHAMT expression pattern is consistent with changes of the JH titer , and recombinant DmJHAMT protein can produce JHs with the presence of its substrates[23] . In addition , the rate of JH in vitro biosynthesis by brain-ring gland complex is higher when JHAMT is over-expressed in the CA[26] . In Bombyx mori , jhamt mRNA level is consistent with JH titer , and the transcriptional suppression of jhamt gene is crucial for down-regulation of JH biosynthesis[11 , 25] . Therefore , JHAMT is a useful and important index for JH . In this study , we explored the role of JH and its signal pathway on sleep regulation through a genetically transgenic system .
To compare sleep-wake profiles between both sexes , we monitored hundreds of wild type flies . Results showed that sleep of males during daytime was much higher than that of females ( Fig 1A ) . Transcriptomic analysis of sleep-deprived and non-deprived rats has shown that sleep–wakefulness states affect gene expression[27] . Thus , we performed a sleep deprivation assay by forcing flies of wild-type w1118 to be awake all night using mechanical vibration , and surprisingly found that the changes of jhamt transcript levels exhibited two opposite tendencies in the male and female flies after 12 h sleep deprivation ( Fig 1B ) . To further elucidate the function of endogenous JH signaling on the regulation of sleep in flies , we used a tissue-specific driver ( Aug21-gal4 ) to over-express JHAMT—a key enzyme synthesizing JH in the CA cells[26 , 28] . Over-expression was confirmed by immunofluorescence with anti-DmJHAMT antibody ( Fig 1C ) . Results showed that flies with up-regulated JHAMT exhibited enhanced sleep differences between both sexes , in which the daytime sleep amount in males showed a significant increase but a significant decrease in females ( Fig 1D and 1E ) , primarily caused by sleep episode duration ( Fig 1F ) . Furthermore , in order to eliminate the possibility that abnormality of metamorphosis development would contribute to non-specific effects on sleep , we applied three different concentrations of exogenous pyriproxyfen–a juvenile hormone analog ( JHA ) - to adult wild-type flies , and then sleep behavior was monitored . Results showed that flies with JHA application exhibited similar sexually dimorphic sleep responses like those above with genetic gain of JHAMT function , with 0 . 1mM JHA as an effective concentration and regulation of sleep showing dose-dependency ( Fig 1G–1I ) . For a loss-of-function analysis , we used a jhamt mutant ( jhamt2 , in which JHAMT is effectively deleted [29] ) , which was further confirmed by immunofluorescence with anti-DmJHAMT antibody , to assess sleep regulation ( Fig 2A ) . With loss of jhamt function , the daytime sleep amount in females significantly increased ( Fig 2B and 2C ) . Although sleep amount of males during daytime showed no significant difference compared to control ( Fig 2B and 2C ) , sleep quality was significantly affected by decreases of the sleep-bout duration ( Fig 2D ) , indicating reduction of sleep quality in the mutant males . The daytime sleep amount and sleep-bout duration in the jhamt-deleted mutants of both males and females was completely or partially rescued by exposing them to 0 . 1mM JHA , in which sleep quality of males increased and of females decreased ( Fig 2B–2D ) . All these results above indicate that males and females exhibit a JH-responsive sexually dimorphic sleep phenotype—a long sleep phenotype in males and a short sleep phenotype in females . An increase of JH increases the sexually dimorphic sleep phenotype , and a decrease of JH decreases this difference and blurs the gender boundaries . In other words , lack of JH will result in feminization of male sleep and masculinization of female sleep . The Drosophila germ cell-expressed ( GCE ) and methoprene-tolerant ( MET ) , both belonging to bHLH-PAS transcription factors , are products of two paralogous genes on the X chromosome . Both proteins potentially mediate the effects of JH as candidate JH receptors[30] . In this study , gce2 . 5k and met27 ( both of them are null alleles ) were used to further investigate the function of the JH signaling pathway on daytime sleep . Results showed that the gce deletion mutant also exhibited sexually dimorphic effects on sleep . Gce mutant male flies exhibited a short-sleep phenotype , while female flies exhibited a long-sleep phenotype . Moreover , administration of 0 . 1mM JHA did not rescue the sleep phenotypes , and even increasing JHA concentration to 1mM did not rescue the sleep phenotype ( Fig 3A and 3B ) . However , in the met mutant , the males and females did not exhibit sexually dimorphic sleep phenotypes , but administration of 1mM JHA did effectively induce sexual dimorphism again ( S1A and S1B Fig ) , indicating the met mutation may be partially involved in a JH-responsive sexually dimorphic sleep regulatory pathway , and it may be also involved in other signaling to affect flies’ daytime sleep . The sleep behavior of Gce2 . 5k but not of met27 phenocopied jhamt2 , suggesting that GCE is the main receptor to regulate Drosophila sexually dimorphic sleep in the JH signaling pathway . To determine how the JH pathway acts on sexual sleep differentiation , we focused on the JH-GCE signaling pathway effects on the sex determination . Sex-lethal ( sxl ) , a key upstream gene of sex-determination in female flies , is well known to establish and maintain sex determination in females[31] , while FruM , a male-specific upstream gene for sex determination , has been shown to contribute to sexual dimorphism by sculpting the sexually dimorphic nervous system[32] . In this study , we compared mRNA levels from both female sxl and tra and male fruM genes among w1118 ( as control ) , jhamt2 and gce2 . 5k , and found their transcript levels in the jhamt and gce mutants almost decreased to half compared with control levels ( Fig 4A , 4B & 4E ) . Because SXL is mainly expressed in female germline cells[33] , we determined SXL protein expression in female ovaries of wild-type w1118 , jhamt2 and gce2 . 5k . Results showed that SXL protein expression was much less in both jhamt and gce deletion mutants than in w1118 controls ( Fig 4C ) . Abnormal expression of SXL in female flies suggested that the female sex differentiation pathway , mediated by SXL and TRA , may be affected in jhamt and gce mutants . We further used elav-gal4 to drive gce dsRNA expression in a large field of central neurons and the results showed that the sleep pattern of female flies was much closer to that of males ( Fig 4D ) . FRUM protein is broadly expressed in the central nervous system of males , and the gce mutation led to reduced FRUM protein in the PI neurons ( Fig 4F ) , which have been reported to be involved in locomotor feminization[16] . Similarly , the density of FRUM also decreased to a certain degree in jhamt2 males ( Fig 4F ) . Furthermore , we specifically used a fru driver ( fru ( 16 ) -gal4 ) to drive gce dsRNA expression and found the male flies’ daytime sleep significantly reduced as assessed by both the daytime sleep amount and the daytime sleep-bout duration ( Fig 4G ) . These results indicate that JH-GCE pathway is important for regulating sexual determination genes and sleep behaviors via related neurons . To further identity the function of sxl , tra and fru genes in sleep regulation , we separately used deletion or insertion mutants blocking expression of sxl , tra and fru to observe effects on sleep behavior . Trans-heterozygous mutant sxl[f2]/sxl[f18] females are viable and fertile , but their sleep pattern is similar to the male pattern with an increased sleep during daytime ( Fig 5A ) , and heterozygous tra mutants ( tra[1]/+ ) also exhibit increased sleep in females ( Fig 5B ) . In contrast , a fru insertion mutation in males caused significant decreases of sleep during the daytime ( Fig 5C ) . To verify the effects of masculinization or feminization , we simultaneously used different drivers ( Aug21-gal4 and Fru ( 16 ) -gal4 ) to drive the expression of traF ( Female-specific protein TRA ) in males . Results revealed that male fly’s daytime sleep pattern is closest to the female fly’s daytime sleep when the traF of females is expressed in fru neurons of males , which is consistent with the importance of FRU-expressing neurons for the male sleep pattern . Expression of traF in the JH synthesis tissue corpus allata ( with Aug21-Gal4 ) also significantly decreased daytime sleep amount and sleep bout duration of the male flies ( Fig 5D ) . As for more genetic methods , FruΔtra , with a modified S exon with tra binding sites mutation , produces female flies exhibiting a male sleep pattern ( Fig 5E ) , and Fru-F , with a deletion of the S exon of the fru male specific splice form , produces male flies exhibiting a female sleep pattern ( Fig 5E ) . Taken together , these results indicate that JH-GCE signaling regulating sleep sexual dimorphism is via the sexual determination genes sxl-tra in females and fru in males , respectively , and these genes seem to be the “switch” genes in females and males to separately control the sexually dimorphic sleep pattern in D . melanogaster . Then , we explored another sex differentiation downstream gene by using a dsx heterozygous mutant . Previous research demonstrated this gene is bifunctional , since male-specific DSX represses female sexual differentiation and vice versa for female-specific DSX[34] . Our results showed that mutation of dsx in both males and females results in an intersexual sleep pattern . Total daytime sleep decreased about 22 . 5% ( P < 0 . 01 ) in dsx[48]/+males and increased about 34 . 0% in females ( P < 0 . 01 ) ( Fig 5F ) , respectively . These results indicate that dsx is important to maintain sleep sexual differentiation . Thus , all sleep analysis in the context of sex differentiation-related genes suggests that JH signaling regulates sleep sexual dimorphism through sex differentiation-related genes . Sleep is usually under control of homeostasis , which responds to sleep need and adjusts sleep intensity , and of the sleep time keeper , or the circadian clock[35 , 36] . Prolonged wakefulness increases sleep pressure which results in a rebound in the next sleep stage[37] , and vice versa . To examine whether the homeostatic system is destroyed by a blocked JH signaling pathway , sleep deprivation was performed during the nighttime and sleep parameters of the next daytime were analyzed using the JH receptor mutant gce2 . 5k and wild-type w1118 control flies . Results showed that a functional sleep homeostatic system was completely preserved in the gce mutant ( Fig 6A and 6B ) . We further examined the circadian parameters of jhamt and gce mutants ( Fig 6C ) and found that the jhamt mutation dampened the flies’ rhythm amplitude and rhythmicity . Furthermore , the gce2 . 5k mutant still exhibited a robust activity rhythm and ~24h periodicity . Based on the circadian phenotypes of jhamt2 and gce2 . 5k , we reasoned that the JH-GCE signaling effects on sleep are circadian rhythm independent .
The insulin system has been reported to be involved in regulation of sleep and locomotor activity [38 , 39] . A previous study showed that HMGCR , an enzyme responsible for the biosynthesis of JH , is controlled by the insulin receptor ( lnR ) [40] . A JH inhibitor fluvastain fed to male flies led to male locomotor activity feminization , and this effect was reversed by application of a JH analog[15] . Ablation of insulin-producing neurons in the adult PI or an insulin receptor mutant were also found to abolish sexual dimorphism in locomotor activity , suggesting that the insulin signaling pathway may work in conjunction with the JH pathway to control sexually dimorphic behaviors[38] . Ecdysone , another major endocrine hormone , has been shown to regulate adult flies’ daytime sleep[41] . However , the effects of ecdysone on sleep did not involve sex differences , although there is a vital relationship between JH and ecdysone for controlling molting and metamorphosis . As a male specific product , sex peptide inhibits female flies’ siesta sleep through copulation behavior , contributing to sexually dimorphic sleep[6] . Recently another paper reported that the male’s dorsal clock neurons are more active than those in females , in correlation with the increased sleep behavior of male flies[7] . In this study , we found that wild-type males synthesized more JHAMT protein in CA cells than females , and notably old flies ( 30th day adults ) seem to synthesize more JHAMT than young adults ( 3rd day adults ) , although it is a small increase in females ( Fig 7A and 7B ) . According to our sleep behavior results , increased JHAMT should promote male sleep and inhibit female sleep . Thus , the results imply that JH may act against sleep decrease in the male aging process . Interestingly , we did find that old male flies exhibit less sleep loss than old females , whose sleep loss reaches almost 50% ( Fig 7C and 7D ) . These phenotypes possibly indicate that aging-induced sleep loss may include effects of JH . JH is an important endocrine hormone , which regulates many physiological and developmental processes through JH receptors . As the JH receptors , MET or/and GCE independently act in different functions[42] . For example , JH regulates the production of female sex pheromones via MET rather than its paralog GCE[43] . However , JH regulates nuclear receptor gene E75A in S2 cells through GCE but not MET[44] . According to the sleep behavioral assays of gce and met mutants , there is no doubt that GCE is the principal receptor for regulation of sexual dimorphism in sleep . Alternatively , met mutation caused sleep length to be significantly longer during the daytime in both sexes , suggesting that met may be involved in other regulatory pathways . A previous study reported that a MET/CYC heterodimer activates circadian rhythm-dependent gene expression in female mosquitoes[45] , suggesting that met may control the circadian rhythm independent of sleep . From humans to insects , morphological and behavioral differences between male and female are ubiquitous . Mating , courtship-related sensory responses and aggression have been demonstrated to be under the control of the sex determination hierarchy[46 , 47 , 48] . Tra initiated by sxl modulates or even drives various female characteristics . A recent paper has reported that myc and tra can promote sexual size dimorphism in fruit flies[49] . In fact , the idea that the sex determination system participates in the regulation of sex-specific size in stag beetles was proposed several years ago , and JH was also shown to play a crucial role[10] . With regard to behavior , TRA can feminize male locomotor behavior in transgenic flies , as described above ( Ref . 15 ) . Fru , a terminal gene of the sex determination hierarchy , orchestrates male-specific neuronal morphology and behavior[50] . Activation of male-specific dsx/fru positive P1 neurons in female brain induces male-like courtship behavior[51] . Another terminal gene ( dsx ) coupled with fru to regulate male sexual behavior[52] . All these sex determination genes orchestrate the characteristics of “maleness” and “femaleness” . In this study , we showed that sexually dimorphic sleep in Drosophila regulated by JH-GCE signaling is produced by the pathway of sex-specific regulatory genes differing between males and females . We propose a model based on the available evidence to explain this complicated sleep sexual dimorphism phenomenon ( Fig 8 ) . The pars intercerebralis sends related signals to the corpora allata . The latter initiates jhamt transcription and activates JH biosynthesis from FA . JH is released and binds to the GCE protein to promote female-specific gene transcription in female D . melanogaster . On the other hand , JH signaling acts on fru , which encodes male specific protein FRUM in the CNS , to maintain the male sleep pattern in the absence of SXL and TRA . DSX may be also responsible for sustaining male and female sleep behaviors . Therefore , the JH signaling pathway plays a necessary role in regulation of sexual dimorphism of sleep in D . melanogaster .
The following strains were used in this study: UAS-jhamt , UAS-gce-RNAi , Aug21-gal4 , jhamt-gal4 , jhamt2 , gce2 . 5k and met27 , which were previously reported ( Ref . 26 , Ref . 29 and Ref . 30 ) . UAS-traF , fru ( 16 ) -gal4 , FruΔtra and Fru-F were obtained from Yi Rao , Yufeng Pan and Chuan Zhou’s labs , respectively . Dsx[48]/TM2 , tra[1]/TM2 , Sxl[f2]/FM7c and Sxl[f18]/FM7c were purchased from the Bloomington Drosophila Stock Center . Flies were reared at 25°C and 65% relative humidity on a standard cornmeal-yeast-agar medium in a 12h light/12h dark cycle . Three- to five-day-old flies were housed in monitor tubes ( 5[W] × 65[L] mm ) with fly food . Experiments were performed in an incubator at a temperature of 25 ± 1°C and a relative humidity of 65% . Light was turned on at ZT0 ( local time 06:30 ) and off at ZT12 ( local time 18:30 ) . The sleep activity was recorded using the Drosophila Activity Monitoring System ( Trikinetics , Waltham , MA ) . The details for the experimental protocol and data analysis were described by Chen et al ( 2013 ) . Sleep was deprived by mechanical vibration of three seconds within each minute interval for the whole night ( 12h ) . All the sleep deprived flies were immediately fixed in liquid nitrogen for qRT-PCR assays . The juvenile hormone analog ( JHA ) pyriproxyfen ( gift from Xiru Liang , China Agriculture University ) was dissolved in 95% ethanol , and JHA-containing food was prepared by adding JHA solution to standard Drosophila food at 50–55°C to the indicated concentrations . Newly eclosed flies were placed in glass tubes with standard medium for three days and transferred to detecting tubes with food containing JHA for behavior monitoring . Three-day-old flies were fixed in 4% paraformaldehyde for 2 h . Then 6 to 8 flies were dissected in phosphate-buffered saline ( PBS ) . They were washed 3 times with PBST ( 0 . 5% Triton X-100 in PBS ) at room temperature . The tissue was soaked in blocking solution PNT ( 1% goat serum in PBS ) for 1 h at room temperature and incubated with primary antibody for 16–24 h at 4°C . After 3 washes , the tissue was incubated in secondary antibody for 2 h at room temperature . The samples were analyzed with Nikon Eclipse TE2000-E and Nikon D-Eclipse ( Nikon , Japan ) confocal microscopes . Anti-DmJHAMT antibody was provided by Dr . Niwa . Anti-FruMale antibody was provided by the Yamamoto lab . Anti-SXL antibody was provided by DSHB at the University of Iowa . RNA was isolated from whole bodies of 3-day-old male and female flies using Trizol Reagent ( TIANGEN , Beijing ) according to the manufacturer’s protocol . RNA was reverse transcribed using Fast cDNA Reverse Transcription Kit ( TIANGEN , Beijing ) . The Quantitative real-time PCR assay was performed using Applied Biosystem Step One Real Time PCR system ( Applied Biosystem , Foster , CA , USA ) , RealMasterMix kit and SuperReal PreMix Plus kit ( TIANGEN , Beijing ) . The sequences of primers are shown in Supplementary S1 Table . Statistical analysis was performed with SPSS statistics 17 . 0 . P values was obtained with t-test and considered significant at P < 0 . 05 and extremely significant at P < 0 . 001 . | Sleep is a very important biological behavior in all animals and takes up around one third of the lifespan in many animals . In both insects and mammals ( including humans ) , sleep differences between male and female ( sexually dimorphic sleep ) have been described over the past decades . However , its internal regulation mechanism is still unclear . The fruit fly Drosophila melanogaster , sharing most sleep characteristics with humans , has been used for sleep studies as a powerful model for genetic analysis . In this study , we reported that Juvenile hormone ( JH ) induces completely different sleep effects between males and females with higher sleep in males and lower sleep in females , while loss of JH function blurs these differences and results in feminization of male sleep and masculinization of female sleep . Further studies indicate that the sexual dimorphism of sleep is generated through the sex differentiation-related genes regulated by JH and its receptor GCE ( germ cell-expressed ) signaling . Furthermore , we found that JH seems to also play an unanticipated role in aging-induced sleep changes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"sleep",
"gene",
"regulation",
"neuroscience",
"animals",
"animal",
"models",
"physiological",
"processes",
"developmental",
"biology",
"drosophila",
"melanogaster",
"model",
"organisms",
"experimental",
"organism",
... | 2018 | Sexual dimorphism of sleep regulated by juvenile hormone signaling in Drosophila |
The neural mechanisms determining the timing of even simple actions , such as when to walk or rest , are largely mysterious . One intriguing , but untested , hypothesis posits a role for ongoing activity fluctuations in neurons of central action selection circuits that drive animal behavior from moment to moment . To examine how fluctuating activity can contribute to action timing , we paired high-resolution measurements of freely walking Drosophila melanogaster with data-driven neural network modeling and dynamical systems analysis . We generated fluctuation-driven network models whose outputs—locomotor bouts—matched those measured from sensory-deprived Drosophila . From these models , we identified those that could also reproduce a second , unrelated dataset: the complex time-course of odor-evoked walking for genetically diverse Drosophila strains . Dynamical models that best reproduced both Drosophila basal and odor-evoked locomotor patterns exhibited specific characteristics . First , ongoing fluctuations were required . In a stochastic resonance-like manner , these fluctuations allowed neural activity to escape stable equilibria and to exceed a threshold for locomotion . Second , odor-induced shifts of equilibria in these models caused a depression in locomotor frequency following olfactory stimulation . Our models predict that activity fluctuations in action selection circuits cause behavioral output to more closely match sensory drive and may therefore enhance navigation in complex sensory environments . Together these data reveal how simple neural dynamics , when coupled with activity fluctuations , can give rise to complex patterns of animal behavior .
Even in the absence of environmental cues , neurons receive and produce a barrage of fluctuating , ongoing signals . These fluctuations are both deterministic , reflecting a neuron’s embedding within complex dynamical networks , and random , arising from stochastic noise sources at synapses and ion channels [1 , 2] . Although the influence of these fluctuations on peripheral sensory processing is well studied [3–6] , very little is known about how they may affect central circuits [7] . Action selection ( AS ) circuits [8] , including ‘command’ neurons that drive behavior from moment to moment [9–11] , may be particularly susceptible to activity fluctuations: they represent information bottlenecks where a relatively small number of neurons can have a disproportionately large influence on actions . The sensitivity of AS circuits to internally generated fluctuations in neural activity is suggested by ecological studies showing how intermittent patterns of walking and resting in animals [12] are well characterized by random walk models [13] . Similarly , behavioral transitions in C . elegans can be effectively captured using a tunable stochastic term within a deterministic mathematical framework [14] . While progress is being made [11 , 15 , 16] , in vivo investigation of the dynamics of complex AS networks remains challenging . In this light , computational modeling can serve as an excellent starting point for generating theoretical predictions that guide in vivo studies . In particular , tools that exploit the power of neural network optimization and dynamical systems analysis [17] are gaining attention [18 , 19] for their ability to elucidate animal behavior [20 , 21] and the activity of neural ensembles [22 , 23] . In this study we used neural network optimization to infer the dynamics of AS circuits driving the locomotor walking patterns of Drosophila melanogaster . Drosophila is an attractive model organism for this type of investigation since its behaviors are increasingly well-described [24 , 25] . Previous studies of Drosophila locomotor patterning have predominantly focused on walking because this behavior has reproducible statistics and can be measured at high-throughput [26–29] . Importantly , due to their relatively small number of neurons as well as the availability of powerful genetic tools , Drosophila AS circuits are under intense investigation [11 , 16 , 30 , 31] . This raises the possibility of testing and further constraining computationally derived theoretical predictions . Several models may explain how fluctuations in AS circuits influence neural activity and behavior . In the simplest , membrane potential fluctuations in AS neurons directly impact the firing of these neurons . Consequently , exceptionally high intensity fluctuations might cause command neurons to fire and initiate actions more frequently . However , this simple feed-forward framework ignores the highly interconnected nature of neural circuits within the central brain . Therefore , more complex dynamical models incorporating feedback may be more appropriate . However , the dynamical features that make central circuits more or less susceptible to the influence of activity fluctuations are unknown . These may include the location and number of stable and unstable equilibrium points in neural activity phase space . To address this question we developed a method for automatically generating neural network models that reproduce measured animal behaviors . Our modeling approach relies on Continuous-Time Recurrent Neural Networks ( CTRNNs ) : dynamical systems that share important properties with biological neural circuits [19 , 32] . These models are consequently more informative of in vivo circuit dynamics than other simple models like non-neuronal Markov [33 , 34] and random walk schemes [12 , 13] . We emphasize that the resulting neural networks are not intended to map directly onto the anatomy of Drosophila AS circuits . Instead they reveal emergent dynamics that represent theoretical predictions about in vivo circuit function . To generate a behavioral dataset for constraining our models we first measured Drosophila basal ( i . e . , sensory deprived ) and odor-evoked locomotor patterns . Next , to explore how neural activity fluctuations might be used to drive basal locomotion in sensory deprived Drosophila , we generated populations of neural network models whose virtual locomotor outputs reproduce the basal locomotor statistics of sensory deprived Drosophila . We next identified which of these models could also match , without changing their underlying dynamics , the odor-evoked locomotor patterns of genetically distinct Drosophila strains . Using dynamical systems analysis , we discovered that models that best reproduce Drosophila basal and odor-evoked locomotor patterns ( i ) require neural activity fluctuations and ( ii ) exhibit feedback-driven multistable dynamics that reorganize in response to sensory stimulation .
Our modeling approach relies on optimizing neural network parameters to match Drosophila data . Therefore , we quantified Drosophila locomotion with high temporal resolution by developing a high-throughput system combining synchronized video-capture at 20 frames per second ( fps ) [35 , 36] , computer-controlled odor delivery ( Fig 1A ) , and behavioral tracking [37] ( Fig 1B ) of the position and orientation of individual flies within a planar arena . Using this system we could study basal locomotion in the absence of visual [24 , 25] , olfactory [38] , gustatory [39 , 40] , and time-varying mechanosensory/auditory [41] stimuli . In addition , to capture Drosophila behaviors driven by sensory cues [24 , 25] , we used a system of valves to deliver precisely timed and spatially homogeneous odor stimuli ( 10% acetic acid [42] ) . Using these tools , we could acquire enough behavioral data to detect patterns in the highly variable behaviors of individual animals [43] . We performed two experiments for each individual fly ( Fig 1C ) . In the first ‘odor impulse’ experiment we tracked 60 s of basal locomotion in the absence of any sensory stimulus , followed by 30 s of locomotor responses to uniform odor exposure , and finally 90 s of post-odor basal locomotion . In the second ‘odor aversion’ experiment we tracked locomotion for 2 min while presenting the aversive odorant on alternative sides of the arena in four separate 30 s periods . We performed experiments under dim far-red illumination , a wavelength of light for which flies are insensitive [44] , to minimize the influence of visual cues on behavior . While much information can be extracted from our measurements , we focused on the presence or absence of walking bouts since these most directly reflect the activity of AS circuits rather than downstream central pattern generators that control leg coordination and walking speed [45 , 46] . As in previous studies [47 , 48] , we classified locomotor behaviors as intermittent walking or stationary intervals ( Fig 1D ) [12] by applying a cutoff to walking speed data ( S1A and S1B Fig ) . As expected , basal locomotor behaviors for individual flies were unpredictable ( Fig 1E ) and characterized by bursts of locomotor activity separated by longer intervening periods of inactivity [26 , 28] . Therefore , to reveal patterns behind these highly variable behaviors , we averaged walking/stationary time-series across 225 genetically identical flies of a single , Canton-S strain . Prior to odor stimulation , flies exhibited a high basal locomotor frequency: many animals walked in the absence of salient sensory cues ( Fig 1F , green , ‘Basal locomotion’ ) . Upon odor presentation , locomotor frequency increased rapidly ( Fig 1F , black ) . When the odor was removed ( S1C Fig ) , locomotor frequency decayed ( Fig 1F , cyan , ‘Decay’ ) . Surprisingly , basal locomotor frequency did not simply return to the pre-odor rate but continued decaying to a substantially lower level ( Fig 1F , magenta , ‘Reduced basal locomotion’ ) . To examine the variation in these complex odor-evoked locomotor patterns , we tracked approximately 200 individuals from each of 98 genetically-distinct , inbred fly strains from the Drosophila melanogaster Genetic Reference Panel ( DGRP ) [49] . These experiments resulted in a behavioral dataset comprising 20 , 223 animals . Indeed , as for Canton-S flies ( Fig 1F ) , across most of the DGRP strains ( Fig 2A , see S1 Table for ‘RAL’ ( Raleigh ) strain IDs ) we observed ( i ) basal locomotion ( Fig 2D ) , ( ii ) post-odor decay of locomotion for periods ranging over an order of magnitude ( Fig 2A and 2B ) , and ( iii ) in most strains , reductions in post-odor basal locomotion ( Fig 2A and 2C , post-/pre-odor frequency < 1 ) . This rich behavioral diversity might reflect random experimental variation or , alternatively , intrinsic biological differences between each strain . To distinguish between these possibilities , we examined the reproducibility of locomotor patterns and found that average basal locomotion was highly consistent for each strain ( Fig 2D ) . Similarly , the time-course of odor-evoked and post-odor locomotion more closely matched between flies of the same strain ( Fig 2E , red ) than between flies of different strains ( Fig 2E , blue ) . Importantly , these simple locomotor characteristics are linked to more complex , ethologically relevant behaviors: median odor aversion ( Fig 2F ) was significantly correlated with basal locomotor frequency across all strains ( Fig 2G , Pearson’s correlation coefficient R = 0 . 65 , P < 10−4 ) . We next asked to what extent fluctuations in AS neural activity can explain these common and reproducible locomotor properties , and what the underlying neural dynamics might be . To address these questions , we built neural network models that were constrained by the requirement to reproduce Drosophila basal and odor-evoked walking patterns . We took a three-step approach for generating and studying AS network models ( Fig 3 ) . First , we generated a population of models ( Fig 3B ) whose virtual locomotor outputs matched the statistics of basal locomotion for a single Canton-S Drosophila strain ( Fig 3A and 3C ) . Second , from these models we identified those that could also reproduce the time-course of odor-evoked locomotion for three genetically distinct Drosophila strains ( Fig 3D ) . Finally , we examined how the emergent dynamics of these networks allow them to reproduce Drosophila behavior ( Fig 3E ) . For the first step , we reasoned that basal locomotion in a sensory-deprived environment would most closely reflect the unperturbed , ongoing activity of Drosophila AS circuits . Therefore , we used these behaviors as a target dataset for neural network generation . In our initial experiments , we observed that some flies could remain stationary for over 20 min . Therefore , to capture the complete range of behavioral intervals , we acquired 5 h of basal locomotor sequences from Canton-S strain flies ( Fig 3A ) . The exact time-courses of individual fly behaviors depend on many , often unknown , factors . Therefore , we aimed to generate network models that could reproduce the duration of walking and stationary intervals rather than exact walking trajectories . Although flies spent more time near the arena edges [47] ( S2 Fig ) , walking and stationary interval durations were only very weakly correlated with arena location ( S2B–S2E Fig ) : the distance correlation [50] between interval start/end locations and interval durations were ~0 . 18 and ~0 . 08 for walking and stationary intervals , respectively ( S2B–S2D Fig ) . These values roughly correspond to a 20% and 10% correlation between walking and stationary bout durations and locations ( S2E Fig ) . Thus , locomotor patterns in our sensory-deprived environment were largely uncoupled from the arena geometry . Our model-discovery method was based on stochastic parameter optimization and therefore required well-defined quantitative metrics for comparing candidate network models with Drosophila behavior ( i . e . , a cost function that guides the search for models ) . Additionally , to efficiently generate models , small changes to model parameters must result in similarly small changes in these quantitative metrics ( i . e . , a smooth fitness landscape ) [51] . Therefore , we normalized histograms of walking and stationary interval durations in two ways . First , each histogram bin was multiplied by its own duration to ensure that more frequent , short-duration locomotor bouts were not over-valued . Second , empty bins were removed from each histogram by using variable bin widths ( S3 Fig; see Materials and Methods ) . The resulting histograms provided a quantitative measure reflecting Drosophila basal locomotor patterns ( Fig 3A , bottom ) . To generate network models we employed a well-established neural network modeling framework , the Continuous-Time Recurrent Neural Network ( CTRNN ) [20] . CTRNNs are an intermediate representation of neural circuits that do not model precise ionic conductances or action-potential generation but retain the dynamical characteristics of neural circuits . Therefore , the emergent dynamics of generated network models , rather than their precise connectivity , are the instructive features [52] . Our CTRNN models were fully connected with recurrent and reciprocal connections between neurons , an intrinsic tau defining the time-scale of activity , and a bias input that constitutively drives the activity of each neuron . Depending on the experiment , our models could also have a Gaussian noise input—representing ongoing fluctuations in neural activity arising from both deterministic network dynamics as well as stochastic neuronal noise—and could have inputs representing olfactory sensory drive ( Fig 3B ) . For each model , one neuron was selected prior to parameter optimization as the output neuron ( NOUT ) driving locomotor behavior . If this output neuron’s activity exceeded a threshold , the virtual fly was walking . Otherwise , the virtual fly was stationary . Using this modeling framework ( Fig 3B ) , we developed an automated pipeline to generate models whose virtual locomotor walking and stationary bouts had the same durations as the Canton-S strain basal locomotor bouts ( Fig 3A ) . We used an iterative optimization algorithm , Particle Swarm Optimization [53] , to define all network parameters ( e . g . , edge weights , tau , bias inputs ) for multiple models in parallel ( Fig 3C ) . To assess a new model , the optimization algorithm simulated it ( i . e . , a virtual fly ) 100 times . An output threshold was then applied to the activity of the model’s output ( NOUT ) , resulting in a binary ( walking or stationary ) time-series . We then aggregated the walking and stationary interval durations from this time-series and compared these histograms to the target Canton-S basal locomotion histograms . The optimization process allowed us to discover model parameters that minimize the difference between virtual basal locomotor patterns and Canton-S basal locomotor patterns . This value , ‘Difference from Drosophila data’ , is a non-linear distance metric . It is therefore most intuitively understood by comparing it to values obtained when the full Drosophila dataset is compared to subsets of the same data ( S4A Fig ) . We generated models 50 times for each network size ( 1–5 neurons ) either in the absence or presence of neural activity fluctuations . This resulted in 500 candidate models . In the second step , we further filtered this population of models by identifying those that could also replicate the complex odor-evoked locomotor patterns of three genetically distinct DGRP strains ( Fig 2A ) . To mimic olfactory stimulation , we added virtual odor inputs to each model ( Fig 3D ) . In the final step , we analyzed the emergent dynamics of models that best matched both basal and odor-evoked locomotor patterns by ( i ) identifying the most common neural activity levels for each model using a ‘trajectory density’ representation and ( ii ) performing dynamical systems analysis of each model in the absence of activity fluctuations to identify equilibrium points in phase space: activity levels that the network tended to settle towards ( Stable ) or move away from ( Unstable ) ( Fig 3E ) . This revealed how activity fluctuations and dynamical properties allow models to reproduce complex Drosophila locomotor patterns . Using this approach we first asked if neural activity fluctuations were required to match the ongoing locomotor patterns of sensory-deprived Drosophila . Specifically , we tested if activity fluctuations were required by neural networks to reproduce Canton-S strain basal locomotor statistics ( Fig 3C ) . Indeed , we found that fluctuations and network dynamics were both required by models to match these in vivo data ( Fig 4A and 4B ) . Neither network dynamics alone ( Fig 4A , n = 250 models with 1–5 neurons , P < 0 . 001 , Wilcoxon Rank Sum Test; S4C–S4E Fig ) , nor a threshold applied to fluctuations in the absence of a network—the simplest , feed-forward AS model—performed as well ( S4B Fig ) . Fluctuation-driven networks with as few as two neurons accurately reproduced long and short time-scale Drosophila locomotor intervals ( Fig 4B & S5A Fig ) . Notably , many of these two-neuron networks had similar dynamics . Each had two stable equilibrium points ( S5B Fig ) and could be further classified post-hoc based on the frequency with which neural activity visited each stable point ( S5C Fig ) . Importantly , for all classes these equilibria did not represent a trivial mapping of two stable points onto two behavioral states ( walking and stationary ) : both equilibrium points were below the threshold for walking ( S5 Fig ) . Instead , in a manner akin to stochastic resonance in peripheral sensory pathways [4 , 6] , walking bouts were engaged when activity fluctuations caused neural activity near the Up state to rise above the threshold for walking . These results reveal how surprisingly compact fluctuation-driven neural network models can reproduce complex Drosophila basal locomotor statistics spanning both long and short time-scales . To identify the most explanatory of these network models , we tested their ability to match an unrelated behavioral dataset: the time-course of odor-evoked locomotor patterns across genetically distinct Drosophila strains . Of the original 98 DGRP strains , we selected three that spanned a large proportion of the behavioral variation that we observed ( Fig 4C ) . Next , to keep their emergent dynamics unchanged , we left all network parameters fixed for the best performing network of each dynamical class ( S5 Fig ) while optimizing odor input strength and the locomotor output threshold to best match the time-course of odor-evoked locomotor patterns for each Drosophila strain ( Fig 3D ) . The best Class 1 model ( Fig 4D ) could faithfully reproduce the time-course of odor-evoked locomotor patterns for every Drosophila DRGP strain . It exhibited pre-odor basal locomotion , sharp increases in locomotor frequency at odor onset followed by a slow decay , and reduced post-odor basal locomotion ( Fig 4E ) . Notably , not all models were as effective; the best Class 2 model failed to replicate odor-response dynamics ( S6 Fig , Root-Mean-Square Error or RMSE > 0 . 08 ) . The capacity for a given model to reproduce odor-evoked locomotor patterns was consistent across all three Drosophila strains ( S6 Fig ) . In addition to fluctuations with Gaussian statistics—a standard modeling approach ( e . g . , [54] ) –our best Class 1 model could also match DGRP strain A locomotor patterns when driven by fluctuations with Power law , or Ornstein-Uhlenbeck ( OU ) statistics [54–56] ( S6C and S6D Fig ) . We next investigated how our best Class 1 model ( S5D and S5E Fig ) reproduced Drosophila basal and odor-evoked locomotor patterns ( Fig 4E ) . We closely examined neural activity trajectories over time and identified several key roles for fluctuations . First , in the absence of fluctuations or sensory input , network activity remained trapped within stable equilibria below the threshold for walking ( Fig 5A , top & S1 Video ) . By contrast , in the presence of fluctuations , neural trajectories could periodically and unpredictably transit between stable equilibria and sometimes exceed the activity threshold for walking ( Fig 5A , bottom & S2 Video ) . Second , in our models , fluctuations were partially responsible for delayed changes in the dynamics of neural activity during and following odor removal . Rather than returning rapidly to stable equilibrium levels ( Fig 5A , top ‘Individual network activity’ ) , fluctuations caused neural activity to take a more tortuous path to these equilibria ( Fig 5A , bottom ‘Individual network activity’ ) . When averaged across a population of virtual flies , this results in a decay of locomotor frequency following odor stimulation ( Fig 5A , compare top and bottom ‘Population average’ ) . Fluctuations by themselves were not sufficient , however , to explain the reduction in basal locomotion following odor stimulation . Traditionally , reductions in neural activity are often attributed to physiological depression due to over-stimulation [57] . However , we observed that several DGRP strains showed little ( S7 Fig , RAL371 ) to no ( S7 Fig , RAL642 ) odor-evoked increases in locomotion but still exhibited reductions in locomotor frequency following odor presentation . Since our models could reproduce post-odor reductions in locomotor frequency without physiological depression , we used these models to investigate how changes in neural dynamics might account for shifts in basal locomotion . In our best Class 1 model matched to Drosophila strain A ( RAL57 ) , we discovered that odor stimulation caused a dramatic shift in network dynamics: the multistable network became monostable with a single subthreshold stable equilibrium point ( Fig 5D , ‘Odor Impulse’ ) . Therefore , although neural activity was initially pushed above the threshold by the odor ( Fig 5C , ‘Odor impulse’ , white arrowhead ) , its subsequent attraction to this new equilibrium point resulted in a decay of locomotor frequency even during odor stimulation ( Fig 5C , ‘Odor impulse’ , red arrowhead ) . When the odor was removed , although the network was once again multistable ( Fig 5D , ‘Decay’ ) , neural activity remained trapped near the odor-induced equilibrium point and took a long time to return to the original , basal equilibria ( Fig 5C , ‘Reduced basal locomotion’ , red arrowhead ) . This was due to both the diffusing influence of activity fluctuations as well as the structure of phase space . The same mechanisms allowed our best model to match Drosophila strain B ( RAL790 ) ( S8A–S8C Fig ) . Interestingly , even when two stable equilibrium points were retained , a substantial shift in the position of one stable point also resulted in decay dynamics matching those of Drosophila strain C ( RAL707 ) ( S8D–S8F Fig ) .
We have combined high-throughput behavioral analysis with automated neural network optimization to generate models that can reproduce complex Drosophila locomotor patterns . The resulting models , while not intended to inform the topology of Drosophila AS circuits , represent predictions about their emergent dynamics [19] . The key feature that allowed network models to reproduce Drosophila locomotor patterns was their dependence upon neural activity fluctuations . At first glance this may seem unsurprising given the complex nature of the behavioral data . However , fluctuations driving behavior in a simple feed-forward manner were insufficient ( S4 Fig ) . Instead fluctuations required coupling to neural dynamics with two attributes . First , our best models exhibited multistable dynamics reminiscent of persistent Up and Down states in vertebrate striatal [58] and cortical neurons [59] . Like stochastic state switching in genetic circuits [60 , 61] and stochastic mathematical models of C . elegans behavioral transitions [14] , in our models fluctuations allowed neural activity to escape stable equilibria and to rise above the threshold for walking . Fluctuations near the Up state led to rapid bursts of walking while residence near the Down state led to longer periods of inactivity [28] . This is strikingly similar to stochastic resonance mechanisms observed in the sensory periphery [4 , 6] . There , noisy fluctuations uncover otherwise subthreshold sensory information . Similarly , in our networks , we found that fluctuations make it possible for weak sensory input to drive locomotion ( Fig 6A ) . More generally , as for neurons in visual cortex [5] , we observed that activity fluctuations linearize an otherwise nonlinear relationship between sensory drive and behavioral output ( Fig 6B ) . Second , in our best models , odor stimulation drove changes in network dynamics by shifting the position and , sometimes , number of stable equilibrium points . After the odor was removed the dynamics returned to the pre-odor state . However , neural activity trajectories were delayed in returning to these original basins of attraction due to both the structure of phase space as well as the diffusive influence of fluctuations . This represents an alternative dynamical mechanism for shifting circuit activity that complements well-studied molecular mechanisms like physiological depression [57] . Dynamical properties of a network can be pushed into a new regime through stimulation and , although the network is identical before and after stimulation , it produces a very different output . Therefore , the difference in pre- and post-odor Drosophila locomotor frequency may be explained by changes in the dynamic trajectories of a fixed system , without any modifications of synaptic strength . One limitation of our study is the reliance upon one type of neural network model . CTRNN models are widely used and well-justified [19] but we expect that follow-up work using models with more [62] or less [63] detailed neural implementations can test the robustness of our predictions . In particular , highly-constrained models have been indispensible for understanding anatomically well-described systems like the pyloric network of the crustacean stomatogastric ganglion [64 , 65] . Although the anatomy and physiology of Drosophila AS circuits are not sufficiently well-characterized to build such detailed models , the body of anatomical and physiological data is growing [11 , 16 , 30 , 31] . This information will help to constrain neural network topologies [66–68] and to reveal how anatomical motifs contribute to the computation of Drosophila action timing . Our dynamical models inform a long-standing debate about the relative influence of neural fluctuations on animal behavior [12 , 24] . Unlike in peripheral sensory circuits , fluctuations in central circuits may largely arise from deterministic signals that occur naturally within highly interconnected networks of neurons . Intriguingly , our models predict that fluctuating activity in central action selection circuits may act in a stochastic resonance-like manner [5] to linearize the relationship between sensory drive and behavioral output . This suggests a potentially beneficial role for neural fluctuations in increasing the dynamic range of sensory responses in complex environments .
Drosophila Canton-S strains were used in odor-impulse ( Fig 1 ) and basal locomotion experiments ( Fig 3A ) . Drosophila melanogaster Genetic Reference Panel ( DGRP ) [49] strains were used in odor-impulse experiments ( Figs 2 & 4C and 4E ) . Experimental arenas ( 50 mm x 10 mm enclosures with a height of 1 . 3 mm ( Fig 1A–1C ) ) were designed using the 3D CAD software , SolidWorks ( Dassault Systèmes , Waltham , Massachusetts , USA ) and CNC machined from polyoxymethylene and acrylic glass . To backlight the arenas , we used a white LED panel ( Lumitronix , LED-Technik GmbH , Hechingen ) filtered with far-red semitransparent film ( Eastman Kodak , Rochester , NY USA ) , a color for which fruit flies are visually insensitive [44] . For olfactory stimulation , we used air bubbled ( Messer Schweiz AG , Lenzburg , Switzerland ) through either water or 10% acetic acid and controlled using Mass Flow controllers ( PKM SA , www . conab . gov . br ) at a regulated flow rate of 500 mL/min via computer controlled solenoid valves ( The Lee Company , Westbrook , CT , USA ) . We used a custom-fabricated circuit board and software [35] ( sQuid , http://lis . epfl . squid/ ) to simultaneously control valves and acquisition cameras ( Allied Vision Technologies , Stadtroda , Germany ) . We measured the flow of odor using a miniPID ( Aurora Scientific Inc . Aurora , Ontario , Canada ) . We performed experiments on adult female Drosophila raised at 25°C on a 12 h light:12 h dark cycle at 2–5 days post-eclosion . Experiments occurred either the morning or late afternoon Zeitgeber Time . Prior to experiments , flies were starved for 4–6 h in humidified 25°C incubators . For odor stimulation experiments , we measured the locomotor behaviors of between 131 and 242 flies ( median 205 flies ) . 98 DGRP strains were screened over the course of approximately 1 year . To minimize the effects of weekly and seasonal variation , we randomly selected and simultaneously screened groups of ~20 strains at a time . We repeated measurements for a single strain ( RAL208 ) four times over the course of the screen to confirm season-independent behavioral reproducibility . For basal locomotion behavior experiments , we recorded ten Canton-S strain flies for 30 min each , 5 h in total in a temperature-controlled room at 25°C under low red light illumination without air flow . Prior to the odor impulse experiment , flies were allowed to acclimate to the arena for 1 min . Subsequently , flies were first exposed to air throughout the arena for 1 min , then 10% acetic acid for 30 s , and finally , air for 90 s . Following an additional resting period with air flow for 90 s , we began the odor aversion experiment during which 10% acetic acid was presented on one side of the arena for 30 s and air on the other . This pattern alternated for an additional three cycles ( Fig 1C , ‘Odor aversion’ ) . We measured each fly’s position over time using Ctrax and Matlab ( The Mathworks , Natick , Massachusetts , USA ) Behavioral Microarray software scripts [37] . Afterwards we discretized the speed of a fly into a binary time-series using a hysteresis threshold . Based on previous studies [26 , 28 , 47 , 48] and confirmed by our own measurements , we considered a fly to have begun walking when its speed exceeded 1 mm/s . For walking flies , we considered locomotion to have terminated when the speed decreased below 0 . 5 mm/s ( a conservative value chosen to reduce the effects of measurement noise ) . We could thus classify speed in a binary fashion: walking or stationary ( Fig 1D ) . When averaged over a population of flies , we obtained a ‘Locomotor frequency’: the proportion of active flies at a given time point ranging from 0 when no flies are walking , to 1 when all flies are walking ( Fig 1F ) . To calculate the reproducibility of basal locomotor frequencies for genetically identical groups of flies , we randomly sampled a group of 65 flies ( selected as 50% of the flies from the strain with the smallest number of flies ) from the same strain . We repeated this sampling 100 times per strain to measure the mean and standard deviation ( Fig 2D ) . To calculate the correlation between odor-response time-courses for fly strains , we randomly sampled two populations ( groups A and B ) of 65 flies ( 0 . 5* the minimum population size ) from each strain . We then normalized Odor impulse traces ( 58th– 200th s of the odor impulse experiment ) between 0 and 1 . Comparisons were performed either between groups from the same strain or from different strains . Each comparison was performed 100 times and the mean R2 value was plotted ( Fig 2E ) . To calculate odor aversion , for each fly we measured the proportion of time spent in the air zone minus the time spent in the odor zone over the course of the odor aversion experiment . This was divided by the total time of the aversion experiment yielding a value between -1 ( always in the odor ) and 1 ( never in the odor ) ( Fig 2F and 2G ) . To assess the effects of chamber geometry on the durations of Drosophila walking and stationary bouts , we computed the distance correlation ( DC ) [50] between either ( a ) walking or ( b ) stationary intervals start/end positions and their corresponding interval durations ( S2B–S2D Fig ) . To increase the power of our analyses , we aggregated data-points by their positions with respect to one arena quadrant of the arena . To do this , data-points were folded twice—once along the Y-axis and a second time along the X-axis—to virtually aggregate them within one quarter of the arena . Consequently , all points near the four arena corners were considered near one another regardless of their corner of origin . For the sake of clarity , this repositioning is not shown in S2B–S2D Fig . To provide a reference metric for data with no correlation , we shuffled one of the variables and recomputed the DC . We repeated this process 100 times for each correlation . Additionally , to gain an intuitive understanding of DC values we took this shuffled dataset and introduced known correlations to incrementally larger subsets . We generated a dendrogram representation of the correlation between odor impulse time-series across all 98 DGRP strains ( Fig 4C ) using an agglomerative hierarchical clustering algorithm . The algorithm performed single-linkage clustering using a distance function of 1 minus the sample correlation between points . The length of each branch represents the correlation between the odor impulse time-series of two strains of flies . For subsequent model matching we selected at random one strain from each of the following correlation intervals: ρ≤ 0 . 9 , 0 . 9≤ρ≤0 . 95 , ρ>0 . 95 ( Fig 4C ) . We focused on only three DGRP strains due to the prohibitive computational time and resources required to optimize populations of virtual flies for each strain . For modeling we used Continuous-Time Recurrent Neural Networks ( CTRNNs ) . This modeling framework was chosen for its ability to mimic the dynamics of biological neural circuits [17] . A CTRNN with M neurons N1 , N2 , … , NM is defined by a system of ordinary differential equations ( ODE ) : dxidt=Fi ( t , x1 , x2 , … , xN ) = 1τi ( −xi+ ∑j=1Mwji⋅σ ( xj+bj ) +Ii ) , i={1 , … , M} The state of a neuron Ni is defined by the variable xi and is updated by an increment dxi inversely proportional to the time constant τi ∈[0 . 05 , 50] . The output of a neuron Ni is obtained by evaluating a sigmoid transfer function σ ( x ) = 1 ( 1+e−x ) on the state xi added to a constant bias bi ∈ [−10 , 10] . Neurons Nj and Ni are connected with synaptic links of weight wij ∈ [−20 , 20] . Furthermore , each neuron can receive an optional input Ii ( e . g . , odor input ) . We tested models up to five neurons in size since even three neurons are sufficient to exhibit a wide variety of dynamical behaviors including chaos [69] . To investigate the capacity of activity fluctuations alone to match Drosophila basal locomotion , we optimized a threshold ranging from -4σ to 4σ directly upon a Gaussian noise source [2 , 70] . For network models without fluctuations ( S4A Fig ) , the fluctuation input is set to zero ( Ii = 0 ) . For fluctuation-driven models , each neuron receives Gaussian noise with standard deviation wNOISE , i ( Ii = wNOISE , I G , where G ~ N ( 0 , 1 ) follows a Normal distribution ) . To test the effects of different noise sources , we substituted Gaussian noise with either 1/fα Power law noise ( CNOISE , https://people . sc . fsu . edu/~jburkardt/m_src/cnoise/cnoise . html ) , or Ornstein-Uhlenbeck ( OU ) noise . OU noise was implemented following the standard formulation of an OU process: dxt=θ ( μ−xt ) dt+σdWt x ( 0 ) =x0 where Wt represents the Wiener process . Using the best Class 1 model , we performed the odor impulse experiment while also optimizing the Power law parameter ( α ) , or OU parameters ( σ and θ ) . We simulated CTRNNs using a custom high-performance C++ implementation . Our implementation used an approximation of the sigmoid function σ ( x ) [71] to speed-up simulations . Furthermore , to decrease the computational load of the simulations the noise value G was generated every TNOISE ∈[0 . 01 , 1] s . For intermediate time-steps the noise value G was interpolated . Although this introduced correlations in the noise , the time-scale at which the noise value changed was orders of magnitude smaller than the time-scale at which the slowest dynamics occurred ( hundreds of seconds ) . ODEs regulating the evolution of the CTRNN were integrated using ODEINT [72] , a publicly available solver for ODE and a Runge-Kutta 4th-order method at a constant integration time-step of 10 ms ( five times smaller than the smallest time constant of a neuron ) . During the simulation of a neural network model , the trajectory of a model’s activity evolves over time from an initial condition , represented by the neuron states xi ( t0 ) , to eventually reside within the dynamical regime of the model ( e . g . , an equilibrium point , a limit cycle , etc . ) . In our experiments , we took two precautions to discard the long transients that sometimes occurred as trajectories passed from their initial positions into the model’s dynamical regime . First , we found the equilibrium points of the model ( i . e . , dxidt=0 ) . Then we generated initial conditions in the neighborhood of identified equilibrium points by sampling from a multivariate Gaussian distribution having an identity covariance matrix centered at the equilibrium points . Second , at the beginning of each simulation we integrated the model for 5 min of real time ( 3∙104 time-steps ) to discard dynamics during transit from initial conditions . To generate a binary time-series equivalent to Drosophila walking and stationary bouts , we applied a threshold THR ∈ ( 0 , 1 ) to the output of a neuron , arbitrarily chosen to be N0 ( referred to in the text as NOUT ) . Whenever the output of this neuron was greater than the threshold ( σ ( x1+b1 ) ≥ THR ) , the virtual fly was walking and otherwise it was stationary . We optimized neural network model parameters using a stochastic optimization method for tuning model parameters in an iterative manner [53] . First , we generated a population of models of a given size ( e . g . , three neurons ) . Next , we measured the activity of these models and transformed these into binary time-series comprising walking and stationary bouts using a threshold . Finally , walking and stationary bout durations were measured and aggregated into weighted variable bin-width histograms for walking or stationary intervals . Bin-widths were derived from the Drosophila target dataset . We compared these histograms to target histograms measured from Canton-S flies . After assessing this population of models , parameters were adjusted towards those of the best performing models in this and previous iterations . The stochastic nature of this process ensured that final models were not identical to one another . This process was repeated until model performance converged . We then studied the topological and dynamical properties of the best models found . In more detail , the NP = {wij , τi , bi , wNOISE , i , THR , TNOISE| i , j ∈ 1 , … , M} parameters of the CTRNN models were optimized using Particle Swarm Optimization ( PSO ) [73] . We used standard parameter settings c1 = 2 , c2 = 2 . The inertia parameter ω of the algorithm was modified during an optimization run , following an update rule ω ( t ) = 0 . 9 − 0 . 7 * tT to favor global search at the beginning of the optimization process and local search towards the end , where t is the current iteration and T = 200 is the maximum number of iterations . PSO operated concurrently on a set of M = 50 solutions . Therefore , a total of 104 solutions were evaluated during each optimization run . Each function evaluation required between 11 s and 30 s of computational time . We optimized CTRNN models on a cluster ( http://hpc . epfl . ch ) , using two nodes with 48 cores AMD Opteron 6176 ( Magny-Cours ) 2 . 3 GHz and 192 GB of memory . To optimize the odor input strength and output threshold of our best models to match Drosophila odor impulse locomotor dynamics , we measured the activity of the model’s output during 60 s of no stimulation ( basal locomotion ) , 30 s of odor stimulation , and then 120 s of no stimulation . We repeated this experiment while iteratively optimizing the few free parameters ( odor input strength per neuron , and output threshold ) to minimize the Root-Mean-Square Error ( RMSE ) between the target Drosophila odor-response time-series ( average of ~200 flies ) and the model’s odor-response time-series ( average of 200 virtual flies ) . The cost function assigns a score to each model evaluating how well it captures Drosophila locomotor patterns by comparing histograms generated by the model with histograms generated from Canton-S strain data . The comparison of these histograms is a crucial aspect of cost function design . While it is possible to use standard statistical tests such as distance measures between empirical cumulative distributions of data ( e . g . , the Kolmogorov-Smirnov test ) , these statistical tests can mislead the optimization process by assigning reduced importance to rare events . For distributions of time durations , it is evident that these approaches would fail , since rare events ( e . g . , long walking or stationary periods ) would be effectively ignored when comparing distributions . Therefore , we generated “weighted” histograms in which each bin was weighted by the duration it represents . For example , 10 walking events of 1 s duration and 1 event of 10s duration , would classically be represented as two bins of different “height” ( 10 and 1 respectively ) . In our weighted histograms these two bins have the same height ( 1s∙10 = 10s∙1 ) . We also wanted to remove empty bins . To do this we generated variable bin-width histograms . The boundaries of each bin for walking and stationary interval histograms were determined using Canton-S basal locomotion data ( S3 Fig ) . We used the same bin boundaries when evaluating each neural network model . For each cost function evaluation , we simulated a model K = 100 times . The model was started from K different initial conditions and simulated for 60 min of real time . For each of the K simulations , we selected at random with equal probability either the first or second 30 min of simulation , to mitigate overfitting of model behavior to the same trajectory and to foster model unpredictability . Each simulation produced a binary time-series representing walking ( 1 ) or stationary ( 0 ) behavior in a virtual fly . Thus , we computed the histogram for walking and for stationary periods using the data from all the selected K chunks . The generated histograms Hs , w for walking bouts and HS , I for stationary bouts were compared to their respective target Drosophila histograms HT , W and HT , I obtaining the distance between the histograms dw and dI . The distance measure between a target and simulated histogram is defined as: d ( HT , HS ) = ∑i=1B |R⋅hS ( i ) −hT ( i ) |⋅tB ( i ) B is the number of bins in the histograms , hs , ( i ) and hT , ( i ) returns the count for bin i in the synthetic Hs and target histogram HT and tB ( i ) returns the interval duration represented by bin i , here corresponding to the lower boundary of the bin . The scale factor R reconciles data obtained from simulations to available Drosophila data . We tested K = 100 simulated initial conditions per cost function evaluation . Therefore , R = 0 . 1 since we used data from 10 Canton-S strain flies . The cost function f maps a model m to a cost function value in [0 , ∞] , f:m → [0 , ∞] . For the sake of simplicity , we presented a normalized cost function value F . F is obtained by normalizing the cost function using the value FNorm that a virtual fly would have if it is always walking or always stationary such that F ( m ) =f ( m ) FNorm= dW+dIFNorm . A value of 0 corresponds to a perfect match , a value of 1 corresponds to the score of a virtual fly that is always walking or always stationary . Intermediate values ranging between 0 and 1 correspond to plausible distributions . Values higher than 1 generally result from models with periodic dynamics at very high frequencies . To derive an intuitive scale for cost function values , we evaluated values resulting from comparing subsets of Drosophila data with the full Drosophila dataset ( S4 Fig ) . We generated subsets of data by selecting at random the desired number of flies F and replicating the data from each selected fly 10/F times , rounded up to the closest integer . In cases with too much data ( F is not a divisor of 10 ) , we randomly removed walking or stationary bouts until we obtained a dataset with the same length as the full Drosophila basal locomotion dataset . We computed the boundaries of variable bin-width weighted histograms using a vector v containing walking or stationary interval durations from 5 h of Canton-S basal locomotion . This routine took as inputs the minimum resolution r of a bin ( the minimum separation between boundaries ) and the minimum count c of events to generate a bin . Next it generated histogram boundaries by recursively splitting the initial single-bin boundaries [0 , max ( v ) ] into smaller bins containing a minimum of c events and having minimum duration of r seconds ( S3 Fig ) . First , we applied this procedure to get bin boundaries for Drosophila walking and stationary bout duration histograms ( HT , W , HT , I ) . Then we used these same bin boundaries to compute both histograms from network model simulations ( HS , W , HS , I ) . Rather than studying their topologies ( e . g . , connectivity weights ) , we classified models by their dynamics [74–76] . In this way the behavior of a model with n neurons can be understood by observing the time evolution of its trajectory through an n-dimensional neural activity phase space . By studying the unfolding of phase space trajectories , one can identify common behavioral motifs among network models with widely different parameters . Using this formalism , features in phase space ( e . g . , attractors , limit cycles , and deterministic chaos ) provide a clear language with which to interpret and compare different models [74] . This dynamical systems perspective has been successful in classifying both artificial neural networks and biological neural populations [23] . To analyze the global dynamical behavior of our models and to classify closely related ones , we performed stability analysis on our models in the absence of Gaussian noise . First , the m equilibrium points x1¯ , x2¯ , … , xm¯ of the CTRNN were identified by numerically finding the roots of the system of differential equations F ( x¯ ) =0 using the multi-dimensional root finder provided by the Gnu Scientific Library ( http://www . gnu . org/software/gsl/ ) . The Jacobian matrix J of a CTRNN is defined as: J ( x ) = ( ∂F1 ( x ) ∂x1⋯∂F1 ( x ) ∂xM⋮⋱⋮∂FM ( x ) ∂x1⋯∂FM ( x ) ∂xM ) ∂Fi ( x ) ∂xj={wjiτiexj+bj ( 1+exj+bj ) 2if i≠j−1τi+wjiτiexj+bj ( 1+exj+bj ) 2if i=j We studied the stability of the CTRNN by linearizing the system in the neighborhood of each equilibrium point and computing the eigenvalues of the Jacobian matrix J of the CTRNN for each equilibrium point x¯ by solving det ( J ( x¯ ) −λΙ ) =0 . For a classification of stability given the equilibrium points’ eigenvalues refer to [74] . We obtained neural activity trajectory density maps by discretizing a plane described by two neuron states ( xi , xj ) into a grid of 103 × 103 cells ranging over the state values [−50 , 50] . Then we counted how many times a trajectory ( its projection onto ( xi , xj ) ) entered each cell . Density maps were generated by initializing models from 104 random initial conditions . The color of a cell in the trajectory density plot is related to the logarithm of the probability that a neural activity trajectory passes through that cell . To quantify differences in the dynamical behavior of two-neuron fluctuation-driven models and to classify them , we generated 1000 initial conditions around each stable equilibrium point and let trajectories evolve for 30 min . We then counted how many times a trajectory switched from one equilibrium point to the other . Trajectories of Class 1 models switched many times during each 30 min period while Class 2 models switched more rarely . Trajectories of Class 3 models did not switch equilibrium points . We computed Lyapunov exponents for models in the absence of fluctuations ( i . e . , Gaussian noise ) by integrating the variational equations dδdt of the CTRNN together with the original system: dδdt= ( dδ11dt⋯dδ1Mdt⋮⋱⋮dδM1dt⋯dδMMdt ) = J ( δ11⋯δ1M⋮⋱⋮δM1⋯δMM ) Following a standard procedure [77] , we integrated the original system together with the variational equations for TLYAP = 1000 time-steps . Then , we orthonormalized the perturbations using the Gram-Schmidt algorithm and computed the full spectrum of M Lyapunov exponents λ1 ≥ λ2 …≥ λM . The Kaplan-Yorke dimension [78] was then computed as DKY= k+∑i=1kλi|λk+1| , where k is the largest integer such that ∑i=1kλi≥0 . | The brain is never quiet . Even in the absence of environmental cues , neurons receive and produce an ongoing barrage of fluctuating signals . These fluctuations are well studied in the sensory periphery but their potential influence on central circuits and behavior are unknown . In particular , activity fluctuations in action selection circuits—neural populations that drive an animal’s actions from moment to moment—may strongly influence behavior . To shed light on the influence of activity fluctuations on action timing , we developed a computational approach for automatically generating neural network models that reproduce large-scale , high-resolution behavioral measurements of freely walking Drosophila melanogaster . We found that models require stochastic activity fluctuations to reproduce complex Drosophila locomotor patterns . Specific fluctuation-driven dynamics allow these models to produce short and long bouts of locomotion in the absence of sensory cues and to reduce locomotor activity after sensory stimulation . These results support a role for ongoing activity fluctuations in the timing of animal behavior and reveal how behavioral shifts can be brought about through changes in the dynamics of neural circuits . Thus , simple dynamical mechanisms may underlie complex patterns of animal behavior . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Fluctuation-Driven Neural Dynamics Reproduce Drosophila Locomotor Patterns |
Membrane bioenergetics are universal , yet the phospholipid membranes of archaea and bacteria—the deepest branches in the tree of life—are fundamentally different . This deep divergence in membrane chemistry is reflected in other stark differences between the two domains , including ion pumping and DNA replication . We resolve this paradox by considering the energy requirements of the last universal common ancestor ( LUCA ) . We develop a mathematical model based on the premise that LUCA depended on natural proton gradients . Our analysis shows that such gradients can power carbon and energy metabolism , but only in leaky cells with a proton permeability equivalent to fatty acid vesicles . Membranes with lower permeability ( equivalent to modern phospholipids ) collapse free-energy availability , precluding exploitation of natural gradients . Pumping protons across leaky membranes offers no advantage , even when permeability is decreased 1 , 000-fold . We hypothesize that a sodium-proton antiporter ( SPAP ) provided the first step towards modern membranes . SPAP increases the free energy available from natural proton gradients by ∼60% , enabling survival in 50-fold lower gradients , thereby facilitating ecological spread and divergence . Critically , SPAP also provides a steadily amplifying advantage to proton pumping as membrane permeability falls , for the first time favoring the evolution of ion-tight phospholipid membranes . The phospholipids of archaea and bacteria incorporate different stereoisomers of glycerol phosphate . We conclude that the enzymes involved took these alternatives by chance in independent populations that had already evolved distinct ion pumps . Our model offers a quantitatively robust explanation for why membrane bioenergetics are universal , yet ion pumps and phospholipid membranes arose later and independently in separate populations . Our findings elucidate the paradox that archaea and bacteria share DNA transcription , ribosomal translation , and ATP synthase , yet differ in equally fundamental traits that depend on the membrane , including DNA replication .
Reconstructing the traits of the last universal common ancestor ( LUCA ) requires constraining the relationships between the three domains of life , the archaea , bacteria , and eukaryotes . Recent phylogenetic studies show that eukaryotes are secondarily derived: they are genomic chimeras , arising from an endosymbiosis between a bacterium and an archaeal host cell [1]–[5] . The divergence between the two primary domains , the archaea and the bacteria , is now seen as the deepest branch in the tree of life [1] , [6]–[8] . The properties of LUCA are most parsimoniously those shared by bacteria and archaea . This leads straight to a serious paradox . Archaea and bacteria share core biochemistry , including the genetic code , transcription machinery , and ribosomal translation [9] , but differ for unknown reasons in fundamental traits including cell membrane [10] and cell wall [11] , glycolysis [12] , ion pumping [13] , and even DNA replication [14] . The differences in membrane lipids may be the key to this major unsolved problem in biology . Phospholipid side chains are typically isoprenoids in archaea and fatty acids in bacteria [15] . While this could reflect adaptive evolution [16] , archaea and bacteria also differ in the stereochemistry of the glycerol-phosphate headgroup [10] . Archaeal lipids have an sn-glycerol-1-phosphate ( G1P ) headgroup , while bacteria use the mirror structure sn-glycerol-3-phosphate ( G3P ) ( Figure 1 ) . There is no persuasive selective explanation for these opposite stereochemistries [10] , [13] , [17] . The enzymes involved , glycerol-1-phosphate-dehydrogenase ( G1PDH ) in archaea and glycerol-3-phosphate-dehydrogenase ( G3PDH ) in bacteria , bear no phylogenetic resemblance , suggesting they arose independently [10] . If so , then LUCA did not possess a modern membrane—a seemingly improbable conclusion , given the central importance of membranes to cells [10] , [17] , [18] . Set against this paradoxical difference in membrane composition is the universality of membrane bioenergetics [19] . Essentially all cells power ATP synthesis through chemiosmotic coupling , in which the ATP synthase ( ATPase ) is powered by electrochemical differences in H+ or Na+ concentration across membranes [20] . The ATPase is universally conserved [21] and shares the same deep phylogenetic split as the ribosome , implying that both were present in LUCA [22]–[24] . The deepest branches in the tree of life are entirely populated by autotrophs [1] , [6] , [7] , [12] , [25] , which also depend on chemiosmotic coupling to drive carbon metabolism via proteins such as the energy-converting hydrogenase ( Ech ) and ferredoxin [26] . But there are serious objections to the idea that LUCA was chemiosmotic . Pumping protons across membranes requires sophisticated proteins , which are only useful in membranes impermeable to protons [27] . Unlike the ATPase , no ion pumps are universally conserved [13] . The pathways for heme and quinone synthesis ( the major cofactors of respiratory proteins ) also differ in archaea and bacteria , although their distribution is complicated by lateral gene transfer , as is reconstruction of the phylogenetic origins of respiratory ion pumps [13] . But it seems likely that both lipid membranes and active pumping are evolutionarily distinct in archaea and bacteria [9] , [11] . It is hard to reconcile these fundamental differences with the universality of the ATPase . On the face of it , LUCA was chemiosmotic , yet did not have a modern phospholipid membrane or active ion pumps . A possible resolution is that LUCA exploited natural ( geochemically sustained ) proton gradients [18] , [28] , [29] . However , the hypothesis that natural proton gradients could drive carbon and energy metabolism in LUCA , in the absence of active ion pumps , faces a serious drawback . Because fluids are electrically balanced , the transfer of H+ ions down a concentration gradient , from an acid solution into a cell , transfers positive charge into the cell , generating a membrane potential that opposes further influx . The system swiftly reaches electrochemical ( Donnan ) equilibrium , in which electrical charges and concentration differences are offset [30] . Equilibrium is death: natural proton gradients could only drive carbon and energy metabolism in LUCA if such equilibrium is avoided—in effect , if protons accumulating inside a cell can leave again . Membrane permeability could be critical to maintaining disequilibrium in any system with continuous flow , as leaky membranes impose less of a barrier to the continued flux of H+ , OH− , and other ions [19] . The feasibility of this hypothesis depends on the dynamics of ion fluxes that are unknown . We have therefore built a model to estimate quantitative differences in free energy ( −ΔG ) across lipid membranes exposed to natural proton gradients . We consider a cell exposed simultaneously to alkaline fluids and relatively acidic water ( Figure 2 ) . Our model is independent of any particular setting , but requires continuous laminar flow with limited mixing ( as found in microporous alkaline hydrothermal vents [18] , [19] , [24] , [31]–[33] and potentially other environments ) , allowing sharp gradients of several pH units to be maintained across short distances of 1–2 µm . In general , we assume that the external pH does not change on either side of the cell , as external fluids are replenished by continuous flow from large reservoirs ( e . g . , hydrothermal fluids or the ocean ) , but we do also consider mixing . Protons enter the cell through membrane proteins , and directly through the lipid phase of the membrane . The overall rate of proton influx depends on the difference in proton concentration and electrical charge ( upon proton entry ) between the outside and inside of the cell , the kinetics of the membrane protein ( e . g . , ATPase ) , the number of membrane proteins ( given as a proportion of the surface area ) , the proton permeability of the lipid phase of the membrane , and the rate of loss of protons from inside the cell ( see Materials and Methods ) . For simplicity , we assume that gradient-exploiting membrane proteins are only present on the acid face of the cell . Proton loss from inside the cell therefore depends on the rate of influx of OH− from alkaline fluids , which neutralize protons within the cell , and the rate of loss of protons across the lipid phase to the alkaline exterior ( Figure 2 ) . We also consider membrane permeability to Na+ , K+ , and Cl− ions , which move charge , and hence influence the electrochemical potential difference and the rate of proton flux . By calculating the overall proton flux on the basis of these parameters , we estimate changes in the steady-state proton concentration inside the cell relative to the outside , giving the free energy ( −ΔG ) available to drive carbon and energy metabolism . Our findings allow us to propose a new and tightly constrained bioenergetic route map leading from a leaky LUCA dependent on natural proton gradients , to the first archaea and bacteria with highly distinct ion-tight phospholipid membranes . These bioenergetic considerations give striking insights into the nature of LUCA , and the deep divergence between archaea and bacteria .
The model shows that cells with 1% ATPase in a proton-tight membrane with glycerol-phosphate headgroups ( giving an H+ permeability <10−5 cm/s , like extant archaea and bacteria [34] ) , collapse natural proton gradients within seconds ( Figure 3A and 3B ) . The magnitude of the pH gradient depends on the environmental setting . To constrain possibilities we considered pH values commensurate with alkaline hydrothermal vents , but the same principles apply to any other setting with dynamic pH gradients across short distances . The early oceans may have been mildly acidic , as low as pH 5 , and alkaline fluids as high as pH 11 [35] but we conservatively set a 3 pH-unit gradient , with the “acid” at pH 7 and alkaline fluids at pH 10 . Nonetheless , collapse of the gradient was evident in proton-tight membranes across a range of gradients ( Figure 3B ) . Protons enter through the ATPase faster than they can exit or be neutralized by OH− , so H+ influx rapidly reaches electrochemical equilibrium . In contrast , leaky protocells ( equivalent to fatty-acid vesicles without glycerol phosphate headgroups ) in a 7∶10 pH gradient with 1% ATPase in the membrane retain nearly all the free energy available , having a −ΔG only ∼17% lower than an open system ( i . e . , a single membrane containing the same number of membrane proteins , separating a continuous flux of acid and alkaline fluids; Figure 3A ) . This is because proton flux through the ATPase is ∼4 orders of magnitude faster than through the lipid phase , even with a high proton permeability of 10−2 cm/s ( based on the kinetics of proton-flux through the ATPase , see Materials and Methods and Table S1 ) . Leaky cells in natural proton gradients of 3 pH units therefore have sufficient free energy to drive ATP synthesis . Even leaky cells are sensitive to the amount of membrane protein , with higher proportions of ATPase collapsing the gradient ( Figure 3C ) . In this case , the rate of H+ entry through ATPase covering 10%–50% of the membrane surface area is substantially faster than the rate of clearance of H+ from inside the cell ( and reaction with OH− ) , collapsing −ΔG . However , 1%–5% ATPase in a leaky membrane ( 10−3 cm/s ) retains a −ΔG of close to 20 kJ/mol ( Figure 3A and 3C ) . With 3–4 protons translocated per ATP synthesized ( Table S1 ) , this gives a −ΔG for ATP hydrolysis of 60 to 80 kJ/mol , similar to modern cells and sufficient to drive intermediary biochemistry , including aminoacyl adenylation in protein synthesis [36] . This assumes the same stoichiometry as the modern ATPase ( 3–4 protons per ATP ) . Because the kinetics of early enzymes would arguably not have been as honed by evolution as their modern equivalents , we used 10% of modern proton flux rates . However , this difference in efficiency actually has limited impact on the model compared with modern flux rates ( Figure S1 ) ; increasing the stoichiometry of the ATPase has a similarly small effect ( Figure S2 ) . We did not estimate rates of ATP synthesis , as that would require additional assumptions about concentrations of ATP , ADP , and phosphate , as well as the rates of ATP consumption and growth; these are almost impossible to constrain at present . The same principles apply to carbon metabolism . We consider whether the membrane protein Ech could drive carbon reduction by H2 in natural proton gradients . Ech uses the proton-motive force to drive carbon metabolism in some archaea and bacteria via the reduction of ferredoxin [26] . As with the ATPase , cells with 1%–5% Ech in the membrane retain most of the free energy available from a 7∶10 pH gradient ( Figure 3D ) . Higher concentrations of Ech ( 10%–50% ) collapse −ΔG even more than the ATPase , as the rate of proton flux through Ech is double that of the ATPase , and its surface area is slightly smaller , so there are more proton pores per unit surface area ( Table S1 ) . Such high concentrations of Ech or ATPase are in any case improbable , and not relevant to modern cells , but demonstrate the range of conditions in which natural gradients can in principle drive carbon and energy metabolism . Given a 7∶10 pH gradient , it is therefore feasible to have 1%–5% Ech and 1%–5% ATPase in the membrane , driving both carbon and energy metabolism in cells with leaky membranes . But incorporation of either G1P or G3P glycerol-phosphate headgroups ( found in archaea and bacteria respectively ) , or racemic mixtures of archaeal and bacterial lipids ( which , surprisingly , are as impermeable to protons as standard membranes [37] ) , are not favored because they reduce the proton permeability of the membrane and so collapse the energetic driving force . Glycerol-phosphate headgroups in particular decrease proton permeability , as they prevent fatty acid flip-flop across the membrane ( see Discussion ) . If leaky cells with low amounts of ATPase and Ech ( 1%–5% ) are viable in natural proton gradients , but cells with phospholipid membranes are not , then the evolution of active pumping becomes a paradox: pumping protons across a proton-permeable membrane does not increase free energy ( −ΔG ) , because the protons immediately return through the lipid phase of the membrane . We demonstrate this using a model of a simple H2-dependent proton pump ( equivalent to Ech operating in reverse , as found in some simple bacteria and archaea [26] ) . We find that in a 7∶10 pH gradient −ΔG falls as membrane permeability decreases from 10−2 to 10−6 cm/s ( Figure 4A ) . −ΔG here depends on two factors: active pumping and the natural pH gradient . As membrane permeability falls , the contribution of the natural pH gradient also falls , undermining −ΔG . In contrast , the benefit of pumping increases , as fewer protons return through the lipid phase . The balance between these two factors depends on the strength of pumping ( which equates to the number of pumps , i . e . , % surface area ) . However , even when the pump occupies 5% of the membrane surface area , pumping H+ gives no advantage until a modern permeability of 10−5 cm/s , i . e . , there is no benefit to improving permeability across 1 , 000-fold ( Figure 4A ) . Thus , there is no selective pressure to drive either the origin of pumping or the evolution of modern proton-tight membrane lipids in natural proton gradients . Pumping Na+ works better across leaky membranes ( Figure 4B ) , as lipid membranes are ∼6 orders of magnitude less permeable to Na+ than to H+ ( due to fatty acid flip-flop; see Discussion ) [34] . However , as with pumping H+ , −ΔG falls as the membrane becomes less permeable , because the contribution of the natural gradient also declines , giving no continuous selective advantage to pumping Na+ . With a proton permeability <10−5 cm/s , there is no advantage to pumping Na+ at a pump density of 1%–5% surface area compared with leaky protocells lacking a pump . Pumping Na+ therefore offers an initial advantage , but there is no sustained selection pressure for tightening membrane permeability to modern values . Neither is there any advantage in the absence of a natural pH gradient . This would apply to the evolution of chemiosmotic coupling in any setting that lacks natural gradients . Under this condition , pumping either H+ ( Figure 4C ) or Na+ ( Figure 4D ) offers a steadily amplifying advantage as membrane permeability falls . However , without an external pH gradient , −ΔG is low , the rise with reduced permeability is meager , and remains well below the 15–20 kJ/mol required by modern cells to drive processes like aminoacyl adenylation for protein synthesis [36] . Cells with permeable membranes ( 10−2–10−4 cm/s ) are therefore unlikely to be viable unless powered by some other means [23] , [27] . Hence in either the presence or absence of pH gradients , there is no sustained selection pressure to drive the evolution of either active pumping or modern membranes . Our model shows that leaky membranes were necessary to survive in natural proton gradients but that pumping protons across such leaky membranes is fruitless . Yet free-living cells require ion-tight membranes and active pumping for bioenergetics . What drove this evolutionary change ? We hypothesize that a necessary first step was adding Na+ as an additional “promiscuous” coupling ion . A non-electrogenic sodium-proton ( 1Na+/1H+ ) antiporter ( SPAP ) , found widely in cells , could in principle use a natural H+ gradient to generate a biochemical Na+ gradient . Exchanging Na+ for H+ does not alter membrane potential directly , but the difference in lipid permeability of the two ions alters ion flux , with significant effects on −ΔG . Because lipid membranes are ∼6 orders of magnitude less permeable to Na+ than to H+ [34] , fewer Na+ ions can pass through the lipid phase of the membrane , so the Na+ gradient does not dissipate as quickly . As a result , Na+ flux becomes more tightly funneled through membrane proteins , improving the coupling of the membrane without changing its chemistry [19] . Because the H+ gradient is sustained geochemically , SPAP simply adds a Na+ gradient to the natural H+ gradient . Taking advantage of mixed Na+/H+ gradients requires promiscuity of membrane proteins for both ions , which is indeed the case for several contemporary bioenergetic proteins , including the ATPase [38] and Ech [26] ( see Discussion ) . SPAP increases proton influx , initially lowering −ΔG ( Figure 5A ) . However , the coupled extrusion of relatively impermeable Na+ ions increases −ΔG by ∼60% within minutes in a 7∶10 gradient , saturating when SPAP covers ∼5% of the membrane surface area ( Figure 5A ) . Importantly , the free energy available from pH gradients declines in more acidic conditions . −ΔG is greatest with a 7∶10 gradient , lower at 6∶9 , and nearly zero with a 5∶8 gradient , despite the three-order-of-magnitude correspondence ( Figure 5B ) . This asymmetry arises because H+ and OH− flux through the membrane depends on concentrations as well as gradient size [39] . Comparatively high acidity and low alkalinity increases H+ influx but hinders OH− neutralization , collapsing the H+ gradient . Because Na+ extrusion through SPAP depends on the natural H+ gradient , SPAP increases −ΔG in relatively alkaline regions ( pH 7–10 and 6–9 ) but has little effect on −ΔG in more acidic regions ( pH 5–8 ) , making acidic regions less favorable for colonization , even with SPAP . When the rate of H+ influx does not collapse the proton gradient , SPAP significantly increases −ΔG , allowing survival in shallower pH gradients ( Figure 5C ) . If a −ΔG>15 kJ/mol is needed for growth , 5%–10% SPAP allows cells to grow in 50-fold weaker gradients ( e . g . , 8 . 5∶10; Figure 5C ) , a significant ecological advantage , facilitating spread . This general principle holds whatever the actual value of −ΔG needed for growth in early cells . The advantage offered by SPAP also applies to fluctuations in gradient size ( e . g . , due to mixing of fluids ) . −ΔG plainly fluctuates with the pH front even in the presence of SPAP; but SPAP still increases −ΔG even with considerable fluctuations in pH ( Figures S3 and S4 ) . Crucially , SPAP is also a necessary preadaptation for the active pumping of protons , and for decreasing membrane permeability towards modern values . Whereas pumping H+ in the absence of SPAP gives no sustained benefit in terms of −ΔG , the presence of SPAP in a leaky membrane allows pumping of H+ to pay dividends . −ΔG now markedly increases with decreasing permeability ( Figure 6A ) , for the first time giving a sustained selective advantage to higher levels of pumping and tighter membranes . As in the absence of SPAP , −ΔG depends on two factors: the power of the pump ( which varies with the proportion of surface area covered ) and the natural pH gradient . As membrane permeability falls , the contribution of the natural pH gradient also falls . While 1% pump cannot sustain −ΔG when the contribution of the gradient is lost , 5% H+ pump gives a steadily amplifying advantage to lowering membrane permeability ( Figure 6A ) . Much the same applies to pumping Na+ ( Figure 6B ) . The lower permeability of Na+ gives an initial benefit to pumping this ion , but this is lost as the membrane becomes tighter , even with 5% pump ( Figure 6B ) . This lower efficacy is due to the much higher external concentration of Na+ . With active pumping , tighter membranes , and SPAP , cells could colonize more acidic regions ( Figure S5 ) , regions with weaker gradients ( Figure 6C ) , and ultimately survive in the absence of a gradient altogether ( Figure 6D ) . With no external pH gradient , SPAP interconverts efficiently between H+ and Na+ , making it feasible to pump either ion ( Figure 6D ) . These cells are now modern in that they have a fully functional chemiosmotic circuit and proton-tight membranes , and hence could evolve the traits required to leave the natural gradients for the external world . We propose that this process occurred independently in divergent populations that had spread widely using SPAP to colonize regions with weak gradients ( see Discussion ) . These independent populations subsequently evolved into the two main branches of early life , the archaea and bacteria [1] .
Our model suggests a resolution to the long-standing paradox that membrane bioenergetics are universal , but membranes are fundamentally different [19] . In so doing , the model gives a striking insight into the deep evolutionary split between archaea and bacteria . It reveals that the late and divergent evolution of impermeable membranes could have arisen as a simple outcome of LUCA's exploitation of natural proton gradients . Our model applies in principle to any environment in which sharp differences in proton concentration are sustained over short distances , one concrete example being alkaline hydrothermal vents [18] , [24] , [31]–[33] . Given the membrane proteins Ech and ATPase , we show that natural proton gradients could have sustained both carbon and energy metabolism in LUCA ( Figure 3C and 3D ) . However , to do so , LUCA had to have very leaky membranes , the only way to avoid deadly electrochemical equilibrium ( Figure 3A ) . Our results indicate that LUCA did not have modern phospholipids . The addition of glycerol-phosphate headgroups is specifically precluded by the requirement for high proton-permeability in natural gradients ( Figure 3A ) . Addition of a glycerol-phosphate headgroup reduces proton permeability substantially , as the polar headgroup cannot cross the hydrophobic interior of the membrane [40] . In contrast , lipid membranes composed of mixed amphiphiles , including fatty acids , have much greater proton permeability , through “flip-flop . ” In flip-flop , protonation of a negatively charged fatty acid eliminates its charge , allowing the neutral residue to migrate across the hydrophobic membrane to the inside [41] . Deprotonation on the relatively alkaline interior rapidly dissipates proton gradients , explaining the high proton permeability of fatty acid vesicles [41] . Flip-flop is not possible with Na+ , which remains ionic in the presence of a negatively charged amphiphile , hence its lower permeability [34] . Our results indicate that LUCA was sophisticated in terms of genes and proteins , but did not have a modern phospholipid membrane . However , LUCA must have had a stable lipid bilayer membrane composed of mixed amphiphiles , probably including fatty acids and isoprenes ( some of which are found in both archaea and bacteria [15] ) . A lipid bilayer membrane is undoubtedly necessary for the function of membrane proteins such as the ATPase and Ech [42] . The actual permeability of membranes is difficult to determine experimentally , as H+ permeability depends in part on the permeability of counter-ions , and therefore varies with the composition of solutions used in measurements . Values of phospholipid membrane H+ permeability range from 10−4 cm/s [43] to 10−10 cm/s [44] , [45] , with a consensus favoring a value of between 10−4 to 10−6 cm/s [34] . The H+ permeability of fatty acid vesicles is higher , in the range of 10−2 to 10−3 cm/s or even greater [41] . These values are for standard temperature , 25°C ( 298 K ) . Both H+ and Na+ permeability rise substantially with temperature , by approximately 1 order of magnitude for every 20°C increase between 20°C and 100°C , although the actual values depend on membrane composition [45] . The membrane permeability also depends on the kinetics of membrane proteins , which likewise vary with temperature . We have used standard temperature for enzyme kinetics . How these values would vary with temperature is difficult to estimate , as the kinetics of enzymes adapted to low temperatures would differ from those in thermophiles if placed in the same membrane at the same temperature . However , our simulations of efficiency and stoichiometry ( Figures S1 and S2 ) suggest that the effect should be substantially less than that of lipid permeability . We are therefore confident that our results apply generally , despite these uncertainties . We stress that our argument relates to the principle of energy transduction in natural proton gradients , not to the specific values used for membrane permeability . The key point is that leaky membranes were essential to transduce natural proton gradients , and there was no advantage to be gained by the evolution of proton-tight phospholipid membranes , whether at low or high temperatures . This leads to a paradox . Pumping either H+ or Na+ over leaky membranes gives no sustained advantage when membrane permeability is lowered over 1 , 000-fold ( Figure 4A and 4B ) . That precludes the evolution of either active ion pumps or modern proton-tight membranes in a LUCA dependent on natural proton gradients . We hypothesize that the evolution of a SPAP was the key innovation that favored the independent evolution of active ion pumps and phospholipid membranes in bacteria and archaea . SPAP has two major effects that made this possible . First , SPAP favors divergence , through adding a Na+ gradient to the geochemically sustained H+ gradient . Because lipid membranes are much less permeable to Na+ ions , these preferentially flow back through membrane proteins , thereby increasing free-energy availability by up to 60% ( Figure 5A ) . For this additional Na+ gradient to be useful , membrane proteins must be promiscuous for Na+ and H+ , which is the case for some primitive ATPase enzymes [38] and for Ech [26] . While the ATPase generally specializes either for H+ or Na+ today , only a few amino acid changes are required to switch from one form to the other [46] . Phylogenetic trees of the ATPase suggest that the H+-dependent and Na+-dependent forms are interleaved , implying greater promiscuity in early evolution [24] . The reason probably relates to the close similarity in ionic radius and charge of Na+ without its hydration shell ( the form in which it usually passes through membrane proteins ) and the hydronium ion , H3O+ ( the form in which H+ is most commonly found in solution ) . Thus it is likely that addition of a Na+ gradient to a natural H+ gradient by SPAP would indeed increase the free energy available to the cell as a usable electrochemical difference . This enabled cells to survive in 50-fold lower gradients ( Figure 5C ) , or with intermittent gradients and mixing ( Figures S3 and S4 ) , facilitating spread and divergence . Second , SPAP gives a continuous selective advantage to actively pumping protons even across a leaky membrane ( Figure 6A ) . This advantage amplifies steadily as membrane permeability decreases , all the way towards values for largely impermeable modern membranes ( Figure 6A ) . Our results lead us to suggest that the SPAP is ancestral and must have been present in LUCA . Phylogenetic analysis is consistent with this prediction . BLAST [47] results show a match for archaeon Methanocaldococcus jannaschii's Mj1275 SPAP to an equivalent or very closely related protein in at least one member of 35 out of all 37 prokaryotic phyla reported to date ( Table S2 ) . The two bacterial clades with a missing match are to date single-member phyla whose only known species may have either lost the gene over time , had it diverge beyond observable similarity to the M . jannaschii ortholog , or simply have not been fully annotated in the databases yet . This confirms our prediction of the universality of SPAP in spite of the stark dissimilarity in membranes , and paves the way for closer phylogenetic analysis of these antiporters and related proteins . We note that the early operation of SPAP would have the effect of lowering the intracellular Na+ concentration substantially below ambient seawater concentration , explaining how cells that evolved in the ocean could nonetheless be optimized to low intracellular Na+ and high K+ concentration . The operation of antiporters ( and possibly symporters ) , driven by natural proton gradients , could in principle have modulated intracellular ionic composition to the low-Na+–high-K+ characteristic of most modern cells , leading to selective optimization of protein function without the need for a specific terrestrial environment with a particular ionic balance [27] . These considerations are also consistent with the universality of SPAP across prokaryotic phyla . Our analysis demonstrates that active ion pumps almost certainly arose after SPAP , and only then did selection favor the evolution of ion-tight membranes with glycerol phosphate headgroups . Given that SPAP in itself facilitated the spread and colonization of regions with shallower ( Figure 5C ) or more intermittent gradients ( Figures S3 and S4 ) , pumping is expected to arise independently in more than one population , as observed [13] , [19] . Only when active ion pumping had evolved was there any benefit to incorporating glycerol-phosphate headgroups , thereby reducing membrane permeability ( Figure 6A ) . Phospholipid biosynthesis involves nucleophilic attack on the prochiral carbonyl center of dihydroxyacetone phosphate [10] . This can be achieved from either side of the molecule , giving rise to opposite stereochemistries of the central carbon in glycerol phosphate ( Figure 1 ) . The enzymes involved , G1PDH in archaea and G3PDH in bacteria appear to have taken these alternatives by chance in independent populations that had already evolved distinct ion pumps . Thus we posit that the ancestors of archaea and bacteria evolved both ion pumps and phospholipid membranes independently , the latter on the basis of a simple binary choice in the orientation of nucleophilic attack on dihydroxyacetone phosphate . We conclude that the membranes of LUCA were necessarily leaky , composed of mixed amphiphiles ( including fatty acids ) but lacking glycerol-phosphate headgroups . Fatty-acid vesicles have long been considered plausible protocells because of their simplicity , stability , and dynamic ability to grow [48]–[50] , but are generally thought unsuitable for chemiosmotic coupling due to their high proton permeability [27] , [51] . Leaky membranes have therefore generally been interpreted in terms of heterotrophic origins of life [52] . In contrast , we find that high proton permeability was in fact indispensable to drive both carbon and energy metabolism in natural proton gradients , consistent with autotrophic origins; and this requirement for leaky membranes in turn precluded the early evolution of phospholipid membranes ( Figure 7 ) . Our model offers a selective basis for the universality of membrane bioenergetics and the ATPase , while elucidating the paradoxical differences in membranes and active ion pumps . The deep disparity between archaea and bacteria in carbon and energy metabolism [19] , [53] , and in membrane lipid stereochemistry [10] , reflects two independent origins of active pumping in divergent populations ( Figure 7 ) . The conclusion that LUCA had leaky membranes , and that modern phospholipid membranes evolved later and independently in archaea and bacteria , provides a framework for interpreting other dichotomies between archaea and bacteria . The late and independent evolution of glycolysis but not gluconeogenesis [12] is entirely consistent with LUCA being powered by natural proton gradients across leaky membranes . Several discordant traits are likely to be linked to the late evolution of cell membranes , notably the cell wall , whose synthesis depends on the membrane [11] and DNA replication [14] . In the latter case , the fingers-thumb-palm motif at the active site of DNA polymerase enzymes [54] and the structure of the replication fork [55] are superficially similar in archaea and bacteria , yet most proteins involved in DNA replication , including the principal replicative polymerases , bear no phylogenetic resemblance [14] , [56] , [57] . That implies either independent origins [14] or inscrutably deep divergence compared with the plainly homologous transcription and translation machinery [56] , [57] . Because the bacterial replicon is attached to the plasma membrane during cell division [58]–[60] , this complex presumably arose after ( or coevolved with ) the bacterial membrane , which must have driven a deep phylogenetic disparity , even if DNA replication had arisen in LUCA . Thus key facets of the fundamental split between archaea and bacteria could be linked to the late origin of phospholipid membranes , for these bioenergetic reasons . While it is difficult to prove that these bioenergetic factors really did account for the deepest branch in the tree of life , they do offer a robust and testable framework that can explain the paradoxical character of LUCA and the stark differences between archaea and bacteria .
Cells were modeled half embedded in the alkaline fluid , with the other half exposed to the comparatively acidic fluid . This produced an inward proton gradient from the acidic side , sustained by the constant neutralization with OH− from the alkaline side ( Figure 2 ) . Only the two external pH values are fixed; the internal pH is then arrived at in response to the fluxes of H+ and OH− across the membrane , which in turn depends on permeability , the respective concentrations of each ion , and flow through the membrane proteins . Equation 1 describes the various ways in which protons could enter or leave the cell at every time step: by simple diffusion across the membrane on either side , and through any of the membrane proteins , namely the ATPase , SPAP , pump , or Ech . ( 1 ) Total concentrations of H+ and OH− were calculated at every time step by neutralization and equilibration to the dissociation constant of water . External fluids were assumed to be part of comparatively large bodies of water , with their acidity and alkalinity sustained by large-scale geological or meteorological processes; thus their concentrations of H+ , OH− , and other ions were assumed constant . Analogous equations were used for other ions . Table S1 describes the parameters chosen for the results presented in the text , unless otherwise stated . We anticipate that enzymes could not have reached their current reaction rate values at the early stages of evolution that we are considering , so for the results presented in the main text we have consistently used 10% of the current turnover rates referenced in Table S1 . A series of results using modern ( 100% ) turnover rates are presented in Figure S1 for comparison . Membrane flux JS of a neutral substance S was modeled using a traditional passive diffusion equation [61] ( 2 ) where PS is the permeability of the substance , A is the area of the membrane , and [S]ext and [S]int are the external and internal concentrations respectively . To account for the effect of membrane potential Δψ on the behavior of charged particles , ion diffusion was modeled using the Goldman-Hodgkin-Katz flux equation [39] , [62] ( 3 ) where zs is the charge of the substance , F and R are the Faraday and gas constants , respectively , and T is the temperature . Electrical membrane potential Δψ was in turn modeled using the Goldman-Hodgkin-Katz voltage equation [39] , [62] ( 4 ) for the concentration of each cation and anion present . Internal protons and hydroxide were equilibrated using the dissociation constant of water . The available free energy ΔG from the H+ gradient was modeled with the traditional equation used by Mitchell [20] ( 5 ) An analogous equation was used for the Na+ gradient . The power of ATP to catalyze biochemical reactions in the cell comes not specifically from hydrolysis of the molecule itself but from the degree to which the ATP/ADP ratio is shifted from thermodynamic equilibrium; that is , the energy available from ATP hydrolysis varies with the ATP/ADP ratio [30] . The equilibrium constant and thus the energy required for ATP synthesis depends on the concentrations of ADP , phosphate , and magnesium ion , as well as pH [20] , [30] , but with the exception of pH these values are unknown for the systems modeled , as are rates of ATP hydrolysis . We have therefore used Equation 5 to calculate the size of the electrochemical gradient ( ΔG ) as a function of the H+ and Na+ gradients and the electrical membrane potential ( Δψ ) . The steady-state ΔG in turn gives an indication of how far from equilibrium the ATP/ADP ratio could be pushed . With 3–4 protons translocated per ATP , a steady-state ΔG of −20 kJ/mol is large enough to drive the ATP/ADP ratio to a disequilibrium of 10 orders of magnitude , equivalent to that found in modern cells [30] . We calculated steady-state ΔG as a function of the size of the H+ and Na+ gradients and the electrical membrane potential ( Δψ ) between the acid fluid and the inside of the cell . These factors in turn depend on steady-state rates of proton flux into and out of the cell via the lipid phase of the membrane ( specified by its H+ and Na+ permeability and surface area ) and through the ATPase . We calculated the maximum flux of H+ or Na+ flux through the ATPase on the basis of the maximum possible number of ions translocated per second . Maximum ion flux is based on the reported maximum turnover rate of ATPase ( Table S1 ) , i . e . , the maximum number of ATP molecules that each ATPase unit can synthesize in one second when operating at top speed , multiplied by 3 . 3 , the number of H+ or Na+ required to synthesize 1 ATP ( Table S1 ) . This number was then multiplied by the number of ATPase units in the system , estimated from the membrane surface area assigned to this protein in each simulation ( e . g . , 1% , 5% , etc . ) and the reported surface area of the membrane-integral FO subunit ( Table S1 ) . We further assumed that the actual flux rate of H+ and Na+ through the ATPase would also depend on the driving force itself , ΔG , i . e . , the size of the H+/Na+ gradient and the electrical membrane potential ( Δψ ) . We assumed that the ATPase obeys hyperbolic Michaelis-Menten dynamics , commonly the case in enzyme kinetics [63] and reported for the ATPase [64] , such that H+/Na+ flux asymptotically approaches the maximum turnover rate when the driving force is large , again assuming that flux rate is unconstrained by ADP availability . Thus , increasing ΔG beyond a threshold cannot increase H+/Na+ flux beyond the maximum turnover rate , so flux rate must saturate . The hyperbolic curve was modeled to reach saturation slightly beyond −20 kJ/mol , a gradient large enough to drive the ATP/ADP ratio to 10 orders of magnitude disequilibrium in modern cells [30] and equivalent to a membrane potential of around 200 mV , close to a maximum for modern lipid membranes , given the low capacitance of thin lipid membranes . This number , between zero and one , was finally multiplied by the maximum flux of H+ or Na+ , described above , to determine the influx of each of the two ions through the ATPase . When added to H+/Na+ flux rates across the lipid phase , the steady-state H+/Na+ flux through the ATPase gave a steady-state ΔG available to drive ATP synthesis . Full promiscuity of the ATPase to Na+ and H+ was assumed , with preference of one ion over the other depending solely on their respective gradient sizes . The Ech was modeled analogously . SPAP was modeled to respond to the H+ and Na+ gradients , exchanging ions in the direction determined by the larger of the two gradients . Δψ was assumed to affect SPAP speed but not direction [65] . Since the H+ gradient is reversed on the alkaline side , we assumed the SPAP , ATPase , and Ech operated only on the acidic side . The pump was modeled as a generic system able to extrude either H+ or Na+ , dependent on the concentration of hydrogen gas ( H2 ) , and responding to the opposing gradient , thus making it easier to pump protons against an alkaline fluid , and more difficult against an acidic fluid . A running example of the code can be found at http://www . ucl . ac . uk/~rmhknjl/research/membranedivergence This code can be run directly from any typical computer with an Internet connection . Additionally , it can be downloaded and run locally ( at no significant increase in speed ) from http://github . com/UCL/membranedivergence The primary amino acid sequence of the M . jannaschii Mj1275 Na+/H+ antiporter ( SPAP ) was obtained from the NCBI protein sequence database . Mj1275 is one of three known SPAP genes in archaeon M . jannaschii , the other two being Mj0057 and Mj1521 [66] . The first belongs to the NapA family , while the latter two are in the NhaP family . Phylogenetic analysis was performed on these three genes as well as the two common Escherichia coli SPAP genes , NhaA and NhaB [67] , [68] , using the NCBI-BLASTp server [47] with standard parameters , filtering for each prokaryotic phylum ( considering each of the proteobacteria as a separate clade ) . Results for Mj1275 showed the highest hit rate ( Table S2 ) , possibly hinting that it is closest to the ancestral form of the SPAP . Results for the other genes are not shown . | The archaea and bacteria are the deepest branches of the tree of life . The two groups are similar in morphology and share some fundamental biochemistry , including the genetic code , but the differences between them are stark , and rank among the great unsolved problems in biology . The composition of cell membranes and walls is utterly different in the two groups , while the mechanism of DNA replication seems unrelated . We address a specific paradox , giving new insight into this deep evolutionary split: membrane bioenergetics are universal , yet the membranes themselves are not . We resolve this paradox by considering the energetics of a hypothetical last universal common ancestor ( LUCA ) in geochemically sustained proton gradients . Using a quantitative model , we show that LUCA could have used proton gradients to drive carbon and energy metabolism , but only if the membranes were leaky . This requirement precluded ion pumping and the early evolution of phospholipid membranes . We constrain a pathway leading from LUCA to the deep divergence of archaea and bacteria on the basis of incremental increases in free-energy availability . We support our inferences with comparative biochemistry and phylogenetics , and show why the late evolution of modern membranes forced divergence in other traits such as DNA replication . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biochemistry",
"astrobiology",
"cell",
"biology",
"earth",
"sciences",
"genetics",
"biology",
"and",
"life",
"sciences",
"chemistry",
"microbiology",
"computational",
"biology",
"evolutionary",
"biology",
"physical",
"sciences",
"biophysics",
"geochemistry"
] | 2014 | A Bioenergetic Basis for Membrane Divergence in Archaea and Bacteria |
The regulation of mucosal immune function is critical to host protection from enteric pathogens but is incompletely understood . The nervous system and the neurotransmitter acetylcholine play an integral part in host defense against enteric bacterial pathogens . Here we report that acetylcholine producing-T-cells , as a non-neuronal source of ACh , were recruited to the colon during infection with the mouse pathogen Citrobacter rodentium . These ChAT+ T-cells did not exclusively belong to one Th subset and were able to produce IFNγ , IL-17A and IL-22 . To interrogate the possible protective effect of acetylcholine released from these cells during enteric infection , T-cells were rendered deficient in their ability to produce acetylcholine through a conditional gene knockout approach . Significantly increased C . rodentium burden was observed in the colon from conditional KO ( cKO ) compared to WT mice at 10 days post-infection . This increased bacterial burden in cKO mice was associated with increased expression of the cytokines IL-1β , IL-6 , and TNFα , but without significant changes in T-cell and ILC associated IL-17A , IL-22 , and IFNγ , or epithelial expression of antimicrobial peptides , compared to WT mice . Despite the increased expression of pro-inflammatory cytokines during C . rodentium infection , inducible nitric oxide synthase ( Nos2 ) expression was significantly reduced in intestinal epithelial cells of ChAT T-cell cKO mice 10 days post-infection . Additionally , a cholinergic agonist enhanced IFNγ-induced Nos2 expression in intestinal epithelial cell in vitro . These findings demonstrated that acetylcholine , produced by specialized T-cells that are recruited during C . rodentium infection , are a key mediator in host-microbe interactions and mucosal defenses .
The recently revealed degree of integration between the nervous and immune systems are remarkable [1] . While it is well accepted that neurotransmitters can act on immune cells to alter cell activation and consequently host immune response , recent evidence demonstrates that select immune cell populations not only respond but can also produce neurotransmitters . Among these immune cells are the CD4+ T-cells that express choline acetyltransferase ( ChAT ) , the enzyme required for acetylcholine ( ACh ) biosynthesis [2–4] . These T-cells are crucial intermediaries between the nervous and immune system , functioning to relay neuronal signals and prevent aberrant immune cell activation . Neural inhibition of inflammation can inhibit innate immune cell function in preclinical models of inflammatory bowel disease [5] , rheumatoid arthritis [6] , ischemia reperfusion injury [7 , 8] , and post-operative ileitis [9] . Immune regulation in this pathway requires norepinephrine ( NE ) released from neurons to activate β2 adrenergic receptors ( β2AR ) on ChAT+ T-cells causing the release of ACh [2] . Mucosal immunity is crucial to restricting access of commensal and pathogenic bacteria to the host . Host defenses are comprised of overlapping mechanisms that bind , flush away , exclude , or kill pathogenic enteric bacteria [10] . These roles are in part fulfilled by differentiated intestinal epithelial cells ( IECs ) that not only act as a physical barrier , but also produce and release mucus [11] , bactericidal antimicrobial peptides [12 , 13] , and free radicals such as nitric oxide ( NO ) that are bactericidal or bacteriostatic [14 , 15] . Loss of these protective mechanisms can result in aberrant immune responses to otherwise innocuous commensal bacteria , increased mucosal inflammation , or susceptibility to infection . In addition , mucosal homeostasis and host-resistance to pathogens is dependent on composition of the intestinal microbiota , with bacterial species that can reduce , or enhance susceptibility to pathogens including Citrobacter rodentium [16–18] . Physiological processes that govern these mechanisms of host defense and host-bacterial interactions are therefore paramount to the health of the host . In the gastrointestinal tract , ACh enhances mucosal protection by controlling IEC functions ranging from release of mucus and antimicrobial peptides to increasing ion and fluid secretion [12 , 19 , 20] . Together , these mechanisms of mucosal defense maintain homeostatic interactions between the host and commensal microbiota , while limiting access of pathogens such as C . rodentium . Although the source of ACh regulating these functions of IEC has long been attributed to ChAT+ secretomotor neurons within the gastrointestinal tract , we and others have previously described ChAT+ T-cells that serve as essential non-neuronal sources of ACh [2–4] . This unique source of ACh appears to participate in mucosal immunity and host commensal interactions . As evidence of this , conditional ablation of ChAT in T-cells was found to reduce production of antimicrobial peptides in the small intestine of naïve mice , and induce changes in the jejunal but not colonic microbiota composition [13] . Despite these key observations , the role of ACh released from specialized T-cells during enteric infection is unknown . With these issues in mind , we have used ChAT-GFP reporter mice , and conditional ablation of ChAT in T-cells to assess the role of T-cell derived ACh in host mucosal immune function during C . rodentium infection . Using this approach , we have identified that ChAT+ T-cells are recruited to the colon during C . rodentium infection , and that conditional ablation of ChAT in T-cells significantly increases C . rodentium burden in the colon . This increased susceptibility to infection is due to decreased expression nitric oxide synthase isoform 2 in IEC , with ACh acting to enhance IFNγ-induced gene transcription .
Mice used in this study are on a C57BL/6 background and were originally purchased from Jackson laboratories ( Bar Harbor , ME ) , including CXCR5-/- , ChAT-GFP ( B6 . Cg-Tg ( RP23-268L19-EGFP ) 2Mik/J ) ) , ChATf/f and LCK . Cre to establish a breeding colony . ChAT T-cell conditional knockout ( cKO ) mice were produced by breeding ChATf/f and LCK . Cre mice to generate LCK . Cre- ChATf/f ( WT ) and LCK . Cre+ ChATf/f ( cKO ) mice . This breeding scheme permitted use of littermate cKO and WT controls . At 6–8 weeks of age , mice were gavaged with either LB , or Citrobacter rodentium ( 108 CFU ( colony-forming unit ) , strain DBS100 , generously provided by Dr . Andreas Baumler ) . In a subset of experiments , colitis was induced by administration of dextran sodium sulfate ( DSS , 3% v/v ) in the drinking water for 5 days followed by normal water for 3 days as previously published [21] . All procedures were approved by the Institutional Animal Care and Use Committee at UC Davis under protocol number 20170 in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals . Mice were euthanized by CO2 asphyxiation followed by cervical dislocation according to American Veterinary Medical Association guidelines for collection of tissues . IEC and lymphocytes were isolated from the colonic lamina propria according to a standard protocol [22 , 23] . In brief the colons from mice were removed , cut open along the mesenteric boarder , washed in PBS , before being place in Hanks buffered salt solution ( HBSS ) with EDTA ( 5 mM ) to remove IEC . For experiments to ascertain the amount of NOS2 expressed by IEC , dissociated cells suspensions were incubated for 20 min with Fc blocking ( CD16/32 ) antibody ( TONBO Biosciences , San Diego , CA ) and subjected to staining using Live/Dead aqua viability stain ( ThermoFisher Scientific , Waltham , MA ) , anti- CD45-FITC ( ThermoFisher Scientific ) and -EpCAM-BV421 ( BD Biosciences , San Jose , CA ) . IEC were then subjected to fixation and permeabilization using Fix/Perm buffer , and washed with permeabilization according to manufacturer’s instructions ( BD Biosciences ) . Intracellular staining was then conducted using anti-NOS2-APC ( ThermoFisher Scientific ) . Lymphocytes were isolated from colonic tissues subjected to removal of IEC as above , followed by digestion of 1 cm cut fragments in Collagenase IV ( 40 mg/ml , Sigma Aldrich ) . Released lymphocytes were incubated in RPMI media ( 10% FBS Pen/strep L-glutamine ( ThermoFisher Scientific ) ) supplemented with PMA/Ionomycin ( Cell Stimulation Cocktail 1X , eBioscience ) , with Golgi-Stop ( BD Biosciences ) for 4 hours . Samples were subjected to staining with Live/Dead Aqua ( ThermoFisher Scientific ) , anti -CD3-BV605 ( BioLegend , San Diego , CA ) , anti-CD4-APC ( TONBO Biosciences ) , followed by fixation and intracellular staining for with anti -IL-22-PerCPeFluo710 ( ThermoFisher Scientific ) , -IL-17A-PE ( BD Biosciences ) , -IFNγ-PECy7 ( ThermoFisher Scientific ) and -GFP-AF488 ( BioLegend ) . All samples were then acquired on a BD LSRII , and assessed using FlowJo ( Treestrar , Ashland OR ) . A 1 cm segment of colon was cut and placed into a pre-weighed microcentrifuge tube to determine tissue weight . Samples were homogenized in sterile PBS using a 5 mm sterile stainless-steel bead ( Qiagen , Germantown MD ) in a bead beater ( Qiagen ) . Sample homogenates were then diluted in sterile PBS , and ( 100 μL ) plated onto MacConkey agar plates , and colonies counted after 16 h of incubation at 37°C . Analysis of gene expression was performed by quantitative real-time PCR ( qRT-PCR ) as previously described [3] . Briefly , a 1 cm long segment of colon was homogenized in Trizol ( Invitrogen , Carlsbad , CA ) using a 5 mm stainless steel bead in a bead beater ( Qiagen ) . RNA was extracted as directed by manufacturer’s instructions , with isolated RNA dissolved in ultrapure H2O ( Invitrogen ) . Synthesis of cDNA was performed using an iSCRIPT reverse transcriptase kit ( Bio-Rad , Hercules , CA ) , and Real time qPCR was performed for the following targets using the indicated primer pairs from Primerbank [24]: IL-1β forward 5’-CTGTGACTCATGGGATGATGATG-3’ , reverse 5’-CGGAGCCTGTAGTGCAGTTG-3’ , IL-6 forward 5’- TAGTCCTTCCTACCCCAATTTCC-3’ , reverse 5’- TTGGTCCTTAGCCACTCCTTC-3’ , IL-17A forward 5’- TTTAACTCCCTTGGCGCAAAA-3’ , reverse 5’-CTTTCCCTCCGCATTGACAC-3’ , IL-22 forward 3’-ATGAGTTTTTCCCTTATGGGGAC-5’ , reverse 3’-CTGGAAGTTGGACACCTCAA-5’ , IFNγ forward 5’- GCCACGGCACAGTCATTGA-3’ , reverse 5’- TGCTGATGGCCTGATTGTCTT-3’ , Tnfα forward 5’-CCCTCACACTCAGATCATCTTCT-3’ , reverse 5’-GCTACGACGTGGGCTACAG-3’ , NOS2 forward 5’-GTTCTCAGCCCAACAATACAAGA-3’ , reverse 5’- GTGGACGGGTCGATGTCAC-3’ , Chi3L3 forward , CTCTGTTCAGCTATTGGACGC , reverse 5’- CGGAATTTCTGGGATTCAGCTTC-3’ , MRC1 forward 5’- CTCTGTTCAGCTATTGGACGC-3’ , reverse 5’- CGGAATTTCTGGGATTCAGCTTC , Retnla forward 5’- CCAATCCAGCTAACTATCCCTCC- 3’ , reverse 5’- ACCCAGTAGCAGTCATCCCA -3’ , Arg1 forward 5’-CTCCAAGCCAAAGTCCTTAGAG-3’ , reverse 5’-AGGAGCTGTCATTAGGGACATC-3’ , Cxcl13 forward 5’- GGCCACGGTATTCTGGAAGC-3’ , reverse 5’- GGGCGTAACTTGAATCCGATCTA-3’ , Ccl1 forward 5’- GGCTGCCGTGTGGATACAG-3’ , reverse 5’- AGGTGATTTTGAACCCACGTTT-3’ , Ccl8 5’- TCTACGCAGTGCTTCTTTGCC-3’ , reverse 5’- AAGGGGGATCTTCAGCTTTAGTA -3’ , Ccl19 forward 5’- GGGGTGCTAATGATGCGGAA-3’ , reverse 5’- CCTTAGTGTGGTGAACACAACA-3’ , Ccl21 forward 5’- GTGATGGAGGGGGTCAGGA-3’ , reverse 5’- GGGATGGGACAGCCTAAACT-3’ , Defa3 forward 5'- GAGAGATCTGGTATGCTATTG-3' , reverse 5'- AGCAGAGTGTGTACATTAAATG-3' , Defa5 forward 5'- TCAAAAAAGCTGATATGCTATTG-3' , reverse 5'- AGCTGCAGCAGAATACGAAAG-3' , Defa20 forward 5'- GAGAGATCTGATATGCTATTG-3' , reverse 5'- AGAACAAAAGTCGTCCTGAG-3' , Defa21 forward 5'- GAGAGATCTGATCTGCCTTTG-3' , reverse 5'- CAGCGCAAAAAAGGTCCTGC-3' , Defa22 forward 5'- GAGAGATCTGATCTGCCTTTG-3' , reverse 5'- CAGCGCAAAAAAGGTCCTGC-3' , Defa23 forward 5'- GAGAGATCTGGTATGCTATTG-3' , reverse 5'- AGCAGAGCGTGTATATTAAATG-3' , Defa24 forward 5'- GAGAGATCTGGTATGCTATTG-3' , reverse 5'- AGCAGAGCATGTACAATAAATG-3' , Defa26 forward 5'- ATTGTAGAAAAAGAGGCTGTAC-3' , reverse 5'- AGCAGAGTGTGTACATTAAATG-3' , Itln1 forward 5'- ACCGCACCTTCACTGGCTTC-3' , reverse 5'- CCAACACTTTCCTTCTCCGTATTTC-3' , Reg3g forward 5'- CCTCAGGACATCTTGTGTC-3' , reverse 5'- TCCACCTCTGTTGGGTTCA-3' , Lyz1 forward 5'- GCCAAGGTCTACAATCGTTGTGAGTTG-3' , reverse 5'- CAGTCAGCCAGCTTGACACCACG-3' , Actb forward 5’-GGCTGTATTCCCCTCCATCG-3’ , reverse 5’- CCAGTTGGTAACAATGCCATGT-3’ . CIIta forward 5’-TGCGTGTGATGGATGTCCAG-3’ , reverse 5’-CCAAAGGGGATAGTGGGTGTC-3’ , Irf1 5’-ATGCCAATCACTCGAATGCG-3’ , reverse 5’-TTGTATCGGCCTGTGTGAATG-3’ . Amplification and data acquisition were conducted using a QuantStudio6 ( ThermoFisher Scientific ) . Data were analyzed using the delta delta CT method normalizing gene expression to Actb in each sample followed by normalization to experimental control sample . Fresh fecal samples from mice were collected and stored at -80°C until analysis . Pellets were extracted with nano-pure water ( 10 mg/mL ) and gently agitated overnight at 4°C . The homogenized samples were centrifuged at 21 , 000 g for 5 min . Supernatants ( 100 μl ) were transferred and centrifuged at 21 , 000 g again for 20 min . For each sample , 20 μl of the supernatant was mixed with 20 μl of 100 mM N- ( 3-Dimethylaminopropyl ) -N0-ethylcarbodiimide hydrochloride ( 1-EDC HCl ) ( Sigma-Aldrich , cat . # E7750 ) in 5% pyridine ( Sigma-Aldrich cat . # 270407 ) and 40 μL of 200 mM 2-Nitrophenylhydrazine ( 2-NPH ) ( Sigma-Aldrich , cat . # N21588 ) in 80% acetonitrile ( ACN ) ( Sigma-Aldrich ) with 50 mM HCl . The mixture was incubated at 40°C for 30 min . After reacting , 400 ml of 10% ACN was added to the solution . Then 1 μl of the solution was injected into an Agilent 6490 triple quadruple mass spectrometer for analysis . Chromatographic separations were carried out on an Agilent C18 stationary phase ( 2 . 1 x 50 mm , 1 . 8 um ) column . Mobile phases were 100% ACN ( B ) and water with 10% ACN ( A ) . The analytical gradient was as follows: time = 0 min , 10% B; time = 4 . 5 min , 10% B; time = 10 min , 35% B; time = 10 . 1 min , 85% B; time 11 . 6 min , 90% B; time 12 min , 90% B . Flow rate was 0 . 3 ml/min and injection volume was 1 μL . Samples were held at 4°C in the autosampler , and the column was operated at 40°C . The MS was operated in selected reaction monitoring ( SRM ) mode , where a parent ion is selected by the first quadrupole , fragmented in the collision cell , then a fragment ion selected for by the third quadrupole . Product ions , collision energies , and cone voltages were optimized for each analyte by direct injection of individual synthetic standards . The MS was operated in positive ionization modes with the capillary voltage set to 1 . 8 kV . Source temperature was 200°C and sheath gas temperature 200°C . Gas flow was 11 L/min , sheath gas flow was 7 L/min , and collision gas flow was 0 . 2 mL/min . Nebulizer pressure ( nitrogen ) was set to 25 psi . Argon was used as the collision gas . A calibration curve was generated using authentic standards for each compound . Colonic tissue specimens were fixed in 10% normal buffered formalin for 24 h prior to gradual dehydration in ethanol , embedded in paraffin and 6 μm thick cross sections were cut onto glass slides . Slides with tissue sections were de-paraffinized and rehydrated according to standard protocols , stained with hematoxylin and eosin to allow for evaluation of histopathology . Crypt lengths were measured using bright field microscopy on these sections with FIJI ( Fiji Is Just ImageJ ) [25] , measuring at least 20 crypts per animal . Slides with colonic tissue section were also used for confocal analysis with antibodies raised against specific proteins of interest according to standard protocols . In brief , after slides were de-paraffinized and rehydrated , antigen retrieval was performed in citrate buffer ( 10 mM , pH 6 . 0 , 30 min . , 95°C ) . After blocking in 5% BSA and normal donkey serum , samples were incubated in primary antibody overnight ( 16 h 4°C ) . Primary and secondary antibodies used are detailed in Table 1 . Slides were washed extensively ( 3 x 5 mins ) in TBS-tween20 and incubated in appropriately labeled secondary antibodies ( Invitrogen ) for 1 h at room temperature , washed , counterstained with DAPI in TBS-tritonX100 0 . 1% v/v , washed and mounted in Prolong gold ( Invitrogen ) . Staining using anti-mouse CDH1 ( E-cadherin ) was revealed using a mouse on mouse kit according to manufacturer’s instructions ( Vector laboratories , Burlingame , CA ) . Slides were imaged on a Leica SP8 STED 3X confocal microscope with a 63X 1 . 4 NA objective . Areas larger than the field of view of the objective were acquired using a tiling approach , whereby adjacent images were acquired with a 10% overlap . Analysis of standard confocal data sets was performed by opening Leica image format files in Imaris Stitcher ( v9 . 0 , Bitplane Scientific ) to fuse overlapping fields of view together . These reconstructed areas were then analyzed using Imaris software . Expression of NOS2 in IEC was determined by creating a mask based on regions of CDH1 staining ( i . e . IEC ) that contained DAPI+ cells . This defined region was then interrogated for the number of IEC present , and the intensity of NOS2 staining . Counting of DAPI+ Ki67+ IEC , or T-cells ( CD3+ DAPI+ ) cells were performed in a similar manner in 3–5 fused fields of view from each animal counted . Following excision of the intestine , segments of colon were cut along the mesenteric border to allow for mounting in the Ussing chamber ( Physiologic Instruments , San Diego , CA ) . Tissues were maintained in oxygenated Kreb’s buffer consisting ( in mM ) of: 115 NaCl , 1 . 25 CaCl2 , 1 . 2 MgCl2 , 2 . 0 KH2PO4 and 25 NaHCO3 at pH 7 . 35 ± 0 . 02 and maintained at 37°C . Additionally , glucose ( 10 mM ) was added to the serosal buffer as a source of energy , which was balanced osmotically by mannitol ( 10 mM ) in the mucosal buffer . Agar–salt bridges were used to monitor potential difference ( PD ) across the tissue , and to inject the required short‐circuit current ( Isc ) to maintain the PD at zero by an automated voltage clamp . Data from the voltage clamp ( i . e . Isc , and PD ) was continuously acquired using acquisition software ( Physiologic Instruments ) . Baseline Isc values were obtained after equilibrium had been achieved approximately 15 min after the tissues were mounted . Isc , an indicator of active ion transport , was expressed in μA/cm2 . After tissues reached stable short-circuit current for 15 minutes , stimulation of ion secretion was induced by addition of the muscarinic receptor agonist carbachol ( 20 μM ) . After returning to baseline the adenylate cyclase activating compound Forskolin ( [FSK] , 20 μM ) was added and response recorded . CMT-93 cell line ( ATCC cat . CCL-223 ) from an induced carcinoma of mouse rectum was cultured in DMEM supplemented with 10% heat-inactivated fetal bovine serum ( FBS supplemented with Pen Strep ( ThermoFisher Scientific cat . 15070063 ) , and maintained at 37°C in a humidified atmosphere with 5% CO2 . Cells were seeded in a 6-well plate at a density of 1×106 cell/well and incubated for 48 h . The medium was replaced with serum-free DMEM for 2 h . Then , cells were treated with IFNγ 1 ng/ml ( Peprotech cat . 315–05 ) , Carbachol ( 100 μM , Sigma Aldrich ) or both for 3h . Cells were collected and stored at -80°C on Trizol . RNA extraction , cDNA synthesis and qPCR were performed as described above . Data were analyzed using one-way analysis of variance ( ANOVA ) in Prism ( Graphpad , San Diego CA ) , with a P value of less than 0 . 05 denoted as significant .
Only sparse numbers of ChAT+ T-cells have been observed in the intestine of naïve mice [13] , however the potential role of ChAT+ T-cells in the mucosal immune response during enteric bacterial infection has not been established . To assess if ChAT+ T-cells are recruited during infection , ChAT-GFP+ mice were infected with C . rodentium and the number of CD3+ ChAT-GFP+ T-cells determined by confocal microscopy on days 6 , 10 , 21 , and 30 post-infection ( p . i . ) . Mice infected with C . rodentium had a significant increase in the number of CD3+ChAT-GFP+ T-cells in the colon beginning 10 days p . i . which persisted until 30 days p . i . ( Fig 1A & 1B ) . In order to characterize these recruited cells , flow cytometry was conducted on isolated lamina propria lymphocytes ( LPL ) . These colonic lamina propria ChAT-GFP+ T-cells ( Single , Live , CD3+ , CD4+ ) 10 days post-C . rodentium infection produced IFNγ , IL-17A , and IL-22 ( Fig 2A ) . Quantification revealed that ChAT-GFP+ T-cells express more IFNγ , IL-17A , and IL-22 by MFI ( mean fluorescence intensity ) compared to ChAT-GFP- T-cells ( Fig 2B ) . Despite this , it is important to note that the frequency of ChAT-GFP+ T-cells actively producing IFNγ and IL-17A were significantly less compared to ChAT-GFP- T-cells . ChAT-GFP+ IL-22+ T-cell population appears to be persistent in the naïve colon and does not increase significantly during infection . These data demonstrate that the ChAT-GFP+ T-cells are not unique to Th1/Th17/Th22 T-cells subsets , and can be polarized to these three different phenotypes ( Fig 2 ) . In contrast to the recruitment induced by C . rodentium infection , induction of colonic inflammation by the chemical irritant DSS failed to increase the number of ChAT-GFP+ T-cells compared to naïve control , despite evidence of overt inflammation ( S1A Fig ) . Together , these results suggest that ChAT+ T-cells are a specific component of the host response to C . rodentium infection and their recruitment is driven by specific signals and rather than a simple consequence of intestinal inflammation . The functional role of T-cell-derived ACh during C . rodentium infection was determined using a T-cell conditional knockout ( cKO ) approach . Accordingly , infected ChAT T-cell cKO mice had increased CFU/g of C . rodentium in colonic tissue at day 10 p . i . as compared to infected WT mice ( Fig 3A ) . To determine if the increased bacterial burden of C . rodentium resulted in altered localization of the pathogen in the colon , confocal microscopy analysis using antibodies directed against C . rodentium was performed ( Fig 3B ) . Compared to WT , we observed increased C . rodentium in the colonic lumen , adjacent to IEC ( CDH1+ DAPI+ ) , and the presence of microcolonies within the colonic crypts in ChAT T-cell cKO mice . Despite the increased bacterial burden in ChAT T-cell cKO mice , no significant increase in the number of proliferating ( DAPI+ CDH1+ Ki67+ ) IEC cells ( Fig 3D &3E ) , histopathological damage , or crypt hyperplasia was observed compared to infected WT mice ( Fig 3C ) . Together these data indicate that T-cell derived ACh is a critical component of host defense during C . rodentium infection but does not influence epithelial barrier integrity or effect crypt hyperplasia . To assess what factors could contribute to recruitment of these CD3+ ChAT+ T-cells during C . rodentium infection , we performed qRT-PCR for chemokines that are cognate ligands for previously identified receptors expressed by this population of T-cells [4 , 26] . The pattern of Cxcl13 expression closely mirrors the recruitment of ChAT+ T-cells ( S2A Fig ) , with significantly increased expression beginning 10 days and lasting until day 30 p . i . while significantly increased expression of Ccl1 , Ccl8 , Ccl19 and Ccl21 occurred between 21 and 30 days p . i . , well after recruitment of ChAT+ T-cells began . Critically , infection of mice deficient in CXCR5 , the cognate receptor for CXCL13 , did not experience increased C . rodentium bacterial burden or pathology ( S2B Fig ) . In addition , assessment of intestinal physiology using Ussing chambers revealed no significant differences in conductance , baseline or evoked short-circuit current responses to carbachol or forskolin in naïve WT or ChAT T-cell cKO mice ( S3 Fig ) . To determine the immunological consequences of conditional ablation of ChAT in T-cells during C . rodentium infection , qRT-PCR was conducted on colon from LB control and infected WT and ChAT T-cell cKO mice for proinflammatory gene expression . At day 10 p . i . , expression of Il-1β , Il-6 , and Tnfα were significantly increased in C . rodentium infected mice compared to LB control mice ( Fig 4 ) . Expression of these cytokines was significantly enhanced 10 days p . i . in the ChAT T-cell cKO mice compared to WT infected animals . In contrast , expression of Ifnγ , Il-17a , Il-22were increased 10 days p . i . to a similar extent in WT and ChAT T-cell cKO mice ( Fig 4 ) . These data indicate that ablation of ChAT in T-cells can significantly alter the host immune response to C . rodentium , but in a manner that does not alter local Th1 , Th17 , or Th22 responses . As ChAT T-cell conditional knockout mice were previously observed to have reduced expression of antimicrobial peptides [13] , we questioned if this could result in an increased C . rodentium burden . Using qRT-PCR we observed no significant differences in antimicrobial peptide expression in the small intestine or colon ( S4 Fig ) in naïve WT and ChAT T-cell cKO mice . As expected , colonic expression of RegIIIγ was significantly increased after C . rodentium infection in both WT and ChAT T-cell cKO mice , however there was no difference between the two genotypes in the terminal ileum or colon ( S4A & S4B Fig ) . As the commensal microbiota actively produces bioactive metabolites , we assessed if production of short-chain fatty acids ( SCFA ) was different in WT compared to ChAT T-cell cKO mice . Mass spectrometry revealed significant changes in specific SCFA during C . rodentium infection . Significantly reduced lactic acid was observed in infected WT but not in ChAT T-cell cKO mice . Butyric acid was significantly enhanced in both WT and ChAT T-cell cKO infected mice compared to uninfected WT or cKO control mice . While significantly increased production of pyruvic acid was detected in the feces from uninfected ChAT T-cell cKO mice , infection reduced the concentration of this metabolite to levels observed in uninfected or C . rodentium infected control mice . ( S5 Fig ) . Together these findings indicate that the increased C . rodentium burden in ChAT T-cell cKO mice was not due to an inability to produce antimicrobial peptides or alterations in SCFA produced by the microbiota . The increased expression of certain pro-inflammatory cytokines coupled with increased colonic C . rodentium burden in ChAT T-cell cKO mice lead us to ascertain if innate effector responses were intact . First , we considered if lack of T-cell derived ACh could increase differentiation of alternatively activated macrophages , disrupting the ability to mount and effect innate responses to C . rodentium . As indicated in Fig 5 , no significant differences were noted in arginase1 ( Arg1 ) , mannose receptor C-type 1 ( Mrc-1 ) , chitinase-like 3 ( Chi3l3 ) , or resistin-like molecule α ( Retnla ) expression by qRT-PCR in colonic tissues between WT and ChAT T-cell cKO mice . Expression of Nos2 ( “iNOS” ) however was significantly abrogated 10 days p . i . in ChAT T-cell cKO mice compared to infected WT . These data indicate that lack of T-cell derived ACh does not increase alternatively activated/M2 macrophage polarization , but significantly impacts the expression of Nos2 . As numerous cell types can express NOS2 , we assessed the localization and quantity of NOS2 protein by confocal microscopy on colonic tissue from infected WT and ChAT T-cell cKO mice and uninfected controls . As indicated in Fig 6 , IEC ( CDH1+ DAPI+ ) were the predominant cell type that were immunoreactive of NOS2 during C . rodentium infection . In keeping with the qRT-PCR data , C . rodentium induced NOS2 expression was significantly reduced in ChAT T-cell cKO compared to WT mice ( Fig 6A ) . Quantification of NOS2 expression in IEC further demonstrate reduced NOS2 expression in C . rodentium infected ChAT T-cell cKO mice ( Fig 6B ) . This reduced ability to increase NOS2 expression in IEC during 10 days p . i . was further validated by flow cytometry conducted on IEC ( Single , live , CD45- , EpCAM+ ) from naïve and infected WT and ChAT T-cell cKO mice ( Fig 6C & 6D ) . These data demonstrate that T-cell deficiency in ChAT significantly impairs C . rodentium induced increases of NOS2 expression in IEC . As ACh has been previously demonstrated to induce NOS2 expression in lung epithelial cells [27] , we sought to determine if ACh could induce similar effects in IEC . The mouse colonic epithelial cell line CMT-93 was treated with IFNγ ± carbachol ( ACh mimetic ) , with Nos2 , irf1 , and CIIta expression assessed by qRT-PCR . As expected , stimulation with IFNγ ( 1 ng/mL , 3 h , time and dose determined empirically ) induced expression of Ciita , Irf1 , and Nos2 . Co-treatment with carbachol further significantly increased expression of Nos2 compared to IFNγ alone , but did not enhance Ciita or Irf1 expression ( Fig 7 ) . Treatment with carbachol alone failed to significantly increase expression of any of the target genes . These results suggest that cholinergic signaling in IEC can synergistically enhance select IFNγ induced genes including Nos2 .
The finding that the nervous system is an active participant during inflammation has been an unexpected and intriguing finding . At the interface between these two systems is a unique type of T-cells that can produce and release ACh in response to sympathetic neurotransmitters [2 , 28] . Although the predominant focus on these ChAT+ T-cells has been on their ability to reduce the severity of disease in a number of clinically relevant immunopathologies [5–9] , ChAT+ T-cells can also help to establish host-commensal interactions . While a prior study demonstrated increased diversity of the small intestinal microbiota in ChAT T-cell cKO mice , due to reduced antimicrobial peptide expression [13] , the role of these cells during enteric bacterial infection was not known . Using a combination of ChAT-GFP reporter and ChAT T-cell cKO mice , our studies are the first to demonstrate recruitment of ChAT+ T-cells and a functional role for these cells during C . rodentium infection . These recruited ChAT-GFP+ T-cells do not appear to be restricted to a unique Th subset , with ChAT-GFP+ T-cells found to produce IFNγ , IL-17A , or IL-22 in agreement with prior studies [3] . As we and others have previously demonstrated that ChAT+ T-cells express the chemokine receptors CXCR5 [4] and CCR8 [29]; we sought to characterize the production of cognate ligands to these receptors during C . rodentium infection , Cxcl13 and Ccl8 respectively . Our analysis indicates that Cxcl13 but not Ccl8 expression is induced beginning 10 days p . i . until day 30 p . i . , a period during infection that closely mirrors when the number of ChAT+ T-cells increased in the colon . This temporal pattern of chemokine expression is corroborated by other studies demonstrating increased Ccl8 during C . rodentium infection [30] . CXCL13 is well established as critical to organization of secondary lymphatic organs [31] , tertiary lymphoid tissues and recruitment of IL-22 producing ILC3 [32] and can be induced by vagal nerve stimulation [33] . Despite this , our studies using CXCR5 KO mice indicate that this signaling axis is either not critical or functionally redundant with respect to the host response to C . rodentium 10 days post-infection . The importance of ACh derived from T-cells to host mucosal immune response during enteric bacterial infection was determined using a ChAT T-cell conditional knockout . Highlighting the host protective role of ACh , we observed an increased bacterial burden following enteric C . rodentium infection in ChAT T-cell cKO compared to WT mice . This increased bacterial burden in ChAT T-cell cKO mice was associated with significantly increased expression of the pro-inflammatory cytokines Il-1β , Il-6 , Tnfα , with equivalent expression of Ifnγ , Il-17a , or Il-22 compared to infected WT mice . Loss of T-cell derived Ach however did not impinge on IL-22 production , typically produced by ILC or T-cells in response to C . rodentium infection [34] . Together these findings , supported by the literature , suggest that ChAT+ T-cells are important in eliciting host-protective responses . Mucosal immunity is comprised of a multitude of overlapping mechanisms that serve to protect the host from pathogens including the production and secretion of antimicrobial peptides . T-cell derived ACh has been implicated in regulation of host-microbial interactions at the mucosal surface by controlling antimicrobial peptide production . Conditional ablation of ChAT in CD4+ cells using CD4 . Cre ChATf/f mouse line resulted in reduced lysozyme , defensin A , and ang4 expression in the small intestine , consequently increasing the diversity of commensal microbiota in the jejunum but not the cecum , or colon [13] . In contrast to these findings we noted no significant reductions in antimicrobial peptide expression in ChAT T-cell cKO compared to WT mice . As expected [35] , expression of RegIIIy was significantly enhanced in WT and ChAT T-cell cKO mice during C . rodentium infection irrespective of genotype . These data suggest that the increased bacterial burden in ChAT T-cell cKO mice was not due to a deficit in antimicrobial peptide expression . Host production of free radicals including NO are critical factors in protection against several bacterial pathogens [36–38] . NO also functions as a short-lived cell signaling molecule and is produced by three distinct isoforms of nitric oxide synthase that are each uniquely regulated in a tissue- or context-dependent manner [37] . In contrast to the constitutively expressed NOS found in endothelium or neurons , bacterial products or inflammation can induce NOS2 expression in a variety of cell types [37 , 39] . Infection with C . rodentium increases NOS2 expression , functioning to limit bacterial burden and disease [14 , 40] . In agreement with this literature , our data demonstrate that IEC in the colon are the predominant cell type expressing NOS2 during C . rodentium infection in WT mice . Conditional ablation of ChAT in T-cells , however , resulted in significantly reduced Nos2 expression compared to WT mice . Confocal microscopy on colonic tissue sections and flow cytometry experiments confirmed NOS2 expression was significantly increased in colonic IEC of WT mice , but not in ChAT T-cell cKO mice during infection . Together these data demonstrate that lack of T-cell derived ACh significantly reduced the induction of NOS2 in IEC during C . rodentium infection . Although NOS2 expression is characteristically elicited by IFNγ-induced activation of STAT1-dependent gene transcription [41] , expression of this cytokine was not affected by ChAT T-cell deficiency . Additionally , we observed that acetylcholine mimetics significantly enhance IFNγ-induced Nos2 expression in IEC in vitro , in agreement with previously reported experiments in lung epithelial cells [27] . There are striking similarities in the aberrant host response to C . rodentium infection in ChAT T-cell cKO and the previously described Nos2-/- mice [14] . For example , both mouse lines exhibit increased bacterial burden at day 10 p . i . without resulting in increased mortality or enhanced colonic histopathology [14] . Although Nos2 deficiency in mice , or inhibition of NO production increases Th17 differentiation [42] , no significant increase in Il-17a expression was observed in the colonic tissue from C . rodentium infected ChAT T-cell cKO mice compared to WT mice . This is likely due to the short half-life of NO in biological fluids [43] , and the expression of NOS2 in colonic IEC far from differentiating T-cells in draining lymph nodes . Our data further substantiate the unique role of ACh producing ChAT+ T-cells in modulating immune function . These unique T-cells appear to function as a critical component of the mucosal immune system , limiting the number and detrimental effects of enteric bacterial pathogens . How these specialized T-cells that are recruited to the colon , become activated , and release ACh during C . rodentium infection warrants future study . Given the requirement for NE signaling through the β2AR receptor on ChAT+ T-cells in septic shock [2] , activation by the sympathetic innervation is a strong possibility . Supporting this contention , Salmonella typhimurium induces activation of the sympathetic innervation within the small intestine , and the release of NE adjacent to muscularis macrophages [44] . While it is tempting to speculate that infection induced activation of a neuronal circuit is host protective , it is important to note that host NE induces bacterial expression of virulence genes by enteric pathogens such as C . rodentium [45] and enterohemorrhagic Escherichia coli [46] . Future studies will only further illuminate the integrated nature of the nervous system and immune system with ChAT+ T-cells as a critical node mediated host protection during enteric bacterial infection . | The nervous system is an active contributor to the regulation of immune responses . Prior studies have identified a unique CD4+ T-cell population that can relay signals from the sympathetic nervous system . These specialized T-cells express the enzyme choline acetyltransferase ( ChAT ) and produce acetylcholine ( ACh ) . Release of ACh in response to neurotransmitters from the sympathetic innervation was previously shown to aberrant immune cell activation , reducing mortality during septic shock . Also , these CD4+ ChAT+ T-cells were previously found to control host-commensal interactions in naïve mice , but their role during enteric bacterial infection was unknown . Here we demonstrate that infection with C . rodentium induces ChAT+ T-cell recruitment and that expression of ChAT by this T-cell population significantly augments host defenses . These data support a diverse and expanding role of ACh in host immune responses . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [] | 2019 | T-cell derived acetylcholine aids host defenses during enteric bacterial infection with Citrobacter rodentium |
An increased prevalence of epilepsy has been reported in many onchocerciasis endemic areas . To determine the prevalence and distribution of epilepsy in an onchocerciasis endemic region in the Democratic Republic of the Congo ( DRC ) . An epilepsy prevalence study was carried out in 2014 , in two localities of the Bas-Uélé district , an onchocerciasis endemic region in the Orientale Province of the DRC . Risk factors for epilepsy were identified using a random effects logistic regression model and the distribution of epilepsy cases was investigated using the Moran’s I statistic of spatial auto-correlation . Among the 12 , 776 individuals of Dingila , 373 ( 2 . 9% ) individuals with epilepsy were identified . In a house-to-house survey in Titule , 68 ( 2 . 3% ) of the 2 , 908 people who participated in the survey were found to present episodes of epilepsy . Epilepsy showed a marked spatial pattern with clustering of cases occurring within and between adjacent households . Individual risk of epilepsy was found to be associated with living close to the nearest fast flowing river where blackflies ( Diptera: Simuliidae ) –the vector of Onchocerca volvulus–oviposit and breed . The prevalence of epilepsy in villages in the Bas-Uélé district in the DRC was higher than in non-onchocerciasis endemic regions in Africa . Living close to a blackflies infested river was found to be a risk factor for epilepsy .
An association between onchocerciasis and epilepsy was suspected as early as the 1930’s in Mexico [1] and later reports were published showing clustering of epilepsy in several African onchocerciasis foci [2–7] . Ecological studies carried out in onchocerciasis endemic areas in West , Central and East Africa found a strong association between the prevalence of onchocerciasis and of epilepsy [8 , 9] . In previous case-control studies this association was less clear , but this was probably due to shortcomings in study design and the selection of comparison groups [9 , 10] . Moreover there seems to be an association between epilepsy and the degree of infection with Onchocerca volvulus . Indeed , in a study in Cameroon , performed before the introduction of annual ivermectin treatment ( to control Onchocerciasis ) , the prevalence of epilepsy and the community microfilarial load were closely related [11] . Moreover a case-control study demonstrated that the microfilarial loads ( microfilariae per skin snip ) in the epileptic group were significantly higher than in the control group [11] . Nodding syndrome , a neurological syndrome of unknown origin is characterized by episodes of atonic seizures . It has been observed in Tanzania ( Mahenge ) , South Sudan ( mainly in the Western Equatoria State ) and northern Uganda ( the districts of Gulu , Kitgum , Pader and Lamwo ) , where it was also found to be associated with onchocerciasis [12 , 13] . Whether this syndrome is a separate entity or is part of the clinical spectrum of epilepsy associated with onchocerciasis has been a matter of debate for many years [14–16] . In this study , we investigated whether there was increased epilepsy prevalence in an endemic focus of onchocerciasis of the Orientale Province in the Democratic Republic of the Congo ( DRC ) and whether nodding syndrome was part of the clinical spectrum of epilepsy in this area . The study area was chosen because of medical observations reporting cases with nodding syndrome like clinical manifestations in two districts of the Orientale Province , namely the Ituri and the Bas-Uélé district . In the Ituri district , epilepsy was frequently reported in the medical history of young patients with onchocerciasis participating in a Moxidectin clinical trial conducted by WHO/TDR at the General Hospital of Reference Rethy between 2009 and 2012 [17] . In view of these observations , the Health Provincial Division of the Ituri district commissioned a scientific team to conduct an investigation of these cases of epilepsy . In February 2013 , the investigation team led by MM visited the Logo health area in the Ituri District , and identified 6 cases with nodding syndrome like clinical manifestations [18] . In Bas-Uélé district , a team led by GMa and JK of the Congolese Ministry of Health investigated cases of “childhood epilepsy epidemic” in the Liguga health area where the reported O . volvulus nodules prevalence in 2012 was 66 . 7% , and identified also children with nodding syndrome like clinical manifestations . These observations led us to carry out a first exploratory mission to the Bas-Uélé district in the localities of Liguga , Dingila , and Titule in March 2014 followed by a second mission in June 2014 to Dingila and Titule . In this paper we describe the prevalence of epilepsy in Dingila and Titule . Moreover in Titule we examined potential risk factors for epilepsy , in particular the distance of each household to the major river .
Studies were performed in two localities of the Bas-Uélé district in the Orientale Province of the DRC , namely Dingila and Titule . The Bas-Uélé district has a tropical climate with a rainy season from April to November . Over the year , the average temperature in Dingila is 24 . 4°C . The average annual precipitation reaches 1 , 738 mm . January is the driest month of the year ( 34mm ) , while October is the wettest with an average rainfall of 255 mm . Vegetation cover consists mainly of dense tropical rainforest and riparian flora and bamboo patches in wetlands near major rivers , while savannah defines it in its most northern part along the border with the Central African Republic . Buta is the main town of the district , which is split into 11 health zones . The Titule health zone has a population of 73 , 432 inhabitants over an area of 5 , 450 km² with a density of 13 inhabitants/km² . The Titule health zone is divided into 10 health areas among which Titule 1 and 2 , were investigated . The main socio-economic activities are agriculture and fishing . Secondary activities include hunting , gathering , collecting snails , mushrooms , caterpillars and termites , or petty trade , while domestic poultry and small livestock are the main animal husbandry . The diet of the population is based mainly on the consumption of cassava and its derivatives , and plantain bananas . For many years , and especially in the period 1996–2002 , the region experienced episodes of war which caused displacement of people into the surrounding forests . Titule is a locality crossed by the Bima River which is a major tributary and confluate with the Uélé in Liguga . Titule has a population of 11 , 882 inhabitants distributed between Titule 1: 4 , 776 ( Gbulusu: 2 , 462 , Mopemba: 2 , 314 ) and Titule 2: 9 , 220 ( Basayo: 2 , 607 , Mokoa: 3 , 896 , Kpebele: 820 ) . Dingila , was a prosperous city around the 1950’s where a major cotton wool farming company ( the Compagnie de Développement du Nord , ( CODENORD ) ) was located . Following two episodes of war , cotton production has been abandoned since 2001 . The population of Dingila is estimated at 12 , 776 inhabitants . Both villages ( Fig 1 ) are situated close to fast flowing rivers: Dingila ( N3 . 68336—E26 . 03611 ) lies along the Uélé river and Titule ( N03 . 28150 –E025 . 52894 ) along the Bima river . A case of epilepsy was defined as a patient who lost consciousness at least twice with convulsions and without fever or any acute illness [19] . A case of suspected nodding syndrome was defined as a person who presented with episodes of decreased consciousness during which the head dropped forward repeatedly [20] . In June 2014 families in Dingila with members known to have epilepsy were interviewed at home by two staff members of the ivermectin community distributors network ( the relais communautaires ) . In total , 370 cases of epilepsy were recorded as well as information on age and sex of the epileptic patients , and age of the first epilepsy episode . The prevalence of epilepsy was determined using the total number of people living in Dingila as denominator . In June 2014 a household survey was carried out in Titule by two physicians , FT and JK together with a nurse ( FM ) and two ivermectin community distributors . The team , split in two groups each led by a physician , visited every 3rd household in four different directions of the village; the starting point was the Titule 1 health Centre . If household members were not at home , the next home was visited . All household heads and parents of children present in the household were interviewed in their local language . Families with and without epilepsy were geo-located ( handheld Garmin 62Cs GPS; ±4m accuracy ) . For every consenting household , a one page questionnaire was completed ( available as supporting information ) . Age and sex of every household member was recorded . Epilepsy is a condition that is well known by the local population; in the local language it is called “Epupuluga” and in lingala “Malali Ya Ndeke” . Interviews started by asking whether there was a family member with Epupuluga or Malali Ya Ndeke . Thereafter the doctors ( FT and JK ) asked the family members to describe the type of seizures ( or to show what happens during a seizure ) , to report on the precipitant circumstances , the duration of seizures , whether they were associated with uncontrollable tongue biting or passing of urine or stool , whether there were family members with episodes of absence ( sudden episodes of decreased consciousness of sudden onset and short duration ) with or without nodding of the head , whether epilepsy had been treated and what the effect was of anti-epileptic treatment on seizures . Although the answers to these questions were not systematically recorded , the final doctors diagnosis was used in determining the prevalence of epilepsy . For household members with epilepsy confirmed by the doctor , the age of the first epilepsy episode was noted . For those who developed epilepsy in 2014 , the exact month of the first epilepsy attack . For every household member the question was asked whether ivermectin was taken in 2013 ( the last year ivermectin was distributed at the time of the study ) . Of the 2906 cases in the original data file 2874 included valid geographic coordinates which were included in the subsequent spatial analysis . The spatial distribution of epilepsy cases over the study area was mapped by the location of their household of residence using a projected coordinate system . Clusters of increased epilepsy risk within the study area were located and mapped by applying a kernel density estimator to the geographical locations of households [21] , weighted according to their respective prevalence rates . To test whether significant clustering of epilepsy prevalence exist and over what distance clustering occurs , Moran’s I statistic of spatial autocorrelation was calculated over incremental distance intervals of 50m starting at 200m . Before analysis , epilepsy prevalence rates per household were transformed using arcsine square root transformation for proportions . Exposure to riverine habitat was determined as the Euclidian distance ( in meters ) between a household and the nearest flowing river . Since data on local river networks were not available , these were modelled based on a 30m SRTM digital elevation model , freely available online [22] . A model of the local hydrology was derived applying a triangular multiple flow direction algorithm [23] to the elevation data , from which a flow network was extracted . The resulting river network was validated using the point locations of rivers recorded in the field . The average distance between modeled river flows and river locations observed in the field was less than 30 meters , corresponding to the resolution of the original digital terrain model used . This observed accuracy was significantly less than the minimum distance between a household and the nearest river flow ( Two sample t-test: t = 20 . 6 , df = 151 , p <0 . 001 ) . The association between the occurrence of epilepsy cases and past ivermectin treatment , and exposure to riverine habitat was determined by means of a Generalized Linear Mixed Model ( GLMM ) using a logistic link function [24] . Ivermectin use during the last year ( in this survey , 2013 ) and the distance to the nearest river were entered as fixed effect into the model . To account for possible confounding effects due to gender and age , these were added as categorical fixed effects . To account for possible clustering of cases within a household , this was included in the model as a random effect . The study was approved by the Institutional Review Board of the University of Kisangani . Written informed consent was obtained from all families who participated in the study .
Among the 12 , 776 individuals of Dingila , 373 ( 2 , 9% ) individuals with epilepsy were identified; 193 ( 52% ) females and 180 ( 48% ) males; mean age: 18 , IQR 13–22 , range 2–62; mean age at onset of the epilepsy: 11 , IQR 7–11 , range 1–37 . The age distribution of these 373 individuals with epilepsy , and the year and age the epilepsy started are shown in Fig 2 . Three hundred and fifty one households of Titule I and II were surveyed over a period of 9 days; in total 2 , 906 people participated in the survey . Sixty seven ( 2 . 3% ) reported episodes of epilepsy; 36 ( 53% ) were male and 31 ( 46% ) female with a median age of 19 , range 5–75 . In 3 ( 4% ) of them head nodding was reported . In 6 families there were at least two persons with epilepsy and in one family three . All patients with epilepsy had seizures the last 5 years , and all except two during the last two years . The latter never took anti-epileptic treatment but received annual ivermectin treatment . Six individuals developed their first episode of epilepsy within the last 6 months ( annual incidence = ±4 . 12/1000 ) , 4 ( 67% ) of them did not take ivermectin 7 months earlier . The age distribution of epilepsy cases , the year and age of onset of epilepsy are shown in Fig 3 . The highest prevalence of epilepsy was observed in the 10–19 age group ( Table 1 ) . Mapping the number of epilepsy cases per household shows high heterogeneity in epilepsy across Titule village , indicative of spatial clustering of epilepsy cases ( Fig 4 ) . From the spatial autocorrelation analysis over incremental distances it appears that the prevalence of epilepsy within households is significantly clustered at distances of 200–350 meters ( Fig 5 ) . The kernel density estimate using a bandwidth of 300 meters , corresponding to the distance at which spatial clustering occurs , confirms the existence of several spatial clusters of increased epilepsy rates ( Fig 4 ) . Table 2 shows the results of the multiple random effects logistic regression model relating the occurrence of individual epilepsy cases to the past usage of ivermectin , the position of a household relative to a river flow a person’s age ( 0–9 , 10–19 and > 20 years of age ) and gender . Clustering of cases within households was accounted for by adding households as a random effect to the model . A significant increase of epilepsy was observed in persons of 10–19 years of age ( OR: 33 . 3 , p<0 . 0001 ) . Being treated with ivermectin over the preceding year resulted in lower odds of epilepsy ( OR: 0 . 23; p < 0 . 001 ) . Equally , living in a household at larger distance to a river reduced the risk of epilepsy ( OR: 0 . 63; p < 0 . 001 ) .
The prevalence of epilepsy in villages in Bas-Uélé district in the Orientale Province in the DRC was comparable with those reported in other onchocerciasis endemic areas in Africa [8] but higher than the epilepsy prevalence ( 0 . 4–1% ) reported in most studies in the rest of the world [26–30] . In a study of 586 , 607 residents in five Health and Demographic Surveillance System ( DSS ) centers in sub-Saharan Africa , only 1 , 711 ( 0 . 29% ) individuals were diagnosed as having active convulsive epilepsy [31] . Prevalence adjusted for attrition and sensitivity varied between sites: 7 . 8 per 1000 people ( 95% CI 7 . 5–8 . 2 ) in Kilifi , Kenya , 7 . 0 ( 6 . 2–7 . 4 ) in Agincourt , South Africa , 10 . 3 ( 9 . 5–11 . 1 ) in Iganga-Mayuge , Uganda , 14 . 8 ( 13 . 8–15 . 4 ) in Ifakara , Tanzania and 10 . 1 ( 9 . 5–10 . 7 ) in Kintampo , Ghana [28] . Interestingly , the highest prevalence of epilepsy was recorded in the Ifakara DSS in Tanzania , located not far from Mahenge , an onchocerciasis endemic region where nodding syndrome was first reported . The peak incidence of epilepsy in patients in Dingila and Titule was around the age of 14–15 . This is similar with other onchocerciasis endemic African regions and can most likely be explained by the equally high incidence of O . volvulus infection in these age groups and the cumulative nature of the O . volvulus infestation [9] . This peak incidence is in contrast with the epilepsy situation in industrialized countries and in non-onchocerciasis endemic regions in Africa where most onset of epilepsy is observed in the very young ( < 5 years old ) and in the older population [9] . In Cameroun , Pion et al . , proposed to call this form of onchocerciasis associated epilepsy “river epilepsy” [9 , 20] . Although parents reported head nodding in three children , cases of nodding syndrome meeting the 2012 WHO case definition were not observed [20] . Important to note that in Titule , there was not only a high prevalence of epilepsy in the 10–19 age group ( 4 . 7% ) but also in the 20–29 age group ( 4% ) . The latter is in contrast with the epilepsy prevalence reported in onchocerciasis endemic regions where ivermectin was not yet introduced . In a study from 1991 in Kyarusozi sub-county in western Uganda , 91% of the epilepsy cases were below the age of 19 [32] . In contrast , in the Imo river basin in Nigeria where ivermectin was distributed since 1994 , in 2002 , 67% of the patients with epilepsy were 20–29 years old [33] . This epilepsy age shift after the introduction of ivermectin is an argument that ivermectin may reduce the incidence of epilepsy . Our study has several limitations . In most individuals the epilepsy was only reported and not observed . Individuals with epilepsy were not examined by a neurologist and laboratory investigations and Magnetic Resonance Imaging were not performed . In Dingila , only households known to have members with epilepsy were visited . Families who have family members with frequent generalized tonic–clonic seizures are generally known in villages by the relais communautaires who know their communities as they are in frequent contact with families and visit all households annually for the distribution of ivermectin , vaccination campaigns and mosquito net distribution . Nevertheless , it is possible that we underestimated the prevalence in Dingila , because individuals with other forms of epilepsy and with infrequent seizures may have been missed . Moreover in both villages we did not use a validated epilepsy questionnaire as was used in other studies in the past [34] . In Titule , we did not use the recommended two or three-stages standard screening methodology for epilepsy prevalence studies [34] . We conducted the survey differently due to time constraint , and because we had the possibility that 2 medical doctors could screen for and diagnose epilepsy directly in households , located not too far from each other . In our definition of epilepsy we did not specify a time limit for laps between the 2 first seizures , or for the last seizure and we only included convulsive epilepsy . Therefore comparison of our results with published epilepsy prevalence data is difficult . In Dingila , as the diagnosis of convulsive epilepsy was not confirmed by a medical doctor , it is possible that certain other conditions such as psychogenic non-epileptic seizures were misdiagnosed as epilepsy . Moreover to determine the prevalence of epilepsy in Dingila we used a census figure as denominator , that was not verified at the moment of the study by a house-to-house survey . Initially we expected that cysticercosis was a possible cause of epilepsy in the region because many pigs were observed in the villages and because cysticercosis is known to be an important cause of epilepsy in Africa [35] . However in a case control study performed in Titule , none of the cases or controls had Taenia solium antibodies , suggesting that cysticercosis was not an important cause of epilepsy in the region [36] . In Titule , both onchocerciasis and Loa Loa are endemic and ivermectin distribution is given to the population regardless whether a person has a Loa Loa infection . Indeed , following the administration of ivermectin in Loa Loa infected patients , with high filarial loads , there is a higher risk ( 1/10 000 ) of developing serious adverse effects mainly caused by the massive death of Loa Loa filariae in the brain causing an encephalopathy and eventually coma and death [37] . However of the 59 cases and 61 controls in our case control study in Titule only one case and one control had a positive Loa Loa PCR blood test [36] . Therefore Loa Loa co-infection cannot explain the high prevalence of epilepsy in the region . On the other hand , in 78% of cases and 83% of controls O . volvulus IgG4 antibodies were found [36] . Individuals who used ivermectin in 2013 were less likely to have epilepsy compared to those who did not use ivermectin in 2013 . The explanation for this could be that certain individuals in Titule with epilepsy also in the past were not taking ivermectin and therefore were not protected against onchocerciasis and onchocerciasis associated epilepsy , or that people with epilepsy avoid taking ivermectin because they may suspect ivermectin is causing epilepsy . Two ( 33% ) of the 6 patients who developed epilepsy during the last 6 months had taken ivermectin 7 months before . If this information is true , this suggest that if ivermectin offers some protection against onchocerciasis associated epilepsy , this protection may not be complete with only one dose of ivermectin per year . The distribution of epilepsy cases within Titule showed a marked spatial pattern . Epilepsy were observed to cluster both within and between adjacent households located near river flows . A possible explanation for this observed pattern could be the fact that households near river flows are more often exposed to O . volvulus vectors ( Simulium sp . ) which occupy riverine habitats , resulting in increased infestation rates . In conclusion , we documented a high prevalence of epilepsy in the Bas-Uélé district , in villages close to rapid flowing rivers infested by blackflies . Identifying whether an O . volvulus infestation can directly or indirectly cause epilepsy , or whether there could be another agent transmitted by blackflies responsible for the epilepsy require further investigation [16 , 38] . | An increased prevalence of epilepsy has been reported in many onchocerciasis ( river blindness ) endemic areas . In 2014 , an epilepsy prevalence study was conducted in Dingila and Titule , two localities within the onchocerciasis endemic Orientale province of the Democratic Republic of Congo , both within the Bas-Uélé district . The prevalence of epilepsy was 2 . 9% in Dingila , and 2 . 3% in Titule . This is much higher than the epilepsy prevalence in non-onchocerciasis endemic regions in Africa . Epilepsy cases were clustered within and between adjacent households . Living close to a river infested with blackflies ( the main vector of Onchocerca microfilariae ) was found to be a risk factor for epilepsy . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"ecology",
"and",
"environmental",
"sciences",
"onchocerca",
"volvulus",
"rivers",
"sociology",
"tropical",
"diseases",
"geographical",
"locations",
"social",
"sciences",
"parasitic",
"diseases",
"animals",
"onchoc... | 2016 | Prevalence of River Epilepsy in the Orientale Province in the Democratic Republic of the Congo |
Human mobility plays a central role in shaping pathogen transmission by generating spatial and/or individual variability in potential pathogen-transmitting contacts . Recent research has shown that symptomatic infection can influence human mobility and pathogen transmission dynamics . Better understanding the complex relationship between symptom severity , infectiousness , and human mobility requires quantification of movement patterns throughout infectiousness . For dengue virus ( DENV ) , human infectiousness peaks 0–2 days after symptom onset , making it paramount to understand human movement patterns from the beginning of illness . Through community-based febrile surveillance and RT-PCR assays , we identified a cohort of DENV+ residents of the city of Iquitos , Peru ( n = 63 ) . Using retrospective interviews , we measured the movements of these individuals when healthy and during each day of symptomatic illness . The most dramatic changes in mobility occurred during the first three days after symptom onset; individuals visited significantly fewer locations ( Wilcoxon test , p = 0 . 017 ) and spent significantly more time at home ( Wilcoxon test , p = 0 . 005 ) , compared to when healthy . By 7–9 days after symptom onset , mobility measures had returned to healthy levels . Throughout an individual’s symptomatic period , the day of illness and their subjective sense of well-being were the most significant predictors for the number of locations and houses they visited . Our study is one of the first to collect and analyze human mobility data at a daily scale during symptomatic infection . Accounting for the observed changes in human mobility throughout illness will improve understanding of the impact of disease on DENV transmission dynamics and the interpretation of public health-based surveillance data .
Human mobility plays a central role in shaping the structure of transmission networks and in influencing epidemiologic processes such as pathogen introduction , epidemic transmission , and endemic persistence [1–4] . While human mobility can drive transmission across multiple spatial and temporal scales [3 , 5] , it is at the finest scales ( daily , intra-urban human movements ) where epidemic processes occur and emergency public health interventions are usually implemented . Evidence from theoretical models and empirical studies show that individual and/or spatial variability in number and frequency of contacts can lead to transmission heterogeneity , where certain individuals or locations contribute disproportionately to pathogen transmission and epidemic spread [6–8] . Thus , identifying social and behavioral characteristics ( e . g . , mobility patterns , occupations , age classes ) most responsible for such disproportionate contributions has become a public health priority , with significant potential for leveraging the power of public health surveillance programs and targeted disease control [9–11] . Dengue , an acute illness caused by four immunologically related viruses in the family Flaviviridae and transmitted by Aedes spp . mosquitoes ( primarily Aedes aegypti ) , is the most important mosquito-borne viral disease of humans worldwide [12] . Because Aedes aegypti seldom disperse beyond 100 meters , have a propensity for resting and biting inside residential buildings , and bite during the day [13–16] , human movements are key to explaining the urban transmission dynamics of dengue virus ( DENV ) [17–21] . Individual movement patterns can also expand the spatial scale of transmission and lead to significant heterogeneity in transmission patterns by connecting otherwise discrete subpopulations of mosquitoes [3 , 22 , 23] . Extensive movement studies performed in the upper Amazon city of Iquitos , Peru , have shown that while human mobility within a resource-poor urban center is highly unstructured ( with only 38% of participants having regular mobility routines ) , the majority of locations visited are either residential or commercial , with most movements ( 81% ) occurring within 1 km of an individual’s home [24–28] . Moreover , an individual’s risk of DENV infection significantly increased when he or she routinely visited the same residential locations as DENV-infected people , whereas the distance the individual lived from a DENV-infected case was not significant [26] . Such empirical characterizations of fine-scale human mobility patterns and risk of DENV infection have informed complex simulation models that explore the theoretical role of human movement on the spatial and temporal patterns of disease dynamics [18 , 23 , 29] . When mobility-driven contact structure is included in theoretical models , the effect on DENV epidemic transmission is dramatic . Overlapping movement patterns within social groups drive the fine-scale heterogeneity in DENV transmission rates; however , the presence of a mosquito vector can hide the effect of socially structured movements if only spatially aggregated infection dynamics are considered [23] . Such models do not take into account the fact that symptomatic infection may influence mobility , which in turn can influence onward virus transmission and the structure of transmission chains . Research on directly transmitted diseases has demonstrated disease-driven behavior changes [30] and the significant influence they can have on predictions of pathogen spread [31–34] . For DENV , mobility data have been captured for febrile symptomatic individuals and healthy individuals using either retrospective movement surveys [35] or GPS trackers [36] . Febrile DENV-infected individuals visited significantly fewer places , traveled shorter distances , and spent more time at home [35 , 36] . These patterns reveal particularly important information for understanding the complex relationship between symptom severity and human mobility , and to ultimately determine if there is an association between human mobility and infectiousness to mosquitoes . However , because DENV infectiousness peaks at 0–2 days after onset of symptoms and lasts for 4–5 days after onset of symptoms [37–39] , human mobility during the first few days of symptoms could be key to better understand transmission dynamics . The goal of this study , therefore , was to conduct detailed , daily retrospective interviews to measure the mobility behavior of clinically apparent DENV-infected individuals throughout their illness , with the goal of generating mobility metrics that can be used to characterize the absolute and relative impacts of disease on potential exposure to Aedes aegypti mosquitoes .
This study was performed in the Amazon city of Iquitos , Peru . Iquitos is a geographically isolated , tropical urban environment with approximately 430 , 000 inhabitants located along the margin of the Amazon River [40] . The city’s economic structure is highly informal and dynamic , with one-third of economically active individuals either unemployed or informally employed [41] . Iquitos has been the home of extensive , long-term arboviral research led by the University of California , Davis and U . S . Naval Medical Research Unit 6 since 1999 [3 , 24–28 , 42] . Extensive human mobility studies paired with detailed epidemiological data have made Iquitos an informative site for understanding the dynamics of arbovirus transmission . All four serotypes of DENV have been introduced in Iquitos; however , at any particular time virus transmission is usually dominated by a single serotype [42 , 43] . Previous research [27] demonstrated that inhabitants visit an average of 5 . 8 ( ±3 . 6 SD ) locations over a two-week period . While most movement ( ~80% ) occurs within 1 km of their home , inhabitants have highly irregular and temporally unstructured routines that are not dominated by a single location , such as a workplace [27] . The study followed a contact-cluster design in which reverse transcription polymerase chain reaction ( RT-PCR ) positive , or viral nucleic acid test positive , DENV-infected individuals ( index cases ) were captured through community or clinic-based febrile surveillance systems , as described previously [26] . At the time of the initial blood sample , a 15-day retrospective semi-structured movement survey ( RMS ) was administered to the index case to identify the locations they visited in the 15 days prior to diagnosis ( characterizing the “pre-illness” period ) . Consenting individuals ( contacts ) from the index cases’ home and residential locations visited by the index case were then screened for DENV infection using RT-PCR [26] [44] . The RMS was administered to DENV PCR-positive contacts to quantify mobility behavior associated with potential virus exposure . RMSs were developed based on findings of focus groups and validated by comparison with data from people wearing GPS tracking devices [24 , 25] . RMSs capture positional , temporal and behavioral information of routine human mobility . Questions focus on the amount of time an individual spent at home , the visitors they received , and the places they visited . For time spent at home , individuals were asked about the average number of hours spent at home each day of the week , specifically focusing on the period from 5 a . m . to 10 p . m . , which includes the peak landing and biting times for Aedes aegypti [45] . For places visited , information was collected on the type of place visited , when , for how long , and how often in the 15-day period . Trained , local Iquitos residents ( the ‘Movement Team’ ) verbally administered electronic RMSs and recorded the information on tablets in the CommCare application [46] . To track movements of DENV positive participants during their illness , daily interviews using a modified daily RMS ( DRMS ) were conducted in person or by telephone for 7 days following the initial RT-PCR-positive blood test ( S1 File ) . Where participants were not available for daily interviews , information about movements on several days was collected at a single interview . The DRMS asks about the amount of time spent at home the previous day ( s ) and the following information about each place visited during the previous day ( s ) : day visited , place type , location , time of day visited , and time spent . For residential places visited , the DRMS asks whom they were visiting , their reason for visiting , if anyone in the home was ill during the preceding 15 days , and ( for routinely visited houses ) if/why there was any change in the time of visitation , as compared to the “pre-illness” period . During this seven-day period , DENV positive individuals were also administered two Quality of Well-Being surveys ( QWB ) by the Movement Team , one 2–3 days and one 7 days after the initial PCR-positive blood test . The QWB survey is a validated instrument used to measure an individual’s quality of life during chronic illness [47] . Our study was a novel application of the QWB survey to an acute illness . The survey responses were sent to the developers at University of California , San Diego , who used a weighted algorithm to produce one well-being score between 0 . 0 ( death ) and 1 . 0 ( asymptomatic and fully-functioning ) covering the three days prior to each survey date [47] . At a follow-up visit scheduled 30 days after the initial PCR-positive blood test , individuals were given a 15-day ( “post-illness” ) RMS and QWB survey in an effort to record their “baseline” mobility behavior and well-being in the absence of illness . For each study participant , the following variables were computed from the “pre-illness” and “post-illness” 15-day RMS: ( 1 ) total number of locations visited , ( 2 ) proportion of visits to each location type , ( 3 ) total number of houses visited , ( 4 ) proportion of visits to houses of family members vs . houses of friends , and ( 5 ) average proportion of time spent at home per day . Equivalent daily values of these variables were collected for each participant from the DRMSs . Rather than referring to values as occurring on a certain number of days after the PCR-positive blood test , a standardized “day after symptom onset” variable was calculated . Because blood tests were not done on the same day of illness for all participants , DRMSs captured a range of 1–15 days after symptom onset . We focused our analysis on days 1–9 after symptom onset; few individuals had data for days 10–15 after symptom onset . Analysis of mobility data had two main objectives: ( 1 ) comparing healthy ( pre- and post- illness ) mobility to mobility during illness , and ( 2 ) determining if mobility patterns changed during the 9 days after symptom onset . For the first objective , mobility during illness was calculated by averaging a participant’s DRMS for all available time points up to day 9 after symptom onset . Comparisons were done for the following mobility metrics: daily number of locations visited , daily number of houses visited , and proportion of time spent at home . When a metric followed a normal distribution ( assessed via the Shapiro-Wilk test ) , pairwise comparisons were performed with paired t-tests followed by Holm-Bonferroni corrections . When the variable was not normally distributed , the non-parametric Kruskall-Wallis Rank Sum Test and pairwise Wilcoxon Signed Rank Test for paired data were utilized . As many individuals would stop visiting other locations during their illness period [35] , we also analyzed the number of locations , number of houses , and time at home as binary variables , asking if any locations/houses were visited and if any time was spent away from home . These binary outcomes were compared between all possible pairs of time points ( pre-during , during-post , and pre-post illness ) using McNemar’s χ2 test . If locations were visited , further analyses determined what type of locations they were . While these data were subject-correlated across time points , they could not be analyzed as paired data because not all participants visited locations at every time point . Generalized logistic mixed-effects models ( GLMMs ) determined the association between the probability of a location type being visited and the time period being considered ( pre-during-post illness ) , while accounting for repeated measures by using participant ID as the random intercept . Location type was separated into four groups: ( 1 ) house , ( 2 ) health , ( 3 ) education/work , and ( 4 ) other ( e . g . , recreation , church , market , port ) . Similarly , logistic GLMMs determined the association between time period ( pre-during-post symptoms ) and the probability of a specific house type being visited ( e . g . , family versus friend ) . For the second objective , aiming to determine whether mobility patterns changed during the illness period , we calculated mobility metrics for 3-day groups ( days 1–3 , 4–6 , and 7–9 after onset of symptoms ) . Daily data were aggregated into 3-day groups to allow for robust analyses , while also controlling for the dearth of data points on certain days . In particular , the first two days after symptom onset had incomplete information for some participants due to the time required to capture individuals with symptoms , run RT-PCR tests , and obtain confirmed test results . To make pre/post and during-illness data comparable , 15-day RMS values were condensed to give movements over an average 3-day period . Analysis of the number of locations/houses , proportion of location/house types , and time spent at home followed the same steps described above . Comparisons were made between the 3-day groups to determine whether significant changes occur in movement patterns during illness . Further , movements within the 3-day groups were each compared to post-illness mobility . The associations between daily ( DRMS ) mobility patterns and possible predictor variables were examined using Generalized Linear Mixed Models ( GLMMs ) , Generalized Additive Mixed Models ( GAMMs ) , and Generalized Additive Models for Location Scale and Shape ( GAMLSSs ) [48–50] . Best-fit models were determined for each of the following mobility outcomes: total number of locations visited ( count variable ) , relative number of locations visited ( compared to when healthy ) ( continuous variable ) , total number of houses visited ( count ) , relative number of houses visited ( continuous ) , total proportion of time spent time at home ( continuous ) , and relative amount of time spent at home ( continuous ) . For both the total number of locations visited and total number of houses visited , GLMMs and GAMMs with underlying Poisson distributions were compared . An individual’s age , occupation , gender , QWB score , and the “day after symptom onset” were considered as predictor variables , with the best-fit model determined using a AIC and a Chi-square test comparing reduction in residual deviance . The response variable proportion of time spent at home was best characterized by a one-inflated beta distribution , so analysis was done with GAMLSS , as detailed below . Although GLMM and GAMM regressions model the mean ( μ ) value of the distribution of the response variable , GAMLSS allows other distribution parameters to be modeled as a function of explanatory variables . A one-inflated beta distribution has possible values 0<y≤1 and is defined in two parts: the probability that y = 1 ( modeled by the η parameter ) and the probability for 0<y<1 , which is shaped by a traditional beta distribution with parameters mean ( μ ) and shape ( σ ) . Here , the η parameter was the probability an individual stayed at home 100% of the time ( y = 1 ) . If an individual did not stay at home the entire day , the proportion of time that was spent at home ( 0<y<1 ) was determined by a beta distribution with μ and σ . We also considered response variables as relative values in order to control for the individual variation in mobility levels . The number of locations ( houses ) an individual visited on each day during illness was considered relative to the average number of locations ( houses ) they visited pre-illness . Similarly , the number of hours a participant spent at home during each day of illness was compared to the average number of hours that individual spent at home pre-illness . While relative number of houses could not be well explained by a set distribution , both relative number of locations visited and relative amount of time spent at home were best characterized by the logistic distribution . Analysis of these response variables was done with GAMLSS , where both the mean ( μ ) and the standard deviation ( σ ) parameters of the logistic distribution could be modeled as a function of explanatory variables . Best-fit GAMLSS models were chosen using forward and backward selection for each of the explanatory variables . All statistical analyses were performed in R 3 . 3 . 0 statistical computing software [48–51] . The procedures for enrollment of participants , dengue diagnosis , semi-structured interviews , and participant follow-up were approved by the Institutional Review Board ( IRB ) of the United States Naval Medical Research Center Unit No . 6 ( NAMRU-6 ) ( NAMRU6 . 2014 . 0021 ) in compliance with all applicable federal regulations governing the protection of human subjects . IRB relying agreements were established between NAMRU-6 and Emory University , Tulane University , University of California Davis , University of Rhode Island , San Diego State University , and University of Notre Dame . In addition to IRB approval , investigators obtained host country approval from the Loreto Regional Health Department , which oversees health research in Iquitos . Adult study participants provided written informed consent and a parent or guardian provided informed consent on behalf of child study participants .
When comparing healthy ( pre-/post-illness ) and symptomatic time points , there was a significant difference in both the proportion of time spent at home and the average number of locations visited ( Fig 1A and 1B ) . Healthy participants spent 60% of their time at home and visited an average of 1 . 3/1 . 1 ( pre-/post-illness ) locations per day , whereas ill participants spent 74% of their time at home ( Wilcoxon test: p < 0 . 001 ) and visited an average of 0 . 73 locations ( Wilcoxon test: pre-illness: p < 0 . 001; post-illness: p = 0 . 010 ) ( S2 Table ) . Participants were also significantly less likely to visit other houses during illness , as compared to pre-illness ( McNemar’s χ2: p < 0 . 001 ) and post-illness ( McNemar’s χ2: p = 0 . 043 ) ( S3 Table ) . The odds ( adjusted odds ratio/AOR ) of an individual visiting an education/work location during healthy time points ( AOR pre-/post-illness: 2 . 0/4 . 4 ) were significantly greater than during illness ( GLMM: p < 0 . 001; Table 1 , Fig 2 ) . Similar significant differences were seen for visits to “other” place types ( GLMM: p<0 . 001; Table 1 , Fig 2 ) . Conversely , the odds of participants going to a health-related place pre- or post-illness were significantly lower than during illness ( AOR: pre-/post-illness: 0 . 019/0 . 002; Table 1 , Fig 2 ) . Although individuals were more likely to visit a house during the pre-illness time period as compared to during illness ( AOR: 1 . 684; GLMM: p = 0 . 013 ) , there was no significant difference for post-illness ( AOR: 0 . 872; GLMM: p = 0 . 64 ) , where individuals were predicted to visit houses with a mean probability of 21% ( Table 1 , S4 Table ) . During days 1–3 and 4–6 after symptom onset , individuals were significantly more likely to spend all of their time at home , compared to both days 7–9 after symptom onset ( McNemar’s χ2: p = 0 . 046 ) and post-illness ( McNemar’s χ2: days 1–3: p = 0 . 008; days 4–6: p = 0 . 008 ) ( S6 Table ) . There was also a significant difference in the average proportion of time spent for days 1–3 and 4–6 ( 76% ) when compared to both days 7–9 ( 69% ) ( Wilcoxon test: days 1–3: p = 0 . 014; days 4–6: p = 0 . 008 ) and post-illness ( 59% ) ( Wilcoxon test: days 1–3: p = 0 . 005; days 4–6: p < 0 . 001; Fig 3B , S5 Table ) . Individuals were significantly less likely to visit any locations during illness compared to post-illness ( McNemar’s χ2: days 1–3: p = 0 . 001; days 4–6: p < 0 . 001; days 7–9: p = 0 . 008; S6 Table ) . Accordingly , the average number of locations visited was significantly lower on days 1–3 ( paired t-test: p = 0 . 017 ) and 4–6 after symptom onset ( paired t-test: p < 0 . 001 ) when compared to the mean 3 . 4 places visited every 3 days at post-illness ( Fig 3A ) . The average number of locations visited on days 1–3 ( 1 . 5 places/3-days ) was also significantly less than the average number of locations visited on days 7–9 after symptom onset ( 2 . 2 places/3-days ) ( Wilcoxon test: p = 0 . 047; S5 Table ) . When considering the type of location visited ( Fig 4 ) , the three during-illness time points ( days 1-3/4-6/7-9 ) were compared to the post-illness period . Post-illness , the participants were predicted to visit education/work places with a 48% probability , “other” places with a 32% probability , houses with a 20% probability , and health-related places with only a 0 . 2% probability ( S7 Table ) . Compared to post-illness , the odds of an individual visiting an education/work place were significantly lower for days 1–3 ( AOR: 0 . 08 ) , days 4–6 ( AOR: 0 . 22 ) , and days 7–9 after symptom onset ( AOR: 0 . 26 ) ( GLMM: p< 0 . 001; Table 2 ) . Conversely , the odds of visiting a health-related place during illness were significantly higher compared to post-illness ( AOR: days 1–3: 3826; days 4–6: 1041; days 7–9: 365 ) , likely due to the very low probability of a health-related location being visited post-illness when healthy ( GLMM: p < 0 . 001 ) . The likelihood of visiting a house during illness was not significantly different than the likelihood post-illness ( AOR: days 1–3: 1 . 08; days 4–6: 1 . 20; days 7–9: 1 . 64; GLMM: p > 0 . 05 ) . There were also no significant correlations between the illness time point and the odds of visiting a family member’s ( versus friend’s ) house ( Table 2 , S7 Table ) . The best-fitting model to describe the relative number of locations visited was a GAMLSS with a logistic distribution . The μ parameter ( mean ) was best explained by a positive effect of day after symptom onset ( p < 0 . 001 ) and a random intercept for participants , which allowed the mean relative number of locations to vary by participant . The σ parameter ( standard deviation ) was best explained by QWB score ( p < 0 . 001 ) , day after symptom onset ( p < 0 . 001 ) , and an interaction between the two ( p < 0 . 001 ) ( Table 3 ) . For relative amount of time spent at home , the best-fit model was a GAMLSS with underlying logistic distribution , where the μ parameter was best explained by a negative effect of day ( p = 0 . 0011 ) and a random intercept for participants . The σ parameter was best explained by a positive effect of day after symptom onset ( p < 0 . 001 ) ( Table 4 ) . With proportion of time spent at home as the response variable , the best-fit model explains the η parameter as a function of age ( <18 or >18 ) ( p = 0 . 005 ) and an interaction between age and day after symptom onset ( p = 0 . 018 ) . The μ parameter was explained by a random slope of participants over time and the σ parameter was explained by the QWB score and a smoothed function of the day of illness ( Table 5 ) . This suggests that whether an individual spent all ( 100% ) of their time at home was dependent on both their age and the day of illness , whereas the proportion of time spent at home ( when less than 100% ) depended on the day of illness ( p = 0 . 005 ) and how they were feeling ( QWB score ) ( p = 0 . 094; Table 5 ) . While the day of illness did not have an overall effect on the mean proportion of time spent at home ( when less than 100% ) , the random slope for participants suggests that day of illness had a varying effect across participants .
We found that dengue illness affects almost all aspects of an individual’s mobility behavior . During mild symptomatic illness , individuals visited significantly fewer locations and houses and spent significantly more time at home . Further , symptomatic participants visited education/work and “other” locations less often than when they were healthy and visited health locations more often . These results ( 1 ) are consistent with and expand prior evidence indicating that individuals with symptomatic illness move less than healthy individuals [31 , 35 , 52]; ( 2 ) refine estimates of the effects of mild symptomatic dengue illness on movement by quantifying changes before , during and after the symptomatic phase of infection; and ( 3 ) suggest the need to better account for disease-driven mobility behavior changes in DENV transmission models [31 , 53] . The most dramatic changes in mobility occurred during the first 3 days after symptom onset , when significantly fewer locations were visited and significantly more time was spent at home . During days 4–6 and 7–9 after symptom onset , the number of locations visited increased and the proportion of time spent at home decreased . By days 7–9 after symptom onset , the number of locations visited and the time spent at home were no longer significantly different from healthy behaviors . This reduction in mobility during illness , particularly on days 1–3 after symptom onset , could affect an individual’s contribution to onwards DENV transmission . For DENV , viremia reaches levels infectious to mosquitoes a few days prior to symptom onset and peaks at 0–2 days after symptom onset , with titers then lowering by days 4–5 ( although some individuals are still capable of infecting mosquitoes ) ( 37–39 ) . During peak infectiousness , most individuals are spending more time at home and visiting fewer places , thereby reducing the number of distinct Aedes aegypti mosquitoes with whom potential virus-spreading contacts occur . This may allow those few individuals who do not alter their movements to have a more significant role in pathogen transmission during peak infectiousness . During the pre-symptomatic period , however , almost all individuals have high mobility and a viremia level sufficient for virus transmission to mosquitoes [38] . Recent theoretical models of within-host viral dynamics for symptomatic individuals estimate that 24% of onward transmission results from mosquitoes biting during the pre-symptomatic period [54] . When also accounting for mobility changes throughout viremia , many individuals may have their greatest contribution to transmission be during the pre-symptomatic stage . Ten Bosch et . al . also estimated that asymptomatic individuals had only 80% the net infectiousness of symptomatic individuals [54] . This reduction in net infectiousness may be counteracted by the hypothetically unaltered mobility patterns exhibited by asymptomatic ( and minimally symptomatic ) individuals , further increasing the overall contribution of silent transmission . Such potential dynamics emerging from the coupling between individual infectiousness , movement , and disease severity deserve further investigation [55] , because they may help explain the explosive nature of DENV outbreaks and the limitations of vector control in containing virus transmission . Throughout an individual’s illness period , we found that day of illness and the participant’s subjective sense of well-being ( QWB score ) were significant predictors for the relative number of locations visited , as compared to pre-illness . When considering the proportion of time spent at home , an individual’s age and their day of illness were significant in predicting whether they chose to stay at home 100% of the day or not , with children being more likely to stay home all day compared to adults . When an individual chose to spend some amount of time outside their house , the day of illness and the QWB score significantly predicted the proportion of time . Further , the relative amount of time participants spent at home compared to pre-illness was also significantly predicted by the day of illness . Individuals with more severe symptoms and those at the beginning of their illness were more likely to be spending more time at home ( both absolute proportion of time and amount of time relative to pre-illness values ) . Further , when compared to pre-illness , individuals at the beginning of their illness have lower values of relatively visited locations compared to toward the end of illness . One limitation of our study is the reliance on participant recall , which can be subject to recall bias . However , the retrospective semi-structured interview we utilized was previously tested in Iquitos and was found to obtain superior data on activity space , as compared to wearable GPS data-loggers [24] . Further , in the DRMS participants only needed to recall movements over the past 24 hours , making bias less likely . Our study also faced limitations with the number of participants and the ability to measure movement on the first two days after symptom onset . Nevertheless , our study is one of the first to collect human mobility data at a daily scale during symptomatic infection . Future studies could build on our study by collecting detailed mobility data from more individuals with a wider spectrum of symptom severity , including across a wider range of diseases . Future studies should also seek to make coupled measurements of an individual’s infectiousness throughout the course of mobility data collection . Human mobility patterns have played an important role in recent vector-borne disease transmission models [56] . There is , however , an increasing need to include differing mobility patterns when modeling individuals that are ill versus healthy . We demonstrate that individuals with dengue spend significantly more time at home , particularly during the first days after symptom onset when they are most infectious , potentially limiting contact with Ae . aegypti outside their home . When looking at the locations being visited during illness , however , the proportion of houses was consistent throughout and remained similar to the post-illness level . This may be of particular importance for onward transmission given the propensity for Ae . aegypti to bite inside houses [13 , 14 , 57] . The abundance of mosquitoes in both an individual’s home and the houses/locations they visit when infectious will likely determine the effect that reduced mobility has on their overall contribution to DENV transmission . Reduction in mobility patterns when symptomatic could also affect the amount of overlap a social group has in the places they frequent . Given the significant role of socially structured human mobility in determining fine-scale DENV transmission rates [23] , accounting for the dynamic nature of social contacts during a symptomatic DENV infection could allow for more accurate modeling of disease transmission and the design of more efficient disease prevention strategies . | Dengue is the most important mosquito-borne viral disease of humans worldwide . Due to the limited mobility of the mosquitoes that transmit dengue virus , human mobility can be a key to both understanding an individual’s exposure to the virus and explaining the spread of dengue throughout a population . Accurate disease models should include human mobility; however , changes in human movement patterns due to the presence of symptoms need to be taken into account . We quantified the impact of symptom presence on human mobility throughout the infectious period by analyzing a dataset on the daily movements of dengue virus infected individuals . Accounting for these changing patterns of mobility will improve understanding of the complex relationship between symptom severity , human movement , and dengue virus transmission . Furthermore , dengue transmission models that incorporate symptom-driven mobility changes can be used to evaluate scenarios and strategies for disease prevention . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"dengue",
"virus",
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"body",
"fluids",
"pathology",
"and",
"laboratory",
"medicine",
"infectious",
"disease",
"epidemiology",
"pathogens",
"spatial",
"epidemiology",
"microbiology",
"social",
"sciences",
"human",
"... | 2019 | Dengue illness impacts daily human mobility patterns in Iquitos, Peru |
The Government of Senegal has initiated the “Projet de lutte contre les glossines dans les Niayes” to remove the trypanosomosis problem from this area in a sustainable way . Due to past failures to sustainably eradicate Glossina palpalis gambiensis from the Niayes area , controversies remain as to the best strategy implement , i . e . “eradication” versus “suppression . ” To inform this debate , we used population genetics to measure genetic differentiation between G . palpalis gambiensis from the Niayes and those from the southern tsetse belt ( Missira ) . Three different markers ( microsatellite DNA , mitochondrial CO1 DNA , and geometric morphometrics of the wings ) were used on 153 individuals and revealed that the G . p . gambiensis populations of the Niayes were genetically isolated from the nearest proximate known population of Missira . The genetic differentiation measured between these two areas ( θ = 0 . 12 using microsatellites ) was equivalent to a between-taxa differentiation . We also demonstrated that within the Niayes , the population from Dakar – Hann was isolated from the others and had probably experienced a bottleneck . The information presented in this paper leads to the recommendation that an eradication strategy for the Niayes populations is advisable . This kind of study may be repeated in other habitats and for other tsetse species to ( i ) help decision on appropriate tsetse control strategies and ( ii ) find other possible discontinuities in tsetse distribution .
The Niayes of Senegal harbours the most northern and western population of Glossina palpalis gambiensis Vanderplank , which is a major vector of the debilitating diseases Human African Trypanosomosis ( HAT ) or sleeping sickness , and African Animal Trypanosomosis ( AAT ) or nagana ( reviewed in [1] ) . Particular meteorological and ecological characteristics of this area provide great potential for agricultural development in general and animal production ( cattle , donkeys , horses , small ruminants , pigs and poultry ) in particular . Most of these animals are however susceptible to AAT which is seriously limiting the development of efficient and productive , sustainable livestock systems . The socio-economic impact of the disease is therefore dramatic and very often underestimated [2] . In the 1970s and 1980s , it was attempted to eliminate the G . p . gambiensis population from the Niayes mainly using ground spraying of residual insecticides [3] . The tsetse and trypanosomosis problem seemed to have disappeared until flies were detected again in 1998 ( unpublished report of the Direction de l'Elevage - DIREL ) . In 2005 , the DIREL initiated a control campaign called «Projet de lutte contre les glossines dans les Niayes» with the objective of developing a sustainable solution to the tsetse and trypanosomosis problem in the Niayes . The programme is funded by the Government of Senegal and technically and financially supported by the Food and Agriculture Organization of the United Nations ( FAO ) and the International Atomic Energy Agency ( IAEA ) . The project is implemented in the context of the African Union - Pan African Tsetse and Trypanosomiasis Eradication Campaign ( PATTEC ) , a political initiative of the African heads of state that calls for increased efforts to manage the tsetse and trypanosomosis problem . Tsetse populations may be reduced using a variety of techniques , including insecticide impregnated traps and targets , live-baits , sequential aerial spraying , and sterile male releases [4]–[9] . In the past , most control efforts were not implemented according to area-wide principles [10] , [11] , and as a consequence , when the control effort was reduced or stopped , the tsetse populations tended to recover – due to either flies surviving the initial interventions , or migrant flies coming from untreated regions , or both [12] . This has fuelled a debate as to whether in some instances “eradication” , defined by FAO [13] , [14] as the creation of a tsetse free zone , may be more cost effective than “suppression” where tsetse densities are reduced to a level minimizing the risk of disease transmission . A sound decision whether to select an eradication or suppression strategy will be facilitated when the population structure within the target region , in particular the degree of genetic isolation of the target population from its adjacent populations is clearly understood . For isolated populations , eradication may be the most cost-effective strategy , as reported for Glossina austeni Newstead in Unguja Island , Zanzibar [5] . But for most mainland populations of tsetse , the geographical limits of target tsetse populations are less easily defined . Application of population genetics techniques can quantify rates of gene flow between sub-populations [15]–[19] , and guide decisions on the choice of control strategies [20] . The level of isolation of the targeted tsetse populations will be an important parameter to guide the Government of Senegal to select the most optimal control strategy . Here we report population genetic analyses of microsatellite and mtDNA markers combined with morphometrics of G . p . gambiensis populations sampled from the Niayes area and from the nearest population in the south-eastern part of the country ( fig . 1 ) to assess their degree of isolation by measuring gene flow among the different populations . The genetic differentiation of the various G . p . gambiensis populations within the Niayes was also assessed to determine if the different populations of the Niayes can be targeted at the same time ( if it is a single panmictic unit ) , or if a sequential control strategy can be contemplated ( if substantial genetic differentiation between populations is found ) . , which will also depend on their respective history , including effective population sizes and possible bottlenecks .
In the Niayes region , four tsetse populations were sampled using Vavoua traps: Dakar Hann which is a swamp forest harbouring an animal park within the city of Dakar , Diacsaw Peul , an area of riparian thicket where tsetse and cattle are in intense contact , Sebikotan and Pout , which are mango and citrus-tree plantations where tsetse and people are in close contact . The tsetse flies collected in the Niaye area and analyzed in the present study are the sole property of the Senegalese authorities . They were collected by the national veterinary services through official mission orders , in one wildlife park ( Parc de Hann ) and three private sites , with the oral consent of the owners . No written consent is mandatory for tsetse fly collection in Senegal . In the south-eastern part of the country , the area of Missira was sampled: it is the nearest known infested area from the Niayes , according to a detailed tsetse survey implemented as part of the baseline data collection of the Niayes tsetse Control project ( see Fig . 1 ) . Areas between the Niayes and Missira are not favourable for tsetse , which was confirmed by zero tsetse catches ( JB , BS unpublished data ) despite intensive trapping efforts . In total 153 tsetse individuals were analysed originating from Diacsaw Peul ( 22 females ( F ) , 8 males ( M ) ) , Dakar Hann ( 23F , 6M ) , Sebikotan ( 21F , 11M ) , Missira ( 23F , 12M ) , and Pout ( 13F , 14M ) . A portion of the 5′ end of the mitochondrial gene COI was amplified , purified and sequenced using the primers CI-J-2195 TTGATTTTTTGGTCATCCAGAAGT [55] and CULR TGAAGCTTAAATTCATTGCACTAATC using the same conditions reported by [56] . The following statistics were calculated using DNAsp version 4 . 50 . 3 [57]: Hd Haplotype diversity , was calculated using equations 8 . 4 and 8 . 12 in [58] , Pi , the nucleotide diversity , which is the average number of nucleotide differences per site between two sequences , and its sampling variance was calculated using equations 10 . 5–10 . 7 in [58] , K , the average number of nucleotide differences and the total variance of K is ( sampling plus stochastic ) , assuming no recombination were calculated using equations from [59] . FST was calculated according to equation 3 in [60] , HST was calculated according to equation 2–4 , and K*ST according to equations 7–11 in [60] . HST is a haplotype frequency based genetic differentiation statistic that does not take into account the number of differences separating different haplotypes . HST = 1− ( HS/HT ) , where HS is the weighted average of subpopulation genetic diversities and HT is the estimated haplotype diversity of the total population . K*ST = 1− ( KS/KT ) , where K*S is a weighted average of the log corrected average number of sequence difference in the populations being compared , and KT is the average number of difference between sequences . 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 K*ST ) [60] . The average number of nucleotide difference ( equation A3 , [61] ) and its variance ( sampling plus stochastic ) were also calculated using DNAsp version 4 . 50 . 3 . Haplotype trees for 738 nucleotides of COI from G . p . gambiensis ( data from a total of 148 individuals in this study ) were generated using the algorithm of [62] . The TCS 1 . 21 programme was used to estimate the haplotype tree , with the connection limit ( probability of parsimony ) at 95% [63] . The maximum number of connection steps at 95% was 11 . In total , the number of analysed wings was 20 , 21 , 18 , 34 and 18 for Diacsaw Peul , Dakar Hann , Sebikotan , Missira and Pout respectively . These analysed individuals were all also analysed by the molecular markers . Wings were dry-mounted between two microscope slides and scanned with a scanner . From this picture , 10 landmarks defined by vein intersections were recorded as previously described [64] . Each landmark has X and Y coordinates , and the 10 LM defined per wing represent a polygon . After scaling , translating and rotating all these polygons so that they can be compared , data were subjected to generalized Procrustes analysis ( GPA ) [65] , [66] allowing to implement shape variables , here represented by 16 “partial warps” ( PW ) ( including uniform component of shape ) . These PW were used to conduct a discriminant analysis to allow for individual reclassification based on Mahalanobis distances ( noted DM ) , which were calculated between populations . The statistical significance of Mahalanobis distances was estimated by 1 , 000-runs permutation tests [67] . A re-classification score was computed where individuals are assigned to each group and the percentage of good classification was then calculated . To evaluate the correlation between distance matrices we undertook three Mantel tests [68] with the “Mantelize it” option of Fstat 2 . 9 . 4 [27] between DC-S&E and Kst* , between DC-S&E and DM , and between Kst* and DM .
The statistics of genetic differentiation between the Niayes population samples from Sebikotan , Pout , Dakar Hann and Diasca Peuhl and the south-eastern Senegal site Missira based on 738 nucleotides of mitochondrial cytochrome oxidase 1 ( COI ) are shown in table 1 . Statistics of genetic differentiation based on nucleotide sequence and haplotype frequency both indicate a very high level of genetic differentiation between Niayes populations and Missira , since there were no common haplotypes between the two regions ( see Fig 4 ) . Genetic differentiation between Diacsaw Peul and Pout was also significant ( 0 . 05>P>0 . 01 ) . Otherwise , there was no significant evidence for differentiation between populations within the Niayes region . The haplotype and nucleotide diversity of the Niayes populations was very low in all cases ( see table 2 ) . An analysis of the shape of the wings was conducted by using Procustes analysis to compare wing shape among the five different populations . All Mahalanobis distances between populations were significant except between Pout and Sebikotan . This is illustrated by reclassification scores which were over 70% indicating differentiation between populations ( see Table 3 ) . Missira sample showed the best one ( 85% ) confirming this sample was easily distinguished from the others . The correlation between the genetic distances and morphometrric distances were all strong and significant ( see Figure 5 ) . The strong correlation between microsatellites and COI distance matrices was mainly due to the fact that there are two sets of points: those from the Niayes sites and those differentiating the Niayes from Missira ( very strong distances for both markers ) . For the correlation between genetic and morphometric data , Figure 5 shows that around 50% of the variance in Mahalanobis distance can be explained by genetics , meaning that the other 50% are probably explained by environmental parameters .
The results of the genetic and morphometric analyses indicated limited gene flow between the G . p . gambiensis population of the Niayes and those of the main tsetse belt in the south-eastern part of Senegal . Using three different kinds of markers , i . e . microsatellite DNA , mitochondrial DNA and geometric morphometrics , the data led to the conclusion that the G . p . gambiensis population from the Niayes can be considered isolated with very little risk of re-invasion should the population eventually be eradicated . These data corroborate the results of the entomological baseline data collection and observations in the field indicating the absence of G . p . gambiensis in the 120 km long area between Missira and the Niayes ( JB , BS unpublished results . ) . On this basis , eradication of G . p . gambiensis from the Niayes can therefore be recommended as an appropriate control strategy . The data from this study , which was part of a comprehensive baseline data collection effort , confirmed that eradication can be recommended as an appropriate control strategy , and as such the study greatly assisted in the decision making on which strategy to select . Although the isolation of the target population is not an absolute prerequisite for AW-IPM , tackling a continuous pest populations is actually more complex , requiring more resources and a long-term , regional commitment ( for details , see [10] , [11] , [14] . Looking at FST values reported in previous studies on G . p . gambiensis , the one observed in the present work between Missira and the Niayes populations was ten times higher than those observed along 260 km on the Mouhoun river in Burkina Faso [17] , and two times higher than the values observed in Guinea between Loos islands and the continent [64] . These values are of the same order of magnitude as those observed between the two different taxa , G . p . gambiensis and G . palpalis palpalis ( DK , PS , unpublished data ) . It is also noteworthy that some of the microsatellite loci used on the individuals in this study amplified poorly , a behaviour not recorded in earlier studies of G . p . gambiensis [17] , [19] , [64] . This may be an additional argument for genetic divergence , since it is known that in different taxa polymorphisms in sequences flanking the microsatellite may occur [69] leading then to mismatches in the primer binding sites . If gene flow between the sites occurs then shared haplotypes of the mitochondrial gene COI would be expected . In fact , we observed no haplotypes shared between Missira and the Niayes region . Future work should include more detailed examination of potential cryptic species within the distribution area of G . p . gambiensis since such cryptic species have been suspected and shown to exist in closely related taxa [70] , [71] . In case the Government of Senegal decides to integrate the sterile insect technique ( SIT ) as part of the AW-IPM ( area-wide integrated pest management ) approach , it would be advisable that prior to the release of sterile males , experiments be conducted to assess mating compatibility between the G . p . gambiensis from the Niayes and those which should be released if they do not originate from the same area ( e . g . from Burkina Faso ) . Although previous studies with G . p . gambiensis originating from Mali and Burkina Faso revealed no mating barriers between these populations ( G . Mutika , personal communication ) , nor did another study between G . p . gambiensis and G . p . palpalis [72] , these tests would provide evidence that genetic differentiation of the Niayes population has not been accompanied by pre-mating barriers , that would threaten the success of the programme . Similarly , it may also be recommended to release sterile flies coming from Burkina Faso in the Niayes to assess their behaviour and performance ( dispersal , dispersion , mobility , lifespan , mating frequency etc . ) in the natural habitat . Indeed , the particular eco-system of the Niayes , the fragmented nature of the preferred vegetation types of G . p . gambiensis ( mango-tree plantations , Euphorbia spp . hedges ) and an annual precipitation below 400–500 mm might have induced the development of some level of xerotolerance of these relatively small populations [73] , [74] . The second question that has relevance for a future AW-IPM programme was to know whether within the Niayes region , the four populations showed any genetic differentiation , or if they constituted a single , panmictic ( i . e random mating ) population . Low haplotype diversity of the populations of this area is in agreement with a previous Single Strand Conformation Polymorphism-based study on mitochondrial DNA haplotypes in the Niayes region of Senegal [75] . The Niayes population is probably a remnant population , possibly of small size and therefore likely to lose rare haplotypes by genetic drift more rapidly than larger populations . Many ancestral G . p . gambiensis haplotypes have probably been lost from the Niayes region , leaving just two haplotypes remaining today . The low diversity was also observed at microsatellite loci with lower genetic diversity than in Missira . The three different markers used in this study generally showed good agreement in the differentiations observed , as can be seen by the high and significant correlations coefficients computed between them . The only apparent discrepancy between microsatellites on one hand and COI based and morphometrics based distances on the other hand probably comes from the variance expected for paired FST estimates and the small sample sizes , the effect of which is probably increased by historical events such as the probable bottleneck that affected the Dakar-Hann population . This may explain why only microsatellite data suggest that the population from Dakar Hann is so much differentiated from all the others , including the geographically closest ones . From an operational point of view however , the population of Dakar Hann is probably also isolated from the three others , since the bottleneck signature was still visible , suggesting very limited exchanges ( if any ) with the other populations . Field data also corroborate this since the population of Dakar Hann is located within an animal park of this huge city which can be seen as an isolated refuge for this tsetse population . This further suggests that an elimination operation may probably be implemented on this population . The results also suggest that the samples from Diacsaw Peul , Sebikotane and Pout , although showing some genetic differentiation , are not completely isolated from each other . This is also consistent with ecological data since it appears that the maximum distance between the forest patches of this area is less than 2km ( data not shown . ) . To conclude , the use of genetic and morphometric markers has been instrumental in the decision-making process of selecting and developing of an appropriate intervention strategy to create a sustainably zone free of G . p . gambiensis and Trypanosomosis in the Niayes region of Senegal . In the near future , it should be encouraged to carry out such studies prior to the selection of target areas or the choice of control strategies , and these should be part of the overall collection of baseline data ( see ref [76] for recent review ) . In addition , the results obtained here suggest that efforts should be made to look for other genetic discontinuities in G . palpalis s . l . distribution that may be indicative of the presence of cryptic species . | Tsetse flies transmit trypanosomes to humans ( sleeping sickness ) and animals ( nagana ) . Controlling these vectors is a very efficient way to contain these diseases . There are several strategies and methods that can be used for control , each being more or less efficient depending on several factors . The Government of Senegal wants to sustainably eliminate trypanosomosis from the Niayes region by controlling the tsetse vector , Glossina palpalis gambiensis . To reach this objective , two different strategies may be used: suppression ( decrease in tsetse densities ) or eradication ( remove all the tsetse in the region until last one ) . For eradication , the approach has to be area-wide , i . e . the control effort targets an entire pest population within a circumscribed area , to avoid any possible reinvasion . Three different tools ( microsatellite DNA , mitochondrial DNA and morphometrics ) were used , and all showed an absence of gene flow between G . p . gambiensis from the Niayes and from the nearest known population in the south east of the country ( Missira ) . This genetic isolation of the target population leads to the recommendation that an eradication strategy for the Niayes populations is advisable . This kind of study may be extended to other areas on other tsetse species . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"infectious",
"diseases/epidemiology",
"and",
"control",
"of",
"infectious",
"diseases",
"genetics",
"and",
"genomics/population",
"genetics"
] | 2010 | Population Genetics as a Tool to Select Tsetse Control Strategies: Suppression or Eradication of Glossina palpalis gambiensis in the Niayes of Senegal |
Praziquantel ( PZQ ) is the treatment of choice for infections with the liver fluke Opisthorchis viverrini , a major health problem in Southeast Asia . However , pharmacokinetic ( PK ) studies investigating the disposition of PZQ enantiomers ( R- and S-PZQ ) and its main metabolite , R-trans-4-OH-PZQ , in diseased patients are lacking . The implementation of a dried blood spot ( DBS ) sampling technique would ease the performance of PK studies in remote areas without clinical facilities . The aim of the present study is to provide data on the disposition of PZQ enantiomers and R-trans-4-OH-PZQ in opisthorchiasis patients and to validate the use of DBS compared to plasma and blood sampling . PZQ was administered to nine O . viverrini-infected patients at 3 oral doses of 25 mg/kg in 4 h intervals . Plasma , blood and DBS were simultaneously collected at selected time points from 0 to 24 h post-treatment . PK parameters were determined using non-compartmental analysis . Drug concentrations and areas under the curve ( AUC0–24h ) measured in the 3 matrices were compared using Bland-Altman analysis . We observed plasma AUC0–24hs of 1 . 1 , 9 . 0 and 188 . 7 μg/ml*h and half-lives of 1 . 1 , 3 . 3 and 6 . 4 h for R-PZQ , S-PZQ and R-trans-4-OH , respectively . Maximal plasma concentrations ( Cmax ) of 0 . 2 , 0 . 9 and 13 . 9 μg/ml for R-PZQ , S-PQZ and R-trans-4-OH peaked at 7 h for PZQ enantiomers and at 8 . 7 h for the metabolite . Individual drug concentration measurements and patient AUC0–24hs displayed ratios of blood or DBS versus plasma between 79–94% for R- and S-PZQ , and between 108–122% for R-trans-4-OH . Pharmacodynamic ( PD ) in vitro studies on PZQ enantiomers and R-trans-4-OH-PZQ are necessary to be able to correlate PK parameters with efficacy . DBS appears to be a valid alternative to conventional venous sampling for PK studies in PZQ-treated patients .
Opisthorchiasis is caused by the trematode Opisthorchis viverrini , a liver fluke affecting about 8 million people in Southeast Asia , particularly in the Mekong basin [1 , 2] . Infection occurs following consumption of raw or undercooked fish harboring O . viverrini metacercariae [3] . In the early phase , the disease is mostly asymptomatic but in the acute stage periductal fibrosis and liver enlargement are common , mostly a result of inflammation due to worm feeding . The chronic stage triggers severe clinical symptoms including jaundice , biliary obstructions , and cholangiocarcinoma as a serious complication [4–7] . Praziquantel ( PZQ ) is the drug of choice for opisthorchiasis and is manufactured as a racemic mixture of R and S enantiomers . The recommended treatment regimen is 3 oral doses of 25 mg/kg , usually administered between 4 and 6 h apart [8] . The disposition of PZQ is highly influenced by the fasting state , the co-administered food type , as well as the liver function [9 , 10] . Though no studies have been conducted against O . viverrini yet , R-PZQ is considered to be the active molecule in the treatment of schistosomiasis , while the inactive S-PZQ is suspected to be responsible for the bitter taste of the drug and for the mild to moderate adverse events caused by the treatment [11–14] . In humans , PZQ undergoes an enantioselective first-pass metabolism through the cytochrome CYP450 3A4 isoform [15] and is mainly transformed into the monohydroxylated metabolite R-trans-4-OH-PZQ ( R-trans-4-OH ) , while S-PZQ is metabolized to several different monohydroxylated molecules [16–18] . R-trans-4-OH displays minor anthelmintic activity , with an IC50 hundred times higher than R-PZQ against Schistosoma mansoni [19] . The disposition of PZQ enantiomers and metabolites has not yet been studied in opisthorchiasis patients . In fact , the only pharmacokinetic ( PK ) study of PZQ involving patients with opisthorchiasis focuses on the racemic drug [20] . Enantioselective disposition was performed exclusively in healthy volunteers with a low dose [21] . Therefore , studies on the enantioselective disposition of PZQ in the diseased population are warranted for a better understanding of the modalities of drug action and disposition . Dried blood spot ( DBS ) sampling is a microsampling technique , involving the collection of capillary blood through a finger prick . The method is therefore less invasive compared to venipuncture . The blood drops are dried on a filter paper and stored at ambient temperature until assayed . Compared to blood or plasma sampling , this method does not require sample freezing and offers easy handling and storage , hence allowing the performance of PK studies in remote areas without clinical set-ups . Blood quantities withdrawn with DBS are minimal ( 20 μl vs . 3–4 ml for plasma or blood ) , which provides an advantage for research with children . Finally , the ease of sample collection enables including a larger number of patients and is therefore ideal for population PK studies [22–25] . The major caveat when replacing plasma with DBS sampling is the use of a different matrix where the drug partition might not be equivalent [25 , 26] . Validating this alternative sampling technique for future trials hence calls for a formal comparison of drug concentrations measured in plasma and DBS . The aim of our study was to elucidate for the first time the kinetic disposition of both PZQ enantiomers and its main metabolite in O . viverrini-infected patients . Additionally , we assessed the difference between concentrations determined in plasma , blood and DBS sampling for the analysis of PZQ PK profiles using Bland-Altman analysis .
Racemic ( rac ) PZQ was obtained from Sigma-Aldrich ( Buchs , Switzerland ) . PZQ enantiomers as well as the metabolite trans-4-OH were donated by Merck Serono ( Darmstadt , Germany ) . Eleven-fold deuterized PZQ ( PZQd11 , internal standard-IS ) was acquired from Toronto Research Chemicals ( Ontario , Canada ) . The chemical structures of PZQ , PZQd11 and trans-4-OH are depicted in Fig 1 . Acetonitrile , ethanol and methanol of MS grade were purchased from Carl Roth GmBH ( Allschwil , Switzerland ) , and ammonium formate , ammonium acetate and formic acid of MS grade from Sigma-Aldrich ( Buchs , Switzerland ) . Ultrapure water was provided using a Millipore Milli-Q water purification system ( Merck Millipore , Darmstadt , Germany ) . Blank human plasma and blood were supplied in lithium heparin-coated vacutainer tubes ( BD , Allschwil , Switzerland ) from the local blood donation centre ( Basel , Switzerland ) . The plasma , blood and DBS sample collection was performed in the framework of a PK and dose-finding study of tribendimidine against O . viverrini in humans . Ethical clearance was obtained from the ethics committee of Northern and Central Switzerland ( EKNZ reference no . 375/11 ) , and from the National Ethic Committee for Health Research , Ministry of Health ( MoH ) of Lao PDR ( reference no . 009/NECHR ) . The trial is registered at Current Controlled Trials ( ISRCTN96948551 ) . In short , 9 O . viverrini-infected patients were treated with 3 oral doses of 25 mg/kg PZQ , with the second and third dose administered 4 and 8 h after the first dose , respectively . The trial was performed at the Champasak Provincial Hospital in Pakse , Lao PDR , and prior to treatment , a standardized food dish ( rice ) was provided to all patients . Adverse events were monitored at 3 , 24 and 48 hours post-treatment using a standardized questionnaire . Prior to treatment , patients underwent physical examinations and laboratory tests , such as liver and kidney parameters and complete blood counts . About 4 ml of venous blood was collected at 0 , 2 , 4 , 6 , 8 , 8 . 5 , 9 , 10 , 11 , 12 and 24 h after the first dose from the antecubital arm vein through an intravenous catheter into EDTA-coated vacutainer tubes ( BD ) . Within 30 min after sampling , 1 ml of blood was pipetted into a cryotube and the remaining blood centrifuged to obtain plasma . Plasma and blood samples were transported on dry ice to Basel where they were kept at -80°C until analysis . DBS samples were collected at 0 , 4 , 8 , 9 , 11 and 24 h post-first-dose from patients 1 to 5 , and at 0 , 2 , 6 , 8 . 5 , 10 and 12 h post-first-dose from patients 6 to 9 . The samples were obtained by puncturing the middle or ring finger with sterile one-way finger prickers ( Accu-Check Safe-T-Pro Plus , Roche , Switzerland ) . Lithium heparin coated capillaries ( Alere Cholestech LDX , V = 40 μl ) were used to collect and deposit blood on DMPK-C cards ( Whatman , GE Healthcare Life Sciences , Cardiff , UK ) . The cards were dried overnight and stored in plastic bags with desiccant at room temperature . The LC-MS/MS method for the analysis of R- and S-PZQ and R-trans-4-OH and its validation for plasma , blood and DBS is described elsewhere [27] . Briefly , plasma and blood calibration samples were freshly prepared and included in each analytical run by spiking blank samples to reach final concentrations from 2 . 5 to 0 . 01 ( lower limit of quantification-LLOQ ) μg/ml for R- and S-PZQ , and of 25 to 0 . 1 ( LLOQ ) μg/ml for R-trans-4-OH . QC samples were similarly prepared by spiking 6 different blanks to obtain high , medium , low and LLOQ concentrations . For the extraction of analytes , 100 μl of plasma or blood samples underwent protein precipitation with 700 μl of IS solution ( 500 ng/ml IS in pure acetonitrile ) , and were shaken in a thermomixer for 20 min at 25°C and 1400 rpm . DBS samples of 5 mm diameter were extracted with 300 μl of DBS extraction solution ( IS solution: ultrapure water , 4:1 , v/v ) , shaken in a thermomixer for 20 min at 25°C , and sonicated for 40 min prior analysis . A first chromatographic separation was achieved through a column trapping system ( HALO C-18 , 4 . 6 x 5 mm , Optimize Technologies , OR , USA ) using 10 mM ammonium acetate and 0 . 15% formic acid in ultrapure water at a flow rate of 0 . 3 ml/min . After 1 min , the analytes were eluted from the trapping to the chiral column ( Lux Cellulose-2 ( 150x4 . 6mm , 3μm , Phenomenex , CA , USA ) ) with an elution gradient of 70 to 90% B , with mobile phase A consisting of 20 mM ammonium formate in ultrapure water and mobile phase B of pure acetonitrile . Statistical analyses were performed with Prism software ( GraphPad , CA , USA ) . Parasite egg counts were determined with duplicate Kato-Katz smears from two stool samples prior to the treatment and between 19 and 25 days after treatment for the estimation of treatment efficacy . Cure rates were defined as the percentage patients who were egg-negative after treatment . The number of eggs per gram of stool ( EPG ) was evaluated by adding up the egg counts from the quadruplicate Kato-Katz thick smears and multiplying this number by a factor of six . Geometric mean egg counts were calculated before and after treatment to determine the corresponding percentage egg-reduction rate ( ERR ) . To evaluate the reproducibility of the measurements , incurred sample reanalysis ( ISR ) was performed with a total of 170 samples originating from 5 patients in the 3 matrices ( 56% of total sample size ) . The percentage difference between the original and the reanalyzed measurement was calculated as follows: percentage difference= ( repeat−original ) ×100mean ( repeat , original ) As acceptance criterion for ISR , at least 66 . 7% of the samples ( 2 out of 3 ) should not deviate by more than 20% , as recommended in the European guidelines on bioanalytical method validation and the daft of the FDA guidelines [28 , 29] . PK parameters , including the area under the concentration-time curve ( AUC0–24h ) , the maximal concentration ( Cmax ) , the time to maximal concentration ( Tmax ) and the half-life ( t1/2 ) were calculated for each patient with the Excel add-in PKsolver [30] using non-compartmental analysis with the linear trapezoidal rule . Concordance of drug concentrations observed in blood or DBS compared to plasma was evaluated using Pearson’s correlation coefficient and Bland-Altman plots , with percentage ratios between the two matrices ( blood/plasma or DBS/plasma ) plotted against mean concentrations [31–33] . For matrix differences in AUC values , Bland-Altman analysis also applied . The limits of agreement at 95% for the ratios were calculated as follows: limits of agreement=mean percentage ratio ± 1 . 96× standard deviation For the drug concentration data , the calculation of the limits of agreement were adapted to take into account multiple measurements per individuals , following the method described by Bland and Altman [34] using Stata software ( version 12 . 1 , College State , TX , USA ) . The partitioning of PZQ between plasma and erythrocytes was assessed in vitro using human blood from the local blood donation center . Whole blood samples ( hematocrit adjusted to 35% ) were spiked in triplicate with the analytes of interest to reach end concentrations of 0 . 05 and 0 . 5 μg/ml for R- and S-PZQ and 0 . 5 and 5 μg/ml for R-trans-4-OH . The samples were incubated 1 h at room temperature . Hundred microliters of each sample were aliquoted and the remainder was centrifuged at 1500 g for 20 minutes to obtain plasma . Whole blood and plasma samples were extracted using acetonitrile containing IS and analyzed as described above . Analyte peaks were normalized with IS peaks and ratios of blood to plasma were calculated for each concentration and analyte .
A total of 91 plasma and 91 blood samples were collected . For DBS , 45 samples were analysed . Due to technical problems , five DBS samples for patient 9 were collected at 6 , 7 , 8 . 5 , 10 and 12 h post-treatment , and an extra venous blood sample was withdrawn at 7 h post-treatment . To estimate the repeatability of the analytical measurements , an incurred sample reanalysis was performed . Between 88 . 2 and 100% of the samples in plasma , blood and DBS were within the ISR acceptance criterion ( within 20% difference ) . All samples from the 9 patients presenting obvious measurement errors or displaying a high discrepancy between matrices were reanalysed ( n = 21 ) . For R-PZQ , concentrations ranged from 0 . 01 to 0 . 85 μg/ml , to 0 . 90 μg/ml and to 1 . 08 μg/ml for DBS , blood and plasma , respectively . For S-PZQ , the following concentration ranges were observed: 0 . 01–1 . 59 μg/ml in DBS , 0 . 01–1 . 83 μg/ml in blood , and 0 . 02–2 . 34 μg/ml in plasma . The metabolite R-trans-4-OH displayed concentrations ranging from 3 . 01 to 19 . 01 μg/ml in DBS , 1 . 62 to 22 . 05 μg/ml in blood , and 1 . 41 to 17 . 85 μg/ml in plasma . All participants were adults , 3 males and 6 females aged 25 to 46 years with a median weight of 56 kg ( Table 1 ) . Prior to treatment , 8 patients displayed moderate O . viverrini infections ( between 1 , 000 and 10 , 000 EPG ) and 1 patient a heavy infection ( 13 , 920 EPG ) . All patients were asymptomatic . Hookworm co-infections were present in 5 participants ( participants 1 , 4 , 6 , 7 and 8 ) , while patient 2 presented a co-infection with the whipworm Trichuris trichiura . Liver and kidney parameters were in the normal range for all the patients . Blood counts were also normal , except for patient 2 who displayed slightly elevated white blood cell counts ( 11 . 7 * 103 cells/l ) and a moderate anaemia ( hemoglobin concentration = 9 . 2 g/dl ) . All patients were treated as planned and tolerated the treatment well , with the exception of patient 2 , for whom the treatment was interrupted due to adverse events ( vomited within 30 min after the second dose ) . Between 19 and 25 days post-treatment , the participants were screened again for the presence of O . viverrini eggs in stool: all patients were cleared from infection , hence cured ( Table 1 ) . Patient variability in plasma concentrations was high , with patient 2 displaying clearly higher concentrations than the other subjects , despite not taking the last dose ( Fig 2 ) . Median PK parameters calculated from plasma concentrations are summarized in Table 2 . R-PZQ displays the smallest AUC0–24h ( 1 . 1 μg/ml*h ) and a short estimated t1/2 ( 1 . 1 h ) compared to the other analytes . S-PZQ exhibits a nearly 5 x higher Cmax ( 0 . 9 μg/ml ) and an AUC0–24h more than 8 x higher than R-PZQ ( 9 . 0 μg/ml*h ) . Both enantiomers peak at the same time ( 7 h ) . The main metabolite R-trans-4-OH has an increased exposure compared to the parent molecule . For example , its AUC0–24h ( 188 . 7 μg/ml*h ) is 20x greater than S-PZQ and 170x greater than R-PZQ . The metabolite’s estimated t1/2 and Tmax are 6 . 4 h and 8 . 7 h , respectively . Patient 2 displays 2–10 fold higher R-PZQ , S-PZQ and R-trans-4-OH AUC0–24h values compared to the other patients . When comparing the analyte concentrations obtained in the different matrices by Pearson’s correlation coefficient , blood versus plasma and DBS versus plasma data displayed correlation coefficients above 0 . 92 ( all p <0 . 01 , Table 3 ) . The mean concentration curves are consistent for plasma , blood and DBS , as exemplified with patient 1 ( Fig 2 ) . The modified Bland-Altman approach for multiple measurements per individual was used on drug concentrations , although the values obtained with this method did not differ from the conventional approach . The Bland-Altman plots ( Fig 3 ) show that percentage ratios were generally consistent across concentrations . The mean percentage ratios of R-PZQ in blood or DBS compared to plasma , display ratios of 79 . 0 and 89 . 6% , respectively ( Table 3 ) . There is therefore a tendency for plasma samples to have slightly higher concentrations of R-PZQ than blood or DBS . For S-PZQ , the same pattern is observed , with slightly higher ratios than R-PZQ: 93 . 9 and 92 . 1% percentage ratios for blood and DBS , respectively . The metabolite R-trans-4-OH displays on the contrary higher ratios of blood or DBS to plasma of 122 . 0 and 110 . 6% , respectively . However , the 95% limits of agreement ( LoA ) all include 100% , except for R-PZQ in the blood versus plasma comparison ( LoA = 59–99% ) . The LoA intervals are large and range up to 55–133% for the parent enantiomers and 94–145% for the metabolite in the blood-plasma ratios . LoA are slightly larger for DBS-plasma ratios . Bland-Altman plots of AUC percentage ratios between plasma and blood or DBS are consistent across AUC values ( Fig 4 ) . The percentage ratios of blood or DBS to plasma for each PK parameter exhibit values between 80 and 120% , except in t1/2 DBS to plasma ratios for R-PZQ ( 122% ) and R-trans-4-OH ( 75% , Tables 4 and 5 ) . As for drug concentrations , the mean ratios of DBS or blood versus plasma tend to be lower than 100% for the parent enantiomers but higher than 100% for the metabolite . Only the t1/2 DBS to plasma ratios do not precisely follow this pattern , probably because they are calculated with 5 samples per patient instead of 10 , driving therefore a higher estimation error . The LoA of the AUC ratios lie between 64–107% for the parent enantiomers and 87–136% for the metabolite in the blood-plasma ratios with a slightly wider range for DBS-plasma ratios . As for drug concentration results , all the LoA include 100% , except for R-PZQ in the blood to plasma ratio . Blood to plasma ratios were consistent across both concentrations measured . R-PZQ displayed ratios of 83 . 1 ± 3 . 6 and 77 . 7 ± 3 . 9% for 0 . 05 and 0 . 5 μg/ml , respectively . The partition of S-PZQ in plasma was similar to R-PZQ , with ratios of 81 . 3 ± 5 . 4 and 74 . 5 ± 1 . 5% for low and high concentrations , respectively . The metabolite R-trans-4-OH showed a partition in plasma higher than the parent enantiomers , with values of 92 . 0 ± 6 . 0 and 87 . 5 ± 6 . 2% for 0 . 5 and 5 μg/ml , respectively .
PZQ is the only drug available for the treatment of opisthorchiasis , yet surprisingly preclinical and clinical work including PK studies are sparse . We conducted for the first time a PK study in patients infected with O . viverrini treated with three doses of PZQ and studied the enantioselective drug disposition in blood , plasma and DBS . The only other PK study conducted with O . viverrini-infected patients so far examined the kinetic disposition of the racemic drug after a single oral dose of 40 mg/kg [20] . Patients were of similar age and weight as in our study , but with a higher proportion of males . The authors observed a Cmax for racemic PZQ of 0 . 9 and 1 . 1 μg/ml in early ( asymptomatic ) and acute ( moderately advanced ) infection , respectively which does not differ from the value observed in our study , 1 . 1 μg/ml for R-and S-PZQ combined . A half-life value of 2 . 3 and 3 . 8 h previously observed for the racemic parent compound in early and acute infection [20] is as well consistent with a half live of S-PZQ of 3 . 3 h ( R-PZQ is eliminated much faster , due to different enzymes kinetics than S-PZQ ) in our study . Given that a dose of 40 mg/kg corresponds to 53% of the dose administered in our trial , the AUC0–24h range observed in our study ( 6 . 1–26 . 3 μg/ml ) is closer to that previously reported for patients with acute opisthorchiasis ( 2 . 5–15 . 6 μg/ml ) than that in patients with asymptomatic opisthorchiasis ( 1 . 6–5 . 0 μg/ml ) [20] . This result is not surprising , given the disease prevalence in the region and the age of the patients , for which acute cases are expected to be frequent [5 , 35 , 36] . We observed a high variability in analyte concentrations between patients . This is often observed in PK studies with PZQ and is likely due to the high first-pass metabolism of PZQ in the liver or gut , with the activity of CYP 450 being highly dependent on the health , genetic and nutritional status of the patient [12] . Multiple dosing can also add to variability , since differences among patients in absorption and elimination as well as competition/saturation effects are common and exacerbate each other . The high AUC0–24h values of the PZQ enantiomers and metabolite determined for patient 2 ( taking only two doses instead of three ) might be explained by several factors . Firstly , this is the only patient suffering from a co-infection with whipworms and hookworms at follow up . Changes in drug metabolism due to immune reactions due to three co-existing parasites might be possible , as some immunomodulators were found to decrease hepatic activity [37 , 38] . Secondly , patient 2 displays the highest weight to height ratio ( body mass index of 35 . 2 kg/m2 ) , which could lead to an overestimation of the effective drug dose , as it is often the case in overweighed patients [39] . Finally , this patient might have developed liver and intra-hepatic bile duct pathologies , thereby altering drug metabolism , as observed in patients infected with another liver fluke , Fasciola hepatica [37] . Although measurement of liver enzyme parameters and the physical examination did not identify this patient as a symptomatic opisthorchiasis case , ongoing liver pathology can not be ruled out [20] . In fact , the detection of hepatic abnormalities due to opisthorchiasis , such as fibrosis or moderate hepatomegaly , is recommended to be performed via ultrasonography ( not done in the present study ) , as liver enzymes do not seem to be a reliable indicator for the pathology of this disease [7 , 40] . Not surprisingly , patient 2 suffered from adverse events during the treatment course , as high Cmax levels are often correlated with adverse events [41] . It might be worth highlighting that this patient as well all other study participants were cured following PZQ treatment . The high efficacy noticed with a triple dose of PZQ is in accordance with previous studies [42] . The patient with the highest infection intensity ( patient 7: 13 , 920 EPG at baseline ) displayed parent and metabolite AUC0–24h values similar to the other patients with moderate EPG values ( between 1 , 000 and 10 , 000 EPG at baseline ) , hence infection intensity does not seem to correlate with PZQ disposition . The most striking result observed in the disposition of PZQ is the high concentration of R-trans-4-OH , culminating at 13 . 9 μg/ml . For comparison , a study from Lima et al . [21] conducted in healthy volunteers treated with a single oral dose of 25 mg/kg PZQ displayed a 10x lower Cmax of R-trans-4-OH . This finding , which might be explained with changes in metabolism due to the liver disease , raises the question of the role of the metabolite in the opisthorchicidal activity of PZQ . In vitro and in vivo studies should be conducted to assess the activities of R-and S-PZQ and R-trans-4-OH against O . viverrini . The incurred sample reanalysis revealed a proportion higher than 2/3 of the samples falling into the acceptance criterion of deviating no more than 20% . These results demonstrate that the measurements are reliable and that there are no major problems in sample handling , processing or analysis . The hematocrit of 35% used for the calibration line and the 25–50% range of hematocrits used for the quality controls reflects values in our patients ( mean hematocrit of 35 . 5 ± 4 . 1% ) , and more generally values encountered in Southeast Asia . For example , in Thailand , the mean hematocrit in men is between 42 and 47% , while in women it lies between 37 and 39% [43] . All the LoA intervals included 100% , indicating no difference between DBS or blood compared to plasma concentrations , except for R-PZQ in blood . The LoAs observed were wide , which can be explained by the additive measurement errors in each matrix . When validating a bioanalytical method , the accepted measurement variability is of ± 15% . This translates to indicative maximal LoA of 71–129% ( calculated using the conventional Bland-Altman formula with SD = 15% ) , which is broadly similar to the results observed in this study . The wider LoA and confidence intervals for DBS-plasma compared to blood-plasma AUC0–24h ratios reflect the half as small sample size for the estimation of DBS AUC0–24hs compared to blood AUC0–24hs samples . In light of these observations , we estimate that there is a general agreement between matrices and that DBS is a valid surrogate to venous sampling . In the Bland-Altman comparisons of blood versus plasma or DBS versus plasma , R- and S-PZQ displayed drug concentrations and AUC0–24h percentage ratios of around 80% and R-trans-4-OH ratios higher than 100% . The higher concentrations observed for R- and S-PZQ when quantified in plasma compared to blood or DBS might arise from a very high affinity of the drug for plasma proteins . PZQ is highly protein-bound ( ~80% ) [12] . Hence , red blood cells might have a slight diluting effect on PZQ concentrations , depending on the blood hematocrit [26] . For example , tasquinimod , an anticancer drug characterized by a very high plasma binding ( >98% ) , revealed a blood:plasma ratio of 66% [44 , 45] . This phenomenon was also observed in a study comparing DBS and plasma sampling with piperacillin and tazobactam in infants with DBS:plasma ratios between 50 and 60% [46] . Therefore , our results for the parent enantiomers are in line with previous observations in drugs with high plasma affinity and displayed an agreement between plasma and blood or DBS of around 80–90% . In contrast , R-trans-4-OH did not follow such pattern and displayed higher concentrations in blood or DBS than in plasma , which likely indicates a lower affinity for plasma proteins than its parent molecule and a higher repartition in erythrocytes . The in vitro evaluation of PZQ partition between plasma and erythrocytes highlighted a higher affinity of the enantiomers to plasma , which echoes the observations in patients described above . Considering that PZQ is bound to 80% to plasma proteins [12] and that acetonitrile precipitation extracts both the unbound and bound fractions , the blood to plasma ratios between 75 and 83% indicate that penetration of the free fraction into erythrocytes is very limited to almost absent . On the other hand , the metabolite R-trans-4-OH displays in vitro an almost even distribution in all blood compartments , while in patient samples this ratio is slightly more biased towards erythrocytes . In conclusion , we have shown that DBS is a valid alternative to plasma sampling for PK studies with PZQ . Additional studies are warranted to estimate the kinetic disposition of patients after different PZQ dosing schemes and to investigate the PK/PD relationship , in particular the role of R–trans-4-OH in the opisthorchicidal activity of PZQ . | Opisthorchiasis , caused by the food-borne trematode Opisthorchis viverrini , affects more than 8 million people in Southeast Asia , and in its chronic phase it might lead to cholangiocarcinoma . Praziquantel ( PZQ ) is the sole drug available to treat the disease and is administered as a racemic mixture of R and S enantiomers , of which R-PZQ is considered active . As PZQ is rapidly metabolized , its disposition and efficacy in patients might considerably vary according to disease state , sex or age . However , pharmacokinetic ( PK ) studies on the disposition of PZQ enantiomers and its main metabolite , R-trans-4-OH , in diseased patients are lacking . To allow the collection of PK samples in a large number of patients , we implemented a dried blood spot ( DBS ) technique , which is less invasive than venipuncture . The aim of our study is to provide first data on the disposition of PZQ enantiomers and the main metabolite of PZQ in opisthorchiasis patients and to validate the use of DBS over venous sampling . Standard PZQ treatment was administered to nine O . viverrini infected patients , and plasma , blood and DBS were simultaneously collected within 24 h post-treatment . We observed a 100-fold higher disposition of the metabolite compared to R-PZQ , which questions its role in the opisthorchidal activity of PZQ . DBS sampling appears to be a valid alternative to venous sampling and will be a valuable tool for future PK studies in PZQ-treated patients . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"body",
"fluids",
"stereoisomers",
"chemical",
"compounds",
"tropical",
"diseases",
"parasitic",
"diseases",
"isomerism",
"foodborne",
"trematodiases",
"metabolites",
"pharmaceutics",
"neglected",
"tropical",
"diseases",
"pharmacolog... | 2016 | Pharmacokinetic Study of Praziquantel Enantiomers and Its Main Metabolite R-trans-4-OH-PZQ in Plasma, Blood and Dried Blood Spots in Opisthorchis viverrini-Infected Patients |
Phenotypic plasticity is associated with non-genetic drug tolerance in several cancers . Such plasticity can arise from chromatin remodeling , transcriptomic reprogramming , and/or protein signaling rewiring , and is characterized as a cell state transition in response to molecular or physical perturbations . This , in turn , can confound interpretations of drug responses and resistance development . Using BRAF-mutant melanoma cell lines as the prototype , we report on a joint theoretical and experimental investigation of the cell-state transition dynamics associated with BRAF inhibitor drug tolerance . Thermodynamically motivated surprisal analysis of transcriptome data was used to treat the cell population as an entropy maximizing system under the influence of time-dependent constraints . This permits the extraction of an epigenetic potential landscape for drug-induced phenotypic evolution . Single-cell flow cytometry data of the same system were modeled with a modified Fokker-Planck-type kinetic model . The two approaches yield a consistent picture that accounts for the phenotypic heterogeneity observed over the course of drug tolerance development . The results reveal that , in certain plastic cancers , the population heterogeneity and evolution of cell phenotypes may be understood by accounting for the competing interactions of the epigenetic potential landscape and state-dependent cell proliferation . Accounting for such competition permits accurate , experimentally verifiable predictions that can potentially guide the design of effective treatment strategies .
The phenotypic plasticity of many tumors can confound the identification of effective therapeutic strategies [1–4] . For such tumors , even if the cancer cells are isogenic , the cellular composition can be a heterogeneous mix of different cell states ( phenotypes ) that exhibit the capacity for dynamic interconversion . Each phenotype can have a characteristic gene expression profile , drug susceptibility , proliferation rate , and metastatic potential [5] . When this heterogeneous population is challenged with a physical or molecular perturbation , the cell states can rapidly evolve [6] to form a new population distribution better suited to survive the challenge . This adaption may proceed without genetic changes [7–9] . Removal of the challenge can lead to recovery of the original population distribution [5 , 10 , 11] . This behavior bears similarities to that of ‘phenotypic equilibria’ [1 , 12 , 13] . In those systems , if a subset of this population of microstates is physically separated from a stable , heterogeneous population and allowed to expand in culture , the phenotypic heterogeneity of the original culture will recover . This facile adaptability makes plastic tumors challenging to drug-target , and it highlights the importance of quantitative models that can provide predictive and mechanistic insights into the underlying driving force controlling such behaviors . Similarities between steady states in nonequilibrium biological systems and perturbation/relaxation scenarios in classical thermodynamics equilibria have prompted investigations into applying physicochemical models for describing phenotype dynamics within an epigenetic landscape [13–16] . Qualitative descriptive models have been explored for many years , but quantitative and predictive models have only been recently explored [14–20] . In one class of studies , epigenetic landscape models are explored , wherein stable cell states are described as local minima ( attractors ) within a metaphoric energy ( or potential ) cell-state landscape . In such models , the driving forces that influence the cellular composition and population dynamics are the gradients on that surface . As a result , cells tend to gravitate and remain in the local minima of such landscapes . However , in many other cases , this potential landscape does not predict the observed phenotypic heterogeneity [16] , implicating other important factors that can influence the population dynamics are at play . To address this puzzle , we studied highly plastic patient-derived BRAFV600E mutant melanoma cell lines as models of cancer cell phenotypic plasticity . The high rate of both response [21] and resistance development [10] of BRAF-mutant melanoma patients to BRAF inhibitor ( BRAFi ) treatment has made such cell lines important models for understanding challenges associated with targeted inhibitors [9 , 10] . BRAFi can trigger a series of nongenetic cell state changes along the melanocytic lineage towards drug-tolerant and eventually drug resistant states through epigenetic reprogramming . These include the transition of drug-sensitive melanocytic cancer cells into a drug-tolerant neural crest-like phenotype , which , under continued BRAF inhibition , can eventually transition into a fully drug-resistant , invasive mesenchymal-like phenotype [5 , 9–11] . The cell biology of this BRAFi-induced phenotypic evolution has been extensively characterized [5 , 22] , and shown to correlate with what is observed in patient biopsies [9–11 , 22] . However , a quantitative biophysical understanding of this type of epigenetic plasticity has not been fully explored . To this end , we carried out two sets of experiments , integrated with two theoretic approaches , on phenotypically plastic BRAFV600E mutant melanoma cell lines . At the macroscopic level , we measured a kinetic series of bulk transcriptomes over a 2 . 5-month course of low dose BRAF inhibition , during which time the cells evolve from a mostly melanocytic , drug-sensitive phenotype to a mesenchymal , drug-tolerant phenotype . This data set provides input into an information-theoretic surprisal analysis [23] , which is used to identify the relative free energy-like potential over the entire course of cell state transition from drug response to drug tolerance . We also utilized microscopic inputs from flow cytometry to profile , at the single-cell level , the phenotypic evolution of the same system . These phenotypic evolution dynamic data cannot be described with conventional Fokker-Planck equation but can be well recapitulated using a modified Fokker-Planck-type ( FP-type ) kinetic model [17 , 18 , 24] which considered cell-state dependent proliferation differences . The model resolves relative cell state potential and cell-state proliferation differences were quantitatively validated through experiment . We further show that both approaches provide a self-consistent picture in which the combined effects from the relative stability of cellular phenotypes , together with the phenotype-specific net-proliferative rate , act as the drivers to predictably influence the cell population dynamics of drug-induced phenotypic evolution over time . The results provide conceptual guidance for considering effective therapy combinations [5] .
We used two patient-derived BRAFV600E mutant cell lines ( M397 and M229 ) with a prominent melanocytic to mesenchymal phenotypic evolution induced upon BRAF inhibition ( Fig 1A and S1 Fig ) [5] . We characterized this process by both a bulk transcriptome profiling ( S1 Table ) and a flow cytometry phenotyping using two protein markers ( MART-1 and NGFR ) that are established cell phenotype markers for this system [5 , 9 , 25] . The transcriptome was measured at Day 0 ( D0 ) , which served as an untreated control , and at a set of time points following BRAFi ( vemurafenib ) treatment ( S2 Fig ) . Following drug exposure , the relative location of the binning of cell populations expressing different levels of the two markers followed a counterclockwise transition trajectory around the flow cytometry plots ( S1 ( A ) Fig ) , moving from the melanocytic phenotype ( MART-1pos ) towards a transiently enriched slow-cycling neural crest ( MART1neg/NGFRhigh ) population around day15 to day20 ( D15–D20 ) , and eventually terminating at a mesenchymal ( MART1neg/NGFRneg ) phenotype at around day62 ( D62 ) with elevated expression of the mesenchymal marker N-cadherin ( Fig 1A ) . This drug-resistant population stably persisted with extended drug treatment beyond D62 ( S1 ( A ) Fig ) . A similar transition trajectory was also observed for M229 ( S1 ( B ) Fig ) . These drug-induced phenotypic transitions agree with previous literature [5 , 22] . To assess the overall stability and transcriptomic eigenpatterns associated with the cell population distributions at various time points across the drug-induced phenotypic evolution , we first applied surprisal analysis ( Eq 1 ) to the transcriptome time series . Surprisal analysis extends the principles of maximum entropy and was initially formulated to understand the dynamics of nonequilibrium systems [26] . Using the method of Lagrange multipliers , it seeks the maximum entropy of molecules and identifies the global steady state with minimal free energy , as well as a series of time-dependent constraints that prevent the nonequilibrium system from reaching the global steady state [16 , 23 , 26 , 27] . Surprisal analysis has been extended to characterize biological processes in living cells , where it assesses the maximum entropy of the biomolecules within the cell ensemble through using a simplified approximation of quantum state distributions of the molecular species [23] . Consequently , for a system with kinetic transcriptome data as input , it can extract the time-independent gene expression baseline ( the global steady state ) , as well as a series of gene expression modules ( constraints ) that evolve with time [16 , 23 , 26 , 28] . A full derivation and thermodynamic interpretation of Eq 1 is provided within the supplementary materials of previous reports [23 , 28] . In Eq 1 , Xi ( t ) is the measured level of transcript i at time t . This is considered to be the global steady-state level of transcript i ( ln ( Xi0 ( t ) ) =−λ0Gi0 ) , modified by the sum of the contributions arising from the constrained processes . The global steady state resolved by surprisal analysis is the cellular state with maximum entropy . If there were no constraints acting upon the cells , then Eq 1 predicts that the cells would be in the global steady state . However , there are non-zero constraints ( with amplitudes given by the λj values ) , which are biological processes that move the system away from the global steady state . Transcripts associated with a constraint are identified through Eq 1 as lowering the entropy of the system , presumably to maintain one or more biological functions . Data mining the set of transcripts associated with a given constraint can provide insight into those biological functions . Although we do not impose the condition that λ0 is time-independent , we neither expect nor find time-dependence ( the λ0 variation is <0 . 7% ) ( Fig 1B ) . To capture the time evolution of the drug-treated cells , each constrained process is represented by a time-dependent amplitude λj ( t ) and constraint-specific contributions from each transcript Gij . In principle , analysis of the transcriptomic data across the time series from D0 to D73 could resolve several constraints , but we resolve only three for M397 ( S3 and S4 Figs ) . This is illustrated in Fig 1C , where we represent the whole transcriptome data as a self-organized map ( SOM ) [29] . The map structure is determined using all data sets . Each tile represents a minicluster of genes with similar expression kinetics . Gene clusters with related expression kinetics are placed close together , while clusters exhibiting very different kinetics are placed far apart . The tile color encodes the average expression level of the genes in that minicluster at a given time point . For SOMs representing a specific constraint , that average gene expression level is also weighted by the participation of the genes in the constraint , as determined from Eq 1 . The gene expression profile for the global steady state remains unchanged throughout the transition , while the differentially expressed genes ( termed eigengenes elsewhere [30] ) specific to constraints λ1 , λ2 and λ3 vary with time . Summing the global steady state and the three constraints reproduces the map of the measured transcriptome , indicating that , within the noise level of the data , the three major constraints are sufficient to accurately recapitulate gene expression levels globally across the transition . ( Fig 1C and S5 Fig ) . The major biological processes involved in each constraint , at a given time point , can be inferred by enrichment on the gene lists ranked by the constraint-specific contributions from each gene Gij ( Fig 2A , S4 Fig and S2 Table ) , and by the time-dependent amplitude λj ( t ) of that constraint . The first constraint shows monotonically increased amplitude ( λ1 ) along the course of the transition ( Fig 1B ) , with up-regulated mesenchymal signatures , migration , invasiveness and metastasis features , as well as NFκB signaling ( G1 positive processes ) . It also reflects reduced glucose uptake and metabolism , MITF activity , and oxidative phosphorylation ( G1 negative processes ) ( Fig 2A ) . Constraint 2 contains similar transcriptional signatures , but its amplitude ( λ2 ) drops after 3 days of BRAFi exposure and slowly increases at later times ( Fig 1B ) . It points to an elevated MITF activity ( G2 negative process ) and reduced cellular proliferation ( G2 positive process ) at day-3 . This is consistent with previous observations that a brief BRAFi exposure can induce melanocytic differentiation and increased BRAFi sensitivity [31 , 32] . The third constraint mainly involves oxidative phosphorylation and the TCA cycle , and has a near zero amplitude except for day 3 ( Fig 1B ) , implying that initial BRAFi exposure leads to a sharply altered metabolic program . The three major constraints associated with M229 displayed similar dynamics and are functionally similar to those in M397 ( S4 Fig ) , confirming the robustness of the BRAFi induced melanocytic to mesenchymal transition . To get a comprehensive view of the enriched transcriptional program , we plotted the enrichment maps of the GSEA results with respect to relevant gene function categories and highlighted representative gene sets in these categories ( Fig 2B and S6 Fig ) . Overall , these transcriptional signatures are wholly consistent with previous reports [5 , 9–11 , 33] , testifying the validity of our cell line model for recapitulating the known biology of the transition and confirming the power of surprisal analysis for dissecting the underlying biology of the transition . The same biological system was further characterized at the single-cell level using flow cytometry analysis of the established cell-state markers: NGFR and MART-1 . The temporal transcriptomic signatures resolved by surprisal analysis result from the dynamics of the BRAFi-induced phenotypic evolution that can be characterized by MART-1 and NGFR marker proteins [5 , 9 , 25] . As shown in our previous report , these two marker proteins can yield the identical phenotypic classification to that of the whole transcriptome data [5] . Therefore , they can be used as robust phenotype markers during the course of the drug-induced transition ( Fig 1A and S1 Fig ) . To model the single cell data , we conceptualize cell population distributions as single cells moving on a configuration space delineated by the marker proteins . In this space , cell states correspond to stable or metastable attractors of a hypothetical potential landscape [34] . The dynamics of the protein markers for a single cell can be described by the Langevin type equation dz / dt = μ ( z ) + ζ , where z is the concentration vector of the protein markers ( z1 , … , zN ) , μ ( z ) is a drift vector in concentration space that describes all of the deterministic ( non-random ) dynamics and can be determined by the gradient of the potential landscape . The term ζ is the white noise term from random fluctuations in protein expression: 〈ζ ( t ) ζ ( t′ ) 〉 = 2D δ ( t–t′ ) where D is the diffusivity tensor measuring the amplitude of those fluctuations [18] . The potential landscape of a cellular system is context-specific . We hypothesized that drug treatment altered the original drug naïve landscape into a new landscape , which in turn yielded relaxation dynamics as each cell adjusts to this new drift field , potentially with motions towards new attractor states . Analyzing the dynamics arising from a multi-dimensional drift field is , in general , an intractable problem . However , the flow cytometry trajectory ( Fig 1A and S1 Fig ) upon BRAF inhibition suggested the simplification that cell populations may be considered to evolve along a linear chain of a limited number of cell states . Therefore , for computational convenience , we projected the protein concentration vectors of the two dimensional ( 2D ) flow cytometry data into a one-dimensional ( 1D ) representation where the cell populations were constrained to move along in this characteristic 1D trajectory ( Fig 3A ) . The distance along the trajectory x = x ( z ) serves as an effective reaction coordinate of the phenotypic evolution ( see Methods ) . The flow cytometry data do not track an individual cell stochastic trajectory , but rather give statistical snapshots of marker protein expression levels across single cells . Thus , it is natural to transform the single-cell Langevin equation into the Fokker-Planck equation for resolving the time-dependent probability distribution p ( x , t ) along the reaction coordinate [35]: ∂p ( x , t ) ∂t=−∂∂x[μ ( x ) p ( x , t ) ]+∂2∂x2[Dp ( x , t ) ] ( 2 ) Here , drift term μ ( x ) implies that motion along x is influenced by a potential landscape . D is a diffusivity that is assumed , for simplicity , to be a constant independent of x or drug treatment . Even in cases where the diffusivity depends on the reaction coordinate x , a Fokker-Planck ( FP ) equation with constant diffusivity can be obtained by a simple coordinate transformation as shown in Ref . [17] . Because the dynamics under consideration are 1D , the drift μ ( x ) can always be presented as the derivative of a scalar potential U ( x ) =−∫0xμ ( y ) dy . This , in turn , is exactly related to the steady state solution of Eq 2 through a Gibbs relation as limt→∞p ( x , t ) =p ( x ) ∞=Cexp ( −2U ( x ) /D ) where C is a normalization constant . Therefore , one can determine ( up to proportionality to D ) the potential U from measurement of the steady state distribution p∞ ( x ) as U ( x ) =− ( D2 ) lnp ( x ) ∞ . This FP approach has been successfully applied to understanding the population heterogeneity of model biological systems [17 , 18] . Here , we used a variation of this method to measure the diffusivity D = 0 . 35q2 / day ( q the unit length of the reaction coordinate ) from sorting-relaxation experiments in the drug-naïve condition ( S7 Fig , See Materials and methods for details ) . Given this D and flow cytometry measurements of the final steady state distribution p∞ ( x ) upon prolonged drug exposure , we inferred the potential U ( x ) , and equivalently the drift μ ( x ) consistent with this model . To test the validity of the FP model , we performed direct numerical simulation of the FP equation with the inferred μ ( x ) , the diffusivity D , and the measured initial distribution p ( x , 0 ) to calculate the cell population distribution p ( x , t ) for subsequent days , which , as shown in Fig 3B ( FP model ) , are in poor agreement with the experiments ( green lines in Fig 3B and S8 Fig ) . The disagreement indicated the existence of extra factors influencing phenotypic transitions which were not considered in Eq 2 . We hypothesized that the disagreement with experiments arose because the drug would influence not only the cell phenotypic evolution but also the cell autonomous proliferation and survival . In other words , the cells have drug susceptibilities–as reflected by the net effect of cell proliferation and cell killing–that vary along the reaction coordinate . These factors can also influence the phenotypic compositions , but are neglected in Eq 2 . Thus , we modified Eq 2 to include a self-sourcing term: ∂P ( x , t ) ∂t=−∂∂x[μ ( x ) P ( x , t ) ]+∂2∂x2[DP ( x , t ) ]+α ( x ) P ( x , t ) ( 3 ) Here the net growth rate α ( x ) ( the net effect of cell proliferation and cell killing under drug treatment ) was introduced to account for cell state-dependent drug susceptibility . As an additional ansatz , we considered α ( x ) as a double step function taking different values for the intermediate neural crest-like phenotype and late-stage mesenchymal phenotype relative to the early stage melanocytic phenotype . It is worthwhile to note that , with Eq 3 , we were no longer working with a probability distribution p ( x , t ) , but instead a non-normalized population P ( x , t ) . Both the differential drift and self-sourcing term act together to induce the cell number changes that are proportional to the population size of a specific cell state . For direct comparison between the model P ( x , t ) and experimentally accessible p ( x , t ) from flow cytometry data , we simply factored out the norm ( p ( x , t ) =P ( x , t ) /N ( t ) =P ( x , t ) /∫−∞∞dxP ( x , t ) ) . In this model , due to the addition of the self-sourcing term , the Gibbs relation between the drug-induced steady state p∞ ( x ) and the potential U ( x ) used in our analysis of the original FP equation no longer holds . To determine the parameters for this modified model , we therefore resorted to an unbiased numerical search for U ( x ) and α ( x ) that best fit the experimental data . The model prediction was obtained by numerically simulating Eq 3 with the same experimentally measured diffusivity D and the initial distribution p ( x , 0 ) as before , together with all possible U ( x ) and α ( x ) values in the unbiased search . We determine goodness of fit using an un-weighted sum-of-square difference between all the predicted and measured cell population distributions p ( x , t ) . In both cell lines , we were able to find one set of U ( x ) and α ( x ) for the modified FP-type kinetic model that produced the best prediction of population distributions over time . When compared to original FP model , the modified model predictions are in much better agreement to experiments ( red lines in Fig 3B and S8 Fig ) . The agreement appears to confirm the validity of the self-sourcing term in Eq 3 , but the value of that term can be put to an independent experimental test . We treated the state-dependent net growth rate α ( x ) as a concrete prediction of the model , and found it to be in good agreement with experimentally measured cell growth rates: cell populations containing a higher fraction of the mesenchymal phenotype ( day21-78 ) grow faster than those with a lower fraction ( day0-21 ) ( Fig 4 , See Materials and methods ) . The agreements between model-predicted growth rates and experiments ( Fig 4 ) further confirmed the validity of Eq 3 and show that differences in state-dependent growth rates are important in determining the drug-induced phenotypic evolution of the melanoma cells . In addition to predicting proliferation rates , Eq 3 also yielded relative values of the epigenetic potential along the reaction coordinate U ( x ) ( Fig 5A and S9 ( B ) and S10 ( A ) Figs ) , which yields an inference of the stability of different states along the coordinate . The scalar potential landscape was obtained by integration of μ ( x ) from Eq 3 over the reaction coordinate x . The shape of the landscape indicates that the intermediate neural crest-like states ( NGFRpos/MART-1neg ) are more stable than both the MART-1pos melanocytic state and the mesenchymal-like state ( NGFRneg/MART-1neg ) , and thus the intermediate states can be considered as an attractor . However , the net growth rate of those intermediate states is relatively low ( Figs 4 and 5A ) , and so the cells do not naturally populate just that state over the course of long-term drug treatment . As demonstrated in previous work , surprisal analysis of the bulk RNA-seq data can also define a free energy-like potential corresponding to the drug-induced phenotypic evolution [16 , 28] . This potential , for the entire transcriptome of a cell state at time t , is relative to the global steady state , and is given by F ( t ) =∑jλj ( t ) 〈Gj〉 , where 〈Gj〉=∑iXiGij ( See Materials and methods for details ) . It has a direct relationship to the entropy of the transcripts and thus evaluates , at a transcriptional level , the relative stability of a cell state ( see Ref . [23] for theoretic details ) . Here we adopted the same definition to calculate the potential landscape over drug-induced phenotypic evolution in melanoma cells . For M397 , this potential landscape calculated from surprisal analysis , similar to the landscape calculated by the modified Fokker-Planck-type ( Eq 3 ) model , indicates that the cells at D11 and D21 , with mostly neural-crest like phenotypes are more stable than cells at earlier times ( melanocytic phenotypes ) or D73 ( predominantly mesenchymal phenotype ) ( Fig 5B ) . For M229 , cells at D21 with mostly the neural-crest like phenotype are also more stable than the cells at D90 ( predominantly mesenchymal phenotype ) ( S10 Fig ) . Thus , the epigenetic potentials calculated from either surprisal analysis of bulk data or the Fokker-Planck kinetic model from single-cell data yield a consistent picture . Both analyses indicate that neural-crest like cells are more stable than the mesenchymal phenotype . This prediction was experimentally validated by sorting the mesenchymal ( NGFRneg/MART1neg ) subpopulation from the M397 D73 distribution ( S11 Fig ) . We carried out surprisal analysis of transcriptome data from both the segregated mesenchymal subpopulation and the unsorted day-73 population ( a mixture of mesenchymal phenotype and neural-crest phenotype ) . Free energy-like potentials were calculated and found to be consistent with the scalar potentials of both sorted and unsorted populations determined by the modified FP-type kinetic model . The pure mesenchymal phenotype displayed higher potentials than the unsorted cells ( Fig 5C ) . Hence , cell sorting and RNA-seq experiments confirmed the consistence between the two theoretic models , and indicated that the drug-resistant mesenchymal cells are epigenetically unstable relative to the neural crest phenotype .
Heterogeneous cancer cell populations can often exhibit a phenotypic equilibrium and evolution behaviors , meaning that a specific composition comprised of relative abundances of distinct cancer cell phenotypes can be a characteristic of the system , and in the meantime , this characteristic composition will evolve or recover following the application or release of molecular or physical perturbations designed to alter it [2 , 5 , 12–14] . This can , of course , confound the interpretation of responses to drug treatment , but it also provides a compelling biophysical puzzle . Here we investigated two statistical physics models to help build a predictive picture of such phenotypic equilibria . The models respectively utilize macroscopic and microscopic inputs , and we applied them towards understanding the population dynamics of phenotypically plastic patient-derived BRAF-mutant melanoma cancer cells following BRAFi treatment . During a few months period of drug treatment , the cells evolve from drug naïve , drug-sensitive melanocytic-dominated composition to a fully drug-resistant mesenchymal-dominated cell population . In an interesting parallel with state transitions in physical systems , the associated cell state transitions are fully reversible: upon drug removal , the mesenchymal cells revert back to a melanocytic state that is , for all intents and purposes , identical to the initial drug naïve state [5] . The first theoretical model , surprisal analysis , utilizes a bulk transcriptome kinetic series across the drug treatment course to provide a description of the global steady state ( the state of maximum entropy ) and to identify specific , time-dependent constraints that keep the system from reaching that steady state . The weights of the constrained processes can be utilized to generate a free energy-like potential of the cell-state space sampled during drug treatment [16 , 28] . It is worth noting that cells are open systems far from equilibrium . While a significant body of work has demonstrated the apparent parallel between equilibrium and nonequilibrium thermodynamics [36–38] , the potential landscape across the cell state evolution in our study is still a metaphor of the real free energy landscape in an equilibrium system . However , the maximum entropy methods can infer the most probable distribution of a probabilistic system regardless of whether or not it is in equilibrium [39] . Surprisal analysis further extends the principles of maximum entropy to understand particularly small systems that are not in thermodynamic equilibrium [23 , 26 , 40] . Therefore , in analogy to entropy in equilibrium thermodynamics , the entropy ( and free energy-like potential ) of the cellular transcriptome calculated from surprisal analysis can be used to evaluate the overall stability of a cell state [28 , 41] . The second theoretic approach consists of a modified Fokker-Planck-type kinetic model , which takes a kinetic series of single cell flow cytometry data as input . This model considers the Langevin dynamics of self-sourcing single cells moving within a configuration space . That motion is influenced by both ( random ) diffusion and drift along a potential gradient , thus permitting a potential surface of the traversed cell-state space to also be extracted . There are two primary considerations that allow results from these two theories to be directly compared . First , the flow cytometry data and the bulk transcriptome data sets capture the same essential biology . This is obviously not always true . However , for this particular case , the cell phenotype markers NGFR and MART-1 used in the single cell assays are known surrogates for drug-induced changes across the whole transcriptome [5] . It also implies that a more selective subset of the transcriptome might equally well recapitulate the underlying biology , which may be assessed by the contribution scores ( Gij values ) within each respective constraint . Second , the phenotypic evolution the melanoma cells proceeds stepwise from melanocytes → neural crest → mesenchymal phenotypes . This permits the cell response to BRAF inhibition to be considered as time-dependent motion along a linear reaction coordinate , and provides an equivalence between the Fokker-Planck reaction coordinate and the surprisal analysis time coordinate ( Fig 5A and 5B ) . We do not directly compare the y-axes of the two landscapes ( Fig 5A and 5B ) , but only the slopes of the curves . The FP scalar potential and the surprisal analysis free-energy like potential have very different origins . The free energy-like potential is derived by comparing transcriptional profiles at each time point with that of the time-independent global steady state . The FP potential is derived from the drift term of Eq 3 , and is , in fact , the only term in that equation that needs to be fitted , since both cell proliferation rate and diffusion along the FP reaction coordinate can be experimentally determined . However , both theories predict that the most stable cellular population is a largely neural crest phenotype . Surprisingly , that is not the cell population that is ultimately induced by the long-term drug exposure . That population is dominated by a mesenchymal phenotype with a minor neural crest component , and is arrived at through competing interactions . On the one hand , the neural crest phenotype serves as an attractor , but those cells only slowly proliferate . The higher potential mesenchymal cells are more proliferative and that is the dominating factor . This highlights a major difference between open biological systems and equilibrium thermodynamic systems [42] . The analyses presented here for the BRAF-mutant melanoma cells might suggest that identifying drug susceptibilities in each of the cancer cell phenotypes might lead to a more effective therapy . However , such highly plastic cancer cells might eventually switch into cell states that are resistant to even broad combination therapies . A more fruitful approach might be to target those biological mechanisms that underlie the plastic nature of the cells [5 , 43] .
Patient-derived melanoma M397 and M229 cell lines were generated from de-identified patient samples with written consent under UCLA IRB approved protocol # 11–003254 . Cells were cultured at 37°C with 5% CO2 in RPMI 1640 with L-glutamine ( Mediatech , Inc , Manassas , VA ) , 10% fetal bovine serum ( Omega Scientific Tarzana , CA ) , and 1% penicillin , streptomycin and fungizone ( Omega Scientific Tarzana , CA ) . Cells were maintained and tested for mycoplasma as previously described [44 , 45] . Cell lines were periodically authenticated to their early passages using GenePrint 10 System ( Promega , Madison , WI ) . Presence of mutations in the genes of interest was checked by OncoMap 3 or Iontrone , and was confirmed by PCR and Sanger sequencing as previously described [44 , 45] . Vemurafenib ( NC0621949 , Selleck Chemicals LLC ) was dissolved in DMSO at designated concentrations before applying to cell culture media . All cell lines were plated in 10cm dish at 60% confluency and treated with vemurafenib for the specified numbers of days at twice the 50% inhibition concentration ( IC50 ) of each cell line as reported before [5] . At different time points after drug treatment , cells were harvest for RNA-seq and flow cytometry . Cell number was also counted for determining the growth rate . Cell growth rate was fitted as the parameter α in the exponential growth curve equation N ( t ) = N0 · 2 ( α·t ) , where N0 is the cell number at the starting time point , and N ( t ) is the cell number at time t . Cell numbers counted at day 0 , 7 and 21 were used to fit for the proliferation rate at day 0–21 time period , and cell numbers at day 30 , 43 , 66 and 78 were used to fit for the one at day 21–78 time period . At different time points , cells were trypsinized from the dish , spun down and washed with PBS . Cell suspensions were stained for flow cytometry with PE-conjugated NGFR antibody from Biolegend ( San Diego , CA ) . All cells were fixed with Fix-Perm buffer from BD Bioscience ( San Jose , CA ) . Cells were then stained for intracellular Melan-A using FITC conjugated antibody from Santa Cruz ( Dallas , TX ) . Isotypes for mouse IgG1k and mouse IgG1 respectively were used to enable correct gating and to confirm antibody specificity . Blue live-dead staining from Life technologies ( Waltham , MA ) was used to gate live cell events . 10000 alive events were collected for each sample . Flow cytometry analysis was conducted using LSR-II from BD Biosciences ( San Jose , CA ) , and the data were analyzed using FlowJo software ( Tree Star , Inc . , San Carlos , California , USA ) . The standard immunofluorescent protocol was implemented using cells grown on the gelatin-coated glass surface . Briefly , 10 , 000 cells/well were seeded in 96-well glass bottom plates ( Greiner Sensoplate Plus , Cat# 655892 ) coated with 0 . 1% gelatin solution , and grown in culture media to ~70% confluency . Cells were washed twice in PBS and fixed in 4% paraformaldehyde ( PFA ) solution for 10 min . After washing twice in wash buffer ( 0 . 1% BSA in PBS ) , cells were blocked and permeabilized in blocking buffer ( 10% normal donkey serum , 0 . 3% Triton X-100 ) for 45 minutes . After removing the blocking buffer , cells were incubated in primary antibody for 4 hours at room temperature . Mouse monoclonal anti-NGFR antibody ( BioLegend Cat# 345106 RRID:AB_2152647 ) or sheep polyclonal anti-N-Cadherin ( R&D Systems Cat# AF6426 RRID:AB_10718850 ) was diluted to 0 . 25 or 10 μg/mL , respectively , in antibody diluent ( 1% BSA , 1% normal donkey serum , 0 . 3% Triton X-100 ) . After washing twice in wash buffer ( 0 . 1% BSA in PBS ) , cells were incubated in secondary antibody for 1 hour at room temperature . Donkey anti-Mouse IgG , Alexa Fluor 647 ( Thermo Fisher Scientific Cat# A-31571 RRID:AB_162542 ) or donkey anti-Sheep IgG Alexa Fluor 594 ( Thermo Fisher Scientific Cat# A-11016 RRID:AB_2534083 ) was diluted to 4 μg/mL in antibody diluent . After washing twice in wash buffer , cells were counter stained for 5 min with 4' , 6-Diamidino-2-Phenylindole ( DAPI ) diluted to 1 μg/mL in PBS . After washing twice in PBS , the wells were filled with 78% glycerol . Fluorescent images were acquired with a Nikon C2plus confocal microscope ( Ti ) using Plan Apo λ 20× objective ( Nikon Inc . , Melville , NY ) . The microscope was controlled by NIS elements AR software ( 4 . 51 . 00 ) with the following settings: 30 μm pin hole , 12-bit acquisition , 0 . 62 μm pixel size , 60 gain , and laser power of 5% ( 405 nm ) , 0 . 3% ( 561 nm ) , or 0 . 6% ( 640 nm ) . Images were background and contrast adjusted using their respective control wells with no primary antibody staining . Cells treated under specified conditions and time periods were trypsinized to harvest for cell pellets . RNA extraction was performed at cell pellets using AllPrep DNA/RNA Mini kit from Qiagen . Bioanalyzer confirmed correct integrity , the library was constructed and Illumina 50 bps single-end RNA-seq data was collected for the samples described . RNA sequencing was performed using 50 bps single end sequencing on the Illumina HiSeq 2500 platform . Libraries were prepared using the IlluminaTruSeq RNA sample preparation kit per the manufacturer’s instructions . Reads were mapped and aligned to the Homo sapiens NCBI build 37 . 2 reference genome using TopHat2 v2 . 0 . 9 [46] . Expression values in fragments per kilobase of exon per million fragments mapped ( FPKM ) were generated using Cufflinks v2 . 2 . 1 program and Cuffnorm to quantify and normalize aligned reads using the geometric library size normalization method [47] . Heatmap and clustering analysis of transcriptomic datasets was performed via MATLAB . Genes are pre-filtered by RPKM value with criteria of average value greater than 0 . 5 and coefficient of variance greater than 0 . 15 . Filtered gene expression values were standardized across each row ( normalized for each individual gene ) and represented by a redblue colormap . Hierarchical clustering was performed with average linkage and Euclidean distance metric . Whole transcriptomic dataset and fractions of contributions from each constraints are visualized using self-organized mosaic maps with respect to its control via Gene Expression Dynamics Inspector ( GEDI ) [29] . Gene Set Enrichment Analysis ( GSEA ) [48] was performed using GSEA v2 . 2 . 3 software with 1000 permutations and weighted enrichment statistics . GSEA enriched gene sets were visualized as interaction networks with Cytoscape [49] and Enrichment Map [50] . Surprisal analysis was applied as described previously [23 , 28] . The measured expression level of mRNA i at time t , ln Xi ( t ) , was expressed as a sum of a steady state term lnXi0 ( t ) and several constraints λj ( t ) Gij representing deviations from the steady state . Each deviation term was a product of a time-dependent weight of the constraint λj ( t ) , and the time-independent contribution of the transcript to that constraint Gij . To implement surprisal analysis , we computed the singular value decomposition ( SVD ) of the matrix ln Xi ( t ) . As well described previously [23] , the SVD factored this matrix in a way that determined the two sets of parameters that are needed in surprisal analysis: the Lagrange multipliers ( λj ) for all constraints at a given time point , and for all times and the Gij ( time-independent ) transcription patterns for all transcripts i at each constraint j . The free energy-like potential calculation based on the surprisal analysis result was implemented as in Ref . [33] . Briefly , The steady-state expression level of transcript i at time t can be linked to its actual expression level by as Xi0 ( t ) =Xi ( t ) exp ( −∑jλj ( t ) Gij ) . Therefore , as shown in Ref . [33] , surprisal analysis defines the free energy-like potential of a transcript i relative to the global steady state at time t as fi ( t ) =∑jλj ( t ) Gij . Taking all the transcripts into account , the free energy-like potential of the entire transcriptome of a cell state at time t relative to the global steady state is given by F ( t ) =∑iXifi ( t ) =∑jλj ( t ) 〈Gj〉 , where 〈 Gj 〉=∑iXiGij[16] . Natural log transformed transcriptomic dataset and fractions of contributions from each constraints ( λj ( t ) Gij ) calculated from surprisal analysis are visualized using self-organized maps ( SOM ) . Self-organized map visualization of high-dimensional dataset in a form appropriate for human pattern recognition without discarding the global , higher-order information . Here , they present individual samples as a single 2-dimensional heatmap and , at the same time , display high-resolution patterns . Thousands of input genes are assigned to 625 rectangular “tiles” ( SOM nodes ) , each of which represents a mini-cluster of genes , arranged so as to form a pattern within a 2-dimensional mosaic map on the SOM grid . Tile represent most similar clusters will be placed adjacent to each other in the mosaic . Gene Expression Dynamics Inspector ( GEDI ) package is utilized to implement the SOM visualization [29] . The dynamics of the protein markers for a single cell can be described by the Langevin type equation dz / dt = μ ( z ) + ζ , where z is the concentration vector of the protein markers ( z1 , … , zN ) , μ ( z ) is a drift vector in concentration space that describes all of the deterministic ( non-random ) dynamics and can be determined by the gradient of the potential landscape . The term ζ is the white noise term from random fluctuations in protein expression: 〈ζ ( t ) ζ ( t′ ) 〉 = 2D δ ( t–t′ ) where D is the diffusivity tensor measuring the amplitude of those fluctuations [18] . In the case of melanocytic to mesenchymal transition , for computational convenience , we projected the protein concentration vectors of the flow cytometry data into a one-dimensional ( 1D ) representation where the cell populations were constrained to move along in this characteristic trajectory . This converted each snap-shot of cell population distribution from the 2D flow cytometry plot onto a one-dimensional distribution along the linear trajectory . More specifically , we reduced the dimensionality of the flow cytometry data by calculating the principle curve of the full set of measurements using the R package princurve . The data points were projected onto the curve , and the distances of these projected points along the curve were used as the one-dimensional data for the two Fokker-Planck models . These data points were converted into probability density functions ( PDF ) using kernel density estimation . Consider the fact that flow cytometry data do not track an individual cell stochastic trajectory but rather give statistical snapshots of marker expression levels across many single cells . Thus , it is natural to transform the single-cell Langevin equation into the Fokker-Planck equation for resolving the probability distribution of the protein markers . The 1D coordinate ( Fig 3A ) is defined as a reaction coordinate x ( z ) such that the Fokker-Planck ( FP ) equation for the probability distribution p ( x , t ) has the following form: ∂p ( x , t ) ∂t=−∂∂x[μ ( x ) p ( x , t ) ]+∂2∂x2[Dp ( x , t ) ] ( M1 ) Here , x is the 1D flow cytometry ( FC ) coordinate , D is a diffusion constant of the cells along x , and drift term μ ( x ) implies that motion along x is influenced by a potential landscape . In this model , the equilibrium distribution p∞ ( x ) = limt→∞p ( x , t ) and the potential U ( x ) = −∫μ ( x ) dx are connected through the Gibbs relation U ( x ) =− ( D2 ) lnp∞ ( x ) ( M2 ) For the unmodified Fokker-Planck equation , this Gibbs relation was applied to the long-term drug treated cell population distribution data ( day78 for M397 and day60 for M229 ) to infer a potential , whose gradient acted as the drift term driving the dynamic changes of the population distribution . This inferred potential and respective drift term , when coupled with diffusion constant D and the initial ( day0 ) population distribution , generated the prediction results in Fig 3B . With regards to calculating diffusion constant D from cell sorting and relaxation experiments , the diffusion coefficient D was assumed to be a constant value independent of trajectory position x and drug treatment condition for simplicity . Based on this assumption , when calculating the diffusion constant , we used time-series flow cytometry data of cell sorting and relaxation experiments . In these experiments , we sorted out the untreated cells into NGFRpos and NGFRneg subpopulations . Both sorted subpopulations were cultured without drug treatment . At different days after sorting and culturing , the cells were harvested to quantify its abundance of NGFR and MART-1 using flow cytometry as shown S6 Fig . Consider the fact that variations of proliferation rates are small in the untreated cells , the Fokker-Planck model ( Eq M1 ) was considered valid and this data was used as input to fit the diffusion constant D . Varying D as a free parameter , the ( drug-naïve ) potential and hence drift were calculated with Eq M2 , using the original , untreated distribution as p∞ ( x ) . The Fokker-Planck equation with these parameters was simulated , with the initial condition p ( x , 0 ) set by the sorted population distribution . The simulated data were compared to the measured time-series distributions with an unweighted sum-of-squares measure . This measure was then minimized as a function of D , and yielded the best-fit diffusion constant D to be 0 . 35 q2/day where q represents the unit length on the flow cytometry coordinate . For the modified Fokker-Planck-type kinetic model described in Eq ( 3 ) where the same reaction coordinate ( x ) as the unmodified equation is applied , the state-dependent proliferation rate α ( x ) was modeled as piecewise-constant with different values for the melanocytic , neural crest , and mesenchymal cell types . The cutoff locations in terms of the reaction coordinate x were chosen as the two local minima in an observed PDF with the coexistence of all three subpopulations . One can show that the time evolution of the PDF does not depend on an overall constant shift α ( x ) +c , so we set the proliferation rate of the starting melanocytic state to 0 for convenience , as the melanocytic cells were observed to be cytostatic without significant proliferation or cell death upon drug treatment . This then left the proliferation rates of the neural crest ( α1 ) and mesenchymal cells ( α2 ) as two free parameters . Because the Gibbs relationship between the long-time density p∞ ( x ) and the potential U ( x ) no longer held with this nontrivial proliferation rate , we resorted to fitting a cubic spline interpolation for the drift μ ( x ) = −∂U ( x ) / ∂x . Twenty spline points were used , with x values uniformly spaced along the curve and μ values as free parameters . Starting with an estimate of α1 ( x ) = α2 ( x ) = 0 and μ ~ x , we calculated the prediction of this model using FiPy to numerically simulate the forward evolution with initial condition p ( x , 0 ) set by the experimentally measured distribution on day-0 . To compare with the experimental data , we used the L2 norm on the difference between the predicted and experimental probability densities L=∑i∫−∞∞ ( ppred ( x , ti ) −pexp ( x , ti ) ) 2dx as the goodness-of-fit metric . Gradient descent was performed on the proliferation and drift parameters to determine the best-fit values that minimize L . The calculated potential landscape results are robust to small variations in parameters for calculating the principal curve . ( S12 Fig ) . | Cancer cells exhibit varied degrees of phenotypic heterogeneity . These phenotypes , each of them with unique molecular and functional profiles , display dynamic interconversion in response to drug perturbations , and can evolve to form new drug-tolerant phenotypes . Such phenotypic plasticity , in turn , renders tumor cells extremely difficult to treat . To get a quantitative biophysical understanding of the origins of the phenotypic equilibrium and evolution associated with drug tolerance development in highly plastic patient-derived melanoma cells , we employed joint experimental and computational approaches , using either bulk or single cell measurements as input , to interrogate the epigenetic landscape of the phenotypic evolution . We found that the observed phenotypic equilibria were established via competition between state-dependent net proliferation rates and landscape potential . The results reveal how the tumor cells maintain a phenotypic heterogeneity that facilitates appropriate responses to external cues . They implicate that , in certain phenotypically plastic tumor cells , drug targeting the driver oncogenes may not have sustained efficacy unless the phenotypic plasticity of the tumor is co-targeted . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"flow",
"cytometry",
"medicine",
"and",
"health",
"sciences",
"population",
"dynamics",
"biological",
"cultures",
"pharmaceutics",
"genome",
"analysis",
"cell",
"cultures",
"population",
"biology",
"melanoma",
"cells",
"thermodynamics",
"research",
"and",
"analysis",
"m... | 2019 | Phenotypic heterogeneity and evolution of melanoma cells associated with targeted therapy resistance |
Mycobacterium ulcerans disease , or Buruli ulcer ( BU ) , is an indolent , necrotizing infection of skin , subcutaneous tissue and , occasionally , bones . It is the third most common human mycobacteriosis worldwide , after tuberculosis and leprosy . There is evidence that M . ulcerans is an environmental pathogen transmitted to humans from aquatic niches; however , well-characterized pure cultures of M . ulcerans from the environment have never been reported . Here we present details of the isolation and characterization of an M . ulcerans strain ( 00-1441 ) obtained from an aquatic Hemiptera ( common name Water Strider , Gerris sp . ) from Benin . One culture from a homogenate of a Gerris sp . in BACTEC became positive for IS2404 , an insertion sequence with more than 200 copies in M . ulcerans . A pure culture of M . ulcerans 00-1441 was obtained on Löwenstein-Jensen medium after inoculation of BACTEC culture in mouse footpads followed by two other mouse footpad passages . The phenotypic characteristics of 00-1441 were identical to those of African M . ulcerans , including production of mycolactone A/B . The nucleotide sequence of the 5′ end of 16S rRNA gene of 00-1441 was 100% identical to M . ulcerans and M . marinum , and the sequence of the 3′ end was identical to that of the African type except for a single nucleotide substitution at position 1317 . This mutation in M . ulcerans was recently discovered in BU patients living in the same geographic area . Various genotyping methods confirmed that strain 00-1441 has a profile identical to that of the predominant African type . Strain 00-1441 produced severe progressive infection and disease in mouse footpads with involvement of bone . Strain 00-1441 represents the first genetically and phenotypically identified strain of M . ulcerans isolated in pure culture from the environment . This isolation supports the concept that the agent of BU is a human pathogen with an environmental niche .
Buruli ulcer ( BU ) , the third most common mycobacteriosis in humans after tuberculosis and leprosy is an indolent , necrotizing disease of skin , subcutaneous tissue and , occasionally , bones [1] . BU has emerged in recent times as an increasingly important cause of morbidity around the world , and has been reported in 30 countries , mostly in tropical areas [2] . This disease is caused by Mycobacterium ulcerans which is peculiar among pathogenic mycobacteria because it produces a potent necrotizing exotoxin , mycolactone , which is a major virulence factor [3] . Although incompletely understood , the epidemiology of BU strongly associates the disease with wetlands and especially slow-flowing or stagnant water [4]–[6] . Indeed , there is indirect evidence that M . ulcerans is an environmental pathogen transmitted to humans from its aquatic niches; however , modes of transmission are unclear [7] . The initial hypothesis that predatory aquatic insects , including Naucoridae and Belostomatidae , were involved in transmission [8] was later reinforced by reports that the salivary glands of Naucoris were colonized by M . ulcerans when fed on infected grubs , and that bites of infected Naucoris transmitted the pathogen to mice [9] . The observation that non-infected humans exposed to aquatic environments in BU endemic areas have higher titers of antibodies to salivary proteins of Naucoridae and Belostomatidae than BU patients in the same areas [10] shows that these water bugs bite humans in natural settings . However , Naucoridae and Belostomatidae are carnivorous insects that normally feed on other aquatic insects , snails , and small fish and only bite humans incidentally [11] . Thus , the significance of biting by M . ulcerans-colonized aquatic insects in the transmission of BU to humans is unknown , and other forms of transmission , including skin trauma , have been considered [12] . Since the discovery of IS2404 [13] , an insertion sequence with more than 200 copies in M . ulcerans [14] , multiple studies have detected IS2404 in environmental aquatic samples , indicating that M . ulcerans is probably present in such samples , and supporting the concept that the etiologic agent of BU is an environmental pathogen . IS2404 was found in samples of water and detritus from swamps in Australia [15] , [16] , [17] , and in aquatic plants [18] , insects ( Belostomatidae , Naucoridae , Hydrophilidae ) , crustaceans and mollusks ( Bulinus sp . and Planorbis sp . ) , and fish ( including Tilapia sp . ) in western tropical Africa [8] , [9] , [18] , [19] , [20] . More recently IS2404 was detected in mosquitos in Australia [21] . However , the recent discovery of IS2404 in aquatic mycobacteria other than M . ulcerans requires re-evaluation of the use of IS2404 PCR for the detection of M . ulcerans DNA in the environment [22] , [23] and emphasizes the importance of the isolation of M . ulcerans from environmental sources . Numerous extensive studies have failed to isolate M . ulcerans in pure culture from the environment , even in highly endemic areas of BU , e . g . in Uganda [24] , the Democratic Republic of Congo [4] , [5] , [25] and West Africa [19] . Two cultures from salivary glands of wild aquatic insects ( Naucoridae ) collected in BU endemic areas of Côte d'Ivoire were positive for IS2404 and were considered to be related to M . ulcerans; however , no phenotypic characteristics of these isolates were reported other than their virulence for mice [9] . In 2004 , Marsollier et al . obtained IS2404 PCR positive cultures from two samples of aquatic plants collected in a BU endemic area of Côte d'Ivoire [18] . One IS2404 positive BACTEC culture inoculated into mice revealed infection compatible with M . ulcerans . The culture was , however , contaminated by Mycobacterium szulgai and M . ulcerans could not be obtained in pure culture even after passages through mice . As briefly reported previously [26] , a pure culture of M . ulcerans ( isolate 00-1441 ) was obtained from an aquatic insect from Benin . In that report no description was given of the methods employed for the isolation of M . ulcerans 00-1441 and of the phenotypic and genetic characteristics of the isolate . Here we present the detailed results of the isolation and characterization of strain 00-1441 , establishing that this mycobacterium is an African type of M . ulcerans with high virulence for mice . Strain 00-1441 represents the first well characterized M . ulcerans strain isolated in pure culture from an environmental source .
Results of ZN staining , culture and PCR studies for the 5 aquatic specimens are shown in Table 1 . Table 2 shows the results of the mouse footpad inoculation with the BACTEC suspensions ( 98–447 , 97–1455 and 98–443 ) that were positive by IS2404 PCR after inoculation with the aquatic specimens . Specimen 98–447: Histopathologic analysis of one mouse sacrificed after 9 months revealed a few well formed granulomas with minimal necrosis around blood vessels , nerves and in muscle . There were large numbers of beaded AFB in the granulomas . Specimen 97–1455: Of the three mice inoculated with this BACTEC culture ( P1 in Table 2 ) , two were sacrificed 9 months after inoculation . The histopathologic analysis of the footpad of one mouse showed marked necrosis with a mild granulomatous response , inflammation of periosteum and many large clumps of AFB in necrotic areas . The footpad homogenate of the third mouse was positive for AFB and the culture on LJ was positive for mycobacteria ( isolate 99–2832 ) but was lost due to contamination by nonacid-fast bacteria; however , PCR performed on the contaminated culture was positive for IS2404 . This mouse footpad was inoculated in vitro and passaged twice into two groups of three mouse footpads . The second ( P2 in Table 2 ) and third passages ( P3 in Table 2 ) were negative for AFB by ZN staining and by culture . Specimen 98–443: This homogenate of a Gerris sp . aquatic insect ultimately produced the M . ulcerans isolate 00-1441 after culture in BACTEC ( positive for IS2404 ) inoculated in mouse footpads ( P1 ) and followed by two other mouse footpad passages ( P2 and P3 ) . Following the first mouse inoculation ( P1 ) , one animal died after 1 month and the other two were sacrificed 9 months after inoculation . Histopathologic evaluation of the footpad of one of these mice showed granulomatous changes with minimal necrosis around blood vessels and nerves . There were large numbers of scattered , short , beaded AFB in the granuloma . ZN stain and culture were negative for the footpad of the third mouse inoculated in vitro . The suspension obtained from the third mouse was used to reinoculate 3 other mice ( P2 ) . One P2 mouse died after 6 months and the other two were sacrificed 12 months later . The histopathologic study of one footpad showed minimal nonspecific inflammation . The other mouse footpad was negative for AFB and by culture and was used for a third passage ( P3 ) into 3 mice . Two of the P3 mice were sacrificed after 6 months . One footpad used for histopathologic study showed nonspecific inflammation . The other footpad was ZN-negative but gave a positive culture on LJ ( 5 colonies ) after 2 months incubation at 32°C . The isolate ( 00-1441 ) was further analyzed and identified as M . ulcerans ( see below ) . The remaining P3 mouse , sacrificed after one year , did not reveal any histopathologic changes . As previously described for virulent M . ulcerans strains [36] , isolate 00-1441 showed cytotoxic activity against BMDM infected at an MOI 1∶1 as deduced at day 4 post-inoculation from the occurrence of mycolactone-associated cytopathic signs [42] namely , cell rounding , shrinkage and detachment of the macrophages ( Fig . 4 ) .
The prevailing concept that BU is associated with wetlands , especially slow-flowing or stagnant water , implies that M . ulcerans is an environmental pathogen transmitted to humans from particular aquatic niches . Historically , the presence of M . ulcerans in aquatic samples , including water , mud , aquatic plants , aquatic insects , aquatic mollusks , crustacea and small fish , has been inferred from the detection by PCR of the insertion sequence IS2404 , highly represented in the genome of M . ulcerans [14] . All previous attempts to isolate fully characterized M . ulcerans from environmental samples , however , have failed , and recent evidence [22] indicates that IS2404 positivity alone is inadequate to establish the presence of M . ulcerans in environmental samples . M . ulcerans 00-1441 , isolated from a Hemiptera ( Water Strider , Gerris sp . ) collected from a swamp in a BU endemic region ( Zagnanado , Benin ) , represents the first fully characterized culture of the agent of BU from an environmental source . Isolate 00-1441 was identified as M . ulcerans by the following criteria: Additionally , 00-1441 had been previously found to have a mycolate profile pattern similar to that of M . ulcerans , with three types of mycolates , α- , methoxy- , and ketomycolates [26] . Moreover , 00-1441 and the predominant African type share identical profiles for IS2404-Mtb2 PCR [44] , and microsatellite VNTR analysis [45] . Based on nucleotide substitutions at the 3′ end 16S rRNA gene [33] , isolate 00-1441 is an M . ulcerans type 1 strain ( an African type ) . The mutation found at position 1317 ( a T instead of a C ) has not been found previously . Indeed the 3′ end of the 16S rRNA gene of all M . ulcerans strains analyzed in 1996 [33] and of all other mycobacterial species has a C at position 1317 [Blast search on the nucleotide collection ( nr/nt ) database ( NCBI ) using the nucleotide sequence of the 3′end 16S rRNA gene ( nt 1244-1461 ) of M . ulcerans type 1] . In a recent study on 75 M . ulcerans isolates from 17 different countries including 10 African countries ( Angola , Benin , Cameroon , Congo-Brazzaville , Côte d'Ivoire , Democratic Republic of Congo , Ghana , Nigeria , Togo and Uganda ) , a few isolates from patients originating from the Zou and Ouémé valleys in Benin presented a T instead of a C at position 1317 ( Portaels et al . , in preparation ) . Interestingly , strain 00-1441 was isolated from the region ( Zou Department ) where some of these patients lived . The aquatic specimens analyzed in the present study likely contained very few mycobacteria since direct smear examination after decontamination was negative for all specimens and primary cultures positive for mycobacteria other than M . ulcerans ( Table 1 ) produced only 1 to 3 colonies . Moreover , despite the very high sensitivity of the IS2404 PCR [14] , detection of IS2404 in the decontaminated specimens was negative indicating that less than 10 mycobacterial cells were present in each suspension [28] . Culture in BACTEC allowed multiplication of the rare mycobacteria present in the inocula since three of the five BACTEC positive cultures were positive by IS2404 PCR . Our previous attempts to detect M . ulcerans in more than 1000 environmental specimens by culture have revealed numerous environmental mycobacteria belonging to species frequently cultivated from the environment [5] . However , other than the results of Marsollier et al . [9] , [18] and the present study , all attempts to culture M . ulcerans from the environment have failed . As discussed elsewhere [19] , there are several possible explanations for the difficulty in culturing M . ulcerans from environmental specimens , namely: ( i ) These specimens are heavily contaminated with other microorganisms , [5] , [13] , [18] , [24] . This is primarily because the generation time of M . ulcerans is longer than that of most other slow-growing mycobacteria that are abundant in the environment [18] , [19] . In the present study , successive passages in mice of BACTEC cultures may have eliminated mouse avirulent environmental mycobacteria [30] co-existing in the specimen , allowing multiplication of M . ulcerans . ( ii ) All decontamination methods currently available for the isolation of M . ulcerans from contaminated environmental specimens have a detrimental impact on the viability of this pathogen [27] . ( iii ) Since M . ulcerans is sensitive to elevated temperatures [46]–[47] , temperature during transportation of environmental specimens to the laboratory is critical , particularly in tropical areas where ambient temperatures often exceed 32°C . ( iv ) As is the case in the present work , environmental specimens used in attempts to isolate M . ulcerans may contain very few bacilli . ( v ) Additionally , in the environment M . ulcerans may be living in a viable but nonculturable ( VBNC ) state . This state may represent a survival adaptation to overcome adverse conditions , but the organism retains secluded cultural viability and virulence capability [48] , [49] . Most pathogenic bacteria of humans are known to enter the VBNC state , including those in aquatic environments [48] , [50] , [51] . The recuperation of culturability in bacteriological media by mycobacteria in the VBNC state may require a suitable resuscitation medium [52] and BACTEC may serve to resuscitate the VBNC M . ulcerans . Additional experiments are required to test for a VBNC state in environmental M . ulcerans . The strain analyzed in the present study was isolated from a Hemiptera ( Gerris sp . ) . Gerris sp . belongs to the worldwide family of the Gerridae . They are elongate insects with very long mid and hind legs ( Fig . 8 ) . The latter allow them to move rapidly on water surfaces to catch their preys . They live on the surface of quiet waters and are unable to walk on the ground , but can fly from one pond or river to another [53] . Several publications have suggested that Hemiptera ( Naucoridae , Belostomatidae ) may play a role in the transmission of BU to humans [1] , [8] , [9] . The successful cultivation of M . ulcerans from another family of aquatic Hemiptera ( Gerridae ) extends the range of hypothetical hemipteran transmitters . Like other aquatic Hemiptera , Gerridae are aggressive predators of other aquatic organisms such as insects and small fish . However , there are no reports of Gerridae biting humans ( Dethier M , personal communication ) and these insects may be only passive , incidental and transient reservoirs of M . ulcerans without an obvious role in the transmission of BU to humans or other mammals . MS analysis of ASLs confirmed that isolate 00-1441 produced mycolactone A/B . The virulence of M . ulcerans is largely due to the presence of the toxic macrolide , mycolactone [3] . It is now recognized that there is a family of mycolactones produced by M . ulcerans and other related mycobacterial species . Each mycolactone has a distinct structure and mass . However , all isolates of M . ulcerans from Africa produce mycolactone A/B [34] . The demonstration of mycolactone A/B in Gerridae isolate 00-1441 presented here provides additional evidence that this strain is similar to virulent strains isolated from patients throughout West Africa . Like other mycolactone A/B producing M . ulcerans strains [36] , strain 00-1441 proliferates extensively in mouse footpads and produces intense footpad swelling . Moreover , in previous mouse footpad inoculation studies on 11 isolates of M . ulcerans from patients in Benin , 5 of which were from bones of patients with M . ulcerans osteomyelitis [54] , no changes were noted in the bones of the mice feet ( Portaels F and Meyers WM , unpublished observations ) . However , feet of NMRI and BALB/c mice inoculated with 00-1441 showed striking destruction of bone . These data regarding mouse infection suggest that strain 00-1441 is highly virulent for mice . Additional experiments in mice and ex vivo are required to compare the virulence of strains sharing the same “T” for “C” 16S rRNA gene polymorphism at position 1317 and identically treated i . e . , after several passages in mice . Such experiments are underway and will be presented in another publication . In the present study , the main steps followed to cultivate M . ulcerans in pure culture from an aquatic insect are summarized in Fig . 9 . Other methods may also be applied such as cultures from salivary glands of wild Naucoridae [9] or aquatic plants [18] , or other culture procedures such as the Mycobacteria growth Indicator Tube ( MGIT ) system [55] , or other decontamination methods [27] . The growth of M . ulcerans in liquid media can also be confirmed by applying VNTR analysis to the IS2404 positive liquid cultures to differentiate M . ulcerans and other IS2404 positive mycobacteria [23] . This was not done because the technique had not yet been developed when the present study was undertaken . In conclusion , for the first time a genetically and phenotypically identified M . ulcerans has been isolated in pure culture from an environmental source , reinforcing the concept that the agent of BU is a human pathogen with environmental aquatic niches . | Mycobacterium ulcerans infection , or Buruli ulcer , is the third most common mycobacteriosis of humans worldwide , after tuberculosis and leprosy . Buruli ulcer is a neglected , devastating , necrotizing disease , sometimes producing massive , disfiguring ulcers , with huge social impact . Buruli ulcer occurs predominantly in impoverished , humid , tropical , rural areas of Africa , where the incidence has been increasing , surpassing tuberculosis and leprosy in some regions . Besides being a disease of the poor , Buruli ulcer is a poverty-promoting chronic infectious disease . There is strong evidence that M . ulcerans is not transmitted person to person but is an environmental pathogen transmitted to humans from its aquatic niches . However , until now M . ulcerans has not been isolated in pure culture from environmental sources . This manuscript describes the first isolation , to our knowledge , of M . ulcerans in pure culture from an environmental source . This strain , which is highly virulent for mice , has microbiological features typical of African strains of M . ulcerans and was isolated from an aquatic insect from a Buruli ulcer–endemic area in Benin , West Africa . Our findings support the concept that M . ulcerans is a pathogen of humans with an aquatic environmental niche and will have positive consequences for the control of this neglected and socially important tropical disease . | [
"Abstract",
"Introduction",
"Results",
"Discussion"
] | [
"infectious",
"diseases/bacterial",
"infections",
"infectious",
"diseases/neglected",
"tropical",
"diseases"
] | 2008 | First Cultivation and Characterization of Mycobacterium ulcerans from the Environment |
In October 2017 , a blood sample from a resident of Kween District , Eastern Uganda , tested positive for Marburg virus . Within 24 hour of confirmation , a rapid outbreak response was initiated . Here , we present results of epidemiological and laboratory investigations . A district task force was activated consisting of specialised teams to conduct case finding , case management and isolation , contact listing and follow up , sample collection and testing , and community engagement . An ecological investigation was also carried out to identify the potential source of infection . Virus isolation and Next Generation sequencing were performed to identify the strain of Marburg virus . Seventy individuals ( 34 MVD suspected cases and 36 close contacts of confirmed cases ) were epidemiologically investigated , with blood samples tested for MVD . Only four cases met the MVD case definition; one was categorized as a probable case while the other three were confirmed cases . A total of 299 contacts were identified; during follow- up , two were confirmed as MVD . Of the four confirmed and probable MVD cases , three died , yielding a case fatality rate of 75% . All four cases belonged to a single family and 50% ( 2/4 ) of the MVD cases were female . All confirmed cases had clinical symptoms of fever , vomiting , abdominal pain and bleeding from body orifices . Viral sequences indicated that the Marburg virus strain responsible for this outbreak was closely related to virus strains previously shown to be circulating in Uganda . This outbreak of MVD occurred as a family cluster with no additional transmission outside of the four related cases . Rapid case detection , prompt laboratory testing at the Uganda National VHF Reference Laboratory and presence of pre-trained , well-prepared national and district rapid response teams facilitated the containment and control of this outbreak within one month , preventing nationwide and global transmission of the disease .
Marburg Virus Disease ( MVD ) is a severe infectious disease caused by Marburg virus , a member of the Filoviridae family , which also includes Ebola viruses . Marburg virus was first recognized in 1967 when outbreaks of haemorrhagic fever occurred in laboratories located in Marburg and Frankfurt , Germany and Belgrade , Yugoslavia ( present-day Serbia ) [2] . MVD has a high case fatality rate ranging from 32% to 88% [1] . It is transmitted to humans after a spill-over event from a wildlife reservoir such as Rousettus aegyptiacus fruit bats or their faeces or contact with infected primates [3–5] . Human-to-human transmission occurs through direct contact with blood , body fluids , secretions and tissues of infected individuals or dead bodies . Marburg virus can be transmitted sexually and studies of Ebolavirus have shown that viral RNA can be detected in semen for up to 407 days [6] . The incubation period can last from 2–21 days and infected individuals are not viremic until initial symptom onset [7] . Three previous MVD outbreaks have been reported in Uganda [8–10] . The first recorded MVD outbreak was in 2007 , in which three cases and one death were reported [8] . In 2012 , a second MVD outbreak was documented , with a total 26 confirmed and probable cases , of which 15 ( 58% ) were fatal [10] . This outbreak started in Ibanda District and subsequently spread to at least four additional districts including Mbarara , Kabale , Kamwenge and Kampala . A third outbreak was confirmed in 2014 involving a single MVD case identified in the capital Kampala [9] . Also , two additional outbreaks of MVD , one in the Netherlands and another in United States of America have been linked to Uganda [11 , 12] While the viral reservoir for Ebola Virus Disease ( EVD ) has not been definitively determined , one reservoir of Marburg virus has been shown to be R . aegyptiacus bats . R . aegyptiacus bats trapped in Kitaka mine and Python Cave located in the Albertine region of Western Uganda have been shown to be reservoirs of Marburg virus [4 , 5 , 13–15] . Following infection of Dutch and American tourists in 2007 and 2008 , respectively , with Marburg virus after bat exposure at Python Cave [11 , 12] , investigators found that 2 . 5% of the R . aegyptiacus bats in this cave were Marburg virus -positive , using a viral specific PCR assay [4 , 5] . Bats in Python Cave and Kitaka mine have been linked to four MVD outbreaks [8 , 10–12] . Infected R . aegyptiacus bats do not appear to develop clinical symptoms or die as a result of infection with Marburg virus [15–17] . The Viral Haemorrhagic Fever ( VHF ) Surveillance and Laboratory Program located at the Uganda Virus Research Institute ( UVRI ) in Entebbe Uganda , received a blood sample from an individual suspected to be infected with a VHF virus from Kaproron Health Centre IV ( Kween District ) on 16th October 2017 . The sample was submitted by Kween District health team after suspecting a VHF following the death of two people in one family with similar clinical symptoms . The serum sample was tested on the day of arrival by RT-PCR for Crimean-Congo haemorrhagic fever virus , Ebolaviruses ( Bundibugyo ebolavirus , Sudan ebolavirus and Zaire ebolavirus ) , Marburg virus and Rift Valley Fever virus and found to be preliminarily positive for Marburg virus . The sample was re-tested and confirmed positive on 17th October 2017 . The Uganda Ministry of Health ( MoH ) National Task Force ( NTF ) was activated on 18th October 2017 at the Public Health Emergency Operations Centre ( PHEOC ) . A multi-sectoral and multidisciplinary National Rapid Response Team ( NRRT ) was established and deployed to Kween and Kapchorwa districts on 18th October 2017 . The outbreak was officially declared by the Ministry of Health ( MOH ) on 19th October 2017 . District Rapid Response Teams ( DRRTs ) were deployed within the affected Districts on the same day . This multi-sectoral team worked to conduct a rapid outbreak investigation and assessment , and initiated intervention measures . We present results from the epidemiological and laboratory investigations of the MVD outbreak and propose recommendations for future filovirus outbreaks .
Kween district was the epicentre of this MVD outbreak as all the cases lived and worked there . Kween district is adjacent to the Eastern border of Uganda with Kenya and is bordered by Kapchorwa district to the West , Bukwo district to the East , Nakapiripirit district to the north , Amudat district to the northeast , Bulambuli district to the northwest and the Republic of Kenya to the south ( Fig 1 ) . The district is located on the northern slopes of Mount Elgon , and has an average altitude of 1 , 900 metres ( 6 , 200 ft ) above sea level . The estimated population of Kween district is 103 , 300 people ( 2012 census ) . District residents mainly engage in subsistence farming of food crops . The communities also engage in animal husbandry and raise a variety of livestock including cattle , sheep , and goats . The most common domestic animal is the donkey , which is often used for transport . The geography and mountainous terrain of Kween district are host to many bat-inhabited caves that are frequently visited by cattle keepers to collect “salt” rocks to feed their animals . The NRRT held initial briefings involving District Rapid Response Teams ( DRRT ) from Kween and Kapchorwa districts after which District Task Forces were activated . Kapchorwa district was included in this response , all confirmed cases sought treatment at the Kapchorwa district hospital , having been referred from the neighbouring Kween district . Four specialised teams ( sub-committees ) were formed and included; case management and infection control , surveillance and laboratory , social mobilization , and response coordination . The sub-committees were tasked to implement key interventions for the response and obtain clinical , epidemiological , laboratory , socio-cultural , ecological and socio-cultural data in order to better characterize the MVD outbreak in the two affected districts . Each specialised team was comprised of subject matter experts from the national and district levels that engaged in the key activities detailed in the sections below . Each team was led by personnel from the respective districts . Clinicians working with epidemiologists and laboratory experts reviewed the clinical notes and preliminary investigations undertaken on the MVD cases admitted to the health facilities of Kaproron Health Centre IV ( Kween district ) and Kapchorwa Hospital ( Kapchorwa district ) . Patient case histories and physical examination findings were extracted with all the essential clinical and epidemiological information captured onto the national VHF case investigation form . To facilitate the identification of additional cases , as in previous outbreaks , a working MVD case definition [9 , 10] was used to classify cases as either suspected , probable or confirmed . A suspected case was defined as any person meeting one or more of the three following criteria: 1 ) Fever ( ≥37 . 5°C axillary body temperature ) and sudden onset with three or more of the following symptoms: loss of appetite , headache , vomiting , abdominal pain , diarrhea , intense fatigue , myalgia and/or joint pains and history of contact with patient with similar symptoms; 2 ) Sudden onset sickness with or without fever and unexplained bleeding from any of the following sites: gastrointestinal tract ( blood in vomitus ) , gums , nose , eyes genital ( non-menstrual ) and any other body site; 3 ) any unexplained and/or sudden death . A probable case was defined as any suspected case with an epidemiological link to a confirmed case . A confirmed case was defined as any person with either a positive PCR , IgM or IgG ELISA laboratory result for MVD , including retrospectively identified cases . The suspected case definition was disseminated to health facilities in the affected and surrounding districts . Similarly , radio programs were conducted to sensitize the public on the symptoms of an MVD case , possible virus transmission routes and steps to control the spread of the disease . An alert desk equipped with contact information for the District Surveillance Officer ( DSO ) was established to receive and coordinate verification of alerts . An ecology team was assembled comprising of District Veterinary Officers ( DVOs ) , Animal Husbandry Officers , the District Natural Resource Officer , the UVRI VHF program team and other DHT members . Initial investigations centred around the community of Kaptum grazing grounds , Kween District , where the initial reported case resided before falling ill . The team performed an environmental assessment of the grazing grounds using a snowballing approach ( e . g . , community members were asked to identify activities of the initial probable case , including caves he visited , in the month before becoming ill ) and conducted interviews with local community and family members of the initial probable case . Caves were identified in the vicinity of this community used by residents for salt mining . The team investigated these caves to look for the presence of the known Marburg virus reservoir host , R . aegyptiacus bats . Three to five millilitres of blood were obtained from all suspected cases of MVD for laboratory testing using real time RT-PCR and ELISA at the UVRI VHF laboratory according to established protocols [18 , 19] . Briefly , RNA was extracted from whole blood using 5X Magmax™ 96 Viral Isolation kit ( Applied Biosystems Inc . , Vilnius , Lithuania ) according to manufacturer’s instructions . Subsequent RT-PCR assays targeted the VP40 viral gene . ELISA for anti-Marburg IgM and IgG detection was performed using 96-well plates . Unless otherwise stated , all ELISA procedures used 100μl test volume per well format; plates were washed 3 times using 0 . 1% Tween-20 in PBS ( v/v ) between all procedures; and all incubation temperatures were at 37°C for 1h . In addition , all reagents used in all procedures were diluted in PBS containing 5% skimmed milk ( also called serum diluent ) . To perform ELISA , all plates were pre-coated overnight at 4°C with Marburg antigens ( for IgG ) and an anti-human IgM ( mu ) antibody ( for IgM ) in physiological buffered solution ( PBS ) . Samples were pre-diluted to 1:100 in serum diluent before their addition onto pre-coated plates . For IgM , the addition of samples onto plates was followed by the addition of a positive antigen ( 1:2 dilution ) on one ( upper ) half of the plate and a mock antigen ( 1:2 ) on the other half ( lower ) of the plate . This was followed by the addition of a primary antibody , a rabbit anti-Marburg antibody ( in a dilution of 1:1500 ) and then a secondary , horseradish peroxidase conjugated , antibody ( in a dilution of 1:8000 ) . The substrate used was 2 , 2′-azino-bis ( 3-ethylbenthiazoline-6-sulfonic acid ( Kirkegaard and Perry Laboratories , Gaithersburg USA ) read at 410nm . For IgG antibody detection , samples were added to precoated plates followed directly by an anti-human IgG conjugate . Similarly , to IgM , the substrate used was 2 , 2′-azino-bis ( 3-ethylbenthiazoline-6-sulfonic acid ( Kirkegaard and Perry Laboratories , Gaithersburg USA ) read at 410nm . Samples collected during the investigation were transported to UVRI using designated vehicles for transportation of specimens , coordinated by the Uganda Central Public Health Laboratory ( CPHL ) in Kampala . A total of 34 individual blood samples were investigated as suspect cases of MVD . Additionally , 36 blood samples from previously identified close contacts of the initial probable case , suspected to be the index case , of this outbreak . These contacts were not actively monitored through contact tracing and had already passed the 21-day follow-up period prior to confirmation of the outbreak . All samples were sent to the laboratory to assess for evidence of previous or active infection using serological and molecular testing . Any acutely positive samples identified were later sent to the Viral Special Pathogens Branch ( VSPB ) laboratory at CDC in Atlanta , GA , USA for secondary confirmation testing , sequencing and isolation . All contacts of the confirmed and probable cases were identified and listed for follow-up to ensure timely identification and isolation of new cases . Contacts with confirmed or probable cases from their date of symptom onset were defined as people who: 1 ) touched the body fluids of a case ( blood , vomit , saliva , urine , faeces; 2 ) had direct physical contact with the body of a case ( alive or dead ) ; 3 ) or shared the linens , clothes , or dishes/eating utensils of a case and 4 ) slept , ate , or spent time in the same household or room as a case . All contacts listed were followed up on a daily basis for 21 days after the last exposure to a probable or confirmed case . Under the supervision of the DSO , contacts were followed-up by trained Village Health Team ( VHT ) members and/or health workers who in turn submitted daily information detailing the health status of contacts under follow-up . The information collected by VHT members included temperature and MVD symptoms . All contacts under follow-up were advised to remain at home and to report any febrile illness to the designated VHT members or health workers . Any contact developing febrile illness during the follow-up period was reported to the alert desk using a dedicated phone number of the DSO . All contacts completing the 21-day follow-up period without developing disease symptoms were dropped from follow-up and encouraged to resume their normal daily routine . All suspected and confirmed cases were treated in designated cubicles in the MVD treatment units and started on supportive treatment including intravenous fluids , correction of electrolyte imbalances , and treatment for secondary infections . Strict barrier nursing and infection , prevention and control measures were observed in the treatment facility . On-the-job training in personal protective equipment ( PPE ) and infection prevention and control measures were provided for local healthcare providers . During initial epidemiological investigations , many of the identified contacts of the initial probable case were found to have already passed the 21-day follow-up period . In order to ensure no unidentified cases or transmission chains were occurring in the community prior to the investigation teams’ arrival in the district , blood samples were collected from close contacts as defined above . Similar investigations were carried out during previous filovirus outbreaks in Uganda where confirmed cases were retrospectively identified through serological testing of contacts of confirmed and probable cases and initially missed by either surveillance prior to outbreak investigation or during the outbreak [10] . A total of 36 close contacts of the initial probable case were sampled for serological and molecular testing . All data collected from cases and contacts were managed using the VHF EpiInfo application [22] by the district biostatisticians , supported by NRRT members . The VHF EpiInfo software has been previously used to manage MVD outbreaks in Uganda in 2014 [9] and was also adopted for data management during the 2014–15 West Africa EVD outbreak [22] . This software has a number of useful assets with respect to VHF outbreak response including: ( 1 ) ensuring that the correct follow-up period is observed for each contact; ( 2 ) generating daily follow-up lists for active case contacts and updating their status after follow up; ( 3 ) linking contacts with their known source cases; ( 4 ) built-in epidemiological data analysis for real-time situational awareness; ( 5 ) removal of contacts from the follow-up list once they have completed the 21-day follow-up or once the source case to which they are linked tests negative ( 6 ) capability to generate transmission chains based on case linkage information; and ( 7 ) implementation of a large number of complex business rules associated with contact tracing without needing user intervention , thus lessening the amount of data entry errors and mistakes . All case and contact data were entered in the VHF EpiInfo application where real-time analysis was carried during the investigations and presented at the daily district task force update meetings . Data was later exported into STATA software for further analysis . Probable and confirmed cases were compared with suspected cases that tested negative ( controls ) in respect to clinical cases and risk factors using Fisher’s Exact test was used to compute a p-value and statistical significance was considered as a p-value ≤ 0 . 05 . This outbreak investigation is a public health emergency that was approved by the National Task Force ( NTF ) for disease outbreaks in Uganda , and hence it was considered a non-research activity . All persons who were interviewed and whose samples were tested gave informed oral consent , and a parent or guardian of any child participant provided informed consent on the child’s behalf .
The full genomic sequence of MARV ( MBG201708608 and MBG201708609 , Fig 4 , red ) falls into a cluster that consists of MARV sequences isolated from humans and bats in Uganda between 2007–9 and 2014 ( Fig 4 , blue ) but are distinct from Marburg virus sequences collected during a previous outbreak in Kabale , Uganda in 2012 ( Fig 4 , KC545387 and KC545388 ) . The 2017 Marburg virus sequences share a recent common ancestor with other Marburg virus sequences collected from Uganda . These include a human MVD case detected in 2014 living in Kampala and Kasese districts ( KP985768 ) , a virus sequence collected from a miner who worked in the Kitaka mine in July 2007 ( 3 , 7 ) and from virus sequences obtained from bats that were collected from either Python Cave or the Kitaka mine during 2007–2009 . The sequences from the 2017 cases were indistinguishable from each other , suggesting person-to-person transmission . The bats inhabiting the caves visited in Kween district were identified as Rousettus species . However , the ecological investigation team was unable to conclusively identify the bats as R . aegyptiacus since trapping of the bats was not performed . Gross visual and physical identification supports the R . spp . identification . Additional ecological studies are needed to identify the species of Rousettus observed and confirm Marburg virus presence in the bat population inhabiting these caves ( Fig 5 ) .
We describe here the first MVD outbreak reported in Eastern Uganda [8] . Previous MVD outbreaks in Uganda have been linked to Kitaka and Python caves in Western Uganda [4 , 11 , 12] , and one MVD outbreak , detected in Central Uganda , which involved a single case in 2014 [9] . The four cases of this outbreak in Kween district , which involved one cluster of closely-related family members and did not spread to other districts within or outside Uganda . This could be attributed to enhanced capacity of the Uganda National VHF and other MOH surveillance structures and their capability to quickly identify , detect and respond to filovirus outbreaks as has been described previously [23 , 24] . As with other MVD outbreaks in Central and Eastern Africa [25–27] , the initial case in this outbreak , KWN001 , appears to have been linked to activities taking place within a cave harbouring Rousettus bats that are a known reservoir of Marburg virus [4 , 5 , 13 , 15] . KWN001 lived and worked in an area surrounded by cliffs within the Mt . Elgon Forest Reserve range , approximately 800m from a cave inhabited by these bats . It is possible that mining salt and collection of manure from the caves exposed KWN001 to excreta and/or other Marburg virus-contaminated material from these bats , and that this is the most probable initial source of virus exposure in this outbreak . While no previous outbreaks of MVD have occurred in this region of Uganda , R . aegyptiacus habitat has been well-documented throughout the Mt . Elgon region [28] , and a recent filovirus risk mapping exercise has shown that this region is at risk of MVD outbreaks [29] . However , many questions remain regarding the exact mechanism behind spillover events of Marburg virus from the wildlife reservoir to human population . One possible explanation has been the observation that Marburg virus is shed in saliva more than through other routes [15] . Residue of bat saliva on fruits such as mangoes , guavas , and apples can cause a spillover event into humans who may consume these foods without properly washing them . However , in this particular outbreak , we believe that the most likely spillover event occurred as a result of salt mining and manure collection activities performed by KWN001 in a cave occupied by Rousettus spp . bats . Small injuries sustained by KWN001 from pounding rocks contaminated with urine , saliva and feces of bats may have contributed to exposure and infection . Another potential risk factor for KWN001 was direct exposure to bats via hunting , but it remains uncertain if he performed this activity . The observed case-fatality rate ( CFR ) of 75% is higher than what has historically been reported in other parts of Uganda . Apart from outbreaks with single fatal cases [9 , 11 , 25 , 30] , most outbreaks of MVD have reported CFRs of below 50% [8 , 10 , 31–33] . Only two outbreaks have reported a large number of cases; the first occurred in the Democratic Republic of the Congo ( DRC ) in 1998 where 154 cases were recorded , and the second occurred in Angola in 2004 with 254 recorded cases . These outbreaks had CFRs of 83% and 90% , respectively [26 , 27] . Although the CFR in the Kween outbreak is considered high , the rapid recognition of a MVD outbreak and epidemiological response could have resulted in the small number of cases . The high CFR seen in this outbreak might be attributed to the cases seeking healthcare late in the course of illness , at which point supportive care is typically less effective [34] . It is also believed that the all four initially sought treatment from traditional healers , and only later sought care from local clinics , at which point their illness was more severe . Additionally , the high CFR in this outbreak may be attributed to other host factors such as immunological status at the time of infection and host genetic factors since all fatalities were in related family members . Further analysis into other factors , such as viral genetic attributes or route of transmission , contributing high CRF should be investigated . Unlike previous outbreaks of MVD in Uganda in 2007 and 2012 [8 , 10] , clinical symptoms of MVD in this outbreak were slightly different and more severe with all the four cases exhibiting fever , vomiting , abdominal pain , general body weakness , anorexia and joint pain ( Table 1 ) . While not always present early during filovirus infections , bleeding is a common symptom associated with disease progression . In this outbreak , symptoms of bleeding from body orifices were more prevalent than in other filovirus outbreaks , where bleeding symptoms are usually observed in 50% or less of the cases . Additionally , hiccups , which are highly associated with filovirus infections [10] , were found in 50% of cases . Evidence from recent outbreaks of MVD and Ebola Virus Disease ( EVD ) suggests that a minority of infected individuals may develop mild infections that would not otherwise be detected through standard VHF surveillance or outbreak investigation methods [10 , 35] . In one such instance in the October 2012 outbreak of MVD in Kabale , Southwestern Uganda , an investigation identified retrospective cases through serological analysis that suggests the outbreak had actually begun in early June 2012 in neighboring Ibanda district , and not in the Kabale area where the previously-identified initial case was identified [10] . In 2011 during the investigation of an EVD outbreak in Luweero district , Central Uganda , one convalescent case was discovered in a family member of a PCR-confirmed case . This was an indication of past infection not acquired from the confirmed case . Additionally , the convalescent case did not recall having any illness similar to Ebola Virus Disease ( EVD ) in the past , suggesting that mild and undetected infection may have occurred [35] . Lastly , during a 2012 outbreak of EVD in Kibaale district , convalescent cases were retrospectively detected through serological testing of outbreak contacts that were followed but did not develop or report symptoms . Following on this experience , we collected samples from close contacts of the first probable case and identified one person that had both IgG and IgM for Marburg virus . These individuals had already passed their 21-day follow-up observation period recommended after contact with a filovirus case due to the first case being identified outside of this window of time . Therefore , when an outbreak begins after a majority of contacts have passed their 21-day follow-up periods , the testing of close contacts may be useful in the identification of unrecognized unreported cases , thus providing critical epidemiological information to characterize and describe the extent of the outbreak and identify additional contacts requiring follow-up . Marburg virus sequenced from acute cases in this outbreak are indistinguishable ( excluding missing coverage at the 5’ and 3’ gene segments ) , suggesting this outbreak resulted from a single spill over event with subsequent human-to-human transmission ( Fig 4 ) . This is further supported by the transmission chain demonstrating that close contact through nursing care and through participation in funeral rites; these activities were notable risk factors for Marburg virus infection in the epidemiology of the outbreak . Close contact with an infected person has been documented as a major risk factor in the transmission of filovirus during outbreaks . Additional research is needed to characterize the molecular epidemiology and ecology of Marburg virus in this region . We observed that the viral sequences from this outbreak in Eastern Uganda share a common ancestor with viral sequences from the 2014 MVD outbreak in Central Uganda involving a single case [9] . Conversely , the 2017 and 2014 sequences were distinct from the 2012 sequences collected in Southern Uganda ( Kabale district ) suggesting the presence of two unique Marburg virus populations . However , Marburg viruses isolated from populations of R . aegypyiacus bats residing in the same cave can contain sequences similar to these two Marburg virus populations and Ravn , thus making geographic associations of Marburg virus to the local bat populations extremely difficult . This point has been demonstrated in R . aegypyiacus bat populations captured in Kitaka and Python caves in Western Uganda [4 , 13] . Additional research is under way to study the movement of R . aegyptiacus bats , determine their populations in Western Uganda and try to identify other populations they may interact with over large geographic areas in different regions of Uganda , and how this might impact the transmission of Marburg and other viruses . Given the presence of caves inhabited with Rousettus spp . bats throughout different regions in Uganda , and the human-bat interface within caves and the environment , there continues to be need for effective VHF surveillance throughout Uganda . In conclusion , this was the fourth MVD outbreak to be laboratory-confirmed in Uganda , and the first in Eastern Uganda . Uganda was able to rapidly identify , confirm , respond and control the outbreak and prevent nationwide , and possibly international , spread of these viruses . This capacity has been developed over a period of 10 years since the 2007 Bundibugyo Ebola outbreak , after which enhanced VHF surveillance and laboratory detection was established in Uganda . The success of this outbreak response greatly contributed to have limiting the total number of MVD cases to only four . This again demonstrates how continued enhanced surveillance and rapid response for filovirus , as well as other VHFs and emerging zoonotic diseases , can contribute to global health security by controlling outbreaks at the source and preventing them from becoming global epidemics as was seen in the West Africa Ebola epidemic in 2014–15 . | Marburg virus disease ( MVD ) is caused by the virus that belongs to the same family as that of Ebola Virus disease . The disease is characterized by severe clinical symptoms such as high fever , diarrhoea and vomiting , and severe bleeding from most body openings . On average , 54% of the people who get infected with Marburg virus die from it [1] . In October 2017 , Uganda reported an outbreak of MVD in the eastern district of Kween that borders Kenya in the Mount Elgon plains . This was the first time MVD was being detected in this part of the country since previous MVD outbreaks had been reported in the western part of Uganda . The Ministry of Health together with partners instituted rapid outbreak response for control . Investigations revealed a cluster of one family involving four cases whereby only one survived . The outbreak was traced back into bat caves in Kween district . Rock salt mining in the cave led to a spill-over of the virus into the human population with the subsequent person-to-person transmission . Through contact tracing and isolation of the infected people among other approaches , the outbreak was brought under control as explained in this article . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"tropical",
"diseases",
"marburg",
"virus",
"uganda",
"geographical",
"locations",
"vertebrates",
"animals",
"mammals",
"viruses",
"filoviruses",
"rna",
"viruses",
"signs",
"and",
"... | 2019 | Marburg virus disease outbreak in Kween District Uganda, 2017: Epidemiological and laboratory findings |
Human strongyloidiasis varies from a chronic but limited infection in normal hosts to hyperinfection in patients treated with corticosteroids or with HTLV-1 co-infection . Regulatory T cells dampen immune responses to infections . How human strongyloidiasis is controlled and how HTLV-1 infection affects this control are not clear . We hypothesize that HTLV-1 leads to dissemination of Strongyloides stercoralis infection by augmenting regulatory T cell numbers , which in turn down regulate the immune response to the parasite . To measure peripheral blood T regulatory cells and Strongyloides stercoralis larval antigen-specific cytokine responses in strongyloidiasis patients with or without HTLV-1 co-infection . Peripheral blood mononuclear cells ( PBMCs ) were isolated from newly diagnosed strongyloidiasis patients with or without HTLV-1 co-infection . Regulatory T cells were characterized by flow cytometry using intracellular staining for CD4 , CD25 and FoxP3 . PBMCs were also cultured with and without Strongyloides larval antigens . Supernatants were analyzed for IL-5 production . Patients with HTLV-1 and Strongyloides co-infection had higher parasite burdens . Eosinophil counts were decreased in the HTLV-1 and Strongyloides co-infected subjects compared to strongyloidiasis-only patients ( 70 . 0 vs . 502 . 5 cells/mm3 , p = 0 . 09 , Mann-Whitney test ) . The proportion of regulatory T cells was increased in HTLV-1 positive subjects co-infected with strongyloidiasis compared to patients with only strongyloidiasis or asymptomatic HTLV-1 carriers ( median = 17 . 9% vs . 4 . 3% vs . 5 . 9 p<0 . 05 , One-way ANOVA ) . Strongyloides antigen-specific IL-5 responses were reduced in strongyloidiasis/HTLV-1 co-infected patients ( 5 . 0 vs . 187 . 5 pg/ml , p = 0 . 03 , Mann-Whitney test ) . Reduced IL-5 responses and eosinophil counts were inversely correlated to the number of CD4+CD25+FoxP3+ cells . Regulatory T cell counts are increased in patients with HTLV-1 and Strongyloides stercoralis co-infection and correlate with both low circulating eosinophil counts and reduced antigen-driven IL-5 production . These findings suggest a role for regulatory T cells in susceptibility to Strongyloides hyperinfection .
Strongyloides stercoralis infects 60 million individuals in tropical and subtropical areas of the world [1]–[3] . It is unique among helminths in that it can complete its life cycle inside a single human host [4] . The clinical presentation of human strongyloidiasis varies with the status of the host's immunity [5] . Immunocompetent individuals develop a chronic , asymptomatic or mildly symptomatic infection . Patients treated with corticosteroids , cancer patients and persons infected with the Human T-cell-lymphotropic virus 1 ( HTLV-1 ) , may develop an accelerated form of infection termed hyperinfection , characterized by gastrointestinal and pulmonary hemorrhage , and secondary bacterial infections due to large numbers of parasite larvae migrating from the gut through the lung [6]–[9] . Strongyloides hyperinfection has a high fatality rate [10] . Animal models have only provided limited understanding of how the host controls S . stercoralis , because the parasite cannot complete the lifecycle in mice [11] . Though murine models are not ideal for this particular parasite , careful studies have suggested a role for innate and adaptative immune mechanisms of control [12] . The innate response requires eosinophils , which kill the larvae [13] . The cytokine IL-5 is essential for development and activation of eosinophils [14] . The adaptive response involves specific antibody production ( including IgG and IgE ) , and granulocytes [15]–[17] . Neutrophils are more important than eosinophils in killing [15] . However , eosinophils are required to generate an optimal antibody response [14] , serving as antigen presenting cells [18] . Chemokines attracting eosinophils ( eotaxin ) and granulocytes ( IL-8 ) are also needed to kill the larvae , presumably by attracting these granulocytes to kill the parasites [19] . There are few studies on the human immune response to S . stercoralis infection [7] . Since hyperinfection develops in transplant and corticosteroid-treated patients , some thought that the cellular immune response might play an important response in controlling infection . HIV is a retrovirus that causes depletion of the cellular immune response and acquired immunodeficiency syndrome ( AIDS ) . However , even with severe depletion of CD4 T cells , AIDS patients control dissemination of S . stercoralis [20] . Thus , factors other than the cellular immune response are likely more important in control of strongyloidiasis . HTLV-1 is another human retrovirus endemic in Japan , Africa , the Caribbean and South America [8] . In contrast to patients with AIDS , patients co-infected with S . stercoralis and HTLV-1 frequently develop S . stercoralis hyperinfection [9] , [10] , [21] . HTLV-1 and HIV share basic retroviral structures and modes of transmission . Both HTLV-1 and HIV target CD4+ T lymphocytes , but , in contrast to HIV infection where T cells are killed , HTLV-1 activates CD4+ T cells and induces CD4+ T cell proliferation through up-regulation of interleukin-2 ( IL-2 ) and its receptor [8] , [22]–[24] . The cytokine profile of T cells isolated from patients with HTLV-1 includes T helper 1 ( Th1 ) cytokines ( e . g . IFN-γ ) as well as regulatory cytokines ( e . g . IL-10 and TGFβ ) [25]–[27] . Unlike HIV patients , patients with HTLV-1 are not susceptible to intracellular opportunistic infections [8] . However , disseminated strongyloidiasis is more common in HTLV-1-infected patients than in HIV patients [9] , [21] . Regulatory T cells are increasingly recognized as playing a key role in reducing injurious host inflammatory and immune responses [28] . Regulatory T cells control the immune response by different mechanisms: cell-to-cell contact , inhibitory cytokines ( e . g . TGF-β , IL-10 ) and by cytokine deprivation [29] . These cells prevent an excessive immune response and bystander tissue damage during the host response to infections [28] . In murine models of Leishmania major infection , regulatory T cells prevent complete elimination of the parasite , yet parasite persistence is required for maintaining the protective immune response [30] . Some studies demonstrate that nematode infections ( human and mice ) induce regulatory cell expansion [31] . Foxp3 expression levels have been evaluated in patients with HTLV-1 associated myelopathy and asymptomatic carriers with varied results [32] , [33] . Foxp3 levels inversely correlate with the rate at which cytotoxic T cells kill HTLV-1 infected lymphocytes in an ex-vivo model [34] . Regulatory T cells have not been studied in HTLV-1 and S stercoralis infected patients . We hypothesize that HTLV-I leads to dissemination of Strongyloides stercoralis by augmenting regulatory T cells ( Treg ) , which in turn down regulate the immune response against this parasite , allowing for the hyperinfection syndrome .
The Instituto de Medicina Tropical ‘Alexander von Humboldt’ at the Universidad Peruana Cayetano Heredia ( IMT AvH - UPCH ) in Lima , Peru is a national referral center for the study of HTLV-1 and its associated diseases . Patients with strongyloidiasis are routinely tested for HTLV-1 infection . Between November 2005 and August 2007 , all newly diagnosed subjects with strongyloidiasis were invited to participate . Strongyloidiasis diagnosis was confirmed by stool examination . All those with S . stercoralis in stool were tested for HTLV-1 infection by Enzyme-Linked Immuno Sorbent Assay ( Ortho HTLVI/HTLVII Ab-capture ELISA , Ortho-Clinical Diagnostics , USA ) with confirmatory western blot analysis ( INNO-LIA HTLVI/II Score , Innogenetics , Belgium ) . Consenting patients underwent further clinical and laboratory evaluation , including complete blood count ( CBC ) with differential , flow cytometry analysis and immunological studies of peripheral blood mononuclear cells ( PBMCs ) . Demographic and symptoms data were obtained from patient interviews and reviews of clinical records . Stool and blood samples were collected at the outpatient clinic at the time of enrollment . In addition , asymptomatic HTLV-1 infected patients and healthy HTLV-1 negative subjects were enrolled as controls . All participants signed a written informed consent form prior to enrollment in the study . The Institutional Review Board ( Comité Institucional de Ética ) of the Universidad Peruana Cayetano Heredia in Lima , Perú approved the study protocol and consent forms . Routine stool evaluation for ova and parasites included direct examination , and spontaneous and fast sedimentation techniques . Stools were examined for S . stercoralis using the Baermann technique modified by Lumbreras [35] . S . stercoralis parasite load was reported in a semi-quantitative scale as negative , 1+ , 2+ , 3+ or 4+ according to the number of larvae observed under microscopy by personnel who were blinded to the HTLV-1 status of the participant . PBMCs were isolated from heparinized blood by density gradient centrifugation ( BD Vacutainer CPT Cell Preparation Tube with Sodium Heparin , NJ ) . Regulatory cells were defined by staining for CD4 , CD25 , and FoxP3 . PBMCs were first stained using Cy-5 conjugated anti-CD4 and phycoerythrine ( PE ) -conjugated anti-CD25 monoclonal antibodies ( BD Biosciences , San Jose , California , USA ) . After fixing and permeabilizing , the cells were then stained for intracellular FoxP3 using a fluorescein-isothiocyanate ( FITC ) -conjugated anti-FoxP3 monoclonal antibody ( eBiosciences , San Diego , California , USA ) . Cells were analyzed using a FACScalibur flow cytometer ( Beckton Dickinson , Franklin Lakes , New Jersey , USA ) . Regulatory T cells were identified as CD25+ and FoxP3+ cells among CD4+ cells within the lymphocyte gate . Absolute CD4+ cell counts were performed using a 4 color single platform staining of whole blood cells ( anti CD3-FITC , CD4-PE , CD45 PerCP and CD8 APC ) . Flow cytometry analysis used FlowJo software ( V . 8 . 5 Tree Star , Ashland , Oregon , USA ) . S . stercoralis third-stage larvae ( L3 ) were obtained and cleaned as previously described [36] . Briefly , feces from an infected dog was mixed with bone charcoal and cultured at 25°C for 7 days . The L3 were obtained from the fecal-charcoal culture by the use of a Baermann apparatus and allowed to settle in a tube for 30 minutes . The supernatant was then removed and the pellet of L3 was resuspended with an equal volume of 2% liquid low-gelling temperature agar ( Sigma Chemical Co . , St . Louis , MO , USA ) at 37°C . The worm-agar mix was allowed to solidify in the center of a Petri dish and then covered with invertebrate saline . The worms were allowed to migrate out of the solidified agar ( leaving most of the bacteria and fecal debris behind ) at 37°C for 1 hour . The saline with the clean worms was then centrifuged to pellet the worms . The L3 pellet was frozen at −20°C until needed . Crude antigen was obtained by sonication of L3 larvae and centrifugation , supernatant was collected and protein concentration was determined by colorimetric Bradford assay . PBMCs were cultured in RPMI-1640 , 10% Fetal Bovine Serum antibiotic supplemented media ( 37°C in 5% CO2 ) for 72 hours in the presence or absence of 2-ug/ml crude infective stage S . stercoralis larvae ( L3 ) antigen . PBMCs were adjusted to a final concentration of 1×106 cells/ml . Supernatants were collected and stored at −80°C until cytokine analysis . Interleukin 5 ( IL-5 ) was measured by enzyme-linked immuno sorbent assay on the cell culture supernatant following manufacturer's instructions ( BD OptEIA , BD Biosciences , San Diego , California , USA ) . HTLV-1 proviral load in PBMCs was quantified in the strongyloidiasis and HTLV-1 co-infected patients . HTLV-1 proviral load was quantified using a real-time SYBR Green PCR method , as described previously [37] . We compared symptoms and parasite load between patients with and without HTLV-1 by Chi square test . Blood results including regulatory T cell numbers and proportions were compared by ANOVA . The IL-5 responses to antigen stimulation were compared by Mann-Whitney U test . The correlation between the IL-5 responses and regulatory T cells were compared by Spearman's rank test .
Demographic data are presented in Table 1 . The median age for all patients was 42 . 0 years ( range 22 to 77 ) . Patients with co-infection were slightly older than those with S . stercoralis alone ( 45 . 0 vs . 39 . 0 years . p = 0 . 36 ) . Gender distribution was similar for both groups , with 55% male and 45% female . Patients were originally from Andean , Coastal , and Jungle regions in decreasing order of frequency for both groups . Most patients had visited the jungle within three months of onset of gastrointestinal symptoms . Abdominal pain , diarrhea , flatulence , nausea , vomiting , and increasing abdominal girth were the most common complaints in S . stercoralis groups , with over 50% of patients affected . Patients with S . stercoralis and HTLV-1 co-infection reported more symptoms than those only infected with S . stercoralis ( Table 1 ) . S . stercoralis and HTLV-1 co-infected patients showed more parasites in stool samples compared with patients infected with S . stercoralis only . A high parasite burden ( 3+ or 4+ larvae ) was noted in 10 ( 83% ) of twelve of the co-infected patients . Conversely , patients with just S . stercoralis had a low burden of disease ( 84% were 1+ or 2+ ) ( p<0 . 001 , Chi-square test , Figure 1 ) . Other parasites were found in the stool of 51% of all patients , including Blastocystis hominis ( 39% ) , Trichuris trichiura ( 10% ) , hookworms ( 8% ) , Giardia lamblia ( 5% ) , and Ascaris lumbricoides ( 2 . 5% ) . Median hemoglobin and hematocrit were lower in the S . stercoralis/HTLV-1 co-infected group compared to HTLV-1 asymptomatic carriers and S . stercoralis-only patients ( 12 . 8 versus 13 . 8 versus 14 . 1 g/dl , p = 0 . 11; and 38 versus 41 and 42% p = 0 . 054 One way ANOVA ) . Total white blood cell counts and neutrophil counts did not differ significantly among the groups . HTLV-1 asymtpmatic carriers did not show increased eosinophil numbers . Among S . stercoralis infected subjects there was a trend towards lower median eosinophils in patients co-infected with HTLV-1 ( 4% versus 7%; 200 versus 511 cells/mm3 ( Table 2 ) . The median CD4+ and CD8+ T-cell count were similar in strongyloidiasis patients regardless of their HTLV-1 status ( Table 2 ) . By contrast , HTLV-1 co-infected patients had an expanded CD4+CD25hi lymphocyte population . Furthermore , the proportion of CD4+ T-cells with regulatory T-cell phenotype ( CD4+CD25+FoxP3+ ) was significantly increased in the HTLV-1 co-infected patients ( median proportion of CD4+ T-cells that were CD25+FoxP3+: 17 . 9% ) when compared to all other groups ( p<0 . 0001 One way ANOVA , Table 2 , Figures 2 and 3 ) . PBMC's from strongyloidiasis patients produced IL-5 in response to S . stercoralis infective stage larvae crude antigen . However , the IL-5 response was reduced in S . stercoralis and HTLV-1 co-infected patients ( 10 vs 299 pg/ml respectively , p = 0 . 0004 , Mann-Whitney test , Figure 4 ) . An inverse correlation was observed between the number of regulatory T-cells and IL-5 cytokine responses ( Spearman r = −0 . 39 , p = 0 . 03 , Figure 5 ) . Proviral load data were available for 9 of the 13 HTLV-1 co-infected patients ( mean 1566 copies/104 PBMCs; ranged from 1442 to 5460 ) . These patients had a significantly higher mean proviral load than did asymptomatic carriers ( mean 561 copies/104 PBMCs; range 1 to 4773 ) and were similar to HTLV-1-associated myelopathy patients ( mean 1783 copies/104 PBMCs; range 142 to 8641 ) , as reported elsewhere [37] .
In this study , we demonstrate that immunocompetent patients with S . stercoralis infection have lower worm burdens and higher eosinophil counts compared to patients who are co-infected with HTLV-1 and normal controls . The immunocompetent patients also produce IL-5 in response to stimulation with S . stercoralis larval antigens . By contrast , patients with S . stercoralis and HTLV-1 co-infection have reduced IL-5 responses to parasite antigens , which may lead to decreased eosinophil count . We also noted increased proportions of CD4+CD25+FoxP3+ regulatory T cells in patients with S . stercoralis and HTLV-1 co-infection compared to either normal controls or patients infected with either HTLV-1 or S . stercoralis only . The proportion of lymphocytes expressing this phenotype was inversely correlated with antigen-driven IL-5 responses . Thus , augmented regulatory T cell function may explain the defective eosinophil numbers in S . stercoralis patients co-infected with HTLV-1 . HTLV-1 is an important risk factor for Strongyloides dissemination , but the mechanisms are poorly understood [38] . Previous studies have demonstrated lower eosinophil counts , decreased production of IL-5 , and increased production of interferon gamma in patients with S . stercoralis and HTLV-1 co-infection [10] . IL-5 is a key growth and activation factor for eosinophils [39] . Eosinophils play a key role in clearance of other helminths larvae [40] . Several mechanisms may be involved including direct cytotoxicity and antibody dependant cellular cytotoxicity [40] . We reasoned that regulatory T cells might be responsible for the reduced IL-5 and eosinophil responses in S . stercoralis and HTLV-1 co-infected patients . Indeed , the proportion of CD4+ cells that were CD25+FoxP3+ was dramatically increased in patients with S . stercoralis and HTLV-1 co-infection . These levels are higher than have previously been described in humans . Furthermore , those with increased proportions of CD25+FoxP3+ cells had decreased antigen-driven production of IL-5 and lower eosinophil counts . Since HTLV-1 can augment expression of CD25 , we also re-analyzed the data without including CD25 expression in the definition of regulatory cells ( i . e . CD4+FoxP3+ ) . We also performed an analysis using CD127 expression ( low or negative ) to define regulatory cells . In all of these analyses , the results were similar . Taken together , these data strongly suggest that Tregs may suppress antigen driven function of T cells , which influences the function of eosinophils . Pro-viral loads were also elevated in the strongyloidiasis and HTLV-1 co-infected patients , similar to levels increased for other HTLV-1 associated diseases like HTLV-1 associated myelopathy . This suggests that this expanded regulatory T-cell population may be induced by HTLV-1-S . stercoralis co-infection or as a response to the proinflammatory effects of these infections . Further studies are needed to determine whether the expansion is in HTLV-1 infection of the Treg cells . In summary , we have demonstrated that patients with S . stercoralis and HTLV-1 co-infection do not control S . stercoralis infection as well as patients with S . stercoralis only as indicated by the increase in the number of larvae in stool samples . This defect in immune host response is associated with lower eosinophil counts and decreased antigen-driven production of IL-5 . This reduced response is inversely correlated with the proportion of CD4 cells , which are CD4+CD25+FoxP3+ , suggesting a role for these cells in blunting antigen-driven protective responses . Additional data is needed to determine whether regulatory T cells are increased because they are HTLV-1 infected or as a response to parasite infection , and to explore mechanisms for suppression of eosinophil counts and function . | Human strongyloidiasis varies from a mild , controlled infection to a severe frequently fatal disseminated infection depending on the hosts . Patients infected with the retrovirus HTLV-1 have more frequent and more severe forms of strongyloidiasis . It is not clear how human strongyloidiasis is controlled by the immune system and how HTLV-1 infection affects this control . We hypothesize that HTLV-1 leads to dissemination of Strongyloides stercoralis by augmenting regulatory T cell numbers , which in turn down regulate the immune response to the parasite . In our study , patients with HTLV-1 and Strongyloides co-infection had higher parasite burdens than patients with only strongyloidiasis . Eosinophils play an essential role in control of strongyloidiasis in animal models , and eosinophil counts were decreased in the HTLV-1 and Strongyloides stercoralis co-infected subjects compared to patients with only strongyloidiasis . The proportion of T cells with a regulatory cell phenotype was increased in HTLV-1 positive subjects co-infected with strongyloidiasis compared to patients with only strongyloidiasis . IL-5 is a key host molecule in stimulating eosinophil production and activation , and Strongyloides stercoralis antigen-specific IL-5 responses were reduced in strongyloidiasis/HTLV-1 co-infected patients . Reduced IL-5 responses and eosinophil counts were inversely correlated to the number of regulatory T cells . These findings suggest a role for regulatory T cells in susceptibility to Strongyloides hyperinfection . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"immunology/immunomodulation",
"virology/immunodeficiency",
"viruses",
"immunology",
"immunology/immune",
"response"
] | 2009 | Regulatory T Cell Expansion in HTLV-1 and Strongyloidiasis Co-infection Is Associated with Reduced IL-5 Responses to Strongyloides stercoralis Antigen |
Paired Immunoglobulin-like Type 2 Receptor Alpha ( PILRA ) is a cell surface inhibitory receptor that recognizes specific O-glycosylated proteins and is expressed on various innate immune cell types including microglia . We show here that a common missense variant ( G78R , rs1859788 ) of PILRA is the likely causal allele for the confirmed Alzheimer’s disease risk locus at 7q21 ( rs1476679 ) . The G78R variant alters the interaction of residues essential for sialic acid engagement , resulting in >50% reduced binding for several PILRA ligands including a novel ligand , complement component 4A , and herpes simplex virus 1 ( HSV-1 ) glycoprotein B . PILRA is an entry receptor for HSV-1 via glycoprotein B , and macrophages derived from R78 homozygous donors showed significantly decreased levels of HSV-1 infection at several multiplicities of infection compared to homozygous G78 macrophages . We propose that PILRA G78R protects individuals from Alzheimer’s disease risk via reduced inhibitory signaling in microglia and reduced microglial infection during HSV-1 recurrence .
Alzheimer’s disease ( AD ) results from a complex interaction of environmental and genetic risk factors [1] . Proposed environmental risk factors include a history of head trauma [2–4] and infection [5–7] . In recent years , large-scale genome-wide association studies ( GWAS ) and family-based studies have made considerable progress in defining the genetic component of AD risk , and >30 AD risk loci have been identified [8 , 9 , 18–20 , 10–17] . A key role for microglial/monocyte biology in modulating risk of AD has emerged from analysis of the loci associated with AD risk . Rare variants of TREM2 , a microglial activating receptor that signals through DAP12 , greatly increase AD risk [11 , 14] . Beyond TREM2 , a number of the putative causal genes mapping to AD risk loci encode microglial/monocyte receptors ( complement receptor 1 , CD33 ) , myeloid lineage transcription factors ( SPI1 ) , and other proteins highly expressed in microglia ( including ABI3 , PLGC2 , INPP5D , and PICALM ) .
The index variant for the Alzheimer’s disease risk locus at 7q21 is rs1476679 ( meta P value = 5 . 6 x 10−10 , odds ratio = 0 . 91 ) [15] . In addition to reduced disease risk , the C allele of rs1476679 has been associated with age of onset [21] and lower odds of pathologic AD ( plaques and tangles ) in the ROSMAP study [22] . In the 1000 Genomes project CEU population ( phase 3 data ) , there were 6 variants in strong linkage disequilibrium ( r2>0 . 9 ) with rs1476679 ( S1 Table ) . None of the 6 variants were predicted to alter regulatory motifs that might influence gene expression ( Regulome DBscore ≤ 4 ) , but one variant ( rs1859788 ) encoded a missense allele ( G78R , ggg to agg transition ) in Paired Immunoglobulin-like Type 2 Receptor Alpha ( PILRA ) protein . Using a cohort of 1 , 357 samples of European ancestry whole genome-sequenced to 30X average read-depth ( Illumina ) , we confirmed the strong linkage between rs1476679 ( in ZCWPW1 intron ) and rs1859788 ( G78R PILRA variant ) ( S1 Table ) . We hypothesized that PILRA G78R was the functional variant that accounts for the observed protection from AD risk . As expected from the strong linkage disequilibrium ( LD ) between PILRA G78R and rs1476679 ( Fig 1A ) , conditional analysis demonstrated that the 2 variants were indistinguishable for AD risk in individuals of European ancestry . In a cohort of 8060 European ancestry samples ( a subset of samples described in 19 ) , individuals homozygous for R78 ( OR = 0 . 72 ) and heterozygous ( OR = 0 . 89 ) for R78 were protected from AD risk relative to G78 homozygotes . We note that the allele frequency of PILRA G78R varies considerably in world populations . Indeed PILRA R78 is the minor allele in populations of African ( 10% ) and European descent ( 38% ) but is the major allele ( 65% ) in East Asian populations [23] . The index variant in the 7q21 locus ( rs1476679 ) has been associated with expression levels of multiple genes in the region , including PILRB [24 , 25] . However , the strongest cis-eQTL in the region is a haplotype tagged by rs6955367 which has a low coefficient of determination to rs1476679 ( r2 = 0 . 085 , D’ = 0 . 982 ) in Europeans and is more strongly associated with expression in whole blood of multiple genes in the region ( PILRB , STAG3L5 , PMS2P1 , MEPCE ) compared to rs1476679 [26] . Since the PILRB eQTL P value for rs1476679 is not significant ( P = 0 . 31 ) after conditioning rs6955367 ( S2 Table ) in whole blood , we conclude that rs1476679 and rs1859788 are not significant causal eQTLs in the 7q21 region and the observed relationship of these SNPs with PILRB expression is due to the weakly correlated variant rs6955367 ( S1 Fig ) . Of interest , the G allele of rs6955367 ( increased expression of PILRB ) is linked to rs7803454 ( r2 = 0 . 83 ) , a variant associated with increased risk of age-related macular degeneration and suggests the presence of independent effects in the PILRA/PILRB region [27] . Paired activating/inhibitory receptors are common in the immune system , with the activating receptor typically having weaker affinity than the inhibitory receptor toward the ligands . PILRA and PILRB are type I transmembrane proteins with highly similar extracellular domains that bind certain O-glycosylated proteins [28–31] , but they differ in their intracellular signaling domains [32–34] . PILRA contains an immunoreceptor tyrosine-based inhibitory motif ( ITIM ) , while PILRB signals through interaction with DAP12 , which contains an immunoreceptor tyrosine-based activation motif ( ITAM ) . Analysis of PILRA knockout mice suggests that PILRA is a negative regulator of inflammation in myeloid cells [35–37] , with knockout macrophages showing increased production of cytokines ( IL6 , IL-1b , KC , MCP-1 ) in addition to increased infiltration of monocytes and neutrophil via altered integrin signaling . PILRA is known to bind both endogenous ( including COLEC12 , NPDC1 , CLEC4G , and PIANP ) and exogenous ligands ( HSV-1 glycoprotein B ( gB ) ) [30 , 31 , 36 , 38] . Because the G78R ( R78 ( AD protective ) ) variant resides close to the sialic acid-binding pocket of PILRA , we tested whether the glycine ( uncharged , short amino acid ) to arginine ( basic , long side chain amino acid ) substitution might interfere with PILRA ligand-binding activity . All non-human PILRA sequences , as well as all PILRB sequences , encode glycine at this position . We also generated amino acid point variants in and around the sialic acid-binding pocket of PILRA . A residue conserved among PILR proteins and related SIGLEC receptors , R126 in PILRA , is well known to be essential for sialic acid interaction [29 , 31 , 38] and so was not further studied here . Based on their location in the crystal structure , evolutionary conservation [31] , and involvement in binding HSV-1 gB [38] , amino acids R72 and F76 were predicted to be important for ligand binding and were substituted to alanine as positive controls for loss-of-function [31] . In addition , S80 , a residue outside of the sialic acid-binding pocket was substituted to glycine . The R72A , F76A , and S80G mutations have not been detected in human populations ( dbSNP v147 ) . To study receptor-ligand binding , 293T cells were transfected with G78 ( AD risk ) PILRA or variants , and then incubated with purified NPDC1-mIgG2a protein ( Fig 1B ) , followed by flow cytometry to detect PILRA and the NPDC1 fusion protein . Among known PILRA ligands , NPDC1 is expressed in the central nervous system and binds with high affinity to PILRA [31] . Expression of the PILRA variants on the transfected 293T cells was comparable to or greater than G78 ( AD risk ) PILRA ( S2 Fig ) . G78 ( AD risk ) PILRA binding to NPDC1 was considered 100% . Both R72A and F76A mutations severely impaired NPDC1 binding ( ~20% of G78 , p-value < 0 . 0001 ) . The R78 ( AD protective ) variant also showed significantly reduced ligand binding ( ~35% of G78 , p < 0 . 0005 ) , while the G80 mutant was the least affected ( ~60% of G78 , p < 0 . 0001 ) ( Fig 1C and S3A and S3B Fig ) . To further test the hypothesis that the AD protective PILRA R78 variant impacts ligand binding , NPDC1 or alternative PILRA ligands HSV-1 gB and PIANP were expressed on the cell surface of 293T cells , and the binding of purified PILRA protein variants was measured by flow cytometry . PILRA R78 showed reduced binding to the various ligands in these assays as compared to G78 ( Fig 1D to 1G and S4A to S4G Fig ) . These data confirmed that the R78 variant impairs ligand-binding activity of PILRA . A peptide motif for PILRA interaction has been established ( Fig 2A ) that includes an O-glycosylated threonine , an invariant proline at the +1 position , and additional prolines at the -1 or -2 and +3 or +4 positions [31 , 38] . Of note , PILRA is capable of binding murine CD99 and human NPCD1 ( both contain the consensus motif ) , but not human CD99 or murine NPCD1 ( both lack the consensus motif ) , suggesting divergence between human and mouse in the range of endogenous ligands bound by PILRA [31] . We sought to identify novel endogenous PILRA ligands by searching for human proteins with either the PTPXP , PTPXXP , PXTPXP or PXTPXXP motif . A total of 1540 human proteins carry at least 1 of these putative PILRA-binding motifs ( S3 Table ) . Narrowing the search , we considered proteins with the motif that have previously been shown to be O-glycosylated in human cerebral spinal fluid [39] , and measured the binding of these proteins to PILRA variants . By flow cytometry , complement component 4A ( C4A ) bound to G78 ( AD risk ) PILRA in a manner comparable to NPDC1 , while APLP1 and SORCS1 showed relatively little interaction with PILRA ( Fig 2B and S5A and S5B Fig ) . We further demonstrated that the PILRA R78 ( AD protective ) variant has reduced binding for C4A ( Fig 2C and S5C Fig ) . We did not test C4B , but its putative PILRA-binding motif is identical to that of C4A . To understand the conformational changes that might occur in the PILRA sialic acid-binding pocket during receptor-ligand interactions in the presence of G78 ( AD risk ) or R78 ( AD-protective ) variants , we evaluated available experimental crystal structures ( Fig 3A to 3C ) [38 , 40] . Structures of G78 ( AD risk ) PILRA reveal a monomeric extracellular domain with a single V-set Ig-like β-sandwich fold that binds O-glycan ligands ( Fig 3B and 3C ) [38] . By analogy to a molecular clamp , the sialic acid-binding site in PILRA undergoes a large structural rearrangement from an “open” to a “closed” conformation upon binding its peptide and sugar ligands simultaneously ( Fig 3A to 3C ) . The essential R126 side-chain engages the carboxyl group of sialic acid ( SA ) directly in a strong salt bridge ( Fig 3C ) . The CC’ loop which contains F76 and G78 undergoes a large conformational change where F76 translates ~15 Å to participate in key interactions with the peptide of the ligand and abut the Q140 side-chain of PILRA ( Fig 3B and 3C ) . In this ligand-bound “closed” conformation of PILRA , Q140 helps to position R126 precisely for its interaction with SA ( Fig 3C ) . Notably , in the structure of R78 ( AD protective ) PILRA crystallized in the absence of any ligand [40] , the long side-chain of R78 is observed to hydrogen bond with Q140 directly ( Fig 3A ) . This unique R78-Q140 interaction has three important consequences: 1 ) it sterically hinders F76 from obtaining a ligand-bound “closed” conformation , 2 ) it affects the ability of R126 to interact with the carboxyl group of SA by altering the R126-Q140 interactions observed in G78 ( AD risk ) PILRA and , 3 ) it likely alters CC’ loop dynamics , ( Fig 3B to 3C ) . Overall , the structure of the R78 ( AD protective ) variant shows that this single side-chain alteration appears to stabilize the “open” apo form of PILRA and likely alters the conformational sampling of the molecular clamp required to obtain its “closed” form to engage its ligands . We therefore propose that in G78 PILRA ( AD-risk associated ) , the engagement of SA by R126 and peptide by F76 is facilitated by G78 ( Fig 3C ) . However , in the AD-protective PILRA variant R78 , the R78 side-chain competes with the central R126-Q140 interaction and alters the positioning of F76 ( Fig 3A ) , which leads to an overall decrease in PILRA ligand binding . This structure-based hypothesis is consistent with the reduced functional cellular binding observed for the R78 variant ( Fig 1 ) . To further test this model , we generated two additional alanine mutants of PILRA at amino acids predicted to be essential ( Q140 ) or non-essential ( S141 ) for conformational changes associated with ligand interaction . 293T cells were transfected with G78 ( AD risk ) , R78 ( AD protective ) , Q140A and S141A variants of PILRA , and receptor-ligand interaction was measured after incubating cells with soluble NPDC1-mIgG2a . PILRA expression was comparable among variants , matching or exceeding G78 ( AD risk ) expression ( S2 Fig ) . R78 ( 44% of G78 , p = 0 . 02 ) and Q140A ( 22% of G78 , p = 0 . 0004 ) variants showed significantly decreased binding to NPDC1 , while S141A ( 117% of G78 , p = 0 . 5 ) had no significant effect ( Fig 3D and S6A and S6B Fig ) . These data are consistent with the experimental structural models that show the interaction of Q140 with R126 is important for productive sialic acid binding ( Fig 3A to 3C ) . Consistently , the Q140A mutation has a strong effect because the Q140-R126 interaction network is completely abolished . By contrast , the AD-protective R78 variant likely has an intermediate effect since it only modulates the Q140 interaction with R126 , which is expected to only alter the frequency or strength of relevant PILRA-ligand interactions . We next investigated the interaction of PILRA variant and ligands in vitro using surface plasmon resonance ( SPR ) . Human PILRA–Fc variants ( G78 , R78 , or Q140A ) were immobilized on a ProteOn GLC sensor chip and binding of NPDC1-mFc or a control mFc-tagged protein was measured . Qualitatively , NPDC1-Fc bound to the R78 ( AD-protective ) and Q140A ( essential for R126 conformation ) variants to a much lesser extent than to G78 ( AD risk ) PILRA , while control Fc-tagged protein showed no binding ( Fig 3E ) . To further probe the mechanistic basis of R78 ( AD protective ) function and phenotype , a more complete SPR characterization of NPDC1-His binding to PILRA variants was performed ( S6C Fig ) . The affinity of NPDC1 toward R78 ( AD-protective ) PILRA ( 76 . 5 nM ) was 4 . 5-fold weaker than the affinity toward G78 PILRA ( 16 . 8 nM ) . The on-rate constant kon for NPDC1-His binding to R78 ( AD protective ) ( 6 . 8×10+3 M-1s-1 ) was ~3-fold lower than binding to G78 ( AD risk ) PILRA ( 2 . 2×10+4 M-1s-1 ) , while the koff rate constants were comparable ( S6C Fig ) . These results are consistent with the idea that , once engaged , the affinity and disassociation rate of R78-ligand complexes are similar to G78 PILRA , but the frequency with which PILRA can productively engage with ligands is reduced in the R78 ( AD protective ) variant by R78 side chain interactions favoring the apo-state ( Fig 3 ) . Taken together , these data support a structural model in which R78 impairs PILRA-ligand interactions by altering the accessibility of a productive sialic acid-binding conformation in PILRA . Given that PILRA is a known entry receptor for HSV-1 [41] and the R78 ( AD protective ) variant showed reduced binding to HSV-1 gB ( Fig 1F ) , we next determined whether there were differences in HSV-1 infectivity based on PILRA genotype . We isolated and differentiated human monocyte-derived macrophages ( hMDMs ) from five pairs of healthy volunteers homozygous for either the G78 ( AD risk ) or R78 ( AD protective ) PILRA variants ( matched for age , gender and ethnicity ) . hMDMs were infected with HSV-1 at different multiplicities of infection ( MOI ) ( 0 . 01 , 0 . 1 , 1 and 10 ) , and infectivity was measured morphologically by light microscopy , by using an LDH cytotoxicity assay , by measuring intracellular viral DNA and in a viral plaque assay . No notable cytopathic effects were observed in the first 6 h of infection , however at 18 hours post infection , extensive cytopathy was detected in G78/G78 PILRA-expressing hMDMs , including loss of cell shape , increased cell volume , birefringence , and formation of both cell aggregates and multinucleated giant cells ( syncytia ) ( Fig 4A and S7 Fig ) . Cytopathic changes were less pronounced in R78/R78 ( Alzheimer’s protective ) homozygous hMDMs ( Fig 4A and S7 Fig ) . hMDMs from R78/R78 PILRA donors showed significantly less HSV-1-induced cytotoxicity at 18 hrs post infection in the LDH assay at 0 . 01 , 0 . 1 , or 1 MOI ( Fig 4B and S4 Table ) . The difference was no longer significant at 10 MOI or if the infection was allowed to proceed for 36 hrs , except at the lowest MOI of 0 . 01 ( Fig 4B , S8A Fig and S4 and S5 Tables ) . hMDMs from R78/R78 donors showed 5–10 fold decreased amounts of HSV-1 DNA at 6 hrs at all MOIs ( 0 . 01 , 0 . 1 , 1 and 10 ) , and at 18 hrs at lower MOIs ( 0 . 01 and 0 . 1 ) , compared to those from G78/G78 donors ( Fig 4C and S8C and S8D Fig ) . No significant differences in HSV-1 DNA were observed between the two genotypes at 18 hrs at higher doses ( 1 and 10 MOI ) ( Fig 4C and S8D Fig ) , or at 36 hrs for any dose of virus ( S8B and S8E Fig ) . Finally , we measured the amount of infectious HSV-1 virus by harvesting supernatants from HSV-1-infected hMDMs and measuring viral titer by plaque assays on Vero cells . Viral plaque forming units ( PFUs ) were significantly lower after 6 and 18 hrs of infection for all MOIs tested , and at 36 hrs for lower MOIs ( Fig 4D and 4E , and S9 Fig ) . Taken together , these data indicate that R78/R78 macrophages were less susceptible to HSV-1 infection than G78/G78 macrophages .
We show here that PILRA G78R is a likely causal variant conferring protection from AD risk at the 7q21 locus . G78R alters the access to SA-binding pocket in PILRA , where R78 PILRA shows reduced binding to several of its endogenous cellular ligands and with HSV-1 gB . Reduced interaction with one or more of PILRA’s endogenous ligands ( including PIANP and NPDC1 ) could impact microglial migration or activation [35–37] . In fact , microglia up-regulate the expression of the PIANP gene in the PS2APP , 5xFAD , and APP/PS1 mouse models of AD [42–44] . The identification of C4 as a novel PILRA-interacting protein is also intriguing , given the increased expression of C4 in mouse AD models [42] , the increase in amyloid deposition observed when complement activation is inhibited [45] , and the genetic association of complement receptor 1 with AD [46] . Finally , we note that both TREM2 and PILRB function as activating receptors and signal through DAP12 [32 , 34 , 47] . A reduction of PILRA inhibitory signals in R78 carriers could allow more microglial activation via PILRB/DAP12 signaling and reinforce the cellular mechanisms by which TREM2 is believed to protect from AD incidence [48] . The relevant ligands for PILRA/PILRB in vivo and the mechanism by which reducing PILRA-ligand interaction confers protection from Alzheimer’s disease remain to be elucidated . A role for infection in accelerating AD has been proposed , but remains controversial [49] . HSV-1 is a neurotropic virus that infects a large fraction of the adult population and has frequent reactivation events . HSV-1 has been implicated in AD pathogenesis by several lines of evidence , including the presence of HSV-1 viral DNA in human brain tissue [50 , 51] , increased HSV-1 seropositivity in AD cases [52–55] , the correlation of high avidity HSV-1 antibodies with protection from cognitive decline [55] , the binding of HSV-1 gB to APOE-containing lipoproteins [56] , HSV-1-induced amyloidogenic processing of amyloid precursor protein ( APP ) [57–59] , and preferential targeting of AD-affected regions in HSV-1 acute encephalitis [60] . In addition , HSV-1 gD receptors and gB receptor PILRA increase with age in multiple brain regions , including the hippocampus [61] . Additional AD risk loci have been proposed to play a role in the life cycle of HSV-1 [62] , including CR1 , which is capable of binding HSV-1 [63] . The reduced infectivity of HSV-1 in R78/R78 macrophages suggests that brain microglia from R78/G78 and R78/R78 individuals are less susceptible to HSV-1 infection and more competent for immune defense during HSV-1 recurrence . These data provide additional evidence for a key role of microglia in AD pathogenesis and provide a mechanism by which HSV-1 may contribute to AD risk . Inhibiting the interaction of PILRA with its ligands could therefore represent a novel therapeutic mechanism to prevent or slow AD progression .
Blood samples and genotypes from healthy human volunteers from the Genentech Genotype and Phenotype program ( gGAP ) were used in this project . Written consent was obtained from all participants in the gGAP program . The study was reviewed and approved by the Western Regional Institutional Board ( Study Number: 1096262 , IRB Tracking Number: 20080040 ) . The conditional analysis between rs1476679 and rs1859788 was performed using the Genome-wide Complex Trait Analysis ( GCTA ) program’s Conditional & joint ( COJO ) analysis option . This program takes summary statistics as input . We used the summary statistics for rs1859788 , rs1476679 from IGAP stage1 GWAS [15] . The COJO program also needs a reference population to calculate the LD and to perform the conditional analysis . For reference population analysis we used the raw genotype data from ADGC cohort . There were 22 , 255 individuals in this cohort that had the non-missing genotype for the rs1859788 . The ADGC dataset was also used for the minor allele frequency calculations that are provided in the text . The coding sequences ( CDS ) of full length PILRA ( AJ400841 ) , human herpesvirus 1 strain KOSc glycoprotein B ( HSV-1 gB ) ( EF157316 ) , and neural proliferation , differentiation and control 1 ( NPDC1 ) ( NM_015392 . 3 ) were cloned in the pRK neo expression vector . Several PILRA point mutations were generated , including R72A , F76A , G78R , S80G , Q140A and S141A . The PILRA variants were incorporated into a full-length G78 ( AD risk ) PILRA construct by site-directed mutagenesis as per the manufacturer’s recommendation ( Agilent Cat . No . 200523 ) and sequences were verified . A full length myc-DDK tagged PIANP construct was purchased from Origene ( Cat . No . RC207868 ) . Full length complement component 4A ( Rodgers blood group ) C4A ( NM_007293 . 2 ) , extra cellular domain ( ECD ) of amyloid beta precursor like protein 1 ( APLP1 ) ( NM_005166 ) ( 1–580 aa ) and ECD of sortilin-related VPS10 domain-containing receptor 1 ( SORCS1 ) ( NM_052918 ) ( 1–1102 aa ) were fused with C-terminal gD tag ( US6/gD , partial [Human alphaherpesvirus 1 ) ( AAP32019 . 1 ) and GPI anchor in pRK vector . The ECD of all PILRA variants ( 1–196 aa ) and NPDC1 ( 1–190 aa ) were PCR amplified and cloned with C-terminal murine IgG2a Fc tag in a pRK expression vector . ECDs of PILRA variants ( G78 ( AD risk ) , R72A , F76A , G78R , S80G , Q140A and S141A ) and NPDC1 fused to the Fc region of murine IgG2a were expressed in a CHO cell expression system , supernatants collected , protein A/G affinity-purified and verified by SDS-PAGE and mass spectroscopy . 293T cells were transfected with lipofectamine LTX reagent ( ThermoFisher ) with various full-length constructs of PILRA variants ( G78 ( AD risk ) , R72A , F76A , G78R , S80G , Q140A and S141A ) . After 48 hours , the transfected cells were harvested and incubated with soluble mIgG2a-tagged ligand , NPDC1-mFc at 50 μg/ml ( as described above ) for 30 minutes on ice . Cells were then washed and stained with 1 μg/ml chimeric anti-PILRA antibody ( mouse Fc region is substituted to human IgG1 backbone on anti-PILRA antibodies [31] ) on ice for 30 min followed by APC-conjugated mouse anti-human IgG ( BD Pharmingen Cat . No . 550931 ) and FITC anti-mouse IgG2a ( BD Pharmingen Cat . No . 553390 ) secondary antibodies according to manufacturer’s instruction . PILRA-transfected 293T cells were examined by flow cytometry for binding of NPDC1 by measuring the frequency of APC and FITC double-positive cells . Double positive cells were gated on the WT sample and than the gates were overlaid on subsequent samples to maintain the same cell population throughout the experiment . For each PILRA variant , the mean percentage of the number of cells binding to NPDC1-mFC relative to the wild type PILRA binding for each experiment was calculated . In the inverse experiment , 293T cells were transfected with lipofectamine LTX reagent [ThermoFisher] with known full-length PILRA ligand ( NPDC1 , HSV-1gB and PIANP ) and predicted ligand constructs ( SORCS1 , APLP1 and C4A ) ( described above ) . After 48 hours , the transfected cells were harvested and incubated with soluble mIgG2a-tagged variants of PILRA ( G78 ( AD risk ) , R72A , F76A , G78R , S80G ) ( described above ) 50 μg/ml for 30 min on ice . Cells were then washed and stained with FITC anti-mouse IgG2a ( BD Pharmingen Cat . No . 553390 ) secondary antibody according to manufacturer’s instruction . PILRA ligand-transfected 293T cells were examined by flow cytometry for binding to PILRA variants by measuring the frequency of FITC-positive cells . The percentage of mean fluorescence intensity ( MFI ) of PILRA-mFC binding on ligand-transfected cells relative to the wild type PILRA binding for each experiment was calculated . Binding of human NPDC1 . Fc to PILRa-Fc variants was measured by SPR using a ProteOn XPR36 ( Bio-Rad ) . PILRA-Fc WT and variants ( G78R and Q140A ) were immobilized on a ProteOn GLC sensor chip ( Bio-Rad ) by EDC/NHS amine coupling ( 2000–2400 RU’s ) and the chip surface was deactivated by ethanolamine after immobilization . NPDC1-Fc diluted in PBST or a control Fc-tagged protein was injected at a concentration of 100 nM over the immobilized PILRA proteins at room temperature[31] . Healthy human volunteers from the Genentech Genotype and Phenotype program ( gGAP ) were genotyped for rs1859788 ( PILRA G78R ) using custom design ABI SNP genotyping assay with the following primers; Forward primer seq: GCGGCCTTGTGCTGTAGAA , Reverse primer seq: GCTCCCGACGTGAGAATATCC , Reporter 1 sequence: VIC- ACTTCCACGGGCAGTC-NFQ , Reporter 2 sequence: FAM- ACTTCCACAGGCAGTC-NFQ . To control for a possible effect of the eQTL for PILRB , all volunteers selected were homozygous AA ( lower PILRB expression ) for rs6955367 ( http://biorxiv . org/content/early/2016/09/09/074450 ) . Genotype for rs6955367 was determined using an InfiniumOmni2 . 5Exome-8v1-2_A . bpm . Peripheral Blood Mononuclear Cells ( PBMC’s ) were obtained by Ficol gradient from five pairs of homozygous donors for rs1859788 ( one with each genotype AA/GG ) . The pairs of samples were matched for age [± 5 years] , gender and self-reported ethnicity . Monocytes were purified from PBMC’s by negative selection using the EasySep Human Monocyte Enrichment Kit without CD16 Depletion ( 19058 ) , as recommended by the manufacturer . Isolated monocytes were differentiated into macrophages in DMEM + 10%FBS + 1X glutaMax and 100 ng/ml MCSF media for 7–10 days . The gGAP program was reviewed and approved by the Western Regional Institutional Board . Macrophages differentiated from healthy human monocytes were incubated with 10 , 1 , 0 . 1 and 0 . 01 multiplicity of infection ( MOI ) of HSV-1 virus at 37°C for 1 hour with gentle swirling to allow virus adsorption . Cells were washed after 1 hr of adsorption and infection was continued for 6 , 18 and 36 hrs . Supernatant was harvested at 6 , 18 and 36 hrs of infection and cell debris were removed by centrifugation at 3000 rpm for 5 min at 4°C . DNA was isolated from infected cells using the QIAamp DNA mini-kit ( Qiagen Cat . No . 51304 ) . Additional cells were fixed with 4% paraformaldehyde after infection and stained with DAPI for microscopy . The CytoTox 96 Non-Radioactive Cytotoxicity Assay ( Promega Cat . No . E1780 ) was performed on supernatant harvested from HSV-1-infected human macrophages as per manufacturer’s recommendations to measure cell toxicity after HSV-1 infection . For each sample , the percent cytotoxicity was calculated as the ratio of LDH released in culture supernatant after infection to completely lysed cells ( maximum LDH release ) . HSV-1 DNA was quantitated using a custom design ABI TaqMan gene expression assay , with the following primers: Forward primer seq: 5'-GGCCTGGCTATCCGGAGA-3' , Reverse primer seq: 5'-GCGCAGAGACATCGCGA-3' , HSV-1 probe: 5'-FAM-CAGCACACGACTTGGCGTTCTGTGT-MGB-3' . GAPDH DNA was quantitated using ABI endogenous control ( Applied Biosystem Cat . No . 4352934E ) . Amplification reactions were carried out with 5 μl of extracted DNA from infected cells in a final volume of 25 μl with TaqMan Universal PCR Master Mix ( Applied Biosystems Cat . No . 4304437 ) as per manufacturer’s recommendations . HSV-1 DNA ( Ct values ) was normalized to cell GAPDH ( Ct values ) to account for cell number . Virus titers from HSV-1-infected cells were determined following a standard plaque assay protocol [64] . In brief , the plaque assay was performed using Vero cells ( African Green Monkey Cells ) seeded at 1x105 cells per well in 48-well plates . After overnight incubation at 37°C , the monolayer was ~90–100% confluent . Supernatants harvested from HSV-1-infected human macrophages were clarified from cells and debris by centrifugation at 3000 rpm for 5 minutes at 4°C . Virus-containing supernatants were then diluted from 10−1 to 10−8 in DMEM ( 1 ml total volume ) . Growth media was removed from Vero cells and 250 μl of supernatant dilution was transferred onto the cells , followed by incubation at 37°C for 2 hrs with gentle swirling every 30 min to allow virus adsorption , after which the virus-containing media was aspirated . The cells were then overlaid with 2% methylcellulose containing 2X DMEM and 5% FBS and incubated at 37°C . 48 hrs post-infection , plaques were enumerated from each dilution . Virus titers were calculated in pfu/ml . | Alzheimer’s disease ( AD ) is a devastating neurodegenerative disorder resulting from a complex interaction of environmental and genetic risk factors . Despite considerable progress in defining the genetic component of AD risk , understanding the biology of common variant associations is a challenge . We find that PILRA G78R ( rs1859788 ) is the likely AD risk variant from the 7q21 locus ( rs1476679 ) and PILRA G78R reduces PILRA endogenous and exogenous ligand binding . Our study highlights a new immune signaling axis in AD and suggests a role for exogenous ligands ( HSV-1 ) . Further , we have identified that reduced function of a negative regulator of microglia and neutrophils is protective from AD risk , providing a new candidate therapeutic target . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"cell",
"binding",
"flow",
"cytometry",
"cell",
"physiology",
"blood",
"cells",
"medicine",
"and",
"health",
"sciences",
"chemical",
"characterization",
"immune",
"cells",
"neurodegenerative",
"diseases",
"293t",
"cells",
"biological",
"cultures",
"immunology",
"alzheim... | 2018 | Paired Immunoglobulin-like Type 2 Receptor Alpha G78R variant alters ligand binding and confers protection to Alzheimer's disease |
The intracellular bacterial pathogen Shigella infects and spreads through the human intestinal epithelium . Effector proteins delivered by Shigella into cells promote infection by modulating diverse host functions . We demonstrate that the effector protein OspB interacts directly with the scaffolding protein IQGAP1 , and that the absence of either OspB or IQGAP1 during infection leads to larger areas of S . flexneri spread through cell monolayers . We show that the effect on the area of bacterial spread is due to OspB triggering increased cell proliferation at the periphery of infected foci , thereby replacing some of the cells that die within infected foci and restricting the area of bacterial spread . We demonstrate that OspB enhancement of cell proliferation results from activation of mTORC1 , a master regulator of cell growth , and is blocked by the mTORC1-specific inhibitor rapamycin . OspB activation of mTORC1 , and its effects on cell proliferation and bacterial spread , depends on IQGAP1 . Our results identify OspB as a regulator of mTORC1 and mTORC1-dependent cell proliferation early during S . flexneri infection and establish a role for IQGAP1 in mTORC1 signaling . They also raise the possibility that IQGAP1 serves as a scaffold for the assembly of an OspB-mTORC1 signaling complex .
Shigella spp . cause diarrhea and dysentery in humans by invading and spreading through the colonic mucosa . Bacterial invasion of cells , intracellular survival , and aspects of intercellular spread are mediated by bacterial effector proteins delivered into the cell cytoplasm by the Shigella type 3 secretion system . Effector proteins interact with host factors to alter cellular processes or cellular signaling cascades in ways that promote infection . Shigella infection leads to an acute inflammatory response and abscess formation in the colonic mucosa that is accompanied by death of macrophages , leukocytes , and enterocytes [1–7] . Despite this destruction , bacterial replication within the tissue depends in part on the viability of infected cells . Certain Shigella effector proteins promote cell survival . IpgD activates the Akt survival pathway , which delays host cell apoptosis and is associated with an increase in intracellular bacterial replication [8] . OspC3 binds and inhibits caspase-4 , which blocks inflammatory cell death [9] . VirA inhibits both necrotic cell death and autophagy [1 , 10] . The cellular scaffolding protein IQGAP1 participates in the manipulation of the cytoskeleton by Salmonella Typhimurium and enteropathogenic E . coli [11–13] . Here , we demonstrate that IQGAP1 restricts the extent of spread of S . flexneri in cell monolayers and interacts with the Shigella effector protein OspB . OspB has been shown previously to modulate NF-κB activation and phosphorylation of ERK1/2 and activation of cytosolic phospholipase A2 and associated IL-8 secretion and transepithelial polymorphonuclear leukocyte migration [14–16] . We show that like IQGAP1 , OspB restricts the extent of S . flexneri spread in cell monolayers . Early during S . flexneri infection , OspB activates the mechanistic target of rapamycin complex 1 ( mTORC1 ) , a central regulator of cell growth and proliferation known to bind IQGAP1 [17 , 18] . OspB activation of mTORC1 results in increased cell proliferation , dependent on IQGAP1 . Increased cell proliferation occurs differentially at infected foci within cell monolayers . These results identify and characterize a targeted mechanism by which Shigella manipulates host cell proliferation during infection .
IQGAP1 was selected from a pilot siRNA screen designed to identify human proteins that modulate Shigella spread . In this screen , siRNA to IQGAP1 was associated with an increase in the area of wild type S . flexneri strain 2457T spread through HeLa cell monolayers ( IQGAP1 siRNA , 1200 ± 182 a . u . versus control siRNA , 596 ± 42 a . u . , p = 0 . 04 , Student’s two-tailed t test ) , determined by measuring the area of GFP-producing bacteria at individual infectious foci within the monolayer in 384-well format . The impact of IQGAP1 siRNA on area of bacterial spread was validated in 6-well format , where siRNA knock-down of IQGAP1 led to a 1 . 8-fold increase in area of spread of S . flexneri ( IQGAP1 siRNA , 12 ± 1 . 4 versus control siRNA , 6 ± 0 . 4 a . u . , p = 0 . 03 , S1 Fig ) . Upon independently examining the role of IQGAP1 in S . flexneri spread using monolayers of mouse embryonic fibroblasts ( MEFs ) that lack or contain IQGAP1 , we observed a similar 1 . 7-fold increase in area of spread in the absence of IQGAP1 ( Fig 1A ) , together suggesting that IQGAP1 might restrict the extent of bacterial spread . Complementation with Myc-IQGAP1 in trans significantly reduced the area of spread for IQGAP1-/- MEFs ( Fig 1A ) , indicating that the observed increase in spread in the IQGAP1-/- MEFs was due to the absence of IQGAP1 per se . We interrogated whether IQGAP1 might interact with any of 26 effector proteins translocated by the Shigella type 3 secretion system . GST-IQGAP1 pulled down OspB-FLAG from the culture supernatant of S . flexneri that expressed and secreted it ( Fig 2A ) . Pull down by GST-IQGAP1 was specific , as other secreted proteins , including the effector IpgD and the secreted translocon protein IpaC , were not pulled down ( Fig 2A ) , and was independent of other mammalian proteins , as GST-IQGAP1 used in these experiments had been purified from E . coli . To test whether the interaction between OspB and IQGAP1 is dependent on other S . flexneri proteins , we performed co-precipitation analysis after purifying both proteins from E . coli . GST-IQGAP1 , but not GST alone , pulled down His-tagged OspB ( Fig 2B ) . Moreover , GST-OspB pulled down purified IQGAP1 ( Fig 2C ) , each also purified from E . coli . OspB also interacts with endogenous IQGAP1 produced in mammalian cells . GST-OspB precipitated IQGAP1 from lysates of MCF-7 human breast epithelial cells , whereas only a faint band was pulled down with GST alone ( Fig 2D ) . Together , these data indicate that OspB binds IQGAP1 specifically and independently of other S . flexneri or mammalian proteins . Like many scaffolding proteins , IQGAP1 is a large multidomain protein with multiple binding partners ( reviewed in [19] ) . The N-terminal calponin homology domain ( CHD ) interacts with F-actin , leading to actin filament bundling and crosslinking in the cell cortex; the polyproline domain ( WW ) interacts with ERK1 , ERK2 , and together with upstream sequences , with mTOR; the IQ region interacts with calmodulin , MEK1 , MEK2 , EGFR , myosin essential light chain , and S100B; the RAS GTPase-activating protein related domain ( GRD ) interacts with Rac1 and Cdc42; the C-terminal RasGAP domain interacts with βeta-catenin , E-cadherin , CLIP-170 , and APC; and the C-terminal Dia1 binding region interacts with Dia1 ( reviewed in [19] ) . To identify the region of IQGAP1 required for interaction with OspB , we generated defined fragments of IQGAP1 in reticulocyte lysates and tested which were pulled down with GST-OspB ( Fig 2E and 2F ) . GST-OspB precipitated the IQ region ( residues 717–916 ) and amino-terminal half ( residues 2–864 ) of IQGAP1 , as well as full-length IQGAP1 , but not the C-terminal half of the protein ( residues 864–1657 ) or the fragment consisting of residues 2–764 , which lacks the IQ region ( Fig 2F ) . These results indicate that the IQ region , which lies between residues 750 and 865 , is necessary and sufficient for interactions with OspB . The observed interaction between IQGAP1 and OspB raised the possibility that OspB might function in IQGAP1 limitation of S . flexneri spread . In IQGAP1+/+ MEFs , an ospB mutant displayed significantly increased area of spread compared to wild type S . flexneri , which was rescued by complementation with ospB in trans ( Fig 1B , black bars ) . These differences were absent from IQGAP1-/- MEFs ( Fig 1B , gray bars ) , indicating that OspB function in spread depends on IQGAP1 . These findings were recapitulated when OspB was provided to cells in trans . In the presence of IQGAP1 , OspB decreased the area of bacterial spread , independently of whether it was delivered by the bacterium or by transient transfection ( Fig 1C , black bars ) , whereas in the absence of IQGAP1 , the effect of OspB on spread was muted ( Fig 1C , gray bars ) . We used transient transfection for these experiments because when we attempted to generate cell lines that stably expressed OspB , we found that in the presence of IQGAP1 , but not in its absence , introduction of OspB precluded the isolation of stably transfected cells , whereas introduction of vector alone did not . This raised the possibility that the combination of IQGAP1 and OspB perturbed essential cellular pathways . Of note , the efficiency of transient transfection was similar under all conditions . Zurawski et al . [14–16] demonstrated that a S . flexneri ospB mutant is no difference from WT S . flexneri in invasion , apoptosis , or the in vivo Sereny assay , but is defective in transepithelial polymorphonuclear leukocyte migration . It is notable that these authors did not observe an effect of OspB on plaque size; we postulate that this difference with our results may be because Zurawski et al . performed plaque assays in HeLa cells , in which we also do not observe an OspB-associated phenotype ( described below ) . As Shigella spread through monolayers occurs by actin based motility [20] , we compared several parameters relating to this process during infection of IQGAP1-/- versus IQGAP1+/+ MEFs , but found only small differences that did not appear to fully explain the observed differences in spread ( S1 Table ) . Given that IQGAP1 plays a role in maintaining the cytoskeleton and thus cell shape and size , it was conceivable that the increase in S . flexneri spread through cell monolayers observed in the absence of IQGAP1 was simply due to differences in cell size . Alternatively , in the absence of IQGAP1 , the dense network of bundled and cross-linked F-actin at the cell cortex might be reduced , which might facilitate S . flexneri movement into adjacent cells . However , visualization of the actin cytoskeleton and measurements of cell size and cortical actin density showed no difference between IQGAP1+/+ and IQGAP1-/- MEFs ( S2A Fig and S1 Table ) . Intercellular spread of Shigella is enhanced by cadherin-cadherin intercellular junctions [21] . Moreover , in polarized Madin-Darby canine kidney II epithelial cells , suppression of IQGAP1 is associated with reduced E-cadherin at intercellular junctions [22] . However , unlike epithelial cells , in MEFs , the classical cadherin that is expressed is N-cadherin [23] . By detecting all cadherins with a pan-cadherin antibody , we found that the distribution and signal intensity of cadherins in the IQGAP1-/- MEFs was indistinguishable from that in IQGAP1+/+ MEFs ( S2B Fig ) , suggesting that the observed differences in S . flexneri area of spread are unlikely to be due to effects of IQGAP1 on cadherin intercellular junctions and that the impact of IQGAP1 on the levels of N-cadherin at intercellular junctions in MEFs is distinct from its impact on levels of E-cadherin at intercellular junctions in epithelial cells . It remains possible that a part of the IQGAP1 phenotype during S . flexneri infection is due to functions of IQGAP1 that are unrelated to OspB . We considered whether the effect of OspB on the area of S . flexneri spread could be due to effects on cell proliferation . Infection of cells by Shigella species results in cell death , which is seen as cellular debris at the center of infectious foci . We reasoned that if cells containing IQGAP1 and infected with S . flexneri expressing OspB proliferated at an increased rate , then these cells might proliferate sufficiently quickly to replace some of the dying cells within foci of infection , enabling bacteria to spread into these new cells and causing the net area of bacterial spread to be smaller . To assess whether increased cell proliferation per se would lead to smaller net area of bacterial spread , we tested the effect of insulin-like growth factor on the area of bacterial spread . The addition of insulin-like growth factor to IQGAP1+/+ MEFs led to a 1 . 3 ± 0 . 1-fold increase ( p = 0 . 03 ) in the number of cells over 24 hr . Insulin-like growth factor-induced increase in cell proliferation was associated with a 21% decrease in spread of the ospB mutant and an 18% decrease in spread of the ospB mutant complemented with OspB ( Fig 3A ) , indicating that increased cell proliferation in the monolayer is associated with decreased area of S . flexneri spread . To test whether the presence of OspB was associated with increases in cell proliferation in the vicinity of plaques , we imaged random plaques and random areas away from plaques formed by S . flexneri producing or not producing OspB at 24 hrs of infection , then blindly comparing the number of nuclei of intact cells within pre-defined grids placed randomly at the edges of infected foci to those within pre-defined grids placed randomly on uninfected areas of the monolayers ( see Methods and S3 Fig ) . The density of cells at the edges of infected foci ( normalized to the density of cells in uninfected areas ) was significantly more for the OspB-producing strain than for the ospB mutant strain ( p = 0 . 03 , Fig 3B ) . Based on these results , we postulate that the observed increase in cell density is sufficient to explain the observed decrease in area of spread . To more directly measure the effect of OspB on cell proliferation , the rate of proliferation of non-confluent MEFs transiently transfected with OspB was determined . OspB was associated with significantly increased cell proliferation over 24–72 hrs following plating of the cells , with a slight trend apparent as early as 12 hrs , and this increase depended on IQGAP1 ( Fig 3C ) . The efficiency of cell adhesion to the substrate was indistinguishable among all conditions , and the amount of cell death was low and similar for all conditions ( Fig 3C and 3E , and S4A Fig ) . A second measure of cell proliferation , the density at which cells attain growth saturation , was increased 1 . 8-fold ( p<0 . 05 ) in MEFs transiently transfected with OspB ( Fig 3D ) . Again , this increase depended on IQGAP1 . Together , these results establish that OspB enhances cell proliferation in a manner that depends on IQGAP1 . A central regulator of cell growth and proliferation is the kinase mechanistic target of rapamycin ( mTOR ) . mTOR is a component of two protein complexes , mTOR complex 1 ( mTORC1 ) and mTOR complex 2 ( mTORC2 ) , that coordinate cell growth functions in response to growth factors and nutrient availability . Moreover , mTOR interacts with IQGAP1 [17] . Rapamycin , a specific inhibitor of mTOR , completely abolished the increase in proliferation observed in OspB-transfected IQGAP1+/+ cells , whereas it had no significant effect on proliferation of cells lacking either OspB or IQGAP1 ( Fig 3E and 3F and S4B and S4C Fig ) . This effect of rapamycin strongly suggested that OspB enhancement of cell proliferation depends on mTOR . We tested whether OspB activated mTOR by examining the phosphorylation state of S6 kinase 1 ( S6K ) , a substrate of mTOR kinase activity that controls cap-dependent translation elongation . Phosphorylation of S6K on Thr-389 was increased 2-fold ( p<0 . 05 ) in the presence of OspB and in a manner that depended on IQGAP1 ( Fig 4A and 4B ) . Phosphorylation of S6K by OspB was inhibited by rapamycin ( Fig 4C ) . Together , these data establish that OspB activates mTOR . In IQGAP1+/+ MEFs , OspB appeared to sensitize cells to rapamycin . In the presence of OspB , rapamycin dramatically inhibited proliferation of cells , whereas in the absence of OspB , it had no effect ( Fig 3E ) . Moreover , suppression of phosphorylation of the mTOR substrate S6K by rapamycin treatment was enhanced by the presence of OspB ( Fig 4C ) . Detailed analysis of this observation will be explored in future experiments . Since rapamycin binds and inhibits mTORC1 , but not mTORC2 [24 , 25] , our findings suggested that OspB activates mTORC1 and not mTORC2 . To further explore this possibility , we examined phosphorylation of Akt at Thr-308 , which is inhibited by mTORC1 , but not by mTORC2 . In the presence of IQGAP1 , OspB was associated with a 3-fold decrease in phosphorylation at Akt Thr-308 ( Fig 4D and 4E , p<0 . 05 ) . Activation of mTORC2 increases phosphorylation of Akt at Ser-473 [26] , which was unaltered by OspB ( S5A Fig ) . Together , these results are consistent with OspB activating mTORC1 and not mTORC2 . Under the conditions used , GST-OspB did not precipitate mTOR from MCF-7 cell lysates . Akt is itself an upstream regulator of mTORC1 activity and a binding partner of IQGAP1 [27] . This raised the possibility that , even though Akt Thr-308 phosphorylation was decreased by OspB ( Fig 4D ) , OspB might activate mTORC1 via Akt . To directly test this , we examined activation of mTORC1 in the presence of the phosphatidylinositol 3-kinase inhibitor LY294002 , which inhibits Akt activation . In the presence of LY , infection with wild-type S . flexneri induced phosphorylation of S6K , yet phosphorylation of Akt was inhibited ( S5B Fig ) , indicating that mTORC1 is activated independently of Akt . To test the relevance of OspB activation of mTOR during S . flexneri infection , we compared S6K phosphorylation in cells infected with an ospB mutant to that in cells infected with WT S . flexneri or the complemented ospB mutant . At 1 hr of infection of IQGAP1+/+ MEFs , WT or ospB complemented S . flexneri induced more phosphorylation of S6K than the ospB mutant ( Fig 5A ) , indicating that OspB delivered by S . flexneri activates mTOR early during infection . Activation of mTOR during infection depended on IQGAP1 , since S6K phosphorylation remained at baseline in IQGAP1-/- MEFs . For cells that were exposed to the non-invasive S . flexneri mutant BS103 , S6K phosphorylation was lower than baseline , suggesting that the presence of extracellular bacteria might suppress mTOR activation , perhaps by reducing the concentration of amino acids in the extracellular media [28] . Treatment of cells with rapamycin inhibited OspB+ S . flexneri-induced phosphorylation of S6K ( Fig 5B ) , indicating that S6K phosphorylation was mediated by mTOR kinase . Rapamycin had no effect on bacterial growth in vitro ( S6A Fig ) . Tattoli et al . described that S . flexneri inhibits mTOR at 4 hrs of infection via amino acid starvation [29]; these authors did not observe the initial activation of mTOR at 1 hr of infection that we describe here . To explore these differences , we compared mTOR activation at 1 and 4 hrs . of infection of HeLa cells , which were used by Tattoli et al . , as well as of the intestinal epithelial cell line Caco2 and of MEFs ( S6B Fig ) . Consistent with the findings of Tattoli et al . , at 4 hrs , mTOR signaling in each of the three cell lines was inhibited by S . flexneri ( S6A Fig ) . At 1 hr , mTOR activation was observed in Caco2 cells and MEFs , but was not apparent in HeLa cells . The absence of initial activation in HeLa cells appeared to be due to a high level of mTOR activation at baseline ( i . e . , in uninfected cells ) that was not present in the other cell lines . This baseline level of mTOR activation might result from the genetic background of this cancer cell line or have been acquired subsequent to its initial isolation . Taken together , these findings suggest that in cells that display at baseline a low level of mTOR activity , S . flexneri infection induces dynamic changes of mTOR activation . To test whether the increase in S . flexneri area of spread that we observed in cells lacking IQGAP1 or with strains lacking ospB ( Fig 1 ) might be due to decreased mTOR activation , we measured spread in the presence or absence of rapamycin . The presence of rapamycin was associated with an increase in the area of spread of wild-type bacteria to that of the ospB mutant ( Fig 5C ) . Rapamycin had no effect on the spread of the ospB mutant , indicating that a rapamycin sensitive pathway is responsible for the effect of ospB on spread . Moreover , the effect of rapamycin on the spreading ability of the wild type strain was dependent on IQGAP1 . These results are consistent with IQGAP1-dependent activation of mTOR by OspB causing changes in cell proliferation and growth that result in a decrease in area of spread of S . flexneri through cell monolayers .
The mTOR complexes are protein kinases that serve as master regulators and coordinators of cellular growth and metabolism in response to extracellular and intracellular growth factors and nutrient states . Deregulation of the mTOR signaling pathway plays a central role in the transformation and the uncontrolled growth of numerous cancers , as well as roles in diabetes , obesity , neurodegeneration , and aging [30] . In this study , we demonstrate that the Shigella type 3 secreted effector protein OspB activates mTOR activity during cellular infection . Bacterial type 3 secreted effectors alter diverse cellular pathways; to our knowledge , this is the first description of a type 3 effector protein that activates mTOR and thereby promotes host cell survival . mTOR exists in two physically distinct multi-protein signaling complexes , mTORC1 and mTORC2 [30 , 31] . The two complexes share the core kinase mTOR , as well as DEPTOR , mLST8 , TTI1 , and TEL2 . mTORC1 is defined by also containing regulatory associated protein of mTOR ( RAPTOR ) , whereas mTORC2 instead contains rapamycin-insensitive companion of mTOR ( RICTOR ) . The two complexes are generally activated by different sets of signals , and their activation leads to distinct , yet overlapping cellular responses . OspB activation of mTOR appears to be specific for the mTORC1 complex , as OspB-mediated activation is completely inhibited by short-term treatment with the specific mTORC1 inhibitor rapamycin and is associated with decreased phosphorylation of Akt on Thr-308 ( Fig 4 ) . The kinase activity of activated mTORC1 leads to increases in cellular protein synthesis , metabolism , and growth , and to inhibition of autophagy . Indeed , activation of the Akt-mTOR pathway by Salmonella typhimurium in macrophages attenuates the autophagic response [32] . Our data support a model in which the effect of IQGAP1 and OspB on the area of S . flexneri spread through cell monolayers is due to effects on cell proliferation ( S7 Fig ) . Infection of cells by Shigella species results in cell death , which is seen as cellular debris at the center of infectious foci . We propose that soon after initial entry into a cell ( either from the extracellular milieu or via bacterial spread from an infected adjacent cell ) , OspB activates mTORC1 , thereby triggering a cellular pathway that leads to increased cell proliferation in the infected cell and possibly in adjacent cells . Cellular proliferation is sufficient to replace some of the dying cells within foci of infection and results in increased numbers of viable cells within the area of infection and increased density of cells within the focus of infection . Since the length of time the bacterium spends spreading from one cell into the adjacent cell ( the time within the protrusion and the new vacuole ) is substantial—similar to the length of time the bacterium spends in the cell cytosol prior to protrusion formation [33]—the presence of increased numbers of cells within the immediate area of infection slows the net lateral movement of the bacteria through the monolayer . The net effect is that the area of bacterial spread within the monolayer is reduced compared with conditions that lack OspB or IQGAP1 . Multiple findings presented here support this model: ( 1 ) OspB activation of mTORC1 is associated with increased cell proliferation ( Fig 3 ) . ( 2 ) Growth factor-induced increases in cell proliferation are associated with decreased area of S . flexneri spread ( Fig 3 ) . ( 3 ) The area of S . flexneri spread is decreased by the presence of OspB and IQGAP1 ( Fig 1 ) and is increased by the addition of rapamycin in a manner that depends on OspB and IQGAP1 ( Fig 5C ) , indicating that the observed decreases in area of spread are due to IQGAP1-dependent OspB activation of mTORC1 . ( 4 ) OspB induces an increase in cell density at the edges of infected foci ( Fig 3B ) , indicating that the effect of mTORC1 activation on cell proliferation in infected monolayers is observed locally within infected foci . Increased local cellular proliferation may serve to provide additional protective intracellular niches for the organism in areas of cell death within infected tissue . These findings do not exclude the possibility that another effect of mTORC1 activation on cellular function contributes to the effect of OspB and IQGAP1 on area of spread or that mTORC1 activation has other beneficial effects on Shigella infection . A related consideration is that OspB activates NF-κB , a critical cell survival signal [16] . Since the effect of OspB on S . flexneri spread was completely blocked by rapamycin ( Fig 5C ) , the spread phenotype appears to be primarily or exclusively due to mTORC1 activation , rather than NF-κB induced cell survival . Also consistent with this is our observation that OspB had no effect on cell survival per se ( S4A Fig ) . Activation of mTORC1 is observed at 1 hr of infection , with a subsequent decrease in SK6 phosphorylation by 4 hrs of infection ( S5 Fig and [29] ) . Despite the transient nature of S6K phosphorylation , increases in cell proliferation and cell density are observed as early as 24 hrs ( Fig 3C ) . A possible explanation for these findings is that transient activation of mTORC1 is sufficient to trigger a cell proliferation pathway that is sufficiently active to significantly impact the number of cells . Alternatively , mTORC1 activation may be bimodal within the infected cells in the monolayer , dependent on the length of time individual cells have been infected , i . e . , in cells that were infected recently , due either to bacteria entering from the extracellular milieu or to bacterial spread from infected adjacent cells , mTORC1 would be activated , and in cells that have been infected for 4 hrs or longer , mTOR would be inhibited . At any given time , the level of mTORC1 activation in the monolayer as a whole would reflect the mix of all cells , yet proliferation would be activated in all recently infected cells . Consistent with this possibility is our observation that OspB induces increased cell density locally within infected monolayers ( Fig 3A and 3B ) . OspB activation of mTORC1 is enhanced by the scaffolding protein IQGAP1 ( Fig 4 ) , and OspB interacts directly with IQGAP1 independent of other S . flexneri or mammalian proteins ( Fig 2 ) . IQGAP1 links components of multiple cellular signaling pathways [34] . Our findings raise the possibility that , in addition to its previously described roles , IQGAP1 serves as a scaffold for OspB activation of the mTORC1 signaling complex . Our data suggest that the IQ motif of IQGAP1 is required and sufficient for its interaction with OspB ( Fig 2E and 2F ) ; the IQ motif is immediately C-terminal to the WW region of IQGAP1 implicated in interactions with mTOR [17] , raising the possibility that OspB and mTOR are in close proximity when bound to IQGAP1 ( S7 Fig ) . Among known activators of mTORC1 is the MEK/ERK MAP kinase signaling pathway , as ERK inactivates the TSC complex , which inhibits mTORC1 [35] . IQGAP1 is known to bind directly to MEK1/2 and ERK1/2 , modulating their activation [36–38] , and to interact with mTOR [17 , 18] . Of note , the Shigella type 3 secreted effector OspF irreversibly inactivates MAP kinases , including ERK1/2 [39] . However , other data suggest that during infection , OspB is required for maximal phosphorylation of ERK1/2 [14 , 15] , raising the possibility that OspB activation of mTORC1 might occur via ERK1/2 . How OspB activation of ERK1/2 and subsequent ERK1/2-dependent activation of mTORC1 might be reconciled with OspF inactivation of ERK1/2 is at present unclear . Whereas OspB restriction of S . flexneri spread is markedly enhanced by IQGAP1 ( Fig 1 ) , in the absence of IQGAP1 , when present both in trans and delivered by bacterial type 3 secretion , OspB has an effect on bacterial spread ( Fig 1C , left ) . This is mirrored by a slight trend , albeit statistically insignificant , of OspB increasing cell proliferation via mTORC1 in the absence of IQGAP1 ( S4B and S4C Fig , OspB GFP plus DMSO versus OspB GFP plus rapamycin ) . Taken together , these data suggest that when high levels of OspB are present , a small effect on mTORC1 may occur even in the absence of IQGAP1 . This is consistent with our model that IQGAP1 is a platform for OspB activation of mTORC1 . In this model , when OspB is present in low concentrations , IQGAP1 is critical for promoting OspB activity , perhaps by localizing it optimally and/or bringing it into close proximity to its target; in contrast , when OspB is abundant , it can interact with its target even if not bound to IQGAP1 . Although additional work will be required to determine the mechanism of OspB activation of mTORC1 , our data provide additional evidence in support of a scaffolding role for IQGAP1 upstream of mTORC1 signaling . Roles for IQGAP1 as a scaffold in signaling by bacterial pathogens have been proposed previously [11 , 12 , 40] . Together with our findings that Shigella OspB depends on IQGAP1 for activation of mTOR , we speculate that IQGAP1 has been hijacked as a signaling scaffold by multiple pathogens . OspB activation of mTORC1 was completely blocked by rapamycin . Rapamycin , in complex with FK506-binding protein of 12 kDa ( FKBP12 ) , inhibits mTORC1 by binding to the FKB12-rapamycin-binding ( FRB ) domain of mTOR [41] . Recent structural studies suggest that this interaction reduces accessibility of substrates to the catalytic cleft of mTOR [42] . Our observation that OspB appears to sensitize cells to the inhibitory effect of rapamycin ( Figs 3E and 4C ) raises the possibility that OspB promotes the interaction or the specific activity of rapamycin with respect to mTORC1; ongoing studies are investigating these possibilities .
Wild-type S . flexneri serotype 2a strain 2457T [43] , an isogenic ospB deletion mutant strain ( gift from C . Lesser ) , and an isogenic non-invasive strain , BS103 , which is cured of the virulence plasmid [44] , were grown in tryptic soy broth from individual colonies that were red on agar containing Congo red . IQGAP1 expression plasmids have been described [45] . For bacterial expression , ospB was cloned under the control of the native promoter in pACYC184 , and for mammalian expression , into pcDNA3 as a transcriptional fusion to gfp . Each of 26 S . flexneri type III secreted effector proteins and the two secreted translocon pore proteins IpaB and IpaC were tagged at the C-terminus with FLAG and expressed under the control of the impaired Tac promoter in pDSW206 [46] , as described [47] . The coding sequence for ospB::flag was amplified by PCR as a BamHI-EcoRI fragment and inserted into pGEX2T or pRSET-A at BamHI and EcoRI sites . The sequences of all plasmids were verified by DNA sequencing . HeLa cells ( ATCC ) , Caco2 ( ATCC ) , and MCF-7 ( gift of A . Dutta ) cells were maintained in Dulbecco’s modified media ( DMEM ) supplemented with fetal bovine serum ( 10% vol/vol ) . IQGAP1-/- and IQGAP1+/+ MEFs had been previously isolated from IQGAP1-/- and littermate control IQGAP1+/+ mice as described [38]; all MEFs used for experiments were early passage . Cells , seeded in 6-well plates , were transfected with 1 . 5–5 . 0 ug plasmid DNA using Lipofectamine ( Invitrogen , Carlsbad , CA ) , per manufacturer’s directions . Transfected cells were infected 16–72 hr later . Transfection efficiency was determined by quantification of cells expressing GFP in transfections performed in parallel by microscopic and FACS analysis . The transfection efficiency of the OspB GFP plasmid was comparable to that of the GFP plasmid . siRNA targeting IQGAP1 ( SMARTpool M-004694-01 ) was from Dharmacon RNAi Technologies ( Thermo Fisher Scientific ) . Reverse transfection of 8 nM siRNA into HeLa cells was performed with HiPerFECT ( Qiagen ) according to the manufacturer instructions . Cell area of IQGAP1-/- and IQGAP1+/+ MEFs at 50% and 100% confluency was calculated on images of monolayers that had been fixed and stained with phalloidin using IP Lab software . The density of cortical actin was assessed on fixed and phalloidin-stained monolayers at 30% confluence by calculating the ratio of fluorescence in 4 μm x 4 μm regions of the cell cortex to that of total fluorescense of the cellular region on images captured at 225x magnification . The rate of cell proliferation was determined by measuring serum-independent growth of transiently transfected cells . Transfection with the OspB GFP or GFP construct was performed in DMEM without fetal bovine serum . The next day , the cells were trypsinized and seeded at 2x104 per well in 12-well plates in DMEM supplemented with 10% fetal bovine serum ( vol/vol ) for 4 hrs to allow the cells to attach . Thereafter , the medium was replaced with DMEM supplemented with 1% fetal bovine serum ( vol/vol ) . Time zero was defined as 4 hrs after plating; at the indicated times ( 12 hrs and 1 , 2 , and 3 days ) , the cells were washed twice with PBS , trypsinized , and counted using a hemocytometer . The saturation density of cells was determined as above , except that after the 4-hr attachment period , the medium was supplemented with 10% fetal bovine serum ( vol/vol ) , and the cells were counted once on day 3 . Cells were maintained in DMEM supplemented with 10% fetal bovine serum ( vol/vol ) . Twenty-four to 48 hrs after transfection with the OspB GFP or GFP construct , cells were washed once with PBS and harvested for western blot analysis . Where appropriate , rapamycin was added to 10 nM ( unless otherwise indicated ) for 1 hr before washing and harvesting . Where appropriate , LY294002 ( 50 μM , Cell Signaling , #9901S ) was added immediately prior to addition of bacteria . Cells seeded in 6-well plates were routinely infected at a multiplicity of infection ( MOI , bacteria:cell ) of 50–150 with exponential phase S . flexneri . The infected plates were centrifuged at 830 g for 10 min and were returned to 37°C for 1 hr . Where appropriate , rapamycin ( to 10 nM ) or DMSO carrier alone was added to the monolayers at the same time as the bacteria were added . For the analysis of actin tail formation and the formation of cell surface protrusions , after 1 hr of infection , gentamicin ( 25 μg/ml ) , which kills extracellular but not intracellular Shigella , was added to the media . The infection was allowed to proceed for an additional 1 hr , at which time , the cells were fixed in 3 . 7% paraformaldehyde in F buffer ( 5 mM PIPES pH 7 . 2 , 5 mM KCl , 137 mM NaCl , 4 mM NaHCO3 , 0 . 4 mM KH2PO4 , 1 . 1 mM Na2HPO4 , 2 mM MgCl2 , 2 mM EGTA , 5 . 5 mM glucose ) , permeabilized in 0 . 5% Triton X-100 in F buffer ( 20 min , room temperature ) , and blocked in 0 . 1 M glycine in F buffer ( 10 min , room temperature ) . Actin was stained with phalloidin , and DNA was stained with 4' , 6-diamidino-2-phenylindole ( DAPI ) . A minimum of 10 infected cells was imaged for each condition . To determine the efficiency of bacterial spread through monolayers , exponential phase S . flexneri were placed on cells seeded in 6-well plates at a multiplicity of infection ( MOI , bacteria:cell ) of 0 . 002–0 . 02 . The plates were centrifuged at 830 g for 10 min to bring the bacteria into contact with the cells and were returned to 37°C for 1 hr , at which time gentamicin ( 25 μg/ml ) , which kills extracellular but not intracellular Shigella , was added to the media . Monolayers were processed in one of two ways . For most experiments , monolayers were infected with GFP-expressing S . flexneri . These monolayers were imaged at approximately 16 hrs of infection using 4x magnification . For experiments in which rapamycin was used , the Shigella strains did not carry gfp; at 16 hrs of infection , infected monolayers were overlaid with 0 . 7% agarose in DMEM containing gentamicin and the vital dye neutral red and were then maintained at 37°C for an additional 24 hrs , at which time they were imaged on a plate scanner . For each condition , a minimum of 10–15 plaques was measured . To measure bacterial spread in the presence of growth factor , plated cells were allowed to attach for 4 hr in 10% fetal bovine serum ( vol/vol ) and were then serum starved overnight in 1% fetal bovine serum , before proceeding with the infection as described above . At 1 hr of infection , in addition to gentamicin , insulin-like growth factor ( 100 ng/ml ) was added to the media . The monolayers were imaged at 24 h of infection . Infection was performed with GFP-expressing S . flexneri essentially as described above for determining the area of bacterial spread , except that the MOI was 0 . 03 and infected cells were maintained with gentamicin , but not an agarose overlay . At 24 hr of infection , monolayers were stained with Hoechst ( 33342 , Life Technologies ) and were imaged using 4x and 10x magnification . Infectious foci were identified by GFP signal from the bacteria . The uninfected areas of the monolayer that were imaged were selected at random from within each quadrant of the well . Using iVision , 100 x 100 or 200 x 200 pixel boxes were drawn around infectious foci or the noninfectious monolayer , and the number of cells within each box was counted . Plasmids that encode FLAG-tagged type III secreted effector proteins [47] were introduced into wild type S . flexneri . Each strain was grown to exponential phase , at which time IPTG was added to 0 . 1 mM to induce the expression of the FLAG-tagged effector . Cultures were returned to 37°C for 1 hr . Cultures were pelleted and resuspended in 30 mM HEPES ( pH 7 . 7 ) , 137 mM NaCl , 10 mM Congo red , and 0 . 1mM IPTG and incubated at 37°C for 30 mins . Culture supernatants were cleared by two serial centrifugation steps and were supplemented with 1 mM DTT , 1 mM EDTA , 1x Complete , 0 . 01% Triton X-100 , and 300 mM NaCl . GST-IQGAP1 , purified as described previously [45] and bound to glutathione sepharose beads , was incubated with cleared supernatants at 4°C for 2 hrs . Beads were recovered by centrifugation at 2 , 500 g for 5 min and washed three times with wash buffer ( 30 mM HEPES , pH 7 . 7 , 300 mM NaCl , 0 . 01% Triton X-100 ) . Bound proteins were eluted off of beads by boiling in SDS sample buffer . Proteins remaining in the cleared supernatants after incubation with IQGAP1 beads were precipitated with 15% trichloroacetic acid . GST-IQGAP1 was produced in E . coli and isolated using glutathione-Sepharose chromatography essentially as previously described [45] . Where indicated , IQGAP1 was further purified by cleaving GST using tobacco etch virus [48] . GST-OspB was produced and isolated as described for GST-IQGAP1 . His-tagged OspB was purified with Ni2+-nitrilotriacetic acid agarose beads ( Qiagen ) following the manufacturer’s protocol . The size and purity of the GST- and His-tagged proteins were evaluated by SDS-PAGE and Coomassie staining . All proteins were at least 90% pure . For assays with GST-tagged proteins , 3–5 μg purified untagged IQGAP1 was pre-incubated at 4°C for 1 h with 20 μl glutathione beads in 1 ml Buffer A ( 50 mM Tris-HCl , pH 7 . 4 , 150 mM NaCl and 1% Triton X-100 ) containing 1X Protease & Phosphatase Inhibitor Cocktail ( Thermo Scientific ) and 1 mM PMSF . The beads were recovered by centrifugation , and the supernatant was transferred to tubes containing either 30 μl GST-OspB or 20 μl GST . All GST proteins were on glutathione-Sepharose beads . Samples were rotated for 3 h at 4°C , washed 5 times with Buffer A and resolved by SDS-PAGE . The gel was cut at ~100 kDa . The bottom portion was stained with Coomassie blue . The top portion was transferred to PVDF . Membranes were incubated for 1 h at 4°C with Blocking Buffer ( LI-COR ) , then probed with anti-IQGAP1 polyclonal antibodies [45] overnight at 4°C . The membrane was incubated with infrared dye-conjugated ( IRDye ) anti-rabbit antibody for 1 h and antigen-antibody complexes were detected using the Odyssey Imaging System ( LI-COR ) as described [49] . For assays with His-tagged OspB , His-OspB was pre-cleared in Buffer A containing 20 μl glutathione-Sepharose beads as described in the prior paragraph for IQGAP1 . The supernatants were transferred to tubes containing 20 μl GST-IQGAP1 on beads or 20 μl GST beads and incubated as above . Samples were processed by SDS-PAGE . The gel was cut at ~75 kDa . The top portion of the gel was stained with Coomassie Blue . The bottom portion was transferred to PVDF , probed with anti-His monoclonal antibody ( Santa Cruz Biotechnology ) , followed by IRDye-conjugated anti-mouse antibody ( LI-COR ) and imaged as outlined above . To identify the OspB binding site on IQGAP1 , IQGAP1 proteins were generated by Transcription and Translation ( TNT ) kit ( Promega ) . To prepare Transcend Biotin-Lysyl labeled IQGAP1 proteins , we mixed 40 μl TNT T7 Quick Master Mix , 1 μl 1 mM methionine , 1 μg pCDNA-IQGAP1 plasmid , 0 . 5–3 μl Transcend Biotin-Lysyl-tRNA , and added water to 50 μl . Samples were incubated at 30°C for 90 min . 40 μl GST-OspB or GST on glutathione-Sepharose beads was incubated with 20 μl TNT product in 1 ml Buffer B ( Buffer A containing protease inhibitors and PMSF ) and rotated at 4°C for 3 h . After 5 washes with Buffer A , samples were separated by 4–20% gradient SDS-PAGE and transferred to PVDF . Visualization was with Strep-HRP , Chemiluminescient Substrate , and exposure to X-ray film . MCF-7 human breast epithelial cells were grown to 90% confluence , washed twice with ice-cold PBS ( 155 . 6 mM NaCl , 1 mM KH2PO4 , and 2 . 9 mM Na2HPO4 , pH 7 . 4 ) , and lysed with 500 ul of Buffer B . Lysates were processed by sonication for 10 s with a Model 100 Dismembrator ( Fisher Scientific ) , and insoluble material was pelleted by centrifugation for 10 min . Rotation with glutathione-Sepharose beads for 1 h at 4°C was performed to pre-clear the supernatants . Equal amounts of protein lysate were incubated with 4 ug of GST or GST-OspB for 3 h at 4°C . After centrifugation , samples were washed five times with Buffer A and separated by SDS-PAGE . The gel was cut at 100-kDa . The bottom portion of the gel was stained with Coomassie blue . The upper portion was transferred to PVDF and probed with anti-IQGAP1 antibodies . For all other western blot analysis , proteins were separated on SDS polyacrylamide gels , and western blot analysis was carried out using standard procedures and the following antibodies: FlagM2 ( F1804 , Sigma; diluted 1:200 ) , IQGAP1 ( ab33542 , Abcam , Cambridge , MA; diluted to 1 μg/ml ) , phospho-S6 kinase ( Thr-389 ) ( 9234 , Cell Signaling Technology; diluted to 1:1000 ) , total S6 kinase ( 2708 , Cell Signaling Technology; diluted to 1:1000 ) , phospho-Akt ( Thr-308 ) ( 4056S , Cell Signaling Technology; diluted to 1:1000 ) , phospho-Akt ( Ser-473 ) ( 4060S , Cell Signaling Technology; diluted to 1:1000 ) , pan-Akt ( 4691 , Cell Signaling Technology; diluted to 1:1000 ) , peroxidase-conjugated anti-beta actin ( A3854 , Sigma; diluted 1:10 , 000 ) , and horseradish peroxidase conjugated goat anti-mouse secondary ( Jackson; diluted 1:2000 ) . Visualization was performed using SuperSignal West Pico Chemilumnescent Substrate or SuperSignal Femto Chemiluminescent Substrate ( Thermo Fisher Scientific ) , per the manufacturer’s instructions . Densitometry of bands was performed using a Bio Rad Molecular Imager Chemi Doc XRS+ Imaging System and ImageJ software . Microscopic images were collected using a Nikon Eclipse TE300 using the software IP Lab or iVision . For time lapse imaging to determine bacterial speed an image was taken every 5 sec for 15 min . Movies were compiled using Image J and the rate of bacterial movement was determined by tracking individual bacteria for 12 consecutive frames . For each experimental condition in each experiment , speeds were determined for 10 or more moving bacteria . The lengths of actin tails and bacterial protrusions were measured on still images using IP Lab . | During infection , Shigella spp . deliver into the cytoplasm of cells effector proteins that manipulate host cell processes in ways that promote infection and bacterial spread . We have discovered that the Shigella effector protein OspB interacts with the cellular scaffolding protein IQGAP1 . OspB induces increased cell proliferation by activating mTORC1 kinase , a master regulator of cellular growth , in a manner that depends on IQGAP1 . As IQGAP1 has been shown to interact with mTOR and with the mTORC1 activators ERK1/2 , we propose that IQGAP1 serves as a scaffold for OspB activation of mTORC1 . The presence of OspB and IQGAP1 lead to restricting the area of spread of S . flexneri in cell monolayers; our data support a model in which the effect of OspB and IQGAP1 on the area of S . flexneri spread is due to effects on cell proliferation locally within infected foci . As infection of cells and tissue by Shigella spp . leads to cell death , increased local cellular proliferation may serve to provide additional protective intracellular niches for the organism within infected tissue . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2015 | Shigella Effector OspB Activates mTORC1 in a Manner That Depends on IQGAP1 and Promotes Cell Proliferation |
During vertebrate craniofacial development , neural crest cells ( NCCs ) contribute to most of the craniofacial pharyngeal skeleton . Defects in NCC specification , migration and differentiation resulting in malformations in the craniofacial complex are associated with human craniofacial disorders including Treacher-Collins Syndrome , caused by mutations in TCOF1 . It has been hypothesized that perturbed ribosome biogenesis and resulting p53 mediated neuroepithelial apoptosis results in NCC hypoplasia in mouse Tcof1 mutants . However , the underlying mechanisms linking ribosome biogenesis and NCC development remain poorly understood . Here we report a new zebrafish mutant , fantome ( fan ) , which harbors a point mutation and predicted premature stop codon in zebrafish wdr43 , the ortholog to yeast UTP5 . Although wdr43 mRNA is widely expressed during early zebrafish development , and its deficiency triggers early neural , eye , heart and pharyngeal arch defects , later defects appear fairly restricted to NCC derived craniofacial cartilages . Here we show that the C-terminus of Wdr43 , which is absent in fan mutant protein , is both necessary and sufficient to mediate its nucleolar localization and protein interactions in metazoans . We demonstrate that Wdr43 functions in ribosome biogenesis , and that defects observed in fan mutants are mediated by a p53 dependent pathway . Finally , we show that proper localization of a variety of nucleolar proteins , including TCOF1 , is dependent on that of WDR43 . Together , our findings provide new insight into roles for Wdr43 in development , ribosome biogenesis , and also ribosomopathy-induced craniofacial phenotypes including Treacher-Collins Syndrome .
Neural crest cells ( NCCs ) , a transient cell type that is unique to vertebrates , originate from the dorsal aspect of the neural tube during embryogenesis . After undergoing an epithelial-to-mesenchymal transition ( EMT ) , NCCs migrate along well defined pathways , and eventually inhabit peripheral destinations where they differentiate into diverse derivatives , including melanocytes , craniofacial cartilage and bone , smooth muscle , and neuronal lineages . In the head region , cranial neural crest cells ( CNCC ) give rise to nearly all craniofacial structures , including the facial skeleton and the vast majority of facial connective tissues [1] , [2] . Defects in CNCC development are associated with craniofacial malformations , one of the most common of human birth defects [3] . Treacher-Collins syndrome ( TCS ) , an autosomal dominant congenital disorder of craniofacial development , characterized by mandibulofacial dysostosis including cleft palate and hypoplasia of the facial bones , is most commonly associated with mutations in the TCOF1 gene [4] . Treacle , the protein encoded by the TCOF1 gene , is a nucleolar phosphoprotein [5] , [6] that plays a key role in ribosome biogenesis via involvement in both rDNA methylation and rRNA transcription [7]–[10] . Extensive research in the mouse model has shown that mutations in Tcof1 disrupt ribosome biogenesis , resulting in impaired proliferation and subsequent apoptosis of neuroepithelial and NCC precursors , which in turn results in reduced numbers of NCCs migrating into the developing craniofacial complex [10] . Interestingly , inhibition of p53 function can rescue craniofacial abnormalities in mouse Tcof1 mutants , without rescuing ribosome biogenesis defects [11] . The question of how ribosome biogenesis defects can preferentially affect NCC proliferation and differentiation remains to be elucidated . In eukaryotic cells , ribosome biogenesis begins with the transcription of rRNA from rDNA located in the nucleolus , the most prominent visible structure in the nucleus . Ribosome biogenesis is extremely complex , requiring the accurate processing of pre-rRNAs into four different ribosomal RNAs ( 28S , 18S , 5S and 5 . 8S in vertebrates ) and complex formation with about 80 constituent ribosomal proteins . In addition , more than 200 nucleolar ribosome biogenesis factors are required to complete the entire ribosome biogenesis process . All ribosomal RNAs , except the 5S rRNA , are initially transcribed as a 47S polycistronic precursor , which subsequently becomes cleaved , folded and modified into the 28S , 18S and 5 . 8S mature rRNAs prior to being incorporated into functional ribosomes [12] , [13] . The cleavage and modification of rRNA is directed by small nucleolar RNAs ( snoRNAs ) . U3 , one of the most extensively studied snoRNA , is an essential component of the small subunit ( SSU ) processome , a large ribonucleoprotein ( RNP ) complex that is required for the maturation of the 18S rRNA and formation of the ribosomal small subunit ( 40S ) [14] . The SSU processome can be further subdivided to three sub-complexes , UTPB , UTPC and UTPA/t-Utp [14] . The t-Utp complex contains seven proteins ( Utp4 , Utp5 , Utp8 , Utp9 , Utp10 , Utp15 and Utp17 ) [15] , [16] . In yeast , depletion of individual t-Utp members commonly is associated with both pre-rRNA synthesis and processing defects [14] , [16] , [17] , although another group reported that the t-Utp subcomplex plays a role in pre-rRNA stabilization rather than transcription [18] . Although the functions of t-Utp components appear to be conserved in eukaryotes , some human UTP orthologs have not yet been identified [19] , indicating that Utp proteins in higher eukaryotes may have evolved specific functions . Recently , a new metazoan specific protein , NOL11 , has been characterized as a hUTP4 interacting partner via yeast two hybrid ( Y2H ) analysis [20] . Although ubiquitously expressed in virtually all eukaryotic cells , mutations in ribosome biogenesis proteins often result in tissue-specific developmental defects [21] . For example , hUTP4/Cirhin is associated with North American Indian childhood cirrhosis ( NAIC ) [22] . Mutation of zebrafish Bap28 , the ortholog of human UTP10 , results in excess apoptosis primarily in the central nervous system [23] , while mutation in WDR36/UTP21 , a modifier protein to human primary open angle glaucoma ( POAG ) , results in mouse embryonic lethality [24] . Mutation of Wdr36 in zebrafish doesn't produce any obvious defects in the first three days of development , while later developmental defects include small eyes and head combined with upregulation of the p53 stress-response pathway [25] . Developmental defects in other organs have also recently been reported [26]–[28] . Here , we report a novel zebrafish mutant , fantome ( fan ) , characterized by a variety of early developmental defects including eye , hindbrain , forebrain , cardiac , neurocranium , fin , and NCC derived pharyngeal arch cartilage development . NCCs in fan mutants fail to differentiate , and NCC precursors undergo p53 mediated cell apoptosis . fan mutants contain a point mutation in the zebrafish wdr43 gene , which encodes Wdr43 , the ortholog of yeast Utp5 . We demonstrate that , similar to the yeast ortholog , zebrafish and human WDR43 localize to the nucleolus . We also show that the C-terminal of Wdr43 , truncated in fan mutants , mediates its localization to nucleoli , and is both necessary and sufficient to mediate its interaction with other t-UTP subcomplex members . Interestingly , blocking WDR43 expression in HeLa cells results in nucleolar maturation defects , together with abnormal localization of other nucleolar proteins , including TCOF1 . Together , our data suggest that loss function of Wdr43 results in ribosome biogenesis defects that induce the p53 signaling pathway , triggering cell death of NCC precursors . We introduce the fan mutant as a valuable model to provide insight into a variety of human craniofacial neurocristopathy diseases .
The fan mutant , identified in a large scale ENU chemical mutagenesis screen conducted by the Yelick Lab [29] , was notable by its distinctive lack of pharyngeal arch cartilages at 4 days post fertilization ( dpf ) ( Fig . 1D′ , arrow ) . Subsequent developmental analyses showed that fan mutants are first identifiable at 16 hpf by a distinct area of necrotic cells present in the neural ectoderm of the presumptive eye ( Fig . 1A′ , arrow ) . At 24 hpf , a larger area of necrosis was detected mainly in neural tissues ( Fig . 1B′ , bracket ) , while at 30 hpf , fan mutants displayed distinct hydrocephaly in hind brain ventricles ( Fig . 1C′ , arrow ) , incomplete closure of the choroid fissure , and lack of pigmented retinal epithelium in the ventral eye ( Fig . 1C′ , arrowhead ) . Preliminary whole mount in situ hybridization ( WISH ) analyses of NCC marker gene expression revealed that fan mutants exhibited reduced snail2 and dlx2a expression at 12 and 24 hpf , respectively , as compared to age matched wild type sibling ( Fig . 1E′ , I′ vs . E , I ) . At 96 hpf , Alcian blue staining revealed that homozygous fan mutants lacked virtually all pharyngeal arch cartilages ( Fig . 1D′ , arrow ) as compared to age matched wild type siblings ( Fig . 1D ) . Early lethal homozygous recessive fan mutants die at approximately 5–6 dpf , while heterozygous fan embryos and adults appear normal and are viable and fertile . In normal zebrafish development , NCCs originating from the dorsal aspect of the neural tube migrate ventrally to the pharyngeal pouches and give rise to a variety of structures including pharyngeal arch cartilages [30] [31] [32] . To more carefully characterize the pharyngeal arch phenotype observed in fan mutants , we used WISH to examine the developmental expression of additional NCC markers , including: sox9a , essential for proper morphogenesis and differentiation of pharyngeal arch cartilages [33] , [34]; the pan-NCC marker crestin , expressed in pre-migratory and migratory NCCs [35]; hand2 , expressed in branchial arch mesenchyme [36]; and dlx2a , which is expressed in migrating CNC that contribute to the pharyngeal arches [37] . We detected reduced expression of all NCC marker expression in fan mutants as compared to age matched wild type sibling embryos ( Fig . 1F–K′ ) . To better visualize and study NCC defects in fan mutants , we also created a Tg ( fli1a:EGFP ) /fan mutant reporter line , which expresses EGFP in the derivatives of the cranial neural crest until at least 7dpf , and in developing vasculature [38] . We found that fan mutants exhibited fewer GFP-positive NCCs , and abnormal NCC migration and pharyngeal arch formation , as compared to age matched wild type siblings ( Fig . 1L–L′ ) . Using bulk segregant analyses , we determined that the fan locus mapped between SSLPs z8774 and z9831 on zebrafish Linkage Group ( LG ) 17 ( Fig . 2A ) , to an interval of 3 . 93 Mb containing 11 genes . Further analyses of cDNA and genomic DNA sequences of these genes identified a cytosine to thymidine mutation at nucleotide 1066 of the wdr43 gene in this interval , which introduced a premature stop codon at amino acid 356 in exon 9 ( Arg356Stop ) ( Fig . 2A ) . This mutation was confirmed by sequencing full length wdr43 cDNA amplified from fan mutant mRNA , and sequence analysis of exon 9 of the wdr43 gene in PCR amplified genomic DNA isolated from six individual homozygous fan mutants . This mutation was not present in wild type sibling cDNA or genomic DNA , and was detected along with wild type sequence in heterozygous fan family members . Full length zebrafish Wdr43 contains 650 amino acid and is well conserved from yeast to human ( Data not shown ) . Domain analysis of the zebrafish Wdr43 protein reveals that it is composed of three WD40 repeats and one Utp12 domain ( http://pfam . sanger . ac . uk ) ( Fig . 2A ) . The truncated form of Wdr43 encoded in the fan mutant lack the C-term 294 aa , including the Utp12 domain ( Fig . 2A ) . The identified gene mutation in fan/wdr43 mutants was confirmed using two approaches . We used single cell injections of titrated amounts of wdr43 antisense morpholino oligomers ( MOs ) to test whether targeted depletion of Wdr43 in wild type embryos resulted in embryos that phenocopied the fan mutant . We first confirmed the functional targeting of anti-sense wdr43 MOs by demonstrated quenching of the wdr43-GFP mRNA construct fluorescence in vivo ( Fig . S1 ) . We next injected wdr43 MOs into clutches of fan mutants and wild type single cell stage embryos , which were then raised to 3–5 dpf and stained with Alcian blue to examine pharyngeal arch cartilage formation . Our results showed that wild type embryos injected wdr43 MOs exhibited early neural tissue necrosis similar to that observed in fan mutants ( Fig . 2C , E , arrows ) . When wdr43 MOs were injected into single cell stage fan mutant embryos , we observed no apparent exacerbation of the fan mutant phenotype , suggesting that the fan mutation is a functional null ( data not shown ) . WISH analyses of wdr43 MO injected embryos revealed similar reduction in NCC marker gene expression , as observed in fan mutants ( Figure S2 ) . Secondly , we performed rescues by injecting full length wild type wdr43 mRNAs into single cell stage fan mutant and wild type sibling embryos . Analyses of injected embryos at 5 dpf via Alcian blue staining revealed rescue of the pharyngeal arch cartilage formation , although full rescue was not observed ( Fig . 2 , I versus G , F ) . Together , these data provide strong evidence that the identified wdr43 gene mutation results in the fan mutant phenotype . We next examined the developmental expression pattern of wdr43 mRNA via whole mount and sectioned in situ hybridization ( ISH ) . wdr43 is maternally expressed , and maintains a fairly ubiquitous expression pattern during the first 24 hours of development ( Fig . 3A–J ) . wdr43 expression becomes restricted to neural and pharyngeal arch tissues between 24 and 48 hpf ( Fig . 3K ) , consistent with the pharyngeal arch defects observed in fan mutants . Sectioned ISH demonstrated discrete wdr43 mRNA expression in neurepithelium and pharyngeal arches at 48 and 72 hpf ( Fig . 3 , N , N′ , P , P′ , arrows ) , consistent with the observed hindbrain and pharyngeal arch defects observed in fan mutants . Strong expression was also observed in the gut epithelium ( Fig . 3 , N , P ) . We examined the expression of fan mutant wdr43 mRNA using RT-PCR analysis of developmentally staged fan mutant and wild type siblings followed by digestion with Dde I , a unique restriction site introduced by the fan allele . These analyses showed that wild type wdr43 was detectable at all stages examined , and also that mutant wdr43 was detectable in fan mutants at 48 and 72 hpf , and thus was apparently not targeted for nonsense mediated decay ( Fig . 3 , M ) . To better characterize tissue necrosis and cell proliferation in fan mutants , TUNEL assay and phosphohistone H3 ( pH3 ) IF analyses were performed , respectively , on developmentally staged fan mutant and wild type sibling embryos ( Fig . 4 ) . TUNEL revealed significantly upregulated apoptosis in fan mutants at all developmental staged examined ( Fig . 4A , arrows ) . In contrast , cell proliferation , indicated via pH3 antibody staining , was decreased in fan mutants as compared to age matched wild type siblings ( Fig . 4B ) . Quantification of TUNEL and pH3 immunofluorescence showed significantly increased apoptosis in fan mutants as compared to age matched wild type siblings at all stages examined , and significantly decreased cell proliferation at 48 hpf ( Fig . 4C ) . Together , these results are consistent with the observed lack of NCC derived pharyngeal arch tissues in fan mutants . To better understand the molecular nature of mutant Wdr43 protein , we next examined the subcellular localization of wild type and fan mutant zebrafish Wdr43 in human cells . First , we performed immunofluorescence analyses of cultured HeLa cells using the anti-human WDR43 antibody and demonstrated that endogenous human WDR43 localized to nucleoli , as shown by co-localization with the nucleolar marker protein , B23 ( Fig . 5 , A1–A4 ) . Next , we generated N-terminal EGFP-tagged wild type and truncated fan mutant zebrafish wdr43 constructs driven by the CMV promoter , which we transfected into cultured HeLa cells , and then visualized chimeric fusion protein using anti-GFP antibody . The EGFP-tagged wild type Wdr43 protein showed perfect overlapping expression pattern with B23 ( Fig . 5 , B1–B4 ) , consistent with the previously characterized nucleolar localization of the yeast ortholog for Wdr43 , Utp5 [39] . In contrast to the full length EGFP-Wdr43 protein , EGFP-tagged fan mutant Wdr43 ( amino acids 1–364 ) lost its exclusive nucleolar localization , and rather exhibited expression throughout the nucleus ( Fig . 5 , C1–C4 ) . To correlate these in vitro results to in vivo expression studies in zebrafish , the same EGFP-tagged zebrafish wild type and fan mutant wdr43 constructs were injected into single cell stage zebrafish , which were then analyzed via confocal microscopy ( Fig . S3 ) . These analyses also showed that while full length Wdr43 co-localized with mCherry-B23 to nucleoli ( Fig . S3 , A–C ) , truncated fan mutant Wdr43 remained dispersed throughout the nucleus ( Fig . S3 , D–F ) . Utp5 , the yeast ortholog of Wdr43 , has been shown to function in the yeast t-Utp subcomplex , which mediates both pre-ribosomal RNA ( rRNA ) transcription and processing [16] [17] . Previously , it has been shown that yeast Utp5 interacts with Utp4 and Utp15 ( Freed & Baserga 2010 ) [40] . We used yeast two-hybrid analyses ( Y2H ) to examine the protein-protein interactions between yeast and zebrafish wild type and fan mutant Wdr43 with other t-Utp complex proteins . We found that both yeast and zebrafish full length Wdr43 interacted with Utp4 and Utp15 ( Fig . 5D and Fig . S4 ) , consistent with previously published yeast Utp5/Wdr43 binding studies [17] . We also determined that the C-terminal portion of Wdr43 protein is both necessary and sufficient to mediate this interaction . Zebrafish and yeast truncated fan mutant Wdr43 did not bind to either Utp4 or Utp15 , while the C-terminal fragment of Wdr43 alone was able to bind to both Utp4 and Utp15 ( Fig . 5D zebrafish and Fig . S4 yeast ) . Together , these data revealed the conserved interaction of yeast and zebrafish full length Wdr43 proteins with Utp4 and Utp15 , and also suggest that the C-terminal portion of Wdr43 , which contains the Utp12 domain , is required for protein interaction of Wdr43 with other t-UTP subcomplex member proteins . Based on our results , and those of previously published reports , we anticipated that Wdr43 would play an important role in ribosome biogenesis . We therefore investigated pre-rRNA synthesis and processing in 30 hpf and 50 hpf fan mutant and wild type sibling embryos via Northern blot analysis , using a probe specific for the 5′ external transcribed spacer ( 5′ETS ) region of the pre-rRNA at the start site of transcription ( Fig . 6A ) . These analyses showed reduced levels of the primary transcript ( labeled a ) in fan mutants ( M ) at both 30hpf and 50 hpf as compared to that of wild type ( W ) siblings , consistent with a defect in pre-rRNA transcription in fan mutants . Quantification of the ratio of full length primary transcript ( a ) to the processed pre-18S rRNA ( b ) showed reduced pre-rRNA in fan mutants relative to age matched wild type siblings . These results are consistent with previously published results showing that siRNA knockdown of human UTP5 resulted in defects in pre-rRNA transcription and processing [19] , and suggest conserved functions for vertebrate Wdr43 and yeast Utp5/Wdr43 in pre-rRNA transcription [17] . Having defined an important role for zebrafish Wdr43 in ribosome biogenesis , we further examined its function in cultured human HeLa cells , which we transfected with human WDR43 small interfering RNA ( siRNA ) ( Sigma MISSION esiRNA ) to silence WDR43 expression . GFP esiRNA was used as a negative control for these studies . Analysis of WDR43 protein and mRNA expression in siRNA treated cells using both Western blot ( Fig . 6B ) and qRT-PCR ( Fig . 6C ) analyses , respectively , confirmed that endogenous WDR43 expression was significantly reduced with WDR43 siRNA treatment . We next examined how WDR43 depletion affected the localization of another t-UTP complex Wdr43 interacting protein , UTP15 . Due to the lack of available antibody for UTP15 , we transfected N-terminal mCherry tagged zebrafish UTP15 into HeLa cells , and monitored its localization via fluorescent confocal microscopy . Consistent with a role for UTP15 in pre-rRNA processing , we found that mCherry tagged UTP15 localized to nucleoli in control GFP siRNA treated cells ( Fig . 6D ) . In contrast , in HeLa cells depleted of WDR43 using WDR43 siRNA , mCherry-tagged UTP15 did not localize to nucleoli , but rather appeared exhibited a perinuclear expression pattern ( Fig . 6E ) . These results indicate that WDR43 is required for entry into the nucleus , as well as for proper nucleolar localization of UTP15 . It was intriguing to us that many of the observed phenotypes observed in fan mutant zebrafish have also been reported in humans ( and mice ) with mutations in TCOF1/Treacle , the gene commonly mutated in Treacher-Collins Syndrome ( TCS ) . TCS results in aberrant NCC specification and differentiation , craniofacial dysmorphologies , increased cell apoptosis , and upregulated p53 signaling [10] , [11] . Based on these common characteristics , we investigated whether the localization of TCOF1 and other nucleolar proteins was affected in human HeLa cells depleted of WDR43 protein . For reference , we examined the expression of the nucleolar proteins Mpp10 , Nucleolin and Fibrillarin , which have been associated with distinct nucleolar functions . Mpp10 is normally found in the dense fibrillar component ( DFC ) and in the boundary between the DFC and the fibrillar center ( FC ) , sites of rDNA transcription and pre-rRNA splicing and modification by snoRNPs , while Fibrillarin is normally found in association with snoRNAs throughout the DFC [41] . Nucleolin/C23 is normally localized to the outer layer of nucleoli with fainter expression at the center [42] . Our investigation of the expression of these nucleolar proteins , and TCOF1 , in control and WDR43 siRNA treated HeLa cells ( Figure 7 ) showed that TCOF1 exhibited a reduced and perinucleolar expression pattern in WDR43 depleted cells ( Fig . 7 B vs . A , arrows ) . Mpp10 expression appeared reduced , but was expressed throughout the smaller nucleoli ( Fig . 7 B vs . D , arrows ) . Nucleolin , although barely detectable in WDR43 siRNA expressing cell lines as compared to control GFP siRNA treated cells , was also expressed in a perinucleolar fashion , similar to that of TCOF1 ( Fig . 7 F vs . E , arrows ) . In contrast , Fibrillarin expression appeared relatively less affected in WDR43 depleted cells , and was detected throughout the smaller nucleoli ( Fig . 7H vs . G , arrows ) . These results are indicative of disrupted nucleolar organization and rRNA transcription , consistent with the observed defects in pre-rRNA transcription observed in fan mutants . To confirm and more reliably study these observations , we next examined TCOF1 localization in stable HeLa cell lines expressing stable short hairpin RNAs ( shRNAs ) for GFP control and WDR43 ( Fig . 8 ) . We first tested five shRNAs against WDR43 gene and found that all of them showed somewhat reduced WDR43 expression using qRT-PCR analysis ( Fig . S5 ) . Western blot analyses showed that two of the shRNA WDR43 cell lines , sh9 and 2a1 , exhibited the most efficient inhibition of protein levels . As observed in WDR43 siRNA treated cell lines , stable WDR43 shRNA expressing cell lines sh9 and 2a1 exhibited mislocalized TCOF1 expression at the periphery of nucleoli , as compared to control shRNA stable cell lines ( Fig . 8 , E , I , N , O vs . A , M , arrows ) . Together , these results suggest that WDR43 expression is required for proper nucleolar organization and the subnucleolar localization of a variety of nucleolar proteins including TCOF1 , and for optimal pre-rRNA transcription . We also found that WDR43 depletion had an effect on nucleolar size and shape . WDR43 depleted cells had larger numbers of mini nucleoli as compared to control cells , which exhibited fewer numbers of larger sized , mature nucleoli . To quantitate this observation , we monitored nucleolar number and size in control and WDR43 depleted cultured HeLa cells immunostained for TCOF1 ( Fig . 8 ) . We found that in normal and control shRNA cultured HeLa cells , nucleoli reassembled after mitosis , with several small nucleoli fusing into ∼1–4 larger , mature nucleoli per HeLa cell . In contrast , TCOF1 expressing nucleoli failed to fuse together in WDR43 shRNA expressing HeLa cells , but rather remained as small unfused mini nucleoli , or “nucleolar caps” , which also appeared spherical shape as compared to the irregular shaped nucleoli present in control HeLa cell cultures ( Fig . 8 F , J vs . B ) . We also found that the average number of nucleoli was increased , and the number of fused nucleoli was reduced , in WDR43 shRNA expressing HeLa cells as compared to control shRNA treated HeLa cells , as shown using the nucleolar marker B23 ( Fig . 8 , F , J vs . B ) . Quantification of these results revealed that WDR43 depleted cells exhibited increased numbers of smaller nuclei as compared to control cells ( Fig . 8 , Q , R ) . Together , these results demonstrate that depletion of WDR43 , an essential ribosome biogenesis factor , affects nucleolar maturation and assembly . Ribosome biogenesis defects , such as those observed in fan mutants , have been reported to be associated with upregulation of the p53 signaling pathway , and cellular apoptosis [11] , [25] [43] [44] [45] [46] . Based on the increased apoptosis observed in fan mutants , we next examined p53 signaling pathway gene expression in developing fan mutant and wild type sibling embryos . Immunohistochemical analysis using the zebrafish p53 antibody ( kind gift of David Lane ) [47] revealed high levels of expression of p53 in fan mutants , while p53 was virtually undetectable in wild type sibling embryos at 5 dpf ( Fig . 9 , B vs . A ) . Next , we used qRT-PCR analyses to show that the expression of p53 downstream target genes , including the N-terminal truncated p53 isoform delta113p53 , mdm2 , and cyclin G1 were all upregulated in 24 , 48 and 72 hpf fan mutants as compared to wild type sibling controls ( Fig . 9 , C , D and data not shown ) . We next tested whether knockdown of p53 signaling via injection of p53 anti-sense MOs could rescue the fan mutant phenotype . Similar to previous reports in the Treacher-Collins mouse model [11] , we found that the fan mutant pharyngeal , neural and eye defects were largely rescued in p53 MO injected fan mutants ( Fig . 9 , G vs . F , and H ) , and NCC marker gene expression was also rescued in fan mutants ( Fig . S2 ) . TUNEL analyses revealed rescue of apoptosis in p53MO injected fan mutants ( Fig . 9 . K vs . J , I ) , indicating that increased apoptosis observed in fan mutants was mediated via upregulated p53 signaling pathways .
Proper ribosome biogenesis is required for the production of functional ribosomes , the primary site of protein synthesis . Most if not all ribosomal proteins ( RPs ) are thought to be essential for ribosome biogenesis and cell survival . It is therefore surprising that ribosome biogenesis defects caused by mutations in certain RPs can lead to variable and seemingly tissue-specific defects in vertebrate development . For example , mutations in several RPs are associated with human congenital hypoplastic Diamond-Blackfan anemia ( DBA ) , and similar DBA phenotypes were observed when DBA associated RP mutations were expressed in zebrafish [48]–[51] . In addition , several reports on zebrafish ribosome biogenesis protein mutants describe a variety of diverse phenotypes , ranging from tumors , to central nervous system degeneration , to organogenesis defects [23] , [25] , [52] , [53] . Here we present data showing that defects in the zebrafish ribosome biogenesis protein Wdr43 result in early developmental defects in a variety of tissues , including neural , eye and heart and pharyngeal arches , while later developmental defects appear fairly localized to NCC derived pharyngeal arch cartilages . These observations raise the question of how a defect in what is thought to be a universally required ribosomal biogenesis protein , Wdr43 , can result in a rather specific craniofacial tissue-specific phenotype ? One possibility is that there are tissue specific , developmental requirements for ribosome biogenesis proteins . For example , certain ribosome biogenesis factors may have tissue specific , developmental expression patterns . In fact , we show here that the expression of zebrafish wdr43 mRNA becomes localized to neural and pharyngeal arch tissues starting at ∼24 hpf , which is consistent with the observed fan/wdr43 mutant phenotype . It is possible that additional ribosomal proteins that regulate cell cycle or apoptosis may similarly exhibit tissue specific expression patterns . Another theory is that ribosome biogenesis defects and subsequent anticipated reduced protein translation efficiency will most significantly affect those tissues exhibiting a high demand for protein synthesis . This may include NCC and erythropoiesis progenitor cells , although recent data does not find evidence for increased translation in NCC [21] . Finally , decreased efficiency of cellular translation machinery may affect a wide and varied spectrum of translation products in different cell types due to mRNA competition for timely translation , which could result in diverse readouts in different cell types . The craniofacial phenotype exhibited by zebrafish fan mutants resembles the craniofacial malformations observed in individuals with Treacher-Collins Syndrome . The fact that NCC specification and differentiation are similarly affected by mutations in nucleolar proteins – TCOF1/Treacle and a common subunit of RNA polymerases I and III in Treacher-Collins Syndrome [54] [55] , and Wdr43 in fan mutants - leads to the intriguing question of why NCCs may be more sensitive to ribosome biogenesis defects as compared with other tissues . Based on our data presented here and on the published reports of others , we hypothesize that high protein translation levels must be maintained by progenitor and differentiating NCCs in order to support their extensive cell proliferation , migration and differentiation . In Treacher-Collins Syndrome , TCOF1 mutant induced defects in ribosome biogenesis are characterized by stimulation of the nucleolar stress response , which in turn activates the p53 apoptosis pathway , resulting in the depletion of the neural crest precursor pool [10] . We observe a similar upregulation of p53 signaling and depletion of NCCs in fan mutants . Although beyond the scope of the present study , it will be interesting in future studies to compare pre-rRNA transcription , ribosome biogenesis and protein translation efficiency in developmentally staged NCC versus non-NCC populations harvested from fan mutant and wild type siblings . Our Northern blot results indicated that pre-rRNA levels are significantly decreased in developmentally staged zebrafish fan/wdr43/utp5 mutants , consistent with the previously characterized role for yeast Utp5 in pre-rRNA transcription [16] , [19] . We suggest that a variety of nucleosomal proteins are required for optimal pre-rRNA transcription . Novel findings from this report include the fact that blocking WDR43 function in HeLa cells resulted in the distinct mislocalization of nucleolar proteins including UTP15 , Mpp10 , nucleolin and to a lesser extent fibrillarin , suggesting that Wrd43/UTP5 is required for proper subnucleolar organization and function . Interestingly , TCOF1 also mislocalized to the outer periphery of nucleoli , rather than exhibiting its normal distribution throughout the nucleolus , suggesting that WDR43 may also be required for proper TCOF1 subnucleolar localization and function . Although we have not detected direct binding between TCOF1 and WDR43/UTP5 using Y2H , we have detected interactions between WDR43/UTP5 and other rDNA transcription component proteins ( data not shown ) . Together , these results suggest roles for Wdr43/UTP5 in ribosomal protein sub-nucleolar localization and function of other ribosome biogenesis factors , and raise the intriguing possibility that manipulation of WDR43 expression could be used to correct the localization and improve the function of TCOF1 in Treacher-Collins Syndrome patients . Nucleolar mis-localization phenotypes have also been observed in HeLa cells treated with actinomycin D , an inhibitor of RNA Pol I [56] , which is a TCOF1/Treacle interacting protein [57] . It is possible that WDR43 may also function together with TCOF1 and Nopp140 to recruit proteins to the nucleolar organizer regions ( NORs ) and the upstream binding factor ( UBF ) , an RNA PolI transactivator [19] . Wdr43 could also mediate rRNA transcription by binding to rDNA and UBF directly , as shown by other Utps [58] . These functions for Wdr43 remain to be elucidated . We also used both siRNA and shRNA WDR43 silencing methods to confirm the function of WDR43 in nucleolar fusion in cultured HeLa cells . At the present time , mechanisms regulating nucleolar fusion remain poorly understood . It has been shown that after mitosis , multiple small nucleoli form around transcriptionally active NORs , and as cells progress through the cycle , these small nucleoli fuse to form larger nucleoli [59] [60] . Although the mechanism of WDR43 function in nucleolar fusion is not clear , preventing nucleolar fusion may not be common to all ribosome biogenesis protein mutations based on the fact that inhibition of NOL11 resulted in the formation of one large ( not small ) nucleolus [20] . One possible explanation is that WDR43 depletion may result in structural changes to rDNA , which in turn could interfere with nucleolar fusion [61] . Such a proposed function for WDR43 may be dependent or independent of its function in the t-Utp complex . A recent study using Xenopus oocytes showed that nucleoli exhibit fluid dynamics similar to that of liquid droplets , and that nucleolar fusion requires dynamic exchange between nucleoli and the nucleoplasm [62] . In future studies , it will be interesting to determine whether WDR43 is also involved in this process . Recent reports emphasize the apparent tissue specific functions for ribosomal proteins previously thought to exhibit functions in all cells and tissues . The results presented here suggest previously unrecognized roles for Wdr43/UTP5 in craniofacial development . The fact that Wdr43/UTP5 is needed for proper formation of nucleoli and for sub-nucleolar organization and function indicates important roles for Wdr43 as a key participant in ribosome biogenesis . As such , the zebrafish mutant fantome provides a valuable vertebrate developmental model and tool to continue in depth functional studies of RPs and ribosome biogenesis factor proteins in NCC differentiation , including the identification of effective tools for reducing the incidence of craniofacial birth defects .
AB and WIK fantome/wdr43 mutant and wild type zebrafish were raised in the Tufts Zebrafish Facility at 28 . 5°C and developmentally staged as previously described ( Westerfield , M . , 1995 ) . For whole mount in situ hybridization analyses , pigmentation was inhibited by treating embryos with 1-phenyl-2-thiourea ( PTU ) at a final concentration of 0 . 2 mM as previously described ( The Zebrafish Book , U . Oregon Press ) . All experimental procedures on zebrafish embryos and larvae were approved by the Tufts University Institutional Animal Care and Use Committee ( IACUC ) and Ethics Committee . The fan mutant was identified in a large-scale ENU-mutagenesis screen conducted by the Yelick Laboratory [29] . Genetic mapping strains were created by crossing identified heterozygous fan mutants to polymorphic WIK wild type zebrafish . Embryos were collected from pairwise matings of mapping strain fan/WIK heterozygotes , and scored at 48 hpf for fan specific phenotypes . Genomic DNA was extracted from individual fan mutant and wild type embryos , and bulk segregant analyses were performed using primers designed to amplify SSLP markers from the Massachusetts General Hospital Zebrafish Server website ( http://zebrafish . mgh . harvard . edu ) . The fan mutation mapped to zebrafish linkage group 17 ( LG17 ) , to a region spanning 11 genes . Nucleotide sequence analyses of all 11 genes identified a premature stop codon in the wdr43 gene of all fan mutant embryos that was not present in wild type siblings . Whole-mount and sectioned in situ hybridizations were performed as previously described ( Thisse et al . , 1993 ) , using a probe generated via PCR using the following primers ( wdr43-forward: 5′- CAGTGCAACAAAAGTTGGTGA-3′; wdr43-reverse: 5′- AAAGTTCTGGTTGGCTGCA-3′ ) . All other probes were obtained from zfin . org . Embryos were analysed using Zeiss Axiophot and M2Bio microscopes , and imaged using Zeiss Axiophot Imager digital camera ( Munich , Germany ) . Digital images were processed using Adobe Photoshop software . Antisense morpholino oligonucleotides ( MOs ) targeted to the initiation of translation codon of wdr43 mRNA ( 5′TCCGTCCGCCGCCATCTTACCGTTC3′ ) were injected into the yolk of 1 cell stage wild type or fan mutant embryos . 2 nL of MO at a concentration of 10 ng/µL was used to knockdown wdr43 translation . Total RNA was extracted from 20 wild type and 20 fan mutant embryos at 24 , 48 and 72 hpf , or from HeLa cells 48 hours after transfection using RNeasy Plus Kit ( Qiagen , Valencia , CA ) . DNA was removed using the DNA-free DNase Treatment & Removal Kit ( Ambion , Life Technologies , Grand Island , NY ) to remove genomic DNA contamination . cDNA was synthesized using a SuperScript III First-Strand Synthesis System ( Invitrogen , Life Technologies , Grand Island , NY ) with random primers . Gene expression was quantified by qRT-PCR using QuantiTect SYBR Green PCR Master Mix ( Qiagen , Valencia , CA ) and real-time cycler Mx3000P ( Stratagene , Agilent Technologies , Santa Clara , CA ) . Primers for zebrafish p53 isoforms , mdm2 , and cyclinG1 were used as described [63] . The following primers were used to amplify the human WDR43 gene: Forward: CCTTCCGCGCACCTCAGTGGTAC; Reverse: AACTGGCGTTGCATGTCCTGTGA . Primers for β-actin , used to normalize the expression levels , were as described [64] . Yeast two-hybrid assays for interaction between yeast and zebrafish Utp proteins were performed as previously described ( Freed and Baserga , 2010 ) . Briefly , yeast utp5 cDNA encoding full length , N-terminal ( 1–343aa ) and C-terminal ( 344–643aa ) proteins were cloned into the pGADT7 prey vector . Additional yeast UTP genes of the t-Utp subcomplex ( Utp8 , Utp9 , Utp10 , Utp15 and Utp17 ) cloned into the bait vector were as previously described ( Freed and Baserga 2010 ) . Both bait and prey vectors were transformed into AH109 yeast strain and interactions were identified based on the ability of transformants to grow on AHTL dropout medium after 3–5 days of incubation at 30°C . To test the interaction between zebrafish Wdr43/Utp5 , Utp4 and Utp15 , full length zebrafish utp4 and utp15 cDNAs were purchased ( Openbiosystems , Lafayette , CO ) and cloned into pGADT7 prey vector . Zebrafish wdr43 cDNAs encoding full length , truncated fan mutant Wdr43 ( 1–356aa ) and C-terminal portion of Wdr43 ( 357–650aa ) proteins were cloned into the pGABT7 bait vector . Interactions were tested by growth in triple dropout medium after 3–5 days of incubation at 30°C . The Y2H studies of zebrafish Wdr43 protein interactions were tested in both bait and prey constructs . To determine the subcellular localization of wild type and mutant Wdr43 proteins , GFP cDNA was cloned onto the N-terminal end of full length or fan mutant zebrafish wdr43 cDNA under the direction of the CMV promoter , using multi-site Gateway reactions [65] . These constructs were then transfected into HeLa cells using Lipofectamine 2000 reagent ( Invitrogen , Life Technologies , Grand Island , NY ) . After 36 hours of growth at 37°C , transfected cells were fixed with 4% PFA , and subjected to standard immunofluorescence ( IF ) analyses using the anti-B23 antibody ( 1∶200 , Santa Cruz Biotechnology , Inc . , Santa Cruz , CA ) . The rabbit polyclonal anti-WDR43 antibody ( 1∶100 , Abcam , Cambridge , MA ) was used to detect the endogenous WDR43 . To check the expression in zebrafish embryos , the same constructs were injected into single cell zebrafish embryos , which were harvested at 24hpf , and analyzed for GFP expression using a Leica TCS SP2 confocal microscope . For siRNA experiments , HeLa cells were transfected with WDR43 MISSION esiRNA ( Sigma , EHU004691 ) using Lipofectamine 2000 reagent ( Invitrogen , Life Technologies , Grand Island , NY ) . GFP esiRNA ( Sigma , EHUEGPF ) was used as a negative control . Treated cells were harvested 48 hours after transfection for Western blotting and immunofluorescence experiments . shRNA constructs were purchased from Openbiosystems and used to establish stable shRNA expressing HeLa cells following the manufacture's protocol . The following shRNA constructs were used: pGIPZ-WIPI1-2 , RHS4430-98853022; pGIPZ-non-targeting control RMS4348 . Total RNA was extracted from 20 wild type and 20 fan mutant embryos at 30 and 50 hpf using standard Trizol protocol for RNA isolation . Northern blot analysis was carried out as described in Freed et al , 2012 [20] . For each sample , 2 µg of RNA was separated by electrophoresis on a 1% agarose/1 . 25% formaldehyde gel in Tricine/Triethanolamine buffer and transferred to a nylon membrane ( Hybond-XL , GE Healthcare ) . Pre-rRNA species were detected by methylene blue staining and hybridization with a 32P-radiolabelled oligonucleotide probe to the 5′ETS: CGAGCAGAGTGGTAGAGGAAGAGAGCTCTTCCTCGCTCA . Quantification of pre-rRNA processing was performed using Image J ( National Institutes of Health ) . Developmentally staged wild type and fan mutant embryos , fixed and processed for cryosectioning , were sectioned at 10 µm . Apoptosis TUNEL assay was performed using the In situ cell death detection kit , Fluorescein ( Roche Applied Science , Indianapolis , IN , USA ) . Cell proliferation was assayed with phospho-Histone H3 immunofluorescence analysis , using anti- phospho Histone H3 ( Ser10 ) antibody ( Cell Signaling , Danvers , MA ) and anti-rabbit goat antibody conjugated with Alexafluor 594 ( Life Technologies , Grand Island , NY ) . | Here , we describe the identification and characterization of a novel zebrafish craniofacial mutant , fantome ( fan ) , caused by a point mutation in the wdr43 gene . Although previously characterized as UTP5 in yeast , a nucleolar protein functioning in ribosome biogenesis , here we show that Wdr43 also regulates early zebrafish development , including NCC specification and differentiation . Mutations in nucleolar proteins have been found to be causative for a variety of human craniofacial syndromes including Treacher-Collins Syndrome ( TCS ) , often caused by mutations in TCOF1 , which also plays important roles in ribosome biogenesis . However , the underlying mechanisms linking ribosomal biogenesis and NCC specification and differentiation into pharyngeal arch cartilages remains poorly understood . Here we describe the fan/wdr43 mutant phenotype , and present functional characterizations of Wdr43 in craniofacial development . We show that WDR43 is required for the proper nucleolar localization of a variety of nucleolar proteins , including TCOF1/Treacle . These studies provide new insight into ribosomal protein function in early zebrafish development , with focus on NCC derived craniofacial development , as a model for human craniofacial neurocristopathies . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"developmental",
"biology",
"model",
"organisms",
"genetics",
"biology",
"genomics",
"neuroscience"
] | 2014 | Tissue Specific Roles for the Ribosome Biogenesis Factor Wdr43 in Zebrafish Development |
Populations of Human Immunodeficiency Virus type 1 ( HIV-1 ) undergo a surprisingly large amount of genetic drift in infected patients despite very large population sizes , which are predicted to be mostly deterministic . Several models have been proposed to explain this phenomenon , but all of them implicitly assume that the process of virus replication itself does not contribute to genetic drift . We developed an assay to measure the amount of genetic drift for HIV populations replicating in cell culture . The assay relies on creation of HIV populations of known size and measurements of variation in frequency of a neutral allele . Using this assay , we show that HIV undergoes approximately ten times more genetic drift than would be expected from its population size , which we defined as the number of infected cells in the culture . We showed that a large portion of the increase in genetic drift is due to non-synchronous infection of target cells . When infections are synchronized , genetic drift for the virus is only 3-fold higher than expected from its population size . Thus , the stochastic nature of biological processes involved in viral replication contributes to increased genetic drift in HIV populations . We propose that appreciation of these effects will allow better understanding of the evolutionary forces acting on HIV in infected patients .
Genetic drift is defined as stochastic fluctuations in frequencies of alleles in a population . Generally , large populations are less stochastic and undergo less genetic drift than smaller populations . While viruses exhibit very large population sizes , suggesting that the genetic processes in these populations are mostly deterministic , it has been recently appreciated that genetic drift is an important factor in virus evolution . For example , plant viruses undergo severe bottleneck events both when spreading from one plant to another and within an individual infected plant , which leads to frequent founder effects in their populations [1] , [2] , [3] , [4] , [5] , [6] . Significant contribution of genetic drift has also been proposed for evolution of animal and human viruses , such as norovirus [7] , measles [8] , hepatitis B virus [9] , coronavirus [10] , Dengue virus [11] , rabies virus [12] , and hantavirus [13] . However , accurate determination of the role of genetic drift in evolution of animal viruses is complicated , because genetic drift , a stochastic process , is hard to discern from antigenic drift , which is a selection-driven process associated with individual differences in immune responses of infected hosts . Nevertheless , studies aimed at separating the role of the immune response still find a significant influence of genetic drift in evolution of some animal viruses [14] , [15] , [16] . The processes that cause genetic drift in extremely large viral populations have not been thoroughly explored , but they include events that occur during both transmission of the virus from one organism to another and viral replication within a single organism ( intra-host ) . Studies of wheat streak mosaic virus suggest that both mechanisms can contribute significantly to viral evolution [1] , [2] , but whether this conclusion applies to animal viruses is not clear . The majority of the work on genetic drift in animal viruses has focused on large scale viral populations , comparing viruses either in geographically isolated regions or in consecutive epidemics and , therefore , does not distinguish between the intra- and inter-host genetic drift . The animal virus for which intra-host genetic drift has been extensively studied is human immunodeficiency virus type 1 ( HIV-1 ) . Multiple studies observed that genetic drift of HIV-1 within an infected individual is several orders of magnitude larger than would be predicted from the total number of infected cells in the body [17] , [18] , [19] , [20] , [21] . Several models have been proposed to explain the observed high genetic drift , including multiple selective sweeps [22] , metapopulation structure [23] , [24] and rare but severe population bottlenecks [25] . All of these models , however , implicitly assume that viral population replicating under homogenous well-mixed conditions should behave as an ideal population . Ideal populations are expected to undergo a certain amount of genetic drift , but real viral populations are also influenced by to the stochastic nature of the biological processes involved in viral replication . Therefore , real viral populations can be expected to have an excess of genetic drift , even under “close-to-ideal” conditions . The degree to which viral replication process contributes to genetic drift in viral populations is the main interest of our study . Here , we studied genetic drift in HIV populations replicating under the most basic conditions of cell culture . We developed a system that can be used to accurately measure genetic drift occurring in HIV populations . Using this system we show that genetic drift in HIV populations in culture is approximately tenfold higher than expected for an ideal population . Because the increase in HIV genetic drift observed in culture is due to replication process itself , it should also be present in vivo and , therefore , may partially explain the high genetic drift observed in HIV populations in infected people .
Our approach to investigating the impact of genetic drift on HIV was to create viral populations of known size and monitor the degree of variation in the frequency of a neutral allele in these populations . An HIV population carrying a neutral allele at 50% frequency was created by mixing two replication-competent variants of HIV , Vpr-FS and Vpr-FS-StuI ( Figure 1A ) . Both variants carry a frameshift insertion in the vpr gene , resulting in non-functional Vpr protein , which is not necessary for viral replication in cell culture . The insertions in the two variants differ in length by 4 bp , which allows accurate measurement of the frequency of each variant in viral mixtures by the PCR-based GeneScan assay ( see Materials and Methods ) . Neither Vpr-FS nor Vpr-FS-StuI has an advantage for replication in culture , i . e . the variants are selectively neutral ( data not shown ) . Thus , by mixing these two variants in a 1∶1 ratio we created a population of HIV with a known neutral allele present at a 50% frequency . The experimental scheme we used to determine the relationship between population size and genetic drift is shown in Figure 1B . Serial 3-fold dilutions of a 1∶1 viral mixture of neutral variants were prepared and used to infect multiple independent cultures of target cells to create HIV populations of different sizes . All cultures were maintained for 5–14 days until most of the cells in virus-positive cultures were infected ( see Materials and Methods ) . At that point , the cell-free virus from virus-positive wells was collected and analyzed by the GeneScan assay to determine the frequency of the two alleles in each of the replicate cultures . The observed Vpr-FS and Vpr-FS-StuI frequencies in each set of replicates were used to calculate the average variance of the observed frequency from the expected 50% . We used this variance as a measure of genetic drift . In parallel , the size of viral population at the beginning of each experiment was calculated by measuring tissue culture infectious dose 50% ( TCID50 ) , which the amount of virus that infects 50% of the wells at a given dilution ( see figure legend to Figure 1B ) . In several experiments , an infectious center assay was used to confirm the estimated number of infected cells at higher dilutions and the results were always within twofold from predictions based on multiplicity of infection ( data not shown ) . For illustration of the technique , representative data obtained in one genetic drift measurement experiment are shown in Figure 2 . Twelve independent cultures of C8166 cells were infected for each dilution of the viral mixture . As expected , at the dilution with the highest amount of virus ( 9553 infected cells per well ) the frequency of Vpr-FS-StuI was very close to 50% ( Figure 2A ) . When the amount of virus used for infection was decreased , Vpr-FS-StuI frequency showed wider variation . Because the same 1∶1 viral mixture was used for all dilutions , the variation in the frequencies of the two neutral variants must have been caused by genetic drift . At the lowest viral dilution ( less than one infected cell per well ) , only a single variant was observed in each of the five infected cultures; 7 of the cultures were not infected at this dilution . It should be noted that Vpr-FS-StuI frequency was equally likely to increase or decrease relative to the starting 50% , which confirmed that Vpr-FS and Vpr-FS-StuI have identical replicative fitness . These data were used to calculate the variance in frequency of Vpr-FS-StuI in replicate cultures for each dilution ( Figure 2B ) . We used this variance as a measure of the amount of genetic drift in HIV populations . As expected , the variance was lowest ( 0 . 00045 ) for the largest population size , indicating that genetic drift was low in these cultures . Variance increased as the population size decreased , demonstrating the predicted reciprocal relationship between the population size and genetic drift . The variance was not calculated for two dilutions with the lowest population size ( 1 . 5 and 0 . 5 ) , because they contained non-infected wells . Assay variance ( see Materials and Methods ) in this experiment was 0 . 000086 , i . e . was less than 20% of the lowest measured total variance ( data not shown ) . To better understand the sources of genetic drift in our experiments , we compared the observed variance to the variance expected to occur due to genetic drift in an ideal population . Probability theory predicts that , in a single generation , an ideal population of N individuals should undergo genetic drift simply due to stochastic sampling . The variance caused by this drift , to which we refer here as Videal , was calculated from the initial allele frequency p and viral population size N asand plotted on Figure 2B as the thin dashed line . For all tested population sizes , the observed variance in frequency of neutral allele was approximately an order of magnitude higher than Videal , demonstrating that , even under relatively homogenous conditions of cell culture , viral populations do not behave as ideal populations . We asked whether amount of genetic drift in HIV populations can be influenced by culture conditions . Thus , we measured Vpr-FS-StuI variance in cultures where the virus mixture was bound to the Raji-DC-SIGN cells prior to the addition of target C8166 cells . In order to do this , we used Raji-DC-SIGN cells , which cannot be infected by wild type HIV , but can bind the virus through the DC-SIGN molecule on their surface and enhance its infectivity by presenting the virus to the surfaces of uninfected cells [26] , [27] . Raji-DC-SIGN cells were incubated with virus for 1 h and then washed three times with media to remove all unbound virus . The cells were then mixed with the target C8166 cells to mediate infection in trans . Vtotal under these conditions was only 4 . 5 fold higher than Videal ( Figure 4A ) , a statistically significant reduction from the 10 . 5 ratio of Vtotal to Videal observed in direct C8166 infections ( Table 1 ) . In order to test whether genetic drift was affected by pre-bound state of the virus or the simple presence of Raji-DC-SIGN cells in culture , we mixed C8166 cells with Raji-DC-SIGN cells prior to addition of the virus ( Figure 4B ) . Because the simple presence of Raji-DC-SIGN in culture did not affect genetic drift of HIV , as evidenced by 9 . 5-fold increase of Vtotal over Videal ( Figure 4B and Table 1 ) , this result suggested that the reduction of genetic drift was connected to the state of the virus in the beginning of infection . We hypothesized that binding of the virus to virus-presenting cells and removal of the unbound virus resulted in increased synchronization of timing of infection , which caused reduction in genetic drift . We tested whether synchronization of infection has an effect on the amount of genetic drift in viral populations by changing experimental parameters to favor synchronized entry of virus entry into cells . Thus , virus was incubated with target cells for one hour , after which the unbound virus was removed by three washes with media . Synchronization of infection drastically reduced the amount of observed genetic drift ( Figure 4C ) , so that Vtotal under these conditions was only 3-fold higher than Videal . The increase was significantly lower than increase in non-synchronized infections of C8166 cells ( Table 1 ) . Similar results were obtained when the virus was incubated with cells for 3 hours , or when it was pre-bound to C8166 target cells at 8C , or spinoculated ( data not shown ) . Thus , a large proportion of genetic drift in HIV populations was due to stochastic effects associated with non-synchronous infections of target cells .
In this study we have shown that genetic drift of HIV populations existing under relatively homogeneous conditions of cell culture exceeds genetic drift expected for an ideal population by an order of magnitude . A large portion of the observed drift is due to the non-synchronous nature of infection , where a small proportion of virions gains reproductive advantage by quickly infecting their target cells . When infection is synchronized , the observed genetic drift is reduced , but is still approximately 3-fold higher than drift expected for an ideal population . Genetic drift depends on the size of the population in question . However , the definition of an individual and , therefore , of a population size , is somewhat complicated for viruses . Individual virions do not contribute to the next generation unless they infect a target cell . The number of infected cells , therefore , is a better measure of population size , and is quite common in HIV population genetics literature [17] , [18] , [22] . Yet the fact that a single cell can be infected by more than one virion adds some confusion . The latter problem was avoided in our study by starting experiments at low multiplicity of infection ( <0 . 05 ) , which makes double-infection unlikely . We defined the size of viral population as the total number of cells infected in the culture by the virus added at the beginning of the experiment . The reasons for the increased genetic drift in HIV populations are not entirely clear . Our work provides strong evidence for the importance of synchronization of infection for reduction of genetic drift . While we have not measured kinetics of infection , it is logical to assume that some of the infections occurred within minutes of addition of virus to cells , while others occurred hours , or even days , later . Indeed , when input virus was washed off after 1 hr during synchronization experiments , viral titers were reduced 10–100-fold as compared to non-synchronized infection ( data not shown ) . As a result , in non-synchronized infections there exists an “early” virus population , which has a temporal advantage over the “late” viral population . “Early” population produces new virions faster and contributes to the next generation of infected cells more than the “late” population . Therefore , the small size of the “early” population defines to a large degree the amount of genetic drift observed in the total population . The potential influence of such non-discrete generations on genetic drift is a well-described phenomenon in population genetics ( for review see [28] ) . Indeed , we found that genetic drift can be significantly reduced when infections are synchronized . Synchronization removes the “late” viral population and , therefore , the total size of viral population becomes much closer to the size of the “early” population . This results in a more synchronous production of virus particles in the second generation ( the major contributor to genetic drift ) , which becomes discrete reducing the observed genetic drift . Yet even in synchronized infections , genetic drift is higher in HIV populations than in an ideal population . We believe that the reasons for that lie in differences in metabolic state of target cells , their lifespan , or expression levels of positive and negative factors involved in viral replication . These and other individual differences between target cells can introduce stochastic events into the viral life cycle . For example , it has recently been proposed that positive feedback loops in transactivation of RNA synthesis by viral protein Tat lead to stochastic differences in levels of viral gene expression [29] . As a result , the viral population is randomly divided into actively replicating and latent subpopulations . In general , these and other factors may lead to a non-Poisson distribution in the number of progeny per parent and affect the amount of genetic drift in the population [30] . Interestingly , our data showed that genetic drift in a heterogeneous mixture of C8166 and CEMx174 cells was similar to genetic drift in a more homogeneous C8166 culture , which appears to contradict this prediction . However , it is possible that the differences between target cells in those experiments were not sufficient to result in an observable effect . Additional experiments are needed to establish the relevant biological differences of different target cells and the effect of those differences on genetic drift in viral populations . All current models of HIV genetic drift implicitly assume that the process of viral replication itself is stochastic only to the degree of an ideal population of a given size [21] , [22] , [23] , [25] . Our results show that this is not the case . Populations of HIV in cell culture undergo approximately tenfold more genetic drift than would be expected from their population sizes . This increase is not sufficient to explain the several orders of magnitude excess in genetic drift of HIV observed in patients [17] , [19] , [20] , [21] , but it provides experimental evidence for one source of genetic variation in HIV populations . Indeed , the factors that we proposed to explain the increased genetic drift of HIV in culture should play similar , or even larger , roles in HIV populations in patients . There , infections should be less synchronized than in culture and the individual differences between target cells should be larger than in cell lines or highly stimulated PBMCs that we used in our experiments . Our data suggest that the existing models which explain this excess of genetic drift through multiple selective sweeps [22] , metapopulation structure [23] , [24] or rare severe population bottlenecks [25] may overestimate the influence of those factors on HIV population genetics . We believe our findings will allow creation of better models describing forces acting on HIV population genetics in an infected person . Genetic drift is a powerful evolutionary force and understanding the factors contributing to it is crucial for our understanding of HIV evolution .
C8166 and CEMx174 cells , expressing secreted alkaline phosphatase ( SEAP ) , were a kind gift of Dr . R . Desrosiers [26] . Raji-DC-SIGN cells have been previously described as B-THP-1 cells [27] . All non-adherent cell lines were maintained in RPMI 1640 medium ( Invitrogen ) supplemented with 7% bovine growth serum , 2 mM L-glutamine , 100 U/ml penicillin and 100 µg/ml streptomycin . Human 293T cells ( ATCC ) were maintained in DMEM medium ( Invitrogen ) supplemented with 7% bovine growth serum and antibiotics . Human peripheral blood mononuclear cells ( PBMCs ) were isolated from healthy donors using Ficoll gradient and maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum , 2 mM L-glutamine and antibiotics . PBMCs were activated prior to infection by phytohemeagglutinin A ( 10 ug/ml ) and interleukin-2 ( 10 U/ml ) for 3 days . Variants of HIV-1 LAI clone , Vpr-FS and Vpr-FS-StuI , deficient in vpr protein were generated by inserting polylinkers of 16 ( Vpr-FS-StuI ) or 20 ( Vpr-FS ) nucleotides at position 980 of the gene ( details are available upon request ) . To generate stocks of infectious virus , 293T cells were transiently transfected with each proviral clone using TransIT reagent ( Mirus , Madison , WI ) and the produced virus was filtered through 0 . 22 µm filter ( Corning ) to remove cell debris . To generate 1∶1 mixture of Vpr-FS and Vpr-FS-StuI clones , small aliquots of the viruses were mixed in several ratios by volume and used to infect C8166 cells at multiplicity of infection ( MOI ) ∼0 . 1 . Cell-free virus was collected three days later and used to measure the ratio of Vpr-FS to Vpr-FS-StuI in each mix ( described below ) . The results were used to mix the two viruses at a 1∶1 infectivity ratio . The 50 µL aliquots of the mixture were stored at −80°C . Two independent mixtures were created and used for all of the experiments described here . Multiple measurements showed that Vpr-FS-StuI was found in the first and second mixture at 51 . 00% and 49 . 68% frequency , respectively . Both mixtures are referred to as “1∶1 mixture” or “50% mixture” throughout the text , but the actual frequencies were used in calculations of the variance . Viral replication in cultures of C8166 and CEMx174 cells was monitored by an increase in SEAP activity in culture medium using PhosphaLight detection kit according to the manufacturer's instructions ( Applied Biosystems , Foster City , CA ) . Viral replication in PBMCs was monitored by the amount of p24 antigen in cell-free media using a p24 detection kit ( NCI-Frederick ) and QuantaBlu Fluorogenic Peroxidase Substrate kit ( Pierce ) according to manufacturers' instructions . Tissue culture infectious dose of 50% ( TCID50 ) was measured in every experiment by the standard approach of limiting virus dilutions and counting the number of infection-negative wells in each dilution . Rows of cultures that contained virus-negative wells were maintained for 21–28 days to ensure detection of all infected wells . To calculate the number of infected cells at each dilution , TCID50 units were converted into infectious particles per milliliter by multiplying TCID50 by 0 . 7 ( to account for the fact that Poisson distribution predicts 50% negative outcomes at ∼0 . 7 mean ) and adjusting for the dilution factor . In some cases , an infectious center assay was used to measure the number of infected cells in a well after infection . Cells were washed to remove any non-bound virus , serially diluted and mixed with 4×105 non-infected cells in 6 replicates for each dilution . The total number of infected cells in the original cell population was calculated similar to TCID50 calculation . RT-PCR-based GeneScan assay was used to measure the proportions of Vpr-FS and Vpr-FS-StuI clones in cultures as described previously [28] , but with a different set of primers . The primers for PCR were designed to flank the region in the vpr gene containing insertions . The primers were 705-vpr-F1 ( 5′-GCCACACAATGAATGGACACTAGAGC-3′ ) and 710-vpr-R4 ( 5′-6-FAM-ATTATGGCTTCCACTCCTGCCCAAGT-3′ ) . Briefly , virus-containing media was lysed by addition of 0 . 04% Triton X-100 and subjected to RT-PCR using OneStep RT-PCR kit ( Qiagen , Valencia , CA ) . The RT step was performed at 50°C for 30 min followed by an RT-inactivation step ( 95°C for 15 minutes ) and two-step PCR amplification ( 1 minute at 58°C and 15 seconds at 95°C ) for 25 cycles . The PCR product was diluted with water 5–100-fold to get the fluorescent signal into the linear range of the machine , ran on Applied Biosystems sequencing machines , and the data was analyzed with free PeakScanner 1 . 0 software ( Applied Biosystems ) . Due to different insertion lengths , products from Vpr-FS and Vpr-FS-StuI have different length and appear as distinct peaks . The area of each peak is calculated by the PeakScanner and is proportional to the relative amount of each PCR product . In an ideal population , the expected variation in neutral allele frequency , V , for a single generation depends on the initial allele frequency p and the population size N: In an ideal population that changes in size over time , the expected variance of a neutral allele frequency at generation n is: Because HIV populations in culture are growing exponentially , N1≪N2≪…≪Nn . Therefore , the majority of the variance is contributed by the first generation and To measure the actual variance in allele frequency we infected multiple ( 12 to 24 ) cultures of cells with 1∶1 mixture of viruses Vpr-FS and Vpr-FS-StuI . The virus was allowed to spread through the culture for 5–12 days until a majority of the cells were infected . At that point , cell-free virus was collected and the proportion of Vpr-FS-StuI virus was determined by the GeneScan assay . The variance in Vpr-FS-StuI frequency was calculated as , where n is the number of cultures , p0 is the frequency of Vpr-FS-StuI in the original 1∶1 virus mixture and pn is the frequency of Vpr-FS-StuI in culture n at the end of experiment . Variability within GeneScan assay itself also contributed to the observed variance in the Vpr-FS-StuI frequency . To account for that , we measured the assay variation by performing multiple assay replicates on a single randomly chosen sample . Assay variance was calculated as , where pn is the frequency of Vpr-FS-StuI in replicate n and is the average frequency in all replicates . This assay variance was subtracted from the total observed variance to obtain true experimental variance . To evaluate differences in observed variation between different experimental conditions , variance functions in combination with generalized linear models were used to test for differences in fold-change between the variance of each experimental condition in relation to the variance of the C8166 experiment [29] . The use of variance functions and generalized linear models allowed us to analyze the data from all experiments simultaneously , and to adjust for contributions to the overall observed variance due to differential population sizes . In addition , any correlation within data from the same experimental replicate is accounted for . This analysis provides estimates and 95% confidence intervals for the fold-difference in variation , adjusted for different population sizes within each experimental condition in relation to the C8166 experimental condition . Any experimental condition where the confidence interval does not contain the value of one has statistically significantly different variance than the C8166 experimental condition . In addition , the same test was conducted comparing the variance from the synchronized experiment to the single cycle experiment . | Genetic drift can be a strong evolutionary force , especially in small populations . Studies of HIV evolution within a single infected patient suggest that genetic drift plays an important role in the evolution of the virus , despite the large size of the viral population . The factors responsible for the high genetic drift are not known , but several models have been proposed to explain the phenomenon; all of them assume that the viral population is ideal . We measured the amount of genetic drift in HIV populations , replicating in the controlled environment of cell culture . We found that HIV populations exhibit approximately 10-fold more genetic drift than would be expected for an ideal population of similar size . Non-synchronous timing of infection is partially responsible for the increase , but other unidentified factors also contribute . While the increase in genetic drift observed for HIV in culture is not sufficient to explain the several orders of magnitude increase in intra-patient genetic drift , it provides strong experimental evidence for the intrinsic stochasticity of the HIV replication cycle . Understanding the sources of genetic drift is necessary to better understand the evolutionary forces that act upon HIV in vivo . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"virology/virus",
"evolution",
"and",
"symbiosis",
"molecular",
"biology/molecular",
"evolution",
"virology/immunodeficiency",
"viruses",
"virology",
"genetics",
"and",
"genomics/population",
"genetics"
] | 2009 | Genetic Drift of HIV Populations in Culture |
Military conflict has been a major challenge in the detection and control of emerging infectious diseases such as malaria . It poses issues associated with enhancing emergence and transmission of infectious diseases by destroying infrastructure and collapsing healthcare systems . The Orakzai agency in Pakistan has witnessed a series of intense violence and destruction . Military conflicts and instability in Afghanistan have resulted in the migration of refugees into the area and possible introduction of many infectious disease epidemics . Due to the ongoing violence and Talibanization , it has been a challenge to conduct an epidemiological study . All patients were sampled within the transmission season . After a detailed clinical investigation of patients , data were recorded . Baseline venous blood samples were taken for microscopy and nested polymerase chain reaction ( nPCR ) analysis . Plasmodium species were detected using nested PCR ( nPCR ) and amplification of the small subunit ribosomal ribonucleic acid ( ssrRNA ) genes using the primer pairs . We report a clinical assessment of the epidemic situation of malaria caused by Plasmodium vivax ( 86 . 5% ) and Plasmodium falciparum ( 11 . 79% ) infections with analysis of complications in patients such as decompensated shock ( 41% ) , anemia ( 8 . 98% ) , hypoglycaemia ( 7 . 3% ) , multiple convulsions ( 6 . 7% ) , hyperpyrexia ( 6 . 17% ) , jaundice ( 5% ) , and hyperparasitaemia ( 4 . 49% ) . This overlooked distribution of P . vivax should be considered by malaria control strategy makers in the world and by the Government of Pakistan . In our study , children were the most susceptible population to malaria infection while they were the least expected to use satisfactory prevention strategies in such a war-torn deprived region . Local health authorities should initiate malaria awareness programs in schools and malaria-related education should be further promoted at the local level reaching out to both children and parents .
Disease emergence is influenced by both natural and human factors . Among human activities , military conflicts characterized by war and regional tribal and/or sectarian strife have had huge impact by destroying infrastructure and collapsing healthcare systems . The affected region and the people therein face diverse short-term and long-term consequences . Large population displacements present higher risks of infectious diseases , lack of resources , which is often the case within crowded refugee camps with sanitation issues , increased exposure of the population to disease vectors , and destruction of healthcare systems lead to negative consequences [1] . Moreover , if the situation were to prolong , discontinued public and healthcare professional education investment and lack of proper surveillance and control of the disease should produce more severe and chronic outcomes . These outcomes make the people in the affected and nearby regions become newly vulnerable to a variety of communicable diseases and directly associated with the prevalence and emergence of the infectious diseases [1 , 2] . The prolonged Soviet war in Afghanistan between 1979 and 1995 annihilated the malaria vector control programs in the nation , which had been implemented in the 1960s and 1970s . The programs were so successful that Afghanistan had virtually eradicated the disease in the late 1970s [3] . The collapse of the program resulted in the re-emergence of malaria , and turned the region into a malaria endemic area . After the Soviet war , Afghanistan and the nearby Pakistani regions , specifically the Federally Administered Tribal Areas ( FATA , a semi-autonomous territory running along the Pakistan–Afghanistan border ) transformed into religious fundamentalism where tribe heads lost their authority and the rule of militant groups such as the Taliban started . Moreover , the subsequent 9/11 triggered occupation of the United States initiated a sizable population displacement of the refugees to FATA Pakistan . The border-crossing migration of Afghan refugees has overwhelmed the local public health system , and has caused an malaria epidemic [4] . Many epidemiology studies attributed the roughly 24%–36% increase in malaria prevalence in this region to the post-Soviet war influx of Afghan refugees into FATA [5] . Different researchers reported that permanent residents have low susceptibility and high immunity against malaria as compared to Afghan refugees [6] . The proportion of malarial cases due to P . vivax is increasing every year . According to the World Health Organization ( WHO ) report , approximately 75% ( previously 64% ) of the infections are transmitted through P . vivax , whereas 25% ( previously 36% ) are caused by P . falciparum , which are the two most prevailing species in Pakistan [7] . Currently , the P . vivax malaria is accounted in 70% of the malaria burden in Pakistan , and FATA being the most impoverished and extremely underdeveloped area in the Pakistan has the highest malaria burden due to the large Afghan refugees and IDPs [8] . Social development gages are appallingly low . There are only 41 hospitals for a population of 3 . 1 million . The Orakzai Agency ( Fig 1 ) has been one of the most neglected areas in FATA during the multi-decade-long military conflicts in the region . The Orakzai tribes have been linked to and have provided safe haven to the Taliban militant groups since late 2001 . A large part of the Orakzai Agency soon fell under control of the Taliban . The Agency was once home to Hakimullah Mehsud , the Tehrik-i-Taliban Pakistan chief who led militant operations , targeting hundreds of NATO supply vehicles in 2008 and 2009 . The regional medical centers and educational institutions were destroyed by repeated militant attacks [9] . This caused disruption to malaria management efforts as well as the increase of the malaria parasite reservoir in the region . For proper detection and control of malaria , obtaining and analyzing reliable epidemiology data is essential . To the best of our knowledge , however , there has been no report on malaria epidemics and clinical manifestations in Orakzai Agency . It has been a challenge to carry out such studies due to the severity of military conflicts . We report here the first study of epidemiology and clinical burden of malaria in this war-torn and Talibanized region , Orakzai Agency in Pakistan ( the sample collecting region is shown in Fig 1 ) . Keeping in view all the state of affairs and distribution of malaria epidemics in nearby FATA regions , the current study aimed to investigate the epidemiology and clinical manifestations of P . vivax and P . falciparum malaria among male and female patients of different age groups . We also report the analytical , epidemiological and clinical differences between P . vivax and P . falciparum infections .
Orakzai Agency ( FATA in Pakistan , Fig 1 ) is divided into the upper Orakzai and the lower Orakzai . Orakzai is a long neglected area with lack of basic health necessities , occasional engagement of armed insurgents , poor living conditions , limited access to vaccines , limited use of vector control measures , and unequal distribution of economic resources; most importantly , this is the case of approximately 60% of the FATA residents [10] . Furthermore , the medical centers and educational institutions have also been ruined by militant attacks . Healthcare-related non-government organization ( NGO ) activities are not permitted in FATA . In 2012 , according to UNHCR , nearly 758 , 000 internally displaced persons ( IDPs ) fled from their homes as a result of security operations in the FATA [11] . More recently , in June 2014 , Pakistani military initiated operation against militant groups in FATA resulted in approximately 450 , 000 IDPs displacement in the Bannu district [12] . Presently , FATA has the highest burden of different infectious diseases , due to the large Afghan refugees and IDPs . The population suffering from or at risk of contracting malaria significantly increased in the FATA as did the malaria parasite reservoir . This retrospective case-control study was conducted at the District Headquarters Hospital ( DHH ) Kalaya in Orakzai Agency , Pakistan between April 2011 and December 2013 . Patients presented with major clinical symptoms ( fever , headache , cough , dyspnea , vomiting , diarrhea , abdominal pain , and convulsions ) of malaria at different partially functional health care centers were referred to the DHH Kalaya . The major clinical symptoms of malaria were based on WHO criteria [8 , 13] . All children and adults presenting to hospital were screened for study eligibility and were hospitalized . A total of 216 microscopy-confirmed patients aged 1–60 years were evaluated in the study . Demographic and clinical records were collected upon enrollment , and baseline venous blood samples were collected for further biochemical and molecular analyses . Control group was collected at the same site as uncomplicated cases and used for statistical analyses . To exclude the confounding effect of sex , age and locality , control population was matched by sex , age and locality . Pregnant women were not included in the study . The inclusion criteria for patients were as follows: ( i ) prostration ( unable to sit ) , ( ii ) multiple seizures , ( iii ) impaired consciousness , ( iv ) multiple convulsions , ( v ) hyperpyrexia , ( vi ) anemia , ( vii ) decompensated shock , ( viii ) dark urine , ( ix ) jaundice , ( x ) hypoglycaemia , ( xi ) hyperparasitaemia , and ( xii ) respiratory problems [14] . In severe malaria , the level of impaired consciousness was assessed by computing the Glasgow Coma Scale ( GCS ) score ( <11 ) in adults or Blantyre Coma Scale Score ( <3 ) in children [14] . For further investigation , studied patients were divided into groups on the basis of their age and sex . After a detailed clinical investigation of patients , a standardized case report template was designed to compile the complete clinical data of each patient . For children younger than five years , their parents or relatives were asked of their medical history . Baseline venous blood samples were taken for microscopy and nested polymerase chain reaction ( nPCR ) analysis . The initial diagnosis of Plasmodium spp . infection was made by thick or thin smears . Two slides were made from each patient’s blood and both thick and thin films were prepared on the slides in the DHH Kalaya laboratory . Giemsa-stained thick blood smears of patients were examined using Giemsa stain and the parasitemia quantified independently by two skilled microscopists [15] . A thick smear was considered negative if no parasite was seen in at least 200 fields . For molecular analysis , the parasite DNA was extracted from filter papers using the Qiagen DNA extraction kit ( QIAGEN , Valencia , CA , USA ) , following to the manufacturer’s protocol . The Plasmodium species were detected using nested PCR ( nPCR ) and amplification of the small subunit ribosomal ribonucleic acid ( ssrRNA ) genes using the primer pair set A ( 5’-TTAAAATTGTTGCAGTTAAAACG-3’ and 3’-CCTGTTGTTGCCTTAAACTTC-5’ ) for the detection of P . vivax; primer pair set B ( 5’-CGCTTCTAGCTTAATCCACAT AACTGATAC-3’ and 3’-ACTTCCAAGCCGAAGCAAAGAAAGTCCTTA-5’ ) for the detection of P . falciparum; primer pair set C ( 5’-CTGTTCTTTGCATTCCTTATGC-3’ and 3’-GTATCTGATCGTCTTCACTCCC-5’ ) for the detection of P . ovale; and primer pair set D ( 5’-GTTAAGGGAGTGAAGACGA-3’ and 3’- TCAGAAACCCAAAGACTTTGATTTCTCAT-5’ ) for the detection of P . malariae . PCR reactions were carried out on a thermal cycler ( Nyx Technik USA ) , beginning with 5 minutes at 94°C , followed by 25 cycles of 45 seconds at 94°C , 45 seconds at 58°C , and 5 minutes at 72°C for the first round; 30 cycles of 45 seconds at 94°C , 45 seconds at 65°C , and 2 minutes at 72°C was then performed for the second round . The final cycle was followed by an extension time of 5 minutes at 72°C . The amplified PCR products were analyzed by 2%–2 . 5% agarose gel electrophoresis stained with ethidium bromide and visualized on the Bio-Rad gel doc system ( Bio-Rad Laboratories , Hercules , CA , USA ) . Due to the absence of well-equipped laboratory facilities in the DHH Kalaya , PCR and biochemical analyses were carried out at the Kohat University of Science and Technology and Kohat hospital . Statistical analysis was carried out on SPSS version 19 . Means , odds ratios with 95% CIs , and χ2 test of independence were calculated when applicable . Statistica ( version 12 ) was used for box-and-whisker plots . In all the studied parameters , a p value of ≤0 . 05 was considered statistically significant . Patients with prior comorbid conditions were excluded from relevant subanalyses , for example , diabetes mellitus patients were excluded from hypoglycemia analysis . All analyses were repeated after excluding all patients with associated infections and comorbid illnesses . Ethical approval for project activities was provided by the Kohat University of Science and Technology . Written informed consent was obtained from the patients and their parents/guardians before recruitment .
We collected 216 blood samples from febrile patients between April 2011 and December 2013 and screened for malaria . These febrile samples were collected from the DHH Kalaya in Orakzai Agency , Pakistan , where ordinary disease detection and control activities had been halted due to decade-long military conflicts . Many of the febrile patients were known to be displaced Afghan refugees , but such data was not recorded . A total of 216 blood samples were diagnosed positive by the microscopic examination . As baseline patient demographics are shown in Table 1 , 178 of 216 patients were identified by nPCR to have contracted malaria ( mean ± SD age 19 . 9 ± 11 . 9 years ) . Among these , we have found monoinfections of P . vivax and P . falciparum , as well as co-infections of both pathogens in diverse age groups ( Table 1 ) . In our study , 11 of 216 patients were presented with impaired consciousness ( GCS <11 or BCS <3 ) during hospitalization , however they were excluded from the study because all 11 patients showed associated infections and comorbid illnesses such as pneumonia . Furthermore , all 11 patients were not identified by nPCR to have contracted malaria . The diagnosis was initially made by the microscopic examination and nPCR . The microscopic examination identified 175 patients ( 81% ) to be infected by P . vivax , and 31 patients ( 14% ) by P . falciparum , and 10 patients ( 5% ) doubly infected by both P . falciparum and P . vivax ( Fig 2 ) . In contrast , the nPCR examination detected 154 ( 86% ) P . vivax infections and 21 ( 12% ) P . falciparum infections , respectively . This method detected 3 ( 2% ) double infections . However , P . ovale and P . malariae infections were not identified in any of the investigated samples by both test methods . Noticeable discrepancies between microscopic ( 216 patients ) and nPCR ( 178 patients ) detections were observed ( Fig 2 ) . Previous comparative diagnosis studies demonstrated nPCR to produce sensitive and reliable diagnosis results better than other methods including microscopy [16 , 17] . nPCR was acceptable to serve as the reference standard in malaria diagnosis . Therefore , we decided to rely on the nPCR diagnosis . We observed greater prevalence of Plasmodium infection in males ( 70% ) . The further clinical and biochemical tests of patients with P . vivax and P . falciparum infections ( Table 2 ) demonstrated the majority of the subjects ( 80%; n = 142 of 178 ) exhibiting severe malaria complications by World Health Organization criteria . As shown in Table 2 , comparable and similar rates of various complications were observed in both P . vivax and P . falciparum patients . Among 121 febrile patients with severe P . vivax infection ( Table 2 ) , the frequency of complications was as follows: decompensated shock ( n = 64; 53%; p = <0 . 001 ) , hypoglycaemia ( n = 12; 10% ) , anemia ( n = 12; 10% ) , hyperpyrexia ( n = 10; 8% ) , multiple convulsions ( n = 10; 8% ) , hyperparasitaemia ( n = 7; 6% ) , and jaundice ( n = 6; 5% ) . On the other hand , the following frequency of complications was observed in febrile patients with severe P . falciparum infection ( n = 21 of 178 ) : decompensated shock ( n = 9; 43%; p = <0 . 001 ) , anemia ( n = 4; 19% ) , jaundice ( n = 3; 14% ) , multiple convulsions ( n = 2; 9% ) , hypoglycaemia ( n = 1; 5% ) , hyperpyrexia ( n = 1; 5% ) , and hyperparasitaemia ( n = 1; 5% ) . The frequency of complications among all patients who tested positive with malaria ( n = 178 ) were as follows: decompensated shock ( n = 73; 41% ) , anemia ( n = 16; 9% ) , hypoglycaemia ( n = 13; 7% ) , multiple convulsions ( n = 12; 7% ) , hyperpyrexia ( n = 11; 6% ) , jaundice ( n = 9; 5% ) , hyperparasitaemia ( n = 8; 5% ) , and mixed complications ( n = 36; 20% ) more than one criteria . The most common malarial complication caused by P . vivax and P . falciparum was decompensated shock ( p = <0 . 001 ) . Decompensated shock symptom had the highest odds ratio ( OR ) for being reported in patients affected by both malarial species ( Table 2 ) . Hypoglycemia , multiple convulsions and anemia had an OR in the similar range in case of P . vivax infected patients whereas a different pattern of OR was observed in P . falciparum infected patients ( Table 2 ) . Although all other statistical associations held , the strength of association varied . The mean parasite count for P . vivax patients was ( 20241 . 9; p = 0 . 744 ) , which is significantly greater than that of P . falciparum patients ( 11848; p = 0 . 744 ) . The mean illness duration was 5 . 2 ± 2 . 0 days for P . falciparum male patients and 4 . 4 ± 0 . 7 days for females . Similarly , the mean illness duration for P . vivax male patients was 5 . 5 ± 1 . 6 days and 6 . 1 ± 1 . 3 days for female patients . The incidence of both vivax and falciparum malaria gradually increased between the ages of 1–20 years with increasing age ( Fig 3 ) . The prevalence of malaria reached its peak among late teenagers ( the age of 15–20 , see Fig 3 ) . It was also obvious that P . vivax infections were most prevalent in children of the age group between 5 and 15 years old as shown in Fig 3A , while P . falciparum infections in children populations were less prevalent as compared with P . vivax ( Fig 3B ) . However , the most noteworthy characteristic was the drastic decrease in malaria incidence in post-puberty males and females ( 21–60 years ) . This age-dependency was observed in both parasite species infections ( P . falciparum and P . vivax ) . High childhood malaria parasite exposure resulted in children ( 1–15 years ) bearing the brunt of the disease burden ( Fig 3 ) . This group turned out to be the most vulnerable .
Nonetheless , our study has its own limitations including lack of meticulous refugee statistics . Our study failed to estimate the true number of imported malarial cases in Orakzai Agency . Due to the presence of the armed Taliban insurgents in the Upper Orakzai , we were unable to collect the disaggregated data of Orakzai Agency . However , this study is the first report on the epidemic situation and clinical analysis regarding this most neglected region . Furthermore , in our study , children were the most susceptible population to malaria infection whereas they were the least expected to use satisfactory prevention strategies in such a war-torn deprived region . Local health authorities should initiate malaria awareness programs in schools and malaria-related education should be further promoted at the local level reaching out to both children and parents . We conclude that this overlooked distribution of malaria should be considered by malaria control strategy makers in the world and by the Pakistani government . | The malaria epidemic and endemic in Pakistan is a present and ongoing threat to public health which could have an impact in the nearby regions as well . For the first time , we report a clinical assessment of malaria endemicity in the Orakzai Agency , which is Pakistan’s most neglected area due to Talibanization and war in Afghanistan . Febrile patient blood samples of the area were investigated to report the clinical assessment of malaria caused by Plasmodium vivax and P . falciparum infections . The nested polymerase chain reaction ( nPCR ) examination detected 154 ( 86% ) and 21 ( 12% ) P . vivax and P . falciparum infections , respectively . We found worsening hygiene conditions in FATA , likely caused by poor socioeconomics and the collapse of the public health infrastructure . Decompensated shock was a common and prominent clinical feature of malaria among all the clinical presentations caused by both P . vivax ( 53% ) and P . falciparum ( 42 . 9% ) . Our results have significant implications for both public health and malaria control in FATA and Pakistan . Our findings illustrate higher prevalence of malaria in children compared to other age groups . Further research on sensible estimates of refugees is required , as well as resistance to anti-malarials . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [] | 2016 | Epidemiology and Clinical Burden of Malaria in the War-Torn Area, Orakzai Agency in Pakistan |
The polarization of nascent embryonic fields and the endowment of cells with organizer properties are key to initiation of vertebrate organogenesis . One such event is antero-posterior ( AP ) polarization of early limb buds and activation of morphogenetic Sonic Hedgehog ( SHH ) signaling in the posterior mesenchyme , which in turn promotes outgrowth and specifies the pentadactylous autopod . Inactivation of the Hand2 transcriptional regulator from the onset of mouse forelimb bud development disrupts establishment of posterior identity and Shh expression , which results in a skeletal phenotype identical to Shh deficient limb buds . In wild-type limb buds , Hand2 is part of the protein complexes containing Hoxd13 , another essential regulator of Shh activation in limb buds . Chromatin immunoprecipitation shows that Hand2-containing chromatin complexes are bound to the far upstream cis-regulatory region ( ZRS ) , which is specifically required for Shh expression in the limb bud . Cell-biochemical studies indicate that Hand2 and Hoxd13 can efficiently transactivate gene expression via the ZRS , while the Gli3 repressor isoform interferes with this positive transcriptional regulation . Indeed , analysis of mouse forelimb buds lacking both Hand2 and Gli3 reveals the complete absence of antero-posterior ( AP ) polarity along the entire proximo-distal axis and extreme digit polydactyly without AP identities . Our study uncovers essential components of the transcriptional machinery and key interactions that set-up limb bud asymmetry upstream of establishing the SHH signaling limb bud organizer .
An important step during the initiation of vertebrate organogenesis is the setting-up of morphogenetic signaling centers that coordinately control cell specification and proliferation . One paradigm model to study these processes is the developing limb bud and recent studies have revealed how morphogenetic Sonic hedgehog ( SHH ) signaling from the zone of polarizing activity ( ZPA ) and Fibroblast growth factor ( FGF ) signaling from the apical ectodermal ridge ( AER ) coordinate cell specification with proliferation along both major limb bud axes [1] . AER-FGF signaling mainly controls the establishment of the proximo-distal ( PD ) limb bud axis ( sequence: stylopod-zeugopod-autopod ) [2] , while SHH signaling by the polarizing region controls antero-posterior ( AP ) axis formation ( radius and ulna , thumb to little finger ) [3] , [4] . Cells receiving the SHH signal inhibit the constitutive processing of Gli3 to its repressor form ( Gli3R ) and upregulate the expression of the Gli1 transcriptional activator , which results in positive regulation of SHH target genes [5]–[7] . In limb buds of mouse embryos lacking Gli3 , the expression of initially posteriorly restricted genes such as Hand2 , 5′HoxD genes and the BMP antagonist Gremlin1 ( Grem1 ) expands anteriorly from early stages onwards and an anterior ectopic Shh expression domain is established at late stages [8] . However , the resulting digit polydactyly arises in a SHH-independent manner , as limbs of embryos lacking both Shh and Gli3 are morphologically and molecularly identical to Gli3 deficient mouse embryos [9] , [10] . These and other studies indicate that Gli3 acts initially up-stream of SHH signaling to restrict the expression of genes activated prior to Shh to the posterior limb bud [11] and that SHH-mediated inhibition of Gli3R production is subsequently required to enable distal progression of limb bud development [9] . The molecular interactions that polarize the nascent limb bud along its AP axis and activate SHH signaling in the posterior limb bud mesenchyme have only been partially identified . Previous studies implicated the basic helix-loop-helix ( bHLH ) transcription factor Hand2 ( dHand ) in these early determinative processes upstream of SHH signaling [1] . In particular , the development of fin and limb buds of Hand2 deficient mouse and zebrafish embryos arrests at an early stage and no Shh expression is detected [12] , [13] . This early developmental arrest in conjunction with massive generalized apoptosis of Hand2 deficient mouse limb buds precluded an in depth analysis of the molecular circuits and signaling systems that control initiation and progression of limb bud development . Furthermore , transgene-mediated over-expression of Hand2 induces digit duplications in mouse limb buds [14] . The functional importance of Hand2 as a transcriptional regulator in these processes was further corroborated by an engineered mutation that inactivates the Hand2 DNA binding domain in mouse embryos , which results in limb bud defects resembling the Hand2 null phenotype [15] . Cell-biochemical analysis showed that Hand2 interacts with so-called Ebox DNA sequence elements most likely as a heterodimer with other bHLH transcription factors such as E12 [16] , [17] and Twist1 , which is also required for early limb bud development [18] , [19] . Genetic analysis in mouse embryos showed that Gli3 is required to restrict Hand2 expression to the posterior limb bud mesenchyme as part of a mutually antagonistic interaction [11] . This interaction was proposed to pre-pattern the limb bud mesenchyme along its AP axis prior to activation of SHH signaling . However , the functional importance of this pre-patterning mechanism for normal progression of limb development remained unknown . Additional pathways are also required for establishment of the Shh expression domain in the posterior limb bud mesenchyme such as retinoic acid signaling from the flank and AER-FGF8 signaling [20] , [21] . During the onset of limb bud development , the expression of the 5′ most members of the HoxD gene cluster is restricted to the posterior mesenchyme by Gli3 [22] , [23] . During these early stages , the 5′HoxA and 5′HoxD transcriptional regulators are required to activate Shh expression in the posterior limb bud mesenchyme [24]–[26] . Consistent with this genetic analysis , the Hoxd10 and Hoxd13 proteins interact directly with the cis-regulatory region that controls Shh expression in limb buds [27] . This evolutionary conserved cis-regulatory region is called ZPA regulatory sequence ( ZRS ) and is located about 800 Kb up-stream of the Shh gene [28] . Genetic inactivation of the highly conserved core region of the ZRS ( termed MFCS1 ) results in limb bud-specific loss of Shh expression and a Shh loss-of-function limb skeletal phenotype [29] . Interestingly , this limb bud specific cis-regulatory region is absent from vertebrate species that have lost their limbs during evolution [30] . Transgenic analysis in mouse embryos revealed that ZRS-LacZ transgenes recapitulate major aspects of Shh expression in limb buds [28] . However , this study did not reveal specific cis-regulatory elements or sub-regions within the ZRS that regulate transcription , but rather indicated that the entire ZRS is required for correct Shh expression . A recent study shows that the ZRS interacts directly with the Shh transcription unit in both the anterior and posterior limb bud mesenchyme [31] . However , the Shh locus loops out of its chromosomal territory only in the posterior mesenchyme , which results in initiation of transcription . The evolutionary conserved function of the ZRS is underscored by an ever increasing large number of point mutations that are scattered through large parts of ZRS region and cause congenital preaxial polydactylies ( PPD ) in humans and many other mammals [32] . In summary , these studies establish that the far upstream ZRS cis-regulatory region controls Shh expression in different tetrapod species and that point mutations cause PPD , while deletion of the central part of the ZRS results in limbless phenotypes . We have generated a conditional Hand2 mouse loss-of-function allele and use it to study the requirement of Hand2 during limb bud initiation . Inactivation of Hand2 in the forelimb field mesenchyme using the Prx1-Cre transgenic mouse strain disrupts the development of posterior skeletal elements . Complete and early inactivation results in a limb skeletal phenotype identical to limbs lacking Shh . Indeed , establishment of the Shh expression domain in the posterior limb bud is disrupted and early molecular markers of posterior identity are lost , while anterior markers expand posteriorly . This reveals the early requirement of Hand2 for establishing posterior identity and activation of Shh expression . Using specific antibodies , we identify protein complexes containing both Hand2 and Hoxd13 transcriptional regulators in wild-type limb buds . Chromatin immunoprecipitation using Hand2 antibodies reveals the specific enrichment of the ZRS in comparison to adjacent non-ZRS DNA sequences in wild-type limb buds . Functional analysis of the DNA-protein interactions in cultured fibroblasts reveals that Hand2 and Hoxd13 transactivate expression of a ZRS-luciferase reporter construct , while this is partially inhibited by Gli3R , which has been previously shown to interact with 5′Hoxd proteins [33] . Indeed , mouse limb buds deficient for both Gli3 and Hand2 lack AP asymmetry along the entire PD limb axis and display severe digit polydactyly with complete loss of identities . Our study uncovers the interactions of Hand2 with the Gli3 and Hoxd13 transcriptional regulators and the far-upstream ZRS cis-regulatory region that are required to polarize the nascent limb bud mesenchyme and establish Shh expression in the posterior limb bud .
Mouse embryos lacking Hand2 die during mid-gestation due to cardiovascular defects and limb bud development arrests prior to formation of limb skeletal elements [12] , [34] . Therefore , we generated a conditional Hand2 loss-of-function allele by inserting two loxP sites into the locus ( “floxed” allele: Hand2f or H2f ) , which enables Cre-recombinase mediated deletion of the Hand2 transcription unit ( Figure S1 ) . Hand2 was inactivated in the limb bud mesenchyme ( H2Δ ˜Δc; Δc: conditional inactivation of the Hand2f allele ) using the Prx1-Cre transgene , which is expressed in the forelimb field mesenchyme from about E8 . 5 onwards ( 14 somites ) [35] , [36] . The inactivation of Hand2 was verified by monitoring the clearance of Hand2 transcripts and proteins in forelimb buds and mesenchymal cells ( Figure 1A and Figure S2A , S2B , S2C ) . Limb bud specific inactivation of Hand2 ( H2Δ ˜Δc; Figure 1A ) causes distal truncations of the forelimb skeleton and loss of the autopod ( Figure 1B ) . The skeletal phenotypes of Hand2 deficient forelimbs are variable , but the most severely affected cases ( 39% of all limbs , n = 80; Figure S3A , S3D ) are identical to Shh deficient limbs ( Figure 1B ) . Indeed , Shh expression and SHH signal transduction are lacking from a similar fraction of all H2Δ ˜Δc limb buds ( Figure 1C and Figure S3C ) . Therefore , the most severely affected H2Δ ˜Δc limb buds correspond to the limb-specific complete Hand2 loss-of-function phenotype ( Figure 1A–1C and Figure S3 ) . Between two and four digits form in hypomorphic H2Δ ˜Δc limbs ( Figure S3A , S3D ) as a likely consequence of residual Hand2 expression , which triggers SHH signal transduction ( Figure S3B , S3C ) . In the most severely affected forelimb buds , cells along the entire PD axis , but in particular in the distal-anterior mesenchyme are eliminated by apoptosis ( Figure 1D ) , which is distinct from the generalized apoptosis and developmental arrest of mouse embryos lacking Hand2 constitutively ( Figure S1D , S1E ) [12] . In H2Δ ˜Δc forelimb buds , cell death is limited to the core mesenchyme at embryonic day E10 . 0 ( Figure 1D , white arrowhead ) . In contrast , no significant apoptosis is detected in forelimb buds of wild-type and Shh deficient limb buds at these early stages ( Figure 1D , open arrowhead ) . Therefore , Hand2 is required for cell survival upstream of its role in activation of SHH signaling ( Figure 1D , left panels ) . During progression of limb bud development , the apoptotic domain expands distal-anterior in H2Δ ˜Δc limb buds and becomes similar to the cell death domain observed in Shh deficient limb buds ( Figure 1D , middle and right panels ) . In mouse embryos , hindlimb development is delayed by ∼12 hrs and activation of the Prx1-Cre transgene in the posterior mesenchyme is delayed by ∼24 hrs in comparison to forelimb buds [35] , [36] . The resulting ∼12 hrs delay in Hand2 inactivation at equivalent stages in the posterior hindlimb bud allows formation of an autopod with 4–5 digits , while the tarsal bones are always fused ( Figure 2A ) . Furthermore , inactivation of Hand2 specifically in the distal forelimb bud mesenchyme from E10 . 5 onwards no longer alters skeletal development ( data not shown ) . In agreement with the subtle skeletal alterations following Prx1-Cre-mediated Hand2 inactivation in hindlimb buds ( Figure 2A ) Shh remains expressed , albeit at slightly lower levels than in wild-types ( Figure 2B ) . Taken together , these studies show that Hand2 is essential to establish Shh expression in the posterior mesenchyme during initiation of limb bud development . Subsequently , it contributes to transcriptional up-regulation of Shh expression . Our further analysis focused on the most severe , complete Hand2 loss-of-function phenotypes in forelimb buds ( Figure 1 ) . The early essential requirement of Hand2 upstream of SHH in forelimb buds ( for cell survival , Figure 1D ) is further substantiated by molecular analysis , which reveals the lack of Tbx3 and Tbx2 expression [37] in the posterior mesenchyme of H2Δ ˜Δc forelimb buds . In contrast , their posterior expression is initiated but not up-regulated in ShhΔ ˜Δ forelimb buds ( Figure 3A and 3B ) . The expression of 5′HoxD genes is activated but not propagated in Hand2 deficient limb buds ( Figure S4A , S4B ) , likely due to the disruption of SHH signaling ( Figure 1C ) . Concurrently , the expression of anterior genes such as Cry-μ , Alx4 and Gli3 is ectopically activated or expands to the posterior margin in H2Δ ˜Δc forelimb buds earlier and/or more prominently than in ShhΔ ˜Δ limb buds ( Figure 3C–3E and Figure S4C ) . This loss of posterior and gain of anterior molecular markers reveal the early essential requirement of Hand2 for establishing posterior limb bud identity . This analysis ( Figure 1 , Figure 2 , Figure 3 ) led us to consider the possibility that Hand2 might directly transactivate Shh expression , possibly in conjunction with 5′Hox genes , which are essential for Shh activation in mouse limb buds [24] , [26] . Chromatin immunoprecipitation ( ChIP ) studies showed previously that Hoxd13 containing chromatin complexes are bound to the far up-stream ZRS cis-regulatory region that controls Shh expression in limb buds [27] . In addition , Hoxd13 is able to transactivate a ZRS-luciferase reporter construct in transfected cells [27] . Therefore , the potential direct interactions of Hand2 with Hoxd13 proteins and the ZRS were assessed by luciferase transactivation assays in NIH3T3 cells , which are mouse fibroblasts commonly used to analyze the SHH pathway [38] . A luciferase reporter construct encoding the entire ZRS ( ZRS-Luc ) was generated by inserting the ∼1 . 7 kb mouse ZRS region ( Figure 4A and Figure S5 ) [28] upstream of an adenovirus minimal promoter ( for details see Text S1 ) . The basal activity of this ZRS-Luc reporter construct was set to 1 and transfection of either Hand2 ( ∼3-fold ) or Hoxd13 ( ∼6 . 5-fold ) induced luciferase activity and their co-transfection resulted in an ∼10 . 5-fold increase ( Figure 4B ) . In silico analysis revealed 6 bona fide Ebox sequence elements within the ZRS ( Figure 4A and Figure S5 ) . Inactivating point mutations in either individual or several of these Ebox elements reduce the activity of the ZRS , but not in a strictly Hand2-dependent manner as the transactivation by Hoxd13 alone is also affected ( data not shown ) . As Hand2 and Gli3R act in a mutually antagonistic manner during initiation of limb bud development [11] , the potential effects of Gli3R on transactivation were assessed . As neither the Gli3 nor Gli1 activator forms are able to activate the ZRS-Luc reporter on their own ( data not shown ) , the ZRS likely lacks functional Gli binding sites [39] , suggesting that any effects of Gli3R would be indirect . Indeed , co-expression of Gli3R results in significant inhibition of transactivation in the presence of Hoxd13 ( Figure 4B ) , in agreement with the proposal that Gli3R can bind to and potentially antagonize Hoxd13 function [33] . In particular , Gli3R represses Hand2-Hoxd13 mediated transactivation of the ZRS-Luc reporter by ∼50% ( Figure 4B ) . The relevance of these interactions for limb bud development was determined by co-immunoprecipitation ( Figure 4C and Figure S6 ) and ChIP analysis ( Figure 4D and 4E ) . Immunoprecipitation of Hoxd13 proteins in combination with Western blotting reveals the existence of protein complexes containing both Hoxd13 and Hand2 protein in wild-type limb buds ( Figure 4C ) . The likely direct nature of these interactions is supported by efficient co-precipitation of epitope-tagged Hand2 and Hoxd13 proteins from transfected cells ( Figure S6 ) . These experiments establish that Hand2 interacts directly with Hoxd13 but not with Gli3R ( Figure S6 ) , which is relevant with respect to their genetic interaction ( see below ) . As the available polyclonal Hand2 antibodies specifically recognize and immunoprecipitate Hand2 proteins ( Figure S2B , S2C , S2D ) , ChIP on wild-type mouse limb buds was performed [40] to enrich Hand2 containing chromatin complexes and the analysis of three independent , fresh chromatin preparations is shown in Figure 4D and 4E . Conventional PCR using the amplicon “c” ( Figure 4A ) detected this ZRS region in chromatin precipitated with anti-Hand2 antibodies ( lanes α-H2 , Figure 4D ) , while no such amplification was detected when non-specific IgGs were used ( lanes α-IgG; Figure 4D ) . To further analyze this apparent association of Hand2 containing chromatin complexes with the ZRS , three amplicons ( “b” , “c” , “d” ) probing different regions of the ∼1 . 7 kb mouse ZRS ( Figure 4A ) were used for real-time PCR ( Q-PCR ) analysis . In addition , two amplicons located outside the mouse ZRS were chosen as likely negative controls ( non-ZRS amplicons “a” and “e” in Figure 4A and 4E and Figure S5 ) . Indeed , Q-PCR analysis revealed a minimally 14-fold enrichment of the amplicons located within the ZRS in comparison to the adjacent non-ZRS regions ( Figure 4E ) . This enrichment is specific as ChIP using non-specific IgGs resulted in much lower Q-PCR amplification of all five regions . In particular , the enrichment of the ZRS in comparison to flanking non-ZRS regions is highly significant ( amplicons “b” to “d” versus “a” and “e”; p = 0 . 0018 ) , while the variability among the three ZRS amplicons is not significantly different . Interestingly , the ZRS region encompassing amplicon “b” , whose enrichment is most variable , does not encode any bona fide Ebox elements ( Figure 4A and 4E ) . This provides additional evidence for the fact that the interaction of Hand2-containing chromatin complexes with the ZRS may not depend only on Ebox sequences . This ChIP analysis ( Figure 4D and 4E ) provides good evidence that the Hand2-containing chromatin complexes bind to the ZRS cis-regulatory region , but not to adjacent non-ZRS sequences . As embryos lacking Hand2 in limb buds survive to advanced stages ( Figure 1B ) , the functional relevance of the pre-patterning mechanism [11] can now be genetically investigated in Hand2 and Gli3 compound mutant ( H2Δ/ΔcGli3Xt/Xt ) embryos ( Figure 5 , Figure 6 , Figure 7 ) . In contrast to the Hand2 deficiency , H2Δ/ΔcGli3Xt/Xt limbs are severely polydactylous and display little phenotypic variability ( Figure 5A and Figure S7A ) . In addition , the zeugopodal bones and elbow joints appear strikingly symmetrical ( Figure 5A , white and black arrowheads in panel H2Δ ˜ΔcGli3Xt/Xt ) . These limb skeletal abnormalities are much more severe than the ones of Gli3Xt/Xt and ShhΔ ˜ΔGli3Xt/Xt limbs ( Figure 4A , panel Gli3Xt/Xt; see also [9] , [10] ) . While the skeletal elements of H2Δ ˜ΔcGli3Xt/Xt limbs seem to lack AP asymmetry , survival of the zeugopod and autopod progenitors is restored and the primordia are expanded in contrast to H2Δ ˜Δc limbs ( Figure S7B and data not shown ) . Moreover , the Sox9 expression domain , which marks the pre-chondrogenic lineage [41] , is expanded in H2Δ ˜ΔcGli3Xt/Xt limb buds that tend to be larger than normal ( Figure 5B , panel H2Δ ˜ΔcGli3Xt/Xt ) . However , no significant changes in proliferation were observed in H2Δ ˜ΔcGli3Xt/Xt limb buds ( data not shown ) . While the pre-chondrogenic condensations of all major skeletal elements are discernible by E10 . 75 in wild-type and Gli3 deficient limb buds , Sox9 expression remains diffuse and non-polarized in H2Δ ˜ΔcGli3Xt/Xt limb buds ( Figure 5B ) . During autopod development , the pool of Sox9 expressing digit progenitors is significantly expanded in H2Δ ˜ΔcGli3Xt/Xt limb buds in comparison to Gli3 mutants and wild-types ( Figure 5B; compare limb buds at E11 . 5 ) . The apparent symmetry of in particular the zeugopod in the H2Δ ˜ΔcGli3Xt/Xt limbs contrasts with the normal AP asymmetry in Gli3Xt/Xt and ShhΔ ˜ΔGli3Xt/Xt limbs ( Figure 5A ) [9] . This observation indicates that Hand2 and Gli3 participate in establishment of the AP asymmetry of the proximal limb skeleton independent of SHH signaling . Indeed , the expression of Runx2 , which marks proximal skeletal primordia [42] , is altered in double mutant limb buds ( Figure 5C ) . By E12 . 0 , Runx2 is expressed in the presumptive stylopod and zeugopodal domains of wild-type limb buds , while few Runx2 positive cells are detected in Hand2 deficient limb buds ( Figure 5C ) . In contrast , the Runx2 expression domain is expanded and lacks polarity in the proximal part of double mutant limb buds ( Figure 5C , black arrowheads ) . Taken together , these results indicate that the skeletal phenotypes and the severe polydactyly of H2Δ ˜ΔcGli3Xt/Xt limbs arise as a consequence of disrupting AP asymmetry ( proximally as indicated by Runx2 ) and aberrant expansion of the skeletal progenitor pools ( distally as indicated by Sox9 ) . In H2Δ ˜ΔcGli3Xt/Xt limb buds , Shh expression is not detected by in situ hybridization ( Figure 6A ) and its expression is ∼10-fold lower than in wild-types ( Figure 6C ) . Interestingly , the variability in Shh expression following Prx1-Cre mediated inactivation of Hand2 ( Figure 1C , Figure S3B , S3C , S3D , and Figure 6C ) is no longer observed in H2Δ ˜ΔcGli3Xt/Xt limb buds ( Figure 6A and 6C ) , which agrees with the lack of significant variability in the resulting skeletal phenotypes ( Figure 5A ) . This could be linked to the fact that posterior Shh expression is already reduced by ∼50% in Gli3Xt/Xt limb buds ( Figure 6A and 6C ) . The low Shh transcript levels detected in the most severely affected H2Δ ˜Δc and H2Δ ˜ΔcGli3Xt/Xt limb buds ( between 8% and 20% , Figure 6C ) likely reflect basal expression not detected by in situ hybridization ( Figure 1D , Figure 6A; see Discussion ) . BMP4-mediated up-regulation of its antagonist Grem1 in the posterior mesenchyme is essential to initiate the self-regulatory signaling system that promotes distal limb bud development [43] , [44] . In H2Δ ˜Δc limb buds , Bmp4 expression appears not significantly altered , while its expression is slightly reduced in H2Δ ˜ΔcGli3Xt/Xt limb buds ( panels Bmp4 in Figure 6B and 6C ) . In particular , the posterior expression domain in double mutant limb buds appears smaller ( arrowheads , panels Bmp4 in Figure 6B ) , which results in rather symmetrical Bmp4 expression along the AP limb bud axis . Furthermore , Grem1 expression is activated , but not up-regulated and distal-anteriorly expanded in Hand2 deficient limb buds ( panel Grem1 in Figure 6B ) , similar to Shh deficient limb buds [44] . In double mutant limb buds , the Grem1 expression domain appears symmetrical due to its anterior expansion . However , the rather variable Grem1 transcript levels are overall reduced in H2Δ ˜ΔcGli3Xt/Xt limb buds in comparison to wild-type and Gli3 deficient limb buds ( panels Grem1 in Figure 6C ) . Finally , the expression of the direct BMP transcriptional target Msx2 [43] is expanded in H2Δ ˜Δc limb buds , while its expression is significantly reduced in Gli3 deficient and double mutant limb buds as a likely consequence of the alterations in Grem1 ( panels Msx2 in Figure 6B and 6C ) . Taken together , these results corroborate the proposal that the initial phase of Grem1 expression in the posterior mesenchyme depends on BMP4 activity [43] . The rather symmetrical Grem1 expression in H2Δ ˜ΔcGli3Xt/Xt limb buds indicates that the second phase of SHH-dependent distal-anterior expansion of its expression in wild-type limb buds is a likely consequence of SHH-mediated inhibition of Gli3R activity [6] . The lack of discernible AP identities in the autopod of H2Δ ˜ΔcGli3Xt/Xt limb buds ( Figure 7A ) is confirmed by molecular analysis . In agreement with the rather symmetric distribution of Bmp4 and Grem1 in the distal limb bud mesenchyme ( Figure 6B ) , Fgf4 is expressed uniformly by the AER in double mutant limb buds ( Figure 7B ) . The distal expression domains of the Hoxd13 and Hoxa13 genes mark the presumptive autopod territory and are required for specification and expansion of the digit progenitors [45] , [46] . Within the distal mesenchyme of H2Δ ˜ΔcGli3Xt/Xt forelimb buds , the expression of Hoxd13 is anteriorly expanded and appears apolar in comparison to wild-type and Gli3 mutant limb buds ( Figure 7C; best seen in the apical views ) . In addition , the AP asymmetry of the distal Hoxa13 domain is also lost in double mutant limb buds ( Figure 7D; best seen in the apical views ) . The expanded and apolar expression of these genes ( Figure 7B–7D ) together with the alterations in Sox9 , Runx2 ( Figure 5B and 5C ) , Bmp4 and Grem1 ( Figure 6B ) reveal the striking loss of the asymmetrical expression of molecular and cellular markers of the AP axis along the entire PD axis in limb buds lacking both Hand2 and Gli3 .
Our biochemical analysis of chromatin isolated from wild-type mouse limb buds reveals that Hand2-containing chromatin complexes are bound to the ZRS , which is the far upstream cis-regulatory region required for Shh expression in limb buds [28] , [29] . In particular , ZRS sequences are specifically and significantly enriched in Hand2 containing chromatin complexes in contrast to flanking regions . Furthermore , Hand2 is part of Hoxd13 protein complexes in limb buds and in transfected cells , the two proteins transactivate the expression of a luciferase reporter gene in a ZRS-dependent manner . Albeit the fact that such transactivation studies are of somewhat artificial nature , the conclusions reached by this analysis completely agree with the results of our genetic analysis of Hand2 functions during mouse limb bud development . Early and complete genetic inactivation of Hand2 in limb buds disrupts establishment of the Shh expression domain in the posterior limb bud , while either incomplete or temporally delayed inactivation does no longer disrupt initiation of Shh expression ( this study ) . This reveals the early essential requirement of Hand2 for establishment of the posterior Shh expression domain , while subsequently Hand2 appears to contribute to transcriptional up-regulation of Shh expression . This may happen as part of an auto-regulatory loop because SHH signaling in turn up-regulates Hand2 expression most likely via repressing production of the Gli3R isoform [9] , [11] , [49] . The low levels of Shh expression detected by Q-PCR even in the most affected H2Δ ˜Δc and H2Δ ˜ΔcGli3Xt/Xt limb buds , but not in Shh deficient limb buds ( JDB and RZ , unpublished ) are indicative of basal transcription of the Shh locus in the absence of Hand2 , which is not detectable by in situ hybridization ( this study ) . This basal expression may depend on Hox transcription factors [24] , [26] or other regulators of Shh expression in limb buds ( see below ) . However , our study shows that Hand2 is essential to establish and upregulate Shh expression in the posterior mesenchyme , which defines the SHH signaling limb bud organizer [1] . This Hand2-mediated transactivation of Shh expression is a likely consequence of its direct interaction with the ZRS cis-regulatory region and is possibly enhanced by formation of transcriptional complexes with Hoxd13 protein in limb buds . Genetic and experimental manipulation of paired appendage buds in mouse , chicken and zebrafish embryos have begun to reveal the factors required in addition to Hand2 and 5′HoxD genes for Shh activation . In particular , AER-FGF and retinoic acid signaling have also been implicated in the activation of Shh expression [21] , [50] . Deletion of both the HoxA and HoxD clusters in mouse embryos disrupts Shh activation and causes early arrest of limb bud development such that the limb skeleton is truncated at the level of the stylopod [24] , [26] . But in contrast to Hand2 , loss-of-function mutations in these genes alone or in combination do not phenocopy the Shh loss-of-function limb skeletal phenotypes [51] , [52] . The Hand2 protein interacts with Hoxd13 and is part of the chromatin complexes bound to the ZRS in limb buds ( this study ) . However , other transacting factors will likely contribute to ZRS dependent activation of Shh transcription . In fact , the overlap of the Hand2 and Hoxd13 expression domains in the posterior limb bud mesenchyme is much bigger than the initial Shh expression domain . During limb bud initiation stages , the Hand2 and Gli3 expression domains overlap significantly , but then become rapidly mutually exclusive [11] . Therefore , these early dynamic changes in the expression domains of the Hand2 , Gli3 and Hoxd13 transcriptional regulators may well alter their interactions and spatially restrict the formation of transcription initiating/enhancing Hand2-Hoxd13 chromatin complexes at the ZRS to the posterior limb bud ( this study ) . These direct interactions would restrict the up-regulation of Shh expression to the posterior limb bud mesenchyme , thereby establishing the SHH signaling limb bud organizer . A recent study shows that the distant ZRS is in close proximity to the Shh transcription unit in both the anterior and posterior limb bud mesenchyme , but only loops out of its chromosomal territory in the posterior mesenchyme [31] . Interestingly , Shh is apparently transcribed by only a fraction of all ZPA cells at one particular time point , which indicates that the chromosomal conformation dynamics control Shh expression at the cellular level [31] . It is known that Hand2 binds DNA primarily as a heterodimer with E12 and/or the bHLH transcription factor Twist1 [16] , [19] . Interestingly , Twist1 is also required during early limb bud development [18] and point mutations in the human Twist1 gene alter its dimerization with Hand2 , which causes congenital limb malformations [19] . Therefore , these additional factors may also participate in regulation of Shh expression . The expression of Hand2 and 5′HoxD genes is activated in parallel , but then they converge functionally on the ZRS to establish the Shh expression domain in the posterior limb bud ( this study and ref . 24 ) . Furthermore , the establishment of the posterior Tbx2 and Tbx3 expression domains is disrupted in Hand2 deficient limb buds . The cis-regulatory elements controlling their expression are currently unknown , but it has been shown that Tbx2 expression requires the overlying non-AER ectoderm [53] . Additional experimental and genetic evidence indicates that Tbx2 and Tbx3 act likely upstream of Shh to restrict its transcriptional activation to the posterior limb bud margin [53] , [54] . In particular , ectopic expression of Tbx3 in early chicken limb buds induces an anterior shift of the entire limb bud together with transient anterior expansion of Hand2 expression [55] . These studies indicate that Tbx genes are part of the molecular circuits that position the limb bud , specify posterior identity and restrict activation of Shh to its posterior margin . The genetic inactivation of the pre-patterning mechanism in H2Δ ˜ΔcGli3Xt/Xt limb buds disrupts establishment of AP asymmetry and self-regulatory limb bud signaling [43] , while PD axis outgrowth and formation of all three major limb skeletal segments are the likely consequence of uniform AER-FGF signaling [2] . This results in a shortened and symmetric stylopod , zeugopod and a polydactylous autopod with highly dysmorphic digits . Similar to H2Δ ˜ΔcGli3Xt/Xt limb buds , limbs lacking 5′HoxD genes and Gli3 are also severely polydactylous but retain some polarity [56] , [57] . Therefore , the loss of AP polarity along the entire proximo-distal axis is more severe than the phenotypes observed in limb buds lacking Gli3 alone or in combination with genes such as Shh , Alx4 or 5′HoxD genes [9] , [56]–[58] . Over-expression of Hand2 in the entire limb bud mesenchyme results in a duplication of the anterior zeugopod ( ulna ) and posterior autopod ( digits ) [12] , which indicates that disturbing the balance between Hand2 and Gli3 either by gene inactivation or over-expression alters AP polarity . Therefore , the balance of the opposing activities of Hand2 and Gli3R in concert with 5′HoxD genes may control specification of the AP limb axis independent and up-stream of SHH signaling . In mouse limb buds lacking the Plzf zinc finger protein , 5′HoxD genes are uniformly expressed from early stages onwards and AP polarity is partially lost in combination hindlimb digit polydactyly [59] . It remains unclear why the digit polydactyly in H2Δ ˜ΔcGli3Xt/Xt forelimbs is more severe than the one of Gli3Xt/Xt ( and ShhΔ ˜ΔGli3Xt/Xt [9] ) forelimbs . However , in H2Δ ˜ΔcGli3Xt/Xt forelimb buds , the distal expression domains of Hoxa13 and Hoxd13 , which delineate the autopod territory and function in digit development ( see [refs . 24] , [26] for further detail ) are anteriorly expanded in comparison to Gli3 deficient limb buds . Such anterior expansion may point to an enlarged pool of autopod/digit progenitors , which could underlie the more severe digit polydactyly . As discussed before , this expansion of the Hoxa/d13 expression domains and the presumptive autopod territory are a likely consequence of the early loss of AP polarity along the entire PD axis in double mutant forelimb buds in contrast to Gli3Xt/Xt mutants . In particular , the H2Δ ˜ΔcGli3Xt/Xt forelimb skeletons bear some resemblance to the primitive paired appendages of Devonian fish and the polydactylous limbs of early tetrapods [60] . We shows that these rather “primitive” limb structures develop in the absence of pre-patterning ( Hand2 , Gli3 ) and the self-regulatory signaling system that interlinks the SHH , BMP and FGF signaling pathways , which are both key to normal limb skeletal development [1] . During tetrapod evolution , the symmetry of primitive polydactylous autopods from the Devonian period [61] was likely broken by beginning to set-up the regulatory interactions described in this study as they initiate posterior polarity up-stream or in parallel to their requirement for establishment of the SHH signaling limb bud organizer . The establishment of these transcriptional regulatory network acting upstream of SHH signaling might have enabled the development of the more refined and better functional pentadactylous limbs of modern tetrapods .
The generation of Hand2 conditional mutant mice is shown in Figure S1 . Hand2 mouse strains were kept in a mixed 129SvJ/C57BL6 genetic background . For details of the generation and analysis of Hand2 mice and embryos see Text S1 . For IP , fore- and hind-limb buds from E11 . 0 embryos were collected in PBS and lysed in lysis buffer ( Tris-HCl 10 mM pH 8 . 0; EDTA 1 mM; NaCl 140 mM; Triton 1%; SDS 0 . 1%; NaDeoxycholate 0 . 1% ) . Protein lysates ( about 300 mg ) were incubated overnight at 4°C with the anti-Hand2 ( M-19 , Santa Cruz; 1 mg ) and protein G beads were added the next morning for about 5 hours at 4°C . After several washes in lysis buffer , beads were resuspended in Laemmli loading buffer and SDS-PAGE was performed under non-reducing conditions . Goat IgG antibodies were used as control . For Co-IP of endogenous embryonic proteins , 50 limb buds at E10 . 5 were dissected in PBS and processed as described [33] . The Hoxd13 or control rabbit IgG antibodies used for co-IPs were covalently cross-linked to G protein beads and bound proteins were detected with Hand2 antibodies ( AF3876 , R&D System ) . ChIP was performed using wild-type fore- and hindlimb buds at E11 . 0 ( 38–42 somites ) . For each experiment , 85 limb buds were dissected , pooled and the freshly cross-linked chromatin divided among the starting samples . The average size of the DNA fragments in the cross-linked and sonicated chromatin was ∼500–2000 bp . Samples were processed as described [62] with the following modifications: protein G magnetic beads ( Dynabeads , Invitrogen ) were pre-absorbed with goat IgG ( 1–2 mg for 30 ml of beads for each sample ) for minimally 1 hour at 4°C . After washing them with BSA-PBS ( 5 mg/ml ) , the beads were added to the chromatin extracts and gently rocked for 1 hour at 4°C . Afterwards , beads were spun down and the chromatin in the supernatant transferred to a new tube and incubated overnight with Hand2 antibodies ( M-19 , Santa Cruz; 1 mg ) or goat IgG antibodies as control ( 1 mg ) . The following day , 25 ml of beads were added and the DNA-immunocomplexes were precipitated for 4 hours at 4°C . ChIP-enriched DNA samples were amplified by Q-PCR and conventional PCR . To compute the enrichment for a particular amplicon , its values were compared with the ones of a completely unrelated amplicon within the mouse β-actin gene that provides an additional negative control . The β-actin gene is located ∼114 Mb downstream of the ZRS on mouse chromosome 5 . The fold of enrichment was then calculated as the fold of increase in the specific signal in relation to the values obtained when using non-specific goat IgGs for ChIP ( values set arbitrarily at 1 ) . All oligos used are listed in Table S1 . Three ChIP experiments were performed using completely independent and fresh ( i . e . non-frozen ) chromatin preparations . The values obtained were analyzed and the graphs shown in Figure 4D ( means ± standard error ) were drawn using the Prism Graphpad Software ( La Jolla , USA ) . The statistical significance of all results was assessed using the Mann-Whitney test as part of the Prism software package . Mouse NIH3T3 fibroblasts were plated on 24-well plates and transfected using Lipofectamine LTX ( Invitrogen ) including a total of 500 ng of DNA . Reporter constructs were co-transfected with 100 ng of Hand2 and/or Hoxd13 and/or Gli3 expression constructs in combination with a Renilla luciferase vector . A detailed description of the generation of the expression constructs is available in Text S1 . Cells were collected 28–30 hours post-transfection and luciferase reporter assays were performed using the Dual Luciferase Kit ( Promega ) . Each assay was repeated at least 10 times . It is important to note that NIH3T3 cells do not express the endogenous Hand2 , Hoxd13 and Gli3 genes ( data not shown ) . For the co-immuno-precipitation assays in cells see Text S1 . | During early limb bud development , posterior mesenchymal cells are selected to express Sonic Hedgehog ( Shh ) , which controls antero-posterior ( AP ) limb axis formation ( axis from thumb to little finger ) . We generated a conditional loss-of-function Hand2 allele to inactivate Hand2 specifically in mouse limb buds . This genetic analysis reveals the pivotal role of Hand2 in setting up limb bud asymmetry as initiation of posterior identity and establishment of the Shh expression domain are completely disrupted in Hand2 deficient limb buds . The resulting loss of the ulna and digits mirror the skeletal malformations observed in Shh-deficient limbs . We show that Hand2 is part of the chromatin complexes that are bound to the cis-regulatory region that controls Shh expression specifically in limb buds . In addition , we show that Hand2 is part of a protein complex containing Hoxd13 , which also participates in limb bud mesenchymal activation of Shh expression . Indeed , Hand2 and Hoxd13 stimulate ZRS–mediated transactivation in cells , while the Gli3 repressor form ( Gli3R ) interferes with this up-regulation . Interestingly , limb buds lacking both Hand2 and Gli3 lack AP asymmetry and are severely polydactylous . Molecular analysis reveals some of the key interactions and hierarchies that govern establishment of AP limb asymmetries upstream of SHH . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"developmental",
"biology/embryology",
"developmental",
"biology",
"developmental",
"biology/morphogenesis",
"and",
"cell",
"biology",
"developmental",
"biology/pattern",
"formation",
"genetics",
"and",
"genomics/gene",
"function",
"developmental",
"biology/molecular",
"developme... | 2010 | Distinct Roles of Hand2 in Initiating Polarity and Posterior Shh Expression during the Onset of Mouse Limb Bud Development |
A new methodology termed Single Amino Acid Mutation based change in Binding free Energy ( SAAMBE ) was developed to predict the changes of the binding free energy caused by mutations . The method utilizes 3D structures of the corresponding protein-protein complexes and takes advantage of both approaches: sequence- and structure-based methods . The method has two components: a MM/PBSA-based component , and an additional set of statistical terms delivered from statistical investigation of physico-chemical properties of protein complexes . While the approach is rigid body approach and does not explicitly consider plausible conformational changes caused by the binding , the effect of conformational changes , including changes away from binding interface , on electrostatics are mimicked with amino acid specific dielectric constants . This provides significant improvement of SAAMBE predictions as indicated by better match against experimentally determined binding free energy changes over 1300 mutations in 43 proteins . The final benchmarking resulted in a very good agreement with experimental data ( correlation coefficient 0 . 624 ) while the algorithm being fast enough to allow for large-scale calculations ( the average time is less than a minute per mutation ) .
One of the most essential properties of all living organisms is the ability to conduct comprehensive “communication” between its individual components . This includes signal transduction , immune system operation , inhibition or activation of particular functions , assembly of macromolecular structures into molecular machines ( such as ATPase ) , and much more . At the molecular level such communications are carried out via macromolecular binding [1 , 2] . The molecular recognition is affected by multiple factors such as concentration and compartmentalization of the macromolecules , their shapes , charge distribution , conformational flexibility , physico-chemical properties of the interfaces and many others [3–11] . Any change of these characteristics could alter the wild type protein binding and therefore might affect the function of macromolecules . While some of abovementioned factors ( macromolecular and salt concentrations , pH and temperature of the media , etc . ) are results of the cellular function , other characteristics ( physico-chemical properties of interfaces , protein charge distribution , etc ) are largely determined by protein amino acid sequence and structure . Because of that , any alteration of the protein primary structure ( insertion , deletion or amino acid substitution ) may have an effect on macromolecular recognition . Having in mind that in vivo interactions occur in the crowded cellular environment , mutations may not only impact binding affinity but also could perturb protein interaction networks resulting in a loss or gain of interactions . Such changes in binding and interactions are frequently implicated in diseases and understanding of their molecular mechanisms is crucial for deciphering the origin of diseases . In particular , the effect of mutations on binding free energy ( binding affinity ) is considered to be an important component of the overall disease effect [12] . The effect of missense mutations on protein-protein complex formation can be experimentally assessed by various techniques such as isothermal titration calorimetry [13] , FRET [14] , surface plasmon resonance [15] , and many others ( see review [16] ) . However they are time-consuming , expensive to carry out and cannot be applied on a large scale . Despite such limitations , investigators have performed many mutagenesis experiments in the past to determine the effects of point mutations on binding free energy . The results reported in the literature were compiled into useful databases , the most prominent one being Skempi database [17] . Although most of these experiments were carried out on protein complexes that were either easy to manipulate biochemically or were of particular interest for the molecular biology community at that time , still such databases can be considered representative for any other interactions since the biophysical principles governing the binding should be universal . Therefore , these experimentally determined binding free energy changes caused by point mutations can serve as an ultimate benchmark for computational methods aiming at in silico predictions . Obviously , large-scale studies of the effects of mutations on protein-protein binding require computational approaches . Roughly speaking , the existing computational methods can be divided into two main categories: sequence-based and structure-based approaches . The main advantage of sequence-based approaches is that they are fast , but the techniques used for the predictions strongly depend on the training set of data [18] and may be over-fitted [19] . On the other part of the spectrum are structure-based approaches , many of them providing a qualitative estimate ( beneficial/neutral/deleterious ) of the changes in binding affinity upon mutations [20] . Multiple approaches in this category utilize different scoring schemes , solvent models ( implicit/explicit models ) , number of representative structures used in the analysis , Monte Carlo and molecular dynamics sampling methodologies , etc . ( for some examples see [21–29] ) . Among the structure-based approaches the most rigorous ( theoretically exact ) methods are the free energy perturbation ( FEP ) and thermodynamic integration ( IT ) methods [30] . However , they require intensive calculations and cannot be applied for large-scale modeling ( see review [31] ) . Among the structure-based methods , the Molecular Mechanical Poisson-Boltzmann ( Generalized Born ) / Surface Accessible ( MM/PB ( GB ) SA ) approach [32–34] represents a reasonable balance between computational time and details of the modeling . In this approach the binding free energy is calculated as a linear combination of potential energies such as molecular mechanics , polar and non-polar solvation energies . Similarly one can construct a function made of linear combination of weighted terms , either statistically or empirically delivered , to predict binding free energy and the change of it due to mutations [21 , 35] . Hybrid approaches do exist as well [24 , 25] . Some of these approaches emphasize on the importance of taking into account structural ensembles in the modeling [25] , others on the role of water phase and solvation energy [24] . In this paper we introduce a new methodology termed Single Amino Acid Mutation based change in Binding free Energy ( SAAMBE ) , which takes advantage of both approaches: sequence- and structure-based methods . It utilizes MM/PBSA approach along with an additional set of statistical terms delivered from statistical investigation of the physico-chemical properties of protein complexes . The new method was tested against more than 1300 mutations in 43 proteins and resulted in a very good agreement with experimental data ( correlation coefficient 0 . 624 ) while being fast enough to allow for large-scale calculations ( the average time is less than a minute per mutation ) .
As described in the method section , several knowledge-based terms were tested to improve the correlation between predicted and experimental ΔΔG . One of these terms was added in the SAAMBE formula to mimic the effect of the change of conformational entropy caused by mutations ( ΔΔS term ) . Others—because of our previous work as InterfaceMT term in Eqs ( 4 and 8 ) [24] . The third set of terms was introduced in SAMMBE formula due to extensive testing of various physico-chemical characteristics as hydrophobicity ( ΔΔHYDR ) , hydrogen bonds ( ΔHB ) and normalized change of the interface area caused by mutations ( ΔΔSASAInterfaceMT ) . It is understood that there is an overlap between some of these terms and the terms within MM/PBSA-based method , and between themselves alone as well . Hydrogen bond change is partially accounted for in MM/PBSA algorithm via the electrostatic energy term . The InterfaceMT and ΔΔSASAInterfaceMT are also related . However , the overlap is not complete as shown by the provided p-values ( Table 1 ) . The functional form of knowledge-based terms was optimized by trying various forms as explained in the method section . Their optimized forms are the one shown in Eqs ( 8 ) – ( 12 ) . The experimentally measured changes of the binding free energy caused by mutations vary from zero to very large positive values ( +8 . 803 ) and very small negative values ( -3 . 786 ) . It can be anticipated that there may be some structural or sequence characteristics associated with the magnitude of the binding free energy change . To test such a possibility , we first provide the distribution of the absolute changes of experimental binding free energy in sDB dataset ( Fig 2 ) . It can be seen that the cases with absolute binding free energy change of less than 1kcal/mol account for about 50% of the cases . Therefore we chose to split the whole database into two sets with similar number of entries: one set with small effect ( |ΔΔG|<1kcal/mol ) ; and another with large effect ( |ΔΔG|≥1kcal/mol ) . The next step was to determine the probability of mutations to cause “small effect” or “large effect” depending on two characteristics: amino acid type and location of the mutation site at the interfacial regions . With regard to amino acid types , we will consider WT and MT separately as explained below . With regard to interfacial location , we use the definitions provided in the Method section ( COR , SUP , RIM , INT and SUR ) . Furthermore we collect all available substitutions M of a given type X → any residue , where X is a particular amino acid ( for example , Ala , Arg , etc ) . Then we calculate the mean and variance of experimental change of the binding free energy for these M cases . In addition , we introduce an estimation of the probability ( P ) of mutation type X → any to cause large effect by: P ( X→any ) =MlargeM ( 1 ) where Mlarge is the number of cases within M subset for which the absolute change of the binding free energy is larger than 1kcal/mol ( large effect ) ( Fig 3 , left panel ) . Similarly we perform the same analysis for substitutions of ( any → X ) and define the corresponding probabilities P ( any → X ) ( Fig 3 , right panel ) . With respect to mutation site location , we select all available cases K for which the mutation site in the WT is located at Y , where Y is either COR , SUP , RIM , INT or SUR . Then we define a probability of mutations within K to cause large effect as: P ( Y , WT ) =KlargeK ( 2 ) where Klarge are the cases experimentally found to result in absolute binding free energy change larger than 1kcal/mol ( Fig 4 , left panel ) . Since mutations involve amino acids with different side chain length and MT and MT structures are subjected to energy minimization , it is quite likely that mutation site location is different in MT compared with WT . For this reason , the same analysis is done for the MT and the corresponding probabilities are defined as P ( Y , MT ) ( Fig 4 , right panel ) . Fig 3 indicates that there is a tendency for some types of substitutions to cause small , while other to cause large effects on the binding free energy . It can be seen that most of substitutions for Tyr and Gly in WT ( P > 0 . 7 ) cause a big change of binding free energy . Consistent with our previous work [24] , mutations to Pro and Gly also often ( P > 0 . 7 ) cause large changes in binding free energy . These results are not surprising since Tyr is a bulky aromatic polar residue . Two effects may be involved in stabilization of the WT structure by this amino acid: formation of hydrogen bond with other charged/polar residues and noncovalent interactions with aromatic rings of other residues such as Trp and Phe , known as “stacking effect” . Two other residues , Pro and Gly , are considered to be special in terms of their physico-chemical characteristics . Although both of them are most often found in a coil rather than in a sheet or strand , they perform different structural roles . Namely , Gly makes the secondary structure more flexible , while Pro tends to rigidify it . Pro is also a well-known secondary structure element breaker—it forms a turn when being introduced in a helix or a strand . The mutation site location also shows distinctive trend ( Fig 4 ) . There is almost linear correlation between the mean of the absolute binding energy change and the probability ( Eq ( 2 ) ) . Thus , the probability of a mutation located at mutations site , both in WT and MT , to cause large change of the binding free energy gradually increases: SUR → INT → RIM → SUP → COR . These observation and the corresponding probabilities can be used to guide SAAMBE predictions . However , before proceeding further with these possibilities , we should analyze the results presented in Figs 3 and 4 . Essentially , four “flags” were identified with four associated probabilities: residue type in WT and P ( X → any ) , residue type in the MT and P ( any → X ) , mutation site in WT and P ( Y , WT ) and in MT and P ( Y , MT ) . Therefore , a consensus scheme must be developed in order to incorporate these quantities into the SAAMBE algorithm . Further refinement of the classification scheme was done by altering the associated ΔΔGi for cases for which there is no agreement between the four “flags” . For example , if a given mutation Q → P in “k” case in sDB with experimentally determined |ΔΔGk| = 10kcal/mol and the mutation sites are in COR in WT and in SUP in MT . From Figs 3 and 4 the corresponding probabilities of causing strong effect are: P ( Q → any ) = 0 . 2 , P ( any → P ) = 0 . 86 , P ( COR , WT ) = 0 . 68 and P ( SUP , MT ) = 0 . 56 . Based on these probabilities , one expects that any mutation from Q will have little chance to cause strong effect ( P ( Q → any ) = 0 . 2 ) , but the specific case of Q → P was experimentally found to result in a large change ( |ΔΔGk| = 10kcal/mol ) . It can be speculated that this large effect is not caused by the WT residue type , Q residue , but because of the mutant residue P and the location of mutation site . Because of that we will alter the corresponding |ΔΔGk| with respect to each of the 4th flags by applying the following formula: |ΔΔGkaltered| ( forPiset ) ={23⋅∑j=1 , i<>j4Pj⋅|ΔΔGk| , |ΔΔGk|<123⋅∑j=1 , i<>j4 ( 1−Pj ) ⋅|ΔΔGk| , |ΔΔGk|≥1} ( 3 ) where Pj stands for: P1 = P ( Q → any ) , P2 = P ( any → P ) , P3 = P ( COR , WT ) , and P4 = P ( SUP , MT ) . These alterations are done for each entry in sDB and for each set of flags . In the entry “k” , original |ΔΔGk| is larger than 1kcal/mol and therefore in the particular case considered above the second row formula is applied . If the original experimental binding free energy change is smaller than 1kcal/mol , the first row formula is applied . To further quantify the applied alterations , we would like to point out that in the extreme case when all three probabilities are 0 . 5 ( i . e . the initial statistical analysis of sDB shows that the type of mutation has equal chance to cause large and small effect ) , applying Eq ( 17 ) will result in no alteration ( no change ) . The resulting set of |ΔΔGaltered| is termed altered dataset and subsequently was used to recalculate the probabilities P ( Table 2 for residue types and Table 3 for the mutation location ) . The results are shown in Figs 3 and 4 as well . These probabilities and classifications will be used to improve the performance of SAAMBE method . Given a particular mutation ( for example , Q → P ) and its location at the interface ( for example COR in WT and SUP in MT ) we calculate the probability of the mutation to cause large effect as: P=P ( P→any ) +P ( any→A ) +P ( COR , WT ) +P ( SUP , MT ) 4 ( 4 ) Thus , if P ≥ 0 . 5 the mutation is classified as a mutation expected to cause large change of the binding free energy . Otherwise , the mutation is expected to cause a small change . Thus , the final refinement of SAAMBE method is to take advantage of estimated probabilities . For each entry in the tDB we calculated the average probability P and split the database into tDB_small ( P < 0 . 5 ) and tDB_large ( P ≥ 0 . 5 ) . For each of subsets we calculated the change in binding free energy ( Eq ( 12 ) ) and obtained the optimal coefficients of each energy terms in SAAMBE by multiple linear regression analysis . This resulted in two sets of SAAMBE coefficients ( Table 1 ) . For comparison we also provide the optimized weights and the correlation coefficient for the total tDB as well ( Table 1 ) . Comparing the weight coefficients in Table 1 , one can see that there are some energy terms that are important for both subsets ( such as EE , VE , SP , IE , entropy and Interface ) . Most of the mutations in the sDB_small are non-interfacial ( for more than 30% of this subset the WT residue is located in the INT or SUR ) and solvent exposed ( ~50% in RIM ) . Based on the magnitude of the weight coefficients , one can speculate that the changes of the binding free energy might be caused by the slight reorganization of the whole protein-protein complex that is reflected in the ΔΔSASAInterface component energy term as well as the change in nonpolar component of salvation energy ( SN ) . On the other hand most of the mutations in the sDB_large are located at the interface ( 95% are in COR , 5% in SUP area ) . In addition to other energy terms , for the cases of sDB_large , the change in hydrogen bonds network and the change in hydrophobicity also play significant roles . Thus , adding such features into the SAAMBE protocol , namely having different weight coefficients in the SAAMBE formula for mutations expected to cause small/large effect on the binding free energy change , increases the correlation coefficient from 0 . 575 to 0 . 624 ( see Table 1 and Fig 5 ) . To evaluate the performance of the SAAMBE method we analyzed six ROC parameters . The results obtained by the SAAMBE algorithm were compared with those calculated by FoldX and BeAtMuSiC methods for the same tDB . According to the Table 4 the number of true positive predictions is twice as high for the SAAMBE as for the other two algorithms . The total number of false predictions is much smaller for SAAMBE . This indicates that SAAMBE outperforms FoldX and BeAtMuSiC by all six ROC parameters using tDB as a benchmark . In terms of numbers , SAAMBE benchmarking results in: sensitivity , or true positive rate , ( 0 . 87 ) ; NVP , or negative predictive value , ( 0 . 84 ) ; method accuracy ( 0 . 9 ) ; and MCC ( 0 . 84 ) . This proves that SAAMBE can predict with high accuracy not only the direction of the change in binding free energy , but also its magnitude . SAAMBE method was developed and optimized to predict the change of binding free energy for a broad range of mutation types . In this subsection we would like to address the question of how SAAMBE protocol can handle special cases: a ) when the bulky residue is substituted with the small one; b ) when the MT residue is Ala , which is typically used for protein “hot-spot” prediction; and c ) the ability to accurately predict the effect of mutations being in a particular location . We will also compare our results with those delivered from FoldX and BeAtMuSic methodologies ( see Table 5 ) . One of the main considerations in developing SAAMBE algorithm was the requirement of the predictions to be made in reasonable time . We tested the time of the algorithm execution for all entries in sDB . The average time was 0 . 21953 min for one mutation calculation ( SE = 0 . 00316min ) when employing 16 nodes for WT- and MT-complexes minimization and single node for the rest of calculation on Clemson University Palmetto Supercomputer ( http://citi . clemson . edu/palmetto/ ) . We also analyzed the effect of particular parameters such as the number of residues in the complex and the largest dimension for the WT-complex on the time of calculations . It was found that the shape of the protein has no impact on the time of calculations . However the total number of residues in the complex affects the total calculation time ( Fig 6 ) . One can see that the dependence of time of algorithm execution vs the total number of residues in the WT-complex can be described with polynomial ( second power ) function ( R = 0 . 99 , 81 points ) . The free coefficient is 5 . 59E-2 min , the linear and quadratic weights are -5 . 13E-5 min and 1 . 18E-6 min respectively . In this work we described a development of a method , the SAAMBE method , to predict the binding free energy changes caused by single mutations . In developing the method , we were particularly interested in using structural information in conjunction with other types of information . This was motivated by the goal to deliver not only correct predictions of the energy changes , but also to be able to offer an explanation of the reason for the effect . Thus , the algorithm has structure-related components , such as hydrogen bonds , interface area , and interface area change . In addition , the MM/PBSA-based components indicate the importance of the direct interactions to the predicted energy changes . Thus , for any predictions , one can qualitatively describe what the major driving effects are . Furthermore , these energy changes can be compared with experimentally observed quantities or with observation delivered from more rigorous methods as FEP or IT . The essential component of this investigation and development was the treatment of the plausible conformational and ionization changes induced by the binding . It is well understood that the binding introduces conformational and ionization changes , in some cases very small ( almost rigid body binding like lock and key ) , in other cases large conformational changes ( induced fit mechanism ) [37–40] . Some of these changes occur far away from the binding interface and typically involve surface groups [37–40] . However , modeling such conformational changes is not trivial , especially if one aims at relatively fast predictions . Our attempts to model the plausible conformational changes induced by the binding via relatively short MD simulations were unsuccessful . Perhaps longer MD simulations complemented with enhanced sampling techniques are needed , but this is computationally too costly for large-scale predictions . Instead of explicit modeling of conformational and ionization changes induced by the binding , we extend our previous approach to model them in electrostatic calculations via amino acid specific dielectric constant [36] . The motivation is based on the understanding that charged residues have the largest effect on electrostatic potential via their charges and ability to adopt different rotamers in response to the electrostatic field or to change their ionization states . Therefore , charged residues should be modeled with a large dielectric constant . Similarly , polar residues are the second in the list , since they have strong dipole moment and can participate in various hydrogen bonds . The rest of the amino acids , mostly hydrophobic residues , do not have many polar atoms and are typically buried in protein interior ( and therefore packed and not able to sample different rotamers ) and should be modeled with low dielectric constant [36] ( for more details see Figs A and B in S1 Text ) . Indeed , the development reported in this work confirmed the applicability of such an approach and significantly improved the performance of SAAMBE method . Using DelPhi capability to assign different dielectric constants for different amino acids , we demonstrated that charged , polar and other residues should be modeled with dielectric constants 9 , 8 , and 7 , respectively . This proves to be very effective and computationally inexpensive approach to mimic conformational flexibility in the framework of continuum electrostatics . The SAAMBE method is a formula made of linear combination of terms: energy , empirical or statistical terms . The quantities or the physical phenomena described by some of them partially overlap , which can be considered as double-counting . However , the statistical analysis ( p-values in Table 1 ) indicates that their values are acceptable ( for more details see Tables A-D in S1 Text ) . Thus , while there is partial overlap for some terms , because of the simplifications made in modeling these phenomena , different terms capture different components of the process and thus they are almost independent . The weight coefficients in the SAAMBE method were optimized against experimentally determined binding free energy changes of the tDB set . Therefore , the prediction accuracy depends on the training dataset and cases to be tested . It is anticipated that if the newly identified cases to be predicted by SAAMBE protocol do not deviate much from the cases in sDB/tDB , the predictions will be quite accurate . However , it is quite possible as well , that a new case is very different from the cases in sDB/tDB and then the prediction may not be accurate . We plan to continue enriching sDB/tDB and re-adjust the weight coefficients ( if needed ) of SAAMBE method and taking advantage of the computational cost to implement SAAMBE into a webserver .
We compiled a dataset , containing experimentally measured values of changes in binding free energy of protein-protein complexes due to single amino acid substitutions , by combining three sets of data mentioned in the following references: [25] , [23] ( Ala scanning database ) , and Skempi database [17] . To avoid the redundancy , all entries in the initially combined data set were screened to identify identical cases and only one representative was retained in the dataset . Then the dataset was further purged with respect to the experimental value of the binding free energy change . Thus , when several experimental values were available for the same mutation in the same protein-protein complex , and the experimental data variation was smaller than 1 . 5 kcal/mol ( the threshold was empirically selected ) , the entries were fused and the averaged value for the change of the binding free energy was used . If the variation was larger than 1 . 5kcal/mol , the entry was deleted . Furthermore , mutations located in structurally disordered protein segments ( missing coordinates in the PDB file ) were removed from the dataset as well . As a result , the final compiled dataset was comprised of 81 different proteins with the total of 2041 single point mutations . This dataset will be used for the statistical analysis of experimental data and will be referred to as sDB hereafter . However , to construct a dataset for training and testing , we further pruned the entries to remove all structures having heteroatoms ( crystallographic water molecules were not considered heteroatoms ) . The motivation was that while some compounds listed in the heteroatoms section of PDB file may be biologically important , the vast majority of them are crystallographic artifacts ( as ions for example ) . Thus , the resulting pruned database ( tDB ) consists of 1326 single point mutations from 43 proteins . Both datasets are available for download from ( compbio . clemson . edu/databases/sDB , tDB . xlsx ) . We assigned the location of mutated residues in the protein-protein complex based on five categories ( COR , SUP , RIM , INT and SUR ) as previously described [41] by computing the relative solvent accessible surface area ( SASA ) ( the ratio between SASA of a residue in protein and in water ( rSASA ) ; rSASA = 1 corresponding to totally exposed residue in the protein ) of the residue in the monomeric ( rSASAm ) and complex ( rSASAc ) states , as well as their mutual difference ( ΔrSASA = rSASAm − rSASAc ) . Thus residues are considered to be at the interface if they are in COR , SUP and RIM regions; and are away from the interface if they are in SUR and INT regions . RIM and SUR locations indicate that the residue is exposed to the water solvent when the complex is formed . The parameters of each location types are provided in Table 6 . The solvent accessible surface area of a residue was calculated with NACCESS software [42] . The initial crystal structures of the protein-protein complexes were obtained from the Protein Data Bank ( PDB ) [43] . Biological units were retrieved and only chains that belonged to the binding partners were retained for further calculations . Since the initial crystal structures might have regions with missing coordinates , we used the profix module from Jackal package to rebuild these regions [44] . It was done using default parameters and selecting “heavy atoms model” option . At the next step we applied the scap module from the same Jackal package to substitute wild-type residue with the mutant to generate the mutant ( MT ) -complex . To eliminate inconsistency that might be associated with applying scap software we also substituted wild-type residue with the same residue using scap to generate the wild-type ( WT ) -complex . To run scap we applied the following parameters: ( a ) CHARMM22 force field parameters , ( b ) large side-chain Jackal rotamer library was selected for the side-chain refinement , and ( c ) predictions were made applying the scap option utilizing 3 initial structures . Once the WT and MT structures were generated , the missing hydrogen atoms were added to the structures with VMD software ( version 1 . 9 . 1 , topology file from CHARMM27 force field ) [45] . Both WT- and MT-complexes were subjected for independent structural refinement by NAMD ( version 2 . 9 , CHARMM27 force field parameters ) [46] . For the minimization procedure we used Generalized Born implicit solvent model ( GBIS ) , implemented in NAMD . The dielectric constant of the implicit solvent was set to be 80 , and 1 for the protein ( various protein dielectric constants were tested—see Result section ) . We used quick N-steps ( optimum value for N was found to be 5000 , see Result section ) conjugate gradient algorithm implemented in NAMD to obtain the relaxed configuration with optimized geometric and steric clashes . The energy-minimized structures of WT and MT complexes were used to calculate all energy components for both the complex ( bound molecules ) and monomers ( unbound molecules ) . Typically such an approach is refereed as to rigid body approach . The binding free energy was calculated based on modified MM/PBSA method combined with knowledge-based energy terms . The individual energy terms are combined via weighted linear function , typically referred as to linear interaction energy ( LIE ) formula or scoring function . Here we chose to term the method as Single Amino Acid Mutation based change in Binding free Energy ( SAAMBE ) method . It has two major components: ( a ) energy components calculated with MM/PBSA technique and ( b ) knowledge-based terms delivered from statistical analysis of entries in sDB . In developing the SAAMBE protocol , we first define the terms ( E ) that will be used in SAAMBE protocol as follows: ΔΔE= ( EABMT−EAMT−EBMT ) − ( EABWT−EAWT−EBWT ) ( 5 ) where “AB” stands for the protein complex and “A” and “B” notations correspond to the unbound monomers . The superscripts WT and MT refer to wild type and mutant , respectively . Thus , Eq ( 5 ) provides the difference of the contribution ( ΔΔE ) of a particular energy term E to the change of the binding free energy caused by a mutation . It should be reiterated that unbound monomer structures were taken from the complex , thus no structural changes are considered to be caused by the binding . In addition , it should be clarified that these terms ( E ) could be potential energies as in case of MM/PBSA delivered terms , or could be an estimation of the entropy change associated with the binding , or could be a term delivered from statistical analysis , for example . Thus their absolute values and dimensionalities vary drastically , but these differences are absorbed by the weight coefficients in the SAAMBE formula . Since weight coefficients in SAAMBE formula are optimized to result in best match against experimentally determined binding free energy changes , the quantity delivered by SAAMBE formula is termed binding free energy change as well ( ΔΔG ) . Below we describe separately the MM/PBSA and the knowledge-based developments of SAAMBE method . The MM/PBSA-based component of the SAAME method is a linear combination of five weighted energy terms: ΔΔGMM/PBSA=w0+w1⋅ΔIE+w2⋅ΔΔEE+w3⋅ΔΔVE+w4⋅ΔΔSP+w5⋅ΔΔSN ( 6 ) Where ΔIE is the change of the total internal energy of complexes . Other energy terms are: ΔΔEE is the change of Coulomb energy , ΔΔVE is the change of van der Waals ( vdW ) energy , ΔΔSP and ΔΔSN are the changes of polar and nonpolar components of solvation energy calculated with Eq ( 5 ) . wi are the weight coefficients which will be optimized against experimental data in tDB . Below we describe the details of calculations of each energy term in Eq ( 6 ) . ΔIE component was calculated as the energy difference of all internal energy terms ( bonded potential , angle potential , and torsion potentials ) of the WT and MT complexes . Strictly speaking , the change of the internal energy should be calculated with Eq ( 5 ) , but since the bound and unbound structures in SAAME protocol are the same , using Eq ( 5 ) will result in zero change of the internal energy . Because of that ΔIE is taken as the difference of the internal energy of complexes only . Obviously this is inconsistent with MM/PBSA methodology and is uninformative thermodynamic quantity , but was accepted since the benchmarking against experimental data showed that adding such energy term in Eq ( 6 ) improves the quality of the predictions ( see Results section ) . The internal energy was calculated with NAMD . ΔΔVE were calculated with the NAMD program using the WT and MT complexes and separated monomers to deliver the terms described in Eq ( 5 ) . It was done by taking the structures on the monomers from already energy-minimized structure of the corresponding complex . Then , each complex , WT and MT , and each separate monomer , WT and MT , were subjected to one step minimization with NAMD to obtain the corresponding vdW energies . ΔΔEE and ΔΔSP energies were calculated with DelPhi software [47] with the following parameters: linear Poisson-Boltzmann solver , scale 1 grids/Å , perfil 70% and external dielectric constant 80 . The choice of the value of internal dielectric constant requires explanation . As it was mentioned above , SAAMBE protocol is rigid body protocol , i . e . the structures of bound and unbound monomers are identical . However , binding is expected to induce small or large structural changes , which are not taken into account in the model explicitly . In the past , we demonstrated that the effects of structural changes on the electrostatic energy can be mimicked by appropriate dielectric constant by assigning specific dielectric constant values to different protein regions [48] . Although our previous analysis was done for folding free energy changes caused by mutations [48] , the same principle should be valid for protein binding free energy modeling . Thus , in the development of SAAMBE protocol , the protein interior was considered to be inhomogeneous and inhomogeneity was modeled via three different dielectric constants ( ε1 , ε2 and ε3 ) . Thus all charge groups ( Asp , Glu , Lys , Arg and His ) were modeled withε1 , all polar groups ( Ser , Thr , Asn , Gln and Tyr ) with ε2 and the rest of amino acids with ε3 . The values of these residue-specific dielectric constants were systematically varied as discussed in the Results section . DelPhi allows for such multi-dielectric modeling [49] . The polar component of solvation energy was calculated via “corrected reaction field energy” module of DelPhi , for both the complexes and separated monomers and then applying Eq ( 5 ) to obtain the difference ( ΔΔSP ) . The Coulombic energies were also calculated with DelPhi for the complexes and separated monomers and the applying Eq ( 5 ) to deliver ΔΔEE . It should be mentioned that the calculated ΔΔEE is not the standard ΔΔEE in MM/PBSA approaches . It is well-known that electrostatic interactions between covalently bound atoms are already taken into consideration via internal energy terms and should not be part of ΔΔEE ( this is taken care in all MD packages ) . However , taking ΔΔEE from the NAMD output resulted in worse performance of SAAMBE method ( as judged by fitting the predictions against experimental data ) and this was the reason to accept such an inconsistency . The nonpolar component of the solvation energy was calculated via linear formula with respect to SASA of the protein and protein complexes ( Eq ( 7 ) ) . The SASA was calculated with NACCESS software [42] and the corresponding coefficients in Eq ( 7 ) were re-distributed in Eq ( 6 ) as: α takes part in the weight w5 while β is absorbed in the free coefficient w0 . The knowledge-based components were calculated according to the formula: ΔΔGKB=w6⋅ΔΔS+w7⋅ΔΔHYDR+w8⋅ΔHB+w9⋅InterfaceMT+w10⋅ΔΔSASAInterfaceMT ( 8 ) where five additional terms were taken into account: entropy ( S ) , hydrophobicity ( HYDR ) , hydrogen bonds ( HB ) , interface area of the MT-complex ( InterfaceMT ) , and the change of the interface area caused by the mutation normalized to the total interface of the MT-complex ( ΔΔSASAInterfaceMT ) . The entropy of the residues in complex and in the corresponding monomers was estimated based on an empirical formula developed in this work . It is based on the maximal number of side chain rotamers ( R ) taken from Ref . [50] . The maximum number of rotamers for each residue is provided in Table 7 . However , we assume that the ability of given amino acid side chain to sample its maximum number of rotamers will depend on its exposure to the surface , i . e . fully exposed residue with relative SASA ( rSASA ) equal to one will be able to access all rotamers , while completely buried one ( rSASA = 0 ) will be completely rigid adopting a particular rotamer . Having in mind that entropy is proportional to the logarithm of states ( in our case rotamers ) , the corresponding formula for this particular residue is: S=ln[rSASA⋅ ( R−1 ) +1] ( 9 ) Eq ( 9 ) is applied to the complexes and individual monomers and the Eq ( 5 ) is used to deliver ΔΔS . The term accounting for the hydrophobicity was modeled using Wimley-White ( H ) hydrophobicity scale [51] ( see Table 7 ) ( different hydrophobicity scales were tested , but were found to perform worse in benchmarking of SAAMBE protocol against experimental data ) . The empirical formula was developed in this work assuming the following: an amino acid contributes to the hydrophobicity depending on its rSASA . For example , a residue being exposed to the water phase will have large contribution to HYRD while practically zero if buried inside the protein . Having in mind that Hj indexes have opposite signs for hydrophobic and hydrophilic amino acids , such a formulation qualitatively describes the physical basis of the hydrophobic effect . The corresponding formula is: HYDR=∑j=1NHj⋅rSASAj ( 10 ) As above , the formula is applied to the corresponding complexes and separate monomers and then Eq ( 5 ) to deliver ΔΔHYDR . The impact of the mutations on the formation of hydrogen bonds ( HB ) was taken into account as well . We computed the number of HB for WT ( ∑HBA−AWT and ∑HBB−BWT ) and MT ( ∑HBA−AMT and ∑HBB−BMT ) monomers and at the same time the number of hydrogen bonds that were formed between monomers in the corresponding complex ( ∑HBA−BMT and ∑HBA−BWT ) . The first class represents the intra-monomer bonds , and the second inter-monomer bonds . It is assumed that intra-monomer HB change resulting in more HB in the mutant , ΔHB > 0 , will make MT monomers more stable than the WT , and thus might decrease binding free energy . In contrast , ΔHB >0 of inter-monomer HB is expected to increase binding affinity of the MT compared with WT . Because of such considerations , the effect of HB on the binding free energy change was calculated as: ΔHB= ( ∑HBA−BMT−∑HBA−AMT−∑HBB−BMT ) − ( ∑HBA−BWT−∑HBA−AWT−∑HBB−BWT ) ( 11 ) where the HB was counted as cases involving two atoms oxygen acceptor and hydrogen ( except the nonpolar Cα and Cβ hydrogen atoms , HA and HB ) atoms located at distance shorter than 2 . 4 Å . Since the nitrogen acceptor is much weaker than oxygen , for simplicity it was not considered . Similarly the geometry of the hydrogen bond was not taken into consideration . Only polar ( S , T , N , Q , Y ) and charge ( R , H , K , D , E ) amino acids were taken into account . Our previous work [24] indicated that the surface area of the interface in the MT-complex is an important factor in predicting binding free energy changes . Because of that , it is included in this protocol as well and was calculated as the difference in SASA of complex and the sum of each of its parts . ΔΔSASA was calculated with Eq ( 5 ) as the difference in SASA of the complex and monomeric states of MT and WT . In order to quantify the performance of our algorithm and compare it with other methods we evaluated the calculated and experimental values of change in binding free energy due to single point mutation and assigned one of four flags for each entry in the tDB: true positive ( tp ) , true negative ( tn ) , false positive ( fp ) , or false negative ( fn ) . The explanation of the assignment procedure is provided in the Table 8 . The quality of the predictions was described by six parameters: accuracy , precision , sensitivity , specificity , negative predictive value ( NPV ) and Matthews correlation coefficient ( MCC ) [52 , 53]: accuracy=tn + tptn+tp+fn+fp ( 13 ) sensitivity=tptp+fn ( 14 ) specificity=tntn+fp ( 15 ) presicion=tptp+fp ( 16 ) NPV=tntn+fn ( 17 ) MCC=tp⋅tn + fp⋅fn ( tp+fp ) ⋅ ( tp+fn ) ⋅ ( tn+fp ) ⋅ ( tn+fn ) ( 18 ) One of the goals of SAAMBE development is to develop fast algorithm capable of large-scale calculations . Thus , the execution time is an important component of the investigation . The execution time was monitored as a function of the number of amino acids in the corresponding complex ( sequence length ) and as a function of the geometrical shape of the complex ( monitored via the largest dimension of WT complex ) . To verify the agreement between experimental and predicted values of the change of binding free energy due to single point mutation we calculated the Pearson correlation coefficient . In the paper all reported correlation coefficients were significantly different from zero with p-value smaller than 0 . 01 . We also performed five-fold cross validation test for the tDB . It was done by randomly partitioning the tDB into five subgroups of approximately equal size . Each combination of four subgroups was used for training , while the fifth—for testing the model . Then correlation coefficients were averaged over different cross-validated sets . | Developing methods for accurate prediction of effects of amino acid substitutions on protein-protein affinity is important for both understanding disease-causing mechanism of missense mutations and guiding protein engineering . For both purposes , there is a need for accurate methods primarily based on first principle calculations , while being fast enough to handle large number of cases . Here we report a new method , the Single Amino Acid Mutation based change in Binding free Energy ( SAAMBE ) method . The core of the SAAMBE method is a modified molecular mechanics Poisson-Boltzmann Surface Area ( MM/PBSA ) method with residue specific dielectric constant . Adopting residue specific dielectric constant allows for mimicking the effects of plausible conformational changes induced by the binding on the solvation energy without performing computationally expensive explicit modeling . This makes the SAAMBE algorithm fast , while still capable of capturing many of the explicit effects associated with the binding . The performance of the SAAMBE protocol was tested against experimentally determined binding free energy changes over 1300 mutations in 43 proteins and very good correlation coefficient was obtained . Due to its computational efficiency , the SAAMBE method will be soon implemented into webserver and made available to the computational community . | [
"Abstract",
"Introduction",
"Results",
"and",
"Discussion",
"Methods"
] | [] | 2015 | Predicting Binding Free Energy Change Caused by Point Mutations with Knowledge-Modified MM/PBSA Method |
Microalgae are promising microorganisms for the production of numerous molecules of interest , such as pigments , proteins or triglycerides that can be turned into biofuels . Heterotrophic or mixotrophic growth on fermentative wastes represents an interesting approach to achieving higher biomass concentrations , while reducing cost and improving the environmental footprint . Fermentative wastes generally consist of a blend of diverse molecules and it is thus crucial to understand microalgal metabolism in such conditions , where switching between substrates might occur . Metabolic modeling has proven to be an efficient tool for understanding metabolism and guiding the optimization of biomass or target molecule production . Here , we focused on the metabolism of Chlorella sorokiniana growing heterotrophically and mixotrophically on acetate and butyrate . The metabolism was represented by 172 metabolic reactions . The DRUM modeling framework with a mildly relaxed quasi-steady-state assumption was used to account for the switching between substrates and the presence of light . Nine experiments were used to calibrate the model and nine experiments for the validation . The model efficiently predicted the experimental data , including the transient behavior during heterotrophic , autotrophic , mixotrophic and diauxic growth . It shows that an accurate model of metabolism can now be constructed , even in dynamic conditions , with the presence of several carbon substrates . It also opens new perspectives for the heterotrophic and mixotrophic use of microalgae , especially for biofuel production from wastes .
Microalgae are unicellular eukaryote microorganisms that can grow autotrophically using light energy and CO2 . Many species can also grow heterotrophically in darkness on various organic carbon sources , including glucose , or can combine heterotrophy and autotrophy for a mixotrophic growth [1] . Microalgae have been domesticated and used to synthesize many products with industrial applications , such as pharmaceutics or cosmetics ( antioxidants , pigments , unsaturated long-chain fatty acids ) , agricultural products ( food supplements , functional food , colorants ) and animal feed ( aquaculture , poultry or pig farming ) [2] . They are also promising organisms for green chemistry ( bioplastics ) , the environment ( wastewater treatment , CO2 mitigation ) , and even energy production ( biodiesel , bioethanol , hydrogen ) [2] . Autotrophic growth of microalgae is limited by light distribution to all the cells , constraining the cell concentration to below 10 g/l ( for the thinnest and most concentrated cultivation systems ) . Heterotrophic growth does not have this limiting factor and higher biomass density can be achieved [1] , drastically reducing the harvesting costs . In addition , heterotrophic growth is usually faster , reducing the cultivation time [3] . However , industrial production of heterotrophic microalgae is hampered by the high economic and environmental costs of glucose , commonly used as a substrate . One solution is to use the waste from other processes , such as glycerol , acetate ( ACE ) or butyrate ( BUTYR ) , which represent low cost carbon substrates . For instance , dark anaerobic fermentation produces an effluent mainly composed of acetate and butyrate [4] . However , some substrates in waste , such as butyrate , can be inhibitory [5] . Moreover , the successive metabolic switches between different substrates are not well understood and are likely to significantly affect growth . Therefore , this bioprocess still needs to be mastered and optimized to produce microalgae and extract the targeted byproducts on an industrial scale and at a competitive price , with consistent quality and in a sustainable way . In this context , mathematical modeling of the metabolism has proven to be an efficient tool for optimizing growth and increasing the production of target molecules . To date , no models exist for heterotrophic microalgal metabolism dynamically switching between several substrates ( S1 Table ) , including mixotrophic growth in light . So far , only static fluxes have been predicted under constant substrate consumption [6–9] . Representing the dynamic shifts for a blend of substrates typical of real wastewater is a major challenge , since some intracellular accumulation might occur , either during the transition between substrates , or due to the varying nature of the light . As a consequence , the quasi-steady state assumption ( QSSA ) required by most of the existing metabolic approaches may be an invalid hypothesis in this case [10] . The DRUM modeling framework recently proposed in [10] was used here to handle the non quasi-steady state ( QSS ) . It allowed the development of a dynamic metabolic model for Chlorella sorokiniana grown on a single-substrate culture and a mixed-substrate ( acetate and butyrate ) culture , combined with various combinations of light . The model is thus designed to represent autotrophic , heterotrophic or mixotrophic modes under diauxic conditions . Our purpose is to propose a relatively generic model , instantiated and calibrated for C . sorokiniana . According to Baroukh et al . [11] , such a generic model should be applicable to a wide range of microalga species .
The goal of the experiments was to grow Chlorella sorokiniana on a synthetic medium mimicking the digestate composition produced by a dark fermenter processing organic waste . At this stage , the composition of the medium was kept simple , with only the two main organic components—acetate and butyrate [4]–to gain a clear understanding of their effects on microalgae growth . Chlorella sorokiniana was grown both in the dark and in the light ( 136 μE . m-2 . s-1 ) , in axenic conditions at 25°C and constant pH ( 6 . 5 ) in triplicate batches with different initial concentrations of acetate and butyrate ( Table 1 ) . Nitrogen ( ammonium ) and phosphorus were provided in non-limiting concentrations in order to focus solely on carbon metabolism . To ensure that no substrate was favored because of acclimation , the inoculum was grown autotrophically beforehand . See Turon et al . [5 , 12] for more details of the experimental protocols . A detailed description of the metabolic network reconstruction is provided in S1 File . Since Chlorella sorokiniana has not been sequenced yet , no genome-scale metabolic network ( GSMN ) reconstruction was possible . However , the core carbon and nitrogen metabolic networks in the GSMN of previously reconstructed microalgae species are relatively similar [13] . Thus , the conserved core metabolic network was used , containing the central metabolic pathways relevant to mixotrophy and heterotrophy: photosynthesis , glycolysis , pentose phosphate pathway , citric acid cycle , oxidative phosphorylation , and synthesis of chlorophyll , carbohydrates , amino acids and nucleotides . Species-specific pathways such as the synthesis of secondary metabolites were not represented , since these pathways were assumed to represent negligible fluxes compared to the main pathways and thus to have little impact on the metabolism . The reactions involved in macromolecule synthesis ( proteins , lipids , DNA , RNA and biomass ) were lumped into generic reactions . The growth-associated ATP maintenance ( GAM ) was replaced by the value experimentally measured by Boyle et al . ( 2009 ) for growth of Chlamydomonas reinhardtii on acetate . They observed 29 . 890 moles of ATP for 1000g of biomass . In our model , the biomass reaction yields 186g of biomass; the maintenance term is thus 5 . 56 moles of ATP per mol of biomass ( details on how this value was computed are available in S1 File ) . A sensitivity analysis was carried out to assess the impact of this maintenance value on model accuracy ( S1 Fig ) . Results showed that the optimal growth associated yield is 12 . 2mol ATP per mol of biomass , with [0–16 . 7 molATP/mol B] as the 10% interval confidence . The value of Boyle et al [6] falls in this interval , demonstrating that an error of this term in this range has a minor impact on model predictions . The non-growth ATP maintenance ( NGAM ) was assumed negligible . It may explain the slightly higher value of the optimal growth associated yield resulting from optimization , which might also compensate for NGAM . Generally , metabolic modeling relies on the QSSA of the whole metabolic network , where intracellular metabolites cannot accumulate or be depleted [14] . The idea of the DRUM approach is to mildly relax this hypothesis , by splitting the metabolic network into a limited number of sub-networks [10] , for each of which the QSS is assumed . The metabolites situated at the junction between the sub-networks , can therefore have dynamics of accumulation and depletion . The sub-networks are defined by metabolic functions and take into account cellular compartments . This assumption is supported by the idea that cell function and cell compartment are often associated with co-regulation and substrate channeling , which implies synchronicity of reactions and thus quasi-steady state for those reactions [10] . The idea is also to find a network splitting simple enough for explaining the experimental data , so as to avoid overfitting by postulating too many reactions kinetics [10] . Further details on the philosophy behind the network splitting and the DRUM framework are given in the discussion section of Baroukh et al . [10] ( DRUM principles are also summarized in supplementary information ( S1 File , section 8 ) ) . For representing the growth of Chlorella sorokiniana with different inorganic or organic carbon sources , the network was split taking into account the compartments of the cell with a global catabolic or anabolic function ( Fig 1 ) : i ) the glyoxysome for acetate and butyrate assimilation , ii ) the chloroplast for photosynthesis , and iii ) the rest of the reactions for functional biomass production ( synthesis of lipids , carbohydrates , proteins , DNA , RNA and chlorophyll ) . Several splitting were tested , particularly on the transported metabolites between the glyoxysome and the cytosol ( which are not consistent in literature ) . The best fitting results were obtained with these three sub-networks . Apart from inorganic compounds , only succinate ( SUC ) and glyceraldehyde 3-phosphate ( GAP ) were intracellular metabolites ( A ) that potentially accumulate . The idea is that these metabolites , which shuttle between the compartments ( respectively between the glyoxysome and the cytosol and between the chloroplast and the cytosol ) , and which act as intermediate between catabolism and anabolism , are the ones that could act as “buffers” inside the cell . Each sub-network was balanced in cofactors and in chemical elements ( carbon , oxygen , nitrogen , phosphorus , sulfur ) . Each sub-network was then reduced to macroscopic reactions ( MRs ) using elementary flux mode analysis [15] . To compute elementary flux modes ( EFMs ) , the efmtool program was used [16] . For the three sub-networks , the EFM could be computed easily , since their total number was less than 1751 ( it should be noted that an EFM analysis of the full network results in 5105 modes ) . Glyoxysomes are specialized peroxisomes found in plants or microalgae [8] , in which fatty acids ( including acetate and butyrate ) can be used as a source of energy and carbon for growth when photosynthesis is not active . Fatty acids are hydrolyzed to acetyl-CoA , and then transformed into succinate via the glyoxylate cycle . Succinate can then be transformed into a variety of macromolecules for biomass growth , through combinations of other metabolic processes taking part in other compartments of the cell . Reduction of the glyoxysome sub-network yielded two MRs , one for each substrate ( Table 2 ) . Photosynthesis supports the generation of cell energy in phototrophic organisms and contributes to the incorporation of inorganic carbon . The process takes place in the chloroplast and consists of two steps commonly known as the light and dark reactions . The light reaction consists of the generation of cell energy ( ATP , NADPH ) from water and photons , producing oxygen . Thanks to the energy of the light reaction , the dark step reactions incorporate carbon dioxide through the Calvin cycle producing a 3-carbon sugar ( 3-phosphoglycerate , or 3PG ) . Then , 3PG is transformed into glyceraldehyde 3-phosphate ( GAP ) and transported to the cell cytosol . Elementary flux mode analysis of this sub-network yielded only one Elementary Flux Mode ( EFM ) ( Table 2 ) , associated with one macroscopic reaction ( MR3 ) . The stoichiometry of the derived macroscopic reaction is in agreement with the literature: a quota of 8 photons are needed per carbon incorporated [17] . The synthesis reactions of lipids , proteins , DNA , RNA , chlorophyll and carbohydrates were grouped into a functional biomass synthesis sub-network and assumed to be in QSS . This sub-network includes glycolysis , TCA cycle , oxidative phosphorylation , pentose phosphate pathway , nitrogen and sulfur assimilation , carbohydrate synthesis , lipid synthesis , amino acid synthesis and nucleotide synthesis . The reduction of this sub-network yielded 1748 EFMs , which is reasonable given the number of involved reactions ( 143 ) , and much lower than the number of modes for the full network ( 5105 ) . Most of these EFMs ( 1618 ) yielded biomass , while the others correspond to futile cycles . In terms of carbon , once normalized by unit of biomass synthesis flux , the 1618 MRs deduced from the EFMs only differed in their consumption of SUC and GAP and their production of CO2 ( S2 and S3 Figs ) . As in Flux Balance Analysis [18] , we assumed that the cell maximized biomass growth , and hence minimized carbon loss when synthesizing biomass from each substrate . The elementary flux modes with the best SUC/CO2 yield and GAP/CO2 yield were thus chosen ( Table 2 ) . The resulting MR consumes SUC or GAP and NH4 for carbon and nitrogen sources , SO4 and Mg for protein and chlorophyll synthesis and O2 for ATP synthesis through oxidative phosphorylation . At this stage , the macroscopic kinetics of the MRs must be determined in order to simulate the metabolic dynamics [10] . We assumed that only the carbon substrates of each MR were limiting , playing thus a role in the kinetics . Michaelis-Menten kinetics was used for acetate consumption ( Table 2 ) , since experimental data showed no growth inhibition on acetate . Haldane kinetics was chosen for the butyrate consumption reaction ( Table 2 ) , since experimental data showed that butyrate inhibited biomass growth ( growth only possible with maximum 0 . 1 gC . L-1 ( Fig 2C ) in butyrate-only experiments ) . Biomass growth clearly exhibited a diauxic growth for the mixed substrate conditions: acetate was entirely consumed before the butyrate concentration started to decrease ( Fig 3 ) . This diauxic growth is probably the result of transcriptional regulations taking place inside the cell . DRUM framework , can explicitly describe such regulations via an appropriate choice of the kinetics in connection with metabolite accumulation . However , to the best of our knowledge , the transcriptional regulations responsible for this diauxic growth are not known . It is thus premature to propose an explicit mathematical expression for representing this phenomenon; a general kinetic expression for diauxic growth , which implies that regulation is performed by acetate directly on the cell transporter of butyrate , was thus chosen . An inhibitory term of acetate concentration on butyrate consumption kinetics was included in the butyrate uptake kinetics ( Table 2 ) . Linear kinetics depending on the mean light intensity in the reactor was chosen to represent photosynthesis ( Table 2 ) . The mean light intensity was computed using a Beer-Lambert law ( S1 File ) . Linear kinetics with respect to the carbon substrate were chosen for biomass synthesis ( Table 2 ) . Finally , the dynamics of the 172 fluxes in the metabolism can be derived from a system of 14 differential equations comprising 14 metabolites and 5 macroscopic reactions representing 3 compartments: dM′dt=d ( SAB ) dt=K′ . α . B where M’ is the vector of metabolites ( 14x1 ) composed of substrate S , metabolites susceptible to accumulate A ( SUC and GAP ) and functional biomass B; K’ is the reduced stoichiometric matrix ( 14x5 ) and α is the kinetics vector ( 5x1 ) ( Table 2 , S6 Fig ) . The way all the metabolic fluxes are computed from K’ and α is recalled in S1 File . Total biomass X ( g . L-1 ) is computed thanks to a mass balance on the cell: X ( t ) =∑AMA . A ( t ) +MB . B ( t ) with MA and MB the molar masses of metabolites A and B ( for further details , see S1 File section 5 ) . The dynamic model has 10 degrees of freedom , and each degree is represented by a parameter that needs to be calibrated . To estimate the parameters , we minimized the squared-error between simulation and experimental measurements using the Nelder-Mead algorithm [19] ( function fminsearch in Scilab® ) . To reduce the risk of local minima , several optimizations were performed with random initial parameters . Nine experiments were used to estimate the parameters ( Table 1 ) ; the nine remaining experiments were reserved to assess the validity of the model ( Table 1 ) . Results of the parameter identification are presented in Table 3 .
The model simulation accurately reproduces experimental data , even for the validation data sets that were not used for calibration ( Figs 2–4 ) . The diauxic growth is particularly well represented ( Figs 3 and 4D ) , and the transient behavior , together with the final biomass , is correctly predicted ( Figs 2–4 ) , showing that the biomass yields obtained from the metabolic network are accurate . Indeed , one of the advantages of metabolic modeling [20] is the prediction of biomass yields supported by the stoichiometry of the metabolic network . Here , the predicted conversion yield of acetate and butyrate to biomass is 0 . 514 grams of carbon biomass per gram of carbon in the incoming substrate . This yield contributes to correct prediction of the biomass for both acetate ( Fig 2B ) and butyrate ( Fig 2D ) , thus validating the approach . Interestingly , the yields are identical between the two substrates . A possible explanation is the fact that more ATP is required for the transport of butyrate into the cell than for acetate , thus balancing the ATP created when converting butyrate and acetate to succinate . The set of kinetic parameters matches both the single-substrate culture and the mixed-substrate culture . This implies that butyrate has no impact on acetate growth rate . However , the inverse is not true , since the acetate concentration at which butyrate consumption starts ( kD ) is very low ( 5 . 39*10−10 M ) , illustrating the strong diauxic growth that occurs . Even the smallest amount of acetate inhibited butyrate uptake . The maximum acetate uptake rate was higher than the maximum butyrate uptake rate by nearly 15 fold , reflecting the preference of Chlorella sorokiniana for acetate . The non-inhibiting butyrate concentration ( SoptMR2 ) was very low ( 1 . 93*10−5 M ) , which highlights the strong inhibition of butyrate in the medium on its uptake . It also explains why , in the butyrate-only experiments , no biomass growth was observed for butyrate concentrations above 0 . 1 g . L-1 ( Fig 2D ) . In addition to substrates and biomass concentrations , light evolution inside the culture vessel was computed . During the first few days , the average light intensity decreases until equilibrium is reached around 16 μE . m-2 . s-1 ( S1 File section 4 , S8 Fig ) . It represents 11 . 7% of the incident light and is in agreement with the literature [21] . Interestingly , equilibrium is reached faster for mixotrophic growth , particularly on acetate , which supports fast heterotrophic growth ( S8 Fig ) . In addition , the photosynthetic quotient for autotrophic growth varies between 1 . 0 and 1 . 16 , matching the typical range of 1 . 0–1 . 8 for microalgae [6] . The predicted metabolic fluxes ( Fig 5 , S9 Fig ) are in accordance with previous studies [11] . Autotrophy ( S9C Fig ) is characterized by high fluxes in the photosynthetic pathways , which convert light and CO2 to GAP . Beyond these pathways , fluxes drop considerably in terms of absolute magnitude . Upper glycolysis is in the gluconeogenic direction to produce the carbohydrate and sugar precursor metabolites ( Glucose 6-phosphate ( G6P ) , Ribose 5-phosphate ( R5P ) , Erythrose 4-phosphate ( E4P ) ) necessary for growth . In the heterotrophic mode , fluxes are more homogenous among reactions ( Fig 5B , S9A Fig ) . Acetate and butyrate are converted to acetyl-CoA in the glyoxysome ( Fig 5B , S9D Fig ) . Acetyl-CoA is then converted into succinate by the glyoxylate cycle and injected in the TCA cycle . Upper glycolysis also goes in the gluconeogenic direction to produce carbohydrate and sugar precursors . This can be achieved thanks to the anaplerotic reactions that convert oxaloacetate to phosphoenolpyruvate ( PEP ) . Mixotrophy is a mixed combination of the autotrophic and heterotrophic modes ( Fig 5A , S9A Fig ) . For mixotrophic growth on acetate , heterotrophic metabolism is dominant , whereas autotrophic metabolism is dominant for mixotrophic growth on butyrate . This is due to the fact that autotrophic growth is slower than growth on acetate but faster than growth on butyrate . Interestingly , in agreement with the data , the model did not predict any growth on butyrate above 0 . 1 gC . L-1 , and at the same time successfully forecasted growth on 0 . 9 gC . L-1 butyrate in mixed substrate conditions ( Fig 3E ) and on 0 . 3 gC . L-1 butyrate in mixotrophic conditions . Indeed , in these conditions , the first-stage growth on acetate and/or light produces enough biomass to finally consume such an inhibiting quantity of butyrate . The substrate to biomass ( S/X ) ratio is known to be a key process parameter for overcoming the inhibitory effects of the substrate [22] . The model therefore represents a tool to compute and optimize the amount of co-substrate that must be added to overcome the inhibition and consume the butyrate . Different strategies could be tested to achieve a low S/X ratio and accelerate butyrate consumption . The simplest approach would involve adding a non-inhibiting substrate in order to reduce the amount of inhibitory substrate per unit biomass . For example , the addition of 0 . 5 gC . L-1 of acetate for a volume equal to half of the culture volume has been found to eventually lead to the consumption of 0 . 5 gC . L-1 of butyrate in 14 days ( Fig 6B ) , which would not have been possible otherwise ( Fig 6A ) . However , in general , such pure substrate is not available . We therefore simulated the addition of a mix of acetate and butyrate in proportions that are representative of fermentative digestate [4] , for a volume equal to half of the culture volume . On the one hand , the acetate contained in the waste stimulated growth , but since it is associated with addition of butyrate , it also increased inhibition . Simulations show that the inhibition is overcome , but does not lead to the total consumption of butyrate within 15 days ( Fig 6D ) . Furthermore , the mixotrophic potential can be exploited: autotrophic growth can be enhanced by illumination in order to ultimately dilute the inhibitory substrates . Illuminating the algae at an incident intensity of 136 μE . m-2 . s-1 leads to the consumption of the same quantity of butyrate in 13 days , and this delay can be reduced to 9 days using a light intensity of 272 μE . m-2 . s-1 ( Fig 6C ) . Finally , if light is provided at the same time as the addition of fermentative digestate ( for a volume equal to half of the culture volume ) , inhibition can be overcome after 10 days ( Fig 6D ) . The advantage of the DRUM approach is its ability to account for the accumulation of some intracellular metabolites and thus to characterize the time to reach steady state . It can also determine more quantitatively the time scales of flux variations in the cell than earlier frameworks . This analysis was applied to SUC and GAP , which are , in our model , the intermediate accumulating metabolites . Interestingly , SUC actually hardly accumulates in the simulations and rapidly achieves a QSS ( S10 Fig ) where its concentration evolves slowly compared to the other variables in the system ( substrate consumption , biomass formation ) . We developed an algorithm to automatically detect the time needed to reach QSS ( tQSS ) . In the experimental conditions of this study , approximately 3 minutes were necessary for succinate to achieve QSS ( S2 File ) thanks to a higher biomass synthesis rate ( via a high kMR4 ) compared to the substrate assimilation rate , implying that succinate is immediately consumed once it is synthesized from butyrate or acetate . A sensitivity analysis on the parameter kMR4 revealed that the confidence interval of tQSS was [0 . 6; 34] minutes ( model error less than 5% of the minimal error ) ( S1 File section 8 , S2 File ) . After the brief transient succinate step , the QSSA for heterotrophic growth on butyrate and acetate is valid . Therefore , the macroscopic model can be reduced further , by merging reaction MR4 with reactions MR1 and MR2 ( S1 File section 6 ) . The same kinetic parameters can be used for simulation , and the fit is nearly identical ( increase of 0 . 6% of the error ) . As a consequence , results considering QSSA are very close to the ones based on DRUM . GAP , in contrast to SUC , does not reach a QSS rapidly ( S11 Fig ) . First , GAP accumulates at high light intensities , reaching a maximum when average light intensity is approximately 60 μE . m-2 . s-1 . Then , it is consumed at low light intensities , reaching a QSS when average light intensity reaches a steady state at 16 μE . m-2 . s-1 ( S1 File section 8 ) . This suggests that microalgal metabolism in autotrophic and mixotrophic modes only reaches a QSS when average light is constant in the culture media , meaning that growth has ceased . This behavior is similar to that of microalgae grown in day/night cycles [10 , 11] , involving accumulation of carbon-reserve metabolites ( carbohydrates , lipids ) during the day , when the light is intense enough , and re-consumption during the night or at the beginning and end of the day , when light intensity is low . Here , the carbon reserve metabolite is GAP , because only GAP accumulated in the model . Nevertheless , it is probable that carbohydrates and/or lipids also accumulate . Further experiments are required to validate these results more extensively and to determine which carbon-only metabolite is stored inside the cell . To confirm these results , a Macroscopic Bioreaction Model of the system [23] , relying on the QSSA assumption , was developed ( see S1 File section 10 for details on the methodology ) . Without accumulation of SUC , the model error was almost unchanged ( 0 . 06% increase of the error ) . But without the possibility for GAP to accumulate , a 40% increase in the error is observed . This confirms our finding that GAP do accumulate inside the cell at high light intensities to be consumed later at lower light intensities . It is also interesting to note that the MBM approach is sufficient and produces accurate results , for applications in heterotrophy only cultures , without the need for accumulating metabolites . The dynamic metabolic model developed for the heterotrophic , mixotrophic and autotrophic growth of Chlorella sorokiniana on acetate and butyrate achieved a so far unequalled accuracy . The model efficiently fits the dynamic experimental data and correctly predicts the biomass yields for a broad range of experimental conditions . This new powerful simulation tool provides new insight into the mixotrophic microalgal process , and allows us to explore the different possibilities to overcome the inhibition induced by some of the substrates , in particular by adjusting the mixotrophic regimes . The model also highlights the dynamics of some internal compounds , especially under an auto- or mixotrophic regime , while light intensity is slowly affected by an increase in self-shading . As a consequence , the model shows that QSSA is not valid for mixotrophic growth as long as the light is variable in the culture medium . In the future , the model should be extended further in order to handle mixotrophic behavior under periodic light/dark cycles . | Most existing metabolic modeling tools are not suitable for studying diauxic growth with dynamic substrate shifts . This paper describes a successful modeling of Chlorella sorokiniana metabolism , based on 172 reactions and validated by nine independent dynamic experiments ( nine experiments were used for its calibration ) , in which microalgae were grown heterotrophically or mixotrophically on acetate and/or butyrate , in the light or dark . Such an extensive validation has not been performed before for microalgae . It was demonstrated that the model could be used to assess the flux dynamics in the cell . This innovative simulation tool was also used to derive original strategies to bypass the toxicity of some substrates using mixotrophic regimes . | [
"Abstract",
"Introduction",
"Model",
"Results/Discussion"
] | [
"carbohydrate",
"metabolism",
"protein",
"metabolism",
"chemical",
"compounds",
"metabolic",
"networks",
"carbohydrates",
"organic",
"compounds",
"plant",
"science",
"metabolites",
"network",
"analysis",
"photosynthesis",
"computer",
"and",
"information",
"sciences",
"chemi... | 2017 | Dynamic metabolic modeling of heterotrophic and mixotrophic microalgal growth on fermentative wastes |
The mosquito Aedes aegypti ( L . ) is a major vector of viral diseases like dengue fever , Zika and chikungunya . Aedes aegypti exhibits high morphological and behavioral variation , some of which is thought to be of epidemiological significance . Globally distributed domestic Ae . aegypti have often been grouped into ( i ) the very pale variety queenslandensis and ( ii ) the type form . Because the two color forms co-occur across most of their range , there is interest in understanding how freely they interbreed . This knowledge is particularly important for control strategies that rely on mating compatibilities between the release and target mosquitoes , such as Wolbachia releases and SIT . To address this question , we analyzed nuclear and mitochondrial genome-wide variation in the co-occurring pale and type Ae . aegypti from northern Queensland ( Australia ) and Singapore . We typed 74 individuals at a 1170 bp-long mitochondrial sequence and at 16 , 569 nuclear SNPs using a customized double-digest RAD sequencing . 11/29 genotyped individuals from Singapore and 11/45 from Queensland were identified as var . queenslandensis based on the diagnostic scaling patterns . We found 24 different mitochondrial haplotypes , seven of which were shared between the two forms . Multivariate genetic clustering based on nuclear SNPs corresponded to individuals’ geographic location , not their color . Several family groups consisted of both forms and three queenslandensis individuals were Wolbachia infected , indicating previous breeding with the type form which has been used to introduce Wolbachia into Ae . aegypti populations . Aedes aegypti queenslandensis are genomically indistinguishable from the type form , which points to these forms freely interbreeding at least in Australia and Singapore . Based on our findings , it is unlikely that the presence of very pale Ae . aegypti will affect the success of Aedes control programs based on Wolbachia-infected , sterile or RIDL mosquitoes .
The mosquito Aedes aegypti ( Linnaeus ) is the most important arboviral vector in the tropics and subtropics [1] . Diseases transmitted by Ae . aegypti , like dengue fever and Zika , are on the rise [2] , and some are reappearing . For instance , chikungunya has returned to the American tropics in 2013 , after being absent for nearly 200 years [3] . Yellow fever was nearly eliminated thanks to an effective vaccine , but is now resurging in central and south Africa [4] . Such epidemiological trends highlight the need to persist with vector control efforts , which requires a thorough understanding of vector biology . Nearly 60 years ago , Mattingly [5] noted that despite a vast body of literature , few mosquitoes have been”the subject of misconception… . in the minds of the general run of entomologists” like Aedes aegypti [5] . The species has a plethora of historical synonyms [6] , mainly as a result of having extensive variation in body color and scaling patterns [7] which was also thought to correlate with behavioral differences ( e . g . [8] ) . These issues motivated Mattingly [5] to revise the taxonomy of Ae . aegypti and create a foundation for the modern studies of this disease vector . Mattingly [5] proposed the intraspecific classification of Ae . aegypti into three forms . A few years later , McClelland [7] reported a high level of variation in color and scaling within and among Ae . aegypti populations , suggesting that subdivision into forms seems oversimplistic and should be abandoned unless correlation between genetic and color variation can be demonstrated . In their latest review of the Ae . aegypti history , Powell and Tabachnick [9] pointed out that McClelland’s recommendations have often been ignored for the past 45 years , despite the fact that multiple genetic marker systems ( allozymes , microsatellites , nuclear and mitochondrial SNPs ) have failed to find a clear differentiation between forms and markers [10–13] . Recently , Chan et al . [14] suggested that the DNA barcoding technique can be used to distinguish queenslandensis individuals from the type individuals in Singapore . The sequence divergence of 1 . 5%-1 . 9% between the two forms [14] , although lower than a commonly adopted threshold of 3% for species delineation in insects [15] , suggests that the two forms may not freely interbreed . Historical records indicate that the two forms have co-occurred in Singapore and other parts of south-east Asia and Australia for hundreds of generations [5 , 8] . In sympatry , genetic isolation can be maintained largely through pre-zygotic isolation mechanisms like incompatibilities in mating behavior [16] . For instance , molecular forms of the malarial mosquito , Anopheles gambiae , fly together in mating swarms but rarely hybridize due to flight-tone matching between males and females of the same form [17] . Similar incompatibilities in Ae . aegypti would have implications for control strategies that rely on successful mating between the release and target mosquitoes , like Wolbachia-based population replacement and suppression [18 , 19] , releases of sterile males [20] or males with a RIDL construct [21] . To explore this further , we analyzed nuclear and mitochondrial genome-wide variation in the co-occurring pale and type Ae . aegypti from Singapore and northern Queensland ( Australia ) . The RADseq approach we employed allows for detection of genetic structure and ancestry with power unparalleled by previous genetic studies of the Ae . aegypti forms [22] . Any association between genetic structuring ( nuclear/mitochondrial ) and the mosquito color/scaling would provide support for the hypothesis of restricted interbreeding between the type and the queenslandensis form , with implications for the implementation of biocontrol programs to suppress diseases transmitted by Ae . aegypti .
The collection of wild mosquitoes in the study areas does not require specific field ethics approval . The sampling was not conducted on protected land , nor did it involve endangered or protected species . Consent was obtained from residents at each location where collections occurred on private property . In Singapore , all samples were collected as larvae from the domestic breeding containers at nine locations during the second week of April 2015 ( Fig 1 , Table 1 ) . These samples were collected during routine inspection by enforcement officers of the National Environment Agency ( NEA ) , Singapore . Larvae were reared to the adult stage under standard laboratory conditions ( 25° ± 1°C , 80 ± 10% relative humidity and 12 h light/dark cycle ) . In Townsville ( northern Queensland ) , samples were collected as adults using Biogents Sentinel traps placed at 55 locations in January 2014 ( Fig 1 , Table 1 ) . Adult mosquitoes were sexed and identified to form based on the key diagnostic color and scaling features , following Mattingly [5] and McClleland [7] . Eleven out of 44 mosquitoes ( 25% ) from Singapore , and seven out of 99 mosquitoes ( 7% ) from Townsville were identified as the queenslandensis form ( Table 1 ) . An additional four queenslandensis individuals collected in Cairns ( northern Queensland ) in December 2014 were included in the analyses ( Table 1 ) . DNA was extracted from 29 individuals collected in Singapore ( 18 female type , 11 female queenslandensis ) and 45 individuals from northern Queensland ( 17 male type , 17 female type , 11 female queenslandensis ) ( Table 1 ) . Qiagen Blood and Tissue DNA kit ( Venlo , Limburg , NL ) was used to extract DNA from a whole adult mosquito . 100 ng of DNA from each individual was used to construct the double-digest RAD library following a previously validated protocol [22] . In short , 100 units of the two frequently cutting enzymes ( MluCI and NlaIII , New England Biolabs , Beverly MA , USA ) were used to digest 100 ng of DNA during three hours of incubation at 37°C . 100 pM P1 and 300 pM P2 Illumina adapters with customized barcode sequences were ligated to the genomic fragments using 100 units of T4 ligase at 16°C overnight ( New England Biolabs , Beverly , MA , USA ) . Pooled ligations were purified and size selected for fragments 300-450bp in length , using the 2% Pippin Prep cassette ( Sage Sciences , Beverly , MA , USA ) . The final libraries ( one for each geographic region ) were enriched with 12 PCR cycles with standard Illumina primers and then sequenced in two HiSeq2500 lanes with the 100 bp paired-end chemistry . Raw fastq sequences were processed within our customized pipeline . First , all reads were trimmed to the same length of 90 bp and removed if the base quality score was below 13 ( FASTX Toolkit , http://hannonlab . cshl . edu/fastx_toolkit/index . html ) . High quality reads were then aligned to the reference mitochondrial genome [23] and the nuclear genome version AaegL1 [24] using the aligner Bowtie [25] . Uniquely aligned reads were passed to the refmap . pl program that runs the Stacks v . 1 . 35 pipeline [26] . In addition to the samples from Singapore , Townsville and Cairns , we included previously sequenced individuals: 15 from Rio de Janeiro ( Brazil ) [27] , 15 from Gordonvale ( northern Queensland ) , and 15 from Ho Chi Minh City ( Vietnam ) ( S1 Table ) . This was done to compare the extent of genetic structuring within and among samples at a regional and global scale . Sexing of the larval samples from Brazil and Vietnam could not be done based on the external morphological features , so we employed a genetic sexing method based on the presence/absence of the male-specific RAD tags [28] . All 119 individuals were included in the creation of the RAD tag catalogues using the default Stacks parameters in the maximum likelihood model of SNP and genotype calling . The populations module was used to filter the catalogues and export data in the FASTA format ( for the mitochondrial variation ) and the variant calling format ( VCF , for the nuclear variation ) . The mitochondrial haplotype richness within and among groups ( Ae . aegypti forms and geographic regions ) was calculated using the rarefaction method implemented in the program HP-rare [29] . Phylogenetic relationship among mitochondrial haplotypes was estimated with the maximum likelihood approach in the program RAxML ( GTRM + G , rapid bootstrap heuristic algorithm and thorough ML search ) [30] . Haplotypes of three related Aedes species , for which the whole mitochondrial genome sequences were available , served as outgroups: Ae . albopictus ( NCBI: NC_006817 . 1 ) , Ae . notoscriptus ( NC_025473 . 1 ) [31] and Ae . vigilax ( KP995260 . 1 ) [32] . Haplotype sequence of the Ae . aegypti reference line ( Liverpool , NC_010241 . 1 ) was also included in the analysis . Parameters of data quality and diversity ( RAD tag depth , percentage of missing data , heterozygosity averaged per individual ) were compared between females of the two co-occurring forms using independent sample t-test . The level of nuclear genetic structuring was estimated using the non-spatial multivariate method DAPC [33] in the R package adegenet [34] . Rousset’s genetic distance ( â ) and geographic distance between pairs of individuals were calculated in the program spagedi [35] . Color distance between pairs of individuals was treated as a binary value: 0 ( same color/form ) and 1 ( different color/form ) .
From the mitochondrial RAD tag catalogue , we extracted 13 polymorphic tags that were shared between at least 80% of individuals ( 60/74 , Table 1 ) . Tags were distributed across eight different mitochondrial genes ( COXI , Cytb , ATP6 , ND1-2 , ND4-6; S1 File ) . All 13 tags were concatenated into a final 1170 bp long sequence that was treated as a mitochondrial haplotype . We found 24 different haplotypes in samples from Singapore and Townsville . Haplotype richness did not differ between the two forms in either location ( Singapore type = 5 . 13 , queenslandensis = 5 . 07; Townsville type = 4 . 19 , queenslandensis = 5 . 0 ) . Moreover , seven haplotypes were shared between the two forms ( Table 1 , Fig 2 ) . There were 207 distinctive alignment patterns and 8 . 17% of undetermined characters in the dataset consisting of 24 haplotypes from Singapore and Queensland , one from the Liverpool strain and three from other Aedes species ( outgroups ) . A phylogeny based on maximum likelihood revealed two highly statistically supported maternal lineages in Ae . aegypti: a basal clade ( more similar to the outgroups ) and a clade arising from it ( a derived clade ) ( Fig 2 ) . Nucleotide distance ( p-distance ) between the two clades ranged from 1 . 2% to 1 . 6% ( S2 File ) . Importantly , haplotypes of the two Ae . aegypti forms were found in both clades , indicating no association between mitochondrial variation and color variation ( Fig 2 ) . While our results do not support the tentative patterns suggested by Chan et al . [14] , they match those from the most comprehensive mitochondrial phylogeny of the African and global Ae . aegypti generated to date [10] . Using the ND4 variation , Moore et al . [10] showed that Ae . aegypti populations outside Africa represent “mixtures” of mosquitoes from the basal clade and the derived clade , with the basal clade likely originating from West Africa and the derived clade mainly from East Africa . Our analyses of the mitochondrial genome-wide variation revealed the same matrilineage structure in populations from Singapore and northern Queensland ( Fig 2 ) . A lack of mitochondrial distinctiveness between the queenslandensis and the type form is also in line with the findings of Moore et al . [10] , who could not separate the type and formosus forms into distinct mitochondrial clades despite their assumed subspecies rank . We extracted nuclear RAD tags that were shared between at least 80% of individuals in the entire dataset ( Singapore , Townsville , Gordonvale , Ho Chi Minh City and Rio de Janeiro ) . To avoid redundant information from the highly linked markers , we randomly selected one SNP per tag with a minor allele frequency greater than 5% , which gave a total of 16 , 569 markers for downstream analyses . Parameters of data quality and diversity did not significantly differ between the co-occurring queenslandensis and type individuals , including the average: percentage of reads uniquely aligned to the reference genome ( Singapore: t df , 27 = 1 . 46 , p = 0 . 15; Townsville: t df , 26 = 0 . 782 , p = 0 . 44 ) , locus depth ( Singapore: t df , 27 = 1 . 66 , p = 0 . 11; Townsville: t df , 26 = -1 . 73 , p = 0 . 095 ) , percentage of missing data ( Singapore: t df , 27 = -0 . 67 , p = 0 . 51; Townsville: t df , 26 = 0 . 951 , p = 0 . 35 ) , or heterozygosity ( Singapore: t df , 27 = 0 . 46 , p = 0 . 65; Townsville: t df , 26 = -2 . 42 , p = 0 . 023 ) ( Table 1 , S1 Fig ) . Discriminant analysis of principal components ( DAPC ) showed a clear-cut differentiation of mosquitoes based on their geographic origin and not their color . When the entire dataset was considered , Ae . aegypti individuals formed genetic clusters that corresponded to their sampling region ( i . e . Rio de Janeiro , Ho Chi Minh City , Singapore and northern Queensland ) ( Fig 3a ) . The only exceptions were three individuals in Singapore ( K-M ) that formed a distinct genetic group ( Fig 3a ) . They were collected as larvae from the same breeding container , and two were identified as the type form and one as the queenslandensis form ( Table 1 , Fig 3b ) . Given their high relatedness ( Supplemental file 4 ) and shared mitochondrial haplotype , as well as high nuclear differentiation from other mosquitoes in the region , it is likely that individuals K , L and M are offspring of an incursion female ( s ) not local to Australia and Vietnam . These individuals were found near the city port , suggesting a possible route of introduction . Further analysis of genetic structuring within Singapore revealed that family groups were sampled within the breeding containers , some of which had both color forms ( Fig 3b ) . Highly related queenslandensis and type pairs were found at four locations ( Fig 3b ) , including the incursion family group ( K , L , M ) . Most of the related individuals ( 24/28 pairs ) , however , had the same color ( Fig 4 ) . These results suggest that the color/scaling pattern is likely to represent a quantitative trait under some environmental influence ( e . g . temperature , humidity , light , nutrient availability ) . The frequency of the color forms has been shown to vary between the dry and the wet season in Ae . aegypti populations from Surabaya , Indonesia [36] . Also , the dorsal abdominal scaling pattern responds to artificial selection [36 , 37] and multiple QTLs associated with this trait have been recently reported [37] . Individuals with the color/scalling patterns corresponding to the queenslandensis form have also been observed ( albeit rarely ) in our laboratory colonies which are maintained by occasional crossing to field-caught type Ae . aegypti ( Jason Axford , personal communication ) . Individuals from northern Queensland were grouped into three clusters corresponding to the three towns where the sampling occurred ( Fig 3c ) . An exception was one queenslandensis individual from Cairns that was grouped with the type individuals from Gordonvale ( Fig 3c ) . The two forms in Townsville could not be distinguished based on their nuclear genome-wide variation ( Fig 3c ) . We found four pairs of closely related individuals: two queenslandensis and two type pairs ( Fig 4 , S3 File ) . In other words , all related pairs detected in Townsville were of the same form . A lower frequency of related individuals in Townsville when compared to Singapore is not surprising given that different sampling methods were employed in these locations . Collection of multiple larvae from the same breeding container increases the chance of sampling family groups , as seen in Singapore and parts of Rio de Janeiro [22] . On the other hand , when BG-sentinel traps are used , the likelihood of related individuals being collected is low . In Townsville , 12 . 5% of pairs from the same trap were close relatives . Sampling effects are reflected in an elevated level of pairwise genetic distance over geographic distance for mosquitoes from Singapore when compared to Townsville ( Fig 4 ) . Such differences in genetic patterns could be erroneously interpreted as differences in the underlying processes ( e . g . different dispersal rates ) , and highlight that sampling methods are crucial when inferring processes within and among Ae . aegypti populations . In summary , we did not find any association between nuclear genetic variation and color/scaling variation in Ae . aegypti from Singapore and northern Queensland . Our results are unlikely to be caused by a lack of power to detect genetic structure , given that more than 16 , 000 genome-wide SNPs allowed us to delineate family groups at a very fine spatial scale . In fact , several families had the queenslandensis and type members . A recent study of global Ae . aegypti populations at 12 microsatellite loci found that at least in one locality in Africa ( Senegal ) the two established forms ( formosus and type ) are interbreeding with no sign of genetic subdivision when brought into sympatry [11] , so it is not surprising that the type and queenslandensis variety also form one genetic cluster . Our results also help explain the similar vectorial capacity for a dengue 2 viral strain of type and queenslandensis females originating from wild Ae . aegypti in Thailand [38] . In addition to an absence of genetic structuring between the two Ae . aegypti aegypti forms , another line of evidence in support of ongoing interbreeding is the presence of Wolbachia in both forms . We detected this endosymbiotic bacterium in three ( out of four ) queenslandensis individuals from Cairns and 14 ( out of 15 ) type individuals from Gordonvale , using a light-cycler assay for Wolbachia detection [39] . Wolbachia is not naturally found in Ae . aegypti , but was introduced into the populations in Gordonvale in 2011 and Cairns in 2013 in an effort to reduce dengue transmission [40 , 41] . This was done by releasing Wolbachia-infected females and males from a colony that originated from type Ae . aegypti [42] . Because the infection is transmitted from mother to offspring , the only way queenslandensis individuals could have become infected by Wolbachia is by mating with infected , type females . Given the high Wolbachia frequency ( > 85% ) in Cairns and Gordonvale at the time of our sampling [41] , the presence of the infection in 3/4 individuals caught in Cairns , and 14/15 individuals caught in Gordonvale was expected .
Our analyses of mitochondrial and nuclear genome-wide variation and the Wolbachia infection indicate that Ae . aegypti queenslandensis and Ae . aegypti type mosquitoes interbreed freely , at least in Singapore and northern Queensland . These findings are of practical importance for control strategies that rely on successful mating between the released and target mosquitoes . Our results also re-enforce the recommendations by the early taxonomic authorities ( Mattingly and McClelland ) that the extant Ae . aegypti queenslandensis should not be ranked as a subspecies . | Aedes aegypti , the most important vector of dengue and Zika , greatly varies in body color and behavior . Two domestic forms of this mosquito , the very pale queenslandensis and the browner type , are often found together in populations around the globe . Knowing how freely they interbreed is important for the control strategies such as releases of Wolbachia and sterile males . To address this question , we used RAD sequencing to genotype samples of both forms collected in Singapore and northern Queensland . We did not find any association between the mitochondrial or nuclear genome-wide variation and color variation in these populations . Rather , “paleness” is likely to be a quantitative trait under some environmental influence . We also detected several queenslandensis individuals with the Wolbachia infection , indicating free interbreeding with the type form which has been used to introduce Wolbachia into Ae . aegypti populations . Overall , our data show that the very pale queenslandensis are not genomically separate , and their presence is unlikely to affect the success of Aedes control programs based on Wolbachia-infected , sterile or RIDL mosquitoes . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"&",
"Discussion",
"Conclusion"
] | [
"biogeography",
"invertebrates",
"plant",
"anatomy",
"medicine",
"and",
"health",
"sciences",
"ecology",
"and",
"environmental",
"sciences",
"population",
"genetics",
"geographical",
"locations",
"animals",
"wolbachia",
"singapore",
"plant",
"science",
"mitochondria",
"bi... | 2016 | The queenslandensis and the type Form of the Dengue Fever Mosquito (Aedes aegypti L.) Are Genomically Indistinguishable |
Soil-transmitted helminths ( STHs ) affect more than 1 . 5 billion people . The global strategy to control STH infections requires periodic mass drug administration ( MDA ) based on prevalence among populations at risk determined by diagnostic testing . Widely used copromicroscopy methods to detect infection , however , have low sensitivity as the prevalence and intensity of STH infections decline with repeated MDA . More sensitive diagnostic tools are needed to inform program decision-making . Using an integrated product development process , PATH conducted qualitative and quantitative formative research to inform the design and development of a more sensitive test for STH infections . The research , grounded in a conceptual framework for ensuring access to health products , involved stakeholder analysis , key opinion leader interviews , observational site visits of ongoing STH surveillance programs , and market research including market sizing , costing and willingness-to-pay analyses . Stakeholder analysis identified key groups and proposed strategic engagement of stakeholders during product development . Interviews highlighted features , motivations and concerns that are important for guiding design and implementation of new STH diagnostics . Process mapping outlined current STH surveillance workflows in Kenya and the Philippines . Market sizing in 2016 was estimated around half a million tests for lower STH burden countries , and 1–2 million tests for higher STH burden countries . The cost of commodities per patient for a molecular STH diagnostic may be around $10 , 3–4 times higher than copromicroscopy methods , though savings may be possible in time and staffing requirements . The market is highly price sensitive as even at $5 per test , only 27% of respondents thought the test would be used by surveillance programs . A largely subsidized STH control strategy and a semi-functional Kato-Katz test may have created few incentives for manufacturers to innovate in STH diagnostics . Diverse partnerships , as well as balancing needs and expectations for new STH diagnostics are necessary to ensure access to needed products .
Soil-transmitted helminths ( STHs ) remain a massive global health problem . Each year , more than 1 . 5 billion people suffer from infection with these parasitic intestinal worms [1] . STH infections have significant harmful effects on the health and well-being of individuals , especially children , in whom infection may lead to malnutrition , anemia , abdominal pains , stunted growth , and poor cognitive development [2–4] . The focus of the global STH control strategy is reducing morbidity in the most at-risk populations—such as preschool-age children , school-age children , and women of reproductive age—through mass drug administration ( MDA ) of donated deworming drugs , such as mebendazole and albendazole [5] . Program decisions for use of MDA currently rely on detecting the presence and load of helminth eggs in stool samples from at-risk groups . The recommended tests for measuring STH infections during surveillance activities involves microscopic examination of stools for STH eggs by Kato-Katz ( KK ) or mini-FLOTAC methods [6] . Although both copromicroscopy methods are adequate to support program decisions when the prevalence and intensity of infections are moderate to high , they lack sensitivity with light-intensity infections , especially after MDA has reduced infections to low levels [7–10] . PATH supported the efforts of the 2012 London Declaration on Neglected Tropical Diseases ( NTD ) by conducting a diagnostic gap analysis for 9 NTDs: STH , Schistosomiasis , Onchocerciasis , Lymphatic Filariasis , Trachoma , Chagas disease , Leprosy , Visceral Leishmaniasis , and Human African Trypanosomiasis [11] . Based on findings of a lack of sufficient diagnostic tools to address the current control strategy for STH infections , further research and development on a new , more sensitive diagnostic tool was initiated . Building on the identified use cases and findings from the previous STH diagnostic gap analysis , the dominant need was a diagnostic tool to support decision-making for reducing or discontinuing MDA when the intensity of STH infection is low [12] . Concurrently , health impact modelling was performed and suggested that a more sensitive diagnostic could potentially add the greatest value in this setting [13] . By accurately detecting low-intensity infections , a new test may not only help to improve decision-making related to use of MDA but also contribute toward the eventual elimination of STH infections . PATH’s integrated product development process includes an early consideration for user and customer needs , as well as implementation constraints and market dynamics [14] . ( Fig 1 ) Planning for sustainable access throughout the product lifecycle is imperative to achieving public health impact [15] . To overcome the shortcomings of current diagnostic methods for guiding MDA use , PATH conducted formative research to inform the design and development of a new , more sensitive test . This work included both user and market research and evaluated factors such as stakeholders and policies , impact and affordability , and user needs and acceptability . The Access Framework developed by Reich and Frost provided a conceptual framework for this formative research , which identifies four important factors in moving promising innovations across the value chain from development to large-scale impact: architecture , availability , affordability , and adoption [16] . ( Fig 2 ) Key principles of this framework include [17]: The findings from this formative research informed PATH’s development of a new molecular diagnostic test for STH infections . The new test , named Dx4STH for easier reference , is a molecular diagnostic kit product that is rapid and sensitive , using a type of isothermal nucleic acid amplification called recombinase polymerase amplification ( RPA ) to detect four key STH species . Technical aspects of the product development will be discussed elsewhere . The focus here is on the methods and results of the formative research , and the research findings’ influence on PATH’s product development activities . This work may also serve as a foundation for any future research on STH diagnostics to ensure access to needed products .
Formative research using qualitative and quantitative methods was conducted with the objective to investigate factors that could facilitate or hinder end-users’ access to a new , more sensitive test for STH infection . This formative research was conducted before and during product development from roughly 2014 to 2017 . This study was reviewed by the PATH Research Ethics Committee and determined to be not human subjects research . Early on , a stakeholder analysis was conducted focused on the development and implementation of STH diagnostics . Stakeholder analysis is a “process of systematically gathering and analyzing qualitative information to determine whose interests should be taken into account when developing and/or implementing a policy or program”[18] , or product development project in this case . It is also a means to capture the unique attributes of a stakeholder audience , to better communicate and share information about the product or project that is tailored to their unique interests or concerns . Stakeholders were identified through a review of peer-reviewed literature , grey literature , World Health Organization ( WHO ) policies , organization websites , and participant lists from relevant recent meetings . Assessments by the research team were made of their knowledge , interest , position , and ability to effect change , using a three-point scale based on perceived influence and interest . Stakeholders were also grouped into categories ( e . g . , manufacturer , donor ) based on their roles in the development and use of STH diagnostics . The search continued until no new categories were identified . Because stakeholders could fall into more than one category , the analysis was performed at the category level and resulted in a matrix to guide a participatory , consensus-building product development process . Next , semi-structured interviews were conducted with key opinion leaders to better understand STH surveillance needs and activities in relation to surveillance for other neglected tropical diseases ( NTDs ) . These interviews focused on stakeholders involved in programmatic aspects of NTD surveillance . A purposive , snowball sampling strategy was used to obtain a heterogeneous range of responses until saturation was reached [19] . Of the 14 individuals contacted , 10 interviews were conducted . Interview topics included the role of programmatic personnel in surveillance , current NTD surveillance activities , and STH surveillance and tools , including issues related to specimen collection , populations surveyed , instrument platforms used , and time to result . The individuals interviewed had experience with NTD surveillance programs in Africa , Asia , and the Americas . Interviews were recorded , transcribed , and then analysed using thematic content analysis . A codebook was created based on a priori themes and revised during analysis based on emerging themes . Data were analysed by one researcher and reviewed by the research team to discuss findings . Ethnographic research in the form of observational site visits were performed to create process maps for government-led STH surveillance in Kenya and the Philippines [19] . PATH researchers identified and shadowed Ministry of Health-led teams conducting ongoing STH surveillance . Observations were captured using the AEIOU heuristic , which is a method to categorize data based on activities , environments , interactions , objects , and users [20] . Surveillance process maps were drafted , reviewed by field managers , and revised based on feedback . Key existing policies were also identified that support the use of STH diagnostics in the surveillance processes observed . Market research was conducted to quantify potential markets , forecast health and economic impacts of using more sensitive STH diagnostics , estimate costs of goods , and assess willingness to pay . Findings from the health and economic impact modelling can be found elsewhere [13] . To quantify the potential market , country-specific data were used to estimate the total number of tests needed worldwide . Data from United Nations , Department of Economic and Social Affairs was used to determine the total number of children ages 5–14 years per country in 2013 and 2016 [21] . Data from the WHO STH data repository was used to determine the number of SAC requiring PC by country [22] . Calculations were originally determined with data from 2012 , and then updated with 2016 data . Based on WHO guidelines , it was assumed that countries would conduct testing in five to ten schools per district , sampling 50 students per school , across the entire population of SAC because STH infections are not focally distributed [23] . The number of districts per country was calculated by dividing the total number of SAC by 200 , 000 as recommended in the guidelines for sentinel site monitoring . To create a standard method for approximating STH prevalence , the proportion of school-age children requiring PC for STH was determined by comparing the number of SAC requiring PC to the total number of children ages 5–14 years in the country . ( S1 File ) A cost analysis was performed for 2 copromicroscopic methods and 4 molecular methods for detecting STH . Data for the Kato-Katz and FLOTAC method was derived from a 2010 publication and adjusted for 2016 costs [24] . For both the Kato-Katz and FLOTAC methods , the costs of instruments were considered primarily microscopes and assumed to be sunk costs as they already exist in most facilities and have a long life-span , thus they are not included in the analysis . The molecular methods included 2 product concepts in development at PATH , one using Dx4STH with the common DNA extraction kits used with qPCR , and a second method using Dx4STH with a simpler , prototype DNA extraction method that involves a modified alkaline lysis-magnetic bead protocol ( MAL-MB ) . The prototype MAL-MB protocol was developed as a simpler method for DNA extraction from stool requiring less equipment and consumables . Multiparallel qPCR and multiplex qPCR were also included as they were conducted during product development as reference assays . For the molecular methods , all instrument costs including ancillary laboratory equipment such as centrifuges , magnetic racks , and heat baths were incorporated making a conservative assumption that these would be new instruments added to existing facilities where surveillance would take place . Additionally , optimal workflows and staffing for the molecular methods were determined based on protocols used in the PATH laboratory . ( S2 File ) . Estimates were derived for roughly 100 samples per day , based on WHO guidelines . Summary costs were calculated by adding up component costs for instruments and consumables , then dividing by throughput . Staff time and infrastructure costs were not included for any methods , as comparable estimates across the 6 methods were not available . To account for the accuracy of the method , the total cost per accurate diagnosis was determined by dividing the total cost by the accuracy of the method . All details for the costing analysis are provided as supporting information . ( S2 File ) A willingness-to-pay analysis was performed using both a perceived-value pricing ( PVP ) approach and a Gabor-Granger approach [25] . The PVP approach assesses pricing expectations and perceived usefulness of a new STH diagnostic using a scale between 0–100 , 0 indicating “not at all useful” and 100 indicating “extremely useful” . Plotting perceived usefulness against the perceived price estimates the perceived value for money without reference to specific price points . Respondents were provided product primers to inform their assessments . To assess the price sensitivity of the market for a new STH diagnostic , the Gabor-Granger approach was used as a price experiment method [25] . The Gabor-Granger method uses a 5-point scale to assess the highest price a respondent would be willing to pay based on a range of fixed prices shown to the respondent . The two methods were used in combination to more accurately capture potential differences between customers and users . Data for these methods was prospectively collected using semi-structured interviews with 29 respondents across a range of stakeholder groups , and 15 countries in Africa , Asia , and the Americas . This analysis was not designed for statistical significance and the target sample size was 20–35 respondents to achieve a heterogeneous distribution of experience and geographical spread . Participation in any or all questions was voluntary , and no personal identifiable information was shared with the research team . These interviews built on , but were separate from the initial stakeholder interviews conducted , and were focused on assessing the commercial viability of the Dx4STH method .
A total of 70 stakeholder organizations were identified and grouped into seven categories: Donor , Manufacturer , MDA Implementer , Country Program , Researcher/Developer , Policymaker , and Advocacy Program . Because the stakeholder assessment is from the perspective of a product development partner , this category is not included . Advocacy Program was classified as both highly influential and interested because of its role as a champion for change and scientific advancement . Researcher/developer was classified as having medium influence and high interest because years of investigation by dedicated scientists have led to many advances in STH diagnostics despite limited funding . Donors were classified as highly influential with medium interest because donors fund most research and development for STH diagnostics but often consider diagnostics a lower priority within the spectrum of STH control efforts . Policymakers were classified as having medium influence and medium interest due to this category’s role in building consensus and reacting to the needs of the STH community . Manufacturers were classified as having medium influence and low interest because they have the capabilities needed to advance STH diagnostics but lack market incentives . Country programs have low influence and medium interest because their desire for improvements in diagnostics depends on other stakeholders to develop products and create a supportive system for implementation . The MDA implementer has low influence and low interest because the role of diagnostics in MDA distribution is poorly defined . After assessing perceived influence and interest , the results distributed stakeholder categories across 4 quadrants , representing varying strategies for communications of the project and the outcomes , none of which have less value and importance . Elements that vary across the groups include frequency , preferred medium , and complexity of communication . “Manage closely” may involve the greatest effort to fully engage and satisfy , “Keep satisfied” may involve enough effort to keep satisfied without boring or overloading , “Keep informed” may involve adequate informing through regular communication to ensure no issues are present with the details of the work , and “Monitor” may involve some but not so much communication as to overload . ( Fig 3 ) In total , 10 key opinion leaders were interviewed . Open-ended questions were asked about strengths and weaknesses of current STH surveillance and diagnostics . The reasons for doing STH surveillance were noted as determining baseline measures , especially as LF programs are winding down , as well as for informing decisions to reduce or stop treatment . An interviewee commented that STH infections is a low priority for most donors within the NTD landscape , because STH infections is associated with relatively low morbidity despite high prevalence . Although interviewees acknowledged the limitations of the Kato-Katz method for detecting low-intensity infections , they were generally satisfied with using it for STH surveillance . The attributes of Kato-Katz that were most commented on were that it is well known , easy to use , easy to implement at the local level , and it is a quantitative measurement that can speciate the different STH infections . Some interviewees noted that a test with higher sensitivity than Kato-Katz would be useful . It was also noted that any new test should have a quantitative readout because data on the intensity of infection are used in a variety of ways outside of informing MDA frequency , such as evaluating effectiveness of MDAs , measuring morbidity , and other research purposes . Interviewees commented that a qualitative molecular diagnostic tool may be valuable only when STH prevalence dropped well below 10% to 20% , which represents a small segment of the current market . Several interviewees commented on the feature of easy to use , noting that resource requirements ( personnel and infrastructure ) necessary for any new test should be like those for Kato-Katz . Highlighted in Fig 4 are select features , motivations and concerns that are useful to consider when guiding the design and implementation of a new STH diagnostic . ( Fig 4 ) Interviews also explored opportunities and barriers to integration of STH surveillance across the NTDs , and the role diagnostics may play . Achieving sustainable access to new STH diagnostic products through appropriate use and health impact , requires a consideration of the surveillance ecosystem across NTDs because it provides opportunities for synergy , and also challenge in the form of competing interests . Positive attributes were noted by most interviewees such as opportunities for capitalizing on and conserving available resources . However , the challenges associated with integration were also noted . Interviewees noted that implementing change is difficult . Currently , disease control guidelines are separate , agendas are separate , and funding is separate . Although some logistics are similar , such as community-based testing , other activities and required skillsets differ making workforce management more complex . Overall , interviewees considered the integration of STH surveillance with surveillance for other NTDs to be a “nice to have” rather than a high priority . Observational site visits in Kenya and the Philippines were conducted by 2 researchers . WHO policies define many procedures of the surveillance activities , such as sentinel site sampling methodologies to select children per school and schools per evaluation unit to ensure prevalence estimates are robust [5 , 23 , 26] . Some aspects that vary by location are due to workflow logistics and varying cultural sensitivities by country . Each day of surveillance included all necessary procedures for data collection from one school including sample collection , processing , analysis , and reporting . ( Fig 5 ) In the Philippines , collection containers and consent forms were sent home with students the night before allowing parents to help children in sample collection . In Kenya , consent forms were sent home with children prior to sample collection , though collection containers were distributed the morning of the surveillance activity and collected the same day . After filled sample containers were collected from all selected children , specimens were transported to a nearby district lab or clinic facility for processing . Transport time varied day to day as the locations of schools and district facilities were not consistent . All necessary materials were transported by the surveillance teams . The district facilities all had available bench space , electrical supply for a light microscope , and access to water for cleaning slides , though infrastructure and setup varied by location . Processing and analysis were performed concurrently , with several laboratory technicians or surveillance workers preparing the Kato-Katz slides while the microscopists read and recorded the results . The time to read each slide depended on the number and species of eggs detected . After results were recorded on paper forms , data were transferred to electronic forms for dissemination at the national and regional level . At the end of the day , materials were cleaned and repacked for the next day of data collection at another school and district facility . Surveillance teams worked Monday through Friday for several consecutive weeks at a time , before returning to other job functions . Data was gathered for 2012 and 2016 , to incorporate the newest STH disease burden data available . In 2012 , WHO identified that 111 countries required PC for SAC , and in 2016 103 countries required PC for SAC . As a surrogate for STH prevalence , the proportion of SAC requiring PC was calculated by dividing the number of SAC requiring PC by the total number of children between the ages of 5 and 14 years . In 2016 , 4 countries with STH burden data did not have disaggregated UN population data ( Dominica , Marshall Islands , Nauru , and Tuvalu ) so the number of STH tests needed was estimated for 99 countries in 2016 . Based on the guidelines , the total SAC population was divided by 200 , 000 to estimate the number of districts per country [23] . In 2012 , potential market size for all STH surveillance tests ranged from 1 . 8 to 3 . 7 million when sampling five or ten schools per district and 50 children per school . Some countries had both high proportions of SAC requiring PC ( >25% of school-age children ) and high numbers of tests ( >100 , 000 tests ) . These countries included Bangladesh , Brazil , Ethiopia , India , Indonesia , Nigeria , Pakistan , and the Philippines when sampling 5 schools per district , as well as Colombia , DRC , Kenya , Myanmar , and Tanzania when sampling 10 schools per district . In 2016 , potential market size ranged from 1 . 3 to 2 . 5 million tests . The remaining high market size countries were India sampled at 5 schools per district , as well as Indonesia , Nigeria , and Pakistan sampled at 10 schools per district . ( Fig 6 ) To estimate the market size for a diagnostic test that could stop or reduce MDA , the proportion of SAC requiring PC was calculated . WHO guidelines recommend PC for SAC populations with greater than 20% prevalence at baseline . After starting PC , reductions in PC are recommended when the prevalence is less than 20% [23] . The market was estimated for countries with higher STH burden , defined as greater or equal to 20% of SAC requiring PC , or lower STH burden , defined as less than 20% SAC requiring PC . In 2012 , there were 29 lower burden countries and 82 higher burden countries , compared to 15 lower burden countries and 84 higher burden countries in 2016 . The potential market size in 2012 was 355 , 000 tests if five schools were sampled and 710 , 000 tests if ten schools were sampled for lower burden countries , and 1 . 5 million and 3 . 7 million tests for higher burden countries . The potential market size in 2016 was 247 , 000 tests if five schools were sampled and 494 , 000 tests if ten schools were sampled for lower burden countries , and 1 . 0 million and 2 . 1 million tests for higher burden countries . Table 1 shows the results of our analysis comparing the fixed and variable costs of several types of diagnostic tests for STH . These estimates assumed a similar workflow of roughly 100 samples per day . Kato-Katz resulted in the least expensive method at a total cost of $2 . 29 per patient . FLOTAC was more expensive than Kato-Katz due to consumable costs . The molecular methods were on average 4–6 times the cost of Kato-Katz varying from $7 . 19 per patient for multiplexed qPCR to $11 . 91 per patient for multiparallel qPCR . Dx4STH was $9 . 76 per patient when using a comparable DNA extraction method to the qPCR methods and decreased to $8 . 33 per patient when a prototype MAL-MB method was used . The largest cost component for molecular methods are consumables , which comprise over 90% of the overall cost . Optimal staffing requirements for all 6 methods were between 3 and 6 staff . For the molecular methods , the time needed to analyse the samples ranged from 4 . 8 hours for the Simple Dx4STH method and 7 . 1 hours for the multiparallel qPCR method . When considering the accuracy of the method , the total cost per accurate diagnosis increases for all methods to a cost per patient of $0 . 06 for Kato-Katz , $0 . 66 for FLOTAC , $9 . 01 for Simple Dx4STH , $10 . 55 for Spin Dx4STH , $12 . 28 for multiparallel qPCR , and $7 . 41 for multiplex qPCR . ( Table 1 )
In this formative research , an early consideration for user and customer needs , implementation constraints , and market dynamics was used to inform the design and development of a new , more sensitive surveillance test for STH . A focus on access throughout the product lifecycle is central to PATH’s work as a product development partner and is important to achieving health impact through new technologies . Several factors were evaluated here through user and market research , including stakeholder and policy concerns , potential market size , affordability , and acceptability . Similar to needing broad implementation partnership to achieve the control and elimination goals for STH infections [5] , the development and use of new tests for STH will also require broad support across diverse stakeholder groups . Early identification and engagement of key stakeholders is important throughout the product development process . Based on the stakeholder analysis , donors , advocacy programs , and researchers/developers should be engaged with most intensively as they have the most interest and influence in the development of new diagnostics . Obtaining input from manufacturers and policymakers is also key to ensuring sustained access to new diagnostic products in the long term . Specifically , early engagement of manufacturers that can ensure sustained , efficient production and distribution of the final product is particularly critical if meeting reliability and regulatory requirements is a priority . Equally important , country programs should be regularly engaged and informed of progress in diagnostic development as they are the target users , and ultimately , adoption and increased country ownership is dependent on their needs being met [26] . Stakeholder interviews , process mapping , and literature reviews contributed to PATH’s development of target product profiles ( TPP ) for more sensitive STH surveillance tests in 2015 . The process of developing these TPPs for STH , as well as Schistosomiasis and Trachoma , was described in a report ( https://www . path . org/resources/diagnostics-ntd/ ) . Since then , global stakeholders meetings on STH diagnostics occurred in 2016 and 2017 , namely meetings in Ghent , Belgium and Annecy , France [27] . A challenge considered was balancing the diagnostic needs of the shorter-term goal to monitor progress towards WHO 2020 milestones with the diagnostic needs of the longer-term program goal of transmission interruption . Additional TPP development occurred through a workshop of experts and key opinion leaders , led by the Bill and Melinda Gates Foundation [28] . Of the two TPPs that were developed , the authors noted that one TPP had most of its minimum criteria met by the Kato-Katz method , limiting the role for new diagnostics to addressing the second TPP . Two attributes that Kato-Katz does not meet , however , are stringent regulatory standards ( ISO ) and quality assurance requirements , which will also be important for any new STH diagnostic and may be addressed through a central laboratory “command center” to ensure unified provision of quality assured diagnostic services [29] . At the STH Advisory Committee meeting in 2017 , the qPCR platform was highlighted noting its improved sensitivity in field studies . Standardization , preservation , and DNA extraction methods are the focus of ongoing research . Acknowledgement of the following diagnostic development goals was also stated: increased coordination among partners for improved efficiency , early consideration of field use , focus on test cost and ease of use , and a standardization of techniques [30] . The cost and cost-effectiveness of new diagnostics were important considerations from the beginning of the project . Health impact modeling , performed to understand potential benefits resulting from improvements in diagnostic sensitivity , found that improved diagnostics increase the probability of elimination by a control program through optimized MDA frequency [13] . Interestingly , using more sensitive diagnostics would increase disability adjusted life years ( DALYs ) averted , and in some scenarios result in an earlier decision to stop MDA , though most often it may result in more frequent MDA . In either case , resources would be more efficiently deployed to improve health impact [13] . The size of the market for high-sensitivity tests may also be a relatively small portion of the overall market for STH diagnostics . In 2016 , of the countries requiring PC , 15% had a proportion of SAC requiring PC below 20% . From 2013 to 2016 , the number of these lower STH burden countries decreased from 29 to 15 , while the higher STH burden countries changed little . Where STH control programs have not yet achieved high MDA coverage , prevalence and intensities of infection may not be reduced to levels low enough to warrant use of more sensitive and expensive molecular tests . The relatively small potential market for high-sensitivity tests may also restrict interest among commercial manufacturers and result in a higher price per test due to the economies of scale . The Dx4STH method would cost somewhere between multiparallel and multiplex qPCR , and all 4 molecular methods were substantially higher than the costs per patient of the copromicroscopic methods . The costing analysis did not include infrastructure or salary costs because comparable estimates were not available for all 6 methods . Most of the cost of Kato-Katz and FLOTAC are salaries [24] , so estimates of these methods are artificially low here . However , all 6 methods will have costs due to infrastructure and salaries , though these estimates may vary by method and location . More skilled staff may have higher pay per unit of time . Considering the skill level of the staff , number of personnel for an optimal workflow , and duration of work would impact final costs of a diagnostic method . Additionally , more equipment would add more instrument costs but may also decrease staffing requirements and time . Finally , stakeholders’ willingness to pay for a new molecular test may be lower than the anticipated cost of goods . Interviewed stakeholders believed that donors consider STH infections a lower priority within the NTD landscape due to lower morbidity , potentially influencing perceptions of test value . Cost-effectiveness or cost-benefit analysis will be needed to make the case for use of molecular tools in specific situations . Overall , the commercial viability of a molecular diagnostic product for STH may be low initially , making donor engagement essential for further product development and introduction . The perceived usefulness of Kato-Katz combined with the higher price for molecular diagnostics contribute to an overall low willingness-to-pay . If initial test volumes needs are low , the absolute cost increase may be acceptable as efficiencies and savings are optimized . The other challenge is that the price sensitivity of the market for new STH diagnostics may be related to the largely donation driven STH control and elimination strategy , which depends on large-scale donation of drugs by pharmaceutical companies , as well as donor dollars for program activities such as implementation , and monitoring and evaluation . The sustainability of long-term dependence on drug donations is uncertain , making the need for innovative , cross-sector drug development partnerships increasingly important [31] . Similarly , innovative partnerships in diagnostic development are also essential to adjust the failed diagnostics market . During this research , many stakeholders stressed the value of quantitative test results , such as for evaluating treatment effectiveness or estimating morbidity , even though decision-making related to MDA frequency is based on prevalence . Although the Dx4STH test was originally intended to produce only qualitative results to guide decision-making around MDA reductions and stopping [32] , stakeholder input led the team to pivot product development to a quantitative test . Compared with a quantitative polymerase chain reaction assay , a quantitative isothermal assay may have the potential to require less expensive and more field friendly equipment while being just as sensitive . Next steps in product development would include further verification testing of the Dx4STH alpha prototype , final development and verification testing of a sample preparation alpha prototype , as well as usability and validation testing in the field . In addition to product development for Dx4STH , further research is also needed to facilitate product adoption and ensure access to any new STH diagnostic , including the development of laboratory protocols for using new tests , additional market research with updated attributes , technology transfer to one or more manufacturers , regulatory and commercialization planning , and implementation research . Broad and inclusive collaboration across the global STH community will make this work more efficient and effective . There are some limitations to this research . Firstly , the value of formative research is for hypothesis generation . Further evaluative and summative research is necessary to answer questions , some of which may have been highlighted here . Observational site visits were performed in only 2 countries . Differences were identified between these settings , and many more differences are likely to be present when observing more countries . To conduct the market sizing , many assumptions had to be made , such as the frequency of STH surveillance , the distribution of at-risk populations , and the specifics of sentinel site monitoring . The frequency of STH surveillance is unclear as multiple rounds of MDA may be necessary to produce measurable differences . The risk factors for STH infections , such as poor sanitation and hygiene , are not evenly distributed across an entire population of SAC . Therefore , determining the number of districts to include in sentinel site monitoring based on the total school age population may be an overestimate . However , surveillance in other populations such as pre-school age children , women of reproductive age , and community-wide strategies may also adjust the market size . These market size estimates should be updated as new information is available . The costing analysis was based on a combination of published data , process mapping from the observational site visits , and learnings from in house product development for Dx4STH . Assumptions are detailed in supplemental file 2 ( S2 File ) and could be updated as new information is available . Finally , the willingness-to-pay analysis was conducted when Dx4STH was a qualitative test . Some findings suggest there may be more willingness to pay for a quantitative test , though how much is currently unclear . The need for new , more sensitive STH diagnostics has been recognized for some time , and many groups have made substantial progress in developing these new technologies . Formative research has supported PATH’s product development of a new molecular diagnostic kit that addresses user needs and implementation requirements while laying the groundwork for future access . Some barriers to moving further include the need for funding to support continued development and evaluation , as well as revisions to policies for STH surveillance to optimally support the use of newer technologies [26] . The inertia created by the cost-effectiveness of MDA with currently donated drugs and the functioning of the Kato-Katz method up to this point , make changing the status quo more challenging . The desire may be to create high sensitivity tests that function within existing guidelines and still maintain the low cost , low complexity , minimal infrastructure requirements expected from decades of using Kato-Katz . Reconsidering this expectation may speed progress to improve program decision-making , and ultimately , eliminate disease . | Soil-transmitted helminths affect more than 1 . 5 billion people , mostly in very poor regions without proper sanitation . To control infections , countries periodically give deworming drugs to populations at risk , such as school-age children , based on the results of diagnostic testing to determine the prevalence of infection . The current method to detect infection works when infections are prevalent , but is less useful after repeated deworming , when disease prevalence is lower . We conducted formative research—including user and market research—to inform development of a new diagnostic tool with improved sensitivity . Findings from this research highlight pain points with current methods and processes , which are also opportunities for innovation . Access to new STH diagnostics may improve program decision-making and eventually contribute to elimination of these infections . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"education",
"decision",
"making",
"sociology",
"tropical",
"diseases",
"social",
"sciences",
"parasitic",
"diseases",
"neuroscience",
"cognitive",
"psychology",
"cognition",
"neglected",
"tropical",
"diseases",
"extraction",
"tech... | 2019 | Formative research to inform development of a new diagnostic for soil-transmitted helminths: Going beyond the laboratory to ensure access to a needed product |
Polygenic risk scores ( PRS ) are designed to serve as single summary measures that are easy to construct , condensing information from a large number of genetic variants associated with a disease . They have been used for stratification and prediction of disease risk . The primary focus of this paper is to demonstrate how we can combine PRS and electronic health records data to better understand the shared and unique genetic architecture and etiology of disease subtypes that may be both related and heterogeneous . PRS construction strategies often depend on the purpose of the study , the available data/summary estimates , and the underlying genetic architecture of a disease . We consider several choices for constructing a PRS using data obtained from various publicly-available sources including the UK Biobank and evaluate their abilities to predict not just the primary phenotype but also secondary phenotypes derived from electronic health records ( EHR ) . This study was conducted using data from 30 , 702 unrelated , genotyped patients of recent European descent from the Michigan Genomics Initiative ( MGI ) , a longitudinal biorepository effort within Michigan Medicine . We examine the three most common skin cancer subtypes in the USA: basal cell carcinoma , cutaneous squamous cell carcinoma , and melanoma . Using these PRS for various skin cancer subtypes , we conduct a phenome-wide association study ( PheWAS ) within the MGI data to evaluate PRS associations with secondary traits . PheWAS results are then replicated using population-based UK Biobank data and compared across various PRS construction methods . We develop an accompanying visual catalog called PRSweb that provides detailed PheWAS results and allows users to directly compare different PRS construction methods .
The underlying risk factors of genetically complex diseases are numerous . Genome-wide association studies ( GWAS ) on thousands of diseases and traits have made great strides in uncovering a vast array of genetic variants that contribute to genetic predispositions to disease [1] . In order to harness the information from a large number of genetic variants , a popular approach is to summarize their contribution through polygenic risk scores ( PRS ) . While the performance of PRS to predict disease outcomes at a population level has been modest for many diseases , including most cancers , PRS have successfully been applied for risk stratification of cohorts [2 , 3] and recently have been used to screen a multitude of clinical phenotypes ( collectively called the medical phenome ) for secondary trait associations [4 , 5] . The goal of these phenome-wide screenings is to uncover phenotypes that share genetic components with the primary trait that , if pre-symptomatic , could shed biological insights into the disease pathway and inform early interventions or screening efforts for individuals at risk . Phenome-wide studies using PRS rely on an easily-calculated single biomarker that combines information across a spectrum of genetic variants . Additionally , PRS may be routinely available in patients’ electronic health records ( EHR ) in the near future , making analyses based on PRS a useful route for agnostic interrogation of the medical phenome . Existing literature has explored how to construct PRS with respect to a single disease phenotype [6 , 7] . In this paper , we demonstrate how polygenic risk scores can be used in concert with the medical phenome to better understand the etiology of disease subtypes nested within a broad disease classification . This is done by examining the shared and distinct genetic risk factors across the related but heterogeneous disease subtypes and also through our comparison of the secondary associations across the phenome corresponding to subtype specific PRS . In the post-GWAS era and with the availability of large biobank data from multiple sources , there is great interest in using gene-based biomarkers such as PRS for risk stratification and exploration of disease etiology . However , it is not always clear how best to construct a PRS for a particular phenotype . A PRS of the general form ∑i=1Kβi^Gi requires specification of three things: a list of markers G1 , G2 , ⋯GK , the depth of the list or the number of markers ( K ) , and the choice of the weights βi^ . These choices can be based on information extracted from the latest GWAS or GWAS meta-analysis ( when available ) , the NHGRI-EBI GWAS catalog of published results [1] ( when available ) , or summary data for GWAS corresponding to each phenotype , e . g . , from efforts that comprehensively screened the UK Biobank ( UKB ) phenome [8 , 9] . While various methods for constructing PRS have been widely studied for predicting the primary phenotype collected through population-based sampling [6 , 10] , it is unknown how the different PRS will be associated with subtypes and related phenotypes and the associations PRS can help unravel across the medical phenome . The comparative performance of different PRS construction methods may depend on the phenotype of interest . For example , diseases such as depression , which are believed to involve genetic contribution of a large number of genetic variants , might perform differently than diseases such as cancer , which may involve a smaller number of causal variants . We provide important empirical results comparing different PRS construction methods in terms of their associations with secondary and related phenotypes and in terms of the associations they identify across the phenome . In this paper , we first explore strategies for constructing a PRS using markers and weights obtained from either the latest GWAS or the NHGRI-EBI GWAS catalog that have reached genome-wide significance . We compare the PRS in terms of their performance [11] for the three most common skin cancer subtypes in the USA: basal cell carcinoma ( MIM: 614740 ) [12] , cutaneous squamous cell carcinoma [13] and melanoma ( MIM: 155601 ) [14] . We compare the two strategies using an independent biobank of genetic , demographic , and phenotype data collected by the Michigan Genomics Initiative ( MGI ) , a longitudinal biorepository effort within Michigan Medicine ( University of Michigan ) [4 , 15] . Based on these results , we choose a PRS construction strategy for each skin cancer subtype for further analysis . For the chosen PRS corresponding to each skin cancer subtype , we perform a phenome-wide association study ( PheWAS ) relating the PRS to the EHR-based phenome of MGI . We call such a study a PRS-PheWAS [4] . PRS-PheWAS results are then replicated using the population-based UK Biobank data . In order to identify secondary associations that are not driven by the primary phenotype , we perform an additional “exclusion” PRS-PheWAS for each skin cancer subtype in which we exclude subjects with any type of observed skin cancer [4] . These studies demonstrate similarities and differences in PheWAS results for PRS constructed for different disease subtypes , suggesting that PRS constructed for various disease subtypes can provide insight into shared and unique secondary associations . Our results further demonstrate the ability of such studies to reproduce known associations between secondary phenotypes and particular disease subtypes through use of PRS . We then describe an approach for using PRS to more directly evaluate the shared and unique genetic architecture of disease subtypes and identify shared and unique secondary phenotype associations related to this genetic architecture . We define a new PRS for each skin cancer subtype using loci unique to that subtype’s chosen PRS . We further construct a composite PRS for general skin cancer consisting of loci common among all subtypes’ PRS . While merging distinct clinical entities into a compound PRS may seem counterintuitive in terms of specificity , such an approach may increase power to identify dermatological features through PheWAS that are shared by all three subtypes . Such features may help provide insight into the shared biological etiology and disease development across disease subtypes . The NHGRI-EBI GWAS catalog and latest GWAS PRS construction methods are based on published GWAS studies , which only report risk variants that reached genome-wide significance ( usually defined by a P-value threshold of P < 5x10-8 ) . However , it is likely that there are additional risk variants below this threshold that could be associated with the trait but have not reached statistical significance [16] . Incorporating non-significant variants may conceivably improve the predictive power of a PRS but may also add additional random false positive signals , which in turn could dilute the discriminatory power of the true risk variants and diminish any predictive gain [6 , 17] . To explore whether a PRS constructed using additional non-significant loci may perform differently than a PRS using only loci reaching genome-wide significance , we evaluated PRS constructed using publicly available genome-wide summary statistics from the UK Biobank at six different p-value thresholds both in terms of associations with skin cancer phenotypes and in terms of secondary phenotype associations . We further applied LDpred , a tool that adjusts GWAS summary statistics for the effects of linkage disequilibrium [7] , to explore the performance of PRS incorporating the entire spectrum of available genetic information across the genome . There is extensive literature on constructing genome-wide PRS using random effects , shrinkage methods , or thresholding ( our focus ) [7 , 18 , 19] , but none of these methods have been evaluated in a PheWAS setting . In this paper , we focus our attention on skin cancer , but the approaches used in this paper can be applied to other diseases with well-defined molecular subtypes ( for example , breast cancer with ER and HER2-defined subtypes ) . We chose to use skin cancer as a demonstrative example for a variety of reasons . First , our discovery dataset ( MGI ) is particularly enriched for skin cancer cases due to the strong skin cancer clinical program at Michigan Medicine and due to the high rate of surgery for skin cancer patients . MGI primarily recruits participants undergoing surgery and is therefore enriched for cancers and other medical comorbidities when compared to a general population [4] . Additionally , skin cancer has well-defined subtypes , which allows us to explore performance of subtype-specific PRS . Skin cancer also provides a setting in which there may be genetic factors uniquely related to particular subtypes as well as genetic factors that are shared risk factors for all skin cancer subtypes . The various PRS–phenotype associations identified in this paper demonstrate ways to explore shared and subtype-specific phenotypes , and this joint framework may provide an enhanced understanding of the genome x phenome landscape . We introduce an online visual web catalog called PRSweb that provides PRS-PheWAS results for melanoma , basal cell carcinoma , and squamous cell carcinoma . PheWAS results are available using four different PRS construction methods explored in this paper: latest GWAS , NHGRI-EBI GWAS catalog , the UK Biobank GWAS summary statistics using different significance thresholds , and LDpred . The weights and the marker list for each PRS method can be downloaded . Furthermore , PheWAS summary statistics can be accessed from PRSweb ( see Web resources ) , providing future investigators with readily available and useful tools to perform further analyses . Comprehensive phenome-wide and genome-wide analyses of large biobank studies with publicly available summary statistics can be rich resources for PRS construction , especially if the trait-of-interest’s prevalence is high in the biobank . Using PRS , we can synthesize complex genetic information that can then be used to identify these shared genetic components across phenotypes . Compared to prior and existing literature , our contribution is new in four principal directions: ( 1 ) comparing various PRS construction methods in terms of their relationships with related EHR-derived phenotypes and subtypes ( 2 ) comparing PRS associations with secondary phenotypes across the phenome of MGI ( academic medical center ) and UK Biobank ( population-based ) , ( 3 ) developing PRS-based methods for understanding the shared and unique genetic contribution across disease subtypes both in terms of disease biology and in terms of secondary phenotype associations , and ( 4 ) introducing a publicly accessible online visual catalog PRSweb to visually represent the PRS x phenome landscape and access summary data from PheWAS .
For each skin cancer subtype ( melanoma , basal cell carcinoma , and squamous cell carcinoma ) , we generated four different sets of PRS: ( 1 ) based on merged summary statistics published in the NHGRI EBI GWAS catalog [1] , ( 2 ) based on the latest available GWAS meta-analysis [30–32] , ( 3 ) based on linkage disequilibrium ( LD ) clumping and p-value thresholding on publicly available GWAS summary statistics from the UK Biobank data [9] , and ( 4 ) based on reweighting effect sizes of GWAS summary statistics by modeling LD with LDpred [7] . For each of the obtained SNP sets for each trait , we constructed a PRS as the sum of the allele dosages of risk increasing alleles of the SNPs weighted by their reported or reweighted log odds ratios . Restated , the PRS for subject j in MGI was of the form PRSj = ΣiβiGij where i indexes the included loci for that trait , βi is the log odds ratios retrieved from the external GWAS summary statistics for locus i , and Gij is a continuous version of the measured dosage data for the risk allele on locus i in subject j . The PRS variable was created for each MGI and UKB participant . For comparability of effect sizes corresponding to the continuous PRS across cancer traits and PRS construction methods , we transformed each PRS of the corresponding analytical data set to the standard Normal distribution using “ztransform” in the R package “GenABEL” [35] . In this study , we first constructed PRS for skin cancer subtypes using either the latest GWAS or the corresponding entries of the GWAS catalog . To compare the association between PRS and skin cancer phenotypes across different PRS construction methods , we fit the following model for each PRS and skin cancer phenotype: logit ( P ( Phenotypeispresent|PRS , Age , Sex , Array , PC ) ) =β0+βPRSPRS+βAgeAge+βSexSex+βArrayArray+βPC , where the PCs were the first four principal components obtained from the principal component analysis of the genotyped GWAS markers and where “Array” represents the genotyping array . Our primary interest is βPRS , while the other factors ( Age , Sex and PC ) were included to address potential residual confounding and do not provide interpretable estimates due to the preceding application of case-control matching . Firth’s bias reduction method was used to resolve the problem of separation in logistic regression ( Logistf in R package “EHR” ) [36–38] , a common problem for binary or categorical outcome models when a certain part of the covariate space has only one observed value of the outcome , which often leads to very large parameter estimates and standard errors . We then evaluated each PRS’s ( 1 ) ability to discriminate between cases and controls by determining the area under the receiver-operator characteristics ( ROC ) curve ( AUC ) using R package “pROC” [39]; ( 2 ) calibration using Hosmer-Lemeshow Goodness of Fit test in the R package “ResourceSelection” [40 , 41]; and ( 3 ) accuracy with the Brier Score in the R package “DescTools” [42] . These evaluations did not adjust for additional covariates . These metrics were estimated using roughly 2/3 of the matched data as a test set after fitting the above model on the remaining 1/3 of matched data , which we will refer to as the training data . We used these metrics and the logistic regression results to choose a PRS construction method to use for each skin cancer subtype moving forward . We compare these measures for various PRS-phenotype relationships for each phenotype separately , so the comparative performance of these measures is not biased by the different case-control sampling . To explore the impact of incorporating non-significant loci into the PRS construction , we further performed the above analyses with PRS constructed using UK Biobank GWAS summary statistics with different p-value thresholds . Similarly , we compared the LDpred-based PRS that assumed six different fractions of causal variants ( non-zero effects ) in the prior: 100% , 10% , 1% , 0 . 1% , 0 . 01% , and 0 . 001% . For LDpred comparisons we also report Nagelkerke’s pseudo-R2 to be consistent with the LDpred workflow [7] . Using the chosen PRS for each subtype , we conducted two PheWAS to identify other phenotypes associated with the PRS first for the 1 , 578 phenotypes in MGI and then for the 1 , 366 phenotypes from UK Biobank . To evaluate PRS-phenotype associations , we conducted Firth bias-corrected logistic regression by fitting a model of the above form for each phenotype and data source . Age represents the birth year in UK Biobank . To adjust for multiple testing , we applied the conservative phenome-wide Bonferroni correction according to the analyzed PheWAS codes ( nMGI = 1 , 578 or nUK Biobank = 1 , 366 ) . In Manhattan plots , we present–log10 ( p-value ) corresponding to tests of H0: βPRS = 0 . Directional triangles on the PheWAS plot indicate whether a phenome-wide significant trait was positively ( pointing up ) or negatively ( pointing down ) associated with the PRS . To investigate the possibility of the secondary trait associations with PRS being completely driven by the primary trait association , we performed a second set of PheWAS after excluding individuals affected with the primary or related cancer traits for which the PRS was constructed , referred to as “exclusion PRS PheWAS” as described previously [4] . We then constructed new PRS scores representing shared and subsite-unique genetic components and performed a PheWAS for each . To evaluate the impact of the matching in the PRS PheWAS and exclusion PRS PheWAS analyses in more concrete terms , we performed sensitivity analyses in which we conducted the PheWAS analyses using the unmatched data . To evaluate how well prior presence of a secondary diagnosis can identify subjects with increased risk of developing skin cancer , we created a binary variable taking the value 1 if a given subject ( 1 ) was diagnosed with the secondary diagnosis and then diagnosed with skin cancer at least 365 days after or ( 2 ) was diagnosed with the secondary diagnosis and never diagnosed with skin cancer . We then fit a Firth bias-corrected logistic regression of the following form: logit ( P ( Primaryphenotypeispresent|Predictor , Age , Sex , Array , PC ) ) =β0+βPRSI ( Secondarynonskincancertrait ) +βAgeAge+βSexSex+βArrayArray+βPC where Array and PC were defined as before . Unless otherwise stated , analyses were performed using R 3 . 4 . 4 [43] . The online open access visual catalog PRSweb available at http://statgen . github . io/PRSweb was implemented using “Pandas” , a Data Analysis Library , which offers high level performance for large data structures and data analysis in the Python3 environment [44] . In combination with “Jinja2” , a templating language for Python , and “Bootstrap” , a Cascading Style Sheets ( CSS ) framework ( see Web resources ) , static HTML files were compiled and allow easy and fast hosting of all PRS-PheWAS results . The interactive plots are drawn with the JavaScript library “LocusZoom . js” ( see Web resources ) which offers dynamic plotting , automatic plot sizing and label positioning . Data were collected according to Declaration of Helsinki principles . MGI study participants’ consent forms and protocols were reviewed and approved by the University of Michigan Medical School Institutional Review Board ( IRB ID HUM00099605 and HUM00155849 ) . Opt-in written informed consent was obtained .
We first explored the comparative performance of various PRS construction strategies in terms of the resulting PRS associations with related phenotypes in the skin cancer setting . Table 2 provides the results . Using each of the chosen PRS described above ( mPRS , bPRS , and sPRS ) , we tested the association between each PRS and each of the 1 , 578 constructed phenotypes in MGI . For each PRS , the strongest associations were observed with dermatologic neoplasms that included overall skin cancer , melanoma , “other non-epithelial cancer of skin” ( the PheWAS parent category of basal and squamous cell carcinoma ) , and carcinoma in situ of skin . In addition , secondary dermatologic traits such as actinic keratosis ( with parent category “degenerative skin conditions and other dermatoses” ) , chronic dermatitis due to solar radiation ( with parent category “dermatitis due to solar radiation” ) , and seborrheic keratosis were found to be associated with all three PRS ( Fig 1 and Table K in S1 File ) . “Diseases of sebaceous glands” , “sebaceous cyst” , and “scar conditions and fibrosis of skin” were associated with bPRS . mPRS was most strongly associated with the melanoma phenotype ( OR 1 . 48 , 95% CI [1 . 41 , 1 . 56] ) , while bPRS was most strongly associated with basal cell carcinoma ( OR 1 . 65 , 95% CI [1 . 56 , 1 . 75] ) followed closely by “other non-epithelial cancer of the skin” ( OR 1 . 39 , 95% CI [1 . 34 , 1 . 44] ) . sPRS was most strongly associated with overall skin cancer ( OR 1 . 34 , 95% CI [1 . 3 , 1 . 38] ) . The OR of all these phenotypes indicated an increased risk for primary and secondary traits with increasing PRS . To substantiate the detected dermatologic associations , we reiterated the association screen of the three PRS in the matched phenome of the population-based UK Biobank data set ( Fig 1 ) . In general , stronger evidence for association was found in UKB compared to MGI . This may be driven by the larger sample sizes , e . g . a total of 13 , 623 skin cancer cases versus 4 , 503 in MGI . In the UK Biobank phenome , the large majority of the previous associations with dermatologic neoplasms were validated with the exception of the trait “dermatitis due to solar radiation” , which had substantially fewer cases in UKB compared to MGI ( 390 versus 2 , 959 cases ) . Unlike MGI , all three PRS were significantly associated ( at the phenome-wide level ) with “cancer , suspected or other” and “malignant neoplasm , other . ” bPRS and sPRS were both associated with “diseases of the sebaceous glands” and “sebaceous cyst . ” In order to explore whether the identified PRS-phenotype associations were driven by the primary trait used to define the PRS ( for example , as a side effect of treatment given after diagnosis with the primary trait ) , we performed a PheWAS for each PRS in which we excluded subjects who were cases for the primary trait or other skin cancer subtypes [4] . Results are shown in Table K in S1 File and Fig D in S1 Text . Actinic keratosis , a skin condition believed to be a precursor to non-melanoma skin cancers , remained significantly associated with the squamous cell carcinoma PRS in MGI and all three PRS in UK Biobank [46–48] . No other phenotypes were significant for MGI . “Sebaceous cyst” and its parent category “diseases of the sebaceous gland” were significant in the main UK Biobank PheWAS and remained significantly associated with basal cell carcinoma PRS and squamous cell carcinoma PRS in UK Biobank in the Exclusion PheWAS . Appendix 1 in S1 Text provides additional information on a sub-analysis of actinic keratosis as a predictor for future skin cancer . In the PRS-PheWAS analyses , we note a striking overlap in the secondary dermatological traits significantly associated with each of the three PRS ( mPRS , bPRS , sPRS ) . One potential explanation for this is that subjects may have more screening after an initial skin cancer diagnosis . Indeed , many subjects have multiple skin cancer diagnoses ( Fig F in S1 Text ) . Fig 2 shows the number of risk loci shared by different PRS . Six risk loci are shared between the mPRS , bPRS , and sPRS . This observation inspired a follow-up exploration in which we defined a PRS for each cancer subtype using the loci unique to that subtype’s chosen PRS . We call these new PRS scores mPRS-u , bPRS-u , and sPRS-u , which reflect the unique loci in the PRS for melanoma , basal cell carcinoma , and squamous cell carcinoma respectively . We also define a PRS consisting of all loci shared across the three skin cancer subtypes , which we call the shared PRS . Table C in S1 Text shows the association between the various constructed PRS and the skin cancer phenotypes . As with mPRS , mPRS-u was most strongly associated with the melanoma phenotype . The bPRS-u score was similarly most strongly associated with basal cell carcinoma . We note that the melanoma AUC for the mPRS score was 0 . 61 ( 95% CI , [0 . 59 , 0 . 62] ) and is only 0 . 54 ( 95% CI , [0 . 52 , 0 . 56] ) for the mPRS-u score . Similarly , the basal cell carcinoma AUC for the bPRS score was 0 . 64 ( 95% CI , [0 . 62 , 0 . 66] ) and is only 0 . 57 ( 95% CI , [0 . 55 , 0 . 59] ) for the bPRS-u score . The sPRS-u score is not more strongly associated with the squamous cell carcinoma phenotype than the other skin cancer subtypes . For this reason , we do not include this PRS in further analyses . The shared PRS constructed as the unweighted sum of risk alleles of loci present in all three PRS scores ( mPRS , bPRS , and sPRS ) is more strongly associated with all three subtype phenotypes than the overall skin cancer phenotype . Fig H in S1 Text shows PRS-PheWAS results using mPRS-u and bPRS-u . The scores again reveal their subtype specificity , while no notable secondary associations were observed . Although not shown here , additional exploration into the loci identified uniquely for each subtype , e . g . via pathway enrichment analyses , may provide some insight into subtype-specific biological mechanisms . Fig I in S1 Text shows PRS-PheWAS results for the shared PRS . Most strikingly , the shared skin cancer PRS was associated with the top skin cancer and dermatologic traits that were previously found to be associated with the three partially overlapping PRS constructs , suggesting that a shared genetic risk may be driving many of these secondary associations . These six underlying loci ( HERC2 [MIM 605837] /OCA2 [MIM 611409] , IRF4 [MIM 601900] , MC1R [MIM 155555] , RALY [MIM 614663] , SLC45A2 [MIM 606202] and TYR [MIM 606933] ) were previously found to be associated not only with skin cancer traits , but also with pigmentation traits of skin , eyes and hair ( Fig 2; MIM 266300 ) [31 , 49–68] . One of these pigmentation traits , skin tanning ability , the tendency of skin to sunburn rather than to suntan , is a well-known risk factor for all skin cancer traits [68 , 69] . A PRS based on the independent risk variants of a recent GWAS meta-analysis on skin tanning ability [69] was strongly associated with overall skin cancer , melanoma , basal cell carcinoma , and squamous cell carcinoma and even outperformed the constructed PRS in some cases ( Table C in S1 Text ) . Furthermore , the skin tanning ability PRS PheWAS identified a very similar set of traits as the shared skin cancer PRS , but in general displayed stronger associations ( Fig I in S1 Text ) . To explore whether a PRS incorporating non-significant loci will outperform a PRS incorporating only significant loci , we constructed PRS using loci related to the phenotype at six different p-value thresholds based on publicly available GWAS summary statistics from the UK Biobank . Larger p-values indicate greater SNP depth ( with more SNPs being incorporated into the PRS ) . The ICD-code-based collection of UK Biobank GWAS results did not include basal cell carcinoma or squamous cell carcinoma subtypes; rather , it included only the merged trait ‘non-epithelial cancer of skin’ ( Fig B in S1 Text ) . Thus , we limited our assessment of the summary statistics to the overall skin cancer GWAS and the melanoma GWAS ( Table J in S1 File ) . Table D in S1 Text provides the results . As with the other PRS construction methods , the melanoma PRS was most strongly associated with the melanoma phenotype for all p-value cutoffs except 5x10-4 . For this p-value cutoff , the melanoma PRS had similar AUC and OR for the melanoma and basal cell carcinoma phenotypes . This p-value cutoff represents the least conservative inclusion cutoff with 1 , 193 included loci , and its results indicated that inclusion of too many suggestive SNPs at lower thresholds may reduce PRS performance . However , we also note that the most conservative cutoff ( 5x10-9 ) produced a PRS that was based on only six loci , which had a weaker OR and AUC compared to other PRS created with less stringent cutoffs . The best performance in terms of AUC and OR relating to the melanoma phenotype were observed for p-value thresholds 5x10-7 and 5x10-8 , which included 13 and 9 loci respectively . The small number of loci identified by this method at more conservative p-value cutoffs may be driven by the lower sample size for melanoma in the UK Biobank compared to the published melanoma GWAS meta-analyses ( n cases = 2 , 691 and n cases = 6 , 628 respectively ) . We note that the melanoma PRS constructed using the UK Biobank summary statistics produced lower AUC across all p-value thresholds than was seen for the latest GWAS and GWAS catalog PRS construction methods . Among the skin cancer subtypes , the PRS constructed for overall skin cancer was most strongly associated with basal cell carcinoma across all p-value thresholds , with odds ratios ranging from 1 . 4 ( 95% CI [1 . 32 , 1 . 48] ) to 1 . 64 ( 95% CI [1 . 55 , 1 . 74] ) . Among the PRS , the overall skin cancer PRS had the greatest discrimination for the overall skin cancer phenotype . Overall skin cancer and melanoma PRS had similar performance in terms of discrimination for the melanoma phenotype across various depths . The overall skin cancer PRS tended to be more strongly associated with and have similar or slightly better discrimination for the overall skin cancer phenotype compared to the melanoma PRS , indicating that the overall skin cancer PRS was more accurate at predicting the overall skin cancer phenotype than the melanoma PRS . The overall skin cancer PRS had very similar association with and discrimination abilities for the overall skin cancer phenotype across all p-value thresholds except the least conservative ( p = 5x10-4 ) , for which the AUC and odds ratio were smaller . Overall , the highest AUCs and strongest OR signals for both PRS and all skin cancer phenotypes were found at depths of 5x10-7 and 5x10-8 . In addition to associations with the primary and related phenotypes , we explored associations between PRS constructed at various UK Biobank summary statistic depths and secondary phenotypes . Fig J ( overall skin cancer ) and Fig K ( melanoma ) in S1 Text show PRS-PheWAS results in MGI using PRS constructed at different depths . Depths of 5x10-7 and 5x10-8 produced very similar results , and other depths tended to identify fewer phenotypes associated with the corresponding PRS . Phenotypes that were associated with the PRS at other depths had weaker associations . PRS-PheWAS using the melanoma PRS and the overall skin cancer PRS produced somewhat different results . For example , the melanoma PRS at different depths did not identify strong associations with “diseases of sebaceous glands” , which is similar to previous PRS-PheWAS results for mPRS in MGI and UKB . In contrast , the overall skin cancer PRS did identify associations with “diseases of sebaceous glands” or its subcategories for all depths except 5x10-5 and 5x10-4 . Fig L in S1 Text provides some additional information about the impact of depth on p-values for selected secondary associations . We evaluated the performance of PRS constructed using the LDpred algorithm , which incorporates millions of SNPs into the PRS definition . Table E in S1 Text provides results . For the overall skin cancer PRS , a modelled 1% proportion of causal variants produced the best results in terms of AUC and OR with respect to the overall skin cancer phenotype ( OR 1 . 30 , 95% CI [1 . 26 , 1 . 35] ) and AUC 0 . 58 , 95% CI [0 . 56 , 0 . 60] ) . For the melanoma PRS , a modelled 0 . 001% proportion of causal variants produced the best results with respect to melanoma ( OR 1 . 42 , 95% CI [1 . 36 , 1 . 49] ) and AUC 0 . 60 , 95% CI [0 . 58 , 0 . 62] ) . This LDpred melanoma PRS performed slightly better compared to the melanoma PRS constructed using UK Biobank summary statistics at a 5x10-8 depth in terms of associations with the primary phenotype . Using the PRS with a percentage of assumed causal variants producing the best pseudo R2 statistic from Table E in S1 Text , we performed a PRS-PheWAS as shown in Figure M in S1 Text . Table 3 summarizes the secondary phenotypes significantly associated with various PRS at the phenome-wide level . Many general skin cancer phenotypes are strongly associated with nearly all PRS . In particular , actinic keratosis and dermatitis due to solar radiation are associated with PRS for all three disease . In contrast , sebaceous cysts and “diseases of sebaceous glands” are strongly associated with PRS for basal cell carcinoma and squamous cell carcinoma but not with PRS for melanoma . For comparison of the aforementioned PRS-PheWAS results and to provide researchers with resources for future PRS-based analyses , we developed an open access , online visual catalog PRSweb available at https://statgen . github . io/PRSweb that enables interactive exploration of the PheWAS results for each of the skin cancer subtypes and different PRS construction methods explored in this paper , for both the MGI and UK Biobank phenomes . PRSweb shows PRS-PheWAS plots with various choices of PRS in the drop-down menu ( example screenshot in Fig 3 ) and offers downloadable PRS constructs ( list of independent risk variants with corresponding weights ) . Mouse-over boxes offer detailed information about top results if needed , without impeding the overall user experience ( grey box in Fig 3 ) . Enrichment of cases in the upper quartiles of the PRS distribution are presented in forest plots .
PRSweb; https://statgen . github . io/PRSweb University of Michigan Medical School Central Biorepository; https://research . medicine . umich . edu/our-units/central-biorepository UK Biobank; http://www . ukbiobank . ac . uk/ UK Biobank GWAS summary statistics; https://tinyurl . com/UKB-SAIGE TOPMed variant browser , https://bravo . sph . umich . edu/freeze5/hg38/ TOPMed program , https://www . nhlbi . nih . gov/science/trans-omics-precision-medicine-topmed-program Minimac4; https://genome . sph . umich . edu/wiki/Minimac4 BCFtools; https://samtools . github . io/bcftools/bcftools . html KING; http://people . virginia . edu/~wc9c/KING/ FASTINDEP; https://github . com/endrebak/fastindep PLINK; https://www . cog-genomics . org/plink2/ Eagle; https://data . broadinstitute . org/alkesgroup/Eagle/ UCSC Genome Browser; http://genome . ucsc . edu/ R; https://cran . r-project . org/ NHGRI-EBI GWAS Catalog; https://www . ebi . ac . uk/gwas/ dbSNP; https://www . ncbi . nlm . nih . gov/projects/SNP/ Imputation server; https://imputationserver . sph . umich . edu/ Jinja , https://github . com/pallets/jinja Locuszoom , https://github . com/statgen/locuszoom . | In the study of genetically complex diseases , polygenic risk scores ( PRS ) synthesize information from multiple genetic risk factors to provide insight into a patient’s inherited risk of developing a disease based on his/her genetic profile . These risk scores can be explored in conjunction with health and disease information available in electronic medical records . PRS may be associated with diseases that may be related to or precursors of the underlying disease of interest . In this paper , we demonstrate how PRS can be used in concert with the medical phenome to better understand the etiology of disease subtypes nested within a broad disease classification . This is done by examining the shared and distinct genetic risk factors across the related but heterogeneous disease subtypes and also through our comparison of the secondary associations across the phenome corresponding to the subtype specific PRS . We consider several PRS construction methods in our study . This framework of analysis is enabled by access to electronic health records and genetics data . Leveraging and harnessing the rich data resources of the Michigan Genomics Initiative , a biorepository effort at Michigan Medicine , and the large population-based UK Biobank study , we investigated the primary and secondary disease associations with PRS constructed for the three most common types of skin cancer: melanoma , basal cell carcinoma and cutaneous squamous cell carcinoma . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"genome-wide",
"association",
"studies",
"dermatology",
"medicine",
"and",
"health",
"sciences",
"carcinomas",
"cancers",
"and",
"neoplasms",
"oncology",
"skin",
"neoplasms",
"genome",
"analysis",
"malignant",
"skin",
"neoplasms",
"skin",
"diseases",
"basal",
"cell",
... | 2019 | Exploring various polygenic risk scores for skin cancer in the phenomes of the Michigan genomics initiative and the UK Biobank with a visual catalog: PRSWeb |
Curly , described almost a century ago , is one of the most frequently used markers in Drosophila genetics . Despite this the molecular identity of Curly has remained obscure . Here we show that Curly mutations arise in the gene dual oxidase ( duox ) , which encodes a reactive oxygen species ( ROS ) generating NADPH oxidase . Using Curly mutations and RNA interference ( RNAi ) , we demonstrate that Duox autonomously stabilizes the wing on the last day of pupal development . Through genetic suppression studies , we identify a novel heme peroxidase , Curly Su ( Cysu ) that acts with Duox to form the wing . Ultrastructural analysis suggests that Duox and Cysu are required in the wing to bond and adhere the dorsal and ventral cuticle surfaces during its maturation . In Drosophila , Duox is best known for its role in the killing of pathogens by generating bactericidal ROS . Our work adds to a growing number of studies suggesting that Duox’s primary function is more structural , helping to form extracellular and cuticle structures in conjunction with peroxidases .
Over 90 years ago , Lenore Ward first described a dominant mutation , Curly , that causes the wings of Drosophila melanogaster to bend upwards [1] . Since then , Curly has become a ubiquitous second chromosomal marker used by Drosophila geneticists on a daily basis to follow and track mutations . Despite its widespread use , how Curly mutations dominantly alter wing curvature has remained obscure . Waddington first proposed that Curly causes an unequal contraction of the dorsal and ventral wing surfaces during the drying period shortly after flies emerge from their pupal cases [2 , 3] . Others have subsequently demonstrated that comparable alterations in wing curvature can be caused by differential growth of the dorsal and ventral epithelia [4] . Irrespective of the mechanism , that similar wing phenotypes have been described for D . pseudoobscura and D . montium mutants [5] suggests the underlying cause of curly wing formation is evolutionarily conserved among Drosophilids [2] . The major factor limiting our understanding of Curly’s function in wing morphogenesis , however , is the fact that its molecular identity has remained unknown . In this manuscript we uncover the long unknown molecular nature of Curly . We show that mutations in the gene duox cause the Curly wing phenotype . Duox is a member of a highly conserved group of transmembrane proteins collectively referred to as NADPH oxidases . These enzymes function to transfer electrons across biological membranes to generate ROS by transferring electrons from NADPH to oxygen through flavin adenine dinucleotide ( FAD ) and heme cofactors [6] . Several biological functions have been described for Duox . Perhaps the best studied of these in Drosophila is its role in host defense where it is thought to generate ROS to kill pathogens [7] . However , Duox also plays an important role in providing ROS , specifically hydrogen peroxide , for heme peroxidases to catalyze the formation of covalent bonds between biomolecules . In mammals , Duox generates hydrogen peroxide for thyroid peroxidase to catalyze the iodination and crosslinking of tyrosine residues in the formation of thyroid hormones [8 , 9] . Duox is also expressed in tissues other than the thyroid , such as the gastrointestinal tract , where its function is less clear [6] . In insects , worms and sea urchins , Duox participates in the formation of extracellular structures through the crosslinking of tyrosine residues [10–12] . Indeed , instead of its function in generating bactericidal ROS , the tyrosine crosslinking activity of Duox may be the primary ancestral function , as it appears to be conserved across phyla . Here we show that specific mutations in the NADPH binding-domain encoding region of duox cause a Curly wing phenotype . Using Curly , we demonstrate that duox is required during the last day of pupal development to stabilize the wing . Furthermore , through suppression experiments , we identify a novel heme peroxidase , Curly Su ( Cysu ) , that works with duox to adhere the dorsal surface of the wing to the ventral one . Uncovering the molecular identity of Curly not only provides an entry point for the functional understanding of this prominent wing mutant phenotype , but also will allow for the discovery of novel duox interacting genes and regulators through unbiased genetic screens . Only through these approaches can we hope to understand the precise molecular function of Duox in the myriad biological processes in which it is involved .
In the course of genetically following a loss-of-function mutation in the gene duox , duoxKG07745 , using the standard Drosophila genetic tool Curly of Oster ( CyO ) balancer , we noticed that we were unable to recover progeny containing both duoxKG07745 and CyO . Since a ) duox and Curly mutations fail to complement one another and b ) Curly roughly maps to 23A4-23B2 [13] , the chromosomal region containing duox , we wondered whether Curly might be an allele of duox . Balancer chromosomes contain numerous inversions and chromosomal aberrations . To exclude the possibility that our inability to recover progeny was due to some other lesion in the 23A4-23B2 region of CyO , we crossed duox KG07745 to various Curly mutations not associated with CyO [1 , 13 , 14] . Consistent with our earlier results , all the Curly alleles tested failed to complement duoxKG07745 , suggesting that Curly mutations indeed reside within the duox locus ( Fig 1A ) . To provide conclusive evidence , however , that duox and Curly are one and the same , we expressed duox ubiquitously in a Curly mutant background . Ubiquitous expression of duox restored viability , allowing recovery of homozygous Curly mutants ( Fig 1A ) . These results strongly suggest that the Curly phenotype is due to mutations in the duox gene . To precisely determine how Curly mutations alter duox’s nucleotide sequence , we sequenced duox in two Curly mutants: CyK , which was previously generated using ethyl methanesulfonate [14]; and Cy1 , the original spontaneous mutation identified by Ward [1] . Remarkably , we found that the same nucleotide was mutated in both CyK and Cy1 ( Fig 1B ) . This single nucleotide mutation resulted in the conversion of a conserved glycine ( number 1505 ) in the NADPH binding domain of Duox to a serine in CyK and to a cysteine in Cy1 ( Fig 1C , red ) . Among the residues within the NADPH binding pocket of Duox , glycine 1505 is extraordinarily well conserved . It is present in all known NADPH oxidases from yeast to humans ( Fig 1D ) , suggesting that it is functionally important . Therefore a conversion of a conserved glycine to a polar amino acid specifically in the NADPH binding pocket of Duox causes Curly . Taken together , our results demonstrate , after more than 90 years since its discovery , that the Curly phenotype is due to mutations within the duox gene . Mutations in the NADPH binding encoding region of duox cause a Curly phenotype , but how exactly do they influence Duox’s function ? From complementation experiments , it is clear that Curly mutations reduce Duox’s normal function as they are not viable in combination with either a loss-of-function duox mutation or a deficiency uncovering the duox locus ( Fig 1A ) . However , Curly mutants also act dominantly , causing the wings of Curly flies to bend upwards ( hence the name ) in contrast to the straight wings of wild-type or duox heterozygous flies . One possible explanation is that Curly mutations act in opposition to Duox’s normal activity as dominant negatives or antimorphs . This , however , is unlikely because ubiquitous overexpression of wild-type Duox in a Curly mutant background failed to restore normal wing shape ( S1 Fig ) . Another possibility is that Curly mutations increase the normal function of Duox either by increasing expression or constitutively activating Duox . However , this too seems implausible because removing wild-type Duox in Curly mutants worsened ( causing lethality ) rather than ameliorated the viability phenotype ( Fig 1A ) . Instead , the most likely explanation is that Curly mutations are neomorphic , causing a dominant gain-of-function in Duox that is different from its normal function . To further investigate the potential neomorphic nature of Curly mutations , we tested whether Curly mutations require NADPH , the substrate of Duox , to generate a wing phenotype . In the first instance , we attempted to alter the endogenous levels of NADPH by reducing the amount of niacinamide , a precursor of NADPH , in the diet of Curly flies . Interestingly , we found that a reduction of dietary niacinamide caused a dose-dependent decrease in the expressivity of the wing phenotype in Curly mutants ( Fig 2A ) . To test this genetically , we next knocked down CG6145 , which encodes a NAD+ kinase that phosphorylates NAD+ to generate NADP+ [15] . Specific knockdown of CG6145 in the wing using apterous-gal4 ( apGal4 ) strongly suppressed the Curly wing phenotype ( Fig 2B ) . Together these results suggest that Curly mutants require NADP+ and/or NADPH in order to cause changes in wing curvature . Therefore , not only do Curly mutations reduce Duox’s normal function , they also endow it with a new function that requires sufficient substrate to dominantly alter wing shape . Having determined that Curly is most likely a neomophoric allele of duox , we decided to use Curly mutations to explore duox’s function in vivo in the wing . To do this , we generated a mutant form of duox , henceforth referred to as duoxCyK , in which glycine 1505 was mutated to a serine , as in the CyK mutant . In order to express duoxCyK conditionally we fused it to an Upstream Activating Sequence ( UAS ) element so that we could control its expression in a time-dependent and tissue-specific manner using the Gal4 system [16] . To demonstrate that this transgene was functional and able to recreate the Curly phenotype , we expressed it ubiquitously throughout the fly using tubpGal4 [17] . Indeed , ubiquitous expression of duoxCyK , but not wild-type duox , resulted in upward bent wings resembling those of CyK mutants ( Fig 2C and 2D ) . This not only demonstrates that the duoxCyK transgene is functional , but also provides further evidence that Curly is an allele of duox . Duox could be required autonomously within the wing for its formation , or instead act non-autonomously in other tissues , as is the case for curled , a mutation that causes similar changes in wing morphology [2] . To determine where duox is required , we first expressed duoxCyK or duox RNAi in the wing using apGal4 . Expression of duoxCyK , but not wild-type duox , caused the upward wing curvature , whereas knockdown of duox in the wing caused a slight downward curvature or cupping of the wing ( Fig 3A ) . This demonstrates that duox is required autonomously in the wing for its formation . Changes in wing curvature can be caused by differential growth between the dorsal and ventral wing surfaces , or could be caused by changes in cuticle structure [3 , 4] . If duoxCyK were differentially influencing growth , we would expect it to be required early in pupal development , but not later , when proliferation and growth of the ventral and dorsal wing surfaces is largely complete [18] . To determine when in development duoxCyK is acting , we conditionally expressed it at various developmental stages using a heat shock-inducible driver . Expression of duoxCyK , but not duox , on the last day of pupal development , but not before then , resulted in upturned wings ( Fig 3B ) . This suggests that Duox does not influence wing growth and instead perhaps plays a role in the formation of the wing cuticle . Taken together , our results demonstrate that duox acts autonomously in the wing to stabilize the cuticle during the last day of pupal development concurrent with the formation of the wing cuticle . Hydrogen peroxide generated by NADPH oxidases is often used by peroxidases , most frequently heme peroxidases , to crosslink proteins and kill pathogens [6] . Heme peroxidases are essential for NADPH oxidase-dependent crosslinking reactions , but largely dispensable to their other functions [6] . To test whether Duox acts with a heme peroxidase in the wing to crosslink proteins and stabilize it , we individually knocked down all known D . melanogaster heme peroxidases ( Fig 4A ) [19] using RNAi in the wings of duoxCyK flies . If duoxCyK acts alone in the wing and does not participate in crosslinking reactions , then knockdown of heme peroxidases should not affect wing curvature . If , however , duoxCyK requires a heme peroxidase to function , then knockdown should suppress the Curly phenotype . Consistent with this , knockdown of one peroxidase , CG5873 , fully suppressed the Curly phenotype in a manner resembling duox knockdowns , suggesting that Duox functions together with CG5873 to crosslink molecules to form the wing ( Fig 4A ) . As CG5873 is a suppressor of Curly we have named it Curly Su , or cysu for short . If cysu acts with duox to form the wing then one would expect it to: a ) be expressed in the wing on the last day of development–the period in which Duox functions to form the wing; and b ) have a similar phenotype to Duox when silenced in the wing . To test this first prediction , we generated a strain expressing endogenously mCherry-tagged Cysu . Consistent with Cysu functioning in the wing , we observed expression of mCherry-tagged Cysu in the wing on the last day of development using confocal microscopy ( Fig 4B ) . To test our second prediction , we knocked down Cysu in the wing using RNAi . Cysu knockdown caused a slight downturning and cupping of the wing resembling Duox knockdown wings , further suggesting that Cysu and Duox function together to shape the wing ( Fig 4C ) . Interestingly , defects in scutellum and notum formation were also observed in Duox and Cysu knockdowns , suggesting that they might play a broader role in cuticle formation ( Fig 4D ) . Consistent with this , Cysu was expressed in the thorax of wild-type flies , but not Cysu knockdowns ( Fig 4E ) . In summary , Duox and the heme peroxidase Cysu act together to stabilize the wing during development . The Drosophila adult wing is made up of two cuticle panels that are synthesized and secreted in a step-wise fashion by the dorsal and ventral ectodermal epithelial cells toward the end of pupal development [20 , 21] . Upon eclosion , the dorsal and ventral cuticular surfaces of each wing expand and within an hour or so become bonded and adherent to one another [18] . To determine how altering Duox influences wing cuticle formation , we imaged wild-type and duox mutant wings using transmission electron microscopy ( Fig 5 ) . In wild-type wings , the ventral and dorsal cuticles are closely apposed and tightly bonded ( Fig 5A ) . By contrast , in Duox knockdown wings the ventral and dorsal cuticles rarely directly contact one another , and instead are separated by a gap filled with disordered , electron poor material ( Fig 5B ) . Interestingly , Curly mutant wings on the other hand appeared much more similar to wild-type wings with occasional abnormal bunching of the dorsal cuticle ( Fig 5C ) . Whether this pinching and bunching reduces the surface area of the dorsal wing causing the wing to curve upward remains unknown . However , ultrastructure experiments clearly demonstrate a role for Duox in the adhesion and bonding of the two cuticle wing surfaces during wing formation .
Here we have shown that the Curly mutation arises in the NADPH-binding pocket encoding region of duox . Using Curly mutations and duox RNAi , we show that Duox is required within the wing to maintain its shape beginning on the last day of pupal development . Results from our genetic studies suggest Duox does this by supplying hydrogen peroxide to the heme peroxidase Cysu to facilitate the bonding of the two wing cuticle surfaces , likely by physically crosslinking them , during wing formation . In all Curly mutants sequenced , a glycine residue , 1505 , in the NADPH-binding pocket of Duox is mutated . This glycine is present in all NADPH oxidases from microbial eukaryotes to humans , and more broadly in oxidoreductase and ferric reductase NAD-binding domains ( PFAM PF00175 and PF08030 , respectively ) . Though mutagenesis studies have not been conducted on this residue itself , it sits beside an equally conserved cysteine residue , which has been studied in detail because mutations in it cause chronic granulomatous disease in humans [22] . This cysteine residue does not appear to be important for NADPH oxidase assembly or binding NADPH [22 , 23] . Instead it is thought to be required for orienting bound NADPH for efficient electron transfer ( via hydride ) to FAD , and eventually oxygen [22] . Given glycine 1505’s proximity , it is possible that mutations in it similarly affect the transfer of electrons from NADPH to FAD . Consistent with this is the observation that Curly mutants are neither homozygous viable nor viable over a deficiency , suggesting that mutation of glycine 1505 causes a reduction in Duox’s normal function . Although Curly mutations reduce Duox’s normal function , they also endow it with a new function . Precisely what this new function is remains obscure , however it likely requires a source of electrons because altering the NADPH/NADP+ by removing niacinamide from the food or knocking down NAD+ kinase suppressed the wing phenotype . It is known that the expressivity of the Curly wing phenotype can be suppressed by larval crowding and/or starving larvae [24] . Given this , it is possible that reduced uptake of niacinamide is a cause of the decreased expressivity of the Curly wing phenotype in starved larvae . Riboflavin shortage during the larval stage has also been suggested to be a cause of this suppression [2 , 25] . Since riboflavin is a precursor of FAD , a co-factor also necessary for Duox’s function , it too may suppress the wing phenotype by reducing endogenous FAD and in turn reducing Duox’s activity . Regardless , Curly mutations are likely neomorphic and their sensitivity to environmental factors is likely mediated by changes in substrate availability . Duox is required autonomously for wing stabilization . Results from this study and another strongly support this assertion [10] . Expression of duoxCyK or knockdown of duox on the last day prior to eclosion , but not earlier , caused defects in wing morphogenesis . This suggests that Duox and Curly do not influence growth or proliferation of the wing epithelia because these processes are complete by this time [18] . Instead , ultrastructural analysis suggests that Duox plays an important role in forming the cuticle of the wing . In duox knockdowns , frequent gaps between the two wing cuticle surfaces were observed , in contrast to the wild-type wings . Defects in adhesion of the two cuticle surfaces were also apparent in Curly mutants . Unlike wings from duox knockdowns , however , the cuticle surfaces in Curly wings were most often tightly apposed with occasional bunching of the dorsal surface . It is possible that in the Curly mutants this aberrant pinching of the dorsal surface decreases its area relative to the ventral surface causing the wing to bend , as first intimated by Waddington 75 years ago [3] . However , we do not know whether this is the cause of the curling or just a consequence of it . Duox is known to be involved in the formation of extracellular matrices and cuticles [10–12] . Typically , it does this by supplying hydrogen peroxide to heme peroxidases , which use the hydrogen peroxide to perform crosslinking reactions . Consistent with Duox playing a role in crosslinking the cuticle we found that the heme peroxidase Cysu was essential for Duox function in the wing . Duox is unusual among NADPH oxidases in that it contains its own peroxidase homology domain , which in Caenorhabditis elegans and D . melanogaster has been proposed to fulfill the function of heme peroxidases , thereby obviating their need [7 , 26 , 27] . However , given that the peroxidase homology domain of Drosophila Duox lacks many amino acid residues , including the proximal and distal histidines , essential for efficient peroxidase function it is unclear how well it functions in this capacity [27] . Indeed , our results suggest that in D . melanogaster , Duox requires the heme peroxidase Cysu not only for stabilizing the wing cuticle , but also in the formation of the notum and scutellum . These findings point to a more general role for Duox and Cysu in cuticle formation . In Drosophila , Duox has been intensely studied in the context of host defense and gut immunity . In the gut , Duox is thought to generate ROS to kill pathogens; flies that have reduced Duox activity have increased susceptibility to infection [7 , 26] . Upon infection ROS generated by Duox kill pathogens , and possibly signal intestinal epithelial cells to proliferate and renew [7 , 28] . Our results , as well as others , demonstrate that Duox is also critical in the formation of cuticle structures and extracellular matrices [10–12] . It is possible that Duox performs a similar function in the Drosophila intestine , perhaps by forming extracellular barriers or structures to protect against infection . Indeed , Duox in conjunction with heme peroxidases has been shown to form such barriers in guts of ticks and mosquitos [29 , 30] . It would therefore be interesting to explore whether Duox and possibly Cysu are also involved in forming barriers to protect against infection in the Drosophila intestine . Duox is an important protein that has a number of diverse functions , which we are only beginning to understand . Curly mutations provide an excellent opportunity to further explore Duox’s functions by identifying unknown interactors and regulators through unbiased genetic suppressor screens . The identification of Cysu through such an approach demonstrates its feasibility and utility . Such approaches will not only tell us about Duox’s function in the wing , but also about its role in immunity and beyond .
Flies were propagated in polystyrene vials ( 28 . 5 mm diameter ) containing cornmeal molasses yeast medium at 25°C . For most crosses 5 virgin females were mated with 3 to 5 males . Fly strains used are shown in Table 1 . duox and duoxCyK were conditionally expressed using a gal4 driver downstream of the heat-shock ( HS ) protein 70 promoter ( HS-gal4 ) . At various times during development Drosophila were incubated at 37°C for 2 hours to induce expression . To verify the location of P{SUPor-P}DuoxKG07745 we performed inverse PCR . Consistent with the FlyBase report ( FBal0226250 ) the 3’ flanking sequence was at genomic position 2L:2 , 826 , 884 . However , unexpectedly , we found the 5’ flanking sequence to be at position 2L:2 , 755 , 447 . This suggests that P{SUPor-P}DuoxKG07745 contains a deletion that perturbes 17 genes from Bacc to duox . The duox open reading frame was amplified from cDNA and cloned into the pVALIUM22 vector [31] between XbaI and EcoRI restriction sites using standard methods . The CyK mutation was generated by site-directed mutagenesis using a QuikChange site-directed mutagenesis kit . Constructs were integrated into the attP2 site on the third chromosome using phiC31 integrase by BestGene . mCherry-tagged Cysu expressing flies were made by BestGene by injecting mimic construct #1315 into Mi{MIC}CG5873MI11428 ( BDSC 56608 ) [32] . Genomic DNA was crudely isolated by homogenizing one to two flies in 0 . 2 mg/ml Proteinase K ( Roche MC00079 ) , 10 mM Tris pH 8 . 0 , 1 mM EDTA and 25 mM NaCl and incubating for 25 min at 55°C . Proteinase K was subsequently inactivated by boiling the samples for 5 min . duox was then PCR amplified from genomic DNA and sequenced by Genewiz . Flies were raised on a defined medium with various concentration of niacinamide from embryo to adult . A solution was prepared as described in Table 2 and the pH was adjusted to 7 . 0 with NaOH . 20 mg/ml agar yeast culture grade ( Sunrise Science Products 1910 ) was dissolved in the solution by heating before adding 0 . 4 mg/ml cholesterol ( Sigma C3045 ) and niacinamide ( Sigma N0636 ) . Fluorescent images were acquired with a 10X/NA objective on a Zeiss LSM 780 confocal microscope . All other images were obtained with a Zeiss SteREO Discovery . V8 microscope . For transmission electron microscopy , whole flies were immersed in 95% ethanol briefly to get rid of any air bubbles , decapitated and immersed into fixative containing 4% glutaraldehyde in 0 . 1M PIPES buffer , pH 7 . 2 at room temperature for 2 hours , and then overnight at 4°C . Flies were next embedded in 1% agar and post-fixed with 2% osmium tetroxide with 1 . 5% potassium ferricyanide in 0 . 1M PIPES buffer for 1 hour then en block stained with 1% uranyl acetate in ddH2O at 4°C overnight . Samples were dehydrated with ethanol at room temperature before incubation with propylene oxide and embedment in Spurr resin ( Electron Microscopy Sciences , Hatfield , PA ) . 500nm semi-thin sections were stained with 0 . 1% toluidine blue to evaluate the area of interest . 60nm ultrathin sections were cut , mounted on formvar coated slotted copper grids and stained with uranyl acetate and lead citrate by standard methods . Stained grids were examined under Philips CM-12 electron microscope ( FEI; Eindhoven , The Netherlands ) and photographed with a Gatan ( 4k x2 . 7k ) digital camera ( Gatan , Inc . , Pleasanton , CA ) . | Fruit fly geneticists rely on a handful of dominant mutations that modify adult morphology in a way that is easy to spot , like changing the shape of the fly’s wings , eyes or bristles . One of the first such mutants identified in the early days of fly genetics and to this day likely the most widely used mutation , is Curly , which causes an upward curvature in the adult wings . Despite its importance as a marker , the genetic cause of Curly has remained unknown . Here , we reveal that Curly mutations occur in the gene duox , which encodes a ROS-generating enzyme . ROS once thought to be merely harmful by-products of metabolism , can also have beneficial purposes . Here we provide evidence that Duox generates ROS to help form and stabilize the wings of fruit flies . Furthermore , we identify a second enzyme , Cysu , which uses the ROS generated by Duox to crosslink proteins in the wing , thereby stabilizing and shaping its structure . Duox occurs in numerous organisms , including humans and fulfills a number of other functions , in particular in immunity and pathogen defense . With this new knowledge , Curly mutations will provide an excellent tool to study and understand the roles Duox plays in a variety of biological contexts . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Curly Encodes Dual Oxidase, Which Acts with Heme Peroxidase Curly Su to Shape the Adult Drosophila Wing |
Large parts of South Sudan are thought to be trachoma-endemic but baseline data are limited . This study aimed to estimate prevalence for planning trachoma interventions in Unity State , to identify risk factors and to investigate the effect of different sampling approaches on study conclusions . The survey area was defined as one domain of eight counties in Unity State . Across the area , 40 clusters ( villages ) were randomly selected proportional to the county population size in a population-based prevalence survey . The simplified grading scheme was used to classify clinical signs of trachoma . The unadjusted prevalence of trachoma inflammation-follicular ( TF ) in children aged 1–9 years was 70 . 5% ( 95% CI: 68 . 6–72 . 3 ) . After adjusting for age , sex , county and clustering of cases at household and village level the prevalence was 71 . 0% ( 95% CI: 69 . 9–72 . 1 ) . The prevalence of trachomatous trichiasis ( TT ) in adults was 15 . 1% ( 95% CI: 13 . 4–17 . 0 ) and 13 . 5% ( 95% CI: 12 . 0–15 . 1 ) before and after adjustment , respectively . We estimate that 700 , 000 people ( the entire population of Unity State ) require antibiotic treatment and approximately 54 , 178 people require TT surgery . Risk factor analyses confirmed child-level associations with TF and highlighted that older adults living in poverty are at higher risk of TT . Conditional simulations , testing the alternatives of sampling 20 or 60 villages over the same area , indicated that sampling of only 20 villages would have provided an acceptable level of precision for state-level prevalence estimation to inform intervention decisions in this hyperendemic setting . Trachoma poses an enormous burden on the population of Unity State . Comprehensive control is urgently required to avoid preventable blindness and should be initiated across the state now . In other parts of South Sudan suspected to be highly trachoma endemic , counties should be combined into larger survey areas to generate the baseline data required to initiate interventions .
Ocular infection with the bacterium Chlamydia trachomatis causes trachoma , a disease characterized by inflammation of the conjunctiva , most commonly in children , that can lead to scarring , opacity of the cornea and blindness in later life [1] . Trachoma is responsible for an estimated 3 . 6% of all cases of blindness , making it the major cause of infectious , preventable blindness worldwide [2] . The disease is largely found in poor , rural communities in low-income countries , where access to water , sanitation and health care is inadequate [3] , [4] . Fifty-seven countries , mostly in Africa , are currently known to be trachoma endemic [5] . Globally an estimated 40 . 6 million people are living with active trachoma ( trachoma inflammation-follicular ( TF ) and/or trachoma inflammation-intense ( TI ) ) and 8 . 2 million are affected by trachomatous trichiasis ( TT ) [6] , [7] . These recently revised figures differ from 2003 estimates of 84 million cases of active trachoma and 7 . 6 million cases of TT [6] , with the discrepancy being at least partially explained by the availability of better data . The reduction in active disease , however , is also thought to reflect socio-economic development in some of the affected countries and the effective implementation of a comprehensive control strategy [1] , [5] , [7] . This so-called ‘SAFE’ strategy is being promoted by the Alliance for the Global Elimination of Blinding Trachoma by the year 2020 ( GET 2020 ) , established in 1997 by the World Health Organization ( WHO ) . SAFE stands for Surgical correction of trichiasis , Antibiotic treatment , Facial cleanliness and Environmental improvement [8] , [9] . The WHO recommended thresholds to initiate trachoma control through “A” or “S” are a prevalence of TF in children age 1–9 years of ≥10% , or TT prevalence of ≥1% in people age 15 years or above [1] , [10] , estimated at a population administrative level for health care delivery or corresponding to a population size of around 100 , 000 . Apart from providing more up-to-date estimates of the number of trachoma cases , recent reviews also show that a handful of countries carry most of the burden [1] , [5] , [7] . The newly independent Republic of South Sudan is one of the countries worst affected by both active trachoma and trichiasis , and has been highlighted as being in particularly urgent need of large-scale SAFE intervention [11] . Scaling up of trachoma control in South Sudan has , however , been hampered by a limited understanding of the geographical distribution of the disease , limited baseline data from suspected endemic areas and , ultimately , a shortage of funds to conduct surveys and implement control [12] . The majority of control activities have been undertaken east of the Nile , while a number of prevalence surveys [13] , [14] and a recently developed risk map [15] indicate that some areas west of the Nile are in equal need of intervention . Trachoma rapid assessments ( TRA ) were therefore conducted during 2009 to establish whether there was evidence of ongoing trachoma transmission in two of the suspected trachoma endemic states west of the Nile; Northern Bahr-el-Ghazal and Unity State . The latter was identified as being trachoma endemic throughout , with more than 10% of children aged 1–9 years in all of the thirteen surveyed villages showing signs of TF [16] . Prior to scaling up trachoma interventions in Unity , further survey work was required to generate estimates of trachoma prevalence for planning and monitoring purposes as the TRA methodology was not developed to generate these data [17] , [18] . With assistance from the United States Agency for International Development Neglected Tropical Disease ( NTD ) Control Program , a population-based prevalence survey ( PBPS ) was therefore conducted across Unity State to generate the required baseline data prior to intervention and investigate risk factors for trachoma and trichiasis in this region . Whilst risk factors for TF in children have been investigated in detail , few studies have specifically looked at risk factors for TT . Survey work in South Sudan is extremely strenuous , due to the lack of infrastructure and the harsh climate , and financial resources for trachoma-related activities are scarce . To minimize effort and cost it was decided to combine the eight ( of nine ) counties of Unity State for which there were no baseline data under one survey area . The implementation of a prevalence survey at an administrative level larger than a district or county may reduce the cost and time to conduct baseline prevalence surveys , which can enable national trachoma programs to raise funds and implement control measures sooner . The option of modifying the current approach of district-by-district PBPS surveys to target trachoma interventions has been recognised by other investigators ( e . g . [19] ) and some alternative study designs are currently under investigation . Here we sampled fewer sites than if county-by-county surveys had been conducted and used computerized simulations to compare the likely results from varying the number of sites surveyed , and hence the overall sample size , in this PBPS surveys with the aim of informing future sampling strategies in areas suspected to be highly trachoma endemic .
The protocol for the national trachoma prevalence survey , including the procedure for informed consent as outlined below , had received ethical approval from the Directorate of Research , Planning and Health System Development , Ministry of Health , Government of South Sudan ( MoH-GoSS ) , in 2008 . Clearance to conduct the surveys was obtained from the State Ministry of Health , followed by County Health Departments . The study was explained to each member of the selected households and the household head was asked to provide written consent for the entire household to participate in the study . To all illiterate heads of households the consent form was read out , and s/he was requested to consent by providing a thumbprint on the consent form . Due to the large number of individuals that were surveyed , it was considered impractical to obtain written consent from each study participant , and the MoH-GoSS approved this procedure . To document verbal consent , the name of each individual who provided verbal consent was recorded , along with the results from the eye examination . Each inhabitant of the household was requested to provide verbal consent; individuals who did not consent were not examined . Personal identifiers were removed from the dataset before analysis . During March 2010 , a PBPS was conducted across Unity State in the north of South Sudan ( Figure 1 ) . In 2008 , the state population had been estimated at 585 , 802 people ( 2008 Census Data ) , while recent program data from delivery of long-lasting insecticidal nets ( LLINs ) and mass drug administration ( MDA ) indicate that the actual population is around 700 , 000 ( Malaria Consortium , unpublished ) The inhabitants belong to two ethnic groups , the Nuer and the Dinka , both of which are pastoralists and rely on cattle products for most aspects of their daily lives . The PBPS followed the standard MoH-GoSS protocol for a national trachoma prevalence survey [20] , except for two notable exceptions: 1 ) The area surveyed here was nearly the entire state , while previous PBPS only surveyed counties or payams [13] , [14] , [21]–[23] . In South Sudan , the state is the first administrative level , while the county and payam comprise the second and third administrative level , respectively . 2 ) Based on financial and logistical considerations , it was decided to survey a total of 40 sites across the state , whereas the national protocol suggests that 20 sites should be surveyed per county . The rationale for using a larger study area and population in the present PBPS was that Unity State had already been indicated as highly trachoma endemic [14] , [15] , [16] . Survey modification as described above was therefore agreed among partners in South Sudan and with the International Trachoma Initiative to ensure that baseline data to target SAFE interventions could be generated quickly and efficiently without compromising its quality . Unity State is composed of nine counties: Mayom , Rubkona , Pariang , Leer , Guit , Koch , Abiemnhom , Mayendit and Panyijar . Mankien payam in Mayom was surveyed for trachoma in May 2005 [14] and , as no intervention had occurred in the county , Mayom was excluded from the present survey . The total number of villages to be surveyed in each county was determined based on the population size of the county , using population figures from the 2008 census , with the number of village to be surveyed in each county being proportional to its population size . For each county a list of its payams was drawn up to select survey villages . For counties where the number of villages to be surveyed exceeded the total number of payams , one village in each payam was randomly selected and any additional villages were randomly selected from payams with more than ten villages in them . For counties with fewer villages to be surveyed than the total number of payams , payams and the villages to be surveyed within these were randomly selected . Within each village , 20 households were randomly selected , apart from Pariang County where 21 households were sampled in two of the seven study villages . Households were randomly selected using the sketch map and segmentation method [24] . All residents of the household were enumerated , and all those present who gave informed consent were examined . Ophthalmic Clinical Officers and General Clinical Officers working for the State Ministry of Health in Unity were trained by an experienced ophthalmologist from Uganda on the use of the WHO simplified grading system over the course of three days; two additional ophthalmologists from Uganda attended the training , as they were designated to lead the survey teams . The WHO grading system categorizes trachoma infection according to five grades: TF , TI , trachomatous scarring ( TS ) , TT and corneal opacity ( CO ) [25] . Two stages of assessment were used to select the best trainees . In the first stage , trainee examiners identified trachoma grades using the WHO set of trachoma slides . Those examiners who achieved at least 80% agreement then proceeded to the second stage of field evaluation . During field evaluation , a reliability study comprising 50 persons of varying age and sex were selected by the ophthalmologist to represent all trachoma grades . Each trainee examiner evaluated all 50 participants independently and recorded their findings on a pre-printed form . Inter-observer agreement was then calculated for each trainee using the ophthalmologists' observation as the “gold standard . ” Unfortunately none of the trainees reached at least 80% inter-observer agreement in the field evaluation and it was therefore decided that grading was to be conducted by the two experienced ophthalmologists only , with the trained graders providing support during the survey . All inhabitants of selected households who provided verbal consent were examined using a torch and a 2× magnifying binocular loupe . Each eye was first examined for in-turned lashes ( TT ) , and the cornea was then inspected for CO . The upper conjunctiva was subsequently examined for inflammation ( TF and TI ) and scarring ( TS ) . Signs had to be clearly visible in accordance with the WHO simplified grading system in order to be considered present . Trachoma signs only had to be present in one eye for the person to be categorized as suffering from a particular grade of trachoma . Alcohol-soaked cotton swabs were used to clean the examiner's fingers between examinations . Individuals with signs of active trachoma or bacterial conjunctivitis were treated with 1% tetracycline eye ointment and provided with information on face washing and good hygiene practices . Patients with TT or other significant eye conditions were referred to the nearest facility where free surgery was available ( i . e . Bentiu Hospital ) . The data were verified at the end of each survey day and entered on a portable notebook computer using Microsoft Office Excel ( Microsoft Corporation , Seattle , USA ) . The computer was charged during the day by means of a small generator or through mains electricity , where available . Second data entry was conducted by different staff at the Malaria Consortium office after completion of the survey . Range and consistency checks were conducted for all variables in STATA 11 . 1 software ( Stata Corporation , College Station , TX , U . S . A . ) . Data were analysed in STATA 11 . 1 . Individuals with missing data for sex and/or age were excluded from the analysis . Overall and for each county , household attribute data were summarised as the mean ( standard deviation SD ) of the village means of the proportion of households with a particular attribute . The age and sex distribution of the enumerated population was tabulated . Unadjusted prevalence estimates for trachoma signs are presented with corresponding 95% exact binomial confidence intervals ( CI ) . Adjusted prevalence estimates and corresponding 95% CIs were obtained from random effects regression models for TF , TI and TS in children aged 1–9 years , for TF and TS in children aged 10–14 years and for TS , TT and CO in adults ( defined as those aged 15 and above ) . All models adjusted for age , sex and clustering of cases as between-village variation and between-household variation parameters . Models also adjusted for county to account for underlying unmeasured variation . Risk factors for TF in children aged 1–9 and for TT in adults ( 15 years and above ) were investigated using random effects logistic regression . Clustering of individuals with signs of trachoma at household and village levels was investigated using random effects regression models , comparing models with and without between cluster variance parameters using the likelihood ratio test ( LRT ) . Where a LRT p-value was <0 . 1 , the model with the additional random effect parameter was retained to obtain adjusted prevalence estimates of signs of trachoma . A stepwise approach was used to build multivariate models in both risk factor analyses , adjusting for age and sex a priori . To account for unmeasured covariates and underlying variation within counties , the final multivariate model for TF also adjusted for county . Model fit was assessed by the LRT where inclusion and exclusion criteria of LRT p≤0 . 1 were applied . To estimate the number of individuals requiring TT surgery , population figures for Unity reported by the 2008 census were adjusted upwards by 20% , based on field experience with LLIN distribution and MDA in the area indicating that census figures provided an underestimate of the actual resident population . The number of individuals aged below and above 15 years was calculated for each county ( data not shown ) using the relative proportions of people in these age categories as established during a detailed population census conducted as part of a recent MDA round in Mayom county ( Malaria Consortium , unpublished ) . The number of individuals in each of the two age categories was then multiplied by the TT prevalence for each county . To arrive at estimates for Mayom county , TT prevalence figures provided by Ngondi et al . [14] were used . A computerized simulation approach was used to compare the results from the actual survey , which was conducted in forty villages , to the alternatives of sampling twenty or sixty villages . The latter was the maximum number of sites that was deemed feasible to sample ( and actual experience showed that sampling of forty sites was already very challenging ) , while sampling of twenty sites would have resulted from applying the current MoH-GoSS survey protocol to the next administrative level ( i . e . the state instead of the county ) and would have helped to keep costs to a minimum . To compare survey designs , a ‘gold standard dataset’ was generated for the United Nations Office for the Coordination of Humanitarian Affairs ( UNOCHA ) village database , containing information on 823 geolocated settlements in Unity State ( excluding Mayom county ) . In order to generate as realistic ‘gold standard’ data as possible , the spatial characteristics of TF prevalence were assessed by constructing a semi-variogram , which plots the semi-variance – a measure of expected dissimilarity between a given pair of observations – as a function of the distance separating those observation . Prevalence data for active trachoma ( TF and/or TI ) in children aged 1–9 years were used from this study , as well as from other population-based prevalence surveys conducted in South Sudan , were used in the semi-variogram analysis . All surveys used the WHO simplified grading system and were conducted between 2001 and 2010 . This resulted in data from a total of 179 communities in South Sudan , 40 of which were located within Unity State . A logistic transformation was used on prevalence data from these surveys to reduce skew before analysis , and a small constant of 0 . 01 was added to the raw data to avoid transforming zero values . Empirical semi-variograms were then estimated using these transformed prevalence estimates . Prevalence data for all 823 settlements in Unity state were then generated using conditional simulation , which uses the semi-variogram parameters to generate multiple possible sets of data ( realisations ) that maintain the spatial variance characteristics of the source data [26] . One thousand realisations of prevalence data were generated , and back transformed , and those with the median , minimum and maximum overall prevalence were selected . These three realisations allowed an exploration of the performance of survey designs accounting for uncertainty in the simulated dataset . Further information on the use of semi-variograms and conditional simulation in NTD surveys is provided elsewhere [27] . Sampling simulations followed the original PBPS sampling design described in the Unity State survey , selecting a given number of clusters ( proportional to population size ) randomly from each of the included eight countries . Using the simulated ‘gold standard’ data , the precision of the original sampling plan of forty clusters was compared to twenty and sixty clusters . Sampling simulations were repeated 1 , 000 times on each realisation and the range of prevalence estimates resulting from surveying twenty , forty or sixty clusters was recorded . Semi-variogram analyses , conditional simulation and sampling simulations were all carried out using bespoke scripts in R 2 . 10 . 1 [28] .
Overall , households were of a median size of seven inhabitants , but throughout the study area some households were more than twice this size ( Table 1 ) . Data summarised across all villages suggest that few heads of households had received formal education or owned a radio ( overall , approximately 12% and 21% respectively ) , but that three quarters of households owned cattle . Flies were commonly observed within and around living areas or on the eyes of children ( 95% of households ) . Latrine access was poor ( 8% ) and around half of the respondents reported a journey of more than 15 minutes walk to reach water . A total of 5 , 727 individuals were enumerated after exclusion of 13 individuals with missing values for age ( n = 9 ) or sex ( n = 4 ) ( Table 1 ) . The distribution of age by sex within enumerated individuals suggested that the census included similar proportions of male and female children aged 1–9 , and more females than males aged 15 years and above ( Table 1 ) . Ninety percent of enumerated children aged 1–9 years were examined for signs of trachoma . Of the individuals aged 15 years and above more than 80% of women were examined , but only around 45% of male residents were examined ( Table 1 ) . The overall adjusted prevalence of TF in children aged 1–9 years was 71% ( 95% CI: 70–72% ) . Unadjusted and adjusted prevalence and corresponding 95% CIs were the same as integer values ( Table 2 ) . Incorporating children with TI , the prevalence of active trachoma in children aged 1–9 was 84% ( adjusted 95% CI: 83–85% ) . The prevalence of trachomatous scarring in this age group was at most 2% . Amongst 10 to 14 year olds , TF was also highly prevalent ( approximately 42% ) , while scarring was observed in 8% and TT in up to 6% ( Table 2 ) . Adjusted prevalence result suggest between 12% and 15% of those aged 15 years and above had TT , far exceeding the treatment threshold of 1% , and between 5% and 8% were experiencing CO in one or both eyes ( Table 2 ) . For estimation of all prevalence figures , adjusted values were very similar to unadjusted values even in the presence of between-cluster variation at household and village levels . Independently more TF was seen in children aged 3–5 years , compared to those aged 1–2 years , for children with ocular and , or nasal discharge and for children from households with flies in and around the living areas or on the faces of children ( Table 3 ) , after accounting for clustering of TF within households , within villages and adjusting for county . All of these factors remained statistically significant risk factors for TF in children aged 1–9 years in multivariate analyses , which adjusted for these factors , in addition to age and sex a priori and clustering as described . Random effects univariate regression models accounted for clustering at household and village level and suggested increased odds of TT in: i ) people aged 30 and above , ii ) female adults , iii ) people with ocular and , or nasal discharge , and iv ) possibly also people who needed to travel further to reach water ( Table 4 ) . Decreased odds were seen for people: i ) living in larger households , ii ) having access to a latrine , iii ) residing in households that disposed of waste more than 20 metres away , and iv ) radio ownership . After adjustment for age , sex and ocular discharge , nasal discharge was no longer associated with risk of TT ( p = 0 . 206 ) . Following the addition of household size , latrine access and time to water were also no longer significant ( p>0 . 1 ) . A final model for TT included age , sex , ocular discharge , household size and distance to waste disposal ( Table 4 ) . Evidence of between-village variation disappeared from the final multivariate model , but household clustering was still apparent . The total population of Unity was estimated to be 702 , 961 based on recent delivery of interventions by Malaria Consortium . Of these , 47% were assumed to be aged 15 or above . Using TT prevalence figures for each county to calculate surgery needs , this meant that a total of 54 , 178 individuals ( lower and upper bounds = 40 , 327–71 , 119 ) are likely to require TT surgery in Unity State , with the majority ( 91% ) being aged 15 or above . Available community level data from the whole of South Sudan had a median prevalence of 53% ( range 0–100% ) . The semi-variogram generated from these data showed that spatial autocorrelation was present up to approximately one decimal degree ( ∼110 km ) ( Figure 2 ) . The simulated results from each of the alternative sample designs are presented in Table 5 , and confirm a reduction in precision associated with reducing the number of clusters by half and an increase in precision when additional clusters are included . Theoretically the median estimate should be the same for all sampling scenarios of a given realisation . In these simulations , the median estimate for each realisation was slightly higher in designs with a higher number of clusters . This is a function of the PBPS sampling strategy , which places relatively more weight on counties with a smaller population size in cases where the calculated number of clusters is rounded up to equal that in different sampling scenarios . In these simulations , Abiemnhom had a low population proportion ( 0 . 04 ) and a lower predicted prevalence . One site was included from this county in two sampling scenarios ( 40 and 20 clusters ) , giving this site a slightly greater weight in the smaller sample . If sites had been selected randomly from the entire state ( the next highest population level ) , then the simulated median prevalence would have been the same for all sampling scenarios . The overall effect on the results is negligible , and the confidence intervals presented in Table 5 are the key parameter . Importantly , the lower range of the credible interval remains well over the 40% threshold in all scenarios , indicating that any of these strategies would provide good evidence that this area is hyperendemic for trachoma . This is best illustrated in Figure 3 , which compares the three sampling strategies for the median , minimum and maximum realisations using filled density plots .
Data from a previous survey [14] , a national trachoma risk map [15] and TRA [16] had indicated trachoma to be endemic throughout Unity State , but provided an insufficient baseline to initiate interventions . The present study therefore set out to generate the required data . The results established that trachoma prevalence in Unity State far exceeds the WHO recommended threshold of 10% TF prevalence in children aged 1–9 years for initiation of antibiotic MDA [10] and of 1% for TT in adults for surgery provision . Alongside the obvious need for MDA and surgery , both the ‘F’ and ‘E’ components of the SAFE strategy [29] will need to be scaled up to address key trachoma risk factors , hence maximising the impact of intervention and contributing towards sustaining trachoma control once it has been achieved . The very high prevalence of trachoma found in the eight surveyed counties is consistent with results from previous work in Mayom county of Unity State [14] , predictions generated by a national trachoma risk map [15] and findings from most other trachoma surveys in South Sudan [see map in [22]] . The uniformly high prevalence of trachoma found across Unity State is also reflected in the results from the sampling simulation . These provide evidence that more economical survey designs , sampling fewer clusters , can be used for decision making in areas where trachoma is likely to be hyperendemic . This may support a move towards regional level surveys in areas where there is evidence that the prevalence of trachoma is high . Hyperendemic areas , such as Unity State , will have more uniform treatment requirements than areas of lower prevalence , in which individual foci may be above treatment thresholds but overall , the health district does not qualify for MDA . In meso- and hypo-endemic areas , higher resolution surveys will therefore be required to capture high prevalence foci and sufficiently understand spatial variation in disease and treatment needs . Lot Quality Assurance Sampling , which allows decisions on control to be made using small sample sizes , has previously been explored in Malawi [30] and Vietnam [31] and may offer a potential solution . Further exploration of this principle would be possible using data from states to the west of Unity State , where trachoma prevalence is predicted to be much lower [15] . Compared to data from other trachoma endemic areas of Sub-Saharan Africa , the levels of TF and TT found in Unity State were considerably higher . For example , in Kenya's Samburu district , prevalences of 35% TF were reported in children below 10 years and 6% TT in adults above 14 years of age . The same survey covered five other districts and found prevalence of both signs to be lower in all of them [32] . A regional survey in Chad found that TF was present in 31 . 5% of children under 10 years and that 1 . 5% of women over 14 years had signs of TT [33] . In Ethiopia , the national prevalence of TF in children below 10 years has been reported as 26 . 2% , while 3 . 1% of women above the age of 14 years showed signs of TT [34] . These proportions obviously vary considerably throughout the country; an overall prevalence of 32 . 7% TF and 6 . 2% TT has been reported from Amhara regional state [35] , Ethiopia's worst affected region [34] . Unity State , like other parts of South Sudan [13] , [22] , is therefore among the most severely affected by trachoma in Africa . The present study identified a number of risk factors for trachoma in Unity State . For TF , ocular and/or nasal discharge and the presence of flies in and around the living areas or on children's faces were associated with an increased risk of trachoma infection , and children between 3–5 years of age were at highest risk . Risk factors for TT in those aged 15 years and above were age , sex , ocular discharge , number of children residing in the household and time ( as a proxy for distance ) to the nearest water collection site . These observations are consistent with our general understanding of trachoma epidemiology [1] and findings of other studies in South Sudan [36] , [37] and in the region [38]–[41] . While eye discharge was identified as a TT risk factor in adults we think that it is likely to be a consequence of TT rather than a cause and may also be indicative of being generally unwell , which could be due to having eye health problems . Latrine provision and close access to water were both limited throughout the study villages , but the strong associations with increased risk of TT is suggestive of poorer hygiene in those with more pronounced eye problems , as were poor waste disposal practices . MDA of antibiotics is unlikely to have an effect on these risk factors , highlighting the importance of health education and environmental improvements as part of a comprehensive control programme [39] . Adult females , rather than males , were at much higher risk of having TT , which is generally thought to be due to the close contact of women with children , children being the main reservoir of infection [1] . Fewer adult males were examined from the enumerated population and if these unexamined males were unaffected by the later stages of disease , the sex effect seen would be greater . We found that TT increased with decreased household size and distance to waste disposal . Increased TT in adults in smaller households , after adjustment for other factors , may also be indicative of adults with TT living in isolation and poverty rather than a smaller household being a risk factor for TT . The present study has a number of limitations . Unlike other PBPS conducted in South Sudan and elsewhere , relatively few villages were sampled in each county , which may have affected the precision of the prevalence estimates for the State , particularly if there was much variation in prevalence within the survey area . Sampling of a total of 40 sites over a large geographical area was , however , considered justified because GIS-based risk-mapping and TRA data had provided a reasonable indication that Unity State was trachoma endemic throughout [15] , [16] . As indicated by the summary measures of household attributes there was some variability between counties and this will not have been captured in as much detail as sampling of 20 sites per county would have allowed . It is nevertheless thought that the overall results generated by the study are reflective of the population in the study area and suitable for decision-making on intervention , particularly as WHO thresholds are very clearly exceeded at all population levels . In the course of scaling up the SAFE strategy it may , however , be found that TT surgery requirements need to be adjusted up- or downwards . This is because we were unable to examine 55% of the enumerated adult males , meaning that the prevalence of TT , CO and visual impairment in the study area could differ somewhat from the estimates provided here . It was not possible to collect DNA samples from children in order to conduct laboratory testing to detect Chlamydia trachomatis . It has been suggested that in a high prevalence setting , prior to availability of mass antibiotic treatment , a diagnosis of TF may not reflect infection with C . trachomatis in 30% of children aged 1–10 [42] . In Unity State , a reduction of 30% in overall prevalence would still imply that the area is hyper-endemic and in desperate need of MDA . Data used to generate simulated realisations were also based on estimates of TF and/or TI in children aged 1–9 years . The inherent measurement error in using clinical signs as opposed to infection prevalence is also a limitation of the realisations , which were based on real survey data . Additionally , generation of the gold standard database is necessary to evaluate alternative sampling designs , but involves a number of assumptions . Included surveys span nearly a decade and it is possible that the spatial characteristics of infection may have changed over time . However , the lack of MDA in many suspected trachoma endemic areas and the general absence of health infrastructure and services in South Sudan makes it unlikely that the spatial distribution of infection will have changed significantly . Even if there had been changes in the rest of South Sudan over time , sampling simulations were conducted within Unity State and this area will have been most strongly influenced by the recent survey reported here . Much remains to be done if South Sudan is to eliminate blinding trachoma by 2020 . As a first step , a comprehensive intervention program needs to be scaled up across Unity state , now that the baseline data are available . Secondly , there is a need for further large-scale surveys , such as the one reported here , in other states . As much as half of the country may need to be targeted with SAFE interventions [15] , but for many areas this still needs to be confirmed . For large suspected endemic areas with little to no available data , such as Upper Nile State and parts of Warrap , Lakes and Central Equatoria States , a two-step procedure such as that used in Unity State would seem the most effective way to get control activities quickly under way . TRAs should be conducted to determine whether there is evidence of transmission in suspected endemic areas . Where the proportion of children with signs of TF is found to be very high , counties could then be combined under one survey area to more quickly identify SAFE intervention needs . | Large parts of South Sudan are thought to be trachoma endemic but baseline data , required to initiate interventions , are few . District-by-district surveys , currently recommended by the World Health Organization ( WHO ) , are often not financially or logistically viable . We therefore adapted existing WHO guidelines and combined eight counties ( equivalent to districts ) of Unity State into one survey area , randomly sampling 40 villages using a population-based survey design . This decision was based on a trachoma risk map and a trachoma rapid assessment , both identifying the state as likely to be highly endemic . The survey confirmed trachoma as being hyperendemic throughout Unity State , meaning that large-scale intervention should be initiated now . Simulation studies were conducted to determine the likely outcome if fewer ( n = 20 ) or more ( n = 60 ) villages had been sampled , confirming that precision decreased or increased , respectively . Importantly , simulation results also showed that all three sample sizes would have led to the same conclusion , namely the need for large-scale intervention . This finding suggests that district-by-district surveys may not be required for areas where trachoma is suspected to be highly prevalent but that are lacking baseline data; instead districts may be combined into a larger survey area . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"disease",
"mapping",
"infectious",
"diseases",
"survey",
"methods",
"public",
"health",
"and",
"epidemiology",
"epidemiology",
"infectious",
"disease",
"epidemiology",
"neglected",
"tropical",
"diseases",
"trachoma"
] | 2012 | Prevalence of Trachoma in Unity State, South Sudan: Results from a Large-Scale Population-Based Survey and Potential Implications for Further Surveys |
Several genetic variants associated with platelet count and mean platelet volume ( MPV ) were recently reported in people of European ancestry . In this meta-analysis of 7 genome-wide association studies ( GWAS ) enrolling African Americans , our aim was to identify novel genetic variants associated with platelet count and MPV . For all cohorts , GWAS analysis was performed using additive models after adjusting for age , sex , and population stratification . For both platelet phenotypes , meta-analyses were conducted using inverse-variance weighted fixed-effect models . Platelet aggregation assays in whole blood were performed in the participants of the GeneSTAR cohort . Genetic variants in ten independent regions were associated with platelet count ( N = 16 , 388 ) with p<5×10−8 of which 5 have not been associated with platelet count in previous GWAS . The novel genetic variants associated with platelet count were in the following regions ( the most significant SNP , closest gene , and p-value ) : 6p22 ( rs12526480 , LRRC16A , p = 9 . 1×10−9 ) , 7q11 ( rs13236689 , CD36 , p = 2 . 8×10−9 ) , 10q21 ( rs7896518 , JMJD1C , p = 2 . 3×10−12 ) , 11q13 ( rs477895 , BAD , p = 4 . 9×10−8 ) , and 20q13 ( rs151361 , SLMO2 , p = 9 . 4×10−9 ) . Three of these loci ( 10q21 , 11q13 , and 20q13 ) were replicated in European Americans ( N = 14 , 909 ) and one ( 11q13 ) in Hispanic Americans ( N = 3 , 462 ) . For MPV ( N = 4 , 531 ) , genetic variants in 3 regions were significant at p<5×10−8 , two of which were also associated with platelet count . Previously reported regions that were also significant in this study were 6p21 , 6q23 , 7q22 , 12q24 , and 19p13 for platelet count and 7q22 , 17q11 , and 19p13 for MPV . The most significant SNP in 1 region was also associated with ADP-induced maximal platelet aggregation in whole blood ( 12q24 ) . Thus through a meta-analysis of GWAS enrolling African Americans , we have identified 5 novel regions associated with platelet count of which 3 were replicated in other ethnic groups . In addition , we also found one region associated with platelet aggregation that may play a potential role in atherothrombosis .
While platelets play a fundamental role in hemostasis , they are also important in the development of atherosclerosis and arterial thrombosis [1] . An elevated platelet count has been associated with adverse clinical outcomes after thrombolysis or coronary intervention in patients presenting with acute myocardial infarction and moderate reductions in platelet count by thrombopoietin inhibition were associated with reduced thrombogenesis in a primate model [2]–[4] . The heritability of variation in platelet count is substantial with estimates ranging from 54% to more than 80% [5]–[8] . In the GeneSTAR study , a cohort included in the current meta-analysis , the heritability of platelet count is 67% [9] . Like platelet count , an elevated mean platelet volume ( MPV ) is also associated with adverse cardiovascular events and its reported heritability is as high as 73% [8] , [10]–[12] . The heritability of MPV in the GeneSTAR cohort was 71% [9] . Recent genome-wide association studies ( GWAS ) and meta-analyses have identified genetic variants associated with these two platelet traits in Caucasians and a Japanese population [13]–[15] . A recent meta-analysis in the CARe Project , involving genotyping of about 50 , 000 single nucleotide polymorphisms ( SNPs ) in 2 , 100 candidate genes , also reported two genetic variants associated with platelet count in African Americans [16] . The genetic variants reported to date explain only a small fraction of the heritability in platelet count and MPV , providing an opportunity for new studies to discover additional genetic variants of importance [15] . Moreover , African Americans have higher platelet counts than Caucasians and additional genetic variants may contribute to this difference [17] . Because of the different allele frequencies and linkage disequilibrium patterns in populations of European and African ancestry , we anticipated that we might discover new genetic loci associated with platelet count and MPV in an African American population compared to Caucasians [18] . We performed a meta-analysis of 7 GWAS studies that included African-American subjects in the Continental Origins and Genetic Epidemiology Network ( COGENT ) in order to identify novel genetic variants associated with platelet count and MPV .
The first of the novel loci from platelet count meta-analysis is located on chromosome 6p22 . The best SNP ( rs12526480; p = 9 . 1×10−9 ) in this region is located in the intron of the leucine-rich repeat containing 16A gene ( LRRC16A ) . The minor allele ( G ) of rs12526480 was associated with decreased platelet count . Ten additional SNPs in the region had p<10−6 ( Table 2 and Table S2 ) . The LRRC16A gene encodes a protein called ‘capping protein ARP2/3 and myosin-I linker’ ( CARMIL ) , which plays an important role in cell-shape change and motility . Genetic variants in LRRC16A have been previously reported to be associated with serum uric acid levels [20] , nephrolithiasis [21] and markers of iron status [22] but there have been no reports of any association with either platelet count or other platelet phenotypes . In the three European American cohorts , rs12526480 was statistically significant in one cohort ( p = 0 . 01 ) and near nominal significance in the combined meta-analysis ( p = 0 . 06 ) with an effect size and direction similar to that observed in African Americans . In Hispanic Americans , rs12526480 was not significantly associated with platelet count ( Table 3 ) . Given the proximity of the LRRC16A gene to the hemochromatosis ( HFE ) gene and the well-known reciprocal relationship between platelet count and iron stores , we additionally assessed the association between rs12526480 and red cell phenotypes in the COGENT African Americans . There was no evidence of association between LRRC16A genotype and hemoglobin , hematocrit , red cell count or mean corpuscular volume in the 16 , 388 African Americans , nor was there any evidence of association between rs12526480 genotype and serum ferritin in 672 African Americans from CARDIA or 2 , 126 from JHS . Nor did adjustment for red cell phenotype or iron status alter the relationship between platelet count and rs12526480 genotype . Finally , we had uric acid levels available in 943 African Americans from CARDIA; again there was no association with LRRC16A genotype ( Table S5 ) . The second locus is on chromosome 7q11 where two SNPs in intronic regions of the CD36 gene ( rs13236689; p = 2 . 8×10−9 and rs17154155; p = 1 . 1×10−8 ) reached GWAS significance threshold , while 8 additional SNPs had p<10−6 . rs13236689 and rs17154155 are in close linkage disequilibrium ( r2 = 0 . 90 in the HapMap Yoruban population ) . After conditioning on rs13236689 in the association analysis , rs17154155 did not remain statistically significant ( p = 0 . 39 ) . Of the three European American cohorts , rs13236689 was statistically significant in the WHI cohort ( p = 0 . 05 ) but not in the meta-analysis of all three studies ( p = 0 . 07 , Table 3 ) . The CD36 gene encodes a thrombospondin receptor ( platelet glycoprotein IV ) which is present on the surface of platelets and several other cells [23] . rs17154155 has been reported to be associated with platelet function as well as with platelet expression of CD36 [24] , [25] . In the third locus on chromosome 10q21 , 71 SNPs reached GWAS threshold and 57 additional SNPs had p<10−6 . Two non-synonymous common variants of unknown functional significance , rs 10761725 ( resulting in serine to threonine substitution ) and rs1935 ( resulting in glutamate to aspartate substitution ) , in this region also crossed the GWAS threshold . All 128 SNPs in this region appear to be in strong linkage disequilibrium based on Yoruban HapMap data . The most significant SNP in this region , rs7896518 ( p = 2 . 3×10−12 ) , is located in an intron of the jumonji domain containing 1C ( JMJD1C ) gene . SNPs in this region have been reported to be associated with MPV ( rs2393967 ) and with native platelet aggregation in platelet-rich plasma ( rs10761741 in Caucasians and rs2893923 in African Americans ) but not with platelet count [15] , [26] . For rs7896518 , data were available from 2 European American cohorts and meta-analysis found a significant association reaching GWAS threshold ( p = 2 . 61×10−9 ) with similar direction of effect size ( Table 3 ) . The fourth novel locus was located on chromosome 11q13 . The most significant SNP ( rs477895; p = 4 . 9×10−8 ) was in an intron of the BCL2-associated agonist of cell death ( BAD ) gene , while 23 other SNPs had p<10−6 . For rs477895 , all replication cohorts had effect sizes in a direction similar to African Americans and one European American and the Hispanic cohorts reached statistical significance ( p = 4 . 48×10−3 and p = 0 . 04 respectively ) . Meta-analysis of the three European American cohorts also found significant association of rs477895 with platelet count ( P = 1 . 71×10−3 , Table 3 ) . The protein encoded by the BAD gene inhibits the activity of the BCL-xL and BCL-2 proteins and thus has a pro-apoptotic effect [27] . Phospholipase C β3 protein encoded by another gene at this locus , PLCB3 , is also known to be present in platelets and its deficiency results in impaired platelet function in mice [28] . This locus also contains SLC22A11 and SLC22A12 , two genes that encode solute carrier proteins and previous GWAS have found association of genetic variants in these genes with serum uric acid levels [20] . Of the two genes , the transcript of SLC22A11 is present in significant amount in platelets as is the transcript for BAD [29] . Interestingly , a SNP about 20 kbp upstream of SLC22A11 , rs4930420 , almost reached GWAS threshold ( p = 9 . 16×10−8 , r2 with rs477895 = 0 . 21 ) and four additional SNPs in complete LD with rs4930420 ( r2 = 1 ) had p-values<10−6 . By examining the actual linkage disequilibrium patterns in this region in COGENT , and by performing conditional regression analysis in more than 8 , 400 African Americans from the WHI cohort simultaneously adjusting for BAD rs477895 and SLC22A11 rs4930420 , we demonstrate that there are likely at least 2 independent platelet count association signals in this region and that the BAD and PLCB3 polymorphisms appear to represent the same association signal ( Table S6 ) . The fifth novel locus was on chromosome 20q13 where one SNP in the SLMO2 gene exceeded GWAS significance threshold ( rs151361; p = 9 . 4×10−9 ) while 2 other SNPs had p<10−6 . One of these two SNPs was located in the first intron of TUBB1 gene ( rs6070696; p = 2 . 5×10−7 ) and was 16 . 3 kbp downstream of the lead SNP ( YRI HapMap r2 = 0 . 6 ) . The TUBB1 gene encodes a beta1 tubulin , which plays an important role in megakaryopoiesis [30] . All replication cohorts had effect sizes in the direction similar to African Americans for rs151361 but only one European American study reached statistical significance ( p = 0 . 01 ) . The meta-analysis of the three European American replication cohorts also found a statistically significant association between rs151361 and platelet count ( p = 1 . 1×10−3 , Table 3 ) . In addition to identifying novel loci , we also replicated 5 previously reported loci at GWAS significance threshold and 3 other loci that were highly significant in our study but not at GWAS significance level ( Table S7 ) . The strongest signal in our platelet count meta-analysis was from chromosome 6p21 ( SNP with the lowest p-value = rs210134; p = 2 . 3×10−15 ) located in the BAK1 gene , a locus that has been reported previously in Caucasians , Japanese , and African American populations [13]–[16] . We also found strong associations between platelet count and loci on chromosomes 6q23 ( rs9494145; p = 2 . 8×10−9 ) , 7q22 ( rs342293; p = 1 . 6×10−8 ) , and 12q24 ( rs6490294; p = 4 . 8×10−9 ) , all of which have been previously reported for Caucasians but not for African Americans [15] . Finally , we confirmed the association of a genetic variant rs8109288 ( p = 5 . 0×10−10 ) in the tropomyosin 4 ( TPM4 ) gene at chromosome 19p13 that has been previously reported for African Americans in a candidate gene study [16] . In our replication cohorts , rs8109288 was associated with platelet count in meta-analysis of European American cohorts and in Hispanic Americans ( p = 2 . 6×10−8 and 0 . 02 respectively ) . We were also able to confirm the association of all previously reported SNPs ( or a nearby SNP in the same LD block ) with platelet count at a p<0 . 05 ( Table S5 ) . Of the three loci we identified at GWAS significance level for MPV , 2 have been previously reported to be associated with MPV in Caucasians , and one has been reported previously in African Americans . The association which has been previously reported in African Americans was of the A-allele of rs8109288 in TPM4 with increased MPV ( p = 3 . 3×10−9 ) ; the same SNP was also associated with platelet count in this study . TPM4 , a protein with a major role in stabilizing the cellular cytoskeleton , is present in platelets [31] . In the 7q22 region , we found that the SNP with the lowest p-value for MPV ( rs342296; p = 1 . 4×10−11 ) was different from the SNP most associated with platelet count ( rs342293; p-value = 5 . 84×10−11 ) although the two SNPs were only 684 bp apart and are in the same LD block ( r2 = 0 . 92 based on HapMap II YRI ) [15] . We also replicated a locus associated with MPV on 17q11 ( rs11653144; p = 4 . 2×10−8 ) at GWAS significance threshold [15] . Of the 10 additional previously reported loci for MPV , we found statistically significant associations with 7 of them although these associations did not reach GWAS significance threshold ( Table S8 ) . For the loci that we were unable to replicate , we found other nearby SNPs with p<0 . 05 . The direction of effect for all SNPs was not similar to the previously reported study of individuals of European ancestry suggesting that the alleles at the causal loci may be different between the two populations . Three regions ( 7q11 , 7q22 , 10q21 ) containing four SNPs ( rs13236689 , rs342296 , rs342293 , rs7896518 ) have already been shown to be associated with platelet aggregation [24]–[26] , [32] . Therefore , the SNPs with the lowest p-values in each of the remaining 8 regions ( Table 4 ) identified for either platelet count or MPV were examined for their association with platelet aggregation in 832 African-American individuals from the GeneSTAR study . Of the 8 SNPs , 3 were associated with a significant change in agonist-induced platelet aggregation but only one exceeded the Bonferroni-corrected significance threshold of 0 . 005 ( Table 4 ) . The minor allele ( C ) of rs6490294 in the ACAD10 gene ( 12q24 ) was associated with increased ADP-induced platelet aggregation ( p = 0 . 002 ) . Variants in this region have been previously reported to be associated with coronary artery disease [15] . The minor allele ( A ) of the 2nd SNP , rs8109288 , in the TPM4 gene , was associated with decreased arachidonic-induced platelet aggregation ( p = 0 . 03 ) and a trend towards decreased aggregation with ADP ( p = 0 . 09 ) . The minor allele ( G ) of the 3rd SNP , rs151361 , in the SLMO2 gene , was associated with increased ADP-induced platelet aggregation ( p = 0 . 008 ) . The last 2 SNPs were nominally significant but did not exceed the Bonferroni-corrected significance threshold .
We report the first meta-analysis of GWA studies of platelet count and MPV in a large number of African American participants from 7 population-based cohorts . We have identified 5 novel loci associated with platelet count of which three were replicated in the European American cohorts and one in the Hispanic cohort . None of these new African-American platelet loci have been reported previously in any racial group . In addition , we have confirmed that several loci previously reported in Europeans or Japanese are also associated with these platelet phenotypes in African Americans . We have further shown that 3 of the 8 loci ( with one exceeding Bonferroni-corrected threshold ) , for which there have been no previously known association with platelet aggregation , are also associated with differences in platelet function using a subset of our African American sample . Interestingly , the 5 novel platelet count loci are intragenic and 4 of these genes are known to have some role in platelet formation or biology . Platelets are small anucleate blood cells that are released from the cytoplasm of much larger bone marrow precursor cells known as megakaryocytes . One of the novel findings is the association of LRRC16A gene with platelet count . The protein encoded by the LRRC16A gene , capping protein ARP2/3 and myosin-I linker ( CARMIL ) , plays an important role in actin-based cellular processes . Actin filaments are essential for end-amplification of pro-platelet processes during megakaryocyte maturation [33] . CARMIL exposes the barbed ends of actin filaments by binding to and then dislodging the capping protein from the actin filament [34] . Capping proteins are up-regulated during megakaryocyte maturation and LRRC16A is differentially expressed in megakaryocytes compared to other blood cells [35] , [36] . The capping protein binding region of the CARMIL protein resides in the later part of the protein ( 940–1121 amino acid residues ) , which is a highly conserved region from protozoa to vertebrates . The majority of the residues in this region are critical for the anti-capping protein activity of CARMIL [37] . The rs12526480 genetic variant identified in our study is located in the latter part of the gene and may be in LD with a functional mutation in this conserved region . Any mutation that decreases the ability of CARMIL to dislodge capping protein from the barbed ends of the actin filament may result in abnormal megakaryocyte maturation and decreased platelet formation which is consistent with the direction of effect we observed in our study . Another novel finding not reported in earlier GWA studies is the association of platelet count with CD36 , a gene that encodes a receptor present on the surface of platelets , megakaryocytes , and several other cells . CD36 has a wide variety of ligands including thrombospondin [23] . Both CD36 and thrombospondin genes are up-regulated during megakaryocyte maturation and binding of thrombospondin-I to CD36 inhibits megakaryopoiesis , thus potentially providing a feedback mechanism for control of megakaryopoiesis [34] , [36] , [38] . The exact mechanism through which activation of CD36 inhibits megakaryopoiesis is unclear but may involve activation of extrinsic apoptotic mechanisms [39] . The most significant SNP associated with platelet count ( rs210134 in BAK1 ) in our study is in complete LD with the most significant BAK1 SNP reported to be associated with platelet count in individuals of European ancestry ( rs210135 , r2 = 1 with rs210134 in HapMap II YRI , p = 2 . 18×10−14 in the current study ) . While the magnitude of effect is similar , the direction of effect is opposite suggesting that the allele at the causal locus is different in the two ethnic groups . A candidate gene study in African Americans has reported another SNP ( rs449242 , r2 = 0 . 81 with rs210134 in HapMap II YRI ) in BAK1 and the direction of effect is similar to our study ( Table S5 ) [16] . In addition to confirming the association of genetic variants in the pro-apoptotic BAK1 gene with low platelet count , we have identified and replicated a variant in another pro-apoptotic gene , BAD , that is associated with low platelet count . The protein encoded by BAD acts as a sensor for apoptotic signals upstream of BAK and activates BAK through indirect mechanisms [27] . The identification of these two genes in the intrinsic apoptotic pathway highlights the importance of the apoptotic process in modulating platelet lifespan in the circulation , which is one of the mechanisms that regulate platelet count [40] . Interestingly , this region also contains genetic variants associated with serum uric acid levels [20] , however , the mechanism through which uric acid levels may be associated with platelet count remains unclear . Genetic variants in the JMJD1C gene have been previously reported to be associated with MPV in Caucasians but not with platelet count . Conversely , we found several SNPs in this region that reached GWAS significance threshold for association with platelet count but none with MPV and we replicated the lead SNP in European Americans at GWAS threshold . In a GWAS study of platelet aggregation in Caucasians , the minor allele ( T ) of rs10761741 was associated with an increase in epinephrine-induced platelet aggregation in Caucasians [26] . JMJD1C gene is a histone demethylase and appears to be involved in steriodogenesis [41] . In addition to its association with platelet aggregation and MPV , previous GWAS have found genetic variants in this gene to be associated with serum levels of alkaline phosphatase and lipoprotein particle size and content [42]–[44] . In addition to confirming the finding of association of A-allele of rs8109288 in TPM4 gene with lower platelet count [16] and replicating this finding in European Americans , we also confirmed the association of the A-allele of this SNP with increased MPV and found a nominally significant association with decreased platelet aggregation . TPM4 gene expression is higher in megakaryocytes than other blood cells or other hematopoietic cells [35] , [45] . Tropomyosin proteins play a central role in actin-based cytoskeletal changes and there appears to be biological plausibility for an effect of genetic variants on megakaryocyte maturation and platelet aggregation [46] . The final novel locus in the SLMO2 gene was also replicated in European Americans but SLMO2 gene has no known role in megakaryocyte biology . However , the variant is located within 13 kb of the TUBB1 gene , which is essential in the formation of normal mature platelets . The TUBB1 gene encodes beta1-tubulin that is exclusively expressed in platelets and megakaryocytes and forms a component of microtubules [30] . Loss of function mutations in TUBB1 gene have been reported in the literature and result in thrombocytopenia , large platelets , and increased risk of intracranial hemorrhage in men [47] , [48] . The G-allele of the rs1513691 variant is associated with increased platelet count , decreased MPV , and increased aggregation , which may point towards a gain in function mutation in this region . All previously reported loci that were also significantly associated with platelet count or MPV at GWAS threshold in our study have known biological roles in platelet biology . Two of these regions , 6q23 and 12q24 , have pleiotropic effects with the 6q23 region associated with several hematological traits [13] , [15] , [49] and the 12q24 region associated with celiac disease and coronary artery disease [15] . More importantly , we also found that the 12q24 locus was associated with platelet aggregation after Bonferroni adjustment for multiple comparisons and thus may provide a mechanistic explanation of its role in development of coronary artery disease . The GG genotype of the most significant SNP in the 7q22 region , rs342293 , is known to be associated with higher PIK3CG mRNA levels in platelets [32] . SNPs at this locus are also associated with platelet aggregation , pulse pressure , and carotid artery plaque [26] , [50] , [51] . TAOK1 is an important regulator of the mitotic progression and may also play a role in the apoptosis of cells [52] , [53] . Our study included over 16 , 000 participants with platelet count and over 4500 participants with MPV measured and we were able to identify loci that explain between 0 . 16–0 . 33% of the variance in platelet count and loci that explain 1–1 . 5% of the variance of MPV ( Table S9 ) . Overall , the loci we identified explain up to 7% of the variance in platelet count and up to 6% of the variance in MPV , assuming that the each of these loci is independent . However , for both platelet count and MPV , the estimated heritability is >50% . Therefore , for each of these traits , the majority of heritability remains unexplained . One of the limitations of GWA studies is the limited power to detect effects caused by genetic variants with frequency <5% . We hypothesize that a significant proportion of the heritability of platelet count and MPV may be explained by variants with frequency <5% . Alternatively , there may be a large number of additional common variants that affect these traits , but have more modest effects . In conclusion , we have conducted a meta-analysis of GWAS studies of platelet count and MPV in a large African American population and identified novel genetic variants in regions with genes that are likely to have a role in platelet formation . Furthermore , we have replicated 3 of the 5 novel loci in European Americans and one in Hispanic Americans . The novel regions identified may provide a focus for further research in improving our understanding of the biology of megakaryocyte maturation and platelet survival . In addition , we examined the effect of the genetic variants associated with platelet count and MPV on platelet function , and found 3 of these genetic variants to be associated with agonist-induced platelet aggregation of which one crossed Bonferroni-corrected significance threshold . Whether these newly identified genetic variants contribute to the risk of coronary artery disease or myocardial infarction , or to disorders associated with hyper- or hypo-aggregation of platelets , merits further investigation .
The 7 studies included in this meta-analysis belonged to COGENT and enrolled 16 , 388 African American participants . The supplementary text contains a detailed description of each participating COGENT study cohort ( Text S1 ) . All participants self-reported their racial category . Additional clinical information was collected by self-report and clinical examination . All participants provided written informed consent as approved by local Human Subjects Committees . Study participants who were pregnant or had a diagnosis of cancer or AIDS at the time of blood count were excluded . We also excluded subjects who were outliers in the analysis of genetic ancestry ( as determined by cluster analysis performed using principal component analysis or multi-dimensional scaling ) or who had an overall SNP missing rate >10% . Fasting blood samples for complete blood count ( CBC ) analysis were obtained by venipuncture and collected into tubes containing ethylenediaminetetraacetic acid . Platelet counts and MPV were performed at local laboratories using automated hematology cell counters and standardized quality assurance procedures . Methods used to measure the blood traits analyzed in this study have been described previously for ARIC , CARDIA , JHS , Health ABC , WHI , and GeneSTAR [54]–[58] . Platelet count was reported as 109 cells per liter , and was recorded in all 16 , 388 study participants . Information on MPV was available in a subset of 4 , 612 participants from five COGENT study cohorts ( ARIC , GeneSTAR , Health ABC , HANDLS , and JHS ) and was reported in femto liters ( 10−15 L ) . All the phenotypes were approximately normally distributed and we did not perform any data transformations . Genotyping was performed within each COGENT cohort using methods described in Text S1 . Affymetrix chips were used in the ARIC , CARDIA , JHS , and WHI studies and Illumina chips were used in GeneSTAR , HANDLS , and Health ABC . DNA samples with a genome-wide genotyping success rate <95% , duplicate discordance or sex mismatch between genetic estimates of gender and self-report , SNPs with genotyping failure rate >10% , monomorphic SNPs , SNPs with minor allele frequency ( MAF ) <1% , and SNPs that mapped to several genomic locations were removed from the analyses . Because African-American populations are recently admixed , we did not filter on Hardy-Weinberg equilibrium p-value . Instead , significantly associated SNPs were later examined for strong deviations from Hardy–Weinberg equilibrium and/or raw genotype data was examined for abnormal clustering . Participants and SNPs passing basic quality control were imputed to >2 . 2 million SNPs based on HapMap II haplotype data using a 1∶1 mixture of Europeans ( CEU ) and Africans ( YRI ) as the reference panel . Details of the genotype imputation procedure are described further under Supplemental Methods . Prior to meta-analyses , SNPs were excluded if imputation quality metrics ( equivalent to the squared correlation between proximal imputed and genotyped SNPs ) were less than 0 . 50 . Differences in platelet count may affect platelet function and aggregation [59] . In addition , younger platelets have higher MPV than older platelets and are more reactive [60] . We hypothesized that the genetic variants that determine platelet count and MPV may also affect platelet aggregation . To examine this hypothesis , we used agonist-mediated platelet aggregation assays , which can provide information about the different aspects of platelet aggregation . For these assays , platelet aggregation agonists , such as collagen or ADP , are added to whole blood or platelet-rich plasma and platelet aggregation is measured after a specified amount of time ( 300 seconds ) . We performed platelet aggregation assays in the participants of the GeneSTAR cohort . Blood samples were obtained as described above , and platelet aggregation in whole blood was measured as reported previously [57] . Briefly , in vitro whole blood impedance in a Chrono-Log dual-channel lumiaggregometer ( Havertown , Pa ) was performed after samples were stimulated with arachidonic acid ( 0 . 5 mmol/L , intra-assay CV = 24% ) , collagen ( 5 µg/mL; intra-assay CV = 9% ) , or ADP ( 10 µmol/L; intra-assay CV = 46% ) . Maximal aggregation within 5 minutes of agonist stimulation was recorded in ohms . For all cohorts , genome-wide association ( GWAS ) analysis was performed using linear regression adjusted for covariates , implemented in either PLINK v1 . 07 , R v2 . 10 , or MACH2QTL v1 . 08 [61] , [62] . Allelic dosage at each SNP was used as the independent variable , adjusted for primary covariates of age , age-squared , sex , and clinic site ( if applicable ) . The first 10 principal components were also incorporated as covariates in the regression models to adjust for population stratification ( Text S1 ) . For GeneSTAR , family structure was accounted for in the association tests using linear mixed effect ( LME ) models implemented in R [63] . Although the JHS has a small number of related individuals , extensive analyses have shown that results were concordant using linear regression or LME , after genomic control [19] . Therefore , results are presented for JHS using linear regression . For imputed genotypes , we used dosage information ( i . e . a value between 0 . 0–2 . 0 calculated using the probability of each of the three possible genotypes ) in the regression model implemented in PLINK or MACH2QTL ( for cohorts with unrelated individuals ) or the Maximum Likelihood Estimation ( MLE ) routines ( for GeneSTAR ) . For both platelet phenotypes , meta-analyses were conducted using inverse-variance weighted fixed-effect models to combine beta coefficients and standard errors from study level regression results for each SNP to derive a combined p-value and effect estimate [64] . Study level results were corrected for genomic inflation factors ( λGC ) by incorporating study specific λGC estimates into the scaling of the standard errors ( SE ) of the regression coefficients by multiplying the SE by the square-root of the genomic inflation factor . The inflation factors for all completed analyses are presented in Table S1 . To maintain an overall type 1 error rate of 5% , a threshold of α = 5×10−8 was used to declare genome-wide statistical significance . Between-study heterogeneity of results was assessed by using Cochrane's Q statistic and the I2 inconsistency metric . Meta-analyses were implemented in the software METAL [64] and were performed independently by two analysts to confirm results . To examine whether there were any differences between males and females , sex-specific GWAS were conducted in each cohort . The results for each SNP were pooled and heterogeneity of allelic effects between females and males was examined using the meta-analysis methods as implemented in GWAMA software [65] . To assess whether the loci previously reported to be associated with the platelet phenotypes in Europeans , Japanese , and African Americans were replicated in the COGENT African-Americans , we examined the meta-analysis results for each index SNP in the regions previously reported , including consistency of direction of effect . If the reported index SNP was not significant at p<0 . 05 we examined adjacent SNPs and reported the closest SNP with p<0 . 05 along with its distance from the index SNP . To examine the association of genotype on platelet aggregation in the GeneSTAR cohort , linear mixed models were used with additive models adjusting for age and sex , and taking into account familial correlation between the individuals . | The majority of the variation in platelet count and mean platelet volume between individuals is heritable . We performed genome-wide association studies in more than 16 , 000 African American participants from seven population-based cohorts to identify genetic variants that correlate with variation in platelet count and mean platelet volume . We observed statistically significant evidence ( p-value<5×10−8 ) that 10 genomic regions were associated with platelet count and 3 were associated with mean platelet volume . Of the regions that were significantly associated , we found 5 novel regions that were not reported previously in other populations . Three of these 5 regions were also associated with platelet count in European Americans and Hispanic Americans . All these regions contain genes that are either known to have or potentially may have a role in determining platelet count and/or mean platelet volume . We further found that one of these regions was also associated with agonist-induced platelet aggregation . Further studies will determine the exact role played by these genomic regions in platelet biology . The knowledge generated by this and other studies will not only help us better understand platelet biology but can also lead us to the discovery of new anti-platelet drugs . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"medicine",
"genetics",
"genetics",
"and",
"genomics",
"biology",
"hematology"
] | 2012 | A Meta-Analysis and Genome-Wide Association Study of Platelet Count and Mean Platelet Volume in African Americans |
Mutations in the neuron-specific α3 isoform of the Na+/K+-ATPase are found in patients suffering from Rapid onset Dystonia Parkinsonism and Alternating Hemiplegia of Childhood , two closely related movement disorders . We show that mice harboring a heterozygous hot spot disease mutation , D801Y ( α3+/D801Y ) , suffer abrupt hypothermia-induced dystonia identified by electromyographic recordings . Single-neuron in vivo recordings in awake α3+/D801Y mice revealed irregular firing of Purkinje cells and their synaptic targets , the deep cerebellar nuclei neurons , which was further exacerbated during dystonia and evolved into abnormal high-frequency burst-like firing . Biophysically , we show that the D-to-Y mutation abolished pump-mediated Na+/K+ exchange , but allowed the pumps to bind Na+ and become phosphorylated . These findings implicate aberrant cerebellar activity in α3 isoform-related dystonia and add to the functional understanding of the scarce and severe mutations in the α3 isoform Na+/K+-ATPase .
Dystonia is a movement disorder characterized by involuntary sustained or repetitive muscle contractions , causing twisting movements and abnormal postures [1 , 2] . It is usually caused by head injuries , drug side effects , metabolic insult , or genetic alterations , and is thought to involve the neuroanatomic circuitry of the basal ganglia , sensorimotor cortex , brainstem , and cerebellum [3] . While most dystonias are idiopathic , some are familial , and modifications of more than 25 designated genes associated with dystonia ( DYTs ) have been described [4] . Several mutations in the ATP1A3 ( DYT12 ) gene , encoding the neuron-specific α3 isoform of the Na+/K+-ATPase , can cause rapid-onset dystonia-parkinsonism ( RDP ) characterized by an abrupt onset of dystonia and parkinsonian motor-related features [5] , or alternating hemiplegia of childhood ( AHC ) characterized by fluctuating spells of tonic , dystonic , hemiplegic and oculomotor abnormalities [6–8] . A separate mutation in ATP1A3 is responsible for the Cerebellar ataxia , Areflexia , Pes cavus , Optic atrophy , and Sensorineural hearing loss ( CAPOS ) syndrome [9] . All three ATP1A3 disorders are typically triggered by environmental and/or physiological events such as physical exhaustion , temperature changes , emotional stress , or infections , pointing to a broad spectrum of often distinct , yet overlapping , neurological disorders [10] . Intriguingly , missense mutations that cause different amino acid substitutions at the same position in the ATP1A3 gene can cause distinct diseases . A prominent example is amino acid position 801 , where different mutations cause RDP or AHC ( D801Y ) [5 , 11 , 12] or AHC ( D801N , D801E and D801V ) [6 , 13–16] . That position 801 is a hotspot for disease-causing mutations correlates with its crucial role in Na+/K+-ATPase pump function; the aspartate residue there is conserved in all Na+/K+-ATPase isoforms of all animal species , where it alternately coordinates both K+ ions [17 , 18] and two of the three Na+ ions transported [19 , 20] , and is required for enclosure of the K+-ions [21] . In the present study we show that a mouse model with the D801Y disease-mutation ( α3+/D801Y mice ) [22] displayed severe hypothermia-induced dystonia , which correlated with abnormal cerebellar neuronal activity in vivo . In vitro pump characterization revealed that D-to-Y mutant pumps failed to carry out Na+/K+ exchange , but retained the ability to bind Na+ . These data thus provide a heretofore unknown link between hypothermia and dystonia that implicates aberrant cerebellar activity in α3 isoform-related dystonias and provides functional insight into the disease-causing effects of the underlying Na+/K+-ATPase dysfunction .
As the abrupt onset of dystonia in both RDP and AHC patients usually occurs in response to a stressful environmental or physiological event [23] , we subjected α3+/D801Y mice to a variety of such conditions ( Fig 1A ) . However , stress tests that included restraining , tail suspension , randomly timed electric foot shocks , exposure to fox urine , hyperthermia resulting in elevation of body temperature to 40 . 4 ± 0 . 3°C , and forced swimming in warm 35°C water did not result in genotype-specific abnormal symptoms . Even a 2-week chronic unpredictable stress protocol failed to provoke symptoms . In contrast , forced swimming in 5–10°C cold water for as little as 4 min consistently caused severe dystonia-like postures with hyperextended limbs ( Fig 1Bleft ) , and long periods of body twisting and convulsion-like movements ( Fig 1Bright ) in the α3+/D801Y mice ( Fig 1A and S1 Video ) . These symptoms were never observed in WT littermates ( Fig 1Bleft and Supplementary movie 1 ) . The attacks were pulsating in severity and lasted on average 39 . 2 ± 2 . 5 min ( n = 6 ) , after which α3+/D801Y mice fully recovered , with no apparent residual or persisting symptoms . As a result of the 4 min cold-water swim , core body temperature dropped significantly to about 20°C in both α3+/D801Y mice and WT littermates ( α3+/D801Y: 36 . 5 ± 0 . 2°C to 19 . 5 ± 0 . 6°C ( p<0 . 0001 , two-way ANOVA with genotype ( WT versus α3+/D801Y ) and condition ( baseline versus hypothermia ) as main factors followed by Tukey's multiple comparisons test ) ; WT: 36 . 5 ± 0 . 1°C to 19 . 8 ± 0 . 2°C ( p<0 . 0001 ) ) ( Fig 1C ) without any difference in either baseline or post swim body temperature between α3+/D801Y and WT mice ( p>0 . 9999 and p = 0 . 9873 , respectively ) . Since only swimming in cold , and not in temperate water resulted in attacks , we speculated that lower body temperature , rather than the stress of forced swimming , was causative for the attacks . To address this , mice were placed in a freezing cold , -20°C , environment and kept there until attacks developed , or until core body temperature dropped below 20°C . Remarkably , all six α3+/D801Y mice ( but none of the six WT mice ) developed identical attacks to those caused by the cold-water swim ( Fig 1A ) as their body temperature dropped below 20°C ( Fig 1C ) , strongly indicating that lowered body temperature alone induced the attacks . To further rule out a stress aspect of the attacks , α3+/D801Y mice were treated with the alpha-adrenergic blocker prazosin before being subjected to cold-water swim . All prazosin-treated α3+/D801Y mice still developed attacks ( Fig 1A ) with durations and severity identical to those untreated . If hypothermia indeed caused the attacks , rewarming the animal during an attack might be expected to reduce the duration and/or severity of the attack . Indeed , placing α3+/D801Y mice on a 33 . 3°C heating pad immediately attack symptoms began , after induction by a cold-water swim , significantly diminished the duration of the attacks compared to the average time for recovery at room temperature ( heating pad: 11 . 8 ± 0 . 5 min; room temperature: 39 . 2 ± 2 . 5 min ( p = 0 . 0002 , paired t-test ) ) ( Fig 1D ) . To explore the electrophysiological nature of these attacks , electrocorticographical ( ECoG ) recordings and hind limb electromyography ( EMG ) recordings were performed on five α3+/D801Y mice before ( baseline ) and during hypothermia-induced attacks . ECoGs were recorded bilaterally from the primary motor cortex in freely moving animals ( Fig 2A and 2B ) . None of the α3+/D801Y mice showed any epileptic activity during the attacks , and there was no evidence of genotype-related differences in ECoG signals or power spectra between baseline recordings and recordings during attacks ( Fig 2C and 2D , respectively ) . As an experimental control , seizures were induced using a lithium-pilocarpine protocol , which resulted in the expected dramatically different ECoG activity and corresponding power spectra ( Fig 2E ) . Simultaneous EMG recordings from the anterior tibialis and gastrocnemius muscles of the hind limb ( Fig 2F ) revealed a pronounced increase in co-contraction of these muscles during the attacks compared to baseline ( Fig 2G and 2H ) , characteristic of dystonia . This was quantified with representative cross correlograms ( Fig 2I ) . From the absence of epileptic activity in ECoG , and the evidence for co-contractions in the EMG recording , we conclude that the attacks were dystonic in nature . Besides dystonia , patients also suffer from ataxia and other motor-related features . Thus , α3+/D801Y mice and WT littermates were subjected to motor tests . Although the α3+/D801Y mice exhibited normal body posture , normal gait ( p = 0 . 8919 fore base width , p = 0 . 5428 hind base width , and p = 0 . 1856 stride length , t-test ) ( Fig 3A ) , and an absence of hind limb clasping ( p = 0 . 8981 , t-test ) ( Fig 3B ) , we did find that α3+/D801Y mice performed considerably worse than WT littermates in more challenging and stressful motor tests . On the balance beam , α3+/D801Y mice needed more time to cross ( p<0 . 0001 day 1 , p = 0 . 0003 day 2 , and p = 0 . 0335 day 3 , two-way ANOVA with genotype ( WT versus α3+/D801Y ) and time ( days ) as main factors followed by Tukey's multiple comparisons test ) , and had more foot slips ( p<0 . 0001 day 1 , p<0 . 0001 day 2 , and p<0 . 0001 day 3 ) when compared to WT littermates over three consecutive days ( Fig 3C ) ( Supplementary movie 2 ) . The α3+/D801Y mice also took significantly more time to climb a vertical rope than WT littermates ( p<0 . 0001 , t-test ) ( Fig 3D ) , and they exhibited a higher ataxia ratio in the parallel rod floor test ( p = 0 . 0032 , t-test ) ( Fig 3E ) , but their grip strength was comparable to that of WT littermates ( p = 0 . 1358 , t-test ) ( Fig 3F ) . Purkinje cells are thought to express solely the α3 isoform , and not the otherwise ubiquitously-expressed α1 isoform [24 , 25] , and are therefore suggested to be particular highly sensitive to modifications of α3 isoform function [26 , 27] . We therefore investigated whether aberrant cerebellar function could be the cause of the observed motor deficits and inducible dystonia . α3+/D801Y mice express around 80% of the α3 isoform protein compared to WT mice in cerebellum ( p = 0 . 0034 for p0 and p = 0 . 0010 for p70 , t-test ) , and likely as a compensation mechanism , α1 isoform levels are slightly higher in α3+/D801Y mice compared to WT littermates ( p = 0 . 0113 for p0 and p = 0 . 0140 for p70 , t-test ) , both at birth and in adulthood ( Fig 4A ) ( Full length Western blots shown in Supplementary S1 Fig ) . To specifically address the expression of α1 and α3 isoforms in Purkinje cells , immunofluorescence staining of cerebellar slices with antibodies against the α1 and α3 isoforms was studied ( Fig 4B ) . WT Purkinje cells showed no expression of the α1 isoform , whereas strong staining of the α3 isoform was noted in all Purkinje cells . Notably , the same pattern of absence of expression of the α1 isoform was observed in Purkinje cells in the α3+/D801Y mice , suggesting that mechanisms to compensate for loss of α3 isoform activity are lacking in Purkinje cells . Loss of cerebellar neurons has been reported in some patients with ATP1A3 mutations [28 , 29] . We therefore compared gross cerebellar morphology in the α3+/D801Y mice and WT littermates . No visible loss of cerebellar mass was evident ( Fig 4C ) and cerebellar morphology appeared normal in stained slices from both α3+/D801Y mice and control WT littermates ( Fig 4D ) . Furthermore , quantification of Purkinje cells positive for calbindin revealed comparable numbers in α3+/D801Y mice and in WT littermates ( Fig 4E ) ( p = 0 . 59 , t-test ) . Next , to investigate if cerebellar neuronal activity was affected , we performed in vivo single-unit extracellular recordings of Purkinje cells ( Fig 5A ) in awake head-restrained α3+/D801Y mice and WT littermates ( representative raw traces are in Fig 5B ) . Comparing α3+/D801Y mice with WT littermates under baseline conditions , α3+/D801Y Purkinje cells exhibited the same mean firing rate as WT ( α3+/D801Y: 62 ± 6 sp/s; WT: 60 ± 5 sp/s ( p = 0 . 9923 , two-way ANOVA with genotype ( WT versus α3+/D801Y ) and condition ( baseline versus hypothermia ) as main factors followed by Tukey's multiple comparisons test ) ) ( Fig 5Cupper ) and their predominant firing rates were also comparable ( α3+/D801Y: 83 ± 7 sp/s; WT: 76 ± 7 sp/s ( p = 0 . 9693 ) ) ( Fig 5Cmiddle ) . However , when investigating the regularity of the firing pattern using the coefficient of variation of interspike intervals ( CV ISI ) , defined as the standard deviation of ISIs/mean ISI , α3+/D801Y mice exhibited a slightly , but significantly , higher CV ISI compared to WT littermates ( α3+/D801Y: 0 . 80 ± 0 . 05; WT: 0 . 53 ± 0 . 04 ( p = 0 . 0377 ) ) ( Fig 5Clower ) . Moreover , numerous short pauses were evident in the raw recordings from α3+/D801Y mice under baseline conditions , which contribute to the increased irregularity compared to WT ( Fig 5B , blue trace ) . Next we recorded from α3+/D801Y mice undergoing dystonic attacks induced by a 4 min cold-water swim , and also from control WT littermates after identical cold-water exposure . Mean firing rate of the Purkinje cells was unaltered in dystonic α3+/D801Y mice ( dystonic α3+/D801Y: 55 ± 6 sp/s; control WT: 53 ± 4 sp/s ( p = 0 . 9966 ) ) ( Fig 5Cupper ) . However , in contrast to α3+/D801Y mice at baseline , dystonic α3+/D801Y mice exhibited high-frequency burst-like firing episodes of 40–80 ms in length , which occurred episodically throughout the whole duration of the induced dystonic attacks . This was evident from the raw traces ( Fig 5B , green trace ) , as well as from the significantly higher predominant firing rate compared to control mice ( dystonic α3+/D801Y: 155 ± 21 sp/s , with individual cells as high as 367 sp/s; control WT: 80 ± 7 sp/s , with individual cells as high as 144 sp/s ( p = 0 . 0002 ) ) and higher CV ISI ( dystonic α3+/D801Y: 1 . 09 ± 0 . 11; control WT: 0 . 60 ± 0 . 05 ( p<0 . 0001 ) ) . The predominant firing rate and CV ISI for α3+/D801Y mice during dystonia were also significantly higher compared to the same α3+/D801Y mice at baseline ( p = 0 . 0002 and p = 0 . 0152 ) , whereas there was no significant difference in these parameters between WT mice exposed to the cold water ( control WT ) and baseline WT mice ( p = 0 . 9959 and p = 0 . 8977 ) demonstrating that the abnormal Purkinje cell activity in dystonic α3+/D801Y mice depends on the presence of the mutant D801Y α3 isoform and is not merely a response of all mice to hypothermia . Purkinje cells form the sole output from the cerebellar cortex and make strong inhibitory synaptic connections onto the deep cerebellar nuclei ( DCN ) neurons , effectively modulating their activity . As the DCN provide the main cerebellar output , we next recorded DCN neurons ( Fig 5D ) to explore if cerebellar output was altered in α3+/D801Y mice ( representative raw traces are shown in Fig 5E ) Like the Purkinje cells , DCN neurons in α3+/D801Y mice at baseline conditions exhibited no alteration in mean firing rate ( α3+/D801Y: 41 ± 4 sp/s; WT: 47 ± 5 sp/s ( p = 0 . 7693 ) ) or predominant firing rate ( α3+/D801Y: 71 ± 8 sp/s; WT: 55 ± 6 sp/s ( p = 0 . 752 ) ) ( Fig 5Fupper and middle ) . However , α3+/D801Y DCN neurons did exhibit a significant higher CV ISI ( α3+/D801Y: 0 . 81 ± 0 . 06; WT: 0 . 52 ± 0 . 03 ( p = 0 . 0297 ) ) ( Fig 5Flower ) , indicating that cerebellar output was more irregular in α3+/D801Y mice compared to WT littermates . In dystonic α3+/D801Y mice , DCN neurons exhibited periods with high-frequency burst-like firing similar to our observations in Purkinje cells , with the same duration and episodic nature . Their mean firing rate was comparable to that of control WT littermates ( dystonic α3+/D801Y: 50 ± 6 sp/s; control WT: 37 ± 4 sp/s ( p = 0 . 2603 ) ) ( Fig 5Fupper ) , but their predominant firing rate ( dystonic α3+/D801Y: 157 ± 19 sp/s with individual cells as high as 296 sp/s; control WT: 53 ± 7 sp/s with individual cells as high as 105 sp/s ( p<0 . 0001 ) ) and CV ISI ( dystonic α3+/D801Y: 1 . 2 ± 0 . 11; control WT: 0 . 62 ± 0 . 06 ( p<0 . 0001 ) ) were both significantly higher than those of control WT littermates ( Fig 5Fmiddle and lower ) . To elucidate the molecular mechanistic consequences of the RDP/AHC-causing D801Y mutation and to compare it to the AHC-causing D801N mutation , Na+/K+-ATPase-mediated currents were recorded in oocytes expressing Xenopus laevis orthologs of α subunit ATP1A1 , encoding either the homologous WT aspartate , D813 , or the D813Y or D813N mutation that are equivalent to WT D801 , D801Y and D801N , respectively , in the human and rodent α3 isoform , in combination with ATP1B3 β subunit . Xenopus ATP1A1/ATP1B3 pumps were studied because these are believed to be the native isoforms in Xenopus oocytes [30 , 31] , an established system for high-resolution measurements of Na+/K+ pump function . Moreover , because the D801-equivalent aspartate is absolutely conserved in all Na+/K+-ATPase α subunits of all species and plays a crucial role in K+-ion binding , the D-to-Y and D-to-N substitutions may be expected to cause comparable disruptions of function in all Na+/K+-ATPase isoforms . Unlike WT Na+/K+ pumps ( Fig 6A , 6D and 6G ) , neither D-to-Y ( Fig 6B , 6E and 6H ) nor D-to-N ( Fig 6C , 6F and 6I ) mutant Na+/K+-ATPases were able to generate the outward ( positive ) current on exposure to high external K+ ( K+o ) that signifies normal electrogenic extrusion of 3 Na+i in exchange for import of 2 K+o in each ATPase transport cycle . D-to-Y or D-to-N mutant Na+/K+-ATPases similarly failed to demonstrate any K+o-activated outward current when competing external Na+ was absent ( Supplementary S2 Fig ) . However , in the presence of external Na+ but absence of K+o , thereby precluding Na+/K+ exchange even in WT Na+/K+ pumps , both D-to-Y ( Fig 6K and 6N ) and D-to-N ( Fig 6L and 6O ) Na+/K+-ATPases , like WT Na+/K+-ATPase ( Fig 6J and 6M ) , generated robust transient currents in response to membrane potential jumps . Those ouabain-inhibited Na+ charge movements reveal the time course of the major conformational change of Na+-bound phosphorylated Na+/K+-ATPases that , in one direction encloses the three Na+ ions and , in the other , releases them to the cell exterior [32] . Also like WT Na+/K+-ATPases ( Fig 6G and 6J ) , both D-to-Y ( Fig 6H and 6K ) and D-to-N ( Fig 6I and 6L ) Na+/K+-ATPases still generated the small steady inward currents at large negative potentials that reflect pump-mediated import of protons [32]; in WT , that proton current is seen clearly only without K+o ( Fig 6G ) , when overlapping outward Na+/K+ exchange current was absent , but in D-to-Y ( Fig 6H ) and D-to-N ( Fig 6I ) , which are both incapable of generating outward Na+/K+ exchange current , that inward current was evident with or without K+o . Indeed , because these mutant Na+/K+-ATPases are incapable of tightly binding external K+ , they are effectively permanently trapped in the phosphorylated conformations that reversibly release the three bound Na+ one at a time to the extracellular medium; it is precisely those conformations that carry out proton import . Importantly , not all effects of D-to-Y and D-to-N mutations were identical under all conditions , with Fig 6H and 6I ( Supp . S2H and S2I Fig ) suggesting that D-to-Y Na+/K+-ATPases possibly support larger proton influx than D-to-N Na+/K+-ATPases at negative resting potentials .
In this study , we found that the α3+/D801Y mouse model exhibited prolonged episodes of hypothermia-induced dystonia , which usually began abruptly with hyper-extension of limbs and developed into abnormal postures and twisting movements , characteristic of dystonia [1 , 2 , 33] . To our knowledge , hypothermia has not previously been shown to trigger dystonia using animal models . Nevertheless , that altered temperature can induce phenotypes in the α3+/D801Y mice is not completely unexpected , as several clinically reported triggers for RDP , AHC , and CAPOS , involve a change in body temperature [8 , 10] . Prolonged exercise , alcohol consumption , and fever , all of which raise body temperature , are among the most frequently reported triggers of RDP [23] . Likewise , CAPOS can be induced by fever [9] . Furthermore , exposure to both cold and warm temperatures has been reported to trigger attacks in AHC patients [10 , 15] . Nevertheless , given that hyperthermia seems to be a common trigger in ATP1A3 patients , we also tested this in the α3+/D801Y mice by exposing them to a heated environment that raised their body temperature to 40 . 4 ± 0 . 3°C , a physiologically relevant fever level . However , this experimentally induced hyperthermia failed to provoke any attacks in the α3+/D801Y mice despite the facts ( i ) that it induced symptoms of hyperthermia , also seen in WT littermates , including immobility or circling [34] , and ( ii ) that 10- to 14-day-old WT mice have been reported to undergo febrile seizures at body temperatures averaging 41 . 3°C [34] . We found that both Purkinje cells and their synaptic target , the DCN neurons , fire significantly more irregularly in the α3+/D801Y mice compared to WT animals , a finding that correlates with observed motor deficits in the α3+/D801Y mice , which are similarly found in all ATP1A3-related disorders [8] . The irregular firing was further exacerbated , and evolved into abnormal high-frequency burst-like firing , when dystonia was induced in the α3+/D801Y mice , similar to changes in cerebellar activity noted during dystonia induced in WT mice by brain perfusion with low-dose ouabain [35 , 36] . In further support of cerebellar involvement in α3 isoform-related dystonia , heterozygous α3 knock-out mice developed increased symptoms of dystonia after cerebellar perfusion with the excitatory glutamatergic agonist , kainate , and they exhibited enhanced inhibitory neurotransmission in the cerebellar cortex compared to WT mice [24] . Why exactly the cerebellum appears to be highly susceptible to dysfunction upon alterations of α3 isoform activity remains to be firmly established . But a likely explanation is that Purkinje cells , as also shown here , express only the α3 isoform of the Na+/K+ pump and not the otherwise ubiquitously expressed α1 isoform [24 , 25] , and also do not appear to be able to express the α1 isoform as a compensatory response to α3 isoform dysfunction . This special characteristic of Purkinje cells is supported by the finding that shRNA-mediated knock-down of the α3 isoform led to disruption of the intrinsic firing of Purkinje cells , but not of DCN neurons which express both α1 and α3 isoforms , when synaptic inputs were inhibited in vitro [27] . How hypothermia affects the firing of the cerebellar neurons and induces dystonia in the α3+/D801Y mice remains to be elucidated . An earlier study found that the intrinsic activity of Purkinje neurons in culture was increasingly slowed as the cells were cooled below 20°C , with the duration of action potentials increasing as their frequency decreased [37] . This temperature range correlates with the induction of attacks in the α3+/D801Y mice we observed when body temperature fell to 20°C . Furthermore , the spread of the ISI became larger in the cultured Purkinje cells as the temperature was lowered [37] , echoing the increase in CV ISI during hypothermia-induced dystonia in the α3+/D801Y mice . In cerebellar slices , the firing of Purkinje cells was furthermore shown to be particular affected by lowering the temperature [38] , suggesting that Purkinje cells are highly temperature sensitive . Maintenance of normal intrinsic activity of Purkinje cells depends on function of their α3 isoform Na+/K+-ATPases [26 , 27] . Although α3 isoform Na+/K+-ATPases in WT mice are presumably slowed by low temperature , the complete loss of Na+/K+ exchange by α3 isoform Na+/K+-ATPases bearing the D-to-Y mutation in cerebellar Purkinje neurons of α3+/D801Y mice may be expected to impair their ability to adequately sustain electrical activity , compared to WT mice , at similarly low body temperatures . Clinically , AHC and RDP have been considered to be distinct disorders , although with overlapping features [8] . The same appears true for mice , as the α3+/D801Y mice in some aspects phenotypically differ from Mashlool ( α3+/D801N ) mice [39] and Myshkin ( α3+/I810N ) mice [40] that are heterozygous for the AHC mutations , D801N and I810N , respectively ( Table 1 ) . Both Mashlool and Myshkin mice exhibited spontaneous recurrent tonic clonic seizures [39 , 40] . In contrast , although α3+/D801Y mice have a lowered threshold for PTZ-induced seizure , they did not develop spontaneous seizures [22] . Mashlool mice , furthermore , effectively modeling AHC closely by exhibiting hemiplegic episodes of relatively long duration , but also short dystonic spells , upon water exposure [6 , 39] . Here we show that α3+/D801Y mice suffer from abrupt hypothermia-inducible dystonia that recapitulates the abrupt triggerable onset of symptoms in ATP1A3 patients [41] . These induced , and EMG-confirmed , dystonic attacks lasted noticeably longer than those observed in Mashlool mice ( α3+/D801Y: 39 . 2 ± 2 . 5 min; Mashlool ( α3+/D801N ) : 0 . 07 ± 0 . 005 min ) [39]; however , the dystonic attacks in α3+/D801Y mice were not persistent as in RDP patients [23] , and could thus resemble spells noted in AHC patients , except that the induced attacks did not include EMG hemiplegia activity or ECoG seizure activity that could be expected of such episodic spells of AHC patients [8] . This possibly reflects the intermediate nature of the D801Y mutation that has been associated with RDP and AHC [5 , 11 , 12] . To further explore how different mutations might cause this spectrum of neurological features , we made side-by-side comparisons of the effects of the D801Y and D801N mutations on the function of otherwise identical Na+/K+-ATPase isoforms , in the same cells and under the same conditions . We show that the D-to-Y and D-to-N mutations both abolish Na+/K+ exchange , in accordance with the known crucial role of this conserved aspartate residue to coordinate both transported K+ ions in the pump binding sites [19 , 20] , and with previous reports that the D801N mutation abolishes K+ occlusion [21] and K+-activated outward current [42] , and reduced forward cycling [43] . Nevertheless , preservation of Na+/K+-ATPase-mediated transient Na+ charge movements in both D-to-Y and D-to-N Na+/K+-ATPases demonstrates that both mutants remain capable of binding 3 Na+i ions , of consequent Na+-dependent phosphorylation , and of essential Na+-releasing and -rebinding conformational changes [32] , as previously inferred for D801N from ouabain-binding measurements [44] . Intriguingly , the Na+-bound phosphorylated conformations that are shown here to be preserved in the D-to-Y and D-to-N Na+/K+-ATPases , and in which their inability to bind K+ ions dooms these mutants to spend most of their time , comprise precisely those conformations that support pump-mediated proton import [32] . In other words , whereas D-to-Y and D-to-N mutants must both be viewed as loss-of-function in terms of Na+/K+ exchange , they must both be considered gain-of-function in terms of their exclusive occupancy of proton-importing pump states . As the failure of Na+/K+ exchange by the D-to-Y and D-to-N mutations demonstrated here in Xenopus oocytes at room temperature under optimal [K+]o , [Na+]i , and [ATP] conditions is expected to be recapitulated in D801Y and D801N α3 isoform Na+/K+-ATPases under physiological conditions in mice , then α3+/D801Y mice and Mashlool ( α3+/D801N ) mice should both be effectively haploinsufficient in terms of α3 isoform Na+/K+-ATPase-mediated Na+ extrusion . However , we found not all effects of D-to-Y and D-to-N mutations to be identical under all conditions , suggesting that D-to-Y Na+/K+-ATPases possibly support larger proton influx than D-to-N Na+/K+-ATPases at negative resting potentials , like those of neurons . Whether phenotypical differences attributed to D801Y versus D801N mutation in α3 isoform pumps , such as clinical diagnoses of RDP rather than AHC in patients , and phenotypical differences between α3+/D801Y and Mashlool mice , reflect these relatively small functional differences cannot be concluded at present , and this requires further investigation . The conclusion that both D801Y and D801N are loss-of-function mutations for Na+/K+ exchange , but are effectively gain-of-function in that the mutant pumps engage full time in proton import , possibly accounts for the scarcity and severity of these de novo gene alterations and the apparent lack of nonsense mutations ( resulting in haploinsufficiency ) in the ATP1A3 patient group [6 , 10] . If this inference is correct , it has significant implications for future therapeutic intervention in ATP1A3-related diseases in which α3 isoform haploinsufficiency may be less damaging than carrying certain missense gain-of-function mutations . Thus , strategies such as exon skipping and genome editing may be selected to eliminate the mutated disease-causing allele in cases where it is considered more deleterious than loss-of-function alleles .
All experiments were performed on 8–16 week old α3+/D801Y mice and WT littermates on a C57/BL6JRj ( Janvier ) background except the electrophysiological in vivo recordings , which were performed on the C57BL/6JR ( Jackson laboratory ) background . Mice were kept at a daily 12 hour light/dark cycle . Male and female mice were included in balanced numbers . Experimental animal protocols performed at Aarhus University were performed according to the Danish national and Institutional regulations and approved by the Animal Experiments Inspectorate under the Danish Ministry of Justice ( permit numbers 2012-15-2934-00621 , 2013−15−2934−00815 and 2014−15−2934−01029 ) . Experimental animal protocols performed at Albert Einstein College of Medicine were done according to the animal guidelines set by Einstein's Institutional Animal Care and Use Committee . Each condition was tested in at least 5 mice of each genotype and repeated 3 times per mouse . After being subjected to each condition , the mice were place on a cleared table or empty cage where they were video recorded and closely monitored by the experimenter for the occurrence of an attack . Average occurrence of attacks ( % ) and attack duration were subsequently calculated per mouse , per condition . Restraining . Mice were restrained in a 60 mL falcon tube for 10 min . The back of the tube was partially closed off but allowed enough air to prevent suffocation . Tail hanging . Mice were hung from their tail for 6 minutes . Foot shocks . Mice were placed in a custom made plastic box with metal wiring on the bottom connected to a stimulating device . Randomly timed short shocks were given for 5 min . Exposure to fox urine . Mice were put in a closed cage with a tube containing a tissue drenched in fox urine for 10 min . Temperate water swim . A clear plastic box ( 42 x 26 x 18 cm ) was filled with 35°C water to a water height that forced mice to swim without the possibility to touch the box floor . Mice were placed in the water and forced to swim for 10 min . Chronic variable stress . The mice were subjected to one of three different stressors over the course of two weeks . Mice were suspended by their tail for 6 minutes on days 1 , 6 and 9 . The mice were placed in glass jars with 30°C water for 6 minutes on days 2 , 8 and 11 . Finally , the mice were restrained for 60 minutes using DecapiCones ( BrainTree Scientific , Inc . , MA , USA ) on days 3 , 7 and 10 . Cold water swim . A clear plastic box ( 42 x 26 x 18 cm ) was filled with 5–10°C cold water to a water height that forced mice to swim without the possibility to touch the box floor . Mice were placed in the water and forced to swim for 4 min . Mice that showed difficulty staying afloat before the 4 min mark were immediately removed from the water . Rectal body temperatures were measured just after the animals were removed from the water . Cold environment . Mice were placed in an empty clear plastic box and placed in a -20°C environment until they displayed attacks or their body temperature reached <20°C as measured by rectal body temperature . Elevation of body temperature . Mice were placed in a large glass beaker placed in a 43°C warm incubator for 15 min . Rectal body temperature was measured after the animals were removed from the incubator . Mice were anesthetized using isoflurane ( 4% in 1 . 5 L/min O2 for induction and 1–2% in 1 . 5 L/min O2 for maintenance ) after which the skull was exposed , cleaned and treated with OptiBond All-In-One ( Kerr Corporation; Orange , CA , USA ) in order to ensure adhesion of the light cured hybrid composite Charisma ( Heraeus Kulzer; Hanau , Germany ) to later attach implants . In case of preparation for ECoG recordings , four small holes ( 0 . 5 mm in diameter ) were subsequently drilled for implantation of the ECoG electrodes; two above bilateral primary motor cortices ( +1 mm AP and ± 1 mm ML relative to bregma ) and two in the interparietal bone ( -1 mm AP and ± 1 mm ML relative to lambda ) to accommodate the ground and reference electrodes ( Fig 2A ) . Teflon coated silver ball-tip electrodes ( ~200 μm ) , attached to an ECoG headmount ( Pinnacle Technology; Lawrence , KS , USA ) , were carefully inserted into the holes and fixed in place using the hybrid composite Charisma . The rest of the skull was subsequently covered with hybrid composite Charisma to ensure insulation after which the headmount was attached to the skull . Surgical preparation for EMG recordings was similar to the ECoG procedure . In this case coated stainless steel wires were attached to the headmount and were subcutaneously led to the hind limb . In order to avoid too much pressure on the wires at hind limb level , the wires were sutured to a patch of skin in the back and hip before attaching them to the hind limb muscles . The end of the wires were partially stripped and carefully stitched to the anterior tibialis and gastrocnemius muscles of the right hind limb . Ground and reference electrodes were made shorter than the others , partially stripped and left loose at the level of the hip . For in vivo electrophysiological recordings a metal bracket was attached to the front part of the skull using light cured hybrid composite Charisma . A recording chamber was constructed on top of cerebellum with dental cement and the cavity was filled with silicone . Following surgery all animals were given flunixin ( 2 . 5 mg/kg ) and 500 μl saline , after which they were allowed at least four days to recover before being subjected to experimental procedures . All animals were monitored closely for any complications on a daily basis after surgery . Mice were transferred to the test room 1 hour prior to testing for acclimation . Behavioral apparatus were cleaned between tests in 70% EtOH . Gait analysis . Front and hind paws where painted with wet paint and the mouse was allowed to run across a long sheet of white paper . Fore base width , hind base width and stride length were subsequently estimated . Hind limb clasping . Mice were suspended by their tails for 30 sec . Each mouse was evaluated by video analysis and given a score of either 0 ( no abnormal hind limb movement ) or 1 ( abnormal hind limb movement ) in 10 sec intervals allowing a maximum score of 3 . Abnormal hind limb movement was defined as the retraction of either one or both hind limb towards the midline [45] . Balance beam . Mice were tested for 3 successive days on a 1 m long , 9 mm wide suspended wooden beam . Number of slips and the time to traverse the center 80 cm of the beam was evaluated by video analysis and calculated as the mean of three trials . Rope climb . Mice were tested on a 40 cm long and 10 mm in diameter vertical rope . Time to climb was evaluated by video analysis and estimated as the mean of three trials . Parallel rod floor . Mice were tested in the parallel rod floor apparatus ( Stoelting Co , Wood Dale , US ) for 15 min . Anymaze software ( Stoelting Co , Wood Dale , US ) was used to record number of slips and distance traveled . Grip strength . A grip bar was attached to the grip strength meter ( Bioseb , Vitrolles , France ) to allow measurements of grip strength . A mouse was picked up by its tail and lowered until it grasped the grip bar at which point the mouse was pulled away horizontally until its grip was released . Readouts of grip strength given in grams were normalized to the body weight of the mouse and calculated as the mean of five trials . Cerebellum was dissected and lysed in 10 mM Tris , 150 mM NaCl , 2 mM EDTA with 1% IGEPAL and protease inhibitor ( Roche , Basel , Switzerland ) . Lysates were separated by SDS-PAGE and electro-blotted onto nitrocellulose membranes ( Pharmacia-Amersham , Amersham , UK ) . Primary antibodies: anti α1 1:2000 ( a6f-c , Developmental Studies Hybridoma Bank , US ) , anti α2 1:1000 ( 07674 , EMD Millipore , US ) , anti α3 1:1000 ( 06172 , EMD Millipore , US ) , and actin 1:1000 ( A2066 , Sigma -Aldrich , St . louis , US ) overnight at 4°C . Secondary antibodies: horseradish peroxidase-conjugated pig anti-rabbit and pig anti-mouse 1:2000 ( Dako , Glostrup , Denmark ) . Visualization was done using a LAS 3000 imager ( Fujifilm , Tokyo , Japan ) with Amersham ECL Western Blotting Detection Kit ( GE Healthcare , Buckinghamshire , UK ) as detection reagent . Full-length Western blots are shown in Supplementary S2 Fig . Cryo sections ( 15 μm thickness ) were blocked in 5% donkey serum PBS/Triton X-100 0 . 25% for 1 hour at RT . Primary antibodies ( anti α1 ( 1:400 ) ( a6f-c , Developmental Studies Hybridoma Bank ) ; anti α3 ( 1:300 ) ( 06172 , EMD Millipore , US ) ; anti calbindin ( 1:400 ) ( ab82812 , Abcam , Cambridge , UK ) were applied in 1% donkey serum PBS/Triton X-100 0 . 25% overnight at 4°C . Secondary labelling was done with Alexa Fluor fluorescent-conjugated secondary antibodies ( Alexa Fluor 488 donkey anti rabbit ( A21206 , Life Technologies , Carlsbad , CA , USA ) ; Alexa Fluor 568 donkey anti mouse ( A10037 , Life Technologies , Carlsbad , CA , USA ) ( 1:350 ) in 1% donkey serum PBS/Triton X-100 0 . 25% for 1 hour at RT . Hoechst ( 1:10000 ) ( Life technologies , Carlsbad , CA , USA ) in PBS was used to counterstain the nuclei . Sections were mounted using fluorescence mounting medium ( Dako , Glostrup , Denmark ) and analysed on a LSM510 laser-scanning confocal microscope using a 40x C-Apochromat water immersion objective NA 1 . 2 ( Carl Zeiss , Göttingen , Germany ) . Zen 2011 software ( Carl Zeiss , Göttingen , Germany ) was used for analysis and image capturing . Mutagenesis and expression . The substitution C113Y ( equivalent to C101Y in human α3 ) was introduced into Xenopus Na+/K+-ATPase α1 subunits to render them ouabain resistant so their activity could be isolated after silencing endogenous ouabain-sensitive Xenopus Na+/K+-ATPases by maintained exposure to 1 μM ouabain . C113Y α1 subunits ( here designated “wild type” ) then served as templates for D813Y and D813N mutations ( equivalent to D801Y and D801N , respectively , in human α3 ) ; these cysteine and aspartate residues are conserved in all Na+/K+-ATPase α isoforms of all species . All substitutions were by QuikChange ( Stratagene , California , US ) and were verified by sequencing . cDNAs were transcribed in vitro , and 15–45 ng of Na+/K+-ATPase α1 subunit cRNA was coinjected with 5–15 ng of wild-type Xenopus β3 subunit cRNA into defolliculated Xenopus laevis oocytes , which were incubated at 18°C for 3 days before recording . Solutions and Na+/K+-ATPase electrophysiology . External solutions contained 125 mM NaOH or tetramethylammonium ( TMA ) -OH , 120 mM sulfamic acid , 0 or 15 mM K+-sulfamate , 5 mM BaCl2 , 1 mM MgCl2 , 0 . 5 mM CaCl2 , 10 mM Hepes ( pH 7 . 6 ) , plus 1 μM ouabain to inhibit native Na+/K+-ATPases; osmolality was 250–260 mosmol/Kg . To inhibit C113Y-containing ouabain-resistant Na+/K+-ATPases , 10 mM ouabain was directly dissolved into appropriate external solutions . Before recording , [Na+]i was raised by incubating oocytes for ≥2 h in K+- and Ca2+-free solution , containing 95 mM NaOH , 90 mM sulfamic acid , 10 mM TEACl , 0 . 1 mM EGTA , 5 mM HEPES ( pH 7 . 6 ) ; osmolality ~210 mosmol/Kg . Two-microelectrode voltage clamp was used to record currents at 22–24°C in oocytes expressing ouabain-resistant Na+/K+-ATPases , as previously described [30] . Whole oocyte currents were acquired with an OC-725A amplifier ( Warner Instruments , US ) , filtered at 1 kHz , and sampled at 5 kHz with an 18-bit ITC-18 A/D-D/A board controlled by Patch Master 2 . 20 software ( Instrutech; HEKA , US ) . Currents at potentials between -180 and +60 mV , in 20 mV increments , were elicited by 50-ms voltage steps from the -20 mV holding potential; steady-state currents were determined by averaging over the last 10 ms . Currents generated by ouabain-resistant Na+/K+-ATPases were obtained by subtracting current traces recorded in 10 mM ouabain from those recorded in 1 μM ouabain . Na+ charge movement quantities , ΔQ , were obtained as integrals of the 10 mM ouabain-sensitive transient currents at -20 mV upon termination of each voltage step in 125 mM Na+o , 0 mM K+o solution . Data was analyzed with IgorPro 6 ( WaveMetric , US ) and Origin 7 . 0 ( Origin Laboratory , US ) . Recordings were performed in awake , freely moving animals using a Pinnacle data acquisition system ( Pinnacle Technology; Lawrence , KS , USA ) . They were given 1–2 accommodation sessions to allow them to get used to the experimental setup after which experiments started . 10–30 min of baseline activity was recorded in each mouse before being subjected to a cold-water swim . ECoG or EMG activity was subsequently recorded during the attack that followed until they were fully recovered ( 20–50 min later ) . As a control , tonic-clonic seizures were induced using a Lithium Pilocarpine protocol in the same mice . 18–24 hrs prior to seizure induction , mice were given an IP injection of lithium chloride ( 3 mmol/kg; Sigma-Aldrich; St . Louis , MO , USA ) dissolved in 0 . 9M NaCl after which 60–120 mg/kg of pilocarpine hydrochloride ( dissolved in 0 . 9M saline; Sigma-Aldrich ) was given IP to induce the epileptic seizure . Data were sampled at 400 Hz for ECoG and 600 Hz for EMG recordings and filtered online using a 1–150 Hz ( ECoG ) or 1–250 Hz ( EMG ) band pass filter . Signals were visualized and power spectra and cross-correlations were calculated using a custom written LabView algorithm ( National Instruments; Austin , TX , USA ) . A craniotomy was made over the cerebellum and the mice were head restrained via the implanted metal bracket in the recording setup and given one hour to acclimate before recording was initiated . In order to record from α3+/D801Y mice during dystonic attacks , mice were cooled in 6–8°C cold water for 4 min and immediately head restrained and after which the recording was initiated . As a control , the same experiments were done in wild type littermates . Single-cell activity of Purkinje cells and DCN neurons was recorded extracellularly using a tungsten electrode ( Thomas Recording , 2–3MΩ ) , which was advanced into the cerebellum until either the Purkinje cell layers or DCNs were reached ( All coordinates given from Bregma: Caudal 6 . 45–6 . 55 mm , Lateral ±1 . 2 mm , Ventral 1 . 4–2 . 2 mm for Purkinje cell layers and 3–3 . 5 mm for DCN ) . The constructed recording chamber was filled with saline and used for ground connection . Signals were band-pass filtered ( 200 Hz–20 kHz ) and amplified ( 2000× ) using a custom built amplifier and then digitized ( 20 kHz ) using a National Instruments PCI MIO 16 XEI ( National Instruments Corporation , Austin , US ) . Waveforms were sorted offline using characteristics of the spikes such as amplitude and energy and by principal component analysis ( Offline Sorter , Plexon , US ) . Purkinje cells and DCN neurons were identified by location , characteristic firing rate , and the presence of complex spikes for Purkinje cells . Average firing rate ( spikes per sec , sp/s ) , predominant ( mode ) firing rate ( sp/s ) , and coefficient of variation of interspike intervals ( CV ISI ) were calculated using custom made MATLAB software . Statistical analyses were performed in Graphpad Prism software ( GraphPad Software Inc , La Jolla , US ) . Student’s t-test ( unpaired or paired ) was used to determine significance when comparing two groups . Two-way ANOVAs followed by Tukey's multiple comparisons test were used to find statistically significant differences between three or more groups . Statistical analyses for each experiment are indicated in the result section with corresponding p-values . Differences were considered significant if p<0 . 05 . | The neurological spectrum associated with mutations in the ATP1A3 gene , encoding the α3 isoform of the Na+/K+-ATPase , is complex and still poorly understood . To elucidate the disease-specific pathophysiology , we examined a mouse model harboring the mutation D801Y , which was originally found in a patient with Rapid onset Dystonia Parkinsonism , but recently , also in a patient with Alternating Hemiplegia of Childhood . We found that this model exhibited motor deficits and developed dystonia when exposed to a drop in body temperature . Cerebellar in vivo recordings in awake mice revealed irregular firing of Purkinje cells and their synaptic targets , the deep cerebellar nuclei neurons , which was further exacerbated and evolved into abnormal high-frequency burst firing during dystonia . The development of specific neurological features within the ATP1A3 mutation spectrum , such as dystonia , are thought to reflect the functional consequences of each mutation , thus to investigate the consequence of the D801Y mutations we characterized mutated D-to-Y Na+/K+-ATPases expressed in Xenopus oocytes . These in vitro studies showed that the D-to-Y mutation abolishes pump-mediated Na+/K+ exchange , but still allows the pumps to bind Na+ and become phosphorylated , trapping them in conformations that instead support proton influx . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"&",
"methods"
] | [
"medicine",
"and",
"health",
"sciences",
"swimming",
"pathology",
"and",
"laboratory",
"medicine",
"neurodegenerative",
"diseases",
"neuroscience",
"biological",
"locomotion",
"animal",
"models",
"model",
"organisms",
"physiological",
"parameters",
"signs",
"and",
"sympto... | 2017 | Hypothermia-induced dystonia and abnormal cerebellar activity in a mouse model with a single disease-mutation in the sodium-potassium pump |
Mitochondrion-related organelles , mitosomes and hydrogenosomes , are found in a phylogenetically broad range of organisms . Their components and functions are highly diverse . We have previously shown that mitosomes of the anaerobic/microaerophilic intestinal protozoan parasite Entamoeba histolytica have uniquely evolved and compartmentalized a sulfate activation pathway . Although this confined metabolic pathway is the major function in E . histolytica mitosomes , their physiological role remains unknown . In this study , we examined the phenotypes of the parasites in which genes involved in the mitosome functions were suppressed by gene silencing , and showed that sulfate activation in mitosomes is important for sulfolipid synthesis and cell proliferation . We also demonstrated that both Cpn60 and unusual mitochondrial ADP/ATP transporter ( mitochondria carrier family , MCF ) are important for the mitosome functions . Immunoelectron microscopy demonstrated that the enzymes involved in sulfate activation , Cpn60 , and mitochondrial carrier family were differentially distributed within the electron dense , double membrane-bounded organelles . The importance and topology of the components in E . histolytica mitosomes reinforce the notion that they are not “rudimentary” or “residual” mitochondria , but represent a uniquely evolved crucial organelle in E . histolytica .
Mitosomes and hydrogenosomes are mitochondrion-related organelles and found in a phylogenetically broad range of eukaryotes . Since organisms that possess hydrogenosomes or mitosomes do not cluster together in eukaryote phylogenies , it is suggested that secondary losses and changes in mitochondrial functions have independently occurred multiple times in eukaryote evolution [1] . This view largely agrees to the observation that the components and functions of the mitochondrion-related organelles differ between organisms [1] , [2] . Entamoeba histolytica , a widespread intestinal protozoan parasite [3] , possesses highly divergent mitosomes [4]–[6] . We have previously shown that sulfate activation is compartmentalized in E . histolytica mitosomes [5] . As sulfate activation generally occurs in the cytoplasm or plastids in eukaryotes [5] , [7] , its compartmentalization to mitosomes is unprecedented . Mastigamoeba balamuthi , a free-living amoeba that is distantly related to E . histolytica , also possesses mitochondrion-related organelle . An expressed sequence tags ( EST ) project showed that the organism has enzymes for sulfate activation , and one of the enzymes has the putative mitochondrial targeting signal at the amino terminus ( Jan Tachezy , personal communication ) . Trichomonas vaginalis , Giardia intestinalis , and Cryptosporidium parvum , which also possess mitochondrion-related organelles , i . e . , hydrogenosome and mitosome , apparently lack genes encoding these enzymes . Phylogenetic analyses further revealed that E . histolytica appears to have acquired enzymes involved in sulfate activation from distinct prokaryotic and eukaryotic lineages by lateral gene transfer [5] . Therefore , sulfate activation is not a common function of the mitochondrion-related organelles , but may be a unique feature of a lineage E . histolytica and M . balamuthi belong to . Although iron sulfur cluster biosynthesis is shared by aerobic eukaryotes and highly divergent G . intestinalis mitosomes and T . vaginalis hydrogenosomes [8]–[11] , it still remains to be unequivocally determined whether iron sulfur cluster biosynthesis is exclusively compartmentalized to the mitosomes in E . histolytica and M . balamuthi [12] . Sulfate is generally activated in two steps . Inorganic sulfate is converted to adenosine-5′-phosphosulfate ( APS ) , in a reaction catalyzed by ATP sulfurylase ( AS ) , and further converted to 3′-phosphoadenosine-5′-phosphosulfate ( PAPS ) in a reaction catalyzed by APS kinase ( APSK ) . Pyrophosphate concomitantly produced in the first reaction needs to be decomposed to phosphates by inorganic pyrophosphatase ( IPP ) . PAPS acts as a sulfuryl donor to transfer the sulfuryl moiety to various acceptors by sulfotransferases , resulting into the formation of sulfurylated macromolecules such as mucopolysaccharides , sulfolipids , and sulfoproteins [7] , [13] , [14] . Alternatively , activated sulfate ( APS and PAPS ) is reduced and assimilated into cysteine [13] . In addition , activated sulfate is reduced to sulfide and used as a terminal electron acceptor in the anaerobic respiration in sulfate-reducing bacteria [15] . The E . histolytica genome contains 10 potential genes encoding for sulfotransferases , but lacks the enzymes for sulfate reduction [5] . Consistent with this , activated sulfate is predominantly incorporated in sulfolipids in E . histolytica [5] . Sulfolipids are a class of lipids containing sulfur . Among them , sulfoquinovosyldiacylglycerol and sulfolipid-I were well characterized in plastids and Mycobacterium tuberculoris , respectively [16] , [17] . Sulfoquinovosyldiacylglycerol was shown to be involved in photosynthesis . Sulfolipid-I was identified as a virulence factor in M . tuberculoris . However , the structure and function of sulfolipids in E . histolytica remains largely unknown . While the sulfate activation has been demonstrated as the major metabolic pathway in E . histolytica mitosomes [5] , the physiological role of this mitosome-confined pathway remains unknown . In this study , we attempted to uncover the role of mitosomes in E . histolytica by using the parasites in which genes for the enzymes involved in sulfate activation , MCF , and Cpn60 were knocked down by gene silencing . We showed that these mitosomal proteins are indeed important for sulfolipid production and cell growth . We also demonstrated the localizations and topologies of the enzymes involved in sulfate activation , MCF , and Cpn60 in mitosomes by immunoelectron microscopy .
G3 strain and psAP-2 plasmid were kindly given by David Mirelman , Weisman Institute , Israel [18] . An upstream region of ap-a gene was amplified from psAP-2 using 5′-AGCTCTAGACCGCGGCGGCTTGCTGCACCCTTTG-3′ ( forward primer; SacII site is underlined ) and 5′-CTCTGAGCTCGTTTAAAGGCCT<$>\raster="rg1"<$>CATGATTGTTTGTAAGATAT G-3′ ( reverse primer; SacI , and StuI sites are double- , and broken-underlined , respectively ) . PCR product was digested with SacI and SacII , and ligated into SacI- and SacII-double digested psAP-2 vector to produce psAP-2-Gunma . Approximately 380-450-bp fragments corresponding to the 5′ end of the open reading frame of Cpn60 , Mcf , As , Apsk , and Ipp genes , respectively , were amplified by PCR using the following primers sets ( SacI and StuI sites are single- and double-underlined , respectively ) : 5′-AAAGGCCTATGCTTTCATCTTCAAGTCATT-3′ ( forward ) and 5′-GGGGAGCTCTTTGTAATTTTCTTTAATAC-3′ ( reverse ) for Cpn60; 5′-CATCAGGCCTATGATACAAGGTATGACTTATAAACG-3′ ( forward ) and 5′-ACGCGAGCTCCTAGCAGTACCAAAGAATGTATC-3′ ( reverse ) for Mcf; 5′-CATCAGGCCTATGAGCATTCAAGAAAACTTAAACAAC-3′ ( forward ) and 5′-ACTTGAGCTCGGTCAATTTCAATAGTTCCTGAG-3′ ( reverse ) for As; 5′-GATCAGGCCTATGGCTACTGCTAAGATTGCTG-3′ ( forward ) and 5′-GACTGAGCTCGAGGTGGTGGTTCAACAAATTC-3′ ( reverse ) for Apsk; 5′- CATCAGGCCTATGTCAATTACTTCTATTGTCCCC-3′ ( forward ) and 5′-CACCGAGCTCATCAATTGGATCATTATCTCCAGG-3′ ( reverse ) for Ipp . The PCR fragments were digested with StuI and SacI , and ligated into the StuI- and SacI-double digested psAP-2-Gunma to produce the plasmids used for gene silencing . Lipofection of trophozoites and selection of transformants were performed as previously described [5] . The open reading frame of Cpn60 , Apsk , and Ipp was PCR-amplified with primers containing a BamHI restriction site , digested with BamHI , and ligated into BamHI-digested pCOLD1 to yield pCOLD1-Cpn60 , pCOLD1-APSK , and pCOLD1-IPP , respectively . These constructs were introduced into E . coli BL21 ( DE3 ) cells . Expression and purification of the recombinant proteins were performed as previously described [19] . Briefly , E . coli pellet was suspended in 20 ml of lysis buffer ( 50 mM Tris–HCl , pH 8 . 0 , 300 mM NaCl , and 20 mM imidazole ) containing 1% Triton X-100 ( v/v ) , 100 µg/ml lysozyme , and 25 U/ml benzonase . After 15-min incubation at 4°C , the cells were sonicated on ice and centrifuged at 12 , 000×g for 20 min at 4°C . The supernatant was applied on 50% Ni2+-NTA His-bind slurry ( Qiagen , Tokyo , Japan ) . The recombinant protein-bound resin was washed three times with buffer A ( 50 mM Tris-HCl , pH 8 . 0 , 300 mM NaCl ) containing 20–50 mM of imidazole . The bound proteins were then eluted with buffer A containing 100 mM imidazole . Rabbit anti-Cpn60 , APSK , and IPP antisera were custom made by Operon Biotechnologies ( Tokyo , Japan ) . Whole cell lysates of each gene-silenced strain were analyzed by SDS-polyacrylamide electrophoresis ( PAGE ) and immunoblot analysis as previously described [5] . The dilution of the primary antibodies was 1∶1 , 000 for anti-Cpn60 , anti-APSK , and anti-IPP antiserum , and 1∶100 for anti-CP5 antiserum [5] . The Fast SYBR® Green Master Mix ( AB Applied Biosystems , Foster City , CA , USA ) was used for qRT-PCR . RNA polymerase II gene ( Rnapol ) was used as a house-keeping reference gene . Total RNA was extracted using TRIzol® reagent ( Invitrogen , Carlsbad , CA , USA ) . The synthesis of cDNA was performed using the SuperScript III First-Strand Synthesis System ( Invitrogen ) . qRT-PCR was performed using the following primers: 5′-CCTATGAAAATCGATTGGACATTCTATTGCC-3′ ( forward ) and 5′-GCATCACCAGTAGCAAACTTTGTAACTTG-3′ ( reverse ) for As; 5′-GCCCCAATTGCACCATATCGTGAAATTAG-3′ ( forward ) and 5′-GCACATTGATCAACAGACTTACCAGCAG-3′ ( reverse ) for Apsk; 5′-GATCCTCTTGCTCAAAACCATTACATCTG-3′ ( forward ) and 5′-GTCTAACGCCAATTTTGATAACTTCTTTTGAG-3′ ( reverse ) for Ipp; 5′-GCATGTTTTGATTTTGTTGCTCCATTAGTTCC-3′ ( forward ) and 5′-CACTGACTAATGGAACAACTTTGACAAATCC-3′ ( reverse ) for Mcf; 5′-GATCCAACATATCCTAAAACAACA-3′ ( forward ) and 5′-TCAATTATTTTCTGACCCGTCTTC-3′ ( reverse ) for Rnapol . qRT-PCR was performed using StepOne Plus Real-Time PCR System ( AB Applied Biosystems ) with the following cycling conditions: 95°C for 20 s , followed by 40 cycles of 95°C for 3 s , and 60°C for 30 s . All reactions were run in quadruplicate , including reverse transcriptase-minus and cDNA-minus controls . Quantification for each target gene was determined by the ΔCt method with Rnapol as reference gene . Metabolic labeling was performed as previously described [5] with some modifications . Briefly , approximately 3×105 trophozoites were labeled with [35S]-labeled sulfate ( 25 mCi/mmol ) in 1 mL of the BI-S-33 medium either continuously for 2 , 4 , or 8 h , or labeled for 4 h and chased for 4 or 24 h after removing [35S]-labeled sulfate . Cells were collected and lipids were extracted with 0 . 5 mL of methanol and separated on a silica high-performance thin-layer chromatography plate in 35∶65∶8 ( vol/vol/vol ) methanol/chloroform/28% ( w/w ) ammonium hydroxide [20] . Thin-layer chromatography plates were dried and analyzed by autoradiography . E . histolytica transformants expressing epitope-tagged mitosomal proteins were previously established [5] . Approximately 5×105 trophozoites were resuspended in 2 ml BI-S-33 medium and seeded onto a molybdenum disk ( Nissin EM Co . , JAPAN ) in a well of a 24-well plate . After 15-min incubation at 35 . 5°C , the molybdenum disk that amoebas adhered to was removed and immediately immersed in liquid propane at −175°C . The disk was further fixed and sectioned as previously described [21] . The disk was reacted with primary antibody diluted at 1∶2000 ( anti-Cpn60 antiserum ) and 1∶500 ( anti-HA monoclonal antibody ) in phosphate-buffered saline containing 1 . 5% bovine serum albumin for overnight at 4°C . The samples were then reacted with colloidal gold-conjugated anti-rabbit or anti-mouse secondary antibody ( 1∶20 ) for 1 h at room temperature . Samples were examined by electron microscopy at Tokaii Microscopy . , Inc ( Nagoya , JAPAN ) .
To investigate the role of the mitosomes and , more specifically , mitosome-localized sulfate activation pathway , chaperones , and ADP/ATP transporter , we established the E . histolytica strains in which AS , APSK , IPP , MCF , and Cpn60 genes were knocked down by gene silencing [18] , [22] . These gene-silenced strains were designated ASgs , APSKgs , IPPgs , MCFgs , and Cpn60gs strains , respectively . AS , APSK , and IPP are essential components of the sulfate activation pathway , while MCF transports ADP/ATP across the mitosomal membrane and Cpn60 functions as a mitosome-specific chaperone [5] . Reduction of gene expression of each target gene was verified by qRT-PCR in the gene-silenced strains . The amount of the steady-state transcript of the genes involved in sulfate reduction was reduced by 80 . 4–91 . 8% ( Figure 1A ) . The changes of the level of the transcripts of irrelevant genes ranged 0 . 4–1 . 8 fold of the control , but mostly varied only in the range of 0 . 8–1 . 6 fold . In APSKgs , IPPgs , and Cpn60gs strains , the reduction of each target protein was confirmed by immunoblotting ( Figure 1B ) . Although we observed slight variations in the amount of APSK and Cpn60 in the gene-silenced strains ( e . g . , reduction of Cpn60 protein in MCFgs strain ) , these variations did not correlate with the changes in the transcripts . In APSKgs strain , IPP mRNA level was also slightly decreased , while its protein level remained unchanged . We previously showed that in E . histolytica trophozoites , the majority of activated sulfate are incorporated into sulfolipids [5] . So , we first examined the time course of accumulation of sulfolipids in ASgs , APSKgs , IPPgs , and control mock transformants ( G3 strain transfected with an empty vector ) , by metabolic labeling . As shown in Figure 2A , in the control mock transformant , four major groups of sulfolipids ( I-IV ) were detected by thin layer chromatography , similar to HM1 reference strain as described previously [5] . The amount of sulfolipids ( I–IV ) changed differently in the individual gene-silenced strains . However , the trend of the decrease of each sulfolipid was similar among the strains: II and III were highly affected , whereas I and IV were not affected as muh as II and III . Thus , only the total count of labeled sulfated lipids is shown . At 8 h of continuous labeling with [35S]-sulfate , the sulfate activation activity in ASgs , APSKgs , and IPPgs was decreased to 54 . 1±11 . 0 , 49 . 1±15 . 9 , and 24 . 0±0 . 6% , respectively , as compared to the control ( Figure 2B ) . We also examined the stability of the accumulated products by pulse-chase experiment . The degradation kinetics of all transformants was similar ( Figure 2C ) . These results indicate that AS , APSK , and IPP are indeed involved in sulfate activation . We next examined the growth of the gene-silenced strains . Although ASgs , APSKgs , IPPgs , and the mock control showed similar growth pattern , the three former strains showed marked growth retardation . The doubling time of ASgs , APSKgs , and IPPgs was 24 . 9±2 . 7 , 26 . 1±2 . 0 , and 34 . 4±1 . 2 h , respectively , while that of the control was 15 . 0±0 . 8 h ( Figure 2D ) . The degree of the growth inhibition was parallel to the level of repressed sulfate activation activity ( Figure 2B , growth rate: IPPgs<APSKgs = ASgs<control ) . These results indicate that sulfate activation is important for cell proliferation . Chlorate ( ClO3− ) , a known inhibitor for AS in mammals and fungi [23]–[25] , inhibited cell growth of E . histolytica ( IC50 = 10 . 5 mM ) . At this concentration , the sulfate activation activity ( as expressed as the total count of labeled sulfated lipids ) was decreased to 22 . 5% of the control . Furthermore , the apparent IC50 of the recombinant E . histolytica AS by chlorate was determined to be 3 . 55±0 . 25 mM . These results support the premise that sulfate activation plays an important role for proliferation . We also investigated the role of Cpn60 and MCF in sulfate activation . In Cpn60gs and MCFgs strains , the activity of sulfate activation was decreased to 20 . 2±5 . 4 and 34 . 2±1 . 1% , respectively , as compared to the control ( Figure 2B ) . MCFgs and Cpn60gs strains also showed marked growth defect; the doubling time of MCFgs and Cpn60gs strains was 41 . 7±5 . 5 and 82 . 4±14 . 5 h , respectively ( Figure 2D ) . These results indicate that MCF and Cpn60 significantly contribute to sulfolipid synthesis and cell proliferation . However , as growth of these gene-silenced strains was moderately-to-severely affected , the growth retardation should be taken into account for the impaired sulfolipid synthesis in these strains . We investigated the localization and topology of the proteins involved in sulfate activation pathway by exploiting the transformants expressing the HA-tagged proteins ( AS , APSK , IPP , and MCF ) . Immunoelectron microscopy revealed that all the proteins examined are confined to the electron dense double-membrane-surrounded organelles , the size of which are 150–400 nm in diameter ( Figure 3A ) . Double labeling with anti-HA and anti-Cpn60 antibodies showed that these proteins were co-localized with Cpn60 , the authentic marker of mitosomes ( Figure 3B; only AS and MCF were shown ) . While AS , APSK , IPP , and Cpn60 were evenly distributed throughout the luminal ( matrix ) part of mitosomes , MCF was concentrated on the inner membrane of mitosomes . The quantification results for the distribution of AS , APSK , IPP , MCF , and Cpn60 were summarized in Table 1 . These proteins were found to be 140-560-fold concentrated in mitosomes as compared to the cytosol . The number of mitosomes was estimated to be 32 . 1±9 . 7 per section ( 10 sections of 10 cells were examined ) . We estimated by the method described previously [12] that the mitosomes density is about 1 . 57 per µm3 , the number of mitosomes per trophozoite is about 6585 , and the volume percentage of mitosomes is about 1 . 2% , in the E . histolytica transformant lines used in this study .
In our previous report , we showed that mitosomes of E . histolytica uniquely possess sulfate activation pathway , while they have lost most of the functions shared by other aerobic eukaryote mitochondria including TCA cycle , electron transport , oxidative phosphorylation , and β-oxidation of fatty acids [5] . Only three chaperones and two mitochondrial-type transporters aside from the components in the sulfate activation pathway are retained in E . histolytica mitosomes [4] , [5] , [26] . Since the sulfate activation pathway is not typically confined to the mitochondria , and generally present in either the cytosol or plastid in eukaryotes [5] , [7] , the physiological significance of its compartmentalization in E . histolytica remains unknown [5] . In this report , we have provided several lines of direct and indirect biochemical and cell biological evidence that the sulfate activation pathway plays an important role in the production of sulfolipids and the growth of trophozoites . Consistent with this premise , the AS inhibitor , chlorate , inhibited the sulfolipid production in and the growth of E . histolytica . Further supporting the specificity of chlorate to AS , two amino acid residues ( Asn198 and His201 ) of Saccharomyces cerevisiae AS , which were implicated in the chlorate binding [24] , as well as Arg362 , which is located in the highly conserved ISGTxxR motif , are well conserved in E . histolytica AS ( Asn211 , His214 , and Arg375 ) . In addition to the importance of the enzymes in the sulfate activation pathway , we demonstrated that MCF , and Cpn60 also play an important role in cell proliferation . The phenotype of Cpn60gs is most likely attributable to multiple defects as Cpn60 is required for the folding and quality control of mitosome-targeted proteins [27] . Lack of Cpn60 should result in an inability to fold freshly-imported mitosomal proteins and therefore make them functional . This notion likely explains why the knockdown of Cpn60 impaired the cell growth more severely than that of the genes directly involved in sulfate activation . The phenotype of MCFgs strain was probably accounted for the lack of ATP supply required for chaperone functions and for the activation of inorganic sulfate into PAPS in AS- and APSK-catalyzed reactions . The latter possibility was supported by the observation that in MCFgs strain , the activity of sulfate activation was significantly reduced while the amount of the proteins involved in the pathway was not changed . Lack of MCF would impair the ADP/ATP ratio in mitosomes , and , as Cpn60 needs ATP to function , likely cause a similar effect as Cpn60gs . We assume that MCF and Cpn60 are not indispensable for sulfolipid synthesis per se , but gene silencing of these house-keeping proteins resulted in broader effects , and thus severe impairment of mitosome functions . Immunoelectron microscopy revealed that MCF is mainly localized on the inner mitosome membrane . Although membrane topology may need to be further verified , the observed localization of MCF agrees well with its biochemical characteristics , previously demonstrated: ATP import and ADP export [28] . Recently , E . histolytica phosphate transporter ( EhPiC ) has been identified and proposed to transport phosphate released from ATP through hydrolysis by chaperones , i . e . , Cpn60 and mitochondrial HSP70 [26] . It is conceivable that EhPiC transports phosphate generated by IPP in sulfate activation pathway in mitosomes . While the structure of E . histolytica mitosomes revealed by immunoelectron microscopy was far different from the typical eukaryotic mitochondria , it was somehow similar to the mitochondrion-related organelles , described as densely-stained double membrane-bound organelles lacking the typical cristae , in M . balamuthi , in shape , apparent size , number per cell , and structure [29] . Together with the fact that M . balamuthi possesses genes involved in sulfate activation , it is possible that sulfate activation is a unique feature shared only by E . histolytica and M . balamuthi . The number and structure of mitosomes demonstrated in this study was slightly different from previous reports [12] , [30] . This may be due to the heterogeneity of mitosomes in the cells . We have also previously shown that the distribution of AS , APSK , IPP , MCF , and Cpn60 in mitosomes was not uniform [5] , suggesting the heterogeneity of mitosomes in E . histolytica . | The mitochondrion and its related organelles are ubiquitous in all extant eukaryotic cells . The mitochondria are believed to have originated from the endosymbiosis of α-proteobacteria in an ancestral eukaryote , and show diverse structures , contents , and functions . Evolution and diversification of mitochondrion-related organelles remains one of the central themes in biology . Entamoeba histolytica , which causes intestinal and extraintestinal amebiasis in humans , possesses a highly divergent form of mitochondrion-related organelles , named “mitosomes . ” Previously , we demonstrated that sulfate activation is the major function of mitosomes in E . histolytica . As the sulfate activation pathway was discovered only in the cytoplasm and plastids in other eukaryotic organisms , its compartmentalization to mitosomes is unprecedented . In this study , we showed that this pathway is important for sulfolipid synthesis and cell proliferation in E . histolytica . Together , we infer that E . histolytica mitosomes are not just rudimentary or residual mitochondria , but important for proliferation of E . histolytica . Thus , E . histolytica represents a useful model to understand evolutionary constraints of mitochondrion-related organelles in eukaryotes . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"organismal",
"evolution",
"microbial",
"metabolism",
"giardiasis",
"microbiology",
"parasitic",
"diseases",
"parasitology",
"parastic",
"protozoans",
"amebiasis",
"microsporidiosis",
"microbial",
"evolution",
"cryptosporidiosis",
"infectious",
"diseases",
"microbia... | 2011 | Sulfate Activation in Mitosomes Plays an Important Role in the Proliferation of Entamoeba histolytica |
The dynamics of the PPi release during the transcription elongation of bacterial RNA polymerase and its effects on the Trigger Loop ( TL ) opening motion are still elusive . Here , we built a Markov State Model ( MSM ) from extensive all-atom molecular dynamics ( MD ) simulations to investigate the mechanism of the PPi release . Our MSM has identified a simple two-state mechanism for the PPi release instead of a more complex four-state mechanism observed in RNA polymerase II ( Pol II ) . We observed that the PPi release in bacterial RNA polymerase occurs at sub-microsecond timescale , which is ∼3-fold faster than that in Pol II . After escaping from the active site , the ( Mg-PPi ) 2− group passes through a single elongated metastable region where several positively charged residues on the secondary channel provide favorable interactions . Surprisingly , we found that the PPi release is not coupled with the TL unfolding but correlates tightly with the side-chain rotation of the TL residue R1239 . Our work sheds light on the dynamics underlying the transcription elongation of the bacterial RNA polymerase .
The DNA-dependent RNA polymerase is the main enzyme that participates in the transcription process transferring the genetic information from DNA to messenger RNA ( mRNA ) [1] . Crystallographic structures of the multi-subunit RNA polymerases in eukaryotes [2]–[4] and bacteria [5]–[8] engaged in transcription elongation process have been obtained . These atomic-level structures provide static snapshots of the transcription cycle [9]–[14] . In each nucleotide addition cycle ( NAC ) of the multi-subunit RNA polymerase , the post-translocation state first allows the substrate NTP to bind to the active site [6] . Then , a critical domain , named trigger loop ( TL ) , can fold then expel the solvent from the active site [15]–[17] , and finally form direct contacts with the substrate NTP . Substitution of a conserved TL histidine can significantly decrease the polymerization rate [18]–[21] . Recent mutagenesis studies have shed light on the roles of the TL on the nucleotidyl transfer [20] , [21] , and the reverse intrinsic hydrolysis process [22] . Previous MD simulation studies also provided information on TL dynamics and its potential regulatory roles during the translocation process [23] , [24] . After the catalytic reaction , PPi forms and releases from the active site [25] , [26]; then the TL opens and allows the template DNA to translocate so that a new NAC can start . Extensive biochemical and theoretical studies have been performed to understand the specific steps in the NAC , such as motions of the TL [17] , catalysis [26]–[30] , translocation [23] , [24] , [31]–[33] and NTP binding [34] , [35] . PPi release in single subunit T7 RNA polymerase is proposed to be tightly coupled with the translocation [36] but the same coupling is not observed in Escherichia coli ( E . coli ) RNA polymerase [37] . Interestingly , recent fluorescence and biochemical studies found that the PPi release in the E . coli RNA polymerase occurs shortly before or concurrently with the translocation [33] . Nonetheless , the interplay between the PPi release step and the TL opening motions at molecular level is still elusive . Previously , we used MD simulations to study the PPi release in the eukaryotic RNA polymerase II ( Pol II ) [25] . We proposed a hopping model for Pol II in which PPi release was coupled with the TL tip motion through the interactions between the TL residue H1085 and the ( Mg-PPi ) 2− group , and subsequently hopping among several positive charged residues in the secondary channel . Our model further suggested that the PPi release is a fast dynamic process so that it may not be able to induce the fully TL opening motion . A comparison of the secondary channel and TL structure between Pol II and bacterial RNA polymerase ( RNAP ) from T . thermophilus ( Tth ) displays substantial differences ( See Figure 1 ) [3] , [7] . In Pol II , the TL contains a long loop domain ( from the Rpb1 residue T1080 to T1095 ) [3] . However , the TL in RNAP consists of two alpha helices connected by a short turn in the closed state [6] . This structural difference suggests that the dynamics of the TL folding in these two systems are likely to be different . Moreover , in addition to the conserved Tth TL residue H1242 , the Tth TL residue R1239 also interacts with the substrate NTP [6]; this residue is absent in Pol II and mutation of the counterpart residue in E . coli ( R933A ) can reduce the nucleotide addition rate [20] . Moreover , the secondary channel in Tth RNAP is much shorter than that in Pol II ( See Figure 1 ) , and exhibits a different layout of the positively charged residues . Specifically , in Pol II , the four residues , K619 , K620 , K518 and K880 are located at relatively separated sites ( See Figure 1A ) . However , the positively charged residues in Tth RNAP: K908 , K912 , K780 and K1369 are close to each other in a continuous region ( See Figure 1B ) . Given these structural differences , it is of interest to compare the dynamics of PPi release in RNAP with that in Pol II . Although conventional all-atom MD simulations can provide the dynamic information for biological macromolecules at atomic resolution , it is still challenging to capture the biologically relevant timescales in microseconds or even longer . Markov State Models ( MSMs ) constructed from a large number of short simulations provide one way to overcome this timescale gap [38] , [39] . MSMs have been successfully applied to model the long timescale dynamics that cannot be directly accessed by conventional MD simulations in studying the conformational changes of biological macromolecules [39]–[41] , including our previous study of PPi release in Pol II [25] . In this study , in order to reveal the mechanism of the PPi release in RNAP , we constructed a MSM from extensive all-atom molecular dynamics ( MD ) simulations in explicit solvent with a system size of nearly 300 , 000 atoms and aggregated simulation time of ∼1 µs . Our results reveal that the PPi release in Tth RNAP adopts a simple two-state model with a fast dynamics over a few hundred nanoseconds . Surprisingly , we found that the PPi release is not coupled with the secondary structure unfolding of TL but only with the side-chain rotation of the TL residue R1239 .
To study the release mechanism of the ( Mg-PPi ) 2− group in RNAP , we modeled the PPi-bound RNAP complex by directly cleaving the Pα-O bond in the ATP-bound RNAP complex that is derived from the Tth RNAP crystal structure ( See SI Figure S1 for the two structures and the Methods section for the modeling details ) [6] . This modeled PPi-RNAP complex was used as the starting structure for the steered MD ( SMD ) simulations to obtain the initial release pathways . To eliminate the bias in SMD simulations , we have then performed 100 10-ns MD simulations , and these simulations have widely sampled the region in the secondary channel ( See SI Figure S2 ) . Finally , we have constructed a MSM from these simulations to obtain the dynamics and other thermodynamic properties of the PPi release ( See the Methods section for details ) . Our MSM shows that the PPi release in Tth RNAP adopts a simple two-state model . In addition to the initial state with the PPi in the active site ( S1 state in Figure 2A ) , only one additional metastable state is identified ( S2 state in Figure 2A ) , and this state is ∼7-fold more populated than the S1 state ( See Figure 2B ) . The S2 state locates in an elongated region where several positively charged residues can stabilize the ( Mg-PPi ) 2− group . These results contrast with our previous findings that the ( Mg-PPi ) 2− group in Pol II hops through four clearly separated metastable states [25] . When the ( Mg-PPi ) 2− group is in the active site ( See Figure 2C ) , three positively charged β′ residues R1029 , H1242 and R1239 can interact with the negatively charged ( Mg-PPi ) 2− group . The residue R1029 locates at the exit of the active site , and thus it may play similar roles on the PPi release with its corresponding residue K752 in Pol II ( See Figure 2D ) [25] . Interestingly , the location of the conserved TL residue H1242 is different from its counterpart residue H1085 in Pol II , though both of them are in direct contact with the ( Mg-PPi ) 2− group . Both before and after chemistry , H1242 interacts with the Pα-O atom of the NTP in RNAP , whereas H1085 is in contact with Pβ-O atom in Pol II ( See Figure 3 ) [3] , [6] . To achieve this , the H1242 in RNAP has to locate deeper in the active site compared to H1085 . Finally , R1239 in RNAP locates at the same position as H1085 in Pol II , suggesting that these two residues may play similar roles in the PPi release . After escaping from the active site , the ( Mg-PPi ) 2− group reaches the S2 state with an elongated shape . In this state , multiple positively charged residues on the secondary channel ( K780 , K908 , K912 and K1362 ) can provide favorable electrostatic interactions with the negatively charged ( Mg-PPi ) 2− group ( See Figure 2C ) . In contrast , the ( Mg-PPi ) 2− group in Pol II is found to transfer through several hopping sites where groups of positively charged residues are spatially well separated ( See Figure 1A ) [25] . From the S2 state , the ( Mg-PPi ) 2− group will directly enter the solvent . In order to elucidate the specific roles of the three important residues: R1029 , H1242 and R1239 in the PPi release ( See Figure 2D ) , we performed additional mutant simulations starting from several different conformations from the S1 and S2 states . The potential of mean force ( PMF ) profile along the distance between the ( Mg-PPi ) 2− group and the Mg2+A is displayed in Figure 4A . The PMF plot shows two major free energy basins that are consistent with the two metastable states identified by our MSM . The starting structures chosen for the mutant simulations fall into two different regions in the PMF profile ( P1 and P2 sites in Figure 4A ) . The P1 site is located in the S1 state , while the P2 site is located in the S2 state but near to the boundary between the S1 and S2 states . Initial conformations from these two sites allow us to examine the roles of the residues involved in different stages of the PPi release . The mutant simulation results indicate that both residues R1239 and R1029 can facilitate the escape of the ( Mg-PPi ) 2− group from the active site to S1 state ( See Figure 4B ) . Here , we use the distance between the ( Mg-PPi ) 2− group and the Pα atom of the 3′-terminal nucleotide of the RNA chain ( dαβ ) to describe the extent of the PPi release from the active site . In the WT simulations ( See P1 in Figure 4A ) , the ( Mg-PPi ) 2− group can move towards the exit of active site with the dαβ value increasing from 6 Å to around 8 Å ( See the left panel of Figure 4B ) . However , the R1239A and R1029A mutants lead to a weaker tendency for the ( Mg-PPi ) 2− group to escape the active site ( the dαβ value fluctuates around 5 . 5 Å , middle and right panels in Figure 4B ) . On the other hand , the R1029K mutant is shown to have a similar effect to help the ( Mg-PPi ) 2− group to leave the active site as in WT ( see Figure S4A ) . These results indicate that positively charged residues play a crucial role to facilitate the PPi to release from the active site . Notably , the H1242A mutant can dramatically promote the PPi release from P1 site ( See SI Figure S4C ) , suggesting that H1242 may prevent the PPi release from the active site . In contrast , the TL residue H1085 in Pol II was previously found to facilitate the PPi release from the active site [25] . This difference may be due to the different locations of these two residues in the active site . Compared with H1085 in Pol II , H1242 in RNAP locates significantly deeper inside of the active site ( See Figure 3D ) . Therefore , it will be more difficult for H1242 to rotate and help the ( Mg-PPi ) 2− group to leave the active site . Instead , H1242 can provide an attractive interaction to prevent the PPi release . Next , we evaluated the roles of residues R1239 and R1029 in PPi release when the ( Mg-PPi ) 2− group is at the S2 state ( P2 in Figure 4A ) . In the WT system , the ( Mg-PPi ) 2− group fluctuates around its initial location within our simulations at a few nanoseconds , which was also observed in the R1029K mutant simulations initiated from P2 ( Figure 4C ) . Intriguingly , R1029A and R1239A substitutions lead to dramatic , but opposite , effects . The R1029A substitution facilitates the PPi release toward the solvent ( Figure 4C , right panel ) . Combined with the previous observations , we conclude that the R1029 may facilitate the PPi release from the active site but prevents the PPi release when it arrives at the S2 state . Thus R1029 plays a similar role as the corresponding residue K752 in Pol II ( See Figure 2D ) [25] . In contrast , the R1239A substitution drives the ( Mg-PPi ) 2− group back to the S1 state , suggesting that R1239 is critical for the ( Mg-PPi ) 2− group to escape the active site ( middle panel in Figure 4C ) . This indicates that R1239 , rather than H1242 residue , plays the role in PPi release most equivalent to that played by H1085 in Pol II . Compared with its counterpart residue H1085 in Pol II , the R1029 has a longer and more flexible side chain . In addition , it can form a stronger salt bridge with the ( Mg-PPi ) 2− group . Therefore , the side-chain rotation of R1239 alone may be sufficient to facilitate the PPi release .
Based on extensive unbiased MD simulations , we built a MSM for the PPi release in RNAP to elucidate its long timescale dynamics . The MSM identified a two-state model for the PPi release ( See Figure 6A ) . The mutant simulations indicate that the β′ residues R1239 and R1029 can facilitate the escape of the ( Mg-PPi ) 2− group from the active site after the catalytic reaction ( See Figure 4B ) . Then the ( Mg-PPi ) 2− group transfers to the S2 state , where it forms favorable interactions with four positively charged residues on the secondary channel: K908 , K912 , K780 and K1369 ( See Figure 6A ) . More strikingly , our work suggests that the PPi release does not induce the TL unfolding but tightly couples to the side-chain rotation of the TL residue R1239 , which in turn makes the TL tip region more flexible . Furthermore , our control simulations show that TL is stable in Pol II , but can quickly unfold ( within 200 ns ) when exposed to the solvent . We thus speculate that the rotation of R1239 that accompanies the PPi release may allow solvent to re-enter the active site and promote the overall movements of the TL domain; This TL movements would further lead to its exposure to solvent and eventually allow TL unfolding . However , the timescales for this solvent-induced TL unfolding may be significantly longer than that of the PPi release so that we didn't observe it in our simulations . We found that the TL in RNAP may be more difficult to unfold than that of Pol II , since its secondary structures barely unfold upon the PPi release . Therefore , if the open state of the TL is a pre-requisite step for the translocation as recently suggested by both experimental [33] and computational studies [24] , it is intriguing that the transcription rate for bacterial RNAP is much faster than that of Pol II [27] . Despite that the more stable secondary structure of the TL in RNAP may slow down its opening motion , its reverse closing motion may be spontaneous and fast . This fast closing motion may further accelerate the nucleotide addition process to achieve an overall fast transcription rate . Finally , our MSM indicates that the PPi release in bacterial RNAP is faster than that in Pol II [25] . This faster dynamics is due to several factors: First , the secondary channel of RNAP is shorter than the that of Pol II due to the absence of the funnel region , therefore the ( Mg-PPi ) 2− release path is shorter , which leads to a faster PPi release from the active site of RNAP . Next , in RNAP , the PPi only needs to overcome a single free energy barrier before it can be released to the solvent ( See Figure 6A ) . In contrast , the PPi release in Pol II was found to go over multiple free energy barriers sequentially before it could be released ( See Figure 6B ) . Furthermore , in our model for RNAP , the S2 state ( in the pore ) has a population over seven times larger than that of the S1 state ( in the active site ) . Thermodynamically , this difference will favor release of PPi from the active site . However , in Pol II , the equilibrium population of the S2 state ( the first state in the pore ) is comparable to that of the S1 state ( in the active site ) [25] . This difference may be due to the fact that the S2 state in bacterial RNAP is greatly stabilized by four positively charged residues that are spatially close to each other ( K908 , K912 , K780 and K1369 ) , but in Pol II , these positively charged residues locate at relatively separate sites ( See Figure 1A ) . Finally , R1239 in RNAP can substantially facilitate the ( Mg-PPi ) 2− release from the active site all the way to the solvent due to its longer and more flexible side chain ( See Figure 5 ) . However , its counterpart residue for PPi release from Pol II , H1085 , only promotes the PPi escape from the active site to the first metastable state , S1 , rather than all the way to solvent [25] .
In order to obtain the initial PPi release pathway , we applied steered MD simulations [51] to pull the ( Mg-PPi ) 2− group out of the active site . The pulling was performed along three directions with the aim of considering all the possible PPi release pathways . Three groups of residues were used to determine the pulling directions: β′ subunit residues 1136–1145 , 908–914 and 1246–1253 ( named as group I , II and III respectively ) . Two sets of pulling simulations were along the wall of the secondary channel: one was pulled towards the center of the Cα atoms of group I residues , and the other was directed to the center of Cαatoms of both group I and group II residues . The third set of pulling simulations pointed to the center of the Cαatoms of group II and group III residues , and toward the solvent . The external force was only applied on the center of mass of the PPi group with a force constant of 0 . 5 kJ mol−1 Å−2 and pulling rate of 0 . 01 Å/ps . For each pulling direction , five independent steered MD simulations were performed starting from the final snapshot derived from 5 parallel MD simulations of the PPi-bound RNAP complex . We first divided the conformations from SMD simulations into 20 clusters using the K-center clustering algorithm [52] . In the clustering , the distance between a pair of conformations was set to be the RMSD value of three PPi atoms ( the bridge oxygen and two phosphate atoms ) . To compute RMSD , the structure was aligned to the energy minimized PPi-RNAP complex by the Cαatoms of the bridge helix domain . We then randomly selected 5 conformations from each cluster ( a total of 100 conformations ) to conduct unbiased MD simulations . Each simulation was run for 10 ns and the snapshots were saved every 2 ps . Altogether , we obtained an aggregation of ∼1 µs simulations with 500 , 000 conformations . In MSM , the conformational space is divided into a number of metastable macrostates and the fast motions are integrated out by coarse graining in time with a discrete unit of Δt . The model is markovian if Δt is longer than the intra-state relaxation time . In other words , the probability for the system to be at a given state at time t+Δt only depends on the state at time t . In MSM , the long timescale dynamics can be modeled by the first-order master equation . ( 1 ) Where P ( nΔt ) is the state populations vector at time nΔt , and T is the transition probability matrix . Δtis the lag time of the model . To calculate T , one can normalize the transition count matrix generated by counting the number of transitions between each pair of states at the observation interval of Δt from MD trajectories . MSM has been successfully applied to model conformational changes that occur at timescales that cannot be directly accessed by conventional MD simulations such as protein folding [39] , [40] , [53]–[55] . To construct the MSM , we have followed a splitting and lumping procedure [52]: In order to check if the model is markovian , we have plotted the implied timescales ( τk ) as a function of the lag time τ: ( 2 ) where μk is the eigenvalue of the transition probability matrix T with the lag time τ . The implied timescales correspond to the average transition times between two groups of states , and thus indicate the dynamics of the system . If τ is sufficiently large , the model is markovian , and the predicted implied timescales will not change upon the further increase of the lag time . In our system , the implied timescale plots reach the plateau at the lag time of ∼4 ns ( See SI Figure S3A ) . Therefore , we select the lag time of 4 . 5 ns to construct the final MSM . To further validate the model , we predicted the probability for a given macrostate to stay within it after a certain lag-time based on our MSM , and this predicted values are in good agreement with those obtained from the original MD simulations ( See SI Figure S3B ) . In order to investigate the stability of the TL in free solution , we have performed a 300 ns control simulation with the isolated TL domain ( A1225 to A1265 ) in solution ( with ∼6900 atoms , See SI Figure S6B ) . However , it is difficult to extend individual MD simulations of the complete transcription complex ( nearly 300 K atoms ) to hundreds of nanoseconds due to its high computing cost . Therefore , we have also performed simulations with a truncated RNAP complex containing all the motifs surrounding the TL domain ( See SI Figure S7A ) , including β subunit residues 381–569 , 831–1049 , β′ residues 604–794 , 901–1470 , 10 upstream hybrid DNA-RNA base pairs , 6 downstream DNA base pairs and Mg2+A in the active site . The final solvated system only contains ∼118 K atoms , but it still takes more than 2 months to perform one 300-ns simulation using 24 CPU cores . The explicit SPC water model was used for the MD simulations , 1 and 29 Na+ ions were added to neutralize the isolated TL and truncated RNAP model respectively . The other setups for the MD simulations were the same as the seeding MD simulations . We have performed one 300-ns simulation for the isolated TL and the other two 300-ns simulation for the truncated RNAP . For the truncated RNAP model , several terminal residues that are truncated from the complete model were fixed in the simulations in order to avoid undesired unfolding . | Pyrophosphate ion ( PPi ) release is a critical step in the nucleotide addition cycle of transcription elongation . Despite extensive experimental studies , the kinetic mechanism of the PPi release in bacterial RNA polymerases ( RNAP ) still remains largely a mystery . As a cellular machine , RNAP contains more than 3000 residues , and thus it is challenging for all-atom molecular dynamics ( MD ) simulations to directly capture the process of the PPi release . In this study , we have simulated the dynamics of the PPi release at microsecond timescale using the Markov State Models ( MSMs ) built from extensive MD simulations in explicit solvent . MSM is a powerful kinetic network model and can predict long timescale dynamics from many short MD simulations . Our results suggest a simple two-state model for the PPi release in RNAP , which sharply contrasts with the more complex four-state hopping model in the yeast RNA polymerase ( Pol II ) . We also observe a 3-fold faster dynamics for the PPi release in RNAP compared to Pol II , consistent with the faster transcription rate in the bacterial systems . Our results greatly improve our understanding of the PPi release , and also provide predictions to guide future experimental tests . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"biochemistry",
"computational",
"chemistry",
"chemistry",
"theoretical",
"chemistry",
"biology",
"computational",
"biology",
"biophysics"
] | 2013 | A Two-State Model for the Dynamics of the Pyrophosphate Ion Release in Bacterial RNA Polymerase |
Helicobacter pylori are gram-negative bacteria notable for their high level of genetic diversity and plasticity , features that may play a key role in the organism's ability to colonize the human stomach . Homeologous natural transformation , a key contributor to genomic diversification , has been well-described for H . pylori . To examine the mechanisms involved , we performed restriction analysis and sequencing of recombination products to characterize the length , fragmentation , and position of DNA imported via natural transformation . Our analysis revealed DNA imports of small size ( 1 , 300 bp , 95% confidence limits 950–1850 bp ) with instances of substantial asymmetry in relation to selectable antibiotic-resistance markers . We also observed clustering of imported DNA endpoints , suggesting a possible role for restriction endonucleases in limiting recombination length . Additionally , we observed gaps in integrated DNA and found evidence suggesting that these gaps are the result of two or more separate strand invasions . Taken together , these observations support a system of highly efficient short-fragment recombination involving multiple recombination events within a single locus .
Many of the bacterial species that are long-term colonizers of human niches are characterized by high levels of genetic diversity [1] . The gram-negative bacteria Helicobacter pylori , which can induce chronic gastritis , peptic ulceration and gastric malignancies after decades of persistence in the human stomach , displays an exceptional level of genomic diversity [2] . Multiple genetically distinguishable isolates have been observed in a single individual , and it has been postulated that genetic diversity and plasticity greatly facilitate adaptation within a host as well as transmission to future hosts [3]–[5] . The finding of highly efficient in vitro recombination , along with evidence of genetic shuffling and a panmictic population structure , suggest that horizontal genetic exchange amongst diverse strains plays a substantial role in generating this extensive gene pool [6] , [7] . Most wild-type strains of H . pylori are naturally competent for transformation , which occurs via the actions of proteins encoded by the comB locus [8]–[13] . Horizontal transfer of donor chromosomal DNA containing a point mutation conferring a selectable antibiotic resistance phenotype can be observed in the homeologous natural transformation of H . pylori [9] . The genome of the transformed strain includes a point mutation flanked by varying lengths of the donor DNA . Better knowledge of the characteristics of homeologous recombination may aid in understanding how H . pylori persists through changing environments within the human host , as well as serving as a model for other naturally competent organisms . Prior studies , which involved mathematical analysis of H . pylori genomic sequence data , provide evidence that the median size of imported donor fragments appears small ( 412 bp ) compared with Streptococcus pneumoniae , Neisseria meningitidis , and Bacillus subtilis ( 2000 , 5000 , and 10 , 000 bp , respectively ) [14]–[17] . In addition , as little as 5 bp flanking an antibiotic resistance-conferring point mutation and 150 bp flanking an antibiotic cassette are sufficient to allow transformation to antibiotic resistance [8] , [18] . In this study , we present a new approach to studying the products of transformation by taking advantage of the naturally-occurring polymorphisms between two sequenced H . pylori strains . We employed restriction analysis and direct sequencing of recombination products to define the length of an imported sequence flanking a point mutation conferring antibiotic-resistance . Using this approach , we were also able to characterize the fragmentation and configuration of this imported DNA . Additionally , we examined whether our findings could be explained by self-hybridization , methylation , or restriction modification of the recombination substrates . Finally , we were able to confirm and compare the transformation characteristics involving more than one genomic locus .
Under the standardized conditions employed [8] , chromosomal DNA from 26695 StrR cells transformed J99 wild-type cells to StrR with a mean frequency of 4 . 56×10−5 transformants/CFU/µg DNA ( Table 1 ) . These data were consistent with those from previous experiments ( [8] , data not shown ) . This frequency of homologous transformation was approximately 10-fold higher than for homeologous transformation with J99 donor chromosomal DNA [8] . Natural transformation resulted in the incorporation of a point mutation ( A128G in codon 43 of rpsL in strain 26695 donor DNA ) into the recipient strain J99 genome . Our first goal was to determine the length of DNA flanking the point mutation ( conferring streptomycin resistance ) that was incorporated into the recipient genome along with the mutation . Since there is a ∼6% genetic variation within conserved ORFs between the two strains [19] , the flanking DNA includes numerous single nucleotide polymorphisms ( SNPs ) , that can differentiate between the two strains as the source of a mosaic DNA . For the 8790 bp flanking the StrR site , there were 627 SNPs ( 7 . 1% polymorphism ) between the strains ( mean separation 20 . 3±19 . 3 bp ) . Those SNPs that were within one of the pre-designated restriction sites ( Figure 1 ) were identified . In each independent transformant , we sought to determine the presence or absence of such restriction sites at various distances from the StrR mutation . To achieve this , PCR products were amplified using primers flanking the restriction site ( Table S1 ) , and the resulting amplicons were treated with the appropriate restriction enzyme . At distances 100 bp and 399 bp downstream of the StrR ( A128G ) mutation , all transformants from 10 independent transformations of strain J99 contained restriction sites unique to the donor strain , 26695 ( “downstream” or “upstream , ” refers to the direction of replication from the origin ) . However , at 811 bp , 3331 bp , 4668 bp , and 5149 bp downstream of A128G , the number of independent transformants containing the 26695-unique sites fell to 5 , 2 , 1 , and 0 , respectively . At a distance of 172 bp upstream of A128G , 9 of the 10 independent transformants analyzed contained the donor 26695-unique sites . That frequency fell to 5 , 4 and 0 transformants at increasing distances upstream of A128G ( 520 , 834 , and 1350 bp , respectively ) . No two of the 10 studied transformations yielded identical results , confirming that these transformants were truly independent . The presence or absence of a restriction site was unambiguous within any of the independent transformants , making it unlikely that mixed clones were present . For each independent transformant , the restriction sites furthest upstream and downstream from StrR that were still unique to 26695 ( donor ) were identified as proximal to the crossover endpoint . To discriminate the crossover endpoints with greater precision , PCR products were amplified using primers flanking these restriction sites , and sequences of the amplicons were determined . For each , the sequence data revealed a region that was bounded by two SNPs—one unique to 26695 and the other , to J99 . To calculate flanking DNA length , the crossover endpoint was defined as the middle of this region . For the 10 independent transformants , the lengths of flanking DNA ranged from 915–4967 bp ( Table 1 ) . In total , we observed the mean integration of 2058±1494 bp donor DNA into the recipient chromosome . Despite substantial variation among flanking DNA lengths , two independent transformants were found to contain crossover endpoints within a single 15 bp region ( transformants 5 and 6 showed crossovers 505–520 bp downstream of StrR ) ( Figure 2A ) . Further analysis of transformant 4 revealed a 13 bp region ( 4742–4754 bp downstream of the StrR mutation ) that can be characterized as a gap in integration . This region contained a SNP unique to J99 ( recipient ) that was flanked by 2 SNPs unique to 26695 ( donor ) . This result was observed in sequences obtained with both the forward and reverse primers . The integration of DNA was asymmetric with respect to StrR , favoring the downstream side at a ratio of about 2∶1 . Several transformants had integrated flanking DNA with extreme asymmetry , such as transformant 4 , which was characterized by a downstream∶upstream ratio of 24∶1 . To examine the hypothesis that flanking DNA asymmetry is influenced by a region of non-homology between 26695 and J99 occurring ∼3 kb upstream of A128G ( Figure 1 ) , we repeated the transformation experiment but reversed the roles of the strains . Chromosomal DNA from J99 StrR cells transformed 26695 wild-type cells to StrR with an average efficiency of 1 . 67×10−6 transformants /CFU/µg DNA ( Table 1 ) . Natural transformation resulted in the incorporation of a point mutation ( A263G in codon 88 of rpsL in strain J99 ) into the recipient strain 26695 genome . Surprisingly , at a distance as small as 110 bp downstream of StrR , 5 of 10 transformants contained a restriction site unique to recipient strain 26695 , indicating that the contiguous downstream crossover point was very close to the A263G transforming point mutation . Sequence data confirmed this extreme asymmetry with some of the crossover endpoints located only 53 bp downstream of A263G ( Figure 2B ) . Several transformants had very high upstream∶downstream flanking DNA ratios; for example , transformant 14 displayed a ratio of 51∶1 . An average of 2814±1542 bp of flanking DNA was integrated along with the point mutation , with flanking DNA lengths ranging from 634 to 6089 bp . The sequence data for transformants 13 and 15 revealed large gaps of integration , in which the flanking donor DNA is interrupted by significant regions ( up to 1658 bp ) of recipient DNA ( Figure 2B ) . To examine the flanking DNA for additional gaps , DNA from six independent transformants ( 11 , 12 , 14 , 15 , 16 , and 17 ) was completely sequenced within a ∼8 . 5 kb region from 2278 bp upstream to 6239 bp downstream of the StrR mutation . The sequence data confirmed the identity of the polymorphisms that had been deduced by the previous restriction site analysis ( Figure 2B ) . No additional gaps of integration were observed . Thus , out of 20 independent transformants , only 3 were found to contain gaps ( transformants 4 , 13 , and 15 ) . Although several transformants had flanking DNA characterized by extreme asymmetry , across the 10 independent transformants , the mean sizes of the flanking DNA fragments upstream and downstream from A263G were similar ( 1472 bp±965 upstream and 1341 bp±1378 downstream ) . The flanking DNA of transformant 11 extends upstream by nearly 3 kb and borders the region of non-homology between the two strains . Although no significant difference in total flanking DNA length was found between the two strains , recipient strain 26695 contained significantly more upstream flanking DNA than J99 ( Table 1 ) . This is not unexpected since the region of non-homology between the two strains lies upstream in relation to the StrR mutation . To estimate the mean size of imports caused by transformation of chromosomal H . pylori DNA , we performed statistical analysis that takes into account both the uncertainty in the position of the recombination endpoints due to the stretches of sequence identity between J99 and 26695 and the fact that longer imports are more likely to appear in the dataset because they are more likely to include the StrR mutation used to select for strains that have undergone recombination ( see Methods ) . The analysis does not take into account the gaps in integration described above , which are difficult to model statistically . Instead , in a first analysis , we only took into account the stretch of contiguous imported DNA immediately flanking the StrR mutation . This analysis gave a mean import size of 1100 bp ( 95% confidence limits 800–1600 bp ) . A second analysis , treating the whole imported stretch as a single unit , and ignoring the effect on ascertainment of the gaps in integration , yields an estimate for the mean size of the entire region involved in the import of 1300 bp ( 95% confidence limits 950–1850 bp ) . That no two of the transformants yielded identical results , confirms the independence of each event . However , despite significant variation among flanking DNA lengths , two transformants were found to contain crossover endpoints within a single 15 bp region ( transformants 5 and 6 downstream of StrR ) . Two other transformants ( 14 and 20 ) shared an endpoint within 24 bp . Still other transformants ( 17 and 20 ) contained crossover endpoints within a single 66 bp region downstream of StrR . Indeed , the endpoints of many transformants were found to be clustered within short regions—transformants 13 , 15 , and 16 were within a 67 bp region; transformants 14 and 15 were within a 97 bp region; transformants 12 and 15 were within a 93 bp region; transformants 1 and 9 were within a 93 bp region; transformants 15 and 19 were within a 105 bp region; transformants 2 and 5 were within a 105 bp region; and transformants 3 and 7 were within a 108 bp region . Of 48 crossover points in the 20 independent transformants studied in the two experimental protocols , 20 ( 41 . 7% ) recurred within short regions ( <110 bp ) . These short regions did not coincide with regions of putative self-hybridization , as determined by MFold [20] . H . pylori strains contain numerous functional methyltransferases that protect endogenous DNA from endonuclease digestion by methylating specific recognition sequences [21] . Methylated genomic DNA may interrupt donor DNA integration during recombination , generating multiple crossover endpoints within a single short region . To explore the hypothesis that genomic DNA methylation patterns play a role in determining crossover endpoints during homeologous transformation , the locations of the methyltransferase recognition sequences in both 26695 and J99 were mapped alongside the crossover endpoints in the 20 independent transformants . This comparison did not yield any significant correlations ( data not shown ) , suggesting that methylation of neither the recipient chromosome nor the donor DNA significantly interrupts the integration of donor DNA . H . pylori strains 26695 and J99 each encode four different active restriction-modification ( R-M ) systems [22] . Each system consists of an active restriction endonuclease ( RE ) and a methytransferase , both recognizing the same sequence . Since the R-M systems are not conserved between the two strains , the donor DNA in homeologous transformation is susceptible to cleavage by REs active in the recipient strain . To examine the role of endonuclease restriction of donor DNA in homeologous transformation , the locations of donor DNA recognition sequences targeted by active recipient strain REs , for which the corresponding methyltransferase is either absent or non-functional in the donor strain , were mapped alongside the crossover endpoints in the 20 transformants ( Figure 3 ) . In total , 11 of 23 ( 48% ) crossover endpoints were within regions that overlapped with the specific RE recognition sites . To determine whether this was a chance observation , we performed statistical inference modeling , considering the endpoints of the contiguous imported fragment flanking StrR as being γ times more likely to occur at sites that are susceptible to cleavage due to restriction modification than at sites that are not . This analysis yielded an estimate of γ of 32 . However the confidence limits are wide , ranging from 0 to 100 , so that based on these 20 events , we have insufficient power to determine whether the restriction sites are significantly overrepresented at import boundaries . The observation of gaps in integration has been described previously in Streptococcus pneumoniae , and has been attributed to the activity of mismatch repair ( MMR ) proteins [23] . However , H . pylori does not contain homologs to known MMR genes [19] , [24] , and has no MMR activity [25] . Since the efficiency of natural transformation was on the order of 10−6 , the predicted frequency of adjacent independent recombination events would be ∼10−12 . Therefore , we hypothesized that these gaps are the result of DNA repair enzyme activities , rather than adjacent independent recombinations . To examine this hypothesis , we analyzed natural transformation of 26695 cells with a ∼9 kb PCR product ( designated P-StrRif ) amplified via primers flanking the A128G mutation and a point mutation conferring rifampin-resistance ( G1588A in rpoB , resulting in D530V ) . The mutation conferring RifR is separated from the mutation conferring StrR by ∼7 . 2 kb ( Figure 1 ) . Such transformation yielded StrR transformants and RifR transformants with a frequency of ∼10−4 transformants/CFU/ugDNA , but StrRRifR transformants ( containing both mutations ) with a frequency of 4 . 7×10−6 transformants/CFU/ug DNA ( Table 2 ) . These findings suggest a linkage between the two markers . However , since both markers are co-localized on a single PCR product , it is not clear whether this linkage was the result of a single DNA crossover event or multiple ones . To differentiate between these hypotheses , we also transformed 26695 cells with an equimolar mixture of the ∼9 kb PCR products containing either the StrR mutation ( designated P-Str ) or the RifR mutation ( designated P-Rif ) . Under these conditions , transformation efficiencies were similar to those from the single product ( Table 2 ) . The frequency of finding an integrated RifR mutation adjacent to a given StrR mutation ( or visa versa ) was ∼1/130 . That this frequency is several log10 higher than the frequency predicted for two adjacent independent recombinations , suggests that the integration of both markers is mechanistically linked , and that such linkage occurs even if both markers are located on separate DNA molecules . To further explore this linkage , we repeated the experiment using J99 cells and a donor ∼9 kb PCR product derived from a 26695 StrR RifR template . Using restriction analysis ( as described above ) , transformant chromosomal DNA was analyzed for the presence of a SNP unique to J99 ( recipient ) that was approximately equidistant from the StrR and RifR mutations . Such analysis confirmed the presence of the SNP in 6 of 6 transformants , suggesting that within the transformant DNA , a gap in integration exists between the two antibiotic resistance mutations . The presence of such a gap is consistent with a mechanism involving multiple crossover events . This analysis was also repeated after reversing the strains . 26695 cells were transformed with a donor ∼9 kb PCR product derived from a J99 StrRRifR template , yielding StrRRifR transformants which were analyzed for the same SNP . This analysis found that four of the six independent transformants contained the SNP unique to 26695 ( recipient ) , but two transformants had the donor SNP . While this finding does not rule out the presence of a gap in integration between the StrR and RifR mutations in these two transformants , it suggests that multiple mechanisms may be involved in the generation of these genetic products during homeologous transformation . The rpsL gene encodes an abundant ribosomal binding protein , and is located within a highly transcribed region of the genome . To evaluate the products of natural transformation in other less-highly transcribed genomic regions , we studied the region of recG . Previous data involving microarrays have shown that the transcription rate of recG is 15 to 87-fold less than that of rpsL [26] . For our studies , we used the pRecGCat plasmid , as described previously [27] . pRecGCat contains a sequence corresponding to recG of strain J99 that is interrupted by a chloramphenicol acetyl transferase ( cat ) cassette conferring chloramphenicol resistance ( CmR ) . In initial studies , we transformed pRecGCat into wild-type J99 cells to generate strain HPXZ136 . Chromosomal DNA from this strain then was used to transform 26695 cells; 10 independent transformants were isolated , as in the prior experiments . The frequency of transformation was 3 . 0×10−6±2 . 5×10−6 transformants/CFU/µg DNA , which is ∼10-fold less than for the transformations involving the StrR point mutation . The transformation frequency for the control experiment ( J99 recG cat→J99 ) was similar ( 6 . 3×10−6±3 . 1×10−6 transformants/CFU/µg DNA ) , indicating no large restriction barrier for this transformation into 26695 . As in the prior experiments ( see Materials and Methods ) , we performed sequence analysis on 10 independent transformants , yielding the locations of the upstream and downstream crossover endpoints ( Figure 4A ) . In total , in addition to the selectable 884 bp cat cassette , the mean flanking insert size was 3393±1013 bp ( range 1650–4355 bp ) . The average lengths of the upstream ( 1771±936 bp ) and downstream ( 1623±959 bp ) inserted sequences were similar , although the actual fragment lengths varied considerably . This is >1 kb more than the mean length of integrated DNA observed for the StrR mutation . No gaps in integration were observed within any of the 10 transformants , nor was there clustering of crossover endpoints . Strain 26695 contains two active restriction endonucleases ( HpyAII and HpyAV ) for which J99 ( and thus HPXZ136 ) lacks the cognate methylases . Within the range of the cross-overs , there are 19 upstream and 12 downstream sites on the J99 chromosome that would be cleaved by these 26695 REs . In total , 6 ( 30% ) of the 20 endpoints overlap with the specific restriction sites , a finding significantly greater than chance ( p = 0 . 046 ) . In a second experiment involving the recG locus , we used a more complex transformation strategy ( Figure 4B ) . We transformed pRecGCat into wild-type 26695 cells to create strain HPXZ135 . Sequence analysis of HPXZ135 revealed a ∼200 bp segment of 26695 sequence interrupting the J99 sequence immediately upstream of the cat cassette ( Figure 4B ) . This ∼200 segment was designated as the “donor gap . ” We then used chromosomal DNA from HPXZ135 ( containing the “donor gap” ) to transform wild-type J99 recipient cells . 10 independent transformants were isolated , as in previous experiments . This transformation yielded colonies at a mean efficiency of 5 . 60×10−6±6 . 93×10−6 transformants/CFU/µg DNA , similar to the results when HPXZ136 was used . Sequence analysis performed on the 10 independent transformants identified the locations of upstream and downstream crossover endpoints ( Figure 4B ) . In total , we observed a mean integration of 3098±946 bp donor DNA into the recipient chromosome , with similar upstream ( 1695±795 bp ) and downstream ( 1392±1157 bp ) sequence lengths . In each transformant , the architecture of the donor DNA was preserved , with the cat cassette and the “donor gap” present in all transformants ( Figure 4B ) . The transformation fragment variation reflected the location of the crossovers involving the flanking 26695 sequences . As in the prior analysis ( Figure 4A ) , we did not observe any clustering of endpoints . We mapped the sites of known active J99 restriction endonucleases ( Hpy99I , Hpy99II , and Hpy99IV ) whose corresponding methylases are absent in 26695 and thus , HPXZ135 . We found relatively few RE sites flanking the cat cassette , with only 4 sites within 4 kb downstream of the cassette ( 708 , 2097 , 3019 , and 3020 bp from the cassette ) and none upstream . With so few informative markers , we found no relationship between crossover points and restriction sites .
In S . pneumoniae , N . meningitidis and B . subtilis , transformation yields imported DNA lengths of about 2 kb , 5 kb and 10 kb , respectively [15]–[17] . It might have been anticipated that transformation of H . pylori would result in imported DNA lengths at the upper end of this range since H . pylori lacks homologs to DNA recombination/repair proteins [28] , which have been implicated in limiting the length of heteroduplex formation during recombination in both prokaryotes and eukaryotes [29] . However , natural transformation of H . pylori yields insertion lengths near the smaller portion of this range , with a mean integration of ∼2400 bp of donor chromosomal DNA into the recipient chromosome ( Table 1 ) . After statistical treatment of the data , the mean import size was found to be ∼1300 bp , which is consistent with previous reports of small import lengths within naturally occurring mosaics of H . pylori , as well as efficient in vitro transformation with small donor DNA fragments [8] , [14] . Although it has been suggested that digestion of naked DNA within the external environment may play a role in reducing the length of imported fragments , a significantly greater mean import length was not observed despite limiting the donor substrate to freshly purified chromosomal DNA extract . While excessive mechanical shearing of the donor DNA cannot be excluded , it is generally prevented with standard techniques and would not explain the astochastic distribution of crossover endpoints [30] . In other naturally competent organisms , such as N . gonorrhoeae , S . pneumoniae , and B . subtilis , DNA is imported as a single strand that is resistant to cleavage by endonucleases [31]–[33] . However , there is evidence to suggest that natural transformation in H . pylori may be different . Natural competence in H . pylori is associated with proteins at the comB locus—a feature not found in other naturally competent bacteria [11]–[13] . Transformation of H . pylori involving double-stranded donor DNA results in ∼1000-fold higher transformation efficiency than transformation with a single-stranded substrate [8] . Furthermore , evidence of a restriction barrier has been found in H . pylori; for example , strain-specific methylation of donor DNA transforms H . pylori more efficiently than unmethylated DNA that is susceptible to restriction [34]–[36] . If substrate donor DNA exists in the cytoplasm as a double-strand , the action of native strain-specific H . pylori restriction endonucleases would be predicted to curtail import size . In our data , we observed an overrepresentation of crossover endpoints at susceptible restriction sites , which would be the expected consequence of donor substrate DNA restriction . It would be interesting to compare our findings with those in strains that do not harbor R-M systems . The finding of imports with extreme asymmetry with respect to the selective StrR mutation is consistent with the previous observation that H . pylori is successfully transformed by asymmetric donor DNA fragments with minimal homology on one flank and is suggestive of a highly efficient mechanism for recombination [8] . This finding is also consistent with the observation that flanking DNA in one transformant actually abuts the region of non-homology between strains 26695 and J99 . This is unsurprising in an organism that lacks MMR enzymes , since such enzymes would be expected to play a role in the early termination of recombination at polymorphic sites . The observation of gaps of integration has been previously described for S . pneumoniae [23] . In bacterial transformation , initiation of recombination involves the invasion of a single-stranded segment of donor DNA into the recipient chromosome . The finding of gaps of integration has been previously attributed to events that occur following donor strand invasion , such as repair excision of mismatched bases , resulting in the conversion of genetically heterozygous sites to homozygosity [23] , [37] . However , since H . pylori lacks a full complement of mismatch repair enzymes , we hypothesized that these gaps may be due to events occurring before or during donor strand invasion . For example , the initiation of two separate but adjacent strand invasion events would result in two neighboring regions of donor genotype separated by a gap . However , the coincidence of two or more independent strand invasion events within ∼3 kb ( see Figure 2B ) would be unexpected given our sample size , unless the events were mechanistically linked . One example of linkage would involve the origination of both invading strands from a contiguous DNA molecule . However , we found that the simultaneous genomic integration of two genetic markers present on two separate DNA molecules occurred with the same frequency as the integration of markers coexisting on the same molecule ( Table 2 ) . While multiple hypotheses exist to account for such linkage , including local concentration effects due to enzyme recruitment , the uptake of large quantities of donor DNA by the cell , and increased susceptibility of certain genomic regions to recombination , these findings suggest that any explanation for the mechanism of linkage between the two markers must include an account for adjacent strand invasions involving at least two discrete DNA molecules . This finding is certainly consistent with a donor substrate that has been cleaved by R-M system enzymes prior to strand invasion of the chromosome . However , the StrR and RifR mutations are located in highly transcribed essential genes , for which the measured recombination rate may differ from that in other parts of the genome . It might be suggested , that the observation of relatively large gaps within a 9 kb region and the finding of much smaller gaps within a ∼3 kb region ( Figure 2B ) , are the result of two different processes . While the incidence of gaps in the 9 kb region is significantly lower than for the 3 kb region ( an incidence of ∼1/130 for the 9 kb region versus 3/20 for the 3 kb region ) , that may not be unexpected since two simultaneous strand invasions that are far apart may occur much less frequently than strand invasions that are closer together . Furthermore , the finding that the incidence of gaps varies at 9 kb versus 3 kb suggests that location and proximity of markers does indeed play a role , and that the simultaneous integration of both markers is not simply the result of two coincidental but independent events . The finding of significant gaps within the 9 kb region strongly suggests that the distance between the StrR and RifR markers ( ∼7 . 2 kb ) may be too long for a single co-transformation event . Thus , each doubly marked fragment only transforms at a single site , but not at both . These observations set an upper boundary ( of 7 . 2 kb ) for the amount of DNA that can be contiguously integrated from a single site , but may reflect the restriction and methylation differences between the donor and recipient strains . To examine whether the same recombination characteristics are present for non-essential genes , we transformed wild-type 26695 cells with chromosomal DNA from strain HPXZ136 , which contains a cat cassette within the recG gene . Interestingly , the frequencies of transformation between recG and rpsL were similar ( Table 1 ) . Given that the transcription rate of recG is up to 87-fold lower than that of rpsL [26] , this suggests that differences in transcription rate may not have a major effect on transformation efficiency . Genomic sequencing of the resulting transformants showed no evidence of gaps in integration or clustering of endpoints , as had been found flanking the StrR mutation in rpsL . However , there was a significant overrepresentation of endpoints that overlapped with sites in which the incoming J99 DNA would be susceptible to 26695-specific restriction . As a further test , we performed a transformation involving a more complex genomic architecture from strain 26695 into J99 . The frequency of transformation was almost the same as when the simpler architecture was used , indicating no substantial differences whether or not a “donor gap” in the identity of the sequence was introduced . The lack of endpoint clustering was consistent with the very small number of J99-specific restriction sites on the incoming 26695-based donor DNA . Of note , the distribution of flanking DNA was relatively symmetric with respect to the cat cassette , a finding that contrasts with the asymmetric distribution of DNA flanking the StrR mutation . This may be due to the loss of homology ∼3 kb upstream of the StrR mutation in rpsL , a feature not found in recG . One potential limitation of our technique is the lack of fully-quantitative standardization of cell density prior to the introduction of donor DNA . However our method of standardizing cell density has yielded highly consistent transformation efficiencies over time , and the transformation frequencies we have observed are consistent with previous experiments [8] . In conclusion , we have presented a new approach to studying the products of transformation that has yielded evidence of highly efficient short-fragment recombination involving multiple recombination events within a single locus . These events likely involve distinct genetic markers residing on discrete DNA molecules , as is the case following processing by cell-specific restriction endonucleases . While our findings may also be partially explained by the lack of a mismatch repair pathway in H . pylori , it would be worthwhile to consider the role of other proteins that regulate recombination , such as MutS2 [38] . It also would be of interest to consider the role of various type I and type III R-M systems that are present in H . pylori . By phylogeography , 26695 is an hpEurope strain while J99 is of the hpAfrica1 lineage , which suggests that they are not closely related despite commonalities within certain loci ( eg . cag and vacA ) [39] . It may be interesting to explore whether genetic exchange between more-closely related strains , with fewer R-M system barriers , yields different results . Within this context , we have uncovered novel and interesting insights into how recombination occurs around a selectable genetic marker . This will help us to further understand how bacteria are able to employ genetic recombination to efficiently generate and maintain genomic diversification within the population—a feature that allows H . pylori to persistently colonize the human stomach [28] .
The H . pylori isolates used in this study were wild-type strains 26695 and J99 , for which the complete genomic sequences have been solved [19] , [24] . H . pylori cells were grown at 37°C in 5% CO2 for 48 h on Trypticase soy agar ( TSA ) with 5% sheep blood ( BBL Microbiology Systems , Cockeysville , MD ) or Brucella-serum ( BS;BBL ) agar with 10% newborn calf serum ( Serologicals Corporation , Norcross , GA ) plates . Spontaneously streptomycin-resistant ( StrR ) mutants of these strains were selected by plating approximately 1010 cells on TSA medium containing streptomycin ( 10 µg/ml ) . This procedure yielded H . pylori strains that had single point mutations conferring streptomycin resistance in rpsL; for 26695 , the mutation was A128G ( K43R ) , and for J99 , the mutation was A263G ( K88R ) [9] , [40] . For the transformation experiments involving donor DNA , independent transformants were selected on BS plates with streptomycin ( 20 µg/ml ) added . Rifampin-resistant ( RifR ) H . pylori strains were obtained by natural transformation of 26695 cells with PCR products amplified using primers containing an A1589T ( D530V ) point mutation in rpoB that confers rifampin resistance [41] . Each independent transformant was harvested from a single colony growing on a separate plate , to ensure that each transformant represented an entirely independent genetic clone . A chlorampheniol-resistant ( CmR ) J99 strain , HPXZ136 , was generated by transforming wild-type J99 with plasmid pRecGCat which is based on J99 recG [27] , and then selecting one single colony from a BS plate containing chloramphenicol ( 10 µg/ml ) . The cat cassette insertion site in the HPXZ136 genome and its flanking region was confirmed by sequencing . A second transforming DNA was prepared by using pRecGCat ( with J99 recG ) to transform strain 26695 , with the same selection and confirmation as above , creating HPXZ135 . This strain has a 200 bp gap of 26695 sequence within the J99 sequence as a result of the pRecGCat transformation . H . pylori chromosomal DNA was prepared from cells of each independent transformant after 48 h of growth on TSA plates , as described [30] . PCR was performed on the isolated DNA in a reaction volume of 100 µl containing 1U of Taq ( Qiagen , Valencia , CA ) as per manufacturer's protocol . Each pair of primers was selected to exactly match the chromosomal sequences in both J99 and 26695 ( Table S1 ) , and flanked a restriction site unique to that segment for either 26695 or J99 ( Figure 1 ) . Restriction enzyme ( RE ) digestions were performed in a reaction volume of 20 µl containing 2U of the appropriate RE in its buffer at 37°C for 12 h . All PCR products were electrophoresed at 120V for 60 min through 2% agarose gels with 0 . 7 µg/mL ethidium bromide . The desired band was excised from the gel , purified ( Qiagen , Valencia , CA ) , and stored at −20°C until used in transformation studies . The H . pylori cells to be transformed were grown on trypticase soy agar ( TSA ) plates with 5% sheep blood ( BBL ) agar for 48 h , harvested into 1 ml of phosphate-buffered saline ( PBS ) pH 7 . 4 , centrifuged at 850 g for 5 min , and the pellet resuspended in 150 µl of PBS . Each transformation mixture , consisting of 25 µl of recipient cells ( ∼108 cells ) and 15 µl ( at 2 ng/µl ) of donor cell chromosomal DNA , was spotted onto a TSA plate and the plates were incubated for 24 h at 37°C in a 5% CO2 atmosphere . The transformation mixture then was harvested into 1 ml of PBS , and 100-µl aliquots of appropriate serial dilutions inoculated to both TSA ( non-selective ) and antibiotic ( selective ) plates with 20 µg/ml streptomycin , 87 . 5 µg/ml rifampin , or both , and incubated for 5 days at 37°C in a 5% CO2 atmosphere . The number of colonies of transformants and total viable cells were counted and the transformation frequency was calculated as the number of streptomycin-resistant colonies per microgram of DNA per recipient CFU . Transformation using the cat cassette was performed similarly , except that 15 µl ( containing 100 ng ) of donor chromosomal DNA was used . CmR transformants were selected on BS agar plates with 10 µg/ml chloramphenicol . For each recipient strain , a single aliquot of cells harvested from a fully confluent plate , was divided into 25 ul portions , which were each placed in the center of a separate TSA plate . Donor DNA was then introduced to each plate and the plates were incubated overnight . The resulting transformed cells from each plate were then harvested , diluted , and inoculated onto an antibiotic plate . One colony from each antibiotic plate was selected with a sterile pipette , inoculated onto a TSA plate , incubated for 4 days , and harvested . Chromosomal DNA was prepared using a CTAB/Chloroform/Phenol extraction , as described [30] . In transformations utilizing donor PCR products , 9 kb PCR amplicons were first generated using primers flanking the StrR mutation ( A128G ) and a mutation conferring rifampin resistance ( G1588A in codon 530 of rpoB , resulting in D530V ) . Cells of strain 26695 were exposed to either 100 ng of PCR products containing both mutations or an equimolar mixture of two different amplicons ( 100 ng each ) , each containing only one mutation . Serial dilutions were inoculated to TSA plates , 20 µg/ml streptomycin plates , 87 . 5 µg/ml rifampin plates , and to plates containing both streptomycin and rifampin . Each CmR transformant was picked with a sterile pipette , inoculated onto a separate BS agar plate with 10 µg/ml chloramphenicol , and incubated for 3 days . The culture from the selective plate was swabbed onto a TSA plate , incubated for 2 days , and harvested . Chromosomal DNA was prepared using the Wizard genomic DNA purification kit ( Promega , Madison , WI ) . To analyze the region of DNA flanking the cat cassette , PCR was performed using chromosomal isolates from 10 independent transformants . Primers were designed to match the sequences in both 26695 and J99 ( Table S2 ) . The desired PCR products were purified with the QIAquick PCR purification kit ( Qiagen , Valencia , CA ) . For analysis of specific sequences flanking the StrR mutation , 0 . 5–1 kb PCR products were prepared as described above and were sequenced on both strands at the DNA Sequencing Resource Center ( Rockefeller University , New York , NY ) . For analysis of specific sequences flanking the cat cassette , ∼1 kb PCR products were prepared as described above and sequenced on both strands at the High-Throughput Genomics Unit ( University of Washington , Seattle , WA ) . All sequence analysis was performed using Sequencher software ( Gene Codes , Ann Arbor , MI ) . If the distribution of imports is assumed to be exponential with mean length λ , and imports are randomly distributed on the chromosome , then the probability density function of imports of length x containing the StrR mutation will be equal to , since the probability that any given import will include the StrR site is proportional to its length . Not all nucleotides differentiate 26695 from J99 , so the length of each observed import is ambiguous . If an import has possible start positions ranging from a to b and end positions from c to d then the likelihood of observing it is . The overall likelihood of the data given λ is the arithmetic product of the likelihoods of all imports . Restriction modification could result in some sites being more likely to contain the endpoints of imports than others . We model this by assuming that unmethylated sites for which the restriction modification system is present in the recipient strain are γ times more likely to be digested than other sites . The likelihood of observing a particular event of length , starting at and finishing at site , conditional on it containing the StrR mutation at site 0 then is proportional to , where if x is at an unmethylated restriction site and otherwise . This expression is approximated by truncating the sum at values of c and d that bound all observed imports ( for imports from 26695 to J99 and for imports from J99 to 26695 ) . See Table 3 for a list of GenBank ID numbers . | Helicobacter pylori are gram-negative bacteria that have been implicated in human diseases after decades of persistence in the stomach . Known for its high level of genetic diversity , H . pylori is competent to undergo natural transformation , a process in which donor DNA is integrated into the recipient chromosome . To examine the mechanisms involved , we analyzed the DNA imported via natural transformation in an experimental model system . We found variation in the average length of imported DNA fragments , with asymmetry with respect to a selectable marker . We also found evidence that strain-specific restriction endonucleases may limit recombination length . Additionally , we observed gaps in the integrated DNA and provide evidence that these gaps are the result of separate strand invasions . Together , our observations support a highly efficient system of short-fragment recombination involving multiple recombination events within a small region of the chromosome . This helps explain how bacteria are able to employ genetic recombination to efficiently generate and maintain genomic diversification within the population—a feature that helps H . pylori persistently colonize the harsh environment of the human stomach . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"molecular",
"biology/recombination",
"microbiology/microbial",
"evolution",
"and",
"genomics",
"infectious",
"diseases/gastrointestinal",
"infections"
] | 2009 | Natural Transformation of Helicobacter pylori Involves the Integration of Short DNA Fragments Interrupted by Gaps of Variable Size |
Effective antiretroviral therapy ( ART ) dramatically reduces AIDS-related complications , yet the life expectancy of long-term ART-treated HIV-infected patients remains shortened compared to that of uninfected controls , due to increased risk of non-AIDS related morbidities . Many propose that these complications result from translocated microbial products from the gut that stimulate systemic inflammation – a consequence of increased intestinal paracellular permeability that persists in this population . Concurrent intestinal immunodeficiency and structural barrier deterioration are postulated to drive microbial translocation , and direct evidence of intestinal epithelial breakdown has been reported in untreated pathogenic SIV infection of rhesus macaques . To assess and characterize the extent of epithelial cell damage in virally-suppressed HIV-infected patients , we analyzed intestinal biopsy tissues for changes in the epithelium at the cellular and molecular level . The intestinal epithelium in the HIV gut is grossly intact , exhibiting no decreases in the relative abundance and packing of intestinal epithelial cells . We found no evidence for structural and subcellular localization changes in intestinal epithelial tight junctions ( TJ ) , but observed significant decreases in the colonic , but not terminal ileal , transcript levels of TJ components in the HIV+ cohort . This result is confirmed by a reduction in TJ proteins in the descending colon of HIV+ patients . In the HIV+ cohort , colonic TJ transcript levels progressively decreased along the proximal-to-distal axis . In contrast , expression levels of the same TJ transcripts stayed unchanged , or progressively increased , from the proximal-to-distal gut in the healthy controls . Non-TJ intestinal epithelial cell-specific mRNAs reveal differing patterns of HIV-associated transcriptional alteration , arguing for an overall change in intestinal epithelial transcriptional regulation in the HIV colon . These findings suggest that persistent intestinal epithelial dysregulation involving a reduction in TJ expression is a mechanism driving increases in colonic permeability and microbial translocation in the ART-treated HIV-infected patient , and a possible immunopathogenic factor for non-AIDS related complications .
Chronic systemic inflammation , characterized by increased frequencies of activated B and T cells [1] , elevated levels of circulating proinflammatory cytokines and chemokines [2] , and more rapid immune cell turnover [3] , is a hallmark of HIV/SIV infection and a better predictor of disease progression than plasma viral load [4] , [5] . Accumulating evidence suggests that this systemic inflammation plays a role in non-AIDS related comorbidities including cardiovascular diseases [1] , [6]–[8] , liver diseases [2] , [9]–[11] , and neurocognitive decline [3] , [12] , resulting in shortened life expectancy and premature aging in patients treated with long term antiretroviral therapy ( ART ) [4] , [5] , [13] , [14] . In addition , plasma levels of microbial products , such as lipopolysaccharides ( LPS ) and bacterial 16s rDNA , are elevated in chronically HIV-infected individuals and associated with markers of immune activation [15]–[17] , implicating circulating microbial products , due to microbial translocation , as a major cause of HIV-associated systemic inflammation [18] . An association between circulating microbial products and systemic inflammation has been observed in other disease processes such as inflammatory bowel disease [19] , [20] and after laparoscopic surgeries [21] , [22] . Moreover , conditioning regimens for stem cell therapy cause gastrointestinal ( GI ) tract injury that facilitates the translocation of microbial products from the intestinal lumen to systemic circulation , ultimately stimulating the immune system and exacerbating graft-versus-host disease [23] , [24] . Klatt et . al . highlight the association between gut epithelial structural damage , local and systemic microbial translocation , and systemic inflammation , in SIV-naïve pigtail macaques [25] , suggesting microbial translocation and systemic inflammation as direct consequences of damage to the GI tract in the absence of chronic viral infection . The GI tract is a major target site for HIV infection , as the mucosal immune system contains the majority of the body's T cells [26] . In addition , greater than 90% of intestinal CD4+ T cells are CCR5+ [27] , providing a large pool of target cells that are preferentially depleted by HIV . Independent of route of transmission , within weeks of HIV or SIV infection , rapid and severe depletion of intestinal lamina propria CD4+ T cells occurs and persists into the chronic phase of the disease [27]–[29] , with preferential depletion of the Th17 and Th22 subsets [30] , [31] . Significant accumulation of mucosal CD8+ T cells during HIV infection has also been shown [32] , [33]; both effects drastically alter mucosal immune homeostasis . Coincident with early mucosal CD4+ T cell loss , gene expression profiling reveals intestinal barrier dysfunction in primary HIV and SIV infection , as exemplified by down-regulation of genes associated with epithelial maintenance and digestive functions [34] , [35] . Upregulation of genes with intestinal mucosal protective and regenerative activity in elite controllers [34] confirms the pivotal role intestinal mucosal integrity may play in limiting systemic inflammation and controlling disease progression . Intestinal barrier dysfunction , long recognized in HIV patients with advanced disease , includes manifestations of pathogen-negative diarrhea and malabsorption [36] . Indirect assessments of intestinal permeability , through measuring urinary excretion of orally consumed oligosaccharides , demonstrate increased small intestinal permeability in symptomatic AIDS patients and some asymptomatic chronic HIV patients , regardless of therapy status [37] , [38] . Notably , increased small intestinal permeability did not correlate with intestinal structural change [37] , and , through in vitro impedance spectroscopy and flux analysis of duodenal biopsies , was suggested to be due to a leak flux mechanism [39] , alluding to an intestinal barrier defect as a result of tight junction ( TJ ) down-regulation . Our recent clinical report demonstrated increased small intestinal and colonic permeability in HIV-infected patients , which was not corrected by ART , further implicating intestinal barrier dysfunction as an ongoing pathophysiological change in ART-treated patients [40] . Our current study , using human intestinal biopsies , extends evidence for intestinal damage in SIV infection of non-human primates [41] and explores the molecular mechanisms behind increased intestinal permeability in ART-treated HIV+ patients . We hypothesize that HIV-associated dysregulation in intestinal epithelial cells will lead to TJ down-regulation , resulting in persistent intestinal barrier dysfunction in the ART-treated patients , contributing to microbial translocation and systemic inflammation .
The University Hospitals Institutional Review Board ( IRB ) has reviewed the following submission: Principal Investigator: Dr . Alan D Levine , Ph . D . Protocol Title: Loss of Intestinal Barrier Function in HIV Infection UHCMC IRB number: 06-07-31 Submission Type: Continuing Review Review Type: Full Board Date of Committee Review: 04/08/2014 As such , the UHCMC IRB has determined that with respect to the rights and welfare of the individuals , the appropriateness of the methods used to obtain informed consent and the risks and potential medical benefits of the investigation , the current submission is acceptable under Federal Human Subject Protection regulations promulgated under 45 CFR 46 and 21 CFR 50 and 56 . The current expiration date for this study is 04/07/2015 Subjects undergoing routine screening colonoscopies were recruited from the Digestive Health Institute and the Special Immunology Unit of the University Hospitals Case Medical Center , Cleveland , OH , with the exclusion criteria of any known or suspected gastrointestinal disease . After written informed consent was obtained , eight pinch biopsies , two each from the terminal ileum , ascending colon , transverse colon , and descending colon , were obtained from thirty-one patients with HIV ( median age 51 years , interquartile range [IQR] 50–55 years ) and thirty-five healthy controls ( median age 56 years , IQR 50–61 years ) . Peripheral blood was collected immediately following the colonoscopy procedure into EDTA-containing tubes to obtain plasma samples , which were stored at −80°C until assay . Apart from three HIV+ subjects who were simultaneously evaluated for chronic diarrhea , in whom no significant terminal ileal or colonic histopathological findings were identified , there were no reports of diarrhea prior to the colonoscopy preparation regimen for other HIV+ subjects and all controls subjects . HIV+ patients were enrolled notwithstanding their CD4 count and viral load , and all were under ART treatment . Controls were not specifically tested for HIV , but had no reported history of HIV infection . All study protocols were approved by the Institutional Review Board at University Hospitals Case Medical Center . Formalin-fixed , paraffin-embedded , 5-µm biopsy sections from the ascending , transverse , and descending colon were deparaffinized , rehydrated , mounted in Fluoroshield mounting medium with DAPI ( AbCam ) . Two biopsies per donor from each location were analyzed in a total of five HIV+ patients and five healthy controls . Individual images were obtained on a Leica DMI 6000 B inverted microscope using a 20× objective connected to a Retiga EXI camera ( Q-imaging ) , and composite images of each section were generated through stitching . A threshold intensity for excluding the background was established to specifically analyze nuclear staining . Epithelial cell nuclei and lamina propria cell nuclei were identified manually , after which measurements for the total area they occupy were enumerated using Metamorph Imaging Software ( Molecular Devices ) . The built-in count nuclei application module was used to determine epithelial cell numbers by setting an approximate nuclear width at 3–8 µm . Relative epithelial cell abundance ( compared to lamina propria cells ) was determined by the ratio of epithelial nuclei area to lamina propria nuclei area . Luminal barrier coverage , defined as the length of the intact epithelial/luminal border relative to lamina propria cell abundance , was designated as the ratio of border length to lamina propria nuclei area . Epithelial cell packing density was calculated as the number of epithelial cells per 100 µm of intact luminal border . Paraformaldehyde-fixed , frozen , 5-µm biopsy sections from the ascending , transverse , and descending colon were blocked with 10% normal goat serum , incubated with rabbit anti-occludin or anti-ZO1 antibody , and detected with chicken or goat Alexa Fluor 488-conjugated anti-rabbit secondary antibody ( Invitrogen Life Technologies ) . Sections were mounted in Fluoroshield mounting medium stained with DAPI ( AbCam ) , and visualized with the Perkin Elmer Ultraview VoX confocal microscope , using an oil-immersion 100× magnification objective lens connected to a Leica DMI 6000 B inverted microscope . En face and transverse Z-stack images ( 0 . 3 µm thickness ) were obtained using Volocity 6 . 2 ( Perkin Elmer ) . After applying a threshold to eliminate non-specific staining , 3D reconstruction of tight junctions was performed and analyzed using Imaris 3 . 0 ( Bitplane Scientific Software ) . An average of 7 fields of view on the intestinal surface and 5 fields of view in the crypts were imaged and analyzed for each biopsy obtained , two biopsies per location , from a total of three HIV+ patients and three healthy controls . Average fluorescence intensity for occludin or ZO-1 staining was analyzed for each field of view . Snap-frozen biopsy specimens stored at −80°C were homogenized with a bead beater ( Retsch ) for 3 min at a frequency of 30 Hz/second to ensure complete homogenization . Total RNA was extracted using the PureLink RNA Mini Kit ( Invitrogen Life Technologies , Carlsbad , CA ) and quantified with the Nanodrop 2000 ( Thermo Fisher Scientific , Wilmington , DE ) . cDNA was transcribed from 1 µg of total RNA using SuperScript II Reverse Transcriptase ( Invitrogen Life Technologies ) . Transcript levels of human beta-defensin 3 ( hBD-3 ) , E-cadherin , and tight junctional proteins occludin , zona occludens 1 ( ZO-1 ) , claudin-2 , and claudin-4 were determined by SybrGreen-based real-time PCR using CFX96 Real-Time PCR Detection System ( Bio-Rad Laboratories ) . After an evaluation of eight commonly used housekeeping transcripts for genetic stability based on geNormPlus analysis ( Biogazelle ) , β-actin and eukaryotic translation elongation factor 1-alpha 1 ( eef1A1 ) were identified and used as references . Primers used are summarized in Table 1 . Calibrated normalized relative quantities ( CNRQ ) of target genes were determined with qBasePlus ( Biogazelle ) analysis [42] . Total protein was extracted in 60 µl of SDS-RIPA buffer ( 50 mM Tris pH 8 . 0 , 150 mM NaCl , 0 . 3% SDS , 1% Triton X , 1 mM EDTA , 1∶100 protease inhibitors ) from snap-frozen descending colonic biopsies using a bead beater ( Retsch ) for 3 min at a frequency of 30 Hz/second , followed by constant agitation for 2 h at 4°C . Proteins were separated by 10% polyacrylamide gel electrophoresis ( Invitrogen Life Technologies ) and electro-transferred onto nitrocellulose membrane ( Invitrogen Life Technologies ) . After blocking with 5% non-fat milk solution , membranes were probed with rabbit antibodies against glyceraldehyde 3-phosphate dehydrogenase ( GAPDH ) , cytokeratin-18 , occludin ( AbCam ) , claudin-2 , and mouse antibodies against claudin-4 ( Invitrogen Life Technologies ) and β-actin ( AbCam ) . After incubation with HRP-conjugated goat anti-rabbit ( Thermo Fisher Scientific ) or anti-mouse ( AbCam ) secondary antibody , signals were visualized with enhanced chemiluminescence , using West Pico Supersignal ( Pierce ) . Chemiluminescence for all membranes was detected using Hyblot CL Autoradiography Film ( Denville Scientific ) . The amount of protein in each band was quantified by densitometry using ImageJ ( National Institutes of Health ) . Dilution series were electrophoresed to determine optimal loading amounts for each target protein , to guarantee that bands fell within the linear range of detection . Extracts yielding bands that were too underexposed or overexposed were re-electrophoresed after adjusting the loading volume to obtain bands that were accurately quantified . An inter-gel control sample used to normalize intensity variations between gels was electrophoresed on all gels . The densitometric intensity of each target protein occludin , claudin-2 , and claudin-4 was normalized to the intensity of cytokeratin-18 in each extract , to determine epithelial-specific TJ protein levels . Analysis was performed on samples from thirteen HIV+ patients and thirteen healthy controls . Plasma was prepared by centrifugation of EDTA-treated whole blood for 10 min at 1610 g and then frozen at −80°C until assay . Soluble CD14 ( sCD14 ) levels were measured using the Quantikine kit ( R&D Systems ) . Samples were thawed on ice and analyzed in batches in duplicate , background was subtracted , and mean values were reported . Plasma samples were diluted to 10% with endotoxin-free water and then heated to 85°C for 15 min to denature plasma proteins . Plasma levels of LPS were quantified with a commercially available Limulus Amoebocyte Lysate assay ( QCL-1000 , Lonza ) according to the manufacturer's protocol . Samples were analyzed in triplicate; backgrounds were subtracted , and mean values were reported . For evaluation of the histological samples , statistical analysis was performed utilizing a mixed-effects model that took the repeated measurements from the same individual into account , and unstructured covariance structure was used for the inference . Analyses were performed using SAS ( Statistical Analysis System , version 9 . 2 ) . All other analyses were performed using Prism 5 . 0 ( GraphPad Software ) . Relative cell abundance , transcript levels in the terminal ileum and colon , as well as protein levels in the descending colon are represented using box-and-whisker plots constructed using Tukey's method , where outliers are noted as distinct data points . Statistical analysis for mRNA , protein levels , and plasma sCD14 levels was performed on all data points , including outliers , via Mann Whitney U test . To analyze for TJ transcript levels versus gut location ( proximal-to-distal ) , the results were analyzed using the Kruskal-Wallis test , a non-parametric version of the one-way ANOVA , with a post test adjustment for multiple comparisons to evaluate linear trend . Spearman's rank correlation was calculated for immune activation markers versus colonic TJ transcript levels . For samples that were analyzed with multiple comparisons , a False Discovery Rate ( FDR ) analysis , using the Benjamini and Hochberg's approach [43] , was implemented with the SAS procedure PROC MULTTEST . P-values adjusted for FDR are reported . All tests were two-sided and p-values less than 0 . 07 were considered significant . We provide accession numbers from Entrez Gene and UniProtKB/Swiss-Prot , for all genes and proteins mentioned in the text , as follows: ZO-1 ( Gene ID: 7082; UniProtKB AC: Q07157 ) , occludin ( Gene ID: 100506658; UniProtKB AC: Q16625 ) , claudin-2 ( Gene ID: 9075; UniProtKB AC: P57739 ) , claudin-4 ( Gene ID: 1364; UniProtKB AC: O14493 ) , E-Cadherin ( Gene ID: 999; UniProtKB AC: P12830 ) , GAPDH ( Gene ID: 2597; UniProtKB AC: P04406 ) , cytokeratin-18 ( Gene ID: 3875; UniProtKB AC: P05783 ) , hBD-3 ( Gene ID: 414325; UniProtKB AC: P81534 ) , eef1A1 ( Gene ID: 1915; UniProtKB AC: P68104 ) , β-actin ( Gene ID: 60; UniProtKB AC: P60709 ) .
Our clinical study on intestinal barrier integrity in an HIV+ population , in which we measured urinary excretion of orally consumed oligosaccharides [40] , revealed increased permeability in the small intestine and colon . These results demonstrated that increased small intestinal permeability is a result of epithelial damage , while colonic paracellular permeability increased without epithelial damage , suggesting a loss in barrier function between intact epithelial cells . Importantly , increases in intestinal permeability were uncorrected in the ART-treated HIV+ patient population . To investigate the mechanism behind the persistent increase in intestinal permeability in ART-treated patients , we obtained intestinal pinch biopsy and plasma samples from thirty-one ART-treated HIV+ patients and thirty-five healthy controls undergoing screening colonoscopies at the Digestive Health Institute at University Hospitals Case Medical Center . Age range was similar between the control and HIV+ cohorts . More males participated than females in both cohorts , with the HIV+ cohort showing a higher percentage of males . HIV+ subjects have been infected for a median of 13 . 6 years , and reached a median peripheral blood CD4+ T-cell nadir of 176 cells/µl at a median of 6 . 5 years before the time of biopsy . At the time of biopsy , the median viral load and CD4+ T-cell count of the HIV+ cohort were 48 copies/ml and 569 cells/µl respectively . All but four HIV+ patients had undetectable viral load . All HIV+ patients studied have been under treatment with antiretroviral therapy for a median of 11 . 2 years , with uninterrupted treatment for a median of 4 . 1 years prior to study entry . All patients were on a minimum of three antiretroviral drugs , including at least two reverse transcriptase inhibitors and a combination of protease inhibitors and integrase inhibitors , at the time of biopsy . Demographics and clinical parameters of the cohorts are summarized in Table 2 . Analysis of the plasma levels of immune activation marker sCD14 and LPS , as an indicator of microbial translocation , was performed on twenty-one HIV+ patients and twenty-one healthy controls within our cohorts . Since a limited number of small biopsies were obtained per study subject , we were restricted to undertaking a single molecular or histological analysis on a sample obtained from each subject , with concurrent quantitative PCR and immunoblotting performed on biopsies from three HIV+ and seven control subjects . Subjects in each cohort were randomly assigned to various analyses . Figure 1 details the breakdown of biopsy and plasma samples from our cohorts into each analytical method . The intestine is comprised of a one-cell layer thick epithelium , lining the interface with the gut lumen and separating the outside environment from the plethora of immune cells in the lamina propria . To assess an overall loss of epithelial cells as a potential mechanism for increased colonic permeability , we compared the abundance of epithelial cells relative to lamina propria cells in intestinal biopsy samples from HIV+ patients and healthy controls ( Figure 2 ) . The epithelium in the HIV+ gut is grossly intact and continuous , without areas of focal epithelial cell loss , crypt bifurcation , neutrophil-induced injury , flattened epithelium , or ulceration , either on the luminal surface or in the crypts ( Figure 2A ) . Relative abundance of epithelial cells in the HIV+ gut , reflected by the ratio of epithelial cell nuclei to lamina propria cell nuclei , shows no changes in the ascending , transverse , and descending colon ( Figure 2B ) . Similarly , considering the extent of intact epithelial-luminal border as a measure of barrier coverage , the relative length of the epithelium in the HIV gut is not altered at all three colonic sites ( Figure 2B ) . This evidence for no loss in relative epithelial cell abundance and tissue length in the HIV-infected population is confirmed and extended by finding no change in cell density or packing in the HIV epithelium , as assessed by the number of epithelial cells per 100 µm of epithelium ( Figure 2C ) . These results directly indicate no changes in the abundance of epithelial cells relative to lamina propria cells , not the absolute number of epithelial cells . A previous report demonstrating the restoration of CD4+ T cells and increase in CD8+ T cells , both measured in absolute numbers , in the HIV+ gut mucosa after prolonged ART [32] enables us to conclude that the absolute number of intestinal epithelial cells is not decreased in the colon of HIV-infected individuals . No change in epithelial cell packing in the HIV+ intestinal epithelium also indicates that permeability in the HIV gut is not manifest at the cellular level , suggesting an alternative mechanism , at the molecular level , for the increase in colonic permeability in the ART-treated HIV+ individual . Since non-absorbable saccharide probes were utilized in our clinical study , the results reflect an increase in paracellular permeability ( extracellular , within the intercellular spaces between epithelial cells ) , as opposed to solute movement through the transcellular pathway [44] . Paracellular intestinal barrier function is primarily mediated by apically located transmembrane tight junctions ( TJs ) that seal the intercellular space between adjacent epithelial cells . One potential molecular mechanism for increased colonic intestinal paracellular permeability is the disruption of intercellular TJs . In an intact epithelial cell layer , the space in between individual epithelial cells is sealed by the apical junctional complex composed of the TJ and the subjacent adherens junction , with passage through the TJ being the rate-limiting step of the paracellular transport pathway , and for overall transepithelial solute transport [45] . The TJ is a multi-protein complex composed of transmembrane proteins , including members of the claudin family and TJ-associated marvel proteins such as occludin , which form the intramembranous TJ strands , and intracellular scaffold proteins such as ZO-1 , which connect the strand-forming proteins to the cytoskeleton and are important for TJ assembly and regulation [45] . Tissue-specific expression governing the combination of sealing and channel-forming claudin proteins in TJ complexes is the major determinant of distinct barrier properties and selectivity [46] , [47] . To examine the subcellular localization and organization of the TJ complex in the intestinal epithelium , we studied the distribution of TJ components occludin and ZO-1 in the HIV+ intestine , through high magnification ( 100× ) confocal microscopy . The TJ is located in the apical region of the lateral epithelial cell membrane , close to the luminal surface and away from the basolaterally-situated nucleus and the lamina propria . When the healthy intestinal epithelium is viewed in cross-section , both occludin and ZO-1 appear as distinct dot-like or line-forming structures on the luminal border ( Figure 3A ) . When viewed en face , TJs , assembled as a ring-like structure in association with the intracellular perijunctional actinomyosin ring , form a ‘chicken wire’ appearance in the crypts of control colonic tissue ( Figure 3B ) . Intact TJs are seen in the healthy control population throughout the colon spanning the ascending , transverse , and descending segments . In the HIV+ intestine , no obvious changes are seen in occludin and ZO-1 distribution , both on the intestinal surface and in the colonic crypts ( Figure 3A and 3B ) , indicating that epithelial TJ intracellular localization to the lateral plasma membrane is maintained , and that TJs are structurally intact , aligned between epithelial cells , in the ART-treated HIV+ population . The composition of TJs was also examined in these colonic biopsies . The abundance of occludin and ZO-1 were determined and represented by the average fluorescence intensity in stained ascending , transverse , and descending colonic sections . Average fluorescence intensity is shown in Tables 3 and 4 . No significant difference in occludin intensity in HIV+ samples is seen at the intestinal surface or crypt for all three colonic sites examined . Similarly , no significant difference in ZO-1 abundance is seen on the intestinal surface . In the crypt , significant increases in ZO-1 intensity are observed in the transverse HIV+ colon . To investigate whether the expression of the intestinal epithelial tight junctional complex is regulated in HIV+ patients at the transcriptional level , mRNA concentrations for a panel of TJ proteins were determined in intestinal biopsies through quantitative real-time PCR ( Figure 4 ) . mRNA levels were quantified using qBasePLUS software , which expresses transcript levels as calibrated normalized relative quantities , calculated based on endogenous levels of β-actin and eef1A1 as controls , selected based on geNorm analysis . While changes in TJ structure and subcellular localization were undetectable in our system using microscopic immunofluorescence , significant decreases in ZO-1 ( p<0 . 01 ) and occludin ( p<0 . 05 ) mRNA expression in the colon ( Figure 4A ) are observed in HIV+ individuals when compared to healthy control subjects . A significant decrease ( p<0 . 01 ) in the transcript expression level of claudin-2 , a cation-selective channel-forming protein , is also observed in the colon of HIV+ subjects , when compared to the expression level in healthy controls . Similarly , a significant decrease ( p<0 . 01 ) in colonic mRNA expression in the HIV+ patient is detected for another claudin family member , claudin-4 ( Figure 4A ) , which functions predominantly as a sealing protein with controversial anion-channel-forming activity . The decrease in TJ mRNA varies from 1 . 4-fold for claudin-4 to 2 . 7-fold for claudin-2 . We propose that these modest changes in mRNA expression , normalized across a wide stretch of tissue , explain why focal changes in protein expression were undetectable via confocal microscopy examining only a limited number of fields of view . Strikingly , the expression levels of all four TJ mRNAs studied are not changed in the terminal ileum of HIV+ patients as compared to their expression in healthy controls ( Figure 4B ) , suggesting an HIV-associated tissue-specific down-regulation of TJ transcripts in the GI tract , seen only in the colon and absent in the terminal ileum . This decrease in TJ mRNA only in the colon is consistent with our clinical study that revealed HIV-associated increased paracellular permeability in the colon and tissue damage in the small intestine [40] . Since small intestinal and colonic biopsies were obtained from four different locations along the GI tract , namely terminal ileum , ascending colon , transverse colon , and descending colon , we are uniquely positioned to investigate whether decreased colonic epithelial TJ mRNA expression in the HIV+ population is differentially distributed relative to anatomical location ( Figure 5 ) . Using a Kruskal-Wallis analysis with a post test adjustment for linear trend , ZO-1 gene expression shows a significant direct linear increase in transcript level from proximal-to-distal intestine in the healthy control population ( p = 0 . 045 ) . This expression pattern is dramatically reversed in the HIV+ population , where ZO-1 transcript levels demonstrate a significant inverse trend between expression and gut location ( p = 0 . 049 ) ( Figure 5A ) . Comparing ZO-1 transcript levels at each specific intestinal site , we observe a significant decrease in HIV+ individuals as compared to the healthy control population in the more distal portions of the GI tract , namely the transverse ( p = 0 . 05 ) and descending ( p<0 . 01 ) colon , consistent with the concept that there is a continual decrease in ZO-1 mRNA expression in the HIV+ population as one travels distally in the colon ( Figure 5A ) . In contrast to ZO-1 , occludin , claudin-2 , and claudin-4 transcript levels remain relatively constant from proximal-to-distal gut in the healthy control population ( Figure 5B–5D ) . While we do not observe a significant linear trend toward reduction in occludin expression toward the distal colon in the HIV+ population , similar to ZO-1 we do find that occludin transcript level is significantly reduced only in the distal , namely the transverse and descending colon in the HIV+ cohort ( p<0 . 05; Figure 5B ) . In congruence with ZO-1 expression in the HIV+ population , claudin-2 and claudin-4 mRNA expression show a significant linear trend toward reduction from the terminal ileum to descending colon ( p = 0 . 044 and p = 0 . 059; Figure 5C and 5D ) . Examining each colonic location individually , significant decreases in claudin-2 expression are observed in the transverse and descending colon ( p<0 . 05; Figure 5C ) . A similar pattern holds for claudin-4 ( Figure 5D ) , with the HIV+ population showing significantly decreased claudin-4 transcript expression in the transverse ( p<0 . 05 ) and a trend toward a decrease in the descending ( p<0 . 08 ) colon , while claudin-2 and -4 transcript levels remain unchanged in the terminal ileum and ascending colon ( Figure 5C and 5D ) . Overall , HIV infection is associated with a significant modification in the intestinal TJ complex's anatomic expression profile , delineated by a proximal-to-distal decreasing gradient in mRNA expression . To eliminate the potential effects of differences between the percentages of male versus female subjects in the control and HIV+ cohorts , we reanalyzed the expression of intestinal epithelial tight junction transcripts only for the male subjects , which represent the overwhelming majority of HIV+ volunteers . Significant decreases in all four TJ mRNA levels are again observed in the colon of HIV+ males ( Figure S1A ) , while transcript levels are not altered in the terminal ileum ( Figure S1B ) , replicating the observation for the entire HIV+ and healthy control cohorts . Upon examination of TJ expression levels along the proximal-to-distal axis of the gut for male subjects , we reconfirmed our conclusions on HIV-associated modifications of intestinal TJ complex transcript expression profile , reflected by progressive decreases in transcript levels toward the distal HIV+ intestine . We see an increase in mRNA levels as the location varies from proximal-to-distal for ZO-1 and claudin-2 in healthy males , which is reversed in HIV+ males . For occludin and claudin-4 , transcript levels are relatively constant from the terminal ileum to the descending colon of healthy males , while claudin-4 shows a progressive decrease toward the distal colon in HIV+ males . ZO-1 , occludin , claudin-2 , and claudin-4 median transcript levels are all decreased 1 . 5 to 2 . 9-fold in the descending colon of HIV+ males ( data not shown ) . We also recognize that HIV is a highly heterogeneous disease . While all of our patients are on ART , some of the patients did display a detectable viral load , which may influence intestinal permeability or GI function . With only four patients showing detectable viral load , this study was not powered to compare the fully virally-suppressed group to these four subjects , yet we could eliminate the impact of detectable viral load on TJ transcript expression by removing them from the analysis . Decreases with identical degrees of significance in all four TJ mRNA levels are observed in the colon of virally-suppressed HIV+ subjects ( Figure S2A ) , while transcript levels are not altered in the terminal ileum ( Figure S2B ) , replicating the observation for the entire HIV+ cohort . Upon examination of TJ expression levels along the proximal-to-distal axis of the colon in fully HIV-suppressed individuals , we found additional decreases in TJ mRNA expression in the ascending colon , which was not seen in the full HIV+ cohort ( 1 . 8-fold for ZO-1; 1 . 7-fold for occludin; 2 . 3-fold for claudin-2; 1 . 6-fold for claudin-4 ) . To verify the HIV-associated decrease in TJ transcript levels , we measured the protein levels of TJ components occludin , claudin-2 , and claudin-4 in the distal colon , the location that showed the greatest reduction in transcript levels in the HIV patient . Total protein lysates from the descending colon of virally-suppressed HIV+ individuals and healthy controls were subjected to immunoblotting ( Figure 6A ) followed by densitometric analysis . Equal loading of samples was verified using GAPDH and β-actin . To accurately quantify the density of each band , samples were electrophoresed twice , varying the loading amounts if needed to obtain a band intensity that fell within the linear range of detection . To examine the TJ protein levels in epithelial cells specifically , levels of each target protein were normalized to the epithelial cell-specific cytokeratin-18 protein , a major cytoplasmic intermediate filament protein expressed in one-layered internal epithelial tissue [48] . Normalized protein levels of occludin , claudin-2 , and claudin-4 all show a significant 3 . 1 to 3 . 2-fold decrease in the descending colon of the virally-suppressed HIV+ cohort ( Figure 6B ) compared to levels in controls , in agreement with the observed decrease in occludin , claudin-2 , and claudin-4 transcript levels in the distal colon of the HIV+ gut . As we described for TJ mRNA , median protein expression levels for occludin , claudin-2 , and claudin-4 for the entire HIV+ cohort , as well as in the male HIV+ subjects , are similarly reduced 2 . 6 to 3-fold in the HIV+ descending colon ( Figures S3 and S4 ) . Eliminating the limited number of HIV+ patients with diarrhea did not alter these results ( data not shown ) . To investigate whether the transcriptional modification observed in the ART-treated HIV+ population is specific to TJ components , we measured the transcript levels of two epithelial cell-specific proteins that are not part of the TJ complex , namely human beta defensin-3 ( hBD-3 ) and E-cadherin , along the proximal-to-distal gut . hBD-3 is an inducible , broad-spectrum anti-microbial peptide , produced by colonic enterocytes as an innate immune effector to protect against luminal pathogens [49] . E-cadherin is a transmembrane protein component of adherens junctions mediating cell-cell adhesion [50] . Analysis of non-TJ cellular components revealed differing patterns of HIV-associated transcription alteration ( Figure 7 ) . Similar to that observed for TJ transcripts , transcription of hBD-3 shows a significant down-regulation in the colon of HIV+ individuals ( p<0 . 05 ) , but examination of hBD-3 transcription at specific colonic locations revealed a distinct pattern ( Figure 7A ) . Transcription is relatively constant in the healthy control population along the ascending , transverse , and descending colon , while in the HIV+ population , although a significant hBD-3 transcriptional down-regulation is observed only in the distal descending colon ( p<0 . 05 ) , there is no significant linear trend between hBD-3 transcript level and anatomical location ( Figure 7A ) . E-cadherin , on the other hand , shows transcriptional up-regulation in the HIV+ colon ( p<0 . 01 ) , compared to levels in the uninfected colon , in stark contrast to other transcripts studied ( Figure 7B ) . Comparing between the healthy controls and the HIV+ population at each gut location , we observed a significant upregulation of E-cadherin transcript levels at the transverse ( p<0 . 06 ) and descending ( p<0 . 05 ) HIV+ colon . In the terminal ileum and ascending colon , E-cadherin transcript levels are unchanged . E-cadherin transcript levels do not vary with location in the healthy or HIV+ gut . ( Figure 7B ) . Identical results for hBD-3 and E-cadherin transcript levels in the colon are observed when only male donors or virally-suppressed populations were examined separately ( Figures S5 and S6 ) . It has been previously reported that circulating levels of the microbial cell wall constituent , LPS , is elevated in both ART-treated and untreated HIV+ patients [15]–[17] , and we hypothesize that the accessibility of translocated luminal microbial products is mediated by a decrease in intestinal TJ expression . To confirm that an inflammatory mediator indicative of LPS exposure is elevated in our HIV+ cohort , we measured plasma levels of soluble CD14 ( sCD14 ) , an LPS co-receptor that promotes its binding to Toll-like receptor-4 and is shed from activated monocytes . Levels of sCD14 were elevated in samples from HIV+ patients compared to healthy controls ( Figure 8A ) . Furthermore , there are inverse correlations that trend toward significance between levels of LPS or sCD14 and claudin transcript expression in the descending colon of both HIV+ and healthy control subjects ( r = −0 . 79 , p = 0 . 059 for claudin-4 vs . sCD14; and r = −0 . 76 , p = 0 . 073 for claudin-2 vs . LPS; Figures 8B and 8C ) , demonstrating a direct link between TJ gene expression in the distal colon and immune activation in HIV infection .
Microbial translocation from the gut , originating from the enormous quantity of intestinal commensal bacteria , is implicated as a major driver of the chronic systemic inflammation that not only predicts pathogenic HIV disease progression and poor response to ART , but , more importantly , may mediate the immunopathogenesis of non-AIDS morbidities , including cardiovascular , liver , and neurocognitive diseases , that shorten the life expectancy of long-term ART-treated HIV-infected individuals [51] , [52] . Elevated microbial translocation is attributed to the simultaneous effects of intestinal mucosal immunodeficiency and disruption to the epithelial barrier , a hypothesis confirmed in pathogenic SIV infection [41] . Structural damage to the intestinal epithelium has been demonstrated in ART-naïve HIV-infected population [53] , [54] , and we now provide the first direct molecular evidence of gut barrier breakdown in virally-suppressed HIV+ patients , corroborating our clinical data demonstrating persistence of increased intestinal permeability in the ART-treated population [40] . Intestinal epithelial disruption is restricted to the colon and manifests at the molecular level as a down-regulation of the TJ components ZO-1 , occludin , claudin-2 , and claudin-4 , via transcriptional control . The colonic epithelium remains grossly intact , and the packing and relative abundance of epithelial cells are maintained . Moreover , we observed a progressive decline in TJ expression along the proximal-to-distal axis of the HIV+ colon , in contrast to the relatively flat or increasing gradients observed in the healthy intestine . Finally , concurrent alterations in the transcriptional pattern of non-TJ epithelial-specific genes suggest that tight junctional down-regulation in the HIV+ gut occurs as part of an overall intestinal epithelial disruption through modified regulation of transcription . The dramatic and rapid depletion of CD4+ T cells from the intestinal mucosa during HIV and SIV infection led to the speculation that injury to the immune component of the intestinal lamina propria is permissive for increased translocation of microbial products into systemic circulation [18] . Indeed , mucosal immunodeficiency begins in the early phase of HIV or SIV infection and is characterized by a profound and selective depletion of CD4+ T cells within days of infection [55] and preferential loss of IL-22 and IL-17 producing T cells [31] , [56] , impaired neutrophil recruitment and macrophage phagocytic function [41] , and local mucosal inflammation [52] , [57] , [58] . Recent literature indicates that mucosal immunodeficiency and structural epithelial deterioration concurrently drive microbial translocation and HIV progression [59] . Intestinal epithelial disruption occurs in early SIV infection through epithelial cell apoptosis , secondary to interactions with the intestinal epithelial cell-associated alternative SIV coreceptor GPR15/Bob [60] . In humans , epithelial dysregulation begins in primary HIV infection and persists into the chronic phase , with down-regulation of genes involved in epithelial maintenance , growth and differentiation , as well as metabolic and digestive functions [34] , [61] . Our results , while highlighting the significant decrease in TJ mRNA and protein expression in chronic HIV infection , reveal a broader change in intestinal epithelial cell transcriptional regulation , even in the setting of viremic control . Our findings are in agreement with the notion of overall epithelial dysregulation in chronic HIV infection , rather than a specific disturbance in TJs , leading to structural deterioration of the epithelial barrier on the molecular level . Evidence for epithelial barrier disruption contributing to SIV pathogenesis has been demonstrated in the non-human primate model . A comprehensive survey of the entire colonic tissue revealed epithelial barrier breakdown in chronic ( non-AIDS and AIDS ) , untreated SIV-infected rhesus macaques , varying from multifocal colonic epithelial disruptions to epithelial loss and overt ulceration [41] . This breakdown in the epithelium is associated with in situ LPS infiltration into the lamina propria and local immune activation [41] . Our results demonstrating an intact intestinal epithelium at the endoscopic and light microscopic level with no decrease in epithelial barrier coverage in the HIV+ colon is more in agreement with a recent study showing microbial translocation in situ with little morphological evidence of human intestinal epithelial breaches [54] . Together these reports suggest distinct mechanisms between humans and primates for intestinal barrier loss . However , we acknowledge that the studies in humans are limited by a random sampling of the colonic mucosa via a restricted number of biopsy samples . Discrete sites of epithelial barrier loss or ulceration , not visible to the clinician , may have been missed in our study . Thus , we do not dismiss the possible contribution of epithelial barrier breakdown , at the cellular level , to HIV-associated microbial translocation , in addition to epithelial TJ down-regulation . While global intestinal epithelial cell function is compromised during HIV infection , in this report we focus on decreased TJ expression in the HIV+ intestinal epithelium as promoting the translocation of microbial products to the lamina propria and systemically , extending in vitro evidence that demonstrated TJ downregulation as a response of genital and intestinal mucosal epithelium to direct HIV-1 exposure [62] . In contrast , Smith et al . [54] reported increased levels of claudin-2 protein in the ileum and rectum during chronic HIV infection . Similarly , Epple et al . [53] demonstrated mucosal barrier defects in the duodenum of HIV-infected individuals including increased mannitol permeability , decreased claudin-1 , and increased claudin-2 protein expression . These earlier results were obtained from untreated viremic patients . Our findings are complementary , except for claudin-2 , and expand the current understanding of HIV-associated TJ disruptions in patients with effective viral suppression . While Epple et al . showed that duodenal mucosal barrier changes are reversed in the small intestine of antiretroviral-treated patients , we systematically probed for TJ disruptions along the length of the colon , highlighting the greatest down-regulation of TJs , both at the transcriptional and translational levels , in the most distal , descending portion of the colon . These distinct differences in response of the small and large intestines to HIV infection and ART highlight the contributions of apoptosis in the former and paracellular permeability in the latter [25] , [40] , [60] , suggesting unique pathologies of HIV damage along the alimentary canal , and must always be tempered by the sample size and heterogeneity of the populations studied . As noted above , our previous clinical study indicates that the increase in small intestinal permeability seen in virally-suppressed , HIV-infected individuals is primarily a result of epithelial cell damage . Consistent with these clinical results , in this report we find that the TJ complex is down-regulated only in the colon , not the small intestine . In the pre-antiretroviral era , advanced HIV disease was associated with small intestinal structural defects , including villous atrophy and crypt hyperplasia [63] . More recently , multiple studies have shown correlation between disease progression , circulating microbial products due to translocation , and plasma levels of intestinal fatty acid binding protein ( I-FABP ) [64]–[66] , which is a marker of small intestinal epithelial cell apoptosis [67] . These results collectively stress the importance of epithelial cell damage through apoptosis as the predominant mechanism for loss of small intestinal barrier integrity . We also present results pertaining to perturbations in the expression pattern of various epithelial-specific transcripts along the proximal-to-distal axis of the HIV and healthy gut , from the terminal ileum to the descending colon . The distally increasing pattern observed for ZO-1 , for instance , in the healthy population is consistent with findings from gene expression mapping along the normal colon identifying transcripts that are differentially expressed from proximal-to-distal segments , a subset of which demonstrates a gradual monotonic change in expression levels , characteristically an increase toward the distal colon [68] . The other subset identified consisted of transcripts with a dichotomous proximal versus distal colon expression pattern . We postulate that such gene expression profiles along the longitudinal axis of the gut is programmed by embryonic development and established by interactions with the external environment . The small and large intestines are distinct organs in the alimentary canal , with unique functions of digestion and absorption of nutrients , and reabsorption of water and electrolytes , respectively . Consecutive anatomical regions within the colon also perform distinct functions , with the proximal portion relatively more involved in solidification of fecal contents as compared to the distal portion , which is responsible for the transient storage of feces [69] . Indeed , such functional differences mirror and are determined by the various intrinsic differences between the terminal ileum , proximal , and distal colon in terms of embryologic origin , morphology , and proliferative capacity [70] . Interactions with luminal content , dependent on diet , and microbiota further modify the epithelium , the degree to which is dependent on colonic transit time , shown to be slowest in the proximal colon [71] . In addition , luminal microbiota shifts along the length of the colon in quantity and diversity [72] , and metaproteome analysis of the colonic mucosal-luminal interface demonstrate significant anatomic region-related differences in host-microbial interactions [73] . Such variations have pathological consequences , with the classic “two-colon concept” of colorectal carcinoma describing striking differences in clinical , molecular , and epidemiological features of tumors in the proximal and distal colon [74] , and more recent data revealing a gradual increase in the frequency of the CpG island methylator phenotype , microsatellite instability , LINE-1 methylation , as well as BRAF , KRAS , and PIK3CA mutations in tumors along the bowel at different colonic subsites [75] . We acknowledge that our observed HIV-associated intestinal TJ downregulation and gene expression pattern alterations are likely a result of complex interactions between the gut environment and intestinal epithelial cells . While we cannot fully address all of the sources of heterogeneity with this limited sample size , we sought to minimize the potential impact of identifiable confounders , while acknowledging that our methodology involves an appreciable number of multiple comparisons , thus reducing our statistical power . While there can be small differences in intestinal transcript expression patterns in females versus males [69] , our conclusions are maintained in age-matched male subjects in our cohorts , eliminating age and gender as confounding factors for the observed HIV-associated changes . Using a similar strategy , we were able to eliminate detectable viral load as a potential confounder . Gene expression profile along the colonic mucosa is modulated after colectomy surgery to correspond to the new ( proximal or distal ) location [69] , providing evidence that colonic transcript expression levels along the proximal-to-distal axis are responsive to pathophysiological perturbations or insults . In light of the importance of the specific microbiome composition on maintaining the gut's structural barrier , as well as local and systemic immunity [76] , alterations of the luminal microbiome ( dysbiosis ) in association with HIV infection [52] , [77] , [78] in a colonic subsite-specific manner would influence local epithelial function and transcriptional activity , resulting in differential proximal-to-distal TJ component expression patterns between the HIV+ and healthy colon . It is important to recognize that luminal and mucosa-associated enteric bacteria represent two distinct populations , with the mucosa-associated population displaying local heterogeneity [72] , [79] . Studies on stool microbiome reveal increased diversity and altered composition of microbiota in the HIV gut – changes that persist in the ART-treated patient [78] , [80] . Similarly , dysbiotic mucosal-adherent microbiota are observed in the ART-treated HIV-infected patient [81] , paving the way to a systematic study of dysbiosis along the longitudinal axis of the HIV gut , that may directly influence intestinal epithelial cell metabolism and function . In addition , expansion of the enteric virome , associated with a previously undescribed set of viruses , in pathogenic SIV infection [77] raises the possibility of viral contributions to HIV progression and intestinal pathology . HIV-associated enteric dysbiosis may also have consequences on the lamina propria T cell populations . Total bacterial load in the stool of HIV-infected subjects negatively correlates with duodenal T cell activation , while the proportions of Enterobacteriales and Bacteriodales are associated with duodenal CD4+ T cell loss [80] . Enrichment of gut bacteria that catabolize tryptophan via the kynurenine pathway [81] may inhibit the differentiation of Th17 cells , a T cell subset important in maintaining mucosal immunity shown to be depleted in the HIV lamina propria . Recent data from our laboratory demonstrate that T cell activation can modulate intestinal epithelial barrier permeability , suggesting possible contributions of lamina propria T cells on intestinal epithelial TJ regulation . In the ART era , non-AIDS associated complications are now a major clinical focus and concern . Systemic inflammation , possibly initiated by circulating microbial products from the gut lumen , is a likely contributing factor to HIV morbidity . Our demonstration of inverse correlations between markers of immune activation and TJ transcript levels and a proximal-to-distal gradient of decreased TJ expression in the HIV colon provide a needed mechanism for increased intestinal permeability in the well-controlled , ART-treated HIV-infected patient , which contributes to microbial translocation and systemic inflammation . | While antiretroviral therapy for HIV-infected patients is remarkably effective in suppressing viral replication and preventing progression to AIDS , treated patients still have a shorter life expectancy due to increased risks for non-AIDS associated morbidities . Recent data showed that these complications are associated with chronic systemic inflammation , and it is hypothesized that bacterial products breaching the intestinal barrier may cause the inflammation . It is known that HIV induces persistent intestinal mucosal immunodeficiency , but evidence for structural damage to the intestinal epithelium is lacking in the antiretroviral-treated patient population . Here , we characterized the intestinal epithelial damage that leads to increased intestinal permeability in this population . We found that while the colonic epithelial layer is intact microscopically , intercellular tight junctions ( TJ ) are down-regulated at the transcriptional and translational levels . We observed further that TJ transcripts progressively decrease along the proximal-to-distal HIV gut . Concurrent alterations in the levels of non-TJ epithelial transcripts suggest that epithelial cells in the HIV gut are transcriptionally dysregulated . Our data provide evidence that TJ disruption is a novel mechanism for increasing colonic permeability in the antiretroviral-treated HIV patient , which may then result in systemic inflammation and associated complications . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"hiv",
"immunopathogenesis",
"immunology",
"microbiology",
"epithelial",
"cells",
"immunodeficiency",
"viruses",
"gastroenterology",
"and",
"hepatology",
"digestive",
"system",
"infectious",
"diseases",
"medical",
"microbiology",
"hi... | 2014 | Progressive Proximal-to-Distal Reduction in Expression of the Tight Junction Complex in Colonic Epithelium of Virally-Suppressed HIV+ Individuals |
The extensive genetic regulatory flows underlying specification of different neuronal subtypes are not well understood at the molecular level . The Nplp1 neuropeptide neurons in the developing Drosophila nerve cord belong to two sub-classes; Tv1 and dAp neurons , generated by two distinct progenitors . Nplp1 neurons are specified by spatial cues; the Hox homeotic network and GATA factor grn , and temporal cues; the hb -> Kr -> Pdm -> cas -> grh temporal cascade . These spatio-temporal cues combine into two distinct codes; one for Tv1 and one for dAp neurons that activate a common terminal selector feedforward cascade of col -> ap/eya -> dimm -> Nplp1 . Here , we molecularly decode the specification of Nplp1 neurons , and find that the cis-regulatory organization of col functions as an integratory node for the different spatio-temporal combinatorial codes . These findings may provide a logical framework for addressing spatio-temporal control of neuronal sub-type specification in other systems .
The nervous system contains a myriad of different neuronal sub-types , and understanding cell fate specification remains a major challenge . Studies in a number of systems have revealed that neuronal subtype specification relies upon complex cascades of regulatory information , involving spatial and temporal selector genes [1] , onwards to terminal selector genes [2 , 3] , often acting in combinatorial codes [4–6] . With respect to spatial information , the Hox homeotic selector genes , expressed in distinct but partly overlapping domains along the antero-posterior axis of the central nervous system , have been extensively studied for their role in cell fate specification [reviewed in [7 , 8]] . With regard to temporal information , seminal studies in the Drosophila embryonic central nervous system ( CNS ) has identified a temporal cascade , where the sequential expression of the transcription factors Hunchback ( Hb ) , Kruppel ( Kr ) , Pdm2 and Nubbin ( collectively referred to as Pdm ) , Castor ( Cas ) and Grainy head ( Grh ) play out in most , if not all neuroblasts ( NBs ) [reviewed in [9]] . The temporal factors dictate the identity of neurons and glia being specified at different stages of NB lineage progression . Although not conserved in its entirety , research in mammals has pointed to similar temporal progressions , and begun identifying some of the factors involved [reviewed in [10]] . In addition , studies have revealed that the Hox spatial information can converge with temporal cues to thereby specify neuronal subtypes [11] . While these functional genetic studies have provided insight into the genetic mechanisms underlying neuronal subtype specification , it is largely unclear how the broader spatio-temporal cues are molecularly integrated to cause discrete terminal selector gene expression , and how terminal selectors feed forward to final cell identity . The Drosophila ventral nerve cord ( VNC; defined here as thoracic segments T1-T3 and abdominal A1-A10 ) contains ~10 , 000 cells at the end of embryogenesis , which are generated by a defined set of ~800 neuroblasts ( NBs ) [12–16] . The Apterous neurons constitute a small sub-group of interneurons , identifiable by the selective expression of the Apterous ( Ap ) LIM-homeodomain factor , as well as the Eyes absent ( Eya ) transcriptional co-factor and nuclear phosphatase ( Fig 1A ) [17 , 18] . A subset of Ap neurons express the Nplp1 neuropeptide , but can be sub-divided into the lateral thoracic Tv1 neurons , part of the thoracic Ap cluster of four cells , and the dorsal medial row of dAp neurons ( Fig 1A ) [6 , 19] . In line with the distinct location of the Tv1 and dAp neurons , studies have revealed that they are generated by distinct NBs; NB5-6T and NB4-3 , respectively [20 , 21] . A number of studies have addressed the genetic mechanisms underlying the specification of the Tv1 and dAp neurons , and the regulation of the Nplp1 neuropeptide . These have revealed that two distinct spatio-temporal combinatorial transcription factor codes , one acting in NB5-6T and the other in NB4-3 , converge on a common initiator terminal selector gene; collier ( col; Flybase knot ) , encoding a COE/EBF transcription factor ( Fig 1B ) [20–22] . Col in turn is necessary and sufficient to trigger a feed forward loop ( FFL ) consisting of Ap , Eya and the Dimmed ( Dimm ) bHLH transcription factor , which ultimately activates the Nplp1 gene [6] . Strikingly , the combinatorial coding selectivity of the spatio-temporal cues combined with the information-coding capacity of the FFL results in the selective activation of Nplp1 in only 28 out of the ~10 , 000 cells within the VNC . While these genetic studies have helped resolve the regulatory logic of this cell specification event , they have not addressed the molecular mechanisms by which the two different spatio-temporal combinatorial codes intersect upon the col initiator terminal selector , to trigger a common terminal FFL , or the molecular nature of the FFL . To address this issue , we have identified enhancers for Tv and dAp neuron expression for the genes in the common Tv1/dAp FFL: col , ap , eya , dimm and Nplp1 . We generated transgenic reporters for these enhancers , both wildtype and mutant for specific transcription factor binding sites , to test their regulation in mutant and misexpression backgrounds . We also used CRISPR/Cas9 technology to delete these enhancers in their normal genomic location to test their necessity for gene regulation . Strikingly , we find that the distinct upstream spatio-temporal combinatorial codes , which trigger col expression in Tv1 versus dAp neurons , converge onto different enhancer elements in the col gene . Hence , the col Tv1 neuron enhancer is triggered by Antp , hth , exd , lbe and cas , while the dAp enhancer is triggered by Kr , pdm and grn . In contrast to this subset-specific enhancer set-up for col activation , the subsequent , col-driven Nplp1 FFL feeds onto common enhancers in each downstream gene . These findings reveal that distinct spatio-temporal cues , acting in different neural progenitors , can trigger the same FFL by converging on discrete enhancer elements in an initiator terminal selector , to thereby dictate the same ultimate neuronal subtype cell fate .
The Ap neurons constitute a set of interneurons in the Drosophila VNC , out of which the thoracic lateral Tv1 neurons and the dorso-medial dAp neurons express the Nplp1 neuropeptide ( Fig 1A ) [6 , 17 , 19] . Tv1 neurons are generated by NB5-6T , while dAp neurons arise from NB4-3 [6 , 21] . Activation of Nplp1 in Tv1 and dAp neurons is controlled by a shared coherent FFL , consisting of col , ap , eya and dimm , where col is both necessary and sufficient to trigger the FFL [6 , 21] . In contrast , this common FFL is triggered by two different upstream spatio-temporal combinatorial codes , acting in the two different NBs . In NB5-6T this includes the temporal gene castor ( cas ) , the Hox homeotic gene Antennapedia ( Antp ) , the two Hox co-factor genes homothorax ( hth ) and extradenticle ( exd ) , as well as the homeobox gene ladybird early ( lbe ) . In NB4-3 , this includes the temporal genes Kruppel ( Kr ) and pdm2/nub ( pdm ) , as well as the GATA gene grain ( grn ) , ( Fig 1B ) [20–22] . To identify the cell-specific cis-regulatory modules ( CRMs ) that act as enhancers for the five genes in the dAp/Tv1 FFL , we analyzed expression of a number of transgenic lines generated in previous studies [17 , 23 , 24] , as well as an eya-CRM-Gal4 transgene ( provided by T . Lian and D . W . Allan; S1 Fig ) . This resulted in identification of fragments capable of driving reporter gene expression in the Tv1 and dAp neurons . To facilitate mutagenesis of CRMs , we attempted to identify smaller genomic fragments that retained appropriate activity . This resulted in the identification of smaller ( 1–2 kilobases ) CRMs for all genes with the exception of col , where larger fragments were required for proper expression ( Fig 1C–1S , S1 Fig , S1 Data ) . Strikingly , we found that while one enhancer region was sufficient to recapitulate Tv1 and dAp expression of ap , eya , dimm and eya , for col we identified two distinct enhancers , one each for expression in dAp or Tv1 neurons ( Fig 1C–1E and 1T ) . Enhancer studies have revealed that some genes may be controlled by several enhancers with partially redundant function , such as ‘shadow enhancers’ , which act to ensure high-fidelity in gene expression [25] . These shadow enhancers have been identified in a growing number of genes , in particular early developmental regulators [26 , 27] . We wanted to address the importance of the identified enhancers within the context of their normal genomic location . To this end , we used CRISPR/Cas9 technology , with two spaced gRNAs , to delete each of the identified enhancers in the FFL ( Fig 2A; Materials and Methods; S2 Data ) [28] . Focusing on col first , we analyzed the col-dAp-CRM ( generating the colΔdAp-CRM deletion mutant ) and observed that deletion of this enhancer resulted in significant loss of Col , Eya and Nplp1 expression , but as anticipated only in dAp and not in Tv1 neurons ( Figs 2D , 2I , 2J , S2C and S2F–S2H ) . In contrast , we found that deletion of the col-Tv-CRM ( colΔTv-CRM ) did not result in any effect upon Eya or Nplp1 , in either dAp or Tv1 neurons ( Fig 2B , 2C , 2I and 2J ) . We furthermore did not observe any effects on Col expression itself , either within the Ap cluster at AFT or globally ( S2A , S2B , S2D , S2E and S2G Fig ) . As anticipated , we also did not observe any effects on Eya , Nplp1 or Col expression in dAp neurons ( Figs 2B , 2C , 2I , 2J , S2E and S2H ) . Given the specificity of this element when placed in a promoter-lacZ transgenic construct ( Fig 1D and 1J ) , we found this lack of effect surprising . This prompted us to analyze the NB5-6T neuroblast at St14 , right after the onset of endogenous Col expression in this lineage . Measuring Col expression levels in the colΔTv-CRM mutants we did indeed observe a minor but significant reduction in expression ( Fig 2K–2M ) . Next , we analyzed eya , ap , dimm and Nplp1 enhancer deletions ( eyaΔCRM , apΔapS-CRM , dimmΔCRM , Nplp1ΔCRM ) , and observed that all exhibited strong effects . Specifically , as anticipated , all enhancer deletions resulted in significant reduction or loss of expression of the targeted gene , in both Tv1 and dAp neurons ( Figs 2E , 2H and S2I–S2X ) . Moreover , in line with the previous genetic analysis that identified an eya/ap ->dimm ->Nplp1 FFL , deletion of the eya , ap or dimm enhancers all significantly reduced Nplp1 expression ( Figs 2E–2J , S2I–S2N and S2Q–S2U ) . Also in line with this FFL , deletion of ap or eya enhancers did not affect one another’s expression ( Figs 2F and S2J–S2P ) . Within the Ap cluster , while deletion of the eya enhancer affected Nplp1 expression in Tv1 , we did observe Eya expression in two cells in the cluster ( S2J Fig ) . However , analysis at AFT , using Col as a specific Tv1 marker at this stage , revealed that Eya expression was lost from the Tv1 neuron , hence explaining the strong effect on Nplp1 in eyaΔCRM mutants ( S3A and S3B Fig ) . In line with the previous genetic analysis , dimm enhancer deletion did not affect either Eya or Ap expression ( Figs 2H–2J , S2R and S2U ) . Finally , deletion of the Nplp1-CRM did not affect expression of Eya , Ap or Dimm ( Figs 2H–2J , S2S and S2V–S2X ) . We conclude that activation of col in dAp neurons strongly depends upon the col-dAp-CRM element , while in contrast , expression of col in Tv1 neurons may operate via several enhancers , some of which presumably must reside outside of the col-Tv-CRM . In contrast , for the postmitotically expressed terminal selectors ap , eya and dimm , as well as the Nplp1 neuropeptide gene , their expression in Ap neurons appears to be critically dependent upon one discrete enhancer element . Having identified necessary and sufficient enhancers for the genes in the FFL , we proceeded to address the putative molecular connections between the upstream spatio-temporal cues and col , as well as between the FFL genes . This was approached by testing the enhancer transgenes in the pertinent mutant backgrounds , as well as mutating relevant candidate binding sites within each enhancer . Focusing on the col Tv enhancer first , we introduced the col-Tv-CRM-lacZ transgene into the Antp , cas , hth and lbe mutant backgrounds . This resulted in significant reduction of expression in all four cases , when compared to the enhancer transgene in a wild type background ( control ) stained on the same slide , with the strongest effect in cas mutants , which displayed a near-complete loss of expression ( Fig 3A–3F ) . Next , we mutated conserved DNA-binding sequences for Antp , Cas , Hth and Exd within the col-Tv-CRM-lacZ , and integrated these into the same genomic location as the wild type transgenic construct ( Fig 3M; Materials and Methods; S3 Data and S4 Data ) . We assayed β-gal expression in NB5-6T at St14 , and found that all of the four mutated enhancer transgenes displayed reduced expression , when compared to the enhancer transgene in a wild type background ( control ) stained on the same slide ( Fig 3G–3M ) . Next we turned to the col-dAp-CRM-GFP enhancer and introduced it into the Kr , pdm and grn mutant backgrounds . We observed significant reduction in GFP expression in all three mutants , when compared to the enhancer transgene in a wild type background ( control ) stained on the same slide ( Fig 4A–4E ) . The loss of GFP expression in the dAp cells , was accompanied by the loss of Eya expression . Next , we mutated all possible binding sites , conserved and non-conserved , for Kr , Pdm ( POU-HD ) and Grn ( GATA ) ( Fig 4L; Materials and Methods; S3 Data and S4 Data ) . We integrated these mutant transgenes into the same genomic location as the wild type transgenic construct and assayed the expression of GFP expression in dAp neurons at stage AFT . We found that the enhancers mutated for Kr or Pdm displayed reduced number of dAp cells expressing GFP , when compared to the enhancer transgene in a wild type background ( control ) ( Fig 4F–4H and 4J ) . The enhancer transgene mutated for Grn sites did not show a numerical loss of GFP expressing dAp cells , but did however show a significantly reduced level of expression in these cells ( Fig 4I and 4K ) . For analyzing the ap enhancer , we focused on the smaller apS2J-CRM-lacZ transgene , and placed this in the mutant background for Antp , lbe and col . Focusing on the Tv1 neurons , we observed significant reduction in β-gal expression in all three mutants ( Antp , lbe and col ) , when compared to the enhancer transgene in a wild type background ( control ) stained on the same slide ( Fig 5A–5E ) . As anticipated from the selective role of Antp , acting in the thorax , and lbe , in NB5-6T , the dAp neurons were only reduced in the thorax in Antp , and unaffected in lbe ( Fig 5A–5D and 5F ) . In contrast , col mutants affected β-gal expression in both Tv1 and dAp neurons ( Fig 5D–5F ) . Next , we mutated all conserved binding sites for Q50-Homeodomain proteins ( TAAT; affecting both Antp and Lbe ) , Col and Exd ( Fig 5M; Materials and Methods; S3 Data and S4 Data ) . We integrated these mutant transgenes into the same genomic location as the wild type transgenic construct and assayed the expression of β-gal expression in Tv1 and dAp neurons at stage 16 . We found that all three mutated enhancers displayed reduced number of Tv1 and dAp cells expressing β-gal , when compared to the enhancer transgene in a wild type background ( control ) ( Fig 5G–5L ) . We did not analyze the involvement of hth on the ap-CRM ( or eya-CRM ) because previous studies revealed that hth mutants could be fully rescued by re-expression of col [22] . exd mutants must be analyzed both as maternal and zygotic mutants , and we did not attempt to introduce the ap- and eya-CRM transgenes into such backgrounds . Similar to the ap enhancer analysis , we placed the eya-CRM-GFP enhancer in the mutant background for Antp , lbe and col . Focusing on the Tv1 neurons , we observed significant reduction in GFP expression in all three mutants , when compared to the enhancer transgene in a wild type background ( control ) stained on the same slide ( S4A–S4E Fig ) . As anticipated from the selective role of Antp , acting in the thorax , and lbe in NB5-6T , the dAp neurons were only reduced in the thorax in Antp , and unaffected in lbe ( S4A–S4D and S4F Fig ) . In contrast , col mutants affected GFP expression in both Tv1 and dAp neurons ( S4D–S4F Fig ) . Next , we mutated all conserved and non-conserved binding sites for Q50-Homeodomain proteins ( TAAT; affecting both Antp and Lbe ) , Col and Exd ( S4N Fig; Materials and Methods; S3 Data and S4 Data ) . We integrated these mutant transgenes into the same genomic location as the wild type transgenic construct and assayed the expression of GFP expression in Tv1 and dAp neurons at stage AFT . We found that the enhancers mutated for Hox and Col sites displayed reduced expression in both Tv1 and dAp cells , when compared to the enhancer transgene in a wild type background ( control ) ( S4G–S4I , S4K and S4L Fig ) . In contrast , the mutation of Exd sites had no effect in Tv1 neurons , and surprisingly showed up-regulation in dAp neurons ( S4J–S4L Fig ) . For analyzing the dimm enhancer , we placed the dimm-CRM-GFP transgene in the mutant backgrounds for Antp , ap , col and eya . We observed significant reduction of GFP expression in both Tv1 and dAp neurons in all four mutants , when compared to the enhancer transgene in a wild type background ( control ) stained on the same slide ( S4A–S4G Fig ) . Next , we mutated all conserved and non-conserved binding sites for Q50-Homeodomain proteins ( TAAT; affecting both Ap and Antp ) , Col and Exd ( S5O Fig; Materials and Methods; S3 Data and S4 Data ) . We integrated these mutant transgenes into the same genomic location as the wild type transgenic construct and assayed the expression of GFP expression in Tv1 and dAp neurons at stage AFT . We found that enhancer mutants for Hox and Exd displayed reduced GFP expression in both Tv1 and dAp cells , when compared to the enhancer transgene in a wild type background ( control ) ( S5H–S5N Fig ) . In contrast , the Col mutant enhancer showed slightly elevated expression in Tv1 neurons while expression was reduced in dAp neurons ( S5J and S5L–S5N Fig ) . For analyzing the Nplp1 enhancer , we placed the Nplp1-CRM-GFP transgene in the mutant backgrounds for col , ap , eya and dimm . We observed significant reduction of GFP expression in all four mutants , when compared to the enhancer transgene in a wild type background ( control ) stained on the same slide ( Fig 6A–6G ) . Next , we mutated all possible conserved and non-conserved binding sites for Q50-Homeodomain proteins ( TAAT; affecting Ap ) , Col and Dimm ( E-boxes ) ( Fig 6P; Materials and Methods; S3 Data and S4 Data ) . We integrated these mutant transgenes into the same genomic location as the wild type transgenic construct and assayed the expression of GFP expression in Tv1 and dAp neurons at stage AFT . We observed that the enhancers mutated for Hox or Dimm sites displayed reduced GFP expression in both Tv1 and dAp cells , when compared to the enhancer transgene in a wild type background ( control ) ( Fig 6H , 6I and 6K–6O ) . In contrast , the enhancer mutated for Col sites did not display any effect on GFP expression ( Fig 6J , 6L and 6M ) . Previous studies have revealed that combinatorial misexpression of the transcription factors in the Tv1/dAp cascade is able to broadly activate the genes in the FFL [6 , 18 , 21 , 29 , 30] . To determine if such combinatorial ectopic effects could act upon the identified enhancers taken out of genomic context , we misexpressed various combinatorial codes of TFs and studied the effects on the pertinent transgenes . Focusing on the apS2-CRM-lacZ and eya-CRM-GFP , we find broad activation of both transgenes when lbe and col are co-misexpressed ( Fig 7A–7D ) . Similarly , combinatorial misexpression of ap , eya and col could ectopically activate the dimm-CRM-GFP transgene ( Fig 7E and 7F ) . Finally , the Nplp1-CRM-GFP transgene was broadly activated by combinatorial expression of ap , dimm , eya and col ( Fig 7G and 7H ) . In all cases , as anticipated , we observed up-regulation of the endogenous Eya , Dimm and Nplp1 proteins ( Fig 7A–7H ) . These results demonstrate that ectopic activation of the dAp/Tv1 transcriptional program can robustly act upon the identified enhancers even outside their normal genomic context .
We find that the two different spatio-temporal programs converge on col , but on different enhancer elements . However , neither enhancer element gave complete null effects when deleted . Specifically , the 6 . 3kb col-Tv-CRM shows robust reporter expression , overlaps with endogenous col expression , responds to the upstream mutants , and is affected by TFBS mutations . However , when deleted ( generating the colΔTv-CRM mutant ) , it had weak effects upon endogenous col expression in NB5-6T , and no effect upon Eya and Nplp1 expression . Deletion of the col-dAp-CRM ( generating the colΔdAp-CRM mutant ) , gave more robust effects with reduction of Col , Eya and Nplp1 in dAp cells , although the expression was not lost completely . Early developmental genes , which often are dynamically expressed , may be controlled by multiple enhancer modules , to thereby ensure robust onset of gene expression . This has been reported previously in studies of early mesodermal and neuro-ectodermal development , in which several genes i . e . , twist , sog , snail are controlled by multiple distal enhancer fragments , so called “shadow enhancers” , in order to ensure reliable onset of gene expression [25] . The shadow enhancer principle is also supported by recent findings on the Kr gene [27] . Moreover , extensive CRM transgenic analysis , scoring thousands of fragments in transgenic flies , has also supported the shadow enhancer idea , revealing that a number of early regulators , several of which encode for transcription factors , indeed have shadow enhancers [26] . The dichotomy between the col transgenic reporter results and the partial impact on col expression upon deletion of its Tv1 and dAp enhancers , gives reason to speculate that col may be under control of additional enhancers , some of which may be referred to as shadow enhancers . The results on the eya , ap , dimm and Nplp1 enhancer mutants stand in stark contrast to the col CRMs findings . For these four genes , the enhancer deletion resulted in robust , near null effects , on expression . It is tempting to speculate that our findings , combined with previous studies , points to a different logic for early regulators , with highly dynamic patterns , requiring several functionally overlapping enhancers for fidelity , and late regulators and terminal differentiation genes , which may operate with one enhancer that is inactive until the pertinent combinatorial TF codes have been established . Analysis of the ap and eya enhancers indicates that Col directly interacts with these enhancers . Both of these enhancer-reporter transgenes are affected in col mutants , and can be activated by ectopic col . Moreover , mutation of one Col binding site in the ap enhancer and two sites in the eya enhancer , was enough to dramatically reduce enhancer activity . Direct action of Col on ap and eya is furthermore supported by recent data on Col genome-wide binding , using ChIP , which demonstrated direct binding of Col to these regions of ap and eya in the embryo [34] . The regulation of ap is an excellent example of the complexity of gene regulation , and studies have identified additional enhancers controlling ap expression in the wing , muscle and brain [35–38] . In contrast to regulation of ap and eya , a direct action of Col on dimm and Nplp1 is less clear . Analysis of the dimm and Nplp1 enhancers did not reveal perfectly conserved Col binding sites . Mutation of multiple non-perfect Col binding sites in the dimm enhancer did not affect reporter expression in the Ap cluster , but did however reduce levels in the dorsal Ap cells . Mutation of non-perfect Col binding sites in the Nplp1 enhancer had no impact on enhancer activity , neither in Tv1 nor dAp . These findings support a model where Col is crucial for directly activating ap and eya , which in turn directly activate dimm and Nplp1 , with some involvement of Col on dimm ( Fig 8 ) . However , support for a direct role for Col on Nplp1 comes from RNAi studies in larvae or adult flies , showing that knockdown of col resulted in loss of Nplp1 , while Ap , Eya and Dimm expression was unaffected [6 , 39] . It is tempting to speculate that Col regulates Nplp1 not via direct interaction with its enhancer , but rather as a chromatin state modulator , keeping the chromatin around the Nplp1 locus in an accessible state , in order for Dimm , Ap and Eya to be able to access the Nplp1 gene . Support for this notion comes from studies on the mammalian Col orthologue EBF , which is connected to the chromatin remodeling complex SWI/SNF during EBF-mediated gene regulation in lymphocytes [40] . Moreover , the central SWI/SNF component Brahma was recently identified in a genetic screen for Ap cluster neurons , and found to affect FMRFa neuropeptide expression in Tv4 without affecting Eya expression , indicating a late role in Ap cluster differentiation [41] . Alternatively , Col may activate Nplp1 via unidentified , low affinity sites , similar to the mechanism by which Ubx regulates some of its embryonic target genes [42] . Col activates the ap and eya genes . ap encodes a LIM-HD protein , a family of transcription factors well known to control multiple aspects of terminal neuronal subtype fate , including neurotransmitter identity , axon pathfinding and ion channel expression [17 , 43–45] . Our results indicate that Ap in turn acts upon dimm , and subsequently with Dimm on Nplp1 . eya encodes an evolutionary well-conserved phosphatase and does not bind DNA directly , instead acting as a transcriptional co-factor [46] . Eya ( and its orthologues ) have been found to interact with several transcription factors in different systems [46] , but whether it forms complexes with Col and Ap is not known . The final transcription factor in the FFL is Dimm , a bHLH protein . Dimm is selectively expressed by the majority of neuropeptide neurons in Drosophila , and is important for expression of many neuropeptides [6 , 19 , 29 , 47 , 48] . Intriguingly , Dimm is also both necessary and sufficient to establish the dense-core secretory machinery , found in neuropeptide neurons [29 , 48–52] . Based upon these findings Dimm has been viewed as a cell type selector gene [1] or a “scaling factor” [53] , acting to up-regulate the secretory machinery . Here , we find evidence for that Dimm acts directly on the Nplp1 enhancer , and this raises the possibility that Dimm is both a selector gene for the dense-core secretory machinery , and can act in some neuropeptide neurons to directly regulate specific neuropeptide gene expression .
Location marker lines: lbe ( K ) -EGFP [54] . lbe ( K ) -lacZ ( provided by K . Jagla ) [55] . Mutant lines: lbe12C005 ( BL#59385 ) . Df ( lbl-lbe ) B44 ( provided by K . Jagla ) . Antp12 [56] ( provided by F . Hirth ) . Antp25 ( BL#3020 ) . casΔ1 and casΔ3 [57] ( provided by W . Odenwald ) . col1 , col3 [58] [59] ( provided by A . Vincent ) . hth5E04 ( BL#4221 ) . Df ( 3R ) Exel6158 ( BL#7637; referred to as hthDf7637 ) . Df ( 2L ) BSC354 ( referred to as eyaDf ) ( BL#24378 ) . eyacli-IID ( BL#3280 ) . dimmrev4 and dimmP1 ( provided by Douglas W . Allan ) . apP44 [60] . Misexpression lines: UAS-col [59] ( provided by A . Vincent ) . UAS-col-HA and UAS-myc-lbe [30] . UAS-dimm [47] . UAS-eya ( BL#5675 ) . UAS-ap [61] . Driver lines: elavC155 = elav-Gal4 ( BL#458 ) . CRM lines: col-dAp-CRM ( chr . 2:28E7; chr . 3:68A4 ) . col-Tv-CRM ( chr2:25C7 ) . apS2-CRM ( chr2:22A3; chr . 3:62E1 ) . apSJ2-CRM ( chr2:28E7 ) . eya-CRM ( chr . 2:28E7; chr . 3:68A4 ) . dimm-CRM ( chr . 2:28E7; chr . 3:68A4 ) . Nplp1-CRM ( chr . 2:28E7; chr . 3:68A4 ) . CRM mutant lines: colΔTv-CRM , colΔdAp-CRM , eyaΔCRM , apΔapS-CRM , dimmΔCRM , Nplp1ΔCRM . gRNA-lines: vas-Cas9 ( BL#51323 ) . apΔapS-CRM gRNAs ( chr . 3: 68A4 ) . colΔdAp-CRM gRNAs ( chr . 2: 28E7 ) . colΔTv-CRM gRNAs ( chr . 3 68A4 ) . eyaΔCRM gRNAs ( chr . 2:28E7 ) . dimmΔCRM gRNAs ( chr . 2: 28E7 ) . Nplp1ΔCRM gRNAs ( chr . 2: 28E7 ) . Mutants were maintained over GFP- or YFP-marked balancer chromosomes . As wild-type control OregonR was used . Staging of embryos was performed according to Campos-Ortega and Hartenstein [62] . Primary antibodies were: Guinea pig α-Deadpan ( 1:1 , 000 ) and rat α-Dpn ( 1:200 ) [54] . Guinea pig α-Col ( 1:1 , 000 ) , guinea pig α-Dimm ( 1:1 , 000 ) , chicken α-proNplp1 ( 1:1000 ) [6] . Rat α-Nab ( 1:500 ) [20] . mAb α-Eya 10H6 ( 1:250 ) ( Developmental Studies Hybridoma Bank , Iowa City , IA , US ) . Rabbit α-Ap ( 1:1 , 000 ) [35] ( provided by D . Bieli and M . Affolter ) . Chicken α-GFP 1:1 , 000 ( Abcam , ab13970 ) . In brief; all wild-type enhancers , were either PCR amplified ( Expand High FidelityPlus PCR system , from Roche Diagnostics ( Indianapolis , IN , USA ) from the OregonR DNA , or de-novo synthetized at GenScript Inc . ( Piscataway , NJ , USA ) in the case of the col-Tv-CRM enhancer and all other mutant enhancer versions . PCR amplified DNA fragments were cloned into the pCR2 . 1-TOPO® TA vector according to the manufactures protocol ( Invitrogen Life technologies , Carlsbad , CA , USA ) for further cloning steps into the placZ . attB or pEGFP . attB landing site vectors [63] ( provided by K . Basler and J . Bischof ) . Furthermore TOPO clones containing the wild type enhancer sequences were sent to GATC Biotech AG ( Cologne , Germany ) for Sanger sequencing . All synthesized enhancer constructs were delivered in a pUC57 vector and subsequently cloned either into the placZ . attB or pEGFP . attB landing site vectors , and integrated into the fly genome via site directed phiC31 mediated integration [64] at BestGene Inc ( Chino Hills , CA , USA ) or GenetiVision ( Houston , TX , USA ) . The online tool ( http://tools . flycrispr . molbio . wisc . edu/targetFinder/ ) was used to design two protospacers with zero predicted off-targets for each CRM , flanking the 5`and 3`regions of the identified enhancer constructs . Sequences for all protospacers can be found in the supplemental information ( CRISPR ) . Primer design and vector assembly was done according to the protocol found at http://www . crisprflydesign . org/wp-content/uploads/2014/06/Cloning-with-pCFD4 . pdf . PCR was performed using the Expand High FidelityPlus PCR system ( Roche Diagnostics , Indianapolis , IN , USA ) according to the provided protocol with an annealing temperature of +61°C . In order to delete the CRMs identified in this study , the tandem gRNA vector ( pCFD4-U6:1_U6:3 ) ( Addgene # 49411; gift from Simon Bullock ) was used to express two gRNAs simultaneously , which flank the 5´ and 3´ regions of the CRMs . The empty vector served as a template during PCR amplification to introduce the protospacers into the gRNA core sequence and U6-1 and U6-3 promoter regions . PCR products containing the protospacers were cloned into the tandem gRNA vector by ligation independent cloning using Gibson Assembly according to the manufacturers’ protocol ( New England Biolabs Inc . , Ipswich , MA , USA ) . All gRNA vector constructs were Sanger sequenced by use of the M13 for and M13 reverse primers to confirm for the correct insert ( GATC Biotech AG , Cologne , Germany ) . Stable transgenic gRNA flies were generated at BestGene and tandem gRNA constructs containing attB landing sites were landed via phi31 mediated integration on the second or third chromosome on cytolocation 28E7 and 68A4 . Fly stocks mutant for CRMs were created by crossing males of the transgenic tandem gRNA flies to virgins of vas-Cas9 ( BL#51323 ) . Stable stocks mutant for CRMs were tested by PCR by using PCR primers flanking the deleted region . PCR fragments spanning the deleted region were sequenced to confirm deletion ( Supplemental Information CRISPR ) . Zeiss LSM 700 Confocal microscopes were used for fluorescent images; confocal stacks were merged using LSM software or Adobe Photoshop . Statistic calculations were performed in Graphpad prism software ( v4 . 03 ) . Cell counts and reporter ( GFP or β-gal ) measurements were done with ImageJ FIJI and numbers transferred to Graphpad prism . To address statistical significance Student's t-test or in the case of invariant cell numbers contingency tables together with Chi-Square test were used . Images and graphs were compiled in Adobe Illustrator . | The nervous system contains a myriad of different cell types . These are specified by elaborate transcription factor cascades , starting with early factors that provide spatial and temporal information , to late factors that dictate final cell identity . The molecular nature of such cascades is poorly understood in any system . We focus on two related neuropeptide neurons in the Drosophila central nervous system , for which an extensive genetic pathway has been identified . We identify the enhancers for the different genes in the cascade , and conduct an extensive molecular analysis of these . Our findings reveal that different spatial and temporal cues converge on different enhancers of a key initiator terminal selector gene , which then triggers a feedforward cascade of sequential enhancer activation , ultimately landing on the enhancer of the neuropeptide gene . These findings may point to general mechanisms underlying specification of unique neuronal cell fate in many systems . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"cell",
"binding",
"cell",
"physiology",
"invertebrates",
"gene",
"regulation",
"regulatory",
"proteins",
"dna-binding",
"proteins",
"neuroscience",
"animals",
"animal",
"models",
"developmental",
"biology",
"drosophila",
"melanogaster",
"model",
"organisms",
"experimental"... | 2017 | Neuronal cell fate specification by the molecular convergence of different spatio-temporal cues on a common initiator terminal selector gene |
Metagenomic sequencing projects from environments dominated by a small number of species produce genome-wide population samples . We present a two-site composite likelihood estimator of the scaled recombination rate , ρ = 2Nec , that operates on metagenomic assemblies in which each sequenced fragment derives from a different individual . This new estimator properly accounts for sequencing error , as quantified by per-base quality scores , and missing data , as inferred from the placement of reads in a metagenomic assembly . We apply our estimator to data from a sludge metagenome project to demonstrate how this method will elucidate the rates of exchange of genetic material in natural microbial populations . Surprisingly , for a fixed amount of sequencing , this estimator has lower variance than similar methods that operate on more traditional population genetic samples of comparable size . In addition , we can infer variation in recombination rate across the genome because metagenomic projects sample genetic diversity genome-wide , not just at particular loci . The method itself makes no assumption specific to microbial populations , opening the door for application to any mixed population sample where the number of individuals sampled is much greater than the number of fragments sequenced .
Microbial populations exchange homologous genetic material at different rates , dramatically affecting the evolutionary potential of the population . While basal mutation rates can be estimated via long-term within-laboratory evolution experiments [1] , recombination rates are more difficult to infer because they require identification of multiple alleles at multiple loci in multiple individuals . Further , biogeographic barriers and interspecies interactions may lead to qualitatively different effects than growth in axenic laboratory culture , making determination of recombination rates in an organism's natural environment critical to accurate interpretation [2] . For the purpose of this study , we ignore the mechanism behind homologous recombination ( i . e . transformation , transduction , or conjugation ) and focus on its effect on genetic diversity . Much research has investigated human recombination hotspots [3] , yet almost nothing is known about variation in microbial recombination rates within a genome . In specific instances , however , studies have experimentally identified sequence motifs associated with recombination hotspots in some species of bacteria and yeast [4] . Mounting evidence suggests that regions known as CRISPR ( Clusters of Regularly Interspaced Short Palindromic Repeats ) form the basis of a bacterial immune system against phage in which chunks of the phage genome are inserted into the CRISPR region [5] . Thus a reasonable hypothesis would be that these regions or other regions with similar effect might recombine with greater frequency than the rest of the genome . Inference of a genome-wide , fine-scale recombination map requires both extensive genome-wide sampling of the genetic diversity within the population of interest as well as an appropriate population genetic model , neither of which has been previously available for microbial populations . Microbial population surveys have primarily sequenced a small number of loci ( “multi-locus sequence typing” ) [6] , which yield no information about variation in local recombination rate . Current methods tailored to microbial populations rely on low-power summary statistics [7] , [8] , heuristics instead of explicitly modeling the source of the recombining fragments [9] , or parsimony based on manual inspection [10] . A few studies ( e . g . [2] , [11] ) applied a more rigorous likelihood-based approach using a population genetic model ( [12]; discussed more below ) , but these were still able to estimate only a genome-wide average rate of recombination . Recently , large-scale metagenomic sequencing projects have begun to generate genome-wide population samples by sequencing random reads from a pool of DNA extracted from all microorganisms in a given environment . Projects that sample environments dominated by only a few microbial “species” are able to assemble near-complete genomes [13] , [14] , in which the constituent reads contain information about the genetic diversity in the population . Considering the large number of individuals in the sampled community relative to the number of reads sequenced , each read derives almost certainly from a different individual microorganism . With average read depths as high as ten [14] , the resulting data hold rich potential for population genetic analysis [15] , [16] . Given these data , we can make inferences about parameters such as mutation rate and recombination rate . In population genetic theory , the per-generation mutation rate , , and per-generation recombination rate , , almost always appear in conjunction with the effective population size , , as the parameters and . In our microbial context , we assume a single recombination event leads to the replacement of a short tract of sequence , creating two recombination breakpoints . A full likelihood method would yield maximal power by calculating the probability of observing the entire pattern of polymorphism across all samples , given the parameters and . In practice , however , this approach is extremely computationally intensive [17] , and even a recent breakthrough using a Markov chain Monte Carlo technique only extends full-likelihood to input data containing fewer than SNPs [18] . Instead , we follow the lead of previous researchers who sacrificed power for greater practicality by using a composite likelihood method [12] , [19] , [20] that approximates the true likelihood , as detailed in the Methods section . However , metagenomic population samples differ from traditional population samples and , as a result , provide new challenges to estimating recombination . First , the sample size varies according to the read depth at a given location instead of being fixed across all loci . Second , the quality of each base call varies along each read , and the random nature of the metagenomic method prevents independent replication of the sampling and sequencing steps to confirm observed polymorphisms . Finally , linkage information is greatly reduced in that instead of the traditional approach of sampling the same individual at all loci , each fragment of DNA derives from a different individual . Depending on the sequencing technology and whether reads were sequenced in pairs , these data will reveal , at most , linkage within two reads of nucleotides that are separated by a distance generally less than 40 kilobases . As high-throughput sequencing becomes ever cheaper , the number of projects producing this sort of data will only increase . The Human Microbiome Project ( http://www . hmpdacc . org/ ) plans to perform metagenomic sequencing of microbes found at five sites around the body . A particularly intriguing future application will be to sequence mixtures of pathogens sampled from within a single infected human . These data , combined with the methods presented here , will allow inferences about the interplay between the immune response and recombination within pathogens .
Our input data consist of a metagenomic assembly ( i . e . alignment of reads to a scaffold ) , untrimmed FASTA sequences for the reads , quality scores for each base in each read and , if applicable , information about read pairs . We explicitly do not consider any uncertainty in either the assembly or in the quality scores for the practical reason that current assembly algorithms and base callers do not generate this information; however , in principle , our method could be extended to incorporate these sources of uncertainty . Given these data , we wish to estimate two population genetic parameters: and . Following [12] , we assume that each site in the assembly has at most two different nucleotides and arbitrary label these as zero and one . In the rare event that more than two distinct nucleotides are observed , then we again arbitrarily label them zero and one after first grouping the nucleotides into two categories: the most common nucleotide and everything else . In the case of a tie for the most common nucleotide , we pick one at random . Given this labeling , we can represent the state of a read at a given position by 0 , 1 or ? , where the question mark represents missing data . Analogously , we represent the state of a single chromosome at two positions simultaneously: 00 , 01 , 10 , 11 , 0 ? , 1 ? , ? 0 , ? 1 ( ignoring ? ? , since this conveys no information ) . An example is given in Figure 1 and described below . Note that , in a metagenomic context , “a single chromosome” means that both nucleotides are either on the same read or on two paired reads . We assume that the total number of sequenced reads is much less than the total number of cells in the sampled environment such that the probability of two independent ( unpaired ) reads deriving from the same original cell/chromosome is essentially zero . First we outline our notation more formally . The assembly , , extends from position 1 to position and contains information about both the content of the reads and their position . The set of quality scores , , contains one quality score for each base in each read in the assembly . We assume Phred-calibrated quality scores [21] , so any particular quality score , , can be converted into an error probability , , by means of the formula . The configuration for a pair of sites , ( ) , is a vector of eight numbers corresponding to the number of chromosomes observed in each of the eight states ( 00 , 01 , etc . ) . For example , in Figure 1 the configuration of the leftmost pair of polymorphic sites is { , , , , , , , } . In addition to the configuration at pair , we also have the set of quality scores , ( ) . We wish to calculate the likelihood of the observed data , , given the quality scores , , and the population genetic parameters of interest , and . We approximate the true likelihood with the composite likelihood: ( 1 ) in which the two-locus configurations are treated as though they were independent among pairs of sites . We take the mutation rate ( and thereby ) to be constant and independent across all sites in the assembly , conditional on the genealogy . However , the recombination rate between two sites and depends on their distance apart , , as measured by the number of nucleotides separating them . We model recombination in microbial populations as occurring via gene conversion with recombination tract lengths drawn from an exponential distribution [12] , [22] , [23]: ( 2 ) where is the average length of the recombination tract . Theoretically and might be identifiable , but in practice our data are insufficient to separate them . Instead we fix and estimate , similar to the approach taken by McVean et al . [12] . Minor misspecification of will simply rescale , although major misspecification of will also change the right-hand side of ( 2 ) . Now we turn to the likelihood of a single two-locus configuration . We first account for sequencing error by summing over all possibilities for the truth , : ( 3 ) where the sum iterates over all elements of the set of possible two-locus configurations , , and is the average number of reads at each site . The first term inside the sum is the error probability , while the second term is the two-locus likelihood without any error . We assume that sequencing errors cause a switch from 0 to 1 and vice versa: ( 4 ) where and the subscript indexes the same position in the same read in the quality scores , the observation , and the truth . In other words , all mismatches between the truth and observed must be the result of an error , while all matches between the truth and the observed cannot have been caused by an error . Next we account for missing data by summing over all possibilities for the unknown nucleotides in the complete configuration , : ( 5 ) where the sum iterates over all elements of the set of configurations compatible with the observed data , ( i . e . those that satisfy the constraints , etc . ) . The first term inside the sum accounts for missing data , while the second term is the pure two-locus likelihood . If we treat the configurations and as a specific ordering of chromosomes , then this first term has a binary value of 1 for all configurations that match at non-missing positions and 0 otherwise . As a result of our definition for the set , all configurations will match at non-missing positions , so the first term is always 1 . We describe calculation of the second term in the next section below . We arrive at the final composite likelihood equation by taking ( 1 ) and substituting in ( 3 ) , ( 4 ) and ( 5 ) , which leaves us with four nested products and sums of significant size as discussed below . Now we wish to find maximum likelihood estimates to our parameters . Joint maximization of and is computationally impractical . Instead , we perform a two-step estimation procedure in which we first estimate from single sites using a previously-developed method that correctly handles sequencing error [15] and then estimate from pairs of sites by numerically maximizing ( 1 ) while holding . We pre-calculate and store the two-locus likelihoods for all possible complete two-locus configurations without error ( i . e . the second term in ( 5 ) ) for a single sample size , , across a range of values and a single fixed value . We generate this table of likelihoods by running a slightly modified version of the complete program from the LDhat package [12] , which assumes a finite sites Jukes-Cantor style biallelic mutation model and uses the neutral coalescent-with-recombination importance sampling method of Fearnhead and Donnelly [24] . The original complete program computed likelihoods only for configurations in which both sites were observed to be polymorphic; our modification enables the calculation of likelihoods for configurations with one polymorphic site and one fixed site . We deduce the final probability of both sites being fixed by subtracting all other probabilities from 1 . Given this table for a fixed sample size and fixed , we can exactly infer an analogous table for smaller sample sizes and approximately infer a table for different values of . A smaller sample size table can be directly generated for an arbitrary new sample size , ; however , in the interests of clarity , we will describe how to generate a table when , which can be iterated . Let the vector denote a configuration of sample size . Assuming probabilities for ordered configurations ( as generated by complete by default ) , the probability of this new configuration is the sum of the probabilities of , , and . Adjusting the table for a different poses a greater challenge . One option would be to run complete many times to generate tables for different values of , but this would be extremely time-consuming . Our alternative solution takes advantage of the fact that , while strongly affects the relative probabilities among the three broad categories of ( both-sites-polymorphic , one-site-fixed , both-sites-fixed ) , only mildly affects the relative probabilities of different configurations within these categories . The approximate probability of a site being polymorphic under the finite sites mutation model in a sample of size is ( approximate in the sense that this ignores the slight possibility of a site being polymorphic but having back mutations erase all traces of that polymorphism ) . If two sites are independent , then the probabilities corresponding to these three categories of pairs are , , . Now we assume that the ratio between the probabilities of these categories is independent of and approximate the probabilities of configurations under some new by multiplying by ( if both sites are polymorphic ) or ( if one site is fixed ) . If both sites are fixed , then we again deduce the probability by subtracting all other probabilities from 1 . Given these tabulated ( or calculated ) values , we use linear interpolation to arrive at the final probability for a given . Linear interpolation as well as our numerical maximization algorithm require that the likelihood surface be reasonably smooth . The importance sampling algorithm leaves a small amount of error in its estimate of the likelihood , which can lead to small wiggles in the likelihood surface . We solve this problem by smoothing the tabulated values where necessary via cubic splines . Also , for configurations with a single fixed site , the importance sampling algorithm did not reduce the variance in the likelihood below the very low level of the slope across , leading to numerical difficulties performing maximization on a non-smooth likelihood surface . We avoid this problem by making the likelihoods for these configurations constant across by setting them equal to their average value . As alluded to earlier , a brute force implementation of the four nested loops in the composite likelihood function would take time where is the length of the assembly ( or region of interest ) , is the read depth and is the average number of missing nucleotides at each site , assuming a constant read depth . Real metagenomic data have variable read depth , which makes the situation even worse with the sequencing error component ( ) dominating the complexity at high-depth sites ( i . e . where ) . Instead we make several approximations: Given these approximations , a standard desktop computer can perform this estimation for 10 kb of sequence , average depth of 10 and a realistic error distribution in less than one hour . Before we discuss our results , we need to quantify the amount of missing data between a given pair of sites . Define to be the proportion of chromosomes that span a particular pair of sites: , where is the number of chromosomes spanning both sites ( i . e . both sites are covered either by the same read or by paired reads ) and is the average number of chromosomes covering each site separately ( and , respectively ) . The average value of this statistic together with the average sample size provide an indirect measure for the amount of information about recombination captured by pairs of sites within a given dataset . We applied our technique to the first 500 kb of the assembly of Candidatus Accumulibacter phosphatis from a recent metagenomic sequencing project of activated sludge from a wastewater treatment plant [13] . The sludge we analyzed came from a laboratory bioreactor in Madison , Wisconsin that had been seeded from a local wastewater treatment plant . We received the data ( P . Hugenholtz , personal communication ) in the form of a finished assembly consisting of ACE and PhD files covering a megabase scaffold of average depth . Equivalent data in a different form are also available directly from the Joint Genome Institute via the IMG/M system [25] and the NCBI Trace Archive ( genome project id 17657 ) .
We first investigate the information content of a single pair of sites as a function of the amount of missing data . This information sets an upper bound on our estimator's performance since we use the composite likelihood instead of the true likelihood . In particular , the Fisher information , , for a single pair of sites with depth decreases with , although the information only falls off dramatically for ( Figure 2 ) . We find these results encouraging since the average of pairs in the actual sludge metagenome falls just above this threshold at 0 . 21 . Note that the Fisher information holds little meaning on an absolute scale since we calculate the information for a single pair of sites rather than for our actual data with many dependent pairs . Instead , the values in Figure 2 should be interpreted on a relative scale . For instance , for , approximately ten independent pairs with would contain the same information about as a single pair with . The bulk of our analyses rely on simulated data where we know the truth and can evaluate the performance of our estimator . We use the program ms [26] in combination with seq-gen [27] to generate sequences across a 10 kb region under a finite-sites model of mutation ( unless specified otherwise ) and the coalescent with recombination . We simulate recombination as gene conversion with mean tract length fixed at ( see equation 2 ) . The sample size ( i . e . number of simulated chromosomes ) is where is the average read depth and is the length of each read in a read-pair . We transform these sequences into metagenomic-style data by randomly distributing read starts uniformly across the simulated region and trimming each simulated sequence to only be present for the length of three segments: one read , the gap between read pairs , and one read . Our simulation assumes no variation in read length or distance between read pairs . Note that a gap of zero produces the same effect as unpaired reads with double the read length . For results with sequencing error , we assign quality scores from the true Sanger sequencing quality score distribution as determined from the sludge data . A “sequencing error” causes a switch from the true nucleotide to each of the other three with probability 1/3 . Given that we are simulating relatively small datasets with low information content , we occasionally generate an assembly with a maximum likelihood at . We exclude these values from all further analyses , but , for each parameter set , we report the proportion of replicates that yielded infinite parameter estimates either in Table 1 or in the text below . We analyzed the performance of our estimator in the presence of sequencing error across a range of plausible values of ( 0 . 002 to 0 . 04 ) , read lengths roughly corresponding to current Illumina , 454 and Sanger sequencing technologies ( 75 , 500 , 1000 ) and gaps between read-pairs ( 0 , 100 , 500 ) by calculating the root mean squared error ( RMSE ) relative to the true value ( ; Figure 3 ) . Note that while RMSE conveniently summarizes our estimator's sampling distribution , it obscures the inherent asymmetry of the distribution caused by the constraint . A clear trend emerges with lower relative RMSE accompanying increased recombination . The estimator has little bias ( results not shown ) and , for , we are able to reliably estimate within a factor of of the true value . For most parameters , increasing the read length reduces the variance by virtue of increasing , but for larger the results for 1 kb reads appear slightly worse than for 0 . 5 kb reads . Increasing the gap between the paired-end reads increases the variance for all except the very smallest . Intuitively , this makes sense: if all pairs of sites are very close together with low then a recombination event will only rarely occur between them; however , if all pairs are far apart with high then recombination events will saturate between the pairs of sites . With the above results suggesting that longer read lengths do not always yield a better estimate , we decided to directly compare a metagenomic-style sample to a “standard” population genetic sample in which the same individuals are sequenced at all loci . The fair comparison keeps the total number of sequenced bases constant , so we simulate a 10 kb region with either 100 reads of 1 kb each or 10 reads of 10 kb each ( Figure 4 ) . For simplicity , we do not simulate sequencing error . As analyzed in the Discussion , despite the average depth being identical between the two sets of simulations , the metagenomic sample ( on the left ) exhibits less bias and much lower variance than the standard sample ( on the right ) . Next we tested our approximation that adjusts the two-site likelihoods for different values of ( see Methods subsection “Two-locus complete likelihoods without error” ) by fixing and simulating across ranging from 0 . 002 to 0 . 025 while estimating using a two-site likelihood table generated for ( Figure 5 ) . Again we do not simulate sequencing error to focus on the effects of . Here we see that the correction ( on the right in Figure 5 ) works quite well for above the likelihood table's driving value ( i . e . ) and somewhat less well for lower , with 3% of the simulations for giving infinite estimates . However , the uncorrected estimator ( on the left ) is strongly biased , with 98% of simulations for resulting in infinite ( unplotted ) estimates and 26% of those for . No other parameter values yielded any infinite estimates . The low results are exacerbated by the correlation of with the number of polymorphic sites . Lower means fewer polymorphic sites; since the majority of information about recombination rate comes from polymorphic sites , we see a larger variance in our estimate of for low . Finally we apply our estimator to the sludge metagenomics project by sliding a 50 kb window in 25 kb steps across the first 500 kb of the assembly and independently estimating the recombination rate within each window ( Figure 6 ) . All windows produced finite estimates with .
The two-site composite likelihood estimator appears to be better suited for metagenomic samples ( i . e . the purpose of this paper ) than for standard population genetic samples ( i . e . the purposes of [12] , [19] ) as seen from Figure 4 . We believe this results from the balance of two opposing factors: greater linkage ( less missing data ) pushes the advantage toward the standard sample , while a larger genealogy with more independence pushes the advantage back toward the metagenomic sample . For the parameter ranges investigated here , the latter force wins and we see that the estimates for metagenomic samples have both less bias and lower variance for a fixed amount of sequencing . This result makes sense given the nature of the composite likelihood technique in which we treat each pair of sites as though it were independent of every other pair . The more chromosomes that are sampled , the more closely this independence assumption matches reality . An intriguing open question is how the composite likelihood estimator on metagenomic data compares to a full likelihood estimator on standard data , but we do not pursue this topic here . The bias in the standard sample estimates ( Figure 4 ) surprised us given theoretical results that assert consistency for the composite likelihood estimator [28] . However , consistency is an asymptotic feature and does not necessarily hold for finite samples . Indeed , further simulations of standard samples with greater sample depth reduced the bias to essentially zero with depth ( results not shown ) . Given that metagenomic samples appear nearly unbiased with depth , the added independence of the metagenomic sample must allow the estimator to converge faster toward the asymptotic results . Further , in contrast to Hudson's and McVean's programs ( maxhap and LDhat , respectively ) , our method makes use of all pairs of sites , including sites observed to be fixed . We include these sites primarily as a byproduct of properly accounting for sequencing error , but these additional data also help reduce our variance . As a bonus , using all sites automatically makes our pairwise likelihoods true likelihoods , thus fulfilling one of the requirements for Fearnhead's [28] results proving the consistency of the composite likelihood estimator . If fixed sites were not included , then the pairwise likelihoods would need to be made conditional on only using pairs of segregating sites , which becomes computationally challenging when dealing with missing data . In fact , while maxhap and LDhat allow missing entries in their input data , this feature is not described in the accompanying papers [12] , [19] , and these implementations do not properly condition their likelihoods to account for the fact that they only use segregating sites . The only disadvantage of using all pairs of sites is that the likelihood calculation scales linearly with the number of pairs and thus using all pairs takes longer; however , our implementation still runs in a reasonable amount of time on realistic amounts of data ( see “Complexity and approximations” subsection in Methods ) . Real data include sequencing errors , which have the potential to bias population genetic inference and increase the variance of estimators [29] . Trimming the data based on quality scores will help reduce these problems , but the remaining error must still be taken into account . We do not have analytic theory quantifying the amount of bias introduced by sequencing error , but simulations show that unaccounted-for errors produce estimates biased toward a specific finite value of that depends on the read length and gap size ( results not shown ) . Intuitively , sequencing error primarily produces singletons , which yield different configurations depending on the distance separating the two sites with errors . If the two sites are close together , then errors will tend to generate 01 and 10 states . If the two sites are far apart , then errors will tend to generate 1 ? and ? 1 states . The first group of states ( 01 , 10 ) provides evidence for higher recombination since , if both mutations originally fell on the same chromosome ( state 11 ) , then recombination would have been necessary to break them up to be ( 01 , 10 ) . The second group of states ( 1 ? , ? 1 ) provides evidence for lower recombination since this pattern of missing data is more likely to have arisen from ( 11 , 11 ) states , which is suggestive of no recombination , then ( 01 , 10 ) states . Thus sequencing error introduces a highly artificial pattern of configurations , with a combination of evidence for high recombination between close pairs of sites and low recombination between distant pairs of sites leading to a maximum likelihood at an intermediate value . For paired-end reads of 500 bases separated by a gap of 0 , errors drive toward . The striking inverse correlation between the estimates of and from the sludge data ( Figure 6 ) could either be the result of an unknown artifact or a biological reality stemming from a dependence between recombination efficiency and sequence divergence . One possibility for an artifact would be sequencing error not accounted-for in the quality scores ( e . g . a PCR error before sequencing ) . Such errors would certainly lead to increased estimates of , but , on the basis of our simulations , seem unlikely to drive down to 0 . Also , such errors would have to occur non-uniformly across the genome at a granularity of 50 kb , which seems implausible . Another potential source for an artifact is our two-step estimation procedure in which we first estimate without regard to recombination and then estimate conditional on . Again , however , simulations reveal that , while affects the variance of , the estimator is unbiased across all tested and shows no correlation between and ( results not shown ) . Without a clear artefactual explanation , we turn toward biology . Laboratory experiments have shown a negative log-linear dependence between sequence divergence and transformation efficiency [30] , and an analysis of a different metagenomic dataset found a similar dependence between divergence and parsimoniously-inferred recombination events [10] . Our data suggest that this pattern holds at a finer resolution with subtle increases in diversity , as quantified by , leading to lower rates of recombination in a log-linear manner , with the exception of regions in which recombination appears nonexistent ( Figure 7 ) . On an absolute scale , these estimates from the sludge data fall into a plausible range for bacterial populations . For instance , in Campylobacter jejuni [31] and in Neisseria meningitidis ranges from 0 . 00270 to 0 . 034 [11] . However , previous estimates of microbial recombination rates have been based on much smaller amounts of data ( in these examples , bases ) relative to the sludge windows of 50 kilobases . In addition , C . jejuni and N . meningitidis are both pathogens , which makes for a quite different ecological and evolutionary environment than that of the nonpathogenic sludge bacterium A . phosphatis . When the sludge estimates of mutation and recombination are viewed relative to each other , we see that mutation events generally occur more frequently than recombination events ( ) , which places A . phosphatis more toward the clonal end of the bacterial spectrum [32] . Overall , our new estimator produces surprisingly accurate estimates of recombination rate , particularly considering the amount of missing data . The real power of the estimator derives from the greater independence of the genealogies underlying the sample; sequencing error and missing data present hurdles to accessing this information but our estimator has surmounted them . Despite our motivation from microbial populations , our method itself makes no assumptions inherent to microbial populations . For our purpose , a “metagenomic” sample simply means sampling a mixture of a large number of individuals from a single species , in which each read ( or pair of reads ) can be safely assumed to have originated from a different individual . Given the results from the comparison to a standard sample , the metagenomic approach should always be followed to obtain maximal information about recombination for a fixed amount of sequencing . An implementation of our Population genetic Inference In Metagenomics ( PIIM ) method is freely available for download from http://ib . berkeley . edu/labs/slatkin/software . html . | At a broad scale , the exchange of genetic material through homologous recombination ( i . e . what happens in animals during sex ) increases the potential rate of adaptation . Bacteria often reproduce clonally , without recombination , by making exact copies of their genomes , but they also have mechanisms analogous to sex that allow them to recombine sporadically . Despite microbes' critical role at the base of our world's ecosystem , microbiologists know surprisingly little about how microbes grow and evolve outside the laboratory . Metagenomic sequencing projects provide a means to sample the genetic diversity of natural microbial populations and have the potential to reveal much about the ecology and evolution of these populations . Here we present a novel method to estimate the recombination rate from metagenomic data , while explicitly allowing for imperfections such as sequencing error and missing data . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"computational",
"biology/metagenomics",
"genetics",
"and",
"genomics/microbial",
"evolution",
"and",
"genomics",
"genetics",
"and",
"genomics/population",
"genetics"
] | 2009 | Inference of Microbial Recombination Rates from Metagenomic Data |
Soil-transmitted helminths ( STH ) are the most common parasitic infections in impoverished communities , particularly among children . Current STH control is through school-based mass drug administration ( MDA ) , which in the Philippines is done twice annually . As expected , MDA has decreased the intensity and prevalence of STH over time . As a result , the common Kato Katz ( KK ) thick smear method of detecting STH is less effective because it lacks sensitivity in low intensity infections , making it difficult to measure the impact of deworming programs . A cross-sectional study was carried out over a four-week period from October 27 , 2014 until November 20 , 2014 in Laguna province , the Philippines . Stool samples were collected from 263 schoolchildren , to determine the prevalence of STH and compare diagnostic accuracy of multiplex quantitative polymerase chain reaction ( qPCR ) with the KK . A large discrepancy in the prevalence between the two techniques was noted for the detection of at least one type of STH infection ( 33 . 8% by KK vs . 78 . 3% by qPCR ) , Ascaris lumbricoides ( 20 . 5% by KK vs . 60 . 8% by qPCR ) and Trichuris trichiura ( 23 . 6% by KK vs . 38 . 8% by qPCR ) . Considering the combined results of both methods , the prevalence of at least one type of helminth infection , A . lumbricoides , and T . trichiura were 83 . 3% , 67 . 7% , and 53 . 6% , respectively . Sensitivity of the qPCR for detecting at least one type of STH infection , A . lumbricoides , and T . trichiura were 94 . 1% , 89 . 9% , and 72 . 3% respectively; whereas KK sensitivity was 40 . 6% , 30 . 3% , and 44 . 0% , respectively . The qPCR method also detected infections with Ancylostoma spp . ( 4 . 6% ) , Necator americanus ( 2 . 3% ) , and Strongyloides stercoralis ( 0 . 8% ) that were missed by KK . qPCR may provide new and important diagnostic information to improve assessment of the effectiveness and impact of integrated control strategies particularly in areas where large-scale STH control has led to low prevalence and/or intensity of infection .
Soil-transmitted helminth ( STH ) infections are the most common parasitic infections among children worldwide , especially among impoverished communities [1] . This also holds true in the Philippines , where the prevalence in school-aged children reportedly was as high as 67% in 2001 [2] . A subsequent study in 2006 , which served as a baseline for the Integrated Helminth Control Program ( IHCP ) of the Department of Health ( DOH ) , showed a prevalence of 54% for at least one type of STH infection and 23 . 1% for the prevalence of heavy-intensity infections [3] . The primary effort to control STH infections , involving mass drug administration ( MDA ) through school-based deworming with benzimidazole anthelminthics [1 , 4–5] , has increased worldwide over the past ten years . The objective of MDA is to minimize transmission and reduce morbidity , which is associated with heavy infections; however , it must be repeated at regular intervals since re-infection occurs rapidly [6–8] . In the Philippines , a nationwide semiannual school-based MDA targeting pre-elementary and Grades 1–6 pupils ( aged 6–12 years old ) in all public elementary schools has been implemented since 2007 by the Department of Education ( DepEd ) in collaboration with the DOH through its IHCP [9] . To assess the impact of the IHCP , in addition to the baseline nationwide survey of STH infections , a follow-up survey was conducted in 2009 . This survey showed a significant decrease in the prevalence overall ( 44 . 7% ) and of heavy-intensity STH infections ( 19 . 7% ) among school-aged children ( 6–12 years ) [10] . While the prevalence appeared to have been reduced it remained higher than the 20% target recommended by the World Health Organization ( WHO ) to achieve morbidity control , despite several years of MDA . In 2015 , a National Deworming Day programme was established to improve access to and uptake of health interventions for all school-aged children enrolled in public elementary schools in the Philippines [11] . It is important that the prevalence and intensity of STH infection is monitored rigorously to assess the effectiveness of the control programme after repeated rounds of treatment . Hence , there is a requirement to utilize highly sensitive and specific methods for detecting infected individuals . Several microscopy-based techniques are available and widely used for identification and quantification of STH eggs and larvae . The most widely used technique is the Kato-Katz ( KK ) thick smear technique , recommended by the WHO for assessing both the prevalence and intensity of infection in helminth control programmes [12] . However , as KK lacks sensitivity , particularly in areas with a high proportion of light-intensity STH infections [13 , 14–16] , molecular approaches are increasingly being used in monitoring and surveillance [10 , 14 , 17–20] . A number of recent studies have shown that the sensitivity of molecular-based diagnosis is considerably higher than the KK procedure , especially if the infection intensity is low [15 , 19 , 21 , 22] . We conducted a parasitological survey among schoolchildren in the province of Laguna located in the Calabarzon region of Luzon in November 2014 , where the reported STH prevalence in 2002 was 84 . 2% [23] . The aims of this study were to 1 ) quantify the prevalence of STH among elementary schoolchildren–particularly before the implementation of the National School Deworming Day programme on July 29 , 2015; and 2 ) compare the diagnostic performance of KK and a multiplex quantitative real time polymerase chain reaction ( qPCR ) method for the diagnosis of STH infections .
A cross-sectional study was carried out over a four-week period from October 27 , 2014 until November 20 , 2014 in Laguna province , the Philippines , to determine the prevalence of STH infections among schoolchildren using the KK method and a qPCR assay . The study protocol was submitted to and approved by the Institutional Review Board of the Research Institute for Tropical Medicine ( RITM ) with approval number 2013–15 and the QIMR Berghofer Medical Research Institute ( QIMRB ) Human Ethics Committee ( approval number: P1271 ) . Permission was sought from the DepEd prior to the conduct of the study . With the permission obtained , we provided an orientation about the study to the principals of each school involved . Written informed consent was obtained from the parents and/or legal guardians of students invited to participate in the study . The purpose and procedures of the study were also explained to the participating children . At study completion , parasitological results were communicated to all parents , with a recommendation for treatment from the local health centre . The study was undertaken in grade 4 and 5 schoolchildren in ten selected elementary schools across ten municipalities of Laguna Province ( chosen municipalities also correspond to the school districts ) . The selection of the municipalities was based on the rural/urban classification . This classification was based on the number of barangays classified by the Philippine Statistical Authority ( PSA ) as rural/urban . A municipality was classified as rural if majority of the component barangays were classified as rural . The same applies in classifying urban municipalities . Five rural ( Alaminos , Calauan , Liliw , Luisiana , Siniloan ) and five urban ( Cabuyao , Pagsanjan , Pila , San Pablo and Sta . Rosa ) municipalities were randomly selected . In this study , the school district or a cluster of school districts was considered as a homogeneous area [12] since all the selected schools are located in a dry region at low altitude . From each school district , one school was randomly selected . The selected schools were as follows: San Andres Elementary School ( ES ) in Alaminos , San Isidro ES in Calauan , Taykin ES in Liliw , San Buenaventura ES in Luisiana , Buhay ES in Siniloan , Gulod ES in Cabuyao , Sampaloc ES in Pagsanjan , Santo Niño ES in San Pablo , Pinagbayanan ES in Pila and Dita ES in Sta . Rosa ( Fig 1 ) . The selection of the subjects was based on WHO recommended guidelines ( [12] , with some modifications ) . The recommended sample size was 200–250 individuals in each ecologically homogenous area to evaluate the prevalence and intensity of STH infection , with an addition of 30% to compensate for non-compliance in the submission of stool samples . A total of 325 individuals were randomly selected and invited to participate to ensure the enrolment of 250 children . A minimum of thirty-five children per school were invited . For this study , Grade 4 students ( aged 9–10 years ) were targeted; however , in schools where the number of Grade 4 student was less than 35 , Grade 5 students ( aged 10–11 years ) were also included . De-worming in the selected schools is done every January and July of the year . The parasitological survey was conducted in October-November 2014 , three months after the July de-worming round . The coverage of this round is unknown , although the recorded coverage in the neighboring municipality of Calamba was only 35% . Information on the previous parasitological burden has not been examined in this area . Stool cups were provided to the children one week before the survey week , which was allotted during five school days , and the method of collecting the samples was explained . This was to ensure submission of the stool samples on any of the five days of the survey . One stool sample per student was collected . The samples were collected , processed at the school site within two hours after collection , and read the same day using triplicate Kato-Katz ( KK ) thick smears ( 41 . 7 mg of stool/smear ) [24] . A team of trained microscopists read the slides . The microscopists worked independently of each other on the samples assigned to them ( i . e . , one sample examined by one microscopist only ) . The slides were read at the school between 2–4 hours post collection to maximize hookworm diagnosis . The number of STH eggs was counted and recorded for each helminth species separately . To ensure validity and accuracy of the results , 42% percent of all slides were randomly selected and re-examined by a reference microscopist at the National Reference Laboratory for Parasitology at RITM . In addition , a specimen of two to three grams of each stool sample was transferred to a 15 ml plastic tube and stored in 80% ( v/v ) ethanol at 4°C . The ethanol-preserved samples were transported at room temperature to QIMRB ( Australia ) and stored at 4°C prior to DNA isolation and subsequent molecular analysis . DNA was extracted from the stool samples using Maxwell 16 LEV Plant DNA kits ( Promega Corporation , Madison , WI USA ) , in conjunction with the Maxwell 16 robot ( Promega ) . Approximately 200 mg of stool was added to a 2 mL twist cap tube and 500 μl of ROSE buffer [10 mM Tris ( pH 8 . 0 ) , 300 mM EDTA ( pH 8 . 0 ) , 1% w/v sodium dodecyl sulfate ( SDS ) and 1% w/v polyvinylpolypyrrolidone ( PVPP ) ] , and 1 g of 0 . 5 mm silica/zirconia beads ( Daintree Scientific , St . Helens , Australia ) was added to the tube [19 , 25] . Tubes were left overnight at 4°C . The next day , tubes were placed into a Precellys tissue homogenizer ( Bertin instruments , Paris , France ) for 30 seconds at 6500 rpm . Following homogenization , tubes were placed in a heating block for ten minutes at 95°C , vortexed , and then centrifuged for five minutes at 10 , 000 g . Two hundred μl of H20 and 300 μl of the supernatant from the centrifuged stool were added to the first well of the LEV cartridge ( from the Maxwell kit ) . Cartridges were then placed in the Maxwell 16 robot , along with elution tubes containing 50 μl of elution buffer . The DNA extraction program for the plant kit was selected on the robot , and DNA extraction was fully automated from this point . Once completed , cartridges were discarded . DNA concentration and quality was tested using a Biotek powerwave HT Microplate Spectrophotometer . All aliquots of DNA were then diluted by a factor of five and used as the template in the resulting qPCRs . DNA with a concentration of less than 10 ng/μl were not used further . Two qPCR assays were performed , a multiplex designed to identify hookworm ( Ancylostoma spp . and N . americanus ) , A . lumbricoides , and T . trichiura , and a singleplex designed to identify S . stercoralis . The multiplex and singleplex qPCRs were performed utilising previously published primers and probes ( Table 1 ) . The multiplex qPCR was made up to a total volume of 25 μl that contained: 10 μl of iTaq supermix ( Bio-Rad Laboratories , Hercules , California USA ) , 2 . 1 μl of H2O , 3 μl of DNA , 60 nM each of A . lumbricoides primers ( forward and reverse ) and probe ( FAM ) , 200 nM each of Ancylostoma spp . primer and probe ( Cy5 ) , 200 nM each of N . americanus primers , 100 nM of N . americanus probe ( ROX ) , 60 nM each of T . trichiura primers , and 100 nM of T . trichiura probe ( Cy5 . 5 ) . The qPCR was performed on a corbett rotorgene 6000 ( Qiagen , Hilden , Germany ) . Cycling conditions consisted of two minutes at 98°C followed by 40 cycles of 98°C for 20 seconds , 74°C for 20 seconds , and 58°C for 20 seconds , followed by a final dissociation phase of 72°C for five minutes . The singleplex qPCR was made up to a total volume of 16 μl that contained: 8 μl of GoTaq ( Promega ) , 4 . 64 μl of H2O , 100 nM each of S . stercoralis primers and probe . The qPCR was performed on a CFX384 ( Bio-Rad ) . Cycling conditions were the same as for the multiplex qPCR ( above ) . Positive and negative controls were used in each run . Two types of positive controls were used . For Ancylostoma spp . , N . americanus and A . lumbricoides , cloned copies of 300 bp G-block gene fragments ( purchased from Integrated DNA Technologies; IDT , Coralville , USA ) , specific for the gene of interest from each species were used [25] . In addition , DNA samples , extracted from eggs and adults of Ancylostoma spp . , N . americanus , A . lumbricoides , T . trichiura and S . stercoralis , were provided by Professor James McCarthy , QIMRB . Negative controls were no-template controls , where water was used in place of DNA template . Statistical analyses were carried out using Stata version 13 . 1 and Microsoft Excel . Only schoolchildren with a matching set of three slides of KK thick smears and qPCR results were included in the final analysis . The prevalence of helminth infections , including the 95% confidence intervals ( 95% CIs ) derived from the KK , qPCR and the combined results of both techniques , were calculated using the proportion command in Stata . The average number of helminth eggs per gram of stool ( EPG ) was obtained by multiplying the number of helminth eggs recorded in the KK thick smear by a factor of 24 , summing the results and dividing it by the number of slides . Classification into light , moderate and heavy infection intensity was based on the average individual EPG derived from the three KK slide readings , considering thresholds set forth by WHO: A . lumbricoides [light ( 1–4 , 999 EPG ) , moderate ( 5 , 000–49 , 999 EPG ) , heavy ( ≥ 50 , 000 EPG ) ]; T . trichiura [light ( 1–999 EPG ) , moderate ( 1000–9 , 999 ) , heavy ( ≥ 10 , 000 EPG ) ]; hookworm [light ( 1–1999 EPG ) , moderate ( 2 , 000–3 , 999 EPG ) , heavy ( ≥ 4 , 000 EPG ) ] [12 , 28] . The geometric mean EPG ( GMEPG ) in infected persons was also calculated for each STH species . In addition , the classification of infection prevalence using the cycle threshold ( Ct ) values derived from the qPCR data was performed using the cut-offs as previously described [25] . The analysis using the cut-off values for intensity of infection was not calculated as this methodology is still evolving . The diagnostic accuracy parameters including 95% CIs , were calculated using two different approaches . The first used the direct method comparison where the relative sensitivity and specificity of the KK compared to the qPCR was calculated using the qPCR as the reference standard . For the second , the relative sensitivity and specificity of both diagnostics techniques were calculated using the combined results of the KK and qPCR as reference standard . The pairwise agreement between the diagnostic techniques ( KK v qPCR ) was evaluated using Cohen’s kappa statistics at 95% CIs . Only species with above 20% prevalence were analyzed . The k-statistics were interpreted as <0 . 00 , poor agreement; 0 . 00–0 . 20 , slight agreement; 0 . 21–0 . 40 , fair agreement; 0 . 41–0 . 60 , moderate agreement; 0 . 61–0 . 80 , substantial agreement; 0 . 81–1 . 00 , almost perfect agreement [29] . The association of sex , age group and school with the STH prevalence derived from KK , qPCR and the combined results of both diagnostics techniques were analysed by using Chi-square test , and P values <0 . 05 were considered statistically significant .
From the 382 schoolchildren who provided consent , 285 ( 74 . 6% ) submitted stool samples . The qPCR was performed on stool samples from 264 schoolchildren due to the insufficient amount of faeces provided by the other 21 . Of the 264 tested by qPCR , 263 ( 99 . 62% ) had complete data records ( matched triplicate KK thick smears and qPCR results ) . Among these 263 , more than half ( 54% ) were female and the majority ( 56% ) were between 8–9 years old . We compared the prevalence of STH infection based on the results of three KK thick smears , the qPCR and the combined results of both techniques . As shown in Table 2 , 33 . 8% of the schoolchildren studied using KK had one or more STH infection . T . trichiura was the most prevalent , infecting 23 . 6% of the schoolchildren; 20 . 5% had A . lumbricoides , and 1 . 9% were infected with Enterobius vermicularis . Applying the results from the qPCR technique , the prevalence of at least one type of STH infection was 78 . 3% , 60 . 8% for A . lumbricoides , 38 . 8% for T . trichiura , 6 . 8% for hookworm ( for 4 . 6% Ancylostoma spp . and 2 . 3% for N . americanus ) , and only 0 . 8% for S . stercoralis . Almost three times ( 160/54 ) the number of A . lumbricoides infections and 1 . 6 ( 102/62 ) times the number of T . trichiura infections were determined by the qPCR technique compared with the KK method . S . stercoralis , N . americanus and Ancylostoma spp . were detected using the qPCR but not by KK . The number of samples that tested negative by the KK was three-fold ( 174/57 ) higher than the number testing negative by qPCR . Considering the combined results of both techniques , the prevalence of at least one type of helminth infection was 83 . 3% . A . lumbricoides was present in 67 . 7% of the schoolchildren , while 53 . 6% were infected with T . trichiura . The prevalence of A . lumbricoides and T . trichiura by diagnostic technique was stratified according to sex , age group and school ( Table 3 ) . No significant differences were observed between sex , age group and prevalence across the three parameters for both A . lumbricoides and T . trichiura . School prevalence for A . lumbricoides ranged from 10% ( 95% CI: 2 . 19–35 . 51 ) to 50% ( 95% CI: 27 . 68–72 . 31 ) as determined by the KK; 16% ( 95% CI: 5 . 69–37 . 54 ) to 92 . 59% ( 95% CI: 72 . 83–98 . 31 ) by qPCR; and 28% ( 95% CI: 13 . 21–49 . 85 ) to 92 . 59% ( 95% CI: 72 . 83–98 . 31 ) by the combined results of both techniques . For T . trichiura , the school prevalence ranged from 5% ( 95% CI: 0 . 57–32 . 27 ) to 50% ( 95% CI: 27 . 68–72 . 31 ) as detected by KK; 15% ( 95% CI: 4 . 39–40 . 37 ) to 80% ( 95% CI: 58 . 25–91 . 97 ) by qPCR and 20% ( 95% CI: 6 . 99–45 . 36 ) to 80% ( 95% CI: 58 . 25–91 . 97 ) by both techniques . Pinagbayanan ES had the highest prevalence of both A . lumbricoides ( 50%; 95% CI: 27 . 68–72 . 31 ) and T . trichiura ( 50%; 95% CI: 27 . 68–72 . 31 ) following the results of the KK . Meanwhile , San Isidro ES had the highest prevalence of A . lumbricoides infection by qPCR ( 92 . 6%; 95% CI: 72 . 83–98 . 31 ) while San Andres ES had the highest prevalence of T . trichiura infection ( 80%; 95% CI: 58 . 25–91 . 97 ) , also determined by qPCR . The differences observed among schools for the prevalence of A . lumbricoides ( KK P = 0 . 031; qPCR P = <0 . 001 , combined results of both techniques P = <0 . 001 ) and T . trichiura ( KK P = <0 . 001; qPCR P = <0 . 001 , combined results of both techniques P = <0 . 001 ) were statistically significant across the three parameters . The geometric mean EPG values for A . lumbricoides and T . trichiura obtained by the KK technique and the mean Ct values derived from the qPCR are summarized in Table 4 , with the EPG count range ( KK ) , the Ct range found in a single stool sample ( qPCR ) , and the infection intensities among the positives ( KK ) stratified by infection intensity category defined by WHO . According to the WHO-defined infection intensity classification [12 , 28] , the majority of infected schoolchildren included in the final analyses had low intensity infections: 48 . 1% ( 25/52 ) for A . lumbricoides , and 85 . 2% ( 52/61 ) for T . trichiura ( Table 4 ) . Based on the Ct values derived from the qPCR analysis , the mean Ct values were 29 . 48 ( range: 21 . 28–34 . 73 ) for A . lumbricoides and 17 . 34 ( range: 8 . 35–34 . 91 ) for T . trichiura ( Table 4 ) . A comparison of diagnostic accuracy was only performed for A . lumbricoides and T . trichiura because of the low number of individuals harbouring the other helminths . The relative sensitivity of the KK for detecting at least one type of STH infection using the qPCR as the reference standard was 36 . 9% ( 95% CI: 30 . 3% - 43 . 9% ) . Relative specificity was calculated as 77 . 2% ( 95% CI: 64 . 2% - 87 . 3% ) . Further , the relative sensitivity of the KK for detecting A . lumbricoides was calculated as 22 . 5% ( 95% CI: 16 . 3% - 29 . 8% ) and 22 . 3% ( 95% CI: 14 . 7% - 31 . 6% ) for T . trichiura . The calculated relative specificity was 82 . 5% ( 95% CI: 73 . 8%-89 . 3% ) for A . lumbricoides and 75 . 8% ( 95% CI: 68 . 4–82 . 2% ) for T . trichiura . Following the combined results of both techniques ( triplicate KK thick smears and the qPCR ) as reference standard ( Table 5 ) , the relative sensitivity for detecting at least one type of helminth infection was higher using qPCR than the KK ( 94 . 1% by qPCR versus 40 . 6% by KK ) . The relative sensitivity for A . lumbricoides diagnosis was higher for qPCR than KK ( 89 . 9% by qPCR versus 30 . 3% by KK ) . For the diagnosis of T . trichiura , similarly , PCR showed a better relative sensitivity than KK ( 72 . 3% by qPCR versus 44% by KK ) . The calculated relative specificity for both techniques for all species was 100% . Parasite prevalence agreement statistics ( Table 6 ) showed a slight agreement ( κ = 0 . 0425 ) between triplicate KK thick smears and the qPCR for A . lumbricoides egg detection while a poor agreement was found for T . trichiura ( κ = - 0 . 0180 ) . The agreement between the two techniques for each helminth species , however , was not significant ( A . lumbricoides P = 0 . 1624; T . trichiura P = 0 . 6224 ) . For all the parasites analyzed , the qPCR detected a large number of samples not detected by KK ( A . lumbricoides , 124; and T . trichiura , 79 ) . However , a small percentage of samples were observed to be positive by KK for A . lumbricoides 7% ( 18/263 ) and T . trichiura 15% ( 39/263 ) but negative by qPCR . In terms of detecting samples with co-infections , there was a two-fold ( 67/27 ) increase in the number of samples with two or more parasite species detected by qPCR than by KK . Of the 206 schoolchildren positive for at least one type of STH infection detected using the qPCR , 28 . 2% had two different parasites while only 4 . 4% harboured three different worm species ( Table 7 ) . The various helminth parasite combinations are shown in Fig 2 . Co-infections with A . lumbricoides and T . trichiura were the most prevalent ( 24 . 3% ) co-infection observed in this study . Slide reading validation showed an overall sensitivity of 97 . 6% ( 95% CI: 91 . 5% - 99 . 7% and specificity of 85 . 7% ( 95% CI: 67 . 3% - 96% ) for detecting STH . Sensitivity for A . lumbricoides was 96% ( 95% CI: 86 . 3% - 99 . 5% ) , whereas the sensitivity for T . trichiura was 98 . 2% ( 95% CI: 90 . 6% - 100% ) .
The STH prevalence among schoolchildren in Laguna province obtained from this study using the combined results of the KK and qPCR procedures was 84 . 2%; this is similar ( although using more sensitive diagnostics ) to that reported in 2002 in Laguna ( KK only ) when the nationwide school-based MDA was not yet locally implemented [23]; and despite several years of MDA . STH control strategies in the Philippines thus need to be enhanced , including sustaining the national deworming day programme with high coverage levels and regular monitoring of its impact with the use of a more sensitive and specific diagnostic technique . The higher sensitivity of the qPCR as shown in this study would imply that the true STH prevalence is substantially higher in the Philippines and elsewhere than currently assessed . This could have a direct impact on policies and control programmes based on suggested WHO treatment guidelines . Higher prevalence estimates obtained with the qPCR method would , for example , result in an increased number of communities subjectively defined as community category II ( classified as with greater than 50% prevalence of infection and low infection intensities ) , for which community-based MDA is recommended [12] . Furthermore , additional interventions ( e . g . water , sanitation and hygiene ( WASH ) and a proven health education intervention ( e . g . “Magic Glasses” ) ) [35] as part of a multi-component integrated approach will be required to augment MDA for sustainable control and even elimination of STH infections . | Worldwide , two billion people are estimated to be infected with soil-transmitted helminths ( STH ) . These infections are primarily found in low resource settings and can result in cognitive impairment and growth stunting in children . The current control method is by chemotherapy , usually during large-scale mass drug administrations ( MDA ) ; however , this does not prevent re-infection , which can occur rapidly after treatment . The currently used diagnostics lack sensitivity in low intensity infections , resulting in underreporting of STH prevalence . In order to evaluate new control programs aimed at preventing re-infection and decreasing environmental prevalence of STH , more sensitive diagnostics are required . In this study we have shown that qPCR is far more sensitive than the traditionally used Kato-Katz ( KK ) microscopic technique , suggesting a role for qPCR in assessing control interventions . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"invertebrates",
"schoolchildren",
"medicine",
"and",
"health",
"sciences",
"education",
"helminths",
"sociology",
"tropical",
"diseases",
"hookworms",
"social",
"sciences",
"parasitic",
"diseases",
"animals",
"necator",
"americanus",
"ascaris",
"ascaris",
"lumbricoides",
... | 2017 | Status of soil-transmitted helminth infections in schoolchildren in Laguna Province, the Philippines: Determined by parasitological and molecular diagnostic techniques |
Pre-mRNA splicing is a critical step of gene expression in eukaryotes . Transcriptome-wide splicing patterns are complex and primarily regulated by a diverse set of recognition elements and associated RNA-binding proteins . The retention and splicing ( RES ) complex is formed by three different proteins ( Bud13p , Pml1p and Snu17p ) and is involved in splicing in yeast . However , the importance of the RES complex for vertebrate splicing , the intronic features associated with its activity , and its role in development are unknown . In this study , we have generated loss-of-function mutants for the three components of the RES complex in zebrafish and showed that they are required during early development . The mutants showed a marked neural phenotype with increased cell death in the brain and a decrease in differentiated neurons . Transcriptomic analysis of bud13 , snip1 ( pml1 ) and rbmx2 ( snu17 ) mutants revealed a global defect in intron splicing , with strong mis-splicing of a subset of introns . We found these RES-dependent introns were short , rich in GC and flanked by GC depleted exons , all of which are features associated with intron definition . Using these features , we developed and validated a predictive model that classifies RES dependent introns . Altogether , our study uncovers the essential role of the RES complex during vertebrate development and provides new insights into its function during splicing .
Splicing is critical step in eukaryotic gene expression and is an important source of transcriptomic complexity [1] . Splicing is carried out by the spliceosome , a large macromolecular complex that includes five small nuclear ribonucleoproteins ( snRNPs; U1 , U2 , U4 , U5 and U6 ) and hundreds of core and accessory proteins that ensure the accurate removal of introns from pre-mRNAs [2] . Canonically spliced introns are removed through two transesterification reactions during a complex process involving the recruitment and release of multiple core splicing factors [3] . However , many introns are recognized by different mechanisms depending on their specific features . Short introns with high GC content are believed to be spliced through an “intron definition” mechanism , in which initial U1-U2 pairing occurs across the intron . On the other hand , long introns surrounding short exons are recognized and spliced through “exon definition” mechanisms , in which the initial pairing bridges across the exon [4] . While these mechanisms are widely accepted , little is known about the specific factors associated with each process . The pre-mRNA REtention and Splicing ( RES ) complex is a spliceosomal complex conserved from yeast to human . It is organized around the U2 snRNP-associated protein Snu17p/Ist3p ( RBMX2 in human ) , which binds to both the pre-mRNA-leakage protein 1 ( Pml1p; SNIP1 in human ) and bud site-selection protein 13 ( Bud13p; BUD13 in human ) [5–7] . Snu17p interacts directly with the pre-mRNA and with Bud13p , and both proteins have been involved in splicing; on the other hand , Pml1p has been mainly linked to the retention of unspliced pre-mRNA in the nucleus [6–9] . Furthermore , the components of the RES complex cooperatively increase the stability and the binding affinity of the complex for the pre-mRNA [10–13] , highlighting the importance of cooperative folding and binding in the functional organization of the spliceosome [10–12] . In yeast , the RES complex interacts with the 3’end of the intron in the actin pre-mRNA and is required for the first catalytic step of splicing [7 , 13] . Additionally , microarray-based studies have shown a global effect of the RES complex on yeast splicing [14 , 15] . Mutations in the RES-complex genetically interact with other spliceosomal components , and several introns seem particularly sensitive to RES complex loss-of-function , often in association with weaker splice sites [6 , 9 , 13 , 16–18] . Interestingly , disruption of the genes encoding for the three subunits of the RES complex show consistent phenotypes including slow growth , thermosensitivity [6] and alteration in budding pattern [19] . While the RES complex was identified in yeast , its function in vertebrates , the features recognized by this complex and its role during development are unknown . Here , we found that expression of the RES complex is enriched in the CNS during early development . We generated loss-of-function mutants for the three components of the RES complex in zebrafish using an optimized CRISPR-Cas9 gene editing system [20] . The three mutants showed severe brain defects with a significant decrease in the number of differentiated neurons and increased cell death in the brain and the spinal cord . We observed a mild retention across most introns , consistent with a global effect on splicing . However , a subset of introns was strongly affected . Importantly , these retained introns showed the hallmarks of intron definition [4 , 21] , as they: ( i ) were shorter , ( ii ) had a higher GC content , and ( iii ) were neighbored by lower GC content exons . We developed a logistic regression model with these and other genomic characteristics that allowed us to discriminate between RES-dependent and independent introns with high accuracy . Altogether , these results provide new insights into the function of the RES complex and identify the features associated with RES-dependent splicing .
To determine the role of the RES complex during vertebrate development , we first analyzed its expression pattern during development . bud13 , rbmx2 and snip1 ( Fig 1A ) are maternally expressed ( Fig 1B , S1D and S1F Fig ) , and later in development their mRNAs are strongly expressed in the central nervous system ( CNS ) ( 26 hours post fertilization , hpf ) , ( Fig 1B ) suggesting that RES complex may be required for brain development . Next , we generated mutant zebrafish lines for rbmx2 , snip1 and bud13 using an optimized CRISPR-Cas9 system ( S1A Fig ) [20] . We identified a seven-nucleotide deletion in bud13 ( bud13∆7/∆7 ) , a sixteen-nucleotide deletion in rbmx2 ( rbmx2∆16/∆16 ) and eleven-nucleotide deletion in snip1 ( snip1∆11/∆11 ) . These mutations are predicted to cause premature stop codons and disrupt protein function ( Fig 1C–1E ) . Zygotic mutants for the three components of the RES complex showed a Mendelian ratio of homozygous mutant embryos . We observed strong structural brain defects and widespread cell death in the CNS at ~30 hpf in bud13 and ~48 hpf in rbmx2 and snip1 ( Fig 1F–1H; S1B , S1C , S1E and S2B Figs ) . The embryo progressively degenerates and mutants die by 4–5 dpf . Earlier depletion of bud13 gene expression using morpholino antisense oligonucleotide targeting the AUG start site showed a more severe phenotype affecting not only the brain and CNS but also other tissues ( e . g . mesoderm ) ( S2A Fig ) , consistent with a repression of the maternal contribution in the morphants compared to the zygotic mutants . Both , the mutant and morphant phenotypes were specific and fully rescued by injection of the cognate mRNA ( bud13 , rbmx2 or snip1 ) ( Fig 1F–1H ) or human mRNA ( hBUD13 ) up to 6 dpf ( S2B Fig ) . Interestingly , rescuing with lower amounts of hBUD13 mRNA partially phenocopied the rbmx2 and snip1 mutant phenotypes at 48 hpf ( S2C Fig ) . These results suggest that the onset of the zygotic phenotype in rbmx2 and snip1 mutants ( 48 hpf ) is likely due to differences in the maternal mRNA contribution and/or protein stabilities for these genes . Taken together , these results demonstrate that ( i ) the mutant phenotypes are specific to the targeted loci , ( ii ) the RES complex is essential for embryonic development and suggest that ( iii ) the biochemical function of Bud13 , and presumably of the RES complex , may be conserved from human to zebrafish . To determine the role of the RES-complex in splicing and gene expression , we analyzed polyA+ RNA from each mutant at the onset of the phenotype . As a control , we analyzed the transcriptome of stage-matched wild type siblings ( See material and methods ) . We observed 621 up-regulated genes in all three mutants , that were significantly enriched for functions related to i ) cell death ( e . g . p53 , casp8 and puma ) ( S4D Fig ) , consistent with the appearance of apoptosis in the brain ( Fig 2A and S1B Fig ) , and ii ) spliceosomal components , suggesting a compensatory effect upon a general splicing deficiency ( See Materials and Methods for details , S4A , S4C Fig and S1 Table ) . In contrast , 745 genes were consistently down-regulated and were enriched for genes involved in transcriptional regulation ( such as sox19b , atoh7 , pou3f1 ) and nervous system development ( e . g . neurod1/4/6b , sox1a ) ( S4B Fig and S1 Table ) . Importantly , bud13∆7/∆7 mutants showed normal neural induction , morphogenesis and regionalization . For example , key brain areas such as the zona limitans intrathalamica ( ZLI; dorsal shh in the diencephalon ) , the mid/hind-brain boundary ( pax2a expression ) , and rhombomeres 3 and 5 ( krox20 expression ) were properly specified ( S3 Fig ) . This suggests that the zygotic function of bud13 is not required for initiation of neural linage patterning and specification . In contrast , we observed a reduction in the number of differentiated neurons , including both excitatory and inhibitory neuronal populations , with significant decrease in glutamatergic as well as GABAergic neurons in the forebrain in bud13∆7/∆7 embryos at 32 hpf ( Fig 2B and 2C ) . These results suggest that zygotic RES activity is required for neuronal differentiation and/or survival , but not for neural induction and early brain patterning . To determine the global impact of the RES complex mutants on splicing , we analyzed the level of intron retention ( IR ) across the transcriptome . Briefly , for any given intron , the percent intron retention ( PIR ) is calculated as the average number of reads mapping to the 5’ and 3’ Exon-Intron ( E-I ) junctions over the average number of reads mapping E-I junctions plus any Exon-Exon ( E-E ) junction that supports removal of that given intron ( Fig 3C ) [22] . We found that 74–79% of introns showed increased retention ( ΔPIR > 0 ) in the individual mutants compared to wild-type siblings ( 72 , 926 introns , with sufficient coverage , see Methods for details ) . In contrast , a significantly smaller set of introns ( 7–9% ) showed decreased retention in the mutants ( Fig 3A and S7C Fig , P<2 . 2e-16 , Fisher exact test ) . Other types of splicing events were less impacted upon RES depletion ( Fig 3B ) , although a substantial number of exons became skipped ( S7D Fig ) in the loss-of-function mutants , consistent with a general disruption of splicing . Interestingly , the three mutants shared 5 , 339 ( ~35% ) introns with a medium-high level of retention ( ΔPIR>5 ) ( Fig 3D ) , consistent with a common function in the RES complex . This is likely an underestimate because each mutant was analyzed at the onset of the mutant phenotype , which is different across these mutants likely due to different level of maternal recue or protein stability . We observed mild widespread intron retention , yet ~3 . 5% of introns showed strong increased retention across each mutant ( ΔPIR >15 , S7C Fig ) , suggesting that a subset of introns have a stronger dependence for RES function in vivo . This effect was validated by RT-PCR for a subset of candidates for each mutant ( Fig 3E , Fig 3F and S5 Fig ) , including wdr26b and ptch2 , that when mutated in humans cause neurodevelopmental disorders [23 , 24] . Genes with strongly retained introns ( ΔPIR >15 ) in at least two of the three mutants were enriched in transcriptional regulation and DNA binding ( S6A Fig ) , including known regulators of vertebrate development and neuronal differentiation ( e . g . irx1a , smn1 , enc1 , smad4 , tbx2a , and nkx6 . 1/6 . 2; S2 Table ) . Further analyses revealed complex interactions between pre-mRNA splicing and mRNA abundance in response to RES complex depletion . Genes with at least one strongly retained intron ( ΔPIR >15 ) had lower expression in the mutants compared to genes without intron retention ( ΔPIR < 2 ) ( S6B Fig; P<10–7 for the three mutants , Wilcoxon Sum Rank test ) . Conversely , genes that were differentially expressed in the three mutants ( down- or up-regulated ) show significantly higher retention ( higher ΔPIR ) in the mutants compared to their WT siblings ( S6C Fig; P<10–8 for all comparisons in the three mutants , Wilcoxon Sum Rank test ) . Finally , genes with increased expression significantly overlapped with those with at least one strongly retained intron ( ΔPIR>15 ) ( S6D Fig ) . Taken together , our analyses indicate that the RES complex has a global effect on splicing , and is strongly required for a subset of introns in genes involved in transcriptional regulation and neural development . To identify features that are primarily associated with RES complex function , we first defined a set of introns that were confidently dependent on RES for proper splicing ( 1 , 409 “RESdep” introns with ∆PIR>15 , ≥1 . 5-fold net increase in intron reads; see Methods ) ( Fig 3C ) . As a control set ( Ctr ) , we defined 5 , 574 introns with ∆PIR<0 . 5 in all three mutants , and evaluated the enrichment of 44 features , many of which have been previously associated with intron retention [22] ( S3 Table ) . Consistent with the genome-wide patterns ( Fig 4 ) , strongly retained introns were enriched for last introns ( Fig 4A and 4B and S7A Fig ) and introns that do not trigger NMD when mis-spliced ( Fig 4A and 4C and S7B Fig ) , suggesting that their accumulation is in part likely due to reduced degradation of the unspliced transcript isoform . However , a significant fraction of highly retained introns was predicted to elicit NMD upon inclusion . We thus hypothesized that this subset of NMD-triggering introns contains specific features that would maximally associate with RES-dependent mechanisms . Based on this , we also separately analyzed introns that were predicted to trigger NMD ( 574 ) and those that were not ( 569 ) ( see Methods ) . RES-dependent introns ( i ) were significantly shorter than the control set ( Fig 5A , median of 281 nt versus 749 nt in the control , P<0 . 001 , Mann-Whitney-U test ) , ( ii ) had elevated GC content ( Fig 5B , P<0 . 001 Mann-Whitney-U test ) , ( iii ) were flanked by exons with lower GC content ( Fig 5C and 5D , S8C and S8D Fig ) , and ( iv ) had weaker branch point ( BP ) consensus sequences [25 , 26] ( Fig 5G and S8 Fig ) . Remarkably , at the genome-wide level , short introns and introns with high GC content also showed a higher retention across all three mutants compared to wild type siblings ( S9 Fig ) . NMD-triggering introns further showed weaker core splicing signals , including acceptor ( 3’ ) and donor ( 5’ ) splice sites and BP consensus sequences than any other intron set ( Fig 5E–5G ) , providing an explanation for their increased sensitivity upon RES complex disruption despite being subject to NMD . Next , we assessed how genomic and transcript features define the dependence on the RES-complex . We applied a logistic regression model using 30 features as predictor variables ( S3 Table ) , and defined a response variable classifying each intron as RES-dependent or non-dependent . We developed a model using 90% of the RESdep and a size-matched subset of control introns as training set , and validated the model on the remaining 10% of the data ( see Material and Methods ) . We found a high performance in the classification of introns , with an average Area Under the ROC curve ( AUC ) of 0 . 821 ( Fig 6A ) . This model was able to classify the impact of mutating each individual RES component with an AUC of ≥0 . 76 . Subsequently , we analyzed the individual contribution of each feature to the model and their potential to reduce the null deviance ( see Methods ) . Consistent with results in Fig 5 , this analysis identified four important features i ) the ratio between exon and intron length , ii ) the ratio between exon and intron GC content , iii ) gene expression levels and iv ) the position of the intron within the transcript ( last intron effect ) ( Fig 6B and S10 Fig ) . To further test the model on an independent set of introns , we applied the model to 108 , 470 introns ( S6 Table ) without sufficient read coverage across our six RNA-seq samples and were not used for previous analyses . We then selected the top 100 introns predicted to be bud13-dependent and -independent and plotted their ∆PIR values based on RNA-seq data for the bud13 mutant and the corresponding control ( S11A Fig ) . The majority of predicted bud13-dependent introns were substantially retained , whereas introns that were predicted to be unaffected showed a median ∆PIR close to zero ( 0 . 91 ) ( S11A Fig ) . The false prediction rate ( FPR ) is 0 . 34 for the bud13 dependent introns and 0 . 16 for the unaffected introns , consistent with the AUC values reported above . Repeating this independent validation with rbmx2 and snip1 dependent introns ( S11A Fig ) showed FPRs of 0 . 33 and 0 . 03 for rbmx2 , and 0 . 44 and 0 . 04 for snip1 . The lower FPRs for the unaffected introns reflect that predicting unaffected introns can be obtained with higher performance , likely because the data contain a considerably larger amount of unaffected than affected introns . We further validated these predictions by RT-PCR assays for five predicted RES-dependent and five RES-independent introns ( S11B Fig ) . Thus , altogether , our logistic regression analysis can identify introns dependent on the RES complex based on specific features within the genomic locus and the transcript . Finally , to test our logistic regression model in isolation from the genomic context , we assayed four introns , two RES dependent and two RES independent , using minigene constructs including the tested intron plus the two flanking exons ( Fig 6C and 6D and S12 Fig ) . Minigenes were cloned into a transgenesis vector ( S12B Fig ) , and injected into 1-cell stage embryos . To assess the splicing pattern , we carried out a RT-PCR followed by PCR ( S4 Table ) . From the subset of four different introns , three events were validated by the RT-PCR and gel electrophoresis as predicted by our model ( Fig 6A , 6B , 6C and 6D and S12 Fig ) . These introns were in serpinb1l3 and col1a2 ( predicted as RES independent ) and wdr26b ( predicted as RES dependent with ∆PIR>15 ) transcripts . Nevertheless , we detected no intron retention in ptch2 , suggesting that its retention upon RES depletion may depend on its genomic context .
Splicing regulatory information is encoded by multiple sequence features , from the core signals ( splice donor and acceptor and branch point ) to other , less understood , sequence elements [27–29] . Our results identify intronic features that are associated with RES-dependent splicing across the transcriptome . These features can be used to discriminate a large fraction of RES-dependent from independent introns . Although loss-of-function for RES complex components caused mild intron retention across the transcriptome , we observed a subset of introns that were strongly accumulated across mutants of the RES complex . Recent in vitro studies on the single intron of the actin gene in yeast showed that the RES complex binds at the 3’ of the intron , between the BP and the acceptor site [12] . We observed that introns that more strongly depend on the RES complex show weaker BP consensus sequences . Furthermore , this subset of introns were shorter and had higher GC content , an association that is particularly striking in zebrafish , since short introns normally have lower GC content [21] . RES-dependent introns are flanked by exons with a lower GC content than RES-independent introns ( Fig 5 , S9C and S9D Fig ) . Therefore , these observations are consistent with a model whereby RES-dependent introns are mainly spliced through intron definition [4 , 21] . This association is surprising , since biochemical evidence suggest that the RES complex joins the spliceosome after recognition of the splice sites [12] , and that the RES complex is not needed for spliceosome assembly in vitro but for U1 and U4 snRNP dissociation before the first catalytic step [7] . One possible explanation for this apparent discrepancy is that RES complex components play a role in early splice site recognition in vivo and therefore that the biochemical functions reported in a limited set of RNAs reflect limitations of in vitro splicing reactions . Alternatively , the RES complex may not be involved in early splice site recognition , but could be a limiting factor for splice site pairing or other steps in spliceosome assembly progression in vivo , particularly for introns defined by intron definition , highlighting differences in molecular pathways for intron- and exon-defined splicing . These concepts are in line with increasing evidence that splice site selection can be modulated at late stages of spliceosome assembly or even catalysis [14 , 27 , 30–35] . Finally , some RES-dependent introns also have weaker donor and acceptor splice site consensus sequences , and thus are expected to be more sensitive to defects on the splicing machinery . This is consistent with previous studies in yeast , which found that weaker 5’ splice sites increased susceptibility to RES loss-of-function [6] . An unexpected observation from genome-wide analyses of core splicing factor loss-of-function experiments is that each factor seems to differentially affect a specific subset of introns and exons [32 , 36] . This suggests that splicing of each intron in the genome is limited by specific core factors , depending on its combination of sequence features , as we observed for RES-dependent introns . As such , disruption of core splicing factors is predicted to produce unique phenotypes dictated by its expression , and the expression and function of genes that contain the subsets of introns sensitive to that factor . Consistent with this hypothesis , RES complex is required during brain development and neuronal survival , and mis-regulated introns are found in genes with well-known functions in neurodevelopment ( e . g . irx1a , smn1 , enc1 , smad4 , tbx2a , and nkx6 . 1/6 . 2 ) . Specifically , zygotic mutants in bud13 , snip1 or rbmx2 show microcephaly and decreased populations of GABAergic and glutamatergic neurons , despite normal specification and regionalization of the CNS ( Fig 1 , Fig 2 and S3 Fig ) . This phenotype is different from those described for a few other spliceosomal-related mutants in zebrafish [37–40] . For instance , while sf3b1 is required for early neural crest development [40] , loss of another core component of the spliceosome , prpf8 , results in massive neuronal cell death and impaired myeloid differentiation [37] . These differences might be caused by the different half-life of the maternal proteins in the zygotic mutants . Alternatively , different components of the splicing machinery might be essential in a cell-type/tissue specific manner during early development . This may also explain why mutations in specific spliceosomal components cause human diseases with diverse phenotypes , such as Taybi-Linder syndrome , microcephalic osteodysplastic primordial dwarfism type I and retinitis pigmentosa [41–43] . Interestingly , a mutation in human SNIP1 ( p . Glu366Gly ) has been associated with epilepsy and skull dysplasia [44] . Our data shows that human BUD13 can rescue loss of bud13 function in zebrafish , and future studies will be needed to determine whether Bud13 has a conserved function during brain development in humans ( S2B and S2C Fig ) . In summary , we have shown that RES complex disruption in zebrafish hinders splicing , but is not essential for the removal of most introns , indicating that such introns can be efficiently defined and spliced through RES-independent mechanisms . However , we found that a subset of introns is particularly affected by RES complex removal and that those introns display the major hallmarks of splicing through intron definition mechanisms . From a functional perspective , RES-dependent introns are in genes enriched for transcription factors and neurodevelopmental regulatory functions , thus resulting in brain developmental defects in loss-of-function zygotic mutants ( Fig 7 ) . Future studies will be needed to understand how spliceosomal mutations disrupt splicing of different genes by affecting specific limiting steps in pre-mRNA splicing resulting in diverse disease phenotypes .
Fish lines were maintained in accordance with research guidelines of the International Association for Assessment and Accreditation of Laboratory Animal Care , under a protocol approved by the Yale University Institutional Animal Care and Use Committee ( IACUC ) . Wild-type zebrafish embryos were obtained through natural mating of TU-AB and TLF strains of mixed ages ( 5–17 months ) . Selection of mating pairs was random from a pool of 48 males and 48 females allocated for a given day of the month . bud13Δ7/Δ7 , rbmx2Δ16/Δ16 , snip1Δ11/Δ11 were obtained through natural mating of heterozygous bud13+/ Δ7 , rbmx2+/ Δ16 , snip1+/ Δ11 mutants , respectively ( see below gene editing using CRISPR-Cas9 ) . Tg ( dlx6a-1 . 4kbdlx5a/dlx6a:GFP ) lines [45 , 46] were obtained from the laboratory of Marc Ekker and Tg ( vglut:DsRed ) [47] from the laboratory of Joseph Fetcho . Embryos were analyzed using a Zeiss Axioimager M1 and Discovery microscopes and photographed with a Zeiss Axiocam digital camera . Images were processed with Zeiss AxioVision 3 . 0 . 6 . Whole-mount embryos were imaged in vivo by confocal microscopy ( Leica TCS SP8 systems , Yale Center for Cellular and Molecular Imaging . ) Mutant embryos and their wild-type siblings were scored at 30–32 hours post fertilization stage . Embryos were anesthetized using Tricaine and mounted in 0 . 6% agarose at an orientation where the frontal view of the brain was imaged . Embryos were imaged at 40x ( 1 . 3 Oil ) using Z-stacks ranging from 70 . 41–96 . 02 μm , with a z stepping size of 0 . 4 μm . Z-stacks started at the first appearance of the GABAergic cells ( GFP-labeled ) and ended where GABAergic cells ( GFP-labeled ) could no longer be visualized . Each xy plane spanned 227 . 94 μm with a pixel size of 0 . 075 μm . Maximum intensity projections were shown for all confocal images . Images were processed using Fiji [48] , Imaris ( Bitplane ) and Huygens deconvolution software ( Scientific Volume Imaging ) . Figures were assembled using Illustrator ( CC , Adobe ) . To quantify the number of glutamatergic ( labeled in DsRed ) and GABAergic cells ( labeled in GFP ) in the bud13Δ7/Δ7 mutant and their wild-type siblings respectively , two blinded raters segmented the raw z-stack images using ImarisCell module ( Bitplane ) and computationally counted the segmented cells in each channel ( GFP and DsRed ) . Identical thresholds and parameters were applied to all samples for segmentation processing . Since the quantification performed by both independent raters yield consistent fold change in the respective cell counts between the bud13Δ7/Δ7 mutant and their wild-type siblings . Only one set of the analyzed results was displayed . Statistical analyses were conducted using Prism 6 ( Graphpad ) . To visualize apoptotic cells , vital dye acridine orange ( Sigma ) was used in live and dechorionated embryos . Embryos were incubated 2 minutes in PBS pH 7 . 1 with 2 ug/ml of acridine orange in the dark . After 3 brief washes in PBS , the embryos were placed in plates with 1% agarose and viewed with fluorescence microscopy , using the FITC filter set 1 [49] . Zebrafish bud13 , rbmx2 and snip1 ORFs were PCR amplified ( S4 Table ) using cDNA from 2 and 6 hpf zebrafish embryos and cloned in pSP64T ( bud13 ) in pT3TS [50] ( rbmx2 and snip1 ) . bud13 PCR product was cut using NotI and EcoRI and ligated into pSP64T . rbmx2 and snip1 PCR products were cut with NcoI and SacII and ligated into pT3TS . To optimize Kozak sequence , the forward oligonucleotide used for rbmx2 ORF introduced extra aminoacid in the 2nd position ( GGC ) . Human bud13 ORF was cloned in pSCDest [51] using gateway gene cloning system ( Thermo Fisher Scientific ) . Final constructs were confirmed by sequencing . To generate mRNAs , the template DNA was linearized using XbaI ( pT3TS ) , BamHI ( pSP64T ) or KpnI ( pSCDest ) and capped mRNA was synthetized using the mMessage mMachine T3 ( pT3TS ) , or SP6 ( pSP64T and pSCDest ) kit ( Ambion ) , respectively and in accordance with the manufacturer’s instructions . In vitro transcribed mRNAs were DNAse treated and purified using the RNeasy Mini Kit ( Qiagen ) . All mRNA rescued the mutant phenotypes when 50–100 pg were injected in one cell stage embryo . The plasmid ( modified from [52] ) for intron retention validation in vivo , pTol2 ( hsp-MCS-polyA , CMV-eGFP-SV40 ) , is a Tol2 transposon-based , bipartite construct consisting of heat-shock promoter ( hsp ) , a multiple cloning site ( MCS ) , to insert the desired validation cassette , flanked by Xenopus globin UTR and polyadenylation tail ( polyA ) as well as a cis-linked CMV promoter and SV40 poly ( A ) -regulated eGFP reporter . Briefly the plasmid was built as following: pT2 ( kop:Cre-UTRnos3 , CMV:eGFP ) [52] was cut with Kpn1 to remove kop:Cre-UTRnos3 . A synthetic DNA fragment containing HSP , 5’UTR Xenopus globin , MCS and 3’UTR Xenopus globin was obtained from Integrated DNA Technologies ( IDT ) and PCR amplified with specific primers ( S4 Table ) . In-Fusion cloning protocol ( Clontech ) was performed using the cut vector and the PCR product to get the backbone vector pTol2 ( hsp-MCS-polyA , CMV-eGFP-SV40 ) . Cassette for validation were obtained as synthetic DNA fragments ( Integrated DNA Technologies , IDT ) or amplified from genomic DNA ( see S4 Table for details ) and cloned directionally in frame with XhoI and SacII . Final constructs were confirmed by sequencing . A morpholino targeting bud13 mRNA start codon was obtained from Gene Tools and re-suspended in nuclease-free water . 1 nl of morpholino solution ( 0 . 6mM ) was injected into wild-type dechorionated embryos at the one-cell stage . A mix of 4 plasmid ( 15 pg/embryo ) with the desired cassette ( minigene ) , 2 predicted as no retained ( serpinb1l3 , col1a2 ) and 2 predicted as retained ( wdr26b , ptch2 ) upon RES loss-of-function was injected in 1 cell stage embryo together with Tol2 mRNA ( Addgene plasmid #31831 ) ( 33 pg/embryo ) . Embryos were sorted in bud13∆7/∆7 and bud13+/ ? at the onset of the phenotype ( ~ 28–30 hpf ) . Minigene expression under HSP promoter was induced by heat shock during 4 hs and then GFP positive embryos were collected for RT-PCR . PCR was performed using the specific Fw primer an a universal Rv globin ( S4 Table , Fig 6C and 6D and S12 Fig ) . CRISPR-Cas9-mediated gene editing was performed as described previously [53] . Briefly , 3 different sgRNAs ( 20 pg each ) targeting bud13 gene ( S4 Table ) were co-injected together with 100 pg of mRNA coding for zebrafish codon optimized Cas9-nanos in one-cell stage embryos ( S1A Fig ) . Cas9-nanos concentrates gene editing in germ cells and increases the viability of injected embryos [20] . F0 founders were mosaic and they were backcrossed with wild-type fish and then F1 fish were genotyped using their corresponding oligos per target site ( S4 Table ) . Heterozygous adult fish bud13+/ Δ7 ( Fig 1 ) were selected to generate bud13Δ7/Δ7 mutants . Similar approach was followed to generate rbmx2 and snip1 mutants but injecting 2 sgRNAs ( S4 Table and Fig 1 ) . krox20 , shh and pax2a in situ probes were previously described [54–56] . RES antisense and sense digoxigenin ( DIG ) RNA probes were generated by in vitro transcription in 20 μl reactions consisting of 100 ng purified PCR product ( 8 μl ) , 2 μl DIG RNA labelling mix ( Roche ) , 2 μl ×10 transcription buffer ( Roche ) , and 2 μl T7/T3 ( antisense probes ) and SP6 ( sense probes ) RNA polymerase ( Roche ) in RNase-free water and purified using a Qiagen RNEasy kit . In situ protocol was followed as detailed previously [57] . To reduce variability , wild-type sibling and bud13Δ7/Δ7 embryos were combined in the same tube during in situ hybridization and recognized based on their phenotype . Before photo documentation , embryos were cleared using a 2:1 benzyl benzoate:benzyl alcohol solution . Images were obtained using a Zeiss stereo Discovery V12 . Total RNA from 32 hpf bud13 Δ7/Δ7 , 48 hpf rbmx2 Δ16/Δ16 , 48 hpf snip1 Δ11/Δ11 embryos and their corresponding siblings was extracted using Trizol ( ten embryos per condition ) . Strand-specific TruSeq Illumina RNA sequencing libraries were constructed by the Yale Center for Genome Analysis . Samples were multiplexed and sequenced on Illumina HiSeq 2000/2500 machines to produce 76-nt paired-end reads . RNA used for intron retention validation experiments was treated with TURBO DNase ( Ambion ) for 30 min at 37°C and extracted using phenol chloroform . Then , Polyadenylated RNAs were purified using Oligo d ( T ) 25 Magnetic Beads ( Invitrogen ) following manufacter recommendations . cDNA was generated by reverse transcription with random hexamers using SuperscriptIII ( Invitrogen ) . RT–PCR reactions were carried out at an annealing temperature of 59°C for 35–40 cycles . PCR products were run in a 1 . 5% agarose gel . Primers are listed in the S4 Table . For the qPCR experiment , total RNA was extracted as described above . GFP and dsRED mRNAs were used as spike-in RNA controls and 1 μg of total RNA was used to generate cDNA . 5 μl from a 1/50 dilution of the cDNA reaction was used to determine the levels of p53 in a 20 μl reaction containing 1 μl of each oligo forward and reverse ( 10 μM ) ( S4 Table ) , using Power SYBR Green PCR Master Mix Kit ( Applied Biosystems ) and a ViiA 7 instrument ( Applied Biosystems ) . PCR cycling profile consisted of incubation at 50°C for 2 min , followed by a denaturing step at 95°C for 10 min and 40 cycles at 95°C for 15 s and 60°C for 1 min . Primers are listed in S4 Table . Zebrafish embryos or a small amount of tissue from the end of the tail were used to extract DNA [58] . Briefly , embryos or fin clipped were incubated in 80 μl of NaOH 100mM at 95°C for 15 min producing a crude DNA extract , which was neutralized by the addition of 40 μl of 1 M Tris-HCl , pH 7 . 4 ( Sigma-Aldrich ) . 1 μl of this DNA extraction was used as a template for PCR reactions using the primers described in S4 Table . Gene expression levels for each condition were calculated from RNA-seq data using the cRPKM metric ( corrected-for-mappability Reads Per Kilobasepair of uniquely mappable positions per Million mapped reads [59] . For this , a reference transcript per gene was selected from the Ensembl version 80 annotation for Danio rerio using BioMart ( 25 , 935 genes in total , S5 Table ) and uniquely mappable positions for each transcript were calculated as previously described [59] . Quantile normalization of cRPKM values was done with ‘normalizeBetweenArrays’ within the ‘limma’ package . To identify differentially expressed genes , we first filtered out genes that did not have cRPKM > 2 in all sibling control or all mutant samples and genes whose quantification was not supported by at least 50 read counts in at least 1 sample . Next , differentially expressed genes were defined as those that showed a fold change in expression of at least 1 . 5 in all 3 mutants and a fold change of at least 2 in 2 out of the 3 control vs . mutant individual comparisons ( bud13 , rbmx2 and snip1 ) . Gene Ontology analysis was performed with the online tool DAVID ( https://david . ncifcrf . gov/ Version 6 . 8 ) using as background all genes that passed the initial filters ( minimum expression and read count ) . Annotated introns for each reference zebrafish transcript in Ensembl version 80 were extracted and those that overlapped with other genes were removed yielding a total of 182 , 017 valid introns . To calculate the percent of intron retention ( PIR ) for each intron in a given RNA-seq sample , we used our previously described pipeline [22] with the following modification: to calculate intron removal , all exon-exon junctions supporting the splicing of the intron were used and not only those formed between the two neighboring exons . This was done to avoid false positives in the case of introns associated to cassette exons or other alternative splicing events . For all analysis , only introns with sufficient read coverage across the six samples were considered ( at least 15 reads supporting the inclusion of one splice site and 10 of the other , or a total of 15 reads supporting splicing of the intron ) . To define the confident set of highly affected introns , potential false positives were filter out by comparing the density of the mapped reads in the introns bodies in the mutant vs the control . For this purpose , we extracted all intronic sequences and calculated the number of uniquely mappable positions per intron following a similar strategy to that used to calculate cRPKMs [59] ( see above ) . Specifically , every 50-nucleotide ( nt ) segment in 1-nucleotide sliding intronic windows was mapped to a library of full-length intronic sequences plus the whole genome , using bowtie with–m 2 –v 2 parameters ( every intronic segment must map at least twice , to its own individual intron sequence and to the corresponding position in the whole genome ) . Segments that mapped more than twice were considered as multi-mappable positions , whereas those that did not map ( e . g . due to undetermined ( N ) nucleotides in the assembly ) were considered as non-mappable . The number of uniquely mappable positions of an intron is defined as the total number of segments minus multi- and non-mappable positions . Next , each RNA-seq sample was mapped to the same library of intronic plus full genomic sequences using–m 2 –v 2 to obtain the unique intronic reads counts . However , to minimize potential artifacts derived from the heterogeneity of the intronic sequences ( e . g . high number of reads mapping to a transposable element or an expressed nested gene ) , if a given intronic position showed a read count more than five times higher than the median read count of the whole intron , then the read count of this position was set to 5 × median ( if the median read count was 0 , then the maximum read count for any given position was set to 5 ) . Finally , ciRPKM scores were calculated ( corrected-for-mappability intronic Reads Per Kilobasepair of uniquely mappable positions per Million mapped reads ) for each intron and condition by dividing this number of counts by the number of uniquely mappable positions in that intron . With this information , the set of confidently retained introns upon RES complex disruption ( “RESdep” introns ) were defined as those introns with ∆PIR > 15 and at least a 1 . 5-fold net increase in read density in the intron body calculated as: [ciRPKMmut/ ( 100‑PIRmut ) ]/[ciRPKMsib/ ( 100‑PIRsib ) ] in at least 2 out of 3 mutants ( 1 , 413 introns in total ) . As a control , we also define a set of confidently non-retained introns as those with a ∆PIR < 0 . 5 in the 3 mutants ( “Ctr” set; 5 , 577 introns ) . To analyze the number of retained introns in the 3 mutants and the inter-mutant overlaps Euler APE-3 . 0 . 0 software [60] was utilized . To investigate the impact of non-sense mediated decay ( NMD ) on global intron retention upon RES mutation , all introns were separated as last or non-last introns ( of the reference transcript ) and between those predicted to trigger and not to trigger NMD . An intron was predicted to trigger NMD if its retention generated an in-frame stop codon that is located further than 50 nts upstream of an exon-exon junction [61] . By definition , last introns cannot trigger NMD . ∆PIR values were plotted as boxplots for each category , and two-sided Wilcoxon Sum Rank tests were used to evaluate statistical differences between the distributions . To identify features discriminating introns highly retained upon RES depletion from un-retained/un-affected introns , we compared the sets of confidently introns ( “RESdep” in Fig 5 . ) with control introns ( “Ctr” in Fig 5 . ) . Moreover , as introns that are predicted not to trigger NMD are expected to be more often accumulated unspecifically , two subsets for “RESdep” introns were generated: ( i ) those introns in genes with more than five introns , are not the last three introns of the gene , and that are predicted to trigger NMD ( “NMD” , 577 introns ) ; and ( ii ) predicted not to trigger NMD or cause a frame shift upon inclusion , unless they are the last intron of the gene ( “no-NMD” , 569 introns ) . For these different sets of introns , 44 features were extracted ( S3 Table ) , including intron and exon length and GC content , strength of 3' and 5' splice sites , branch point ( BP ) related features , and transcript length , using custom scripts in combination with the following two external tools: MaxEntScan scripts for determining the strength of 3' and 5' splice sites [62]; and SVM-BPfinder software for determining BP related features ( BP strength , distance from BP to 3' splice site , and pyrimidine track length ) [26] . For the latter analysis , the 150 nts upstream of the 3' splice sites were extracted and these sequences were used as input for SVM-BPfinder . Furthermore , we recorded the highest log-score of the SF1 position weight matrix binding model across these 150-nt intronic sequences [25] . We applied logistic regression models to the discrimination between differentially retained introns ( retained ) and non-differentially retained introns ( control ) upon RES mutations . We focused on the set of confidently retained introns ( “RESdep” , 1 , 409 introns ) versus control introns with an absolute ∆PIR < 0 . 5 in the three mutants ( “Ctr” , 5 , 565; we removed 9 introns for which we could not determine all features ) . The binary response variable of the logistic regression models indicates for each intron if it belongs to the retrained or control group . As predictors we used 30 quantitative and qualitative features ( S3 Table ) , including intronic and exonic characteristics , position along the transcript and gene expression in wild type conditions , among others . The binomial logistic regression models were learned using Lasso variable selection [63–65] available in R through library glmnet ( 2 . 0 . 10 ) , and the generalized linear model function glm from the R stats library . To investigate the overall classification performance , we randomly partitioned the data set “RESdep” of retained introns into 90%/10% ( i . e . , 1268/141 ) training/test data , and randomly sampled the same amount of training/test data from the control “Ctr” data set . We used the training data to learn a logistic regression model with Lasso variable selection and tested it on the test data . Next , to evaluate how well this model classifies specific subsets of "RESdep" introns and retained introns specific for each mutation , we applied the model trained with “RESdep” vs “Ctr” data having held fixed its parameters to the classification of the 141 control test-introns vs . 141 retained introns subsampled from the following sets: ( i ) “RESdep_∆PIR10” introns from the “RESdep” set with ∆PIR > 10 in all three mutants ( 871 introns ) ; ( ii ) “NMD” , introns from the “RESdep” set predicted to trigger NMD when retained ( 574 introns ) ; ( iii ) “bud13” , introns with ∆PIR>15 upon bud13 mutation at 32 hpf ( 2 , 363 introns ) ; ( iv ) “rbmx2” , introns with ∆PIR>15 upon rbmx2 mutation at 48 hpf ( 2 , 186 introns ) ; and ( v ) “snip1” , introns with ∆PIR>15 upon snip1 mutation at 48 hpf ( 2 , 675 introns ) . We repeated this procedure , including model training and classification of test data , 10 , 000 times and report average ROC curves ( Fig 6A ) and average model coefficients for each feature extracted from the trained models . These averages indicate the direction of the effect ( e . g . positively [blue] or negatively [red] associated with retention upon RES mutation; Fig 6B; S10B Fig and S3 Table ) . To study the potential of each feature to contribute to the discrimination between the "RESdep" and "Ctr" intron sets , we randomly partitioned the dataset of retained introns into 90%/10% ( i . e . , 1268/141 ) training/test data , and randomly sampled the same amount of training/test data from Ctr . Using the training data , we learned logistic regression models without Lasso variable selection using only a single feature at a time neglecting all other features . The test data were used to determine the AUC . This experiment was repeated 10 , 000 times and we report average AUCs for each feature in Fig 6B and S10B Fig . In addition , the fraction of the null deviance that was reduced by each single-feature model was recorded , and the average reductions of the null deviance for each feature are reported in S3 Table . | RES complex is essential for splicing in yeast but its function and role during vertebrate development are unknown . Here , we combined genetic loss-of-function mutants with transcriptomic analysis and found that a subset of introns is particularly affected in RES complex knock-out background . Those introns display the major hallmarks of splicing through intron definition mechanisms ( short introns , rich in GC and flanked by GC depleted exons ) . Moreover , bud13 , rbmx2 and snip1 mutant embryos showed a marked brain phenotype with a RES-dependent introns enrichment in genes with neurodevelopmental function . Altogether , our study unveils the fundamental role of RES complex during zebrafish embryogenesis and provides new insights into its molecular function in splicing . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"fish",
"neuronal",
"differentiation",
"vertebrates",
"animals",
"cell",
"differentiation",
"animal",
"models",
"osteichthyes",
"developmental",
"biology",
"mutation",
"model",
"organisms",
"experimental",
"organism",
"systems",
"molecular",
"biology",
"techniques",
"embryo... | 2018 | RES complex is associated with intron definition and required for zebrafish early embryogenesis |
In bacterial cells , gene expression , metabolism , and growth are highly interdependent and tightly coordinated . As a result , stochastic fluctuations in expression levels and instantaneous growth rate show intricate cross-correlations . These correlations are shaped by feedback loops , trade-offs and constraints acting at the cellular level; therefore a quantitative understanding requires an integrated approach . To that end , we here present a mathematical model describing a cell that contains multiple proteins that are each expressed stochastically and jointly limit the growth rate . Conversely , metabolism and growth affect protein synthesis and dilution . Thus , expression noise originating in one gene propagates to metabolism , growth , and the expression of all other genes . Nevertheless , under a small-noise approximation many statistical quantities can be calculated analytically . We identify several routes of noise propagation , illustrate their origins and scaling , and establish important connections between noise propagation and the field of metabolic control analysis . We then present a many-protein model containing >1000 proteins parameterized by previously measured abundance data and demonstrate that the predicted cross-correlations between gene expression and growth rate are in broad agreement with published measurements .
Few processes are more fundamental to life than the growth and proliferation of cells . Bacterial cells in particular are highly adapted to grow rapidly and reliably in diverse habitats [1] . Yet , the composition of individual bacteria grown in a constant environment is known to fluctuate vigorously , in part due to the stochastic nature of gene expression [2–5] . Many experimental and theoretical studies have shed light on the origins , characteristics and consequences of this “noisy” expression [2–17] . Still , it remains unknown to what extent , and by what routes , noise in gene expression propagates through the cell and affects the rate of growth [5 , 18 , 19] , which is often considered a proxy for its fitness [18 , 20] . Recently , important progress towards understanding noise propagation in single cells has been made through experiments in which the instantaneous growth of individual Escherichia coli cells was monitored in real time under fixed growth conditions [5 , 21] . Such experiments have revealed large fluctuations in the growth rate , with coefficients of variation of the order of 25% , which in part result from noise in the concentrations of metabolic enzymes [5] . Conversely , growth-rate fluctuations affect the concentrations of individual enzymes , because the cell’s constituents are diluted whenever the cell grows [22] . Such results emphasize that a clear understanding of these processes is complicated by the fact that gene expression , metabolism , and growth are highly interdependent , involving multiple layers of feedback and cellular constraints . This interdependence is also central to a series of recent studies that characterize the average composition and growth rate of Escherichia coli cultures in balanced exponential growth under variation of the growth medium [23–29] . In particular , these experiments have revealed striking linear relations between their mean proteomic composition and their mean growth rate [26–31] . Phenomenological models have demonstrated how such “growth laws” can be understood as near-optimal solutions to constrained allocation problems [20 , 32–34] . These results also stress that global physiological variables and constraints strongly affect the expression of individual genes . As such , both these experiments and the single-cell experiments mentioned above suggest a “holistic” perspective: the behavior of individual components cannot be understood without some knowledge of the cell’s global physiological state [35 , 36] . Here , we present a model of bacterial cells growing under fixed external growth conditions , in which gene expression , metabolism and growth are fully integrated . We offer a highly simplified description that nevertheless imposes several essential global cellular constraints . Both gene expression and growth rate fluctuate due to the stochastic synthesis of many protein species that together control the rates of metabolism and growth . Conversely , the rate of metabolism constrains the protein synthesis rates and the growth rate sets the dilution rate of all proteins . As a result , noise in the expression of each gene propagates and affects the expression of every other gene as well as the growth rate—and vice versa . Below , we first introduce the generic modeling framework and its assumptions . We then make an excursion to the theory of growth control , in order to define growth-control coefficients and establish connections between the propagation of noise and the field of Metabolic Control Analysis . Next , we discuss how the concentration of each protein is affected by the synthesis noise in all other proteins; this exposes a hidden assumption in a standard operational definition of intrinsic and extrinsic expression noise . We subsequently explain the noise modes that characterize the noise propagation between gene expression and growth in the context of a toy model with just two proteins . Lastly , we present a many-protein model that includes 1021 protein species with experimentally measured parameters . We demonstrate that the cross-correlations functions between expression and growth rate predicted by this model capture the main features of published measurements .
We here discuss the key assumptions of the modeling framework ( Fig 1 ) ; see S1 Text , pp . 1–6 for details . We consider a culture of bacterial cells that has reached steady-state exponential growth under fixed external growth conditions . We study fluctuations of gene expression within individual cells in this steady state , and in particular how these fluctuations reverberate through the growing cell . Similar assumptions connecting the increase in biomass , the cellular growth rate , protein synthesis , and growth-mediated dilution were explored in a recent review article [37] . The mass density of E . coli cells is dominated by protein content [38] and under tight homeostatic control [39] . We assume that this homeostasis also eliminates long-lived protein-density fluctuations in single cells . Then , the volume of a cell is proportional to its protein mass M ≔ ∑i ni , where ni is the abundance ( copy number ) of protein i . ( We ignore that different proteins have different molecular weights . ) The instantaneous growth rate is then defined by μ≔M ˙ / M , and the proteome fraction ϕi ≔ ni/M of enzyme i measures its concentration . Differentiation of ϕi with respect to time then yields ϕ i ˙ = π i - μ ϕ i , ( 1 ) where πi is the synthesis rate per protein mass . ( Here we neglect active protein degradation , which on average amounts to about 2% of the dilution rate [40] . ) By definition , proteome fractions obey the constraint ∑i ϕi = 1 . Combined with Eq ( 1 ) this results in μ = ∑ i π i . ( 2 ) That is , the growth rate equals the total rate of protein synthesis . Another key assumption of our model is that the cellular growth rate is an intensive quantity . That is: given fixed mass fractions , the growth rate does not depend on the cell size , as suggested by the observation that individual E . coli cells grow approximately exponentially within their cell cycle [5 , 41] . Based on this , we express the synthesis rate of protein i as: π i = f i μ d ( ϕ ) + N i , ( 3 ) in which μ d ( ϕ ) ≔J / M . ( 4 ) The first term in Eq ( 3 ) is an intensive function; it captures the deterministic effect of the cellular composition ϕ = ( ϕ1 , ϕ2 , … ) on the metabolic flux J that quantifies the rate of biomass production , normalized by the protein mass M . ( Note that , here and below , we use the term metabolism in a broad sense; it is intended to encompass all catabolic and anabolic processes required for biomass production and cell growth , including protein synthesis . ) The coefficients fi specify which fraction of this flux is allocated towards the synthesis of protein species i . Because the fi are fractions , ∑i fi = 1 . The second term of Eq ( 3 ) couples each synthesis rate πi to a zero-mean Ornstein–Uhlenbeck noise source Ni that represents the stochasticity of both transcription and translation [42] . Each noise source is characterized by an amplitude θi and a rate of reversion to the mean βi; the latter’s inverse β i - 1 characterizes the time scale of intrinsic fluctuations in πi . The variance of Ni is given by Var ( N i ) = θ i 2 / ( 2 β i ) . All noise sources are mutually independent , and we neglect other sources of noise , such as the unequal distribution of molecules over daughter cells during cell division ( see Discussion ) . Combining Eqs ( 2 ) and ( 3 ) reveals that μ = μ d ( ϕ ) + ∑ i N i , ( 5 ) which identifies μd ( ϕ ) as the growth rate afforded by a given proteome composition ϕ in the zero-noise limit . Given a function μd ( ϕ ) , Eqs ( 1 ) – ( 3 ) fully define the dynamics of the cell . Below , we focus on the simplest case where , under given environmental conditions , the allocation coefficients fi are constant . This means that the cell does not dynamically adjusts its allocation in response to fluctuations in expression levels . We note , however , that such dynamical effects of gene regulation could be included by allowing the fi to depend on intra- and extra-cellular conditions , and in particular on the cellular composition ϕ . ( See S1 Text , p . 4 . ) We also stress that the allocation coefficients may differ strongly between growth conditions , as demonstrated by the growth laws mentioned above . For example , the fi’s of ribosomal proteins must be considerably larger in media that support a fast growth rate than in media with strong nutrient limitation , because the mean mass fraction of ribosomal proteins increases with the growth rate [30] . Here , however , we describe stochastic cell growth under fixed environmental conditions , so that the ( mean ) allocation of resources is well-defined and knowable in principle—for example through proteomics data . Fig 1 is an illustration of the modeling framework . Noise in the synthesis of a protein species induces fluctuations in its mass fraction ( Eq ( 1 ) ) . Through their effect on metabolism , these fluctuations propagate to the deterministic growth rate μd , which modulates the synthesis of all protein species ( Eq ( 3 ) ) . In parallel , all noise sources directly impact the growth rate μ ( Eq ( 5 ) ) and thus the dilution of all proteins ( Eq ( 1 ) ) . The transfer coefficients C i μ are reminiscent of the logarithmic gains defined in biochemical systems theory , which relate enzyme abundances to the metabolic flux in a given pathway [43] . It has previously been shown that these gains are relevant in the context of noise propagation [44] . Here , however , we consider the growth rate of the cell rather than the flux through a distinct pathway . In this section , we connect the transfer coefficients C i μ to the control of cellular growth and the field of Metabolic Control Analysis ( MCA ) [45 , 46] . In MCA , flux-control coefficients ( FCCs ) C i J are defined that quantify to what extent an enzyme concentration ϕi limits ( controls ) a metabolic flux J [45 , 46]: CiJ≔[ ϕiJ∂J∂ϕi ]ϕ0 . ( 8 ) In direct analogy to this definition of FCCs , the transfer coefficients of Eq ( 7 ) can be interpreted as growth-control coefficients ( GCCs ) that quantify each enzyme’s control of the growth rate . From Eq ( 4 ) a direct link between FCCs and GCCs can be derived ( see also [47] , p . 7 of S1 Text , and S1 Fig ) : C i μ = C i J - ϕ i . ( 9 ) The GCCs are specified by the sensitivity of the growth rate μd ( ϕ ) to changes in the proteome composition ϕ , evaluated in the steady-state mean , ϕ0 . Both the mean composition ϕ0 and the function μd clearly differ between growth conditions; therefore , the GCCs depend on the growth conditions as well . As mentioned , studies on the resource allocation of cells grown under different growth conditions have revealed striking empirical relations between the mean proteome composition and the mean cellular growth rate [26 , 28–30] . Even though these growth laws describe relations between growth rate and composition , they should not be confused with μd . The growth laws describe correlations between the mean composition and the mean growth rate under variation of the growth conditions , whereas μd describes the deterministic effect of the instantaneous composition on the instantaneous growth rate under a particular , fixed growth condition . There is no direct relation between the two . By extension , the growth laws do not directly translate into knowledge on the GCCs . Within the above framework , many statistical properties can be calculated analytically [5 , 42] . In particular , the noise level of the concentration of protein i , quantified by the coefficient of variation ηi , can be expressed as: η i 2 = ∑ j ≠ i ( 1 - ϕ 0 , i ) 2 ϕ 0 , i 2 Var ( N i ) μ 0 ( μ 0 + β i ) + ∑ j ≠ i Var ( N j ) μ 0 ( μ 0 + β j ) . ( 14 ) The derivation is provided in S1 Text , pp . 4–6 . Eq ( 14 ) shows that the coefficient of variation has two components: the first term results from the noise in the synthesis of the protein itself , the second from the noise in the synthesis of all other proteins . Each term is proportional to the variance of the corresponding noise source , but weighted by a factor that decreases with the mean growth rate μ0 and the reversion rate βi of that noise source . This analysis confirms that the inherent noise in the synthesis of one protein affects all other proteins . A fundamental distinction is commonly made between intrinsic and extrinsic noise in gene expression [44] . Intrinsic noise results from the inherently stochastic behavior of the molecular machinery involved in gene expression; extrinsic noise from fluctuations in the intra- and extracellular environment of this machinery . In this sense , the two terms in Eq ( 14 ) can be identified as intrinsic and extrinsic contributions . Complications arise , however , if the standard operational definition of these terms is applied [4 , 6] . This definition considers two identical reporter constructs R and G expressed in the same cell ( Fig 2A ) . Noise sources extrinsic to both reporters affect both reporters identically , inducing positively correlated fluctuations in the concentrations of the reporter proteins . Intrinsic noise sources instead produce independent fluctuations in each concentration . Extrinsic noise is therefore measured by the covariance between both expression levels; intrinsic noise by their expected squared difference . This operationalization , however , implicitly assumes that intrinsic noise does not propagate between the reporters . This assumption is violated in our model because the synthesis of reporter R directly contributes to the dilution of protein G ( Fig 2B ) . Consequently , the covariance between the expression levels has two contributions: Cov ( ϕR , ϕG ) ϕ0 , b2=−2 ( 1−ϕ0 , b ) ϕ0 , bVar ( Nb ) μ0 ( μ0+βb ) ⎵transmissionbetweenRandG+∑j≠R , GVar ( Nj ) μ0 ( μ0+βj ) ⎵othersources , ( 15 ) where the label “b” indicates quantities that are by definition identical for both expression systems . The second term on the right-hand side is positive and stems from noise sources that affect both reporters identically . The first term , however , is negative; it reflects the transmission of noise between reporters R and G . It would be misleading to identify Eq ( 15 ) as the extrinsic component of the noise—it is not even guaranteed to be positive . We conclude that the operational definition is not suitable when noise propagates between arbitrary genes . The circulation of noise in the cell can be studied by measuring cross-correlations between expression and growth rate in single-cell experiments [5] . Interpreting measured cross-correlations , however , is non-trivial . To dissect them , we now discuss a toy version of the model with just two protein species , X and Y . Despite its simplicity , it displays many features seen in more realistic models . Within the linear noise framework , ϕY–μ and πY–μ cross-correlations , respectively denoted RϕY μ ( τ ) and RπY μ ( τ ) , can be calculated analytically [42] . Up to a normalization , the results can be written as: RϕYμ ( τ ) ∝CYμSY ( τ ) ⎵Control+ϕ0 , YAY ( τ ) ⎵Autogenic−∑j=X , Yϕ0 , j[ CjμSj ( τ ) +ϕ0 , jAj ( τ ) ]⎵Dilution; ( 16 ) R π Y μ ( τ ) ∝C Y μ A Y ( − τ ) ⎵ Control +ϕ 0 , Y B Y ( τ ) ⎵ Autogenic +∑j = X , Y − C j μ [ C j μ S j ( τ ) + ϕ 0 , j A j ( τ ) ] ⎵ Transmission . ( 17 ) ( For a full derivation , not limited to the two-protein case , see S1 Text , pp . 5–6 . The two-protein case is discussed further in S1 Text , pp . 8–9 . ) These equations are plotted in Fig 3A and 3B ( see caption for parameters ) . As the equations show , the cross-correlation functions are linear combinations of three functions Si ( τ ) , Ai ( τ ) , and Bi ( τ ) , which are also illustrated in the figure . To aid interpretation , the cross-correlations can be decomposed into four noise modes , as indicated in Eqs ( 16 ) and ( 17 ) . The control mode ( Fig 3C ) reflects the control of enzyme Y on the growth rate . Noise NY in the synthesis of Y causes fluctuations in ϕY , which transfer to the growth rate in proportion with the GCC C Y μ . Because the effect of ϕY on μ is instantaneous , the contribution to the ϕY–μ cross-correlation is proportional to the symmetric function SY ( τ ) . In contrast , the effect of πY on μ involves a delay; hence the contribution to the πY–μ cross-correlation is proportional to the asymmetric function AY ( τ ) . In both cases , the amplitude scales with C Y μ . The autogenic mode ( Fig 3D ) is a consequence of Eq ( 2 ) . Because the growth rate matches the total rate of protein synthesis , noise in the synthesis of Y instantly affects the growth rate , resulting in a noise mode in the πY–μ cross-correlation that is proportional to the symmetric function BY ( τ ) . With a delay , this noise also affects ϕY , adding an asymmetric mode to the ϕY–μ cross-correlation . This mode does not depend on the control of Y; instead , its amplitude is proportional to the mean concentration ϕ0 , Y . The dilution mode ( Fig 3E ) pertains only to the ϕY–μ cross-correlation . It reflects that the growth rate of the cell is also the dilution rate of protein Y ( Eq ( 1 ) ) . With a delay , upward fluctuations in μ therefore cause downward fluctuations in ϕY . A subtle complication is that noise in the synthesis rate of both proteins reaches μ via two routes: through the immediate effect of πY on μ , and through the delayed effect of πY on ϕY , which in turn affects μ in proportion with C Y μ ( see in Eq ( 6 ) ) . Together , these routes result in a mode towards which each protein contributes both a symmetric and an asymmetric function . Lastly , the transmission mode ( Fig 3F ) is unique to the πY–μ cross-correlation . It reflects that all noise sources affect the cell’s composition ϕ and therefore μd; this in turn induces fluctuations in the synthesis rate πY . The noise sources again affect the growth rate via the two routes explained above , causing a symmetric and an asymmetric component to the πY–μ cross-correlation for each protein . The above analysis shows that , even in a highly simplified linear model , the cross-correlations are superpositions of several non-trivial contributions . The intuitions gained from this exercise will be used below when we present the results of a more complex model . In single E . coli cells , the cross-correlations between gene expression and growth rate have been measured by Kiviet et al . [5] . To test whether the above framework can reproduce their results , we constructed a model that includes 1021 protein species with realistic parameters , based on an experimental data set [51] . In the experiments , micro-colonies of cells were grown on lactulose ( a chemical analog of lactose ) and expression of the lac operon was monitored using a green fluorescent protein ( GFP ) reporter inserted in the operon . Because intrinsic fluctuations in GFP expression affect the cross-correlations directly as well as indirectly , through their impact on the growth rate and the expression of other genes , we modeled this reporter construct explicitly ( see Fig 4A , and S1 Text , pp . 9–11 ) . Specifically , the lac operon O was represented as a collection of three proteins Y , Z , and G ( for LacY , LacZ , and GFP ) affected by a shared noise source NO in addition to their private sources NY , NZ , and NG . The GCC of the operon as a whole is the sum of the GCCs of its genes . By varying the mean expression of the lac operon with a synthetic inducer , Kiviet et al measured cross-correlations in three growth states with different macroscopic growth rates: “slow” , “intermediate” , and “fast” [5] . Empirically , the macroscopic growth rate obeyed a Monod law [52] as a function of the mean lac expression . We therefore mimicked the three growth states by choosing their mean lac expression levels and growth rates according to three points on a Monod curve that approximates the empirical one ( Fig 4B , labels D , E , and F ) . Via Eq ( 7 ) , the same curve also is also used to estimate the GCC of the lac operon in each condition . Under “slow” growth conditions , the lac enzymes limit growth considerably ( large GCC ) ; under “fast” conditions , lac activity is almost saturated ( small GCC ) . To choose realistic parameter values for all other proteins , we used a published dataset of measured means and variances of E . coli protein abundances [51] . For each of the 1018 proteins in the dataset , the model included a protein with the exact same mean and variance ( see Fig 4C ) . This uniquely fixed the amplitudes of all noise sources . The GCCs of all proteins were randomly sampled from a probability distribution that obeyed the sum rule of Eq ( 11 ) . ( See Materials and methods , and S1 Text , p . 10–11 ) .
We have presented a model of stochastic cell growth in which the growth rate and the expression of all genes mutually affect each other . Systems in which all variables communicate to create interlocked feedback loops are generally hard to analyze . Analytical results were obtained by virtue of stark simplifying assumptions . Nevertheless , the predicted and measured cross-correlations have similar shapes and show similar trends under variation of the growth rate . That said , a few differences are observed . Chiefly , at slow and intermediate growth rates the model consistently underestimates the decorrelation timescales ( peak widths ) . In the model , the longest timescale is the doubling time; this timescale is exceeded in the experimental data . This suggests a positive feedback that is not included in the model , possibly as a result of gene regulation ( also see S2 Fig ) , or else a noise source with a very long auto-correlation time . Alongside their measurements , Kiviet et al . published their own linear noise model , which fits their data well . In fact , the shapes of the noise modes emerging in that model are mathematically identical to those presented above [42] . Yet , the models differ strongly in their setup and interpretation . Kiviet et al . model a single enzyme E that is produced and diluted by growth . It features only three noise sources: one directly affects the production of E ( “production noise” ) , one the growth rate μ ( “growth noise” ) , and one affects both simultaneously ( “common noise” ) . While these ingredients are sufficient to fit the data , the interpretation and molecular origins of the common and growth noise are left unspecified . In our model , which includes many proteins , similar noise modes emerge without explicit growth or common noise sources . Each enzyme perceives fluctuations in the expression of all genes as noise in the growth rate; this results in a dilution mode similar to that of Kiviet et al . Furthermore , noise in the synthesis of each enzyme instantaneously affects the growth rate ( Eq ( 2 ) ) due to the assumed homeostatic control of protein density . Hence , this noise behaves as a common noise source , which explains why the autogenic mode is mathematically identical to the common-noise mode of Kiviet et al . We conclude that noise in the expression of many enzymes , combined with homeostatic control of protein density , can contribute to the observed but unexplained common- and growth-noise modes . Control coefficients are routinely used in metabolic control analysis [45 , 46 , 54 , 55] and have also been studied in the context of evolutionary optimization [47 , 56] . In our linearized model , GCCs emerged as transfer coefficients , indicating that these quantities also affect the propagation of noise . Conversely , this suggests that GCCs could be inferred from noise-propagation measurements . For example , the Pearson correlation coefficient ( cross-correlation at zero delay ) between ϕi and μ might be used as an indication of control . However , we have seen in Fig 3 that the ϕi–μ correlation involves several noise modes that are independent of the GCC . As a result , the signs of the Pearson correlation and the GCC do not necessarily agree ( see Fig 5A ) . In addition , the intrinsic noise and GCC of the reporter protein can result in a negative cross-correlation even if the operon’s control is positive ( Fig 5B ) . Alternatively , the asymmetry of the control mode in the πi–μ cross-correlation could perhaps be exploited [5] ( S4 Fig ) . Unfortunately , this asymmetry is also affected by other modes , such as the transmission mode , which can mask the effect ( S4 Fig , panel C ) . We conclude that , in any case , such results have to be interpreted with great caution , ideally guided by a quantitative model . Future theoretical work should aim to relax assumptions and remove limitations . The assumed strict control of protein density can be relaxed by allowing density fluctuations . If these are long-lived , they will likely weaken the autogenic mode and introduce new modes of their own . Also , additional noise sources can be included that do not stem directly from protein synthesis . In particular , we ignored noise originating from cell division despite its importance [8 , 57 , 58] . In addition , gene regulation will affect some noise modes; this can be studied by allowing the fi to depend on ϕ . It will also be interesting to include non-protein components of the cell , such as RNAs . A further caveat is that the linear approximation used here is only reasonable if the noise is sufficiently weak . In fact , in the presence of strong non-linearities , the approach may even break down completely . For instance , it has been shown that cellular growth can be stochastically arrested when an enzyme whose product is toxic to the cell is expressed close to a threshold beyond which toxic metabolites build up to lethal doses [59] . In such circumstances , expression level noise in those enzymes can have a highly nonlinear effect on the cellular growth rate , resulting in subpopulations of growth-arrested cells [59] . That said , under more ordinary conditions linear models that describe noise in cellular networks have previously been used to great success [5 , 42] . Throughout this document we have considered noise sources that act on each production rate independently . Alternatively , one could hypothesize that the observed fluctuations in protein concentrations might instead originate from noise in the allocation of the flux—that is , from fluctuations in the allocation coefficients fi . This would be expected under the supposition that ribosomes are always fully occupied and translating at a constant , maximal rate , so that the relative rates of protein synthesis are determined solely by competition between different mRNAs based on their relative abundances and their translation initiation rates . Protein synthesis rates then become intrinsically correlated: an increase in the synthesis rate of one protein requires an simultaneous decrease in the synthesis rates of other proteins . In future work , such alternative models could be explored in detail . Preliminary simulations , however , show a striking symmetry in the ϕi–μ cross-correlation and a consistent asymmetry in the πi–μ cross-correlation ( for details see S1 Text pp . 12–13 , and S5 Fig ) . This can be understood as follows . If an increase in a particular synthesis rate is always compensated by a decrease in other production rates , the noise does not affect the sum of all production rates nor the growth rate instantaneously . Therefore , no autogenic mode should be present . Notably , in our model it is the autogenic mode that explains the asymmetry in the measured ϕi–μ cross-correlations as well as the dominant symmetric mode in the πi–μ cross-correlations under the fast growth condition . We conclude that noise on flux allocation alone cannot readily explain these experimental findings and additional noise sources would have to be included , such as the common noise as defined by Kiviet et al . [5] . Lastly , we hope that this work will inspire new experiments that can confirm or falsify the assumptions and results presented above . In particular , single-cell measurements of mass-density of protein-density fluctuations [60 , 61] could establish whether our assumption of density homeostasis is warranted . Also , additional single-cell measurements could determine whether expression noise indeed propagates between reporter proteins , adding to their covariance , and whether the amplitude of the various noise modes scales with the GCCs and mass fractions as predicted .
The Monod curve ( Fig 4B ) is given by μ0 = μmax ϕ0 , O/ ( ϕhalf+ ϕ0 , O ) , with μ0 the mean growth rate , ϕ0 , O the mass fraction of the lac-operon proteins , μmax = 0 . 8 h−1 , and ϕhalf = 0 . 005 . The three growth states correspond to three points on this curve , with ϕ0 , O/ϕhalf = {0 . 3 , 1 . 3 , 15}; this mass is shared equally among proteins Y , Z , and G . The mass fractions of the remaining proteins matched the proportions of the dataset [51] . The amplitudes of all noise sources were uniquely fixed by the constraints that ( i ) the CV of each Lac protein was 0 . 15 , ( ii ) the amplitude of NO was 1 . 5 times that of NG [4] , and ( iii ) all other CVs agreed with the dataset [51] . All noise reversion rates were set to β = 4μmax . To select the GCCs , we first randomly assigned proteins ( ≈ 25% of the total mass ) to the non-metabolic sector H . After the lac reporter construct was added , the GCC of each protein h ∈ H was set by Eq ( 12 ) . In each growth state , the GCC of the lac operon was calculated from the Monod curve , which yielded C O μ = { 0 . 77 , 0 . 43 , 0 . 063 } . Assuming GFP is non-metabolic and the GCCs of Y and Z are equal , we set C G μ = - ϕ 0 , G and C Y μ = C Z μ = ( C O μ - C G μ ) / 2 . The GCCs of all other proteins were sampled from a probability distribution that respects Eq ( 11 ) and assumes that proteins with a larger abundance tend to have a larger GCC ( see S1 Text , p . 11 ) . | Small as they are , bacterial cells are influenced by random fluctuations in their macromolecular copy numbers . Single-cell experiments have shown a complex interplay between this compositional “noise” and fluctuations in the cellular growth rate . While it is clear that this interplay originates from the tight interdependence of gene expression , metabolism , and growth , the underlying mechanisms are poorly understood . In this paper , we present a mathematical framework that describes compositional noise reverberating through the cell . We identify multiple routes by which noise in the expression of individual genes can propagate through the cell and demonstrate which factors affect each route . In doing so , we establish fundamental connections between the field of metabolic control analysis and the transmission of gene-expression noise . We then present a model tailored to Escherichia coli that includes >1000 genes with expression parameters set by previously measured protein abundances and show that it can reproduce the main features of measured cross-correlation functions between gene expression levels and growth rate . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"cell",
"physiology",
"protein",
"metabolism",
"engineering",
"and",
"technology",
"signal",
"processing",
"enzymology",
"operons",
"cell",
"metabolism",
"noise",
"reduction",
"protein",
"synthesis",
"enzyme",
"metabolism",
"dna",
"enzyme",
"chemistry",
"chemical",
"syn... | 2018 | Noise propagation in an integrated model of bacterial gene expression and growth |
In the developing mammalian visual system , spontaneous retinal ganglion cell ( RGC ) activity contributes to and drives several aspects of visual system organization . This spontaneous activity takes the form of spreading patches of synchronized bursting that slowly advance across portions of the retina . These patches are non-repeating and tile the retina in minutes . Several transmitter systems are known to be involved , but the basic mechanism underlying wave production is still not well-understood . We present a model for retinal waves that focuses on acetylcholine mediated waves but whose principles are adaptable to other developmental stages . Its assumptions are that a ) spontaneous depolarizations of amacrine cells drive wave activity; b ) amacrine cells are locally connected , and c ) cells receiving more input during their depolarization are subsequently less responsive and have longer periods between spontaneous depolarizations . The resulting model produces waves with non-repeating borders and randomly distributed initiation points . The wave generation mechanism appears to be chaotic and does not require neural noise to produce this wave behavior . Variations in parameter settings allow the model to produce waves that are similar in size , frequency , and velocity to those observed in several species . Our results suggest that retinal wave behavior results from activity-dependent refractory periods and that the average velocity of retinal waves depends on the duration a cell is excitatory: longer periods of excitation result in slower waves . In contrast to previous studies , we find that a single layer of cells is sufficient for wave generation . The principles described here are very general and may be adaptable to the description of spontaneous wave activity in other areas of the nervous system .
In the early stages of neural development , when initial sets of connections between neurons are being formed , neural activity helps organize and refine developing circuits . Before the onset of stimulus-driven activity , which helps refine neural organization in later developmental stages , neural circuits generate spontaneous patterns of activity which guide early development [1] . This spontaneous activity has been observed in many areas of the developing nervous system , including the auditory system [2 , 3] , neocortex [4] , hippocampus [5] , spinal cord networks [6 , 7] , brainstem nuclei [8] , and retina [9 , 10] . In the retina , spontaneous activity takes the form of coordinated bursts of spikes in neighboring retinal ganglion cells ( RGCs ) that slowly spread across the retina [10 , 11] . Retinal waves occur in a variety of species before visual experience , including cat , turtle , chick , mouse , and ferret [12] . They have non-repeating boundaries [13 , 14] , propagate with no directional bias , and can initiate at any retinal location [10 , 11 , 13 , 15] . The entire retina is covered in minutes [13 , 14 , 16] . Retinal waves drive activity-dependent organization in the visual system [1 , 11 , 12 , 17] . They have been shown to refine retinotopy in the LGN , superior colliculus , and cortex [18–25] , to drive segregation of the LGN into eye-specific layers [17 , 19 , 22] , and to drive responses in V1 neurons [26] . While the physiological mechanisms underlying retinal waves have been studied extensively [12 , 17 , 27] , there have been few attempts at modeling them . The first model was based on extracellular diffusion of potassium driving RGC activity [28] . Experimental evidence contradicted this premise [13] and another model was put forward , based on random amacrine cell activity and long refractory periods where amacrine cells are non-responsive [14] . Subsequent physiological evidence has shown these assumptions to be invalid , as amacrine cells regularly depolarize during waves and release excitatory transmitter when doing so [29 , 30] . Other limitations are that the model produces non-uniform net coverage of the retina [31] , that it has only been demonstrated to produce waves similar to postnatal day 2 ( P2 ) to P4 ferret , and that the properties of the generated waves , including wave size , frequency , and velocity , can be very sensitive to small changes in network state or parameters [32] . In this study we describe a retinal wave model that is robust to parameter variation and generalizes across species . We make use of the findings that amacrine cells receive input and depolarize during local wave activity [13 , 29 , 30] , that they have variable periods between spontaneous depolarizations [29 , 30] , and that the period between depolarizations appears to be a function of recent local excitation [30] . These observations lead to the central principle behind the model , that the refractory period , or the period between spontaneously occurring bursts in cells , is a function of the amount of excitatory input recently received by the cell . The resulting model produces waves with randomly distributed initiation points and non-repeating borders across a wide range of parameter settings . The velocity , domain size , and interwave interval ( IWI ) of the generated waves can be configured to match those seen in ferret , rabbit , mouse , turtle , and chick retinas . We show that a single homogenous group of cells can produce these wave behaviors , in contrast to claims that such behavior requires two independently functioning cell types [14] . We also show that the model exhibits chaotic behavior , producing seemingly random patterns of waves in the absence of stochastic input . The uniformity of retinal coverage provided by the model and the realistic spatio–temporal patterns of activity should also make it useful as an input to developmental models of the retino–geniculo–cortical pathway [31 , 33] .
The waves produced by the model are qualitatively comparable to published images of physiological waves [13 , 14 , 34 , 35] . Figure 1 shows two examples of wave activity and Video S1 shows 4 min of simulated wave behavior . Waves begin when several nearby amacrine cells are at or near the point of spontaneous depolarization and the excitation produced by the depolarization of some cause premature depolarization in others . If a sufficiently high density of depolarized amacrine cells is present , a wave develops . The wave continues to propagate in all directions where there is a high enough density of amacrine cells capable of depolarizing as a function of the excitatory input from their neighbors . The non-repeating wave behavior occurs because amacrine cells receive differing amounts of input during wave activity , resulting in some cells becoming more refractory than others . Figure 1C shows the threshold and excitation of a cell over time . Amacrine cells near the central regions of a wave receive more input during a wave , and hence become more refractory , than amacrine cells near the wave boundaries . This provides a deterministic and destabilizing force that inhibits the production of repeating wave domains as subsequent waves will more readily “invade” the border areas of a previous domain than central areas . Observations have shown that both the form of the original domain and the timing of these “invasions” determine the extent of the invasion and the subsequent increase in refractoriness for the amacrine cells involved . This turns the largely coherent refractoriness of amacrine cells in the original domain into several incoherent subgroups , inhibiting generation of a subsequent wave capable of following the boundaries of a predecessor . Figure 2 shows 40 sequential waves passing through a randomly selected spot on the retina , giving an example of the variability and non-repeating quality of the waves . The domain size and IWI distributions of simulated and physiologically recorded waves are shown in Figure 3 . Figure 3A shows the distribution of domain sizes and the averaged response of five P2–P4 ferret retinas ( data from [14] ) . The two distributions are very similar . Simulated domains average 0 . 156 ± 0 . 141 mm2 ( median 0 . 119 mm2; the average size of physiological domains was not reported ) . The IWI is defined as the period of time between successive waves passing a given location on the retina . Figure 3B shows plots of IWI distributions measured in the model retina and that are observed experimentally [14] . The model's IWI averaged 117 ± 47 s ( median 116 s ) which is similar to the experimentally measured value of 115 ± 48 s [13] . As with domain sizes , the model and physiological IWI statistics , and the shapes of the IWI distribution , were very similar . Experimentally reported velocities averaged 177 μm/s with a frequency of 3 waves/mm2/s [14] . Corresponding figures for the model were 176 μm/s and 3 . 0 waves/mm2/s . The IWI distribution measured by [13] was based on calcium imaging while other studies of similarly aged ferrets [11] used electrodes and reported notably shorter IWIs ( 90 versus 115 s ) . If RGC spiking , and hence wave activity , were to occur when amacrine cells were active but below the threshold of detection through calcium imaging , that would explain this discrepancy . Indeed , it has been estimated that RGCs which spike less than seven times in a burst may not be detected in calcium imaging studies [36] . To test this , we halved the wave detection threshold in the model . This reduced the average IWI to 86 ± 43 seconds , producing waves with an average velocity of 162 μm/s . This is comparable to electrode recordings of waves that showed an IWI of 90 s and a velocity of 100–300 μm/s [11] . These results suggest that wave activity and RGC spiking occurs in areas of the retina not detected by large-scale calcium imaging . Physiological studies have reported that waves have a random distribution of initiation points [14–16] . Figure 4A shows the distribution of wave initiation points from the model over a 60 min period . With the exception of the edges , initiation points are distributed uniformly across the retina , consistent with physiological studies [14 , 15] . Figure 4C shows a density plot of initiation points produced by the model after 120 h simulated time . Border effects are apparent near the retinal boundary , with each point on the boundary having two to three times the rate of wave initiation as a point in the central area . The model does not represent changing densities of amacrine cells in the peripheral and border areas of the retina , a simplification that might create altered patterns of initiation points compared to the physical retina . Physiological retinal wave studies have not yet addressed possible border effects on wave generation . One application of retinal wave models is for use in modeling development of the retino–geniculate pathway , such as done by Elliott and Shadbolt [31] . Their model required wave activity to be relatively uniform across the retina , as areas of high relative activity would achieve disproportionately large representation in the LGN . In their study they used the most accurate existing retinal wave model [14] and found that it had significant non-uniformities , including areas of high relative activity they termed “hot spots” . The duration of activity of RGCs in the model averaged 96 . 8 ± 25 . 4 s after a period of simulation . While individual cells have been reported to have different firing rates [11] , experimental studies have not reported observations of “hot spots” or other clear non-uniformities in the spatial variability of retinal activity . A test for the current model was to determine how evenly wave activity was distributed across the retina . We found that each location on the retina was active for an average of 95 . 8 ± 3 . 9 s over a 110 min simulation , with no spatial groupings of above or below average activity . There were no suggested acceptable limits for variability by [31] , but the reduction in the standard deviation from 26 . 2% of the mean to 4 . 1% over a nearly identical duration of activity should greatly improve the uniformity of retinotopic organization in such models and may be more representative of actual retinal behavior . Chaotic behavior is generally defined as behavior that is sensitive to small variations in initial conditions and that generates apparent randomness whose origins are entirely deterministic . To investigate this , we initialized the retina as described in Methods and then allowed the state of the model to evolve as a deterministic cellular automaton , where the state at any given time is a strict function of the immediately preceding state , using no added simulated noise . Specifically , P ( Equation 3 ) was constant . Analyses showed no significant differences between waves produced in simulations run this way compared to waves generated with ongoing noise . Table 1G shows the parameters used to reproduce P2–P4 ferret waves statistically similar to those measured by [13 , 14] ( Figure 5 ) . The IWI was 115 ± 46 s , average wave velocity was 183 m/s , and wave frequency was 3 . 0 waves/s/mm2 . This compares with the physiological IWI of 115 ± 48 s [13] , an average velocity of 177 μm/s , and wave frequency of 3 . 0 waves/s/mm2 [14] . The distribution of wave sizes was similar ( compare Figures 3A and 5A ) , with the model producing waves 0 . 16 ± . 16 mm2 in size . To investigate sensitivity to changes in initial conditions , we initialized the retina as above but changed the threshold value of a single amacrine cell by 0 . 1 , which is 4% of the average initial value . We then allowed the model to evolve for 90 min . We repeated this with ten different amacrine cells selected from around the simulated retina and then compared the behavior of a single amacrine ( the center cell ) across each of these simulations ( Figure 5E ) . The timing and intensity of bursting activity for the observed amacrine cell was different in all tested cases . More notable , however , was the collective change in behavior of all amacrine cells , which produced greatly altered wave patterns when viewing the entire simulated retina . Similar behavior occurs when the model is perturbed while it is running . To analyze sequential waves and determine if there were any cyclic patterns , we took snapshots of the retina when a wave passed an arbitrarily chosen point and compared sequences of successive waves ( Figure 2 ) . No patterns were apparent on the short time scales observed ( up to 3 h ) . To explore the possibility of very long cycle times , we ran several very long simulations ( 2 or 3 y simulated time ) and analyzed model output . To do this , we chose 19 amacrine cells uniformly distributed across the retina , and stored the activity pattern of these cells in a list . For each time step that three or more amacrine cells were active , we created a 19 bit integer , one bit for each cell , with a bit set to “1” if the cell was depolarized and “0” otherwise , and stored this value . After the simulation , we took the final 30 patterns and searched for this sequence in the list . For simulations run this way ( n = 3 ) , no match was found , indicating that any cycle behavior is longer than biologically relevant time scales . We used a small model retina for this search ( 0 . 65 mm2 , dT = 100 ms ) on the presumption that a small retina would show shorter cycle times than a large retina . Wave behavior was similar at the beginning and end of this interval both qualitatively and quantitatively ( spatiotemporal properties of waves varied by <5% ) . The activity-dependent refractory period is critical for the production of non-repeating and apparently random behavior . Analysis showed that the model produced non-repeating wave behavior so long as there was a differential in input received by cells during a wave , and hence a variable increase in their refractory periods . Extending the period where this input was summed ( i . e . , when the cells refractoriness increased regardless of whether or not it was active ) or by reducing the parameter H2 to very low values produced stable waves which covered the entire retina and eliminated realistic wave behavior . Such changes also eliminate realistic wave behavior when the model is run with stochastic input ( i . e . , randomly varying P ) . The observations that the model shows sensitivity to small changes in initial conditions and produces apparently random waves that are non-periodic and with randomly distributed initiation points suggest that it has a deterministically chaotic regime . However , we did not carry out strict mathematical tests for the presence of chaos ( calculation of Lyapunov exponents or a complexity analysis of cellular automata [43] ) , and therefore we do not claim to have demonstrated chaos in a mathematical sense . The model's behavior is consistent with “chaotic aperiodic behavior” as described for cellular automata [43] and is present for all time steps tested ( ranging from 5 ms to 200 ms ) . The biological relevance of this behavior is not that the model is chaotic in a mathematical sense , which would be interesting , but that it is chaotic in a practical sense . It uses a simple mechanism which does not rely on underlying stochasticity to reproduce the non-repeating and random waves that are observed in many species and are mediated by different chemical pathways [11] . The model is framed in a simple form in order to focus on general principles behind wave production . One particular simplification regards the duration an amacrine cell depolarizes , which is treated as constant but actually varies , with amacrine cells that depolarize in isolation doing so for relatively brief periods while depolarizations which are coincident with a wave are much longer [30] . To test whether this simplification affected wave production , we modified amacrine cells to have short fixed depolarizations and allowed them to remain depolarized , and hence excitatory , for as long as they had sufficient input from neighboring cells to do so . With this modification , wave behavior still appeared normal , with the model again being able to match the spatiotemporal properties of waves in different species . We targeted reproduction of P2–P4 ferret waves , producing waves with size = 0 . 16 ± 0 . 12 mm2 , velocity = 180 μm/s , IWI = 113 ± 52 , and frequency = 3 . 0 waves/mm2/s ( model parameters: H1 = 5 . 0 , H2 = 0 . 25 , P = 36 , and K = 0 . 3 ) . While waves appeared normal , amacrine cell behavior was notably different after this modification . Cells firing in isolation produced very brief bursts while amacrine cells that contributed to a passing wave depolarized with a duration and magnitude that varied by the cell's position in the wave ( Figure 6 ) .
The main strength of this model is its ability to reproduce the statistical properties of retinal waves seen in several species using only a small set of basic principles consistent with known physiology . Amacrine cells have been observed to spontaneously depolarize [30] , they regularly depolarize during wave activity [29 , 30 , 44] , and they release transmitter when depolarized [45 , 46] , even immediately after passage of a wave [30] , thus contributing to wave activity . The model is based on these observations and makes two additional assumptions: that depolarization is increasingly easy to achieve the more time that has elapsed since the previous depolarization , and that cells which receive more input have longer intervals between spontaneous depolarizations . These assumptions are supported by data in a recent study [30] which showed that amacrine cells have slowly decaying afterhyperpolarizing potentials ( AHPs ) , and they have very short refractory periods when pharmacologically isolated , longer refractory periods when they are spontaneously active while connected to neighbors , and even longer refractory periods when they depolarize during a passing wave . The model presented here differs significantly from previous models of retinal waves [14 , 28] . The most recent and related model [14] was based on the assumption of random depolarization of amacrine cells and it required a second layer of RGCs to filter sparse amacrine cell activity to produce wave-like behavior . The present model is based on deterministic activity-dependent refractory periods and produces spatially dense patterns of depolarized amacrine cells . The principle of activity-dependent refractory periods is very general and is not constrained by the properties of any particular neurotransmitter pathway or cell type . It produces waves with a large range of spatiotemporal properties and could underlie the production of waves at many different stages of development , in different species , and even in different brain area . A further difference is that only a single cell layer is required to produce waves , something that was previously not thought to be possible [14] . It should be stressed however , that while the principles we describe are very general , our model only addresses basic wave behavior that occurs in early development and ignores the emerging complexity of the retina as it matures . There are several ways to implement the basic principles of the model , and we explored some of the possibilities . As described above , amacrine cells were allowed to have both fixed and variable depolarization durations , and the model was run with and without stochastic input . Other strategies that we tested included: producing excitations at random points in the network , as might occur if amacrine cells , or other cells present later in development , were to depolarize spontaneously or to spontaneously release vesicles; using a layer of RGCs to filter amacrine cell activity , similar to [14]; varying the connectivity radius and the connectivity strength; and using continuous ( periodic ) boundary conditions . None of these variations resulted in significantly different behavior , suggesting that the underlying principles of the model are more important for wave generation than their particular implementation . In the model , the magnitude of the threshold regulates the refractory period , and this magnitude depends on recent input to the cell , with cells receiving more input during their periods of activity having higher thresholds . The rate of threshold decay was largely constant , resulting in longer refractory periods in cells that contributed to a wave compared to those that depolarized in isolation . Biologically , this refractory period results from a calcium-dependent potassium current [30] and possibly other factors , such as an activity-dependent variation in intracellular chloride , which has been proposed to drive spontaneous activity in the developing spinal cord [47] . The model does not differentiate between mechanisms contributing to the refractory period and only predicts characteristics of the resulting behavior . More physiologically detailed and species-specific models of the retina will be necessary for understanding the finer dynamics of retinal waves , and more experimental data will be required to adequately constrain such models . Given that calcium imaging appears to not detect all wave activity , and the spatial extent of electrode studies is limited , one experiment that would be very helpful would be simultaneous electrode and calcium imaging recordings over retinal areas sufficiently large to discern waves and their boundaries , as this would determine the frequency of very small patches of activity , how much activity is required before Ca2+ signal detection is possible , and how focused or extensive actual wave activity is in relation to the calcium imaging responses . The model produces output that should be useful in computational studies of the developing retino–geniculate pathway [31 , 33] , since the parameters can be adjusted to produce retinal waves with a wide range of size , velocity , and IWI . The output provides a relatively uniform net retinal coverage and it is simple to convert it to RGC spike trains by using amacrine cell activity as input to integrate and fire neurons . However , there are at least two significant ways in which this model does not conform to experimental observation . First , the duration of RGC bursting is weakly explained by the present model , as during a wave amacrine cell activity at a given location in the retina typically lasts 1–3 s . We have made no attempts to reproduce or account for the burst variability seen between species [12] , including the seconds-long oscillations of excitation observed in turtle retina [15] or the longer burst times observed in older ferret [11] . The behavior of the model suggests that additional factors are behind the long duration bursts seen physiologically , possibly involving input from additional cell types ( e . g . , bipolar cells ) and the use of metabotropic ion channels and/or additional neurotransmitters ( e . g . , GABA ) . Second , simple spiking patterns , as would be produced by integrate and fire neurons , are only seen during early development ( but see [48] for integrate and fire neurons which produce various burst patterns ) . As development proceeds , alpha , beta , and gamma RGCs develop distinct firing patterns [11] and ON and OFF RGCs begin to fire at different rates [49] . Computational studies using the output of this model as input to higher levels of the visual pathway will need to address these factors , as appropriate , according to the particular species and age being modeled . Wave behavior is stable across a wide range of parameter settings ( Figure 7 ) . Using Figure 7 as a guide , model parameters can be manipulated to produce waves quite different from those described here , including waves that slowly progress across all cells , or that produce small groups of excitation that propagate very little . Analysis of the effects of changing different parameters show that the duration an amacrine cell is excitatory ( parameter D ) is the most important factor in regulating the velocity of waves , particularly at slower velocities . When simulating mouse waves , it was necessary to increase D above 2 s to achieve velocities near the 110 um/s reported physiologically [34] . This suggests that the excitatory mechanism used in these mice involves either a slower excitation , such as would be produced by extracellular diffusion of transmitter , or a reduced rate of transmitter degradation , compared to what occurs in other species . Alternatively , mice may have extended durations of amacrine cell depolarization and/or periods of vesicle release . Wave velocity was similarly affected in simulations where amacrine cells were allowed to depolarize for variable durations . Slowing the onset of the AHP , thus prolonging depolarization , reduced average wave velocity . Increasing the speed of AHP onset increased wave velocity . A different situation exists in E14–16 chicks , where extraordinarily fast waves are observed ( . 5–1 . 5 mm/s ) [15 , 41] . Extremely short duration excitations can produce waves this fast , but the calcium imaging response from such brief depolarizations is greatly attenuated , often below the threshold of detectability . A more natural explanation for the increase in wave velocity is that the excitation time constant ( K ) approaches zero , causing the excitatory influence from one cell to be quickly realized in others . Physiologically , the adenosine/cAMP pathway may be related to the time constant . Adenosine has been shown to enhance transmitter release and to modulate neuronal excitability [50] , and both adenosine and cAMP have strong influences on wave velocity [45] . These results suggest that adenosine/cAMP might play a decreasing regulatory role in retinas with higher wave velocities , such as chick . The model makes several experimentally testable predictions . One is that wave behavior is the result of activity-dependent refractory periods in spontaneously active amacrine cells—normalizing threshold changes , such as by making AHP responses nearly uniform across cells , should eliminate non-repeating waves . Related to this , induction of an activity-dependent refractory period in cells which are recurrently connected and spontaneously active should produce non-repeating wave behavior . A second prediction is that wave velocity should be a function of the duration of excitatory influence of an amacrine cell . Manipulating this interval through genetic or pharmacological means should influence wave velocity . Third , wave behavior , as measured by RGC activity , should spatially extend beyond waves detected through large-scale calcium imaging , meaning that waves as defined by spike activity should be bigger than those seen by calcium imaging . The commonalities of retinal wave behavior across species and different anatomical and neuropharmacological pathways suggest an underlying mechanism that is robust and capable of being implemented in many ways . Our model displays this flexibility and has been framed to focus on mechanisms likely to be common among species and not to be constrained by specific physiological implementations or specific neural cell types . Hence the principles that we describe may be applicable to the description of activity in other parts of the brain such as auditory system , spinal cord , neocortex , and hippocampus which , like the retina , also exhibit patterns of coordinated spontaneous activity during development [2–12 , 51] .
In order to better compare amacrine cell behavior to the experimental results of calcium imaging studies , a rough approximation of a calcium response was produced and the spatial patterns of active ( i . e . , depolarized ) amacrine cells were measured . The retina was partitioned into pixels , one for each amacrine cell , with each pixel assigned a luminance value based on the activation level of all cells with dendrites passing through that point in the retina . Each pixel operated as a leaky integrator and had an intensity calculated according to where L is the pixel intensity at pixel i ( bounded on [0 , 1] ) , j is a set of all amacrine cells with dendrite overlapping i , and dT = 100 ms . All dendrites passing through a point on the retina contribute to the calcium response , with the soma generating a stronger response than the dendrite . Because of the short dendritic spread of RGCs [52] , the addition to the calcium signal due RGCs should have minimal effect on the spatial dynamics of the signal . This transformation is a coarse approximation and was not required for the model to produce wave behavior . It was primarily used to smooth wave progression and wave boundaries , assisting in automated wave tracking , and also to make model output resemble the experimental results more closely ( Figure 10 ) . The wave propagation images in Figures 1 , 2 , 5 , and 6 are based on simulated calcium imaging . A wave was detected when a ) the luminance of a pixel exceeded a threshold ( L ≥ 0 . 30 ) ; and b ) the pixel was not adjacent to any pixels that were assigned to a pre-existing wave . The pixel was considered to be part of a wave until its value fell below a lower threshold ( L < 0 . 25 ) . This threshold range was used to minimize pixels near threshold from repeatedly joining a wave when oscillating near threshold . The initialization point of this wave was the centroid of all connected pixels which exceeded the lower detection threshold ( 0 . 25 ) on the first frame the threshold was crossed . Wave velocity was calculated by measuring the distance between the centroid and the farthest point reached by the wave and dividing by the time required to reach that point . The velocity of each wave was stored , and the average of these values calculated . When two waves collided , both were omitted from the calculations as there was no longer an unambiguous starting or most distal point . Analysis did not demonstrate any significant difference between these joined waves and waves which remained independent , so their exclusion should not significantly bias the measurements . Wave size was calculated using the number of connected pixels that crossed threshold during the lifespan of the wave ( each pixel was counted only once ) . To reduce border effects in IWI and retinal coverage calculations , pixels associated with amacrine cells within one dendritic radius of the retinal boundary were omitted from the analysis . The IWI distributions were calculated by measuring the inter-wave interval of all analyzed pixels and storing these values in a histogram . The model is framed in the simplest form we found that produced robust wave behavior . One simplification involved the duration of amacrine cell depolarizations , which was constant in the model , although studies show that it varies , depending on whether the cell depolarizes in isolation or contributes to a wave [30] . To explore if our simplified amacrine cell behavior affected wave production , we modified amacrine cells by allowing them to depolarize for brief fixed intervals and remain depolarized as long as sufficient excitation was present . This was done by ( a ) slowing the onset of threshold increase , where the factors governing the refractory period , such as the AHP , take seconds to be fully realized , thus allowing prolonged depolarizations to occur , and ( b ) not resetting the excitation level ( Xi ) , allowing it to always reflect current input to the cell . As long as excitation was greater than the threshold , the amacrine cell was depolarized and excitatory to its neighbors . A 3 s refractory period was imposed after each fixed depolarization interval to prevent multiple triggerings during a single wave . In these simulations , D was set to 0 . 45 s and the maximum change of threshold ( ΔRi ) per second was limited to 4 . 0 . Larger threshold changes took more than 1 s to be fully realized . Other values for D and threshold onset rates were explored and produced similar results . | Neurons from the immature retina extend axons that make connections in the visual centers of the brain . Chemical markers provide guidance for these axons , but patterned neural activity is necessary to refine their connections . Much of this activity occurs in a distinctive pattern of waves before the retina is responsive to light , but it is not known how these waves are generated . In this study , we describe a simple mechanism that can explain the production of retinal waves . We use the knowledge that immature retinal cells are spontaneously active and show that waves will result if cells that receive more input when they are spontaneously active have longer intervals between activity . The resulting model reproduces experimentally observed waves in a variety of species , including ferret , chick , mouse , rabbit , and turtle , both at the level of individual cells and of the entire retina . The behavior appears intrinsically chaotic and the model is not tied to the properties of any particular biochemical pathway . We suggest that this mechanism could underlie not only the spontaneous patterns of activity that are generated in the retina but other areas of the developing brain as well . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"developmental",
"biology",
"retina",
"ophthalmology",
"ferret",
"computational",
"biology",
"neuroscience"
] | 2007 | Retinal Wave Behavior through Activity-Dependent Refractory Periods |
Apparent occupancy levels of proteins bound to DNA in vivo can now be routinely measured on a genomic scale . A challenge in relating these occupancy levels to assembly mechanisms that are defined with biochemically isolated components lies in the veracity of assumptions made regarding the in vivo system . Assumptions regarding behavior of molecules in vivo can neither be proven true nor false , and thus is necessarily subjective . Nevertheless , within those confines , connecting in vivo protein-DNA interaction observations with defined biochemical mechanisms is an important step towards fully defining and understanding assembly/disassembly mechanisms in vivo . To this end , we have developed a computational program PathCom that models in vivo protein-DNA occupancy data as biochemical mechanisms under the assumption that occupancy levels can be related to binding duration and explicitly defined assembly/disassembly reactions . We exemplify the process with the assembly of the general transcription factors ( TBP , TFIIB , TFIIE , TFIIF , TFIIH , and RNA polymerase II ) at the genes of the budding yeast Saccharomyces . Within the assumption inherent in the system our modeling suggests that TBP occupancy at promoters is rather transient compared to other general factors , despite the importance of TBP in nucleating assembly of the preinitiation complex . PathCom is suitable for modeling any assembly/disassembly pathway , given that all the proteins ( or species ) come together to form a complex .
Eukaryotic genes are thought to be regulated by hundreds of proteins that assemble into pre-initiation complexes ( PIC's ) at promoters using an ordered pathway . One aspect of the PIC assembly pathway involves the recruitment of the general transcription factors ( GTF's ) , such as TBP and TFIIB , by sequence-specific activators . TBP and TFIIB then contribute to the recruitment of RNA polymerase II ( pol II ) and other GTF's , which eventually start transcription . A fundamental question concerning our understanding of gene regulation is the extent to which each assembly and disassembly step is distinct at every gene in a genome . Is the traditional biochemical view that TBP “locks in” or commits to a promoter , and in a recurring manner nucleates PIC formation valid in vivo ? And is the PIC disassembly process in vivo , simply the reverse of the assembly process ? Parts of the assembly/disassembly pathway have been rigorously defined in vitro with a few purified proteins and DNA , and this has provided us with our current parsimonious view of PIC regulation [1] , [2] , [3] , [4] . In no case have assembly or disassembly reactions been reconstituted in a way that fully recapitulates the physiological setting ( presence of sequence-specific regulators , coactivators , specifically positioned nucleosomes , chromatin regulators , GTFs , etc ) at every gene , and so these questions remain open , in regards to the extent to which in vitro defined reactions mimic the in vivo events occurring throughout a genome . The answer to this question is not readily addressed in vivo , since reactions are not as definable nor quantifiable as in vitro biochemical reactions with purified components . Nonetheless , assays do exist for measuring relative levels of protein•DNA complex formation in vivo , and so mechanistic inferences will be sought . The goal here is to evaluate in vivo occupancy data in light of biochemical mechanisms that are intended to reflect the corresponding in vivo reaction . The extent of biological insight is predicated on rather subjective assessments of the assumptions associated with interpretation of in vivo data . Within the context of declared constraints and assumptions , we propose a means to model in vivo protein-DNA occupancy data , so as to better integrate and conceptualize massive genomic datasets . This study is focused on the means of such modeling and the assumptions inherent in the data , using specific examples of PIC assembly . Currently , perhaps the most widely used assay to measure the occupancy of proteins at genes in vivo is the chromatin immunoprecipitation assay ( ChIP ) . In ChIP , proteins are crosslinked to DNA , the protein is then purified , and the bound DNA identified either through directed PCR or through genome-wide detection platforms ( ChIP-chip and ChIP-seq ) . In this way , for example , the relative occupancy level of TBP , TFIIB , pol II , and many other proteins at every promoter in the genome in a population of cells can be assayed . Recent studies using differential ChIP and photobleaching experiments have provided compelling evidence for a dynamic state of PIC components in living cells [5] , [6] , [7] . Therefore , it is now within a conceptual framework to expect factors like RNA polymerase II , TBP , and other GTFs to undergo multiple assembly and disassembly cycles at promoters for each productive transcription event , rather than the traditional simple view that GTF's remain locked in place during multiple transcription cycles . The existence and origins of distinct occupancy levels of PIC components on genes has not been systematically explored , and thus is the impetus for conducting the modeling studies described here . Differential occupancy patterns for the GTFs have been described [8] , and may be caused by gene-specific regulators that influence the recruitment or retention of specific general transcription factors ( among other proteins ) , and thus assembly/disassembly mechanisms might differ from gene to gene ( or sets of genes ) . Here , we attempt to utilize ChIP-chip binding information gleaned at every promoter in the yeast genome to either plausibly infer or exclude PIC assembly/disassembly mechanisms . The major limitation in any such approach is that the number of permutations of possible assembly/disassembly mechanisms exceeds the amount of data available to constrain such mechanisms . In other words , occupancy data , alone , is insufficient to uniquely specify an ordered PIC assembly and disassembly pathway . Imposition of additional constraints ( or assumptions ) , such as predefining either the assembly ( or disassembly ) pathway , might however eliminate certain dissociation ( or association ) mechanisms as incompatible with the data , and thus serves the purpose of plausibly excluding mechanisms rather than uniquely identifying a mechanism . Here , we develop a ChIP modeling program , termed PathCom , in the context of a fixed PIC assembly pathway to infer allowable dissociation mechanisms . We validate the simulation using an existing chemical kinetics simulator COPASI [9] . Within the declared constraints , we discern the compatibility of different PIC disassembly mechanisms at nearly every transcriptionally-active gene in the yeast genome with existing ChIP-chip occupancy data .
The overall goal here is to inter-relate ChIP in vivo occupancy data with biochemical assembly/disassembly mechanisms , in a way that attempts to support or dispute such mechanisms . Such inter-relationships can be complex when one considers that hundreds of proteins are involved in transcriptional regulation . Therefore , we start by modeling only two factors ( the GTF's TBP and TFIIB ) , and increase complexity by adding more GTFs one at a time up to six factors . While we focus on PIC assembly/disassembly mechanisms on a genomic scale , any number of factors and combination of assembly/disassembly steps in gene regulation may be considered , given that all proteins ( or species ) come together to form a complex . TBP ( T ) binds to DNA ( D ) to form a protein-DNA ( TD ) complex , and in the presence of TFIIB ( B ) form a TDB ternary complex ( Figure 1A ) [10] , [11] , [12] . In the presence of sufficient levels of these proteins , their DNA occupancy level will vary from 0% to 100% as dictated by the context of each promoter . In principle , there are two pathways by which TBP and TFIIB assemble step-wise onto DNA ( Figure 1B ) [13]: A ) TBP binds to DNA , then TFIIB binds; or B ) TFIIB binds DNA first , then TBP . Their reversal constitutes two pathways for dissociation . The constant availability of energy to drive directional processes allows the pairing of any association and dissociation mechanism . Consequently , there are four paths by which an in vivo occupancy level is achieved for a two-component reaction . The availability of only two experimental constraints ( TBP and TFIIB occupancy levels on DNA ) is insufficient to specify the predominant association and dissociation pathways . In the absence of a necessary additional experimental constraint , we created a hypothetical constraint for the purposes of modeling , in which we eliminated all but one association pathway . That allowed us to evaluate the two possible dissociation pathways . The reciprocal modeling could also be done , by eliminating all but one dissociation mechanism . Since the purpose of this study is to demonstrate how the modeling works and to discuss its assumptions , caveats , and utility , we illustrate the process using a single association pathway that has good experimental support and model all possible dissociation pathways . Biochemical [1] and crystallographic [13] evidence shows that TBP binds DNA first , followed by TFIIB , which makes cooperative contacts with both TBP and the DNA ( Figure 1A ) . On this basis , we fixed assembly pathway “A” ( Figure 1B ) , which sufficiently constrains the system so that measured TBP and TFIIB occupancy levels can distinguish between the two dissociation pathways , “1” and “2” . In this context , dissociation pathway “1” allows either TBP or TFIIB occupancy to be greater than the other , but pathway 2 is only plausible if TBP occupancy is greater . Using published genomic datasets of TBP and TFIIB occupancy [14] , we modeled four groups of genes , each having either a high ( H ) or low ( L ) experimentally measured level of TBP and TFIIB ( Figure 1C , and Figure S1 ) . These occupancy levels were reproducible and verified by a second data source ( Affymetrix high density tiling arrays ) also present in the previous study ( Figure S2 ) [14] . We chose four subdivisions so as to separately consider different types of occupancy patterns . In principle , each gene could be treated independently . However , grouping of similarly behaving genes had the advantage of creating more robust occupancy values that are based upon hundreds of measurements , rather than just one . Aggregating the data dampened the variability caused by gene-specific differences in crosslinking efficiency and detection . It also served to identify predominant occupancy patterns that might reveal underlying themes in gene regulation . One limitation of such grouping is that it assumes a single underlying mechanism exists for an individual gene and for an entire group of genes , which may be unlikely in detail but reasonable for purposes of demonstration . To compare occupancy levels between proteins , it was necessary to place them on the same scale . We achieved this by scaling ChIP occupancy values ( fold over background ) for each factor from 0% to 100% . Our rationale , assumptions , and method for doing this are described in the Methods section . Figure 1D shows a cluster-plot of the genes with their TBP and TFIIB percent occupancies . Since the “ ( L , L ) ” group ( Figure 1F ) had low levels of both factors , TBP and TFIIB did not substantially occupy these genes . Consequently , modeling would not be informative for this group , and thus was not examined further . In addition , the “ ( H , L ) ” group comprised <1% of all genes , and so it too was not examined further . For the remaining two groups , TFIIB occupancy was greater than TBP occupancy . When assembly pathway A was fixed , in which TFIIB assembles last , then the observed higher level of TFIIB occupancy over TBP can only be accommodated by a situation where TFIIB dissociates last . Thus , for both groups ( ( L , H ) and ( H , H ) ) , the data reject dissociation pathway 2 ( TFIIB dissociates first ) and accept pathway 1 . These outcomes are illustrated in Figure 1D , by the black ( incompatible ) and green ( compatible ) squares . Note that when the alternative assembly pathway B is fixed , both dissociation pathways were compatible . This simple case illustrated how different starting assumptions ( assembly pathway A vs B ) resulted in a different set of compatibility outcomes . From this analysis , several insights were obtained: 1 ) Some occupancy levels simply do not distinguish among mechanisms . 2 ) In contrast to the simplified in vitro derived biochemical mechanism , TFIIB might remain at most promoters after TBP has dissociated ( although TFIIB may nevertheless be dynamic ) . How TFIIB does so is a matter of speculation that the data do not address . Based upon known TBP/TFIIB/DNA biochemical interactions , the notion that TFIIB might dissociate after TBP would seem untenable . However , the additional complexity that exists in vivo might accommodate such a mechanism if other proteins not explicitly defined in this model retained TFIIB at the promoter , after TBP had dissociated . TFIIB engages pol II at promoters via specific interactions [15] , [16] , [17] . Pol II tightly associates with DNA in an “open” promoter complex [18] , [19] , and tends to accumulate at the 5′ ends of genes [14] , [20] , [21] , [22] . If an active mechanism removes TBP , such as through the well-established ATP-dependent mechanism of Mot1 [23] , then TFIIB might remain on promoter DNA via pol II and in the absence of TBP . Towards our goal of modeling the assemblage of many proteins , we next consider a three-factor assemblage . The interaction of TFIIB with pol II ( P ) and TBP is structurally and biochemically well defined [13] , [15] . As in the two-step modeling , based upon biochemical precedent , we constrain the system to the following assembly pathway: TBP → TFIIB → pol II ( Figure 2A , black arrows ) . Since there are three factors , there are six possible dissociation pathways . Modeling three factors through six mechanisms for eight groups of genes became conceptually challenging to work through in the intuitive manner described for two factors . However , we determined that the plausibility of any mechanism could be evaluated by two basic rules: Rule 1: Does the mechanism make it unconditional that one protein's occupancy level must be greater than another ? For example , in the two factor mechanism , if TFIIB enters last and leaves first ( Figure 2B , left path ) , then such a mechanism requires that TFIIB occupancy be less than TBP occupancy . On the other hand , if TFIIB leaves last ( Figure 2B , right path ) , then such a mechanism allows both TBP and TFIIB to occupy the DNA independent of the other . This mechanism will therefore accommodate any occupancy levels observed for these proteins . Rule 2: Does the occupancy of one protein , other than the first and last proteins to assemble , have an occupancy level greater than the summed occupancy of any previously-associating protein and any subsequently-associating protein ? If so , does the mechanism give the possibility that the protein's occupancy is greater than the combined occupancies of these two other proteins ? This rule is applicable towards modeling of more than two factors . When this condition is met , then the protein must at some point occupy DNA without the other two proteins , and thus must be the last of the three to dissociate ( but not necessarily the last to dissociate overall if the mechanism has more than three proteins ) . When iterated over all factors in a mechanism , this rule determines the allowable orders of dissociation . For example , consider a fixed assembly order with TBP first , then TFIIB , then pol II ( Figure 2C ) : If TFIIB occupancy is greater than the sum of TBP and pol II occupancy , then only those dissociation mechanisms that have TFIIB dissociate last are compatible . If this condition is not true , then any dissociation mechanism can be accommodated by this rule , including the ones having TFIIB dissociate last ( but some might be disallowed in the context of rule 1 ) . These two rules , together , determine which dissociation mechanisms will be compatible with the data given an assumed association pathway . Note that depending on the actual percent occupancies , these rules will have varying effectiveness in narrowing down the dissociation mechanisms . If the rank order of observed occupancy is the same as the order of association , then all dissociation mechanisms will work . We transformed these queries into a program termed PathCom ( short for Pathway Compatibility ) , which was used to generate the compatibility chart in Figure 2D ( green = compatible , black = incompatible ) . This software is available in Protocol S1 and Protocol S2 for Windows and Mac users , respectively . Using the rationale from the two-step model , we generated eight groups of genes corresponding to either high or low occupancy of each of the factors ( Figure 2D ) . We sought to validate the approach taken by PathCom , to ensure that it reflected enzymological concepts for which this modeling attempts to emulate . Our validation employed COPASI , a freely available program that simulates biochemical kinetics [9] . Reaction mechanisms and concentrations ( the latter equivalent to the occupancy levels described here ) represent input parameters . For each mechanism and each group of genes , COPASI iteratively “guesses and checks” in an attempt to find a set of rate constants that delivers the observed occupancy levels for TBP , TFIIB , and pol II . It then reports a goodness-of-fit by measuring the square difference between the observed and the optimized occupancies , reporting this as an E-value ( see Methods ) . To maximize the parameter search space and avoid local minima , COPASI imposes some randomness in moving through the decision-making process . Since the system is under-constrained and randomness is involved , each repeated modeling run converges on a different solution for each mechanism ( i . e . , many different combinations of rate constant values can produce the observed occupancy levels , if a solution can be found ) . The values of the underlying rate constants generated by the Parameter Estimator in COPASI are not meaningful; rather the resulting E-value provides a quantitative measure of the suitability of a mechanism to fit the data . Re-running COPASI on the same dataset returns essentially the same E value ( not shown ) . Thus , COPASI provides a robust means of evaluating alternative mechanisms and validating PathCom . Figure 2D shows the compatibility findings of all eight possible clusters using three factors against the six possible dissociation mechanisms using PathCom . Figure 2E shows the corresponding log10 E-value assessments using COPASI . In all cases , the COPASI-reported E-values matched the Boolean decisions made by PathCom ( compare Figure 2D and E ) . Log10 E-values generated by COPASI were bimodal ( Figure 2E , bottom bar graph ) , providing a demarcation between compatible and incompatible outcomes . Thus , the simplified Boolean process associated with PathCom was validated by a kinetic mechanism simulator ( COPASI ) . Importantly , the analysis indicates that given a fixed association mechanism , there are a limited number of dissociation mechanisms ( green squares in Figure 2D ) that can account for the observed occupancy data . Fixing different association pathways generates different mechanism compatibility patterns ( Figure S3 ) . In Figure 2D , clusters of genes that had very few members ( e . g . , ( H , L , L ) and ( H , L , H ) ) , or had very low occupancy of all tested factors ( e . g . ( L , L , L ) ) may not be particularly robust , and thus less reliably interpreted . For the remaining clusters , one to two mechanisms were found to be compatible . A common theme was that TBP dissociated first , then pol II , and then TFIIB , which was consistent with the conclusions drawn from the two-factor assembly analysis described above . In principle , dissociation of pol II may proceed via removal into the bulk nucleoplasm and/or translocation down the DNA upon transcription , where ChIP occupancy would not be detected by microarray probes at the 5′ ends of genes . Consistent with the latter possibility , high transcription frequencies are observed at the ( H , H , L ) set , which has high TBP and TFIIB occupancy but relatively low occupancy of pol II ( Figure 2C ) . These genes are also enriched with pol II in the body of the gene ( not shown ) . The suggestion that TFIIB dissociates after both TBP and pol II dissociation is consistent with some reports in the literature [24] , and suggests that perhaps other factors retain TFIIB at promoters in the absence of TBP and pol II . TFIIB and TFIIF are known to interact with each other [25] , and potentially with activators [24] , [26] , [27] , [28] . We further examined the plausibility that TBP might not be fully bound at “high” occupancy promoters by looking at experimentally determined “digital footprints” of TBP bound at those promoters having the highest TBP occupancy ( Figure S4 ) [29] . Indeed , in all cases , no TBP footprint was detected over the TATA box , which is consistent with the notion that TBP does not fully occupy even its most highly occupied sites . Groups of genes that had very few members ( e . g . , ( H , L , L ) and ( H , L , H ) ) , or had very low occupancy of all tested factors ( e . g . ( L , L , L ) ) are expected to have higher variation , and thus less reliably interpreted . Therefore , these groups were not examined further . For the remaining groups , one to two mechanisms were found to be compatible . A common theme was that TBP dissociated first , then pol II , and then TFIIB , which was consistent with the conclusions drawn from the two-factor assembly analysis described above . As more factors were added to the modeling , and genes grouped according to low or high occupancy levels of each protein , the number of possible groups grew exponentially ( 2n , where is the number of modeled proteins ) . Consequently , membership in each group diminished , some to negligible levels . Those with negligible membership did not represent predominant patterns and may have arisen by chance as a consequence of noisy occupancy levels . Therefore , we combined groups of genes that lacked a viable membership level ( see Methods for membership criteria ) . Using the in vitro model for PIC assembly , we next added TFIIH ( H ) to the mechanism: TBP → TFIIB → pol II→ TFIIH . This mechanism is applicable even if pol II and TFIIH were entering together . As shown in Figure 3A , the groups with the highest membership of genes included those with low TBP occupancy levels , and either low or high levels of the other GTFs ( indicated by asterisks for gene groups that had at least two high occupancy GTFs ) . A group having high levels of all GTFs predominated among those groups having high TBP occupancy , denoted ( H , H , H , H ) . In the context of the modeled assembly pathway , these results suggest that TBP is removed from most measured genes before the other GTFs , except in cases where PIC assembly is maximal . The latter could be interpreted to reflect continuous reloading of TBP , which has recently been shown to be fairly dynamic [6] , [7] . Our modeling studies with PathCom suggest that the most plausible mechanisms for gene groups with abundant membership and at least two high abundance GTFs include early TBP dissociation ( Figure 3B ) . However for one abundant gene set ( L , H , L , H ) , the data are also compatible with an early dissociation of pol II followed by TBP ( or simultaneous with it ) ( Figure 3B , dissociation mechanisms 13 and 14 ) . In the four-factor mechanism , groups having a relatively large gene membership typically were limited to being compatible with only one or two of the 24 theoretically possible dissociation mechanisms ( Figure 3A , compatibility chart ) . Thus , the modeling of more factors increased the number of potential mechanisms in a factorial relationship ( n ! ) with the number ( n ) of proteins being modeled . However , the number of plausible mechanisms remained largely fixed at one to two , with a few exceptions . We next added TFIIF ( F ) ( Figure 4 ) and TFIIE ( E ) ( Figure 5 and 6 ) . While evidence suggests that TFIIF fits into the following fixed assembly pathway ( including simultaneous recruitment with pol II ) [3]: TBP → TFIIB → pol II → TFIIF → TFIIH [1] , [3] the literature reports seeming conflicting evidence for TFIIE entry [1] , [8] , [30] , and thus we chose to pursue to two alternative assembly mechanisms: TBP → TFIIB → pol II→ TFIIF → TFIIE → TFIIH ( Figure 5 ) and one where TFIIE enters prior to pol II ( Figure 6 ) . We focused on the few clusters that had the most members and had multiple factors with high occupancy ( indicated by asterisks ) . These included clusters with 687 , 580 , and 252 members ( Figs . 4 , 5 , and 6 ) . The membership for these particular clusters remained unchanged as more factors were included in the modeling because they failed to generate new gene groups that had sufficient membership to avoid consolidation . Thus , the occupancy data and the associated mechanisms displayed robust consistency as multiple GTF's were added on , which is consistent with them working together in a PIC . The occupancy levels in the five-factor modeling were compatible with mechanisms that had TBP and pol II dissociate early and TFIIB and TFIIF dissociating late ( Figure 4B ) . Interestingly , groups with few genes tended to have a larger number of compatible mechanisms ( more green boxes in Figure 4A ) . While the significance of this is unclear , it might reflect a cellular design that avoids ambiguity in the PIC disassembly pathway . That is multiple , alternative dissociation pathways may be problematic to control . In modeling six factors ( Figure 5 ) , the predominant compatible disassembly pathways for the two alternative assembly pathways retained the dissociation of TBP and pol II as early steps in all mechanisms . Whether we define TFIIE assembly as early ( upper panel ) or late ( lower panel ) , the occupancy data supported the following two predominant dissociation mechanisms: P→T→H→B→ ( E , F ) and T→P→ ( E , F , H ) →B , although when E associated early , the following pathway was also acceptable: T→P→ ( F , H ) → ( E , B ) . Spot checks of our results using COPASI confirmed our findings ( not shown ) .
Genome-wide occupancy data for the many hundreds of proteins involved in gene regulation is now accumulating . One major challenge has been to inter-relate such occupancy data and conceptualize it in light of models about how these proteins function together . Such models , as in the case of the assembly of the transcription machinery at promoters , are derived from biochemical experiments conducted on isolated components of the transcription machinery . The extent to which inferred biochemical mechanisms reflect in vivo processes is not known . We are not aware of any means of modeling genome-wide occupancy data to determine whether it is compatible with biochemical mechanisms . To this end , we developed the software tool PathCom . PathCom is generic in that it will determine whether any number of user-defined mechanisms is compatible with measured occupancy data of any number of relevant proteins . We applied PathCom to transcription complex assembly/disassembly , which has been extensively defined biochemically and for which genome-wide ChIP-chip occupancy data is available for . Biological insight gleaned from the modeling is subject to the veracity of the assumptions regarding what in vivo ChIP occupancy data actually measures , and the quality of the data being modeled . Eukaryotic protein coding genes utilize a common set of general transcription factors to assemble RNA polymerase II at promoters . A long-standing question that biochemistry has attempted to explain is the order of assembly of the transcription machinery and what happens to individual components during multiple transcription cycles . As far as the general transcription machinery is concerned , in vitro ordered assembly starts with TBP followed by TFIIB , then pol II and TFIIF , and then TFIIE and TFIIH [1] , [3] . In vivo ChIP occupancy data alone cannot discern whether such an assembly pathway is correct at any or all genes , and thus is a premise of the modeling example employed here . In the context of such a fixed assembly pathway , we explored different occupancy patterns of the general transcription machinery observed across the yeast genome , and interpret such occupancy patterns to potentially reflect alternative dissociation mechanisms . Should alternative association mechanisms be considered , then alternative dissociation mechanisms are likely . In regards to the genome-wide distribution of the GTF's , we did not see a random partitioning of genes into high vs low occupancy states for each factor . Principal component analysis ( PCA ) indicates the presence of a single major component ( not shown ) , and several minor ones . This would be consistent with the strong tendency of the GTF's to work together . What is interesting about the PCA is that TFIIB , pol II , TFIIF , and TFIIH were the main drivers in the first principal component , despite pol II having relatively low occupancy at the promoter region . TBP contributed the least to the principal components ( Dataset S1 and Figure S5 ) . In addition , we determined whether genes with <10% occupancy or ≥10% occupancy had a tendency toward having TATA versus TATA-less promoters , using data from [31] . We found that approximately 20% of genes with <10% or ≥10% occupancy levels were TATA-containing genes . Therefore , neither group had a bias toward TATA or TATA-less genes . Also we took the very highest TBP binding genes ( at least 50% binding ) and they also had 20% TATA-box genes . It does not seem likely that factor percent binding shows any correlation with the percent of genes that have TATA-boxes or sequence-effects in general . When clustering all GTF's and pol II , three high occupancy states stood out as having a large membership . These included genes with high levels of 1 ) all GTF's , 2 ) all GTF's except TBP , and 3 ) all GTF's except TBP and pol II . The group having high levels of all GTF's was by far the most highly transcribed , which is not surprising . This group included the ribosomal protein genes . However , for the major groups , low levels of TBP were more closely linked to low levels of transcription than the occupancy level of any of the other factors including pol II . This confirms on a genomic scale the earlier notion established on a few genes that TBP recruitment or retention is rate-limiting in transcription [32] . However , since pol II and the other GTF's are commonly found at high levels at many promoters even when TBP levels are low , it also seems likely that steps after TBP recruitment will be rate-limiting at certain genes . Otherwise , a rapid initiation and elongation phase would be expected to result in low pol II occupancy at all promoters . While the number of dissociation mechanisms scale factorially ( n ! ) with the number ( n ) of proteins involved , we did not see an equal distribution of genes into each type of mechanism , and we did not see a corresponding increase in the number of compatible dissociation mechanisms . Instead , the number of compatible mechanisms remained rather fixed at one to two , for a given association mechanism . The general pattern observed for most genes , was that if TBP , TFIIB , pol II , and the other GTFs assembled in the listed order , then the dissociation order was generally TBP , then pol II , then the other GTFs , with the latter being less resolved .
PathCom requires the user to enter occupancies of proteins in a tab-delimited text file followed by the name of the cluster line by line . In a header , before the occupancies are entered , users enter one-letter codes to denote protein identities ( of the user's choice ) followed by a number to indicate the order in which the proteins assemble ( See Text S1 for information how PathCom was designed and how it was intended to be used ) . Below each protein in the header , the user enters the percent occupancies calculated along with the name of each cluster ( or gene ) . After execution , the program then reads each cluster's occupancies on each line . Given the fixed order of association of proteins specified by the user in the header , the program generates all possible dissociation sequences . Note that if the user changes the association order , the pool of dissociation reactions will remain the same , but the numbering of each dissociation reaction will be different , because PathCom uses the specific association to generate the dissociation sequences . The program processes each dissociation sequence , pairing it with the fixed association sequence , and given the rules of compatibility ( discussed in the paper ) , computes whether the input protein occupancies are compatible with the mechanism ( association + dissociation ) it is testing . PathCom processes all possible dissociation sequences for all groups entered . PathCom writes the results to a tab-delimited text file . In this file , the horizontal axis is labeled with every mechanism identification number and the vertical axis is labeled with every cluster name . Also , PathCom writes a file that matches each dissociation sequence with its dissociation sequence identification number . Every time a set of occupancies and a mechanism are compatible , the program reports “−1” , and when they are not , the program reports “0 . ” Results can be clustered through Cluster then visualized graphically in Treeview [33] The code is given in Protocol S1 and S2 for users of Windows and Mac OS , respectively . COPASI conducts chemical kinetic and stochastic simulations [9] , and is freely available for download at www . copasi . org . Reactions were set to be irreversible for simplicity . Initial input protein and DNA concentrations were set to be equal , having an arbitrary value of 10 ( setting the DNA concentration to 1 gave the exact same results in terms of compatibility , Figure S6 ) . Since the observed occupancy levels for a factor represent the sum of all intermediate species having that factor , it was necessary to employ the Parameter Estimation function to optimize this sum , using the free protein concentration equal to ( 1 – Occ/100 ) ×10 , where “Occ” is the measured percent occupancy level , and had a practical lower limit of 0 . 1% ( this formula is only valid when all species concentrations were set to 10 ) . The Parameter Estimator may converge on a local minimum , which may not represent the optimal solution . Running the estimator multiple times alleviated the local minimum , since it employs a random search component . COPASI creates an objective value ( E ) used to measure goodness of fit between simulated and measured values:where “i” represents each of the protein factors involved in the modeling , “w” is the weight that is given to a particular protein in the optimization procedure , which is calculated automatically by COPASI , “x” is the measured occupancy , and “y” is the simulated occupancy . Since COPASI aims to minimize this sum of squares , lower E values ( more negative log10 E ) reflect better congruence between modeled and measured data . Since each modeling run has a manual component and becomes computationally draining with a large number of factors , it became impractical to run COPASI to fully generate the compatibility charts for four or more factors . Nonetheless , we employed COPASI to spot check these charts , and found 100% agreement with PathCom . | For proper cell function , cells need to precisely coordinate the expression of their genes on their DNA at precise times . In order to better understand how the cell works , it is important to understand how , when , and why a cell needs to turn on or off certain genes at certain times . In order to assist the cell to properly express its genes , there are hundreds of proteins that can bind and access DNA . Each protein has a unique function and these proteins assemble together into a very large complex to turn on genes . The assembly of these proteins has defined to some extent , however the whole process of assembly and disassembly of this complex in the cell is still poorly understood . In our modeling analysis , we have attempted to utilize genome-wide binding data to better understand how the transcription machinery that “reads” genes might disassemble , in light of what is known about the assembly process . This knowledge helps us better understand how cells coordinate their on/off-switching of their genes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"genetics",
"and",
"genomics/gene",
"expression",
"genetics",
"and",
"genomics/bioinformatics",
"genetics",
"and",
"genomics/functional",
"genomics"
] | 2010 | Genome-Wide Modeling of Transcription Preinitiation Complex Disassembly Mechanisms using ChIP-chip Data |
Neutrophils form the first line of host defense against bacterial pathogens . They are rapidly mobilized to sites of infection where they help marshal host defenses and remove bacteria by phagocytosis . While splenic neutrophils promote marginal zone B cell antibody production in response to administered T cell independent antigens , whether neutrophils shape humoral immunity in other lymphoid organs is controversial . Here we investigate the neutrophil influx following the local injection of Staphylococcus aureus adjacent to the inguinal lymph node and determine neutrophil impact on the lymph node humoral response . Using intravital microscopy we show that local immunization or infection recruits neutrophils from the blood to lymph nodes in waves . The second wave occurs temporally with neutrophils mobilized from the bone marrow . Within lymph nodes neutrophils infiltrate the medulla and interfollicular areas , but avoid crossing follicle borders . In vivo neutrophils form transient and long-lived interactions with B cells and plasma cells , and their depletion augments production of antigen-specific IgG and IgM in the lymph node . In vitro activated neutrophils establish synapse- and nanotube-like interactions with B cells and reduce B cell IgM production in a TGF- β1 dependent manner . Our data reveal that neutrophils mobilized from the bone marrow in response to a local bacterial challenge dampen the early humoral response in the lymph node .
Lymph nodes ( LNs ) are secondary lymphoid organs where pathogenic antigens are captured and processed , and antigen-specific ( adaptive ) responses are generated . T and B cells arrive to the LNs with the blood flow or via the afferent lymphatics , and occupy highly specialized compartments ( niches ) to differentiate into effector cells [1 , 2] . At the same time , LN residing innate cells shape these adaptive response directly by capturing antigens and either eliminating or presenting them , and indirectly by creating cytokine-rich surroundings [3] . Among the latter , neutrophils are the most dynamic cells mobilized to the LNs following infection or immunization [4 , 5] . While activated neutrophils are known for their capability to either support lymphocyte proliferation and activation [6] or suppress adaptive cell function [7] , the physiological roles of their influx to the LNs following vaccination or during the course of an infection remain only partially understood . Mature neutrophils express Ly6Ghi , CXCR2 , and CXCR4; and reside in the bone marrow ( BM ) niche retained by high concentration of SDF-1α [8] , and in the red pulp of the spleen [9] . During inflammation neutrophils are mobilized to the blood and migrate toward the source of CXC chemokines and other mediators released by affected cells or pathogens [10] to liquidate the source of danger [11] . Concurrently , they infiltrate adjacent lymphoid tissues to perform other highly specialized tasks , often linking innate and adaptive immunity [12] . In challenged LNs , neutrophils support cell-mediated responses during the differentiation of Th1 and Th17 cells , and development of efficient Th2 mediated response [13 , 14] . However , suppressive effect of neutrophils on T cell mediated response have also been shown [15 , 16] . Neutrophils augment antibody production by facilitating marginal zone B cell responses in spleen [17] , and can favor the transition from autoimmunity to lymphoma [18] . Conversely , depletion of neutrophils in mice immunized with protein antigens in adjuvants leads to elevated levels of serum antibodies [19] . The formation of a productive humoral response in LNs depends upon proper B cell trafficking and highly orchestrated intercellular interactions . After B cells exit high endothelial venules ( HEVs ) , they migrate through the medullary region ( MR ) and interfollicular zones ( IFZ ) to populate follicular areas near the subcapsular sinus ( SCS ) [20] . Follicular B cells exposed to cognate antigen migrate to the follicle border to acquire T cell help , and either proceed to the IFZ to differentiate into early antibody secreting cells or re-enter follicles to form germinal centers ( GCs ) . GC B cells clonally expand and differentiate into plasma cells ( PCs ) or memory B cells [21] . Terminal B cell differentiation is accompanied by increasing expression of the transcription factor BLIMP-1 [22] , and often takes place within the IFZ , and along the medullary cords . PCs predominately reside in the MR , or leave the LN to localize in splenic red-pulp or in specialized BM niches [23] . B cell proliferation and maturation can be boosted by cytokines like BAFF , APRIL and IL-6 released by innate cells [24] , or inhibited in T cell contact-depended manner [25] or by cytokines like TGF- β 1 [26] . Sites or niches where recruited neutrophils reside in LN and their regulatory effects on LN B cells are largely unknown . Staphylococcus aureus ( S . aureus ) is a potent human pathogen and the most common cause of skin and soft tissue infections in the USA . The host mobilizes both innate and adaptive immune responses to counter the infection . While neutrophils provide an initial line of defense arriving rapidly at the invasion site , the importance of humoral immunity in pathogen clearance is unresolved [27 , 28] . Some studies dispute its importance emphasizing the role of cellular immunity and in particular the importance of Th1 and Th17 cells [29] . Supporting this view B cell deficiency does not worsen the level of S . aureus bacteremia [30] . Yet multiple bacterial virulence factors specifically target humoral immunity [31] . For example , the humoral immune response is suppressed by S . aureus superantigens , which activate antimicrobial B cell populations triggering activation-induced cell death [32] and S . aureus protective antigens suppress B cell response [33] . LAC is a clone of methicillin-resistant S . aureus ( MRSA ) strain USA300 ( known as Los-Angeles County clone ) that compromises severely both innate and adaptive immunity of the host [34] . Detailed understanding the mechanisms of neutrophil and B cell responses to LAC is an urgent need in order to develop an effective anti-Staphylococcal vaccine strategy [35] . In this study we asked how the massive neutrophil recruitment that occurs during local S . aureus infection might impact the humoral immune response in the draining LN . We analyzed the mobilization of neutrophils to the inguinal LN ( iLN ) challenged with heat-inactivated or live S . aureus using intravital two-photon laser scanning microscopy ( TP-LSM ) . Our in vivo data indicate that the migration areas of mobilized neutrophils and activated B cells in the iLN often overlapped , while neutrophils and B cells established multiple intercellular interactions enriched with F-actin . The early humoral response to S . aureus in the iLN was significantly boosted after neutrophil depletion in vivo , and BLIMP1+ GC B cell numbers were elevated . Shown in vitro , activated neutrophils secreted TGF-β1 , which potently suppressed IgM production by iLN B cells . To specify the origin of neutrophils recruited to the iLN , we performed intravital microscopy of mouse calvarium and demonstrated neutrophil egress from the BM prior to their mobilization to the iLN . Our results suggest that the mobilization of bone marrow neutrophils to LNs following immunization or infection acts to limit the early humoral response .
A previous study had shown neutrophil recruitment to the iLN following the local injection of Complete Freunds’s adjuvant ( CFA ) [36] . CFA is composed of inactivated and dead M . tuberculosis emulsified in mineral oil . It is commonly used to enhance humoral immunity and is part of some induction schemes for triggering autoimmunity in mice . To provide a basis for comparison to S . aureus injected mice , we assessed local neutrophil response following subcutaneous CFA injection near the iLN ( S1A Fig ) . Analysis of cell mobilization kinetics indicated a peak of neutrophil recruitment approximately 4 h after CFA immunization both in the blood and in the iLN that subsided nearly to base line the following day ( Fig 1A ) . Ly6G+/CD11b+ cell population increased 10 fold in the blood ( S1B Fig ) and 8 fold in the iLNs ( S1C and S1D Fig ) . Both the percentage and overall number of B220+ cells also increased in the LN by 24 h after CFA injection , while CD4+ and CD8+ T cells numbers remained unchanged ( S1E Fig ) . Confocal microscopy was used to examine live LN sections from LysM-GFP mice ( MGI:2654931 , S1 Table ) , injected with CFA or PBS . LysM is highly expressed in neutrophils and at lower levels in other myeloid cells; therefore , neutrophils can be distinguished on the basis of their morphology and strong GFP expression [37] . Our analysis showed that GFPhi cells concentrated within the SCS , MR , T cell zone ( TZ ) , and IFZ in CFA immunized iLN at 4 h post-injection ( Fig 1B ) . In contrast , the PBS injected mouse had only a rare GFPhi cell ( Fig 1C ) . A comparison of GFP fluorescence intensities ( neutrophils ) indicated the presence of multiple cells in immunized and only few in control iLN , while B220 fluorescent intensities ( B cells ) were analogous at this time point ( Fig 1D ) . In vivo , GFPhi cells were mobilized to the iLN 2 h after CFA administration , rapidly increasing their numbers , thereafter ( Fig 1E and S1 Movie ) . Neutrophils arrived initially via the SCS and blood vessels; however , they entered iLN parenchyma predominantly by exiting blood vessels . Inside the capillaries , GFPhi cells displayed signs of early leukocyte diapedesis: rolling , adhesion , and arrest ( S1 Movie ) . ILN in PBS-injected control contained only a rare neutrophil after 2 h , no further infiltration was observed , and the microvasculature was free of neutrophils ( Fig 1F , arrows ) . Analysis of normalized mean GFP fluorescence within the HEVs confirmed abundant presence of GFPhi cells only in immunized iLN ( Fig 1G and 1H ) . To determine whether neutrophils could be recruited to B cell follicles , we induced laser damage within a follicle . Between 0 and 1 h neutrophils exited the HEVs near the follicle and migrated directly to injury site forming a swarm ( S2 Movie and S1F and S1G Fig ) . 1 . 5 h later neutrophils left the follicle , perhaps via chemorepulsion [38] , as many recently swarmed cells moved backwards partially clearing the area . These data show that mobilized neutrophils infiltrate the SCS , MR , and IFZ of the iLN . After 2 h of CFA challenge , neutrophils infiltrate the iLN parenchyma arriving from the blood microcirculation . Neutrophils avoid entering iLN follicles; however , a local injury can trigger their immediate entry . Next , we studied neutrophil influx to the iLN in response to a local injection of inactivated S . aureus Wood 46 strain . Analysis of recruitment kinetics was expanded to time points between 0 and 120 min , and at 2 , 3 , 6 , 12 and 24 h . The neutrophils increased in the blood 1 h post-injection , continued to increase reaching a plateau at 6 h , and returned to baseline by 12 h ( Fig 2A , left ) . Over the same interval we detected two waves of neutrophils infiltrating the iLN , 1st between 0 and 60 min with the peak at 30 min , and 2nd between 2 and 24 h with a plateau between 6 and 12 h . Their percentage had returned almost to baseline by 24 h ( Fig 2A , right ) . Epifluorescent microscopy of intact LysM-GFP iLN showed abundant presence of GFPhi cells in the SCS ( white dashed line ) and IFZ ( IF , white arrows ) of immunized iLN at 12 h ( Fig 2B ) . We also analyzed recruitment of neutrophils to distant LNs choosing the axillary and superficial cervical LNs and to the spleen at 4 and 12 h after immunization . Along with the massive influx of neutrophils to the iLN , we detected a significant recruitment to the spleen , but none to distant LNs ( S2A and S2B Fig ) . Shown by intravital TP-LSM of the iLN in immunized LysM-GFP mice , S . aureus bioparticles arrived with the lymph flow ( S3 Movie , Mobilization , red arrowheads ) and partially accumulated in the SCS ( Fig 2C , left panel , red arrowheads ) . The influx of neutrophils to the iLN gradually increased between 2 and 4 h ( Fig 2C , right panel; S3 Movie , Mobilization , green ) . Mobilized from the blood stream ( Fig 2C , blue arrowheads ) , they first infiltrated the MR and the IFZ , then between 2 and 3 h migrated from the parenchymal regions to the SCS , where they phagocytized bacterial particles ( Fig 2D , left panel , blue arrowheads ) . Between 3 and 6 h , neutrophils loaded with bioparticles moved back into the iLN parenchyma and swarmed in the MR and IFZ ( S3 Movie , Swarming , white square ) . The swarms usually accumulated bioparticles trapped by the neutrophils ( Fig 2D , right panel , while squares ) . By 12 h after bioparticle injection , while B cells migrated within the follicle and in the IFZ , many of neutrophils were recruited to the follicle border ( S4 Movie ) . At the follicle border , they formed associations with B cells , mostly in the perivascular regions ( Fig 2E , left ) . In more detail , after exiting HEVs neutrophils encountered B cells that migrated or oscillated along the outer vessel wall . When a B cell appeared in close proximity to a neutrophil , the cells often clustered ( Fig 2E , right ) . Remarkably , while most of neutrophils only transiently interacted with B cells , others formed persistent cell-cell contacts ( S5 Movie ) . In transient interactions , neutrophils usually formed protrusions , wrapped around lymphocytes ( S2C Fig ) , and then left , while B cells responded by attempting to follow the departing neutrophils . Such interactions usually resolved within 10 to 60 sec ( S5 Movie , arrows ) . The persistent interactions typically involved arrested neutrophils that formed tight intercellular contacts with B cells lasting 30 min and longer ( S2D Fig and S5 Movie , white circles ) . We also found formation of multiple cell-cell contacts between recruited neutrophils and CD4+ T cells migrating within TZ and reaching the T-B border ( Fig 2F ) . Quantitative analysis of cell-cell interactions showed both short and long-lasting interactions between neutrophils and B cells ( Fig 2G , upper chart ) , while all interactions of GFPhi cells with CD4+ cells were transient ( Fig 2G , lower chart; S6 Movie , arrows ) . Only GFPlo cells ( DCs ) formed long-term interactions with CD4+ T cells ( S6 Movie , blue circles ) . We also analyzed LN cell populations that predominantly engulfed S . aureus . Approximately 4% of the total LN cells were S . aureus positive at 12 h after injection ( S2E Fig ) and 1% after 24 h . As expected , more than 80% of the positive cells were neutrophils , while the other positive cells included CD169+ and CD169- macrophages along with CD11c+ DCs ( S2F Fig ) . As neutrophils provided the bulk of the clearance , we asked which cell type would clear S . aureus in their absence . For this , we depleted mice of neutrophils followed by bioparticle injection . While in isotype control mice 5–10% of the CD169+ macrophages contained bioparticles , in the depleted mice , this percentage increased to 25–35% ( Fig 2H ) . These results indicate that the local injection of S . aureus induces a rapid recruitment of neutrophils to the adjacent LNs and spleen . Mobilized LN neutrophils swarm and intense neutrophil phagocytosis ensues . Multiple neutrophil interactions with B cells and CD4+ T cells occur . In the absence of neutrophils , LN CD169+ macrophages more actively participate in bacterial particle clearance . Next , we studied neutrophil recruitment to the iLN after local S . aureus infection . LAC-GFP derivative of USA300 was used as a live S . aureus strain . Consistent with earlier observations in CFA and S . aureus bioparticle immunized mice , local LAC-GFP infection caused rapid and massive influx of neutrophils to the iLN ( Fig 3A and 3B ) . Analysis of mobilization kinetics , however , revealed more abundant ( total Ly6G+/CD11b+ cell number per iLN ) and continuous ( percentage over time ) neutrophil influx after the infection comparing to immunization ( Figs 2A , 3A , and S2A ) . While in infected mice the peak of recruitment was observed by 12 h after the infection , neutrophil numbers did not drop by 24 h ( Fig 3A ) . Neutrophil influx to the iLN following infection continued as their numbers were elevated until at least day 7 post-infection ( Fig 3B ) . To visualize early events of neutrophil recruitment to the iLN after local LAC-GFP infection , we performed TP-LSM using dsRed ( MGI:3663358 , S1 Table ) bone marrow chimeras . In these mice all hematopoietic cells express dsRed; however , in infected mice mobilized neutrophils were distinguished by their high intensity of fluorescence ( S3A and S3B Fig ) . Flow cytometry analysis of LAC-infected mice at 24 h after infection ( S3 Fig ) showed that neutrophils mobilized to the blood stream ( S3A Fig ) and to the iLN ( S3B Fig ) were dsRedhi , and B cells dsRedmed/lo ( S3B Fig , lower left panel ) . ILN of PBS injected dsRed bone marrow chimeric mice was free of dsRedhi cells ( S3B Fig , lower-right panel ) . Additionally , neutrophils in LAC-GFP infected iLN were identified due to their distinct morphology and behavior , i . e . size , dynamic migration , swarming and phagocytosis of the bacteria . Imaging neutrophil influx to the iLN between 2 and 12 h after LAC-GFP injection revealed their rapid mobilization from the blood vessels to the MR , TZ , IFZ and eventually to the SCS ( Fig 3C and S7 Movie ) . At indicated time points , Ly6G+/CD11b+ cell population in the iLN of infected dsRed mice increased from 0 . 3 to 3–5% ( Fig 3D ) . Abundant presence of neutrophils was also observed in the iLN at 24 h after the infection using TP-LSM ( S3C Fig ) . Mobilized neutrophils infiltrated the iLN parenchyma where they intensively swarmed and phagocytized LAC-GFP . Many of recruited neutrophils carried GFP+ bacteria while migrating and swarming in the iLN ( Fig 3E , white square; S8 Movie , white circles and cell tracks ) . While not as potent as the S . aureus , local instillation of sheep red blood cells ( SRBC ) near the iLN or injection of a standard protein antigen in alum ( NP-KLH ) also recruited neutrophils to the LN , while their recruitment was slower . Neutrophils were found in SRBC-immunized iLN by day 3 after immunization , and localized mostly in IFZ and at B cell follicle borders ( S3D Fig , arrowheads ) . We also examined uptake of LAC-GFP by SCS macrophages in infected iLN after neutrophil depletion ( Fig 3F and 3G ) . In isotype control mice the majority of LAC-GFP+ cells were neutrophils ( Fig 3F ) . Consistent with previously observed in S . aureus bioparticle-immunized mice , in LAC-GFP infected mice depleted of neutrophils , the CD169+ macrophage population that contained LAC-GFP ( CD169+ within GFP gate of live LN cells ) was increased to 75–80% ( Fig 3G ) . This percentage was elevated comparing to previously observed during immunization ( Fig 2H ) . This data shows rapid and continuous influx of neutrophils to the iLN adjacent to LAC-GFP infection site . While recruited neutrophils rapidly phagocytize the majority of LAC in the iLN , in absence of neutrophils SCS macrophages uptake the bacteria . To investigate neutrophil-B cell interactions observed in immunized mice , we imaged neutrophils and B cells from Lifeact-GFP mice ( MGI:4831036 , S1 Table ) . In these mice , filamentous actin ( F-actin ) can be visualized due to GFP expression during F-actin assembly [39] . Lifeact-GFP mice were immunized locally near the iLN with S . aureus bioparticle , and isolated Ly6Ghi cells and B220+/MHCII+ B cells were studied both in vitro and in vivo . The Ly6Ghi cells formed prominent cellular protrusions that contacted B220+/MHCII+ B cells , after both cell types adhered to ICAM-1/VCAM-1/KC coated plates ( Fig 4A ) . Live cell time-lapse confocal microscopy revealed that the intercellular contacts were enriched with F-actin ( Fig 4B ) , and B cell-neutrophil interactions induced rapid clustering of F-actin at the leading edge of neutrophils ( Fig 4C ) . When bioparticles were added to the co-cultures , Lifeact-GFP neutrophils rapidly acquired F-actin during bioparticle uptake ( S4A–S4C Fig and S9 Movie ) as well as during formation of cell-cell interactions ( S4D–S4F Fig and S9 Movie ) . However , the quantitative analysis of GFP mean fluorescence revealed that F-actin accumulated more rapidly in neutrophils engaging B cells than during bioparticle phagocytosis ( Figs 4D , S4C and S4F ) . B cells in return formed tight membrane associations with neutrophils ( Fig 4E , upper panel ) and fine membrane protrusions or nanotubes ( Fig 4E , lower panel ) . We also imaged F-actin enriched B cell-neutrophil intercellular contacts in vivo . Shown using TP-LSM in mice with adoptively transferred dsRed B cells and Lifeact-GFP neutrophils , F-actin formation initially occurred at neutrophil leading edge and later at cell-cell contact sites ( Fig 4F ) . The majority of observed interactions occurred when both cell types were arrested in perivascular space near the blood vessels ( Fig 4G ) . Quantification of F-actin assembly in Lifeact-GFP neutrophils measured as increase in GFP mean fluorescence showed increases during formation of intercellular contacts , equal or higher to that detected in the same neutrophils detaching from blood vessels post-diapedesis ( Fig 4H ) . These experiments show direct synapse-like and nanotube-like interaction between neutrophils and B cells in immunized mice . These intercellular contacts are enriched with F-actin that accumulates at a cell-cell contact area within seconds . The large influx of neutrophils and their observed interactions with B cells following local injection of S . aureus suggested that these interactions might influence the subsequent humoral response . To test this possibility we depleted neutrophils in vivo and measured antibody production by iLN B cells in mice immunized with S . aureus bioparticles or infected with LAC-GFP . The mice received an intraperitoneal injection of Ly6G-specific antibodies ( 1A8 ) or isotype control antibodies at day -1 , 0 and 1 of immunization/infection with S . aureus . 24 h after first 1A8 injection , neutrophils were mobilized to the blood and LNs in S . aureus immunized isotype control-injected , but not 1A8-injected mice ( S5A–S5C Fig ) . At day 5 , the iLNs in neutrophil-depleted mice were larger , and more heavily vascularized than in isotype control mice ( Fig 5A ) . Analysis of the kinetics of lymphocyte recruitment to S . aureus bioparticle-immunized iLN revealed an increase in B220+ cell population and decrease in CD4+ and CD8+ populations in neutrophil-depleted mice ( Fig 5B ) . B220+ cell numbers increased in neutrophil-depleted mice correlating with the total iLN cell numbers ( S5D Fig ) . We harvested the iLN B cells at days 5–6 post S . aureus injection , cultured them for 3 days and measured the levels of IgG and IgM in the supernatants . We compared amounts of antibodies produced by B cells derived from a single iLN ( S5D Fig ) . In the LNs from mice injected with S . aureus bioparticles , neutrophil depletion caused a 12-fold increase in total IgG and 30-fold increase in total IgM production ( Figs 5C and S5E ) . When quantified as amount of antibodies per B cell number , antibody production was also increased in B cell cultures derived from neutrophil-depleted mice ( S5F Fig ) . Total IgG levels were elevated in the serum of neutrophil-depleted mice starting at day 14 after immunization with S . aureus bioparticles ( Fig 5D ) . In the LNs harvested from LAC-GFP infected mice , neutrophil depletion resulted in over a 100-fold increases in both IgG and IgM production by LN B cells ( Fig 5E ) . Thus , the fold increase in antibody production after neutrophil depletion was higher in LAC-GFP infected mice than in the S . aureus bioparticle immunized mice ( Fig 5F ) . Using LAC or LAC spa lysates as antigens , we found that LAC-specific IgG and IgM responses were elevated in neutrophil-depleted mice ( Fig 5G ) . At day 5 after infection , LAC was found in the LNs of neutrophil depleted mice but not of isotype control-injected mice ( S3G Fig ) . To determine if neutrophil depletion also augmented LN B cells responses to protein antigens we isolated LN B cells 7 days after immunization and measured their secretion of IgG and IgM . In case of SRBCs we measured total IgG and IgM production and for the NP-KLH immunized mice we measured NP-KLH specific IgG and IgM produced by LN B cells . In both instances neutrophil depletion resulted in a higher production of antibody ( S5H and S5I Fig ) . To provide insight into the mechanism by which activated neutrophils suppress LN B cell antibody production we established an in vitro system . We isolated B cells from the iLNs of naïve mice and activated them with either LPS or S . aureus in the presence of absence of neutrophils . In the co-culture we chose a ratio of 10 B cells to 1 neutrophil as that is the approximate ratio of B cells to neutrophils in the immunized iLN . We relied on the ability of LPS or S . aureus to activate both B cell antibody production and to stimulate neutrophils . We found that both inductive signals increased IgM production in the B cell cultures . When neutrophils were present we observed a potent suppression of IgM production ( Figs 5H and S5J , left ) . At the same time , we did not observe such a pronounced reduction of IgA levels in LN B cell cultures in presence of S . aureus bioparticle-activated neutrophils ( S5J Fig , right ) . Seeded at the same cell density , by day 5 B cell numbers in B cell/neutrophil co-cultures were 1 . 5-fold lower than in pure B cell cultures ( S5K Fig ) . Thus , in presence of activated neutrophils , IgM production by total LN B cell cultures was 5-fold suppressed and IgA production 2-fold suppressed ( S5L Fig , left ) . When normalized for B cell number ( production by 1 x 106 B cells ) , IgM production was still 4-fold decreased , and IgA production only 35% decreased ( S5L Fig , right ) . Next , we tried to identify the inhibitor present in the activated neutrophil cultures . As TGF-β1 is known as a potent inhibitor of B cell antibody production [26 , 40] , we added a neutralizing TGF-β1 antibody to B cell-neutrophil co-culture . The suppressive effect of neutrophils was nearly completely reversed ( Fig 5I , left ) . In addition , supernatant from LPS-activated neutrophils ( SN act ) , but not from non-stimulated cells ( SN non-act ) , also suppressed IgM production , and this effect was reversed by adding a neutralizing TGF-β1 antibody ( Fig 5I , middle and right ) . We also verified that LPS-activated neutrophils secrete TGF-β1 , much more than non-activated neutrophils or LPS-activated B cells ( Fig 5J ) . These data indicate that neutrophils mobilized to antigen stimulated LNs can suppress B cell antibody production and suggest that this may occur via neutrophil TGF-β1 production . An increase in humoral immune response in neutrophil-depleted mice infected with live S . aureus is more pronounced than in those immunized with S . aureus bioparticles , SRBC or NP-KLH . To analyze the impact of neutrophil influx on generation of early PC population in the LN we utilized mice expressing a BLIMP1-YFP transgene ( MGI: 99655 , S1 Table ) . BLIMP1-YFP mice were injected with isotype control or 1A8 antibodies and immunized with S . aureus bioparticles near the iLN . Consistent with previous characterization of PC development in BLIMP1-YFP mice [41] , we identified YFPhi cells in the BM and the iLN of S . aureus immunized mice at day 7 , but not at day 3 after immunization . As shown using epifluorescent stereomicroscopy at day 8 after immunization , YPFhi cells localized mostly in the perivascular regions in the MR and IFZ in the iLN of isotype control-injected mice ( Fig 6A , white square ) . In the iLN of neutrophil-depleted mice , YPFhi cells were more abundant in the MR and IFZ , and more tightly packed around the blood vessels within these regions ( Fig 6B , white square ) . Furthermore , YFPmed cells found in B cell follicles were more numerous in the neutrophil-depleted mice ( Fig 6C ) . As shown by flow cytometry analysis the neutrophil-depleted mice had more B220+ cells ( S6A Fig ) and more GL7+Fas+ cells within B220+ gate per iLN than isotype control mice ( Fig 6D and 6E ) . 2–6% of the B220+GL7+Fas+ cells were also BLIMP1-YFP+ ( Fig 6F ) . These cells were enriched in the neutrophil-depleted mice and are likely the same cells observed in LN follicles using TP-LSM ( Fig 6C ) . A typical flow cytometry pattern of the B220+ gated cells analyzed for GL7 and FAS expression from control mice and depleted mice is shown ( Fig 6G ) . In vitro , LPS activated BLIMP1-YFP+ cells established cell-cell contacts with BM derived neutrophils by forming both tight interactions and membrane arms ( Fig 6H ) . Intercellular interactions between neutrophils and BLIMP1-YFP+ cells were also present in S . aureus immunized iLN with dsRed BM derived neutrophils adoptively transferred 12 h prior to imaging ( Fig 6I ) . Flow cytometry analysis of LNs in mice immunized with S . aureus bioparticles confirmed that population of Ly6G+/CD11b+ cells was increased at day 7 after immunization ( S6B and S6C Fig ) . More than 80% of this population represented endogenous neutrophils versus those adoptively transferred prior to imaging ( S6D Fig ) . Shown by TP-LSM in vivo , BLIMP1-YFP+ cells occupied distinctive perivascular niches , and neutrophils accumulated within the perimeter of these niches ( Fig 6J ) . While some of mobilized neutrophils formed short cell-cell contacts with BLIMP1-YFP+ cells along their migratory tracks , others were arrested inside the niches in clusters with BLIMP1-YFP+ cells ( S10 Movie ) . Imaging live sections of immunized BLIMP1-YFP+ iLN at day 7 after S . aureus bioparticle injection has shown similar localization of Ly6Ghi cells to that observed during initial neutrophil recruitment: in the MR , IFZ and TZ , often clustered around blood vessels ( S6E Fig , arrows ) . To analyze kinetics of neutrophil recruitment from the BM to the blood in response to a local immunization , we compared the recruitment rates after subcutaneous injection of S . aureus to those after intravenous KC/AMD3100 injections . Flow cytometry analysis of whole blood revealed a 4-fold increase of the GFPhi cell population in S . aureus and 10-fold increase in KC+AMD3100 injected mice 2 h after injection ( Fig 7A ) , while neutrophil numbers in the blood of PBS injected control remained at a baseline level . Importantly , neutrophil recruitment to KC+AMD3100 reached plateau 1 h after injection , while peak of neutrophil recruitment after S . aureus injection was observed between 3 and 4 h after injection ( Fig 7B ) . Furthermore , we found a 3-fold increase in neutrophil recruitment rate in mice injected with opsonized S . aureus bioparticles comparing to mice injected with non-opsonized bacteria ( Fig 7B ) . In order to demonstrate mobilization of neutrophils following immunization , we imaged mouse calvarium BM [42] in S . aureus bioparticle injected LysM-GFP mice . GFPhi cells appeared in the calvarium microvasculature 1 h after immunization . Between 2 and 4 h after immunization GFPhi cells accumulated in the vascular niche , rolling and adhering to the blood vessel wall of the capillaries , while PBS injected mice had only few such cells ( Fig 6C and S11 Movie ) . 3 h post immunization neutrophils started to egress from the BM niche to the blood stream via the central vein ( Fig 6D ) . Only minor neutrophil recruitment to the microvasculature and central vein was observed in PBS injected control mice after 3 h of imaging likely a consequence of the surgical procedure and imaging . Collectively , these data show that neutrophils are recruited from the BM niche to the vascular niche after subcutaneous injection of S . aureus . Shortly after , they are released to the circulation in a manner , strongly suggesting cellular uptake of S . aureus followed by chemoattractant release to the blood stream .
This study provided time-lapse analysis of neutrophil influx to the LN , assessed their role in the development of early humoral response after local S . aureus immunization and infection , and specified BM origin of LN neutrophils . We have used TP-LSM to show in vivo that neutrophils infiltrate the MR and IFZs in bacterial pathogen challenged LN , the areas where B cells migrate and differentiate into PCs . Within the LN , B cells and neutrophils exhibit short- and long-term interactions . The analysis of mice depleted of neutrophils and in vitro studies defined a suppressive role of activated neutrophils during the initial LN humoral response likely via their production of TGF-β1 . Neutrophil influx to adjacent LNs during local infection was more abundant and continuous than during immunization with S . aureus bioparticles or CFA . Since live replicating bacteria were expected to cause stronger innate response , we used 100 fold less CFU than inactivated bacterial particles . In either immunized or infected mice , neutrophils arrived to the LNs with the blood flow , crossed the HEVs and entered the LN parenchyma , migrating along the blood vessels [43] to invade the MR and IFZ . Shown in BLIMP1-YFP mice , neutrophil extravasation and migration occurred mostly within the niches filled with BLIMP1+ cells . When co-localized , neutrophils and B cells formed cell-cell interactions and multicellular complexes . The majority of interacting cells were localized along the follicle border , an important site for B cell-T cell interactions [44] . Neutrophils avoided crossing the follicle border and entering the follicle unless laser damage was triggered . Still unidentified molecular signals retain neutrophils outside the CXCL13-rich environment , yet these signals are clearly subordinate to those guiding neutrophils to an inflammatory site . Together with establishing cell-cell interactions , neutrophils actively cleared S . aureus in the LNs , and swarmed in the IFZ and at the inner surface of the SCS , a process often followed by accelerated neutrophil lysis and NETosis [45] . The effects myeloid cells have on B cell differentiation , maturation and antibody production may be beneficial due to their release of TNF family members BAFF and APRIL [17]; or alternatively , they may be suppressive by their production of prostaglandins [46] or via mechanism not completely understood [47] . We find that LN humoral response increases in neutrophil depleted mice immunized with inactivated S . aureus or infected with LAC-GFP . Using LAC lysates instead of LAC-GFP verified that IgG and IgM responses were not GFP-specific . As LAC spa is an isogenic protein A mutant , it was used to overcome the protein A-dependent non-specific interaction of S . aureus with the Fc part in antibodies [48] . Therefore , using LAC or LAC spa confirmed that in LAC-GFP infected mice detected IgG and IgM responses were S . aureus specific . Neutrophil influx to immunized LNs could suppress B cell responses in several ways . Most obvious is the removal of antigen [49] . Our data shows presence of LAC in the LNs at day 5 after infection in mice that were depleted of neutrophils during infection . Another possible mechanism may involve SCS macrophages . CD169+ macrophages uptake antigens delivered with lymph flow and present them to B cells [50] . Neutrophil swarming and microbial-caused death can lead to the loss of SCS macrophages; and depletion of granulocytes rescues the macrophage layer [4] . Neutrophil influx could damage the SCS macrophages and reduce antigen presentation to B cells resulting in less efficient B cell activation . Supporting this views , our data shows that SCS macrophages more actively participate in bacterial clearance in the LN lacking neutrophils . Finally , the recruitment of activated neutrophils into the LN may directly limit the expansion and/or differentiation of antigen stimulated B cells by producing suppressive cytokines . In this study we show that neutrophils from S . aureus immunized mice establish cellular protrusions to reach for B cells , and B cells display formation of nanotube-like structures , thus direct interactions may expose B cells to neutrophil-released cytokines . We also show that activated neutrophils secrete TGF-β1 , which can suppress antibody production by iLN B cells . In addition , the increased number of B cells in the immunized LN following neutrophil depletion is consistent with either a direct or indirect suppressive effect of neutrophils on humoral response . A recently published study used BLIMP1-YFP mice to show that myeloid cells shape the formation of the humoral response [47] . Diphtheria toxin-mediated ablation targeted at Ccr2-expressing myeloid cells resulted in an enhanced number of antibody secreting cells in the LN . In contrast , depleting Ly6G+ cells had little impact on the number of antibody secreting cells in the LN . Our study differs from this study in several ways . First , and most importantly , we depleted neutrophils prior to and during immunization while the other study [47] depleted myeloid cells on day 4 and 6 post immunization . Our data shows that abundant neutrophil influx to the LN occurs as early as 2 to 12 h after immunization or infection . The early arriving neutrophils are likely those that can suppress LN antibody responses . In addition , S . aureus may recruit more neutrophils into the lymph node and amplify the magnitude of the immune response . Mobilized Ly6Ghi cells abundantly co-localized and interacted with BLIMP+ cells in the IFZ , the site of the extra follicular antibody response , and later in the MR where PCs are localized . Besides measuring early IgG and IgM production in neutrophil depleted mice , we used the BLIMP1-YFP mice to assess the numbers of emerging antibody secreting cells . Depletion of neutrophils prior to immunization led to increased numbers of GC B cells and a subset of GC B cells expressing BLIMP1 . Together , our data suggest direct involvement of neutrophils in control of the humoral response in LNs . Since establishing intercellular contacts with innate cells is a key regulator of antigen-specific B cell differentiation [51] , the long-lasting interactions between neutrophils and BLIMP1+ cells we observed in vivo may represent immune synapse-like formations . Imaging calvarium BM revealed rapid mobilization of neutrophils in response to S . aureus . The mobilization of BM neutrophils to an inflammatory site via antagonistically regulated CXCR2/KC and CXCR4/SDF-1α chemokine axes is well documented [52 , 53] . However , mature neutrophils can also reside in spleen [17] and migrate between lymphoid compartments during inflammation [9] . Here we show in vivo that the egress from the BM precedes neutrophil influx to the iLN . Additionally , our studies indicate that BM derived Ly6G+ cells home to immunized iLN when injected intravenously . Our results demonstrate a BM origin for many of neutrophils recruited to the immunized iLN . Whether neutrophils can also be recruited from the spleen remains unclear . The more efficient neutrophil recruitment that occurred with opsonized bioparticles suggests that the cell-mediated uptake of bacteria by macrophages and perhaps other innate cells lining the SCS contributes to the BM neutrophil recruitment . While neutrophils are imperative as immediate innate defense against S . aureus , we show that their abundant recruitment to the LNs during local infection or immunization may reduce the efficient development of a specific humoral response . We speculate that the neutrophil influx reduces the exposure of pathogenic antigens to adaptive cells in the LN by phagocytizing bacteria and thus masking the antigen from SCS macrophages . In addition , activated neutrophils exhibit a clear direct suppressive effect on the differentiation of naive LN B cells to antibody secreting cells , which is likely mediated by the release of soluble factors like TGF-β1 and by the direct cell-cell interactions . Thus , while neutrophils may enhance splenic marginal zone B cell responses their recruitment to local LNs is detrimental for local antibody response . As most vaccines are delivered by subcutaneous injection our findings are relevant to immunogen design and the choice of vaccine adjuvants . Adjuvants are necessary to activate innate cells to achieve optimal antibody responses , but as our study indicates there is a potential downside as an overly exuberant neutrophil recruitment will impair antibody responses . Furthermore , directly targeting local TGF-β production would likely augment B cell antibody responses . Our study also raises the possibility that B cell-neutrophil interactions may impact neutrophil function . Future investigations should provide a better understanding of the mechanisms that link the innate and adaptive humoral responses against pathogens such as S . aureus .
All animals were bred and housed under pathogen-free conditions and used according to the guidelines of the Animal Care and Use Committee ( NIH ) . The LysM-GFP mice were provided by Dr . Hyeseon Cho and Dr . Ron Germain ( NIAID ) with permission from Dr . Thomas Graf ( Center for Genomic Regulation , Barcelona , Spain ) . The BLIMP1-YFP mice were provided by Dr . David Fooksman ( Skirball Institute of Biomolecular Medicine , New York , USA ) with permission from Dr . Dimitris Skokos ( The Rockefeller University , New York , USA ) . The C57BL/6 GFP-Lifeact mice were provided by Dr . Roland Wedlich-Soldne ( Martinsried , Germany ) . DsRed mice ( Jax Stock 006051 ) were received from Dr . Taha Bat and Dr . Cynthia Dunbar ( NHLBI ) . CD45 . 1 C57BL/6 mice were purchased from the Jackson Laboratory ( Bar Harbor , ME ) . DsRed and BLIMP1-YFP bone marrow chimeric mice were generated in animal facility within Comparative Medicine Branch ( NIH/NIAID ) as described in S1 Text . All targeted mouse genes are listed in S1 Table . All experiments were performed using sex and age matched animals , typically between 6 to 10 weeks old . S . aureus LAC clone of strain USA300 ( pulsed-field type USA300 ) was obtained from NARSA ( Network on Antimicrobial Resistance in Staphylococcus aureus ) . LAC-GFP clone was generated as a derivative of the USA300 LAC clone constitutively expressing genome-encoded GFP . For the integration of the gfp gene , whose DNA sequence was optimized for AT-rich Gram-positive bacteria ( gfpopt ) , on the chromosome of LAC clone , first overlap extension PCR was used to create blaZ-gfpopt . The constitutively active beta-lactamase promoter was amplified from S . aureus N315 genomic DNA with primers BlaZFw ( ATGCGGATCCCTAACAATAGAAATATAAAACAAAAGC ) and BlaZ-GfpRv ( AATTCTTCTCCTTTTGACATAATAAACCCTCCGATATTAC ) and gfp-opt was amplified from plasmid pSW4-GFPopt [54] with BlaZ-GFPFw ( GTAATATCGGAGGGTTTATTATGTCAAAAGGAGAAGAATT ) and GFPRv ( ATGCCTGCAGTTACTTATATAATTCATCCAT ) . A fusion product of the two PCR fragments was amplified with primers BlaZFw and GFPRv and cloned into plasmid pLL29 [55] BamHI and SalI restriction sites , resulting in plasmid ( pLL29-blaZ-gfpopt ) . Plasmid pLL29-blaZ-gfpopt was phage-transduced into LAC clone as described previously [55] . The integration of pLL29-blaZ-gfpopt into the ϕ11 attachment site of LAC clone was confirmed using primers scv4 ( ACCCAGTTTGTAATTCCAGGAG ) paired with scv10 ( TATACCTCGATGATGTGCATAC ) and primer scv8 ( GCACATAATTGCTCACAGCCA ) paired with scv9 ( GCTGATCTAACAATCCAATCCA ) . Expression of GFPOPT in USA300 LAC clone was confirmed by fluorescence microscopy ( excitation/emission at 470 ± 20 nm/ 515 nm , respectively ) . LAC S . aureus USA300 LAC spa ( an isogenic protein A mutant ) was a kind gift from Prof . A . Prince ( Columbia University , New York , NY ) . Glycerol stocks of S . aureus USA300-derivative LAC-GFP were grown to mid-exponential growth phase ( for min of 2 h ) in 50 ml of TSB at 37 °C with shaking at 180 rpm . Bacteria were harvested and washed and resuspended in sterile PBS prior to injections . CFU counts from infected LNs were performed as described [56] . Shortly , mice were euthanized and LNs were harvested . One LN of each mouse was placed into a 2-ml tube containing 1 ml of sterile PBS with 500 mg of 2 mm borosilicate glass beads ( Sigma ) . The LNs were homogenized in a Fast Prep bead beater ( Thermo Savant ) at 6 m/s for 20 s . The homogenates were diluted in PBS , plated onto TSB plates , and incubated overnight at 37°C for CFU counting . CFA containing heat killed M . tuberculosis ( H37Ra ) , 0 . 85 mL paraffin oil and 0 . 15 mL mannide monooleate ( Sigma Aldrich ) was injected in amount of 50 μl per iLN . S . aureus Bioparticles were purchased from Life Sciences ( Molecular Probes , Cat # . S2851 , S23371 , S23372 ) . The reagent consists of heat-killed bacteria Wood 46 strain without protein A , unlabeled or fluorescently labeled with either Alexa Fluor 488 or Alexa Fluor 594 . Alexa Fluor dyes of S . aureus Bioparticle conjugates are bound to the surface of bacterial cells , but not internalized . Prior to use , the Bioparticles were coated with S . aureus opsonizing reagent that contains rabbit polyclonal IgG antibodies specific for S . aureus and RIA-grade bovine serum albumin to block nonspecific binding ( Molecular Probes , Life Sciences , Cat # . S2860 ) . Opsonized S . aureus bioparticles were injected in amount of 2 x 107 bacterial particles per mouse ( 1 x 107 per iLN ) . NP-KLH ( Biosearch Technologies ) precipitated in Alum ( Thermo Scientific ) was given at amount of 25 μg per iLN . Sheep red blood cells ( SRBC , Lonza ) were administered in amount of 1 x 108 cells per iLN . S . aureus USA300-derivative LAC-GFP was injected at 1–3 x 105 CFU per mouse ( 0 . 5–1 . 5 x 105 CFU per iLN ) in 200 μL of sterile PBS . All antigens were injected subcutaneously within 1 cm of the adjacent iLN ( S1A Fig ) . Neutrophil influx to the iLN was monitored using TP-LSM or confocal microscopy between 0 and 24 h; and in the iLN , blood , draining LNs and spleen using flow cytometry between 0 to 24 h , or days 1 to 5 after immunization . Whole blood was collected from mouse tails ( tail snip ) , and iLN cells or splenocytes were isolated from euthanized mice . LysM-GFP cells were analyzed directly , and C57BL/6 cells were immunostained using fluorochrome conjugated anti-Ly6G/ ( clone RB6-8C5 ) , -B220 , -CD11c , -GL7 , -Fas , -CD4 , -CD8 , -CD169 , and—F4/80 antibody ( eBioscience ) . Flow cytometry was performed using FACS Canto II flow cytometer with FACS Diva 6 . 2 software ( BD Biosciences ) . Immune cell subsets were isolated using magnetic negative selection system with dynabeads M-280 streptavidin and magnetic particle concentrator ( Invitrogen ) as previously described [57] . BM derived neutrophils were isolated from mouse femur and tibia using anti B220 , -CD38 , -CD138 , -CD11c , -CD4 and-CD8; and B cells were isolated from mouse spleen by using anti CD11c , -Gr-1 , -CD4 and-CD8 biotinylated antibody ( BD Pharmingen ) . Isolated cells were cultured in complete lymphocyte medium ( DMEM supplemented with 10% FBS , 25 mM HEPES , 50 μM β-ME , 1% Pen/Strep/L-Glu and 1% Sodium Pyruvate ) in humidified CO2 incubator . Cells were allowed to recover for 30 min , and labeled with CellTracker™ Blue , CMAC , CellTracker™ Green , CFDMA or CellTracker™ Red , CMTPX ( Molecular Probes , Invitrogen ) according to the manufacturer’s protocol . Labeled cells were administered intravenously into recipient mice ( 5 x 106 cells per mouse ) via tail vein injection , and imaged between 16 and 24 h later . Isolated mouse LNs were sliced into 250 μm sections using Leica VT1000 S Vibrating Blade Microtome ( Leica Microsystems ) . Live cell imaging of immunostained sections was performed using Leica SP8 inverted 5 channel confocal microscope equipped with a motorized stage and 2 HyD ultra-sensitive detectors ( Leica Microsystems ) . Images of whole LNs were tiled using Leica Application Suite ( Leica Microsystems ) and processed using Imaris ( Bitplane ) software . For live cell imaging , BM-isolated neutrophils and LN-isolated B cells were cultured for 2 h on ICAM-1 + VCAM-1 + KC ( Recombinant Mouse ICAM-1/CD54 Fc Chimera , CF; Recombinant Mouse VCAM-1/CD106 Fc Chimera , Recombinant Mouse CXCL1/KC CF; R&D Systems ) coated glass-bottom dishes ( No 1 . 5 coverglass; MatTek ) . Live cells were stained in complete medium with fluorescently labeled anti-Ly6G and anti-B220 correspondingly ( BD Pharmingen ) . Confocal imaging was performed using Leica SP8 equipped with incubation chamber ( CO2 , 37°C ) for live cell imaging ( Pecon ) . Images were processed using Imaris ( Bitplane ) software . Detailed description of confocal microscopy setup is provided in supporting information ( S2 Text ) . Immunized mice were injected intravenously with 1% EB solution in PBS ( Evans blue dye , Sigma Aldrich ) at 1 ml/kg . Mice were euthanized , iLNs exposed on a skin flip , kept moisturized with PBS and imaged immediately after exposure . Fluorescent and bright field images of intact mouse iLNs were collected using motorized stereomicroscope Leica M205 ( Leica Microsystems ) equipped with 1x objective . GFP/YFP were excited at 488 nm and EB at 561 nm . Images were processed using Leica Application Suite ( Leica Microsystems ) and Imaris ( Bitplane ) software . All imaging experiments were performed at Biological Imaging Section ( NIH , NIAID ) using Leica SP5 inverted confocal microscope ( Leica Microsystems ) equipped with dual Mai Tai lasers as previously described [57] . Mouse surgery for imaging the iLN was performed according to the Cold Spring Harbor protocol [58] modified for the inverted microscope setup . For imaging neutrophil recruitments from the BM mice were injected intravenously with KC+AMD3100 ( AMD 3100 octahydrochloride; Recombinant Mouse CXCL1/KC CF; R&D Systems ) . Mouse calvarium BM was imaged as described [42] , using upright microscope setup and a custom-made stage with the head holder ( NIH Division of Scientific Equipment and Instrumentation Services ) . Post-acquisition image processing was performed using ImageJ ( National Institutes of Health ) , Imaris ( Bitplane ) and Huygens ( Scientific Volume Imaging ) software . Detailed description of the imaging technique is provided in supporting information ( S3 Text ) . Neutrophils were depleted in vivo as described [59] using anti Ly6G functional grade antibody 1A8 ( eBioscience/BioLegend ) . Briefly , animals were injected intraperitoneally with 100 μg of isotype control rat immunoglobulin G ( eBioscience/BioLegend ) or 1A8 antibody at days -1 , 0 and 1 of immunization . Efficiency of depletion was monitored by flow cytometry analysis or by TP-LSM , and typically represented > 90% in blood , spleen and draining LNs . At days 5 to 7 of immunization ( 6 to 8 of depletion ) mice were sacrificed and the iLNs harvested . B cells isolated from a single iLN were cultured in complete lymphocyte medium for 72 h , and the supernatants were collected . Antibody concentration in the supernatants was measured with commercial ELISA kits ( Mouse IgG total ELISA Ready-SET-Go , Mouse IgM total ELISA Ready-SET-Go; eBioscience ) according to the manufacturer's protocol . LAC-specific IgG and IgM were measured using plates coated with bacterial lysates . Total lysates from LAC and LAC spa were prepared as previously described [60] with the following modifications . Bacteria were grown as above , pellets were resuspended in 1 ml of sterile PBS and incubated 30 minutes at 37 °C in the presence of Halt protease inhibitor single use cocktail ( Thermo Scientific ) and lysostaphin . The digested lysates were transferred to 2-ml Lysing Matrix B vials ( MPbio ) and homogenized in a Fast Prep bead beater ( Thermo Savant ) at 6 m/s for 20 s . Protein concentrations were determined with the Quant-iT assay kit ( Life Technologies ) . IgG and IgM specific for NP-KLH were measured by ELISA using plates coated with NP-KLH . For antigen-specific ELISAs plates were coated at protein concentrations 10 μg/mL ( LAC or LAC spa lysates ) and 1 μg/mL ( NP-KLH ) in PBS overnight and blocked with 1% BSA in PBS . Neutrophils were isolated from the BM and cultured in for 24 h in presence of 2 μg/mL LPS ( from E . coli , Serotype R515 ( Re ) , TLR grade , ENZO Life Sciences ) or 1 x 106 particles/mL S . aureus bioparticles . Supernatants were added to freshly isolated LN B cells . Isolated iLN B cells were cultured in complete lymphocyte medium at initial concentration between 5 x 105 to 2 x 106 cells/mL . B cells and neutrophils were added to the cultures in ratio 10 B cells: 1 neutrophil , for 5 days in presence of LPS or S . aureus bioparticles . Alternatively , supernatants from either activated or non-activated neutrophil cultures ( 25% of culture medium ) were added to B cells . Final concentration of LPS was 2 μg/mL , and S . aureus 1 x 106 particles/mL in all cultures . Neutralizing anti TGF-β1 ( TGF-beta 1/1 . 2 Polyclonal antibody; R&D Systems ) were added at final concentration 1 μg/mL , and TGF-β1 ( 50 ng/mL ) . TGF-β1 in neutrophil and B cell cultures was measure using commercial ELISA kit ( eBioscience ) . IgM and IgA levels in the supernatants were measured with commercial ELISA kits ( eBioscience ) according to the manufacturer's protocol . The animal experiments and protocols were performed according to the regulations of NIAID Division of Intramural Research Animal Care and Use Committee ( DIR ACUC ) . Animal Study Proposal LIR 16 entitled “Analysis of Innate Immune Function in Mice” that covers this work was approved by the NIAID DIR ACUC on Dec 1st , 2011 as the initiation date , and has been reviewed annually . The NIAID DIR ACUC as a part of the NIAID DIR Animal Care and Use Program , as part of the NIH Intramural Research Program ( IRP ) , complies with all applicable provisions of the Animal Welfare Act and other Federal statutes and regulations relating to animals; and is guided by the "U . S . Government Principles for the Utilization and Care of Vertebrate Animals Used in Testing , Research , and Training" . The policies , procedures and guidelines for the NIAID IRP are explicitly detailed in NIH Policy Manual 3040–2 , “Animal Care and Use in the Intramural Program” ( PM 3040–2 ) and the NIH Animal Research Advisory Committee Guidelines ( ARAC Guidelines ) that are posted on the NIH Office of Animal Care and Use public website at: http://oacu . od . nih . gov . The statistical significance was evaluated by subjecting the data to a Student's t-test using GraphPad software . Values are presented as means ± SD or means ± SEM as indicated . * , P<0 . 05; ** , P<0 . 01; *** , P<0 . 001 . | Highly antibiotic resistant Staphylococcus aureus ( S . aureus ) are an important human pathogen and major cause of hospital acquired infections . An early host defense mechanism against bacterial infection is neutrophil recruitment , which helps eliminate the bacteria at the site of invasion . However , unless quickly neutralized , pathogens such as S . aureus can gain access to nearby lymph nodes via draining lymphatics . Lymph nodes protect the host by mobilizing additional resources that limit further pathogen dissemination . These include recruitment of neutrophils to the lymph node to directly target pathogens and the initiation of adaptive immune mechanisms , such as the humoral immune response , which transforms B lymphocytes capable of making pathogen specific antibodies into antibody producing plasma cells . Using a mouse model that allows direct visualization of lymphocytes , neutrophils , and fluorescently-labeled S . aureus in lymph nodes , we document the rapid appearance of bacteria in the lymph node following local S . aureus infection . We characterize the dynamic influx of neutrophils that occurs as a consequence and reveal direct B cell-neutrophil interactions within the lymph node parenchyma . We find that while lymph node neutrophils rapidly engage bacteria , they limit the subsequent humoral immune response likely by producing Transforming Growth Factor-β1 , a factor known to limit B cell responses . These finding have important implication for our understanding of B cell responses against potent pathogens such as S . aureus and for the design of effective vaccines . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Neutrophil Recruitment to Lymph Nodes Limits Local Humoral Response to Staphylococcus aureus |
Quantitative viral outgrowth assays ( QVOA ) use limiting dilutions of CD4+ T cells to measure the size of the latent HIV-1 reservoir , a major obstacle to curing HIV-1 . Efforts to reduce the reservoir require assays that can reliably quantify its size in blood and tissues . Although QVOA is regarded as a “gold standard” for reservoir measurement , little is known about its accuracy and precision or about how cell storage conditions or laboratory-specific practices affect results . Owing to this lack of knowledge , confidence intervals around reservoir size estimates—as well as judgments of the ability of therapeutic interventions to alter the size of the replication-competent but transcriptionally inactive latent reservoir—rely on theoretical statistical assumptions about dilution assays . To address this gap , we have carried out a Bayesian statistical analysis of QVOA reliability on 75 split samples of peripheral blood mononuclear cells ( PBMC ) from 5 antiretroviral therapy ( ART ) -suppressed participants , measured using four different QVOAs at separate labs , estimating assay precision and the effect of frozen cell storage on estimated reservoir size . We found that typical assay results are expected to differ from the true value by a factor of 1 . 6 to 1 . 9 up or down . Systematic assay differences comprised a 24-fold range between the assays with highest and lowest scales , likely reflecting differences in viral outgrowth readout and input cell stimulation protocols . We also found that controlled-rate freezing and storage of samples did not cause substantial differences in QVOA compared to use of fresh cells ( 95% probability of < 2-fold change ) , supporting continued use of frozen storage to allow transport and batched analysis of samples . Finally , we simulated an early-phase clinical trial to demonstrate that batched analysis of pre- and post-therapy samples may increase power to detect a three-fold reservoir reduction by 15 to 24 percentage points .
The latent HIV-1 reservoir that persists following treatment with suppressive ART exists primarily in resting CD4+ T cells and is an obstacle to eradicating HIV-1 [1–4] . There are substantial ongoing efforts to eliminate or reduce the size of this reservoir [5–7] . Evaluating such efforts requires assays that can reliably quantify its size in blood and tissues in order to monitor its changes during curative intervention strategies . Replication-competent HIV-1 can be measured by QVOA . These terminal dilution assays place known numbers of resting CD4+ T cells in culture wells , usually in serial dilutions of cells that cover several orders of magnitude , with replicate wells at each dilution . The CD4+ T cells are activated before co-culture with cells that are highly permissive for primary strains of HIV . The propagation of HIV replication is detected by an assay for either p24 antigen or HIV RNA in the supernatant of these co-cultures over a two- to three-week period [8] . Each well is read as negative or positive , which means that replication-competent virus was present in at least one of the cells in the well . The number of infectious units per million cells ( IUPM ) is then estimated by maximum likelihood assuming single-hit Poisson dynamics [9] . This approach has represented the “gold standard , ” because it measures replication-competent virus in latently infected cells , which is crucial because the majority of integrated HIV-1 DNA is replication-defective [10–12] . Use of QVOA presents both practical and statistical challenges , many of which are attributable to the rarity of the target entities: often on the order of only one CD4+ T cell in a million is positive for replication-competent virus by a single-round QVOA . These challenges include: 1 ) QVOA requires that large volumes of blood be collected to generate large numbers of Ficoll-purified peripheral blood mononuclear cells ( PBMC ) , which are generally further processed without freezing/thawing , into input resting CD4+ T cells , 2 ) each assay takes weeks to complete and is expensive ( ∼$3 , 000 ) , 3 ) substantial personnel time is required for cell purification , culture and monitoring supernatants for HIV replication , limiting test throughput to two to four QVOAs per lab per week , 4 ) not all replication-competent virus is detected by a single QVOA [11] , and 5 ) different laboratories employ varying methods [8] . In addition , the performance characteristics of QVOAs performed within and between labs have not been carefully evaluated , a gap that this study was designed to address . The rarity of infectious units complicates the analysis of performance characteristics , because even split samples from the same collection can have large relative differences in the true numbers of infectious units that they contain . In addition , the number of infectious units in positive wells is not known , which also adds to the assays’ variability . We describe in the next section Markov-chain Monte Carlo ( MCMC ) methods that we developed to account for these inevitable background sources of variation while estimating additional variation , including batch effects and inter-lab variation , as well as assessing the impact of freezing PBMC samples on assay performance . We also describe simulations to validate these methods and to assess the implications of the parameter estimates . We then present the results of our method-validation simulations , estimated model parameters based on the results of four QVOAs applied to 75 split samples , and simulation results evaluating some implications of the models , before concluding with some additional discussion .
Participants in the RAVEN project are enrolled and followed as part of the UCSF OPTIONS and SCOPE programs , with specific consent for apheresis collections and testing for this study as approved by the UCSF Committee on Human Research ( IRB ) # 10-03244 . The five ART-suppressed HIV-1 infected participants for the current study were selected to have diverse replication-competent reservoirs based on QVOA results from a previously published study [10] . Leukapheresis collections from five HIV+ participants and one uninfected control participant were divided into 12 aliquots ( control ) or 15 aliquots ( each HIV+ participant ) . Each aliquot contained roughly 300—750 million PBMC as requested by the testing labs . Three aliquots from each HIV+ participant comprised the fresh panel; all other aliquots were stored at −180° C and comprised the frozen panel . Four labs participated in the study: University of Pittsburgh ( U . Pitt . ) , University of California San Diego ( UCSD ) , Johns Hopkins University ( JHU ) , and Southern Research ( SR ) . One aliquot per participant from the fresh panel was distributed for immediate testing to three of the labs ( all except SR ) . The frozen panel included 18 uniquely coded liquid nitrogen-cryopreserved PBMC aliquots that were distributed to all four labs for testing 4—12 months after freezing , depending on lab testing capacity . Aliquots were blinded as to participant and aliquot identity , except that all labs had knowledge that the negative control was not included in the fresh panel . Within each lab , the frozen panel was analyzed in balanced batches of two aliquots each , designed to enable measurement of both within-batch and between-batch variation ( Fig 1 ) . Labs thawed aliquots ( if frozen ) , isolated CD4+ T cells ( CD4s ) , and performed QVOA per lab protocol ( S2 Table ) . S3 Table provides the well configurations ( cell input counts and number of replicate wells ) used for each aliquot . While QVOA output is typically reported as an estimated infection frequency and confidence interval ( expressed as infectious units per million , or IUPM ) , all labs reported individual well outcomes ( positive or negative for viral outgrowth ) . Reporting at this higher level of granularity allowed for more accurate statistical modeling . Individual laboratory reports were unblinded , checked for transcription errors if there were discrepancies in IUPMs calculated by the RAVEN statistical team and IUPMs reported by the laboratories , and then compiled for statistical analysis . S4 Table reports the resulting dataset . The JHU lab used two different protocols for the fresh and frozen panels: For the fresh panel , viral outgrowth was measured at day 7 and 14 of coculture for all aliquots , but continuation to day 21 was contingent on the results at day 14 . For the frozen panel , viral outgrowth was measured at days 7 , 14 and 21 for all aliquots . Owing to this variance in methods , we present analysis of the data in two ways , differing in treatment of the JHU lab: Our primary analysis used cumulative QVOA results through day 21 measurements from the JHU frozen panel and no measurements from the JHU fresh panel , while our secondary analysis used day 14 measurements from both panels . As some wells require the full 21 days for outgrowth to be evident [13] , the day 21 measurement is more sensitive , and so the primary analysis may yield more relevant characterization of QVOA precision . The secondary analysis , however , draws upon three labs instead of two for the fresh/frozen comparison and may yield more relevant characterization of the effect of cryopreservation . Unless otherwise stated , all experimental results reported draw upon the primary analysis . As noted in the Introduction , some variability in measured IUPM is unavoidable even for a perfect assay , due to Poisson sampling variation and uncertainty about the number of infectious units that were present in positive wells . We therefore developed a statistical model that estimates additional sources of variation beyond this unavoidable background . The design permitted identification of extra variation at the aliquot level , batch-to-batch variation within each lab , and lab-to-lab variation . Below are details of the statistical model that accomplishes this task , along with our methods for fitting the model . We did not include the control participant in this modeling .
We first tested the MCMC method in simulations of the frozen panel at a single lab ( Methods: “Validation by simulation: Single lab” ) . We found that fitting a model that forces aliquot- and batch-level random effects to be positive led to over-estimates of these effects , particularly when the simulated effect was moderate or nonexistent ( median biases up to +0 . 132 log 10 , S5 Table ) . Analysis by maximum likelihood ( Methods: “Maximum likelihood estimation” ) gave poorer results . These findings indicated that priors for the random effect variances would be suitable only if they included mass at zero . We therefore implemented priors with 50% point masses at zero by using MCMC to fit an ensemble of eight separate model components , together allowing for the eight possible combinations of presence/absence of the three random effects ( aliquot-level , batch-level , and lab-level , see Methods: “Markov-chain Monte Carlo estimation” ) . In simulations mimicking the 4-lab experimental design ( Methods: “Validation by simulation: Multiple labs” ) , we found that MCMC fits of the ensemble model estimated all fixed effects and the between-lab random effect very well . Median biases in these estimates were between −0 . 013 and +0 . 012 log 10 , and 95% CI coverage was between 93 . 2% and 95 . 0% ( Table 2 ) . The model produced similar estimates for all priors tested . The HC ( 0 , 1 ) prior did slightly outperform the other three priors tested ( S6 , S7 and S8 Tables ) , and so it was chosen for all analyses of experimental data . The method also estimated the combined effect of aliquot-level and batch-level variation well ( median bias −0 . 024 log 10 , 95% CI coverage 96 . 6% ) . Yet it was more difficult to disentangle variation between these two levels: Aliquot-level variation was overestimated slightly ( median bias +0 . 043 log 10 ) , batch-level variation was substantially underestimated ( median bias −0 . 202 log 10 ) , and both had coverage slightly lower than expected . This difficulty likely arises from experimental constraints that limited batches to only one or two aliquots each . This mis-assignment of batch-level variation to the aliquot level may also explain bias and low coverage for the combined effect of aliquot-level and lab-level variation . An experimental design with larger batches may be less vulnerable to these issues . To investigate whether MCMC fitting of the ensemble model might lead to a false conclusion that variation is present at a given level , we simulated data for which one or more of the three sources of variation was removed . When all three sources were removed from simulations , MCMC rarely produced estimates surpassing the level of variation observed in the experimental data ( posterior probability weight typically no more than 0 . 1% , bottom row of S9 Table ) . When one or two sources of variation were included , MCMC often misidentified the source of variation , though misidentification as aliquot-level variation was less common and typically produced estimates smaller than the experimental estimate . Even when one source was misidentified as another , estimates of total variation coming from all three levels performed well . Although the goal of each lab’s assay is the same—to quantify infectious provirus infecting resting CD4+ T cells—infection levels reported by UCSD were consistently higher than those reported by the other three labs ( Fig 2 ) . We used the Bayesian model to estimate the systematic effect of assay characteristics and lab practices , measured as fold-change from U . Pitt . as reference ( Table 3 ) . When accounting for differing cell counts in each assay , as well as excess random variation at the aliquot , batch , and lab levels , we found that UCSD reported IUPMs averaging 9 . 2-fold higher than those reported by U . Pitt . ( 95% CI 3 . 8—24 ) . For the other two labs , credible intervals for this systematic effect spanned 1 . This result is not surprising , given methodological differences in each assay . U . Pitt . , JHU , and SR each recorded a well as positive for viral outgrowth if levels of viral protein p24 measured by ELISA exceeded that of a threshold reference sample . UCSD , on the other hand , used viral RNA detection . While RNA detection is more sensitive than the p24 assay [22] , it may have greater potential for false positive results . In fact , two wells of a single negative coded control aliquot were reported as positive ( Fig 2 ) . Both the systematic effect and assay input cell counts determine the frequency of all-negative assay results . While SR’s experimental protocol was modeled on that of JHU ( S2 Table ) , they reported more all-negative aliquots ( 4 of 15 versus 1 of 15 aliquots for JHU ) . This observation reflects the fact that SR generally recovered fewer resting CD4s from each aliquot than JHU did for input into QVOA ( average of 12 . 7 versus 26 . 8 million cells per aliquot , S3 Table ) , and it might also reflect a systematically lower IUPM scale ( half that of JHU , although credible intervals overlap ) . In addition to this systematic lab effect , we determined that random variation in excess of the baseline Poisson-binomial model was likely ( posterior probability 0 . 85 ) at all three levels—between-aliquot , between-batch , and between-lab . Excess variation in at least one level was a near-certainty ( posterior probability = 1 − 10−22 ) . Table 4 summarizes estimates of excess variation at each level . As study design was limited to batches of only one or two aliquots , it is not easy to disentangle variation at the aliquot and batch levels; estimates are inversely correlated with one another ( slope −0 . 69 , S1 Fig ) . Combining variation at both of these levels , we estimate that two aliquots , studied in two different batches at the same lab , are expected to vary 2 . 0-fold in excess of Poisson variation ( 95% CI 1 . 6—2 . 7 ) . Two aliquots , studied at two different labs , are expected to vary 2 . 3-fold in excess of Poisson variation ( 95% CI 1 . 8—3 . 5 ) , an estimate obtained by combining variation at all three levels . When accounting for excess variation , all credible intervals estimated for the effect of cryopreservation spanned 1-fold change , or the absence of an effect ( Table 5 ) . When using results from all three labs that tested fresh and frozen panel aliquots ( SR did not test fresh PBMC samples ) to estimate a single effect size , we estimated between 0 . 56- and 1 . 97-fold change in infection frequency compared to fresh aliquots . In the context of the large ( > 100-fold ) reductions sought by latency-reducing therapies , this effect is not major . Credible intervals were wider when the effect was estimated for each lab separately , and the interval was particularly wide for JHU ( 25-fold difference between top and bottom of interval ) , owing to the larger number of all-negative aliquots at this lab . Running MCMC analysis on each lab separately suggested that each lab’s assay offered a similar level of precision . The median estimates for aliquot- and batch-level variation for each lab fell within the 95% credible interval of the joint estimates in Table 4 , and none were more than 21% away from the corresponding median estimate ( S10 Table , S2 Fig ) . Precision of early readout ( day 14 ) from JHU assays was , however , estimated to be lower than that of the other assays: There was combined 3 . 5-fold variation at both levels ( 95% CI 2 . 0- to 10 . 7-fold ) , which is 69% higher ( 95% CI 10% smaller to 458% higher ) than the joint lab estimate that included data from the later JHU readout . This difference in precision may reflect the fact that allowing more time for exponential growth leads to a stronger p24 signal and clearer distinctions between positive and negative wells . S11 Table provides full joint posteriors for each separate analysis . Neglecting the strong evidence supporting excess random variation at multiple levels ( Table 4 ) can generate misleading interpretations about precision of lab procedures . To demonstrate the relevance of accounting for excess random variation , we recomputed all estimates in a model excluding this excess variation ( Tables 3 and 5 , right columns ) . While median estimates did not change greatly , credible intervals shrank by 2- to 10-fold for each parameter . Neglecting this excess variation therefore overstates certainty in parameter estimates . One particular effect of this error in our experiment would be to conclude , rather strongly ( p < 0 . 001 ) , that cryopreservation increases observed infection frequencies in the JHU lab . Paying close attention to the sources of random and systematic variation can help in choosing assays for and optimizing design of clinical trials for latency-reducing therapies . In the next two sections , we demonstrate how simulations based on the parameter estimates described above may guide this effort . While each lab’s assay aims to measure the replication-competent HIV latent reservoir , protocols differ among them: U . Pitt . , JHU , and SR use a p24 antigen test to detect viral outgrowth , while UCSD uses an RNA PCR test; U . Pitt . , JHU , and SR use PHA and gamma-irradiated PBMCs to stimulate resting cells , while UCSD uses antibody to CD3/CD28 bound to the culture plate ( S2 Table ) . Additionally , target cells added to propagate virus differ among labs . These protocol differences may explain both systematic and random variation between labs ( Tables 3 and 4 ) . We may think of these labs as measuring different aspects of latency , each with a valid claim to being a meaningful measure , with experimental and biological motivations for specific protocol choices . In the absence of an external standard defining latent reservoir size , we can nonetheless address how sensitivity of an assay affects its accuracy . Specifically , assays that use more input cells overall or have systematically high IUPMs ( high fold-change in Table 3 ) will have improved sensitivity . By drawing from the joint posterior distribution ( S11 Table ) and simulating data for each draw , we investigated how sensitivity relates to the accuracy of measuring small reservoirs ( Methods: “Characterizing accuracy of assays” ) . For typical infection frequencies ( 1 IUPM on the U . Pitt . scale , higher or lower for the other assays according to systematic effect , Table 3 ) , all four assays have nearly identical accuracy , as measured by median absolute error from a consensus standard ( Fig 3 ) . As infection frequency declines from this level , error increases for all four assays , most sharply for SR , but gradually for UCSD and JHU . In the case of SR simulated at 0 . 1 IUPM on the U . Pitt . scale ( median of 0 . 039 IUPM on the SR scale , Table 3 ) , for a majority of parameter draws , a majority of simulated assay outcomes are all-negative , resulting in infinite error on the log scale . In the case of UCSD and JHU , improved accuracy has costs: The more sensitive RNA-based assay used by UCSD may have a higher false positive rate ( 2 of 18 control wells with one million cells were reported as positive , compared to none of the 193 control wells with one million cells among the other three assays , S4 Table ) , and JHU exhausted a larger sample ( 26 . 5 million cells simulated , versus 8 . 6 to 12 . 5 million for the other assays ) . If the JHU assay is simulated with fewer cells , bringing it in line with the other assays , its accuracy profile lies between that of U . Pitt and SR ( S3 Fig ) ; overlapping credible intervals in accuracy reflect overlapping credible intervals in systematic lab effects . If false positives did in fact occur in HIV+ samples studied by the UCSD RNA-based assay , it would be reflected in our estimates as both improved sensitivity for UCSD ( higher β2 ) and reduced correlation between labs ( higher σc ) . We used simulations to investigate the possible tradeoff between sensitivity and specificity , treating the JHU assay as a gold standard ( not subject to lab-based random effect ) and subjecting UCSD to the total between-lab variation ( applying a random effect with standard deviation 2 × σ c to each participant ) . If an equal number of cells is made available to both assays , then the extra sensitivity of RNA readout outweighs the cost of extra variability for particularly small reservoirs ( IUPM of 0 . 2 or less on the U . Pitt . scale , Fig 4 ) . Uncertainty in estimates of between-lab variation , UCSD systematic effect , and JHU systematic effect , however , make definitive comparisons difficult; credible intervals for the UCSD/JHU accuracy difference overlap zero for all IUPMs simulated . If the JHU assay instead uses roughly three times as many cells as UCSD , as in the experimental study , then it does not suffer the same disadvantage at the low IUPMs simulated and enjoys a somewhat larger advantage at higher IUPMs . The accuracy of any two assays may be compared head-to-head by the same simulation method , judging their performance against a consensus standard ( applying σc equally to both assays , S12 and S13 Tables ) or against a chosen standard assay ( applying 2 × σ c to other assays , S14 and S15 Tables ) . Cryopreserving specimens allows samples taken at different points in time to be thawed and analyzed together in the same batch . This strategy eliminates the estimated 1 . 8-fold batch variation that would otherwise affect longitudinal comparisons in clinical trials , potentially boosting power to detect reduction in latency . To investigate this possibility , we simulated and analyzed data for two hypothetical latency-reducing therapies with strong effect ( 10-fold reduction in latency ) and weak effect ( 3-fold reduction ) ( see Methods: “Simulation of clinical trial” ) . For each therapy , we simulated both a p24 assay based on JHU protocol , which uses all available cells from a participant , and an RNA assay based on UCSD protocol , which uses a fixed number of cells regardless of availability . Consistent with JHU assay cell counts ( S3 Table ) , we supposed that roughly half as many cells would be recovered from cryopreservation as would be available fresh . For the weaker therapy , batching substantially improved power to detect and accuracy to measure latency reduction following treatment ( Table 6 ) . Improvement was greater for the UCSD protocol , with power increasing from 58 . 6% to 82 . 6% and median absolute error declining from 0 . 14 log10 to 0 . 10 log10 . Although freezing-thawing sacrificed half of the resting cells available to the JHU protocol , the benefits of batching overcame this deficit , leading to a 15 percentage-point increase in study power . For the stronger therapy studied in a smaller cohort , batching provided a modest benefit to the UCSD protocol , but reduced cell availability in the JHU protocol eliminated this benefit completely . Moreover , both protocols underestimated the effect of treatment ( −7% to −34% bias , Table 6 ) . This outcome resulted from our use of an imputed IUPM in the case of negative assay results , a common practice in the field [23] . Strong therapy that reduces latency 10-fold can often result in true IUPMs well below the imputed value . In the batched JHU protocol , 35% of samples returned a negative assay , leading to the largest underestimate of treatment effect . To provide an alternative to the t-test that avoided imputation , a maximum likelihood estimate of effect size was also computed . When this method was used , batching improved both power and accuracy in all cases , but the resulting confidence intervals had poorer coverage ( S16 Table ) . None of the three software packages that we tried produced fully satisfactory results when applied to the experimental data ( Methods: “Maximum likelihood estimation” ) . The best log-likelihood obtained was from a model in Stata that excluded between-lab variation . When this variation was included , the log-likelihood closely approached that of the best model before ending with an error . In SAS , the model fitting completed for essentially the same model , with σc = 0 , but with most parameters having standard errors equal to zero . In R , the model converged to substantially different parameter values , with a much worse log-likelihood . Aside from the lack of between-lab variation , which may reflect the bias toward zero seen in single-lab simulations ( see Methods: Maximum likelihood estimation ) , many results by maximum likelihood were similar to the MCMC results given above . The cryopreservation effect and systematic lab effects were all within 10% of MCMC posterior medians . The typical between-aliquot excess variation was estimated to be 1 . 61-fold ( vs 1 . 50-fold by MCMC ) , with a p-value of 0 . 0003 . The typical between-batch excess variation was estimated to be 1 . 43-fold ( vs 1 . 77-fold by MCMC ) , with p = 0 . 098 . A likelihood ratio test for at least one source of excess variation had p = 1 . 4 × 10−13 .
This study provides evidence that QVOA assays have extra variation , beyond what is theoretically inevitable , at three levels: between split samples even in the same batch , between batches run using the same assay , and between different assays . Results for stored frozen samples did not appear to differ systematically from those on fresh samples . We developed and validated methods for fitting detailed statistical models of assay variation . We are now using these methods to evaluate faster and cheaper alternatives to the classical QVOA assays evaluated here [27] . | The latent reservoir of resting CD4+ T cells is a major , if not the primary , obstacle to curing HIV . Quantitative viral outgrowth assays ( QVOAs ) are used to measure the latent reservoir in ART-suppressed HIV-infected people . Using QVOA is difficult , however , as the fraction of cells constituting the latent reservoir is typically about one in one million , far lower than other infectious disease biomarkers . To study reliability of these assays , we distributed 75 PBMC samples from five ART-suppressed HIV-infected participants among four labs , each conducting QVOA and following prespecified sample batching procedures . Using a Bayesian statistical method , we analyzed detailed assay output to understand how results varied within batches , between batches , and between labs . We found that , if batch variation can be controlled ( i . e . , a lab assays all samples in one batch ) , typical assay results are expected to differ from the true value by a factor of 1 . 6 to 1 . 9 up or down . We also found that freezing , storing , and thawing samples for later analysis caused no more than a 2-fold change in results . These outcomes , and the statistical methods developed to obtain them , should lead towards more precise and powerful assessments of HIV cure strategies . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [] | 2019 | Assessing intra-lab precision and inter-lab repeatability of outgrowth assays of HIV-1 latent reservoir size |
Heterochromatin and associated gene silencing processes play roles in development , genome defense , and chromosome function . In many species , constitutive heterochromatin is decorated with histone H3 tri-methylated at lysine 9 ( H3K9me3 ) and cytosine methylation . In Neurospora crassa , a five-protein complex , DCDC , catalyzes H3K9 methylation , which then directs DNA methylation . Here , we identify and characterize a gene important for DCDC function , dim-3 ( defective in methylation-3 ) , which encodes the nuclear import chaperone NUP-6 ( Importin α ) . The critical mutation in dim-3 results in a substitution in an ARM repeat of NUP-6 and causes a substantial loss of H3K9me3 and DNA methylation . Surprisingly , nuclear transport of all known proteins involved in histone and DNA methylation , as well as a canonical transport substrate , appear normal in dim-3 strains . Interactions between DCDC members also appear normal , but the nup-6dim-3 allele causes the DCDC members DIM-5 and DIM-7 to mislocalize from heterochromatin and NUP-6dim-3 itself is mislocalized from the nuclear envelope , at least in conidia . GCN-5 , a member of the SAGA histone acetyltransferase complex , also shows altered localization in dim-3 , raising the possibility that NUP-6 is necessary to localize multiple chromatin complexes following nucleocytoplasmic transport .
The densely staining regions of eukaryotic chromosomes , referred to as heterochromatin , typically contain repetitive , A:T rich DNA , and are characterized by low gene density , reduced genetic recombination , di- or tri-methylation of lysine 9 on histone H3 ( H3K9me2 or H3K9me3 ) , and DNA methylation [1–4] . Heterochromatin is critical for centromere and telomere function , and is largely responsible for the silencing of transposable elements [1 , 4] . Unlike the situation in animals and plants , which require DNA methylation for normal development [5 , 6] , the fungus Neurospora crassa does not require DNA methylation for viability [4 , 7] . Neurospora has characteristics of heterochromatin found in higher eukaryotes and is convenient for genetic and biochemical studies . These traits have led to the identification of genes involved in establishing , maintaining , and regulating DNA methylation and other features of heterochromatin [8] A single DNA methyltransferase ( DNMTase ) , DIM-2 , is responsible for all DNA methylation in vegetative tissue of Neurospora [9] . DIM-2 directly interacts with heterochromatin protein-1 ( HP1 ) [10] , which binds to H3K9me3 [11 , 12] . The histone methyltransferase ( HMTase ) DIM-5 [13 , 14] is responsible for trimethylation of H3K9 . In vivo , DIM-5 activity depends on all members of the five protein complex , DCDC ( DIM-5/-7/-9 , CUL4 , DDB1 Complex ) but DIM-7 alone appears sufficient to target DIM-5 to incipient heterochromatin regions [8 , 15] . DCDC resembles Cullin-4 E3 ubiquitin ligase complexes , with the WD-40 protein DIM-9 being the putative DCAF ( DDB1/CUL4 associated factor ) , which is normally expected to recognize substrates destined for ubiquitination . However , results of recent studies indicate that DCDC does not function as a canonical ubiquitin ligase [16] . Thus , important questions regarding how DCDC and other members of the heterochromatin/DNA methylation machinery function and are controlled remain unanswered . To improve our understanding of the control of DNA methylation and heterochromatin formation , we characterized the Neurospora dim-3 strain , which shows a substantial loss of DNA methylation [7] . Genetic mapping , whole genome sequencing , and complementation tests identified the causative mutation in the nup-6 gene , resulting in a critical change in the eighth ARM repeat of NUP-6 ( Importin α ) . This protein ( also known as Srp1p in yeast and karyopherin α in humans ) is the canonical nucleocytoplasmic transport adaptor . Importin α binds “cargo” proteins to be transported into the nucleus , complexes with Importin β , and then is shuttled through the nuclear pore complex to the nucleoplasm [17–19] . We found that dim-3 strains have a drastic reduction in global H3K9me3 , indicating that NUP-6 is required for proper DCDC function . Nuclear transport and interactions of critical DCDC components appear normal in dim-3 strains but at least two DCDC components ( DIM-5 and DIM-7 ) are mislocalized from heterochromatin . Curiously , the SAGA histone acetyltransferase , but not DCDC components , showed increased localization in euchromatin of a dim-3 strain . Altogether , our results reveal a nuclear transport-independent role of NUP-6 in localizing chromatin complexes to sub-nuclear targets .
The dim-3 gene was genetically identified in a brute-force screen for methylation defects following N-methyl-N’-nitro-N-nitrosoguanidine mutagenesis [7] . Southern blots probed for the representative interspersed heterochromatic regions 8:G3 , 8:A6 , and 2:B3 [20] in a histidine auxotrophic dim-3 strain illustrate the substantial DNA methylation reduction caused by the allele ( Fig . 1A ) . Genome-wide bisulfite sequencing ( BS-Seq; Rountree and Selker , in preparation ) demonstrated that the residual DNA methylation in a dim-3 mutant is distributed normally , or largely normally , to heterochromatic regions ( Figs . 1B and S1 ) . Dim-3 was mapped to the right arm of Linkage Group V , and high-throughput sequencing identified two point mutations in the open reading frame of gene NCU01249 , E396K and R469H; both were confirmed by Sanger sequencing . NCU01249 encodes NUP-6 ( NIH GenBank accession EAA31416 . 1 ) , which is predicted to form a structure similar to yeast Importin α [21] , a protein that is highly conserved in eukaryotes [19] . Neurospora NUP-6 includes an N-terminal ∼80 amino acid Importin β-binding domain ( IBB ) , followed by ten ∼40 amino acid ARM repeats ( Fig . 1C ) , each of which should fold into a triple-alpha helical bundle to form a binding pocket for the nuclear localization signals ( NLS ) of cargo proteins ( S2A Fig . ; [22] ) . ARM repeats 1–9 are thought to recognize the NLS [22] while cargo release factors interact with the less-conserved 10th ARM repeat [23] . The E396K and R469H changes in nup-6dim-3 are in the putative 8th and 10th ARM repeats , respectively ( Figs . 1C and S2A ) , and should not destabilize the NUP-6dim-3 protein nor impact its nuclear shuttling ( S2B-S2C Fig . ) . Interestingly , dim-3 strains have a minor growth defect ( S2D Fig . ) and are homozygous sterile , indicating that one or both of the amino acid substitutions compromise an important cellular process . To confirm that nup-6dim-3 causes the observed DNA methylation loss in dim-3 strains , we introduced a wild type ( WT ) nup-6 gene ( nup-6+ ) at an ectopic locus ( his-3 ) of a dim-3 strain . Ectopic nup-6+ restored global DNA methylation to WT levels at all the representative heterochromatic loci tested ( Fig . 1A ) , suggesting that nup-6dim-3 was indeed responsible for the Dim phenotype . To determine if one or both of the mutations cause the phenotype , we replaced the endogenous nup-6+ allele with engineered mutant alleles . As expected , the strain with the reintroduced nup-6dim-3 allele showed reduced DNA methylation , although somewhat less so than the original dim-3 isolate ( Figs . 1D and S3 ) . Reintroduction of the R469H mutation did not affect DNA methylation , whereas the reintroduced E396K mutation caused a loss equivalent to that observed with the reintroduced nup-6dim-3 allele , implicating this change as the causative mutation ( Figs . 1D and S3 ) . As expected , the nup-6 gene appears essential , because strains with the gene deleted are only viable as heterokaryons , containing both nup-6+ and Δnup-6::hph nuclei [24] . Thus , there is no reason to expect that the E396K creates a null allele . While comparing DNA methylation levels in various dim-3 strains , we noticed that the histidine-requiring dim-3 strains had an exacerbated reduction in DNA methylation , leading us to systematically test the possible effect of histidine on DNA methylation in dim-3 and dim-3+ strains . We found that histidine supplementation decreased DNA methylation at all heterochromatic regions tested in a dim-3 strain , as demonstrated by Southern blotting ( Fig . 2A ) and genome-wide bisulfite sequencing ( BS-Seq; Figs . 1B and S1 ) , but has no marked effect on DNA methylation levels in a dim-3+ strain . We tested several possibilities to determine the source of this histidine effect . The "cross pathway control" system causes de-repression of several amino acid biosynthetic pathways concomitant with changes in the level of an individual amino acid [25] . We tested its involvement in the histidine effect by checking for possible consequences of added arginine , lysine , and tryptophan , but found they neither caused loss of DNA methylation nor restored it when added with histidine ( S4 Fig . ) . RNAi components are not required for DNA methylation [11] , but histidine supplementation and exposure to other DNA damaging agents , induces expression of the Neurospora Argonaute QDE-2 ( EAA31129 . 2 ) and increases the levels of qiRNAs ( QDE-2-interacting small RNAs; [26] ) . We tested the possible involvement of DNA damage and RNAi in the histidine effect by growing a dim-3 strain in medium containing histidine ( his ) , hydroxyurea ( HU ) , or ethyl methane sulfonate ( EMS ) . Only histidine addition reduced DNA methylation in dim-3 strains ( S5A Fig . ) , and this effect also occurred in a Δqde-2 strain ( S5B Fig . ) , indicating the histidine effect in dim-3 strains involves neither DNA damage nor qiRNAs . Considering that NUP-6 is critical for nuclear transport and that histidine does not further deplete H3K9me3 levels ( below ) , we investigated whether the histidine effect results from altered nuclear transport and/or sub-nuclear localization of DIM-2 ( AF348971 . 1 ) or HP1 ( AY363166 . 1 ) , which operate downstream of H3K9me3 . We found no significant difference in levels of DIM-2-3xFLAG in nuclei from a dim-3 strain grown in the presence or absence of histidine relative to WT ( Fig . 2B ) , suggesting that neither histidine nor the nup-6dim-3 allele impacts DIM-2 nuclear shuttling . In addition , cytological analyses of HP1-GFP ( S6 Fig . ; [11] ) and DamID analyses of HP1-DAM ( below ) revealed no evidence of a transport defect in dim-3 strains . We next tested if histidine perturbs sub-nuclear targeting of DIM-2 or HP1 in a dim-3 strain . The localization of many heterochromatin-specific proteins , including DIM-2 , are not readily detected by standard chromatin immunoprecipitation ( ChIP ) , perhaps due to transient chromatin interactions , but can be detected by DamID [8] . Therefore , we fused the DNA adenine methyltransferase ( dam ) gene to the downstream ends of the dim-2 [27] and hpo ( encoding HP1 ) genes , expressed these constructs in dim-3 and dim-3+ strains grown in the presence or absence of supplemented histidine , and tested their localization by digestion of genomic DNA with the GAmTC-specific restriction endonuclease DpnI followed by Southern blotting . DIM-2-DAM was found to localize to heterochromatin in both dim-3+ and dim-3 strains grown in minimum medium ( Figs . 2C and S7A ) . In contrast , in a dim-3 strain grown with histidine , DIM-2-DAM showed reduced localization to heterochromatin , while histidine did not alter DIM-2-DAM localization in a dim-3+ strain ( Figs . 2C and S7A ) . Unlike the situation with DIM-2 , histidine did not reduce , and in fact slightly increased , the heterochromatic localization of HP1-DAM ( Figs . 2D and S7B ) . These findings suggest that in dim-3 strains , histidine may compromise the direct interaction between DIM-2 and HP1 that is necessary for DNA methylation in Neurospora [10] . To determine whether the reduced DNA methylation observed in dim-3 strains reflects a loss of H3K9me3 , we assessed global H3K9me3 levels by western blotting . H3K9me3 was greatly reduced in dim-3 strains compared to that in a WT strain , and was reestablished after introduction of an ectopic nup-6+ gene ( Fig . 3A ) , indicating that NUP-6 is required for normal levels of H3K9me3 . Considering that added histidine reduced DNA methylation levels in dim-3 strains ( Fig . 2 ) , we checked if histidine exacerbates the H3K9me3 loss . Western blots showed no additional loss of H3K9me3 ( Fig . 3B ) , supporting the notion that histidine reduces DNA methylation by compromising DIM-2 localization in a dim-3 background . To determine if the residual H3K9me3 , like DNA methylation , is found in normal heterochromatic regions in dim-3 strains , we carried out H3K9me3-specific ChIP with high throughput sequencing of associated DNA ( ChIP-seq ) , and despite low signal , found an apparently equivalent distribution of this chromatin mark as in a dim-3+ strain , indicating that the remaining H3K9me3 is correctly localized to heterochromatin in a dim-3 strain ( Figs . 3C and S8 ) . Presumably , decreased DNA methylation in dim-3 strains is due to the dramatic reduction of H3K9me3 in heterochromatin . To determine if the H3K9me3 loss in dim-3 strains might compromise heterochromatin-associated silencing , we tested the expression of drug resistance markers integrated at telomeric ( telVR::hph ) and centromeric ( cenVIR::bar ) sites , which are silent even in the absence of DNA methylation , i . e . in a dim-2 mutant [27 , 28] . The dim-3 allele de-repressed the telVR::hph marker as evidenced by growth in the presence of hygromycin ( Fig . 3D ) , indicating that heterochromatin in dim-3 is compromised , although this de-repression is not as striking as in an hpo mutant , which fully de-represses the marker [28] . The dim-3 allele did not de-repress the cenVIR::bar marker ( Fig . 3D ) , consistent with the existence of a DNA methylation-independent mechanism for silencing centromeric heterochromatin [27] . Considering that the canonical function of NUP-6 is to transport proteins into the nucleus , we tested if a dim-3 strain has diminished nuclear transport of proteins required for heterochromatin formation and of a protein known to depend on the Importin α/β import system . Our observations that DIM-2 and HP1 localize normally in the nucleus ( Figs . 2B and S6 ) and that dim-3 strains have only a minor growth defect ( S2C Fig . ) did not support the idea that dim-3 strains are defective in nuclear shuttling . Nevertheless , because the dim-3 mutant shows a striking loss of H3K9me3 , we tested whether nuclear transport of DCDC components ( DIM-5 [AF419248 . 1] , DIM-7 [AL513463 . 1] , DDB1 [EAA33111 . 1] , DIM-9 [XP_956278 . 2] , and CUL4 [XP_957743 . 2] ) is impaired . To monitor nuclear transport , nuclei were isolated from dim-3+ and dim-3 strains bearing DCDC members individually 3xFLAG-tagged at their endogenous loci [8 , 15] and protein levels were assessed by western blotting ( Figs . 4A and S9 ) . Nuclei preparations were shown to be clean of cytoplasmic contamination by probing western blots for phosphoglycerate kinase ( α-PGK , EAA33194 . 1; Fig . 4B ) . We found that the nuclear level of every component of DCDC was undiminished by the dim-3 mutation ( Figs . 4A and S9 ) . Indeed , some of the components actually showed increased nuclear abundance . DIM-5 and DIM-7 showed equivalent levels in dim-3+ and dim-3 strains while the nuclear levels of DDB1 , DIM-9 and CUL4 were higher in the dim-3 strain ( Figs . 4A and S9 ) . Moreover , an examination of the distribution of DCDC components between the nuclear and cytoplasmic fractions revealed no increase in the relative amount of the proteins in the cytoplasmic fraction ( Fig . 4B ) . DIM-7 , which is an exclusively nuclear protein , showed no change in abundance or localization . Similarly , while the nuclear level of DIM-5 increased slightly in the dim-3 strain ( Fig . 4A ) , the relative nuclear/cytoplasmic distribution of the protein was equivalent in dim-3+ and dim-3 strains . DIM-9 , which is predominantly nuclear in dim-3+ strains , showed an increase in both the cytoplasmic and nuclear fractions of the dim-3 strain ( Fig . 4B ) . It is interesting that the nuclear levels of DDB1 , DIM-9 , and CUL4 were elevated in dim-3 strains . The increase in CUL4 only occurred when this strain is supplemented with histidine ( compare Figs . 4B and S9 ) , which is consistent with the induction of a DNA damage response , as previously documented [26] . The basis for increased DDB1 and DIM-9 levels in dim-3 strains is not clear but may reflect a feedback mechanism to regulate the amount of functional DCDC . Results of qRT-PCR analyses of dim-7 , ddb1 , and dim-9 transcripts in dim-3+ and dim-3 strains did not reveal RNA differences that could account for the differences in protein levels ( S10 Fig . ) , suggesting the effect is at the translational or posttranslational level . To test if the dim-3 mutation impacted nuclear transport of a protein with a nuclear localization signal ( NLS ) known to be bound and transported by Importin α , we overexpressed a GFP reporter construct with an N-terminal SV40 monopartite NLS [22 , 29–32] in dim-3+ and dim-3 strains and examined nuclear GFP signal by fluorescence microscopy . We found that 91 . 6% of dim-3+ cells had strong GFP signal inside their nuclei , and this result was mirrored in a dim-3 strain , where 92 . 5% of cells had nuclear GFP ( Fig . 4C ) . Thus , NUP-6dim-3 effectively transports canonical nuclear cargo . Given that all DCDC members , as well as DIM-2 and HP1 , are transported into the nucleus in dim-3 strains , we next considered the possibility that NUP-6dim-3 somehow interferes with DCDC assembly . To test this hypothesis , we monitored DCDC component interactions in dim-3 and dim-3+ nuclei by co-immunoprecipitation ( co-IP ) assays with 3xFLAG-tagged DCDC members . We began by analyzing the interaction between DIM-7 and DIM-5 . Equivalent levels of DIM-7-3xFLAG were recovered from dim-3 and dim-3+ nuclei ( Fig . 5A ) , consistent with our finding that DIM-7 levels in the nucleus are not affected by the dim-3 mutation ( Figs . 4A and S9 ) . Equivalent levels of DIM-5 were found in co-IPs of both backgrounds , implying that DIM-7 binding to DIM-5 is not compromised in dim-3 cells ( Fig . 5A ) . Since the DIM-5 interaction with the DCDC members appears mediated through DIM-7 [15] , we assessed DIM-5 binding to DIM-9-3xFLAG in dim-3+ and dim-3 nuclei . Because more DIM-9-3xFLAG was recovered from dim-3 nuclei than from dim-3+ nuclei ( Fig . 5A ) , we normalized the amount of DIM-5 purified to the DIM-9-3xFLAG bait levels . The DIM-5 level directly correlated with the DIM-9-3xFLAG level ( Fig . 5A ) , indicating that DIM-5 is not limiting and this interaction is not compromised in dim-3 cells . We also monitored DIM-9-3XHA binding to DDB1-3xFLAG . Equal levels of DDB1-3xFLAG were purified from dim-3+ and dim-3 nuclei , and the DIM-9-3xHA interaction appeared normal ( Fig . 5B ) , suggesting no disruption in DCAF-substrate adaptor binding by the dim-3 mutation . We then monitored the nuclear interaction between DIM-7-3xFLAG and DIM-9-3xHA , which were previously shown to directly interact [15] . DIM-9-3xHA showed an increased interaction with DIM-7-3xFLAG in dim-3 nuclei ( Fig . 5C ) , potentially because DIM-9 levels are increased in dim-3 strains ( Fig . 5C , input ) . In summary , all examined DCDC members showed normal or increased interactions in dim-3 nuclei , implying that the H3K9me3 loss is not due to impaired DCDC assembly in dim-3 strains . It remained possible that the DCDC is not properly localized to heterochromatin in dim-3 strains . To investigate this possibility , we examined the localization of DIM-5 , DIM-7 , and DIM-9 in dim-3 strains by DamID . DIM-5-DAM was previously shown to localize to heterochromatin , and this localization was dependent on DIM-7 [8] . We found that the dim-3 mutation caused substantially reduced localization of DIM-5-DAM at the three representative heterochromatic regions tested ( 8:A6 , 2:B3 , and 8:G3; Figs . 6A and S11A ) . This reduction was not exacerbated by histidine ( S11B Fig . ) , in contrast to the case for DIM-2 localization ( Fig . 2 ) . Probing for euchromatin regions ( hH3 and pan-1 ) revealed that DIM-5-DAM is not misdirected to euchromatin ( Figs . 6A and S11A ) . The C-terminus of DIM-7 is critical for normal function [8] , such that DAM-tagged DIM-7 does not fully complement dim-7 mutations . Nevertheless , as with DIM-5-DAM , we found that association of DIM-7-DAM with the heterochromatin regions ( 8:A6 , 2:B3 , and 8:G3 ) was also markedly reduced in the dim-3 background ( Figs . 6B and S11C ) , suggesting that NUP-6dim-3 impacts targeting of DIM-7-DAM to heterochromatin . In contrast , DamID of DIM-9-DAM did not reveal marked differences between the dim-3+ and dim-3 strains ( Figs . 6B and S11C ) . We note that DIM-7-DAM and DIM-9-DAM may show a greater association with the euchromatic gene tested ( hH3 ) than did DIM-5-DAM , perhaps reflecting an unknown role of these proteins outside of heterochromatin . The interaction of DIM-7-DAM , but not DIM-9-DAM , with the euchromatin marker was reduced in the dim-3 strain ( Figs . 6A and S11C ) . To confirm that DIM-7-DAM is mislocalized , we expressed DIM-7-GFP in dim-3+ and dim-3 vegetative tissue . The level of DIM-7-GFP produced from the native dim-7 promoter was not cytologically detectable , leading us to express it under the control of the stronger ccg-1 promoter . Despite overexpression , DIM-7-GFP co-localized with HP1-mCherry ( S12 Fig . ) , at least in the cells examined ( conidia ) , suggesting that DIM-7-GFP behaves normally to mark heterochromatic regions in vivo despite the importance of the DIM-7 C-terminus [8] . In 93 . 5% of dim-3+ cells DIM-7-GFP formed compact foci , typically at or near the nuclear periphery ( Figs . 6C and S12 ) , consistent with DIM-7-GFP marking heterochromatic regions . Interestingly , approximately half of dim-3 nuclei examined also showed such foci ( Fig . 6D ) . It would be interesting to learn whether the observed difference between dim-3 and dim-3+ cells reflect differences in cell cycles of these strains . Unfortunately , because no genetic method to synchronize Neurospora has been developed , our studies were limited to unsynchronized cells . DIM-7-GFP appears equivalently expressed and transported into nuclei of dim-3+ and dim-3 strains ( S13 Fig . ) . Thus , the dim-3 mutation seems to partially perturb normal localization of DIM-7 fusion proteins within the nucleus . The cytological mislocalization of DIM-7-GFP in a dim-3 mutant led us to examine the distribution of NUP-6 and NUP-6dim-3 as well . Our expectation , based on work in other systems [31 , 33 , 34] , was that NUP-6 would be largely associated with the nuclear periphery . Indeed , at least in conidia , the majority of NUP-6-GFP was nuclear , but consistent with its role in nuclear transport , a small percentage was observed in the cytoplasm . Most ( 75 . 3% ) dim-3+ cells showed NUP-6-GFP localization near the nuclear membrane ( NM ) , forming either foci or a contiguous ring surrounding the genomic DNA ( Fig . 7A , upper panel; additional examples of NUP-6-GFP localized at the nuclear periphery are provided in S14 Fig . ; Z-stack fluorescent images in S1 Movie ) . We note that the vast majority of the NM-associated NUP-6 does not co-localize with HP1-marked heterochromatin ( S15 Fig . ) although some overlap of fluorescent signals was occasionally observed ( yellow arrows ) leaving open the possibility that NUP-6 is transiently associated with heterochromatin . A minority of dim-3+ cells ( 24 . 7% ) show diffuse NUP-6-GFP in the nucleus ( Fig . 7A , lower panel ) . We do not know the basis for the variability in nuclear localization of NUP-6+-GFP but it is noteworthy that , because of technical limitations , our study used unsynchronized cells , leaving open the possibility that the protein is differentially distributed in the cell cycle and that the dim-3 mutation affects the cell cycle . In contrast to the case in dim-3+ cells , a majority of dim-3 conidia ( 73 . 1% ) showed diffuse localization of NUP-6dim-3-GFP in the nucleus ( Fig . 7B , lower panel; additional examples of dispersed , nuclear NUP-6dim-3-GFP are provided in S16 Fig . ; also evident in Z-stacks of NUP-6dim-3-GFP , S2 Movie ) . Nevertheless , 26 . 9% of dim-3 conidia still showed compact structures at the nuclear periphery , similar to those observed with wild type NUP-6 , consistent with the partial phenotype of dim-3 strains . The NUP-6+-GFP foci were found at the nuclear envelope , as shown by co-localization with a tagged component of the nuclear pore complex , NUP-84-mCherry ( [35 , 36]; XP_964074 . 1; Fig . 7C ) ; NUP-6dim-3-GFP was not seen to localize at the nuclear envelope ( Fig . 7D ) . Thus , both NUP-6 and DIM-7 appear to be partially mislocalized from the nuclear envelope in the dim-3 background , raising the possibility that mislocalization of NUP-6 leads to mislocalization of DIM-7 and other members of DCDC . The results presented above suggest that NUP-6 plays a role in the normal localization of DCDC at heterochromatin . To explore the possibility that NUP-6 is also involved in localizing proteins associated with euchromatin , we examined the histone acetyltransferase complex SAGA , which is normally associated with a subset of gene promoters [37 , 38] . We built C-terminal DAM fusions for two components of the complex , GCN-5 ( EDO65389 . 2 ) , the acetyltransferase subunit , and TAF-5 ( XP_961292 . 2 ) , a structural subunit , and tested their localization by DamID at the euchromatic am ( amination deficient ) gene ( EAA32325 . 1 ) and the hpo gene . Relative to the situation in the wildtype strain , the dim-3 strain , GCN-5-DAM showed increased association to am and hpo , while TAF-5-DAM showed no marked change in its association with these genes ( Fig . 7E ) . As expected , these SAGA members were not found at heterochromatin ( Fig . 7F ) . We conclude that NUP-6 differentially influences the localization of members of DCDC and SAGA , representatives of chromatin complexes normally found in heterochromatin and euchromatin , respectively .
Our characterization of the Neurospora dim-3 mutation , which causes a partial loss of DNA methylation [7] , led us to discover an unexpected function for the nuclear transport protein NUP-6 ( Importin α ) . We mapped dim-3 to a region of Linkage Group V and determined that a mutation causing a glutamic acid to lysine substitution ( E396K ) in nup-6 is responsible for the methylation defect . NUP-6 is the Neurospora homolog of the Xenopus / Drosophila Importin α , yeast Srp1p , and human Karyopherin α . As first demonstrated with Xenopus Importin α [39] , this protein serves as an adaptor for classical nucleocytoplasmic transport , shuttling proteins from the cytoplasm to the nucleus through the nuclear pore complex [17–19] . We found that in addition to its defect in DNA methylation ( Fig . 1A-B ) , dim-3 strains have an even more pronounced deficiency in H3K9me3 ( Fig . 3A-B ) and show reactivation of the silenced telomeric telVR::hph marker ( Fig . 3D ) , indicating that heterochromatin is compromised in dim-3 strains . Interestingly , results of DamID suggest that both HP1 and DIM-2 can appropriately localize to heterochromatin in dim-3 strains grown in minimal medium ( Fig . 2C-D ) . Apparently the reduced H3K9me3 levels in dim-3 strains still allow nucleation of near-wild type levels of HP1 , which in turn recruits DIM-2 , resulting in only moderately reduced DNA methylation ( Fig . 2A ) . Histidine supplementation exacerbates loss of DNA methylation in the dim-3 mutant ( Fig . 3A ) and results in mislocalization of DIM-2 ( Fig . 2C ) without causing further reduction of H3K9me3 or loss of HP1 association ( Figs . 3B , 2D and S7B ) . This curious histidine effect , which is not found with wild type strains and is independent of induction of DNA repair processes by histidine [26] and other agents , likely results by compromising the DIM-2—HP1 interaction or by perturbing DIM-2 activity . Considering that H3K9me3 depends on a five protein complex , DCDC ( DIM-5/-7/-9 , CUL4 , DDB1 Complex ) , it seemed likely that the reduced tri-methylation of H3K9 caused by the dim-3 mutation resulted from defective nuclear transport of at least one member of the DCDC . Surprisingly , we observed normal , or increased , nuclear levels of all DCDC components ( Figs . 4A-B and S9 ) . Similarly , HP1 and DIM-2 , which are downstream of H3K9me3 in the DNA methylation pathway , showed no reduction in nuclei ( Figs . 2B and S6 ) . Likewise , a canonical NLS-containing import substrate was effectively transported into dim-3 nuclei ( Fig . 4C ) . Together , these observations suggest that NUP-6dim-3 has no nucleocytoplasmic transport defect . Because NUP-6dim-3 does not appear to compromise nucleocytoplasmic transport of the DNA methylation machinery , we investigated whether NUP-6 is involved in targeting DCDC to heterochromatin . DamID experiments revealed that DIM-7 and DIM-5 association with heterochromatin is reduced in a dim-3 strain ( Fig . 6A-B ) . Moreover , cytological examination of DIM-7-GFP showed a marked reduction in foci in the dim-3 background ( Fig . 6C-D ) despite being equivalently expressed and transported in the different backgrounds ( S13 Fig . ) , suggesting that NUP-6 is necessary for proper localization of one or more members of the DCDC . This mislocalization presumably results in other defects noted in dim-3 strains , including the large reduction in H3K9me3 ( Fig . 3A-C ) , the reduced HP1-GFP localization ( S6 Fig . ) , and the loss of telVR::hph silencing ( Fig . 3D ) . Interestingly , despite being mislocalized , the DIM-5 and DIM-7 interactions with other DCDC components appear normal ( Fig . 5 ) , suggesting that formation of this complex does not depend on localization at its normal sub-nuclear site . Since the remaining H3K9me3 in dim-3 is only found at its normal locations ( Fig . 3C ) , it is conceivable that the mislocalized DCDC does not tri-methylate H3K9 because the DCDC requires some signal from the sequences that underlie heterochromatin for its activity . The observation that DIM-9 localization to heterochromatin was only minimally disrupted in a dim-3 strain ( Fig . 6B ) , raises the possibility that DIM-9 , DDB1 , and CUL4 use a different mechanism from that of DIM-5 and DIM-7 for heterochromatin targeting to form the DCDC . We conclude that dim-3 strains are not defective in nuclear transport , but are partially defective in localizing some of the heterochromatin machinery . We were curious whether euchromatin machinery might also be affected , either positively or negatively . DamID analyses of two members of the SAGA histone acetyltransferase complex revealed increased localization of one of the proteins ( GCN-5 ) to euchromatin loci in the dim-3 background ( Fig . 7C ) . Interestingly , studies in Drosophila demonstrated an interaction between Importin α and the chromatin insulator protein CTCF [40] . There have been previous clues from S . cerevisiae and Xenopus of nuclear transport-independent roles of Importin α homologs . In yeast , srp1 was originally isolated as a dominant suppressor of ribosomal RNA-specific Polymerase I mutations [33] , yet two of the characterized srp1 alleles had markedly different phenotypes that showed intragenic complementation [41 , 42] . The srp1-31 allele is defective in the nuclear transport of mitotic control proteins [43 , 44] , whereas the srp1-49 allele is proficient for nuclear transport , but shows defects in nucleolus formation and proteasome function [41 , 42] . In Xenopus , Importin α inhibits the onset of mitosis by binding the mitotic-promotion factor TPX2 in non-mitotic cell cycle stages [32] . Nuclear transport-independent functions of NUP-6 may require a stable localization to specific sub-nuclear regions . In Xenopus , nuclear membrane ( NM ) associated Importin α promotes nuclear envelope formation independent of the nuclear transport components Importin β , Ran , or CAS [45] , and in yeast , Srp1p is in proximity of the NM [31 , 33 , 34] . We observed a similar localization for NUP-6+-GFP in Neurospora conidia ( Fig . 7A ) , while NUP-6dim-3-GFP was predominantly found away from the NM ( Fig . 7B ) ; the localization of NUP-6-GFP at the NM may vary with stages of the cell cycle and might be differentially controlled in individual cells , possibilities we cannot currently address . The apparent association of NUP-6 with the NM might be mediated through an interaction with the nuclear pore complex ( NPC ) , as we found that NUP-6+-GFP co-localized with the NPC member NUP-84 ( Fig . 7C ) . An association of NUP-6 with the inner NM could provide a platform to target chromatin complexes ( Fig . 7G , left ) . We speculate that the dim-3 change in NUP-6 compromises this targeting ( Fig . 7G , right ) . Of possible relevance to our findings suggesting a nuclear transport-independent function of NUP-6 / Importin α for heterochromatin formation and DNA methylation , the yeast NPC member Nup170p was recently reported to directly interact with chromatin , including the nucleolar-ribosomal DNA repeats and sub-telomeric regions to repress transcription [46] . Other nucleocytoplasmic-independent roles for the nuclear trafficking machinery have been suggested , including targeting Argonaut proteins to cytoplasmic P bodies for miRNA-mediated gene silencing by the human Importin β homolog , Importin-8 [47] . It is also noteworthy that NPC members , including Nup98 , Sec13 , and Nup50 , have been found in NPC-independent nucleoplasm “pools” that directly interact with promoters of developmental or cell cycle genes to activate transcription and increase epigenetic memory through a transport-independent mechanism [48–50] . It will interesting to learn if NUP-6 , nucleoporins , or other components of nucleocytoplasmic transport machinery or the NPC mediate heterochromatin formation , DNA methylation , and other functions through transport-independent processes in higher eukaryotes .
N . crassa strains were grown , maintained , and crossed following standard protocols [51] . All nucleic acid manipulation protocols are given in S1 Text . Gene replacement constructs were generated as described [52] with modifications detailed in S1 Text . Genomic DNA isolation ( modified from [53] ) and Southern blotting [54] were performed as previously described . Primers to amplify DNA from methylated or control regions are listed in S2 Table or [20] . Whole genome sequencing was performed as previously described [55] . ChIP-seq was performed as previously described [56] . To display ChIP-seq data , sequencing reads were mapped to the Neurospora genome using Bowtie 2 [57] . Bam files were converted to read density tdf files spanning 200bp windows using the count function of Integrative Genomics Viewer’s ( IGV ) IGVtools , and were displayed using the IGV ( www . broadinstitute . org/igv/; [58] ) . H3K9me3 ChIP-seq from a wild type strain has been previously described [56]; for the H3K9me3 ChIP-seq from dim-3 , 11 , 662 , 061 reads ( 77 . 63% of total ) mapped at least once to the Neurospora genome . The Y-axis shows the number of mapped reads at each position multiplied by 106 and divided by the total number of mapped reads [58] , providing normalization between samples . For bisulfite sequencing , genomic DNA was isolated from strains grown at 32°C with shaking for ∼48 hours in Vogel’s liquid medium , and 50ng was treated with bisulfite using the EZ DNA Methylation Lightning kit ( ZYMO Research ) . Libraries were prepared for sequencing with the EpiGnome Methyl-Seq Kit and EpiGnome Index PCR Primers ( Epicentre ) , purified with Agencourt AMPure XP beads ( Beckman Coulter ) , Qubit HS Assay quantified ( Life Technologies ) , visualized on the Fragment Analyzer ( Advanced Analytical ) , and sequenced on a Illumina HiSeq 2000 ( www . illumina . com ) at the UO Genomics Core Facility as single-end 100nt reads . The BRAT-BW software package ( compbio . cs . ucr . edu/brat/; [59] ) was used to prepare and map the reads to the N . crassa OR74A ( annotation NC12 ) genome , which was converted to a four stranded reference genome to permit bisulfite mapping . BRAT-BW acgt-count “-B” option cytosine-only files produced for the forward and reverse strand reads were merged . A python script ( bidensity ) was written to calculate the average 5mC level over a specified sliding window across the genome , producing a . wig file for IGV display . For qRT-PCR analyses , RNA was isolated as previously described [60] , Qubit-RNA assay quantified ( Life Technologies ) , and equal levels of RNA were DNAse treated ( Invitrogen ) and the first strand cDNA was synthesized with Superscript III Kit ( Life Technologies ) following the manufacturers’ protocol . Diluted cDNA was used for triplicate quantitative real-time PCR experiments using FAST SYBR Green master mix ( KAPA ) with the primers 4701/4702 ( dim-7 ) , 4661/4662 ( ddb-1 ) , and 4663/4664 ( dim-9 ) , and normalized to actin ( Accession number U78026 . 1; primers 3209 , 3210 ) ( S2 Table ) on a Step One Plus Real Time PCR System ( Life Technologies ) . Immunoprecipitation and western blotting was performed as previously described [52] using antibodies listed in S1 Text . Extraction of histones was performed as previously described [61] . Data files of H3K9me3 ChIP-seq and Bisulfite-seq have been deposited to GEO ( http://www . ncbi . nlm . nih . gov/geo/ ) under the accession numbers ( GSE61173; ChIP-seq ) and ( GSE61174; BS-seq ) . The accession number GSE61175 reports both data sets . | The epigenetic information contained in chromatin is essential for development of higher organisms , and if misregulated , can lead to the unregulated growth associated with human cancers . Chromatin is typically classified into two basic types: gene-rich 'euchromatin' , and gene-poor heterochromatin , which is also rich in repeated DNA and 'repressive chromatin marks' . As in humans and eukaryotes generally , heterochromatin in Neurospora crassa is decorated with DNA methylation and histone H3 lysine 9 ( H3K9 ) methylation , but unlike the case in mammals , loss of these epigenetic marks does not compromise viability . In Neurospora , the DCDC , a five-member Cul4-based protein complex , trimethylates H3K9 . Little information is available on the regulation of DCDC or similar complexes in other organisms . Using forward genetics , we identified a novel role for Importin α ( NUP-6 ) for the function of DCDC . Although NUP-6 typically functions in nucleocytoplasmic transport , the dim-3 strain , which contains an altered nup-6 gene that reduces DNA methylation and H3K9me3 , shows normal nuclear transport of the heterochromatin machinery and a canonical transport substrate . Two DCDC members are mislocalized from heterochromatin in the dim-3 mutant , signifying that NUP-6 may be important for targeting key proteins to incipient heterochromatic DNA . The euchromatic complex SAGA has increased euchromatin localization in dim-3 , suggesting that NUP-6 may localize multiple chromatin complexes to sub-nuclear genomic targets . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2015 | Neurospora Importin α Is Required for Normal Heterochromatic Formation and DNA Methylation |
West Nile Fever is a zoonotic disease caused by a mosquito-borne flavivirus , WNV . By its clinical sensitivity to the disease , the horse is a useful sentinel of infection . Because of the virus’ low-level , short-term viraemia in horses , the primary tools used to diagnose WNV are serological tests . Inter-laboratory proficiency tests ( ILPTs ) were held in 2010 and 2013 to evaluate WNV serological diagnostic tools suited for the European network of National Reference Laboratories ( NRLs ) for equine diseases . These ILPTs were designed to evaluate the laboratories’ and methods’ performances in detecting WNV infection in horses through serology . The detection of WNV immunoglobulin G ( IgG ) antibodies by ELISA is widely used in Europe , with 17 NRLs in 2010 and 20 NRLs in 2013 using IgG WNV assays . Thanks to the development of new commercial IgM capture kits , WNV IgM capture ELISAs were rapidly implemented in NRLs between 2010 ( 4 NRLs ) and 2013 ( 13 NRLs ) . The use of kits allowed the quick standardisation of WNV IgG and IgM detection assays in NRLs with more than 95% ( 20/21 ) and 100% ( 13/13 ) of satisfactory results respectively in 2013 . Conversely , virus neutralisation tests ( VNTs ) were implemented in 33% ( 7/21 ) of NRLs in 2013 and their low sensitivity was evidenced in 29% ( 2/7 ) of NRLs during this ILPT . A comparison of serological diagnostic methods highlighted the higher sensitivity of IgG ELISAs compared to WNV VNTs . They also revealed that the low specificity of IgG ELISA kits meant that it could detect animals infected with other flaviviruses . In contrast VNT and IgM ELISA assays were highly specific and did not detect antibodies against related flaviviruses . These results argue in favour of the need for and development of new , specific serological diagnostic assays that could be easily transferred to partner laboratories .
West Nile fever ( WNF ) is a zoonotic disease caused by West Nile virus ( WNV ) . It can cause a severe neuro-invasive disease ( WNND for WN neuro-invasive disease ) in 1–10% of infected horses . WNF cases in horses have to be reported to international organisations ( the World Organisation for Animal Health ( OIE ) and the European Commission for European countries ) . The virus is primarily transmitted by mosquitoes of the Culex genus and is amplified by wild bird reservoirs . WNV circulation has been regularly reported in the Mediterranean basin , as well as in Eastern and Central Europe since 1998 . WNV outbreaks have increased substantially since 2010 in Europe [1] . The endemic circulation of WNV in several European countries ( Romania , Italy and Spain ) and regions ( the Balkans ) argue for reinforced surveillance of this disease [2] . WNV surveillance varies among European countries , ranging from clinical surveillance of horses or humans to active surveillance of birds or other infected species through regular serological screening and/or active WNV detection in trapped mosquitoes . This active surveillance is promoted in endemic areas where horses could have been previously infected or vaccinated against WNV [3] . Nevertheless , due to its clinical sensitivity to WNV infection , the horse is a sentinel whatever the surveillance system used . Horse cases of WNV can usually be diagnosed before human cases [4 , 5] . In such a context , the improved detection of WNV infection in this species would be extremely helpful . In the case of WNND , WNV diagnosis is hampered by the short duration and low level of WNV viraemia . Consequently , serological diagnostic tests are preferred to confirm WNV infection in horses . The evidence of IgM antibodies in serum or cerebrospinal fluid ( CSF ) or the increase in Immunoglobulin G ( IgG ) titres in 2 serial samples obtained 2–3 weeks apart is sufficient to confirm WNV infection in horses . Many different serological tools are available to diagnose or screen for WNV antibodies [6] . The most commonly used are virus neutralisation tests ( VNTs ) , including the plaque reduction neutralisation test ( PRNT ) or micro-virus neutralisation test ( micro-VNT ) , immunofluorescence assays ( IFAs ) , and enzyme linked immunosorbent assays ( ELISAs ) . While rapid tests such as ELISAs and IFAs are preferred because of their sensitivity , reproducibility and affordability , WNV VNTs are still gold standard tests and offer high diagnostic specificity . Noteworthy , seropositive tests should be interpreted with care due to frequent cross-reactions among flaviviruses , especially in rapid serological tests . Indeed , WNV belongs to the Flavivirus genus and many related flaviviruses are reported in Europe such as the mosquito-borne Usutu ( USUV ) and Bagaza viruses ( BAGV ) , as well as the tick-borne encephalitis viruses ( TBEVs ) , Louping ill virus ( LIV ) or Meaban virus [6–10] . Infection by such flaviviruses has been shown to induce antibodies that generate positive results in rapid serological diagnostic tests . Recently , USUV—a flavivirus that originated in Africa—has been isolated in Austria , Belgium , the Czech Republic , Hungary , Italy , Germany , Greece , Poland , Switzerland and Spain . It has also emerged in France and the Netherlands in 2015 and 2016 respectively [8 , 11–15] . BAGV—also termed Israel turkey meningoencephalitis virus ( ITMV ) —was first isolated in 1960 in Israeli turkeys with neurological symptoms , and has more recently emerged in Europe and more specifically in Spain in 2010 , where it was found in partridges and pheasants [7] . TBEV and LIV are pathogens borne by Ixodes ticks . TBEV has been responsible for major encephalitis outbreaks in humans in northern Europe [16 , 17] . LIV is a zoonotic pathogen causing encephalitis mainly in sheep and cattle , and is concentrated in Ireland and the United Kingdom [6 , 18] . The risk of the emergence of new flaviviruses in Europe such as the Japanese encephalitis virus ( JEV ) or Zika viruses should not be neglected . JEV genomic RNA was detected in dead birds in Tuscany around 1997–2000 and in mosquitoes in Northern Italy in 2010 [19 , 20] . Such a circulation of varied flaviviruses in Europe and the large cross-reactivity of rapid WNV serological tools should be considered when performing serological WNV assays on horses . All positive serological results using rapid assays should be confirmed by a more specific VNT with the viruses known to circulate in the area . WNV surveillance in animals is supported by a network of cooperating laboratories in Europe . In 2008 , the European Commission appointed the French agency for food , environmental and occupational health and safety ( ANSES ) as the European Reference Laboratory ( EURL ) for equine diseases with the remit of harmonising and improving the diagnosis of equine diseases in Europe . Two inter-laboratory proficiency tests ( ILPTs ) were performed in 2010 and 2013 by the network of European laboratories . This study was designed to acquire insights into the performance of serological methods and protocols applied in Europe to detect WNV ( OIE manual , chapter 2 . 1 . 20 ) [21] and to evaluate the homogeneity of results among National Reference Laboratories ( NRLs ) in European and Mediterranean countries .
Seventeen European Union NRLs participated in the ILPTs on WNV serology in 2010 and 21 participated in 2013 ( Fig 1 and Table 1 ) . Participants also included reference laboratories from Morocco ( in 2010 and 2013 ) , Bosnia and Herzegovina ( 2013 ) , other European participants ( French , German , Irish , Italian or Spanish laboratories involved in WNV surveillance ) as well as participants producing commercial WNV ELISA kits ( 1 in 2010 and 3 in 2013 ) . In all , 21 and 35 participants were included in the 2010 and 2013 ILPTs respectively ( S1 File ) . A number code was assigned to each participating laboratory to ensure a blind analysis of the ILPT results . The panel of samples consisted of 16 and 15 sera in 2010 and 2013 respectively , with a minimum of 100 μl of horse serum per sample . Horse sera with or without antibodies directed against WNV or other flaviviruses were deliberately chosen . Samples were heat-inactivated at 56°C for 30 minutes , aliquoted and stored at -20°C until shipment to avoid degradation of the antibodies . The homogeneity of the serum preparation was confirmed on 5 panels by ELISA tests prior to shipment . Panels were randomly coded and shipped in dry ice to participants ( Table 2 ) . Participants were given 4 weeks to send back their results and were asked to store the samples either at -20°C or 4°C and to avoid repetitive thawing/freezing cycles . The stability of the panels at the end of the ILPT period was tested on one panel stored at 4°C throughout the ILPT period . The participants were asked to test the samples with the diagnostic methods they routinely applied in their laboratories . They could use OIE- or EURL-recommended methods , commercial kits according to the manufacturer’s instructions or in-house methods . The OIE gold standard serological assay is the PRNT with a threshold plaque reduction level of 90% ( PRNT90 ) [21] . The micro-VNT is a modification of the PRNT and allows a larger number of samples to be screened using cell microplates [25] . Different ready-to-use diagnostic ELISA kits for veterinary purposes were commercially available . Two commercial competition kits—the ID screen WNV competition kit ( ID Vet ) and the Ingezim WNV Compaq kit ( Ingenasa ) —had been developed before 2010 and could be used to detect anti-E antibodies in multiple species . Both ELISA kits use plates pre-coated with WNV recombinant antigens and measure the competition between antibodies present in the animal serum tested and a monoclonal anti-E antibody conjugated to horseradish peroxidase ( HRP ) . These competitive ELISA kits detect virtually every Ig isotype , but are mainly used to detect IgG and are classified herein among the IgG detection tools . IgMs are specifically detected by IgM Antibody Capture ELISAs ( MAC-ELISA ) . Equine anti-WNV IgMs present in horse sera bind to anti-horse IgM antiserum coated on ELISA plates . This binding is revealed by the addition of a positive antigen ( recombinant WNV antigen ) , a monoclonal antibody directed against WNV antigen and conjugated to HRP and a chromogenic substrate . While the IDEXX IgM WNV Ab kit was available prior to 2010 , the Ingezim WNV IgM and ID screen WNV IgM capture kits were marketed after 2010 , so were only used during the 2013 ILPT . A statistical analysis on 2010 and 2013 ILPT results was carried out on the kits used for both ILPTs , to analyse the effect of the country and/or the effect of the batch on the result . A linear mixed model was used . The results obtained , i . e . the %S/N values with the ID screen WNV competition kit or ISR values with the IDEXX IgM WNV Ab kit , was the dependent variable . The country and the batch number were treated as fixed effects , the serum number being treated as a random effect . Log-likelihood tests were used to analyse the effect of the country and of the batch . This analysis was performed separately on results obtained for the detection of the anti-E antibodies ( ID screen WNV competition kit was the only kit used in both ILPTs ) and on results obtained for IgM detection ( IDEXX IgM WNV Ab kit being the only test used in both ILPTs ) . Statistical analyses were conducted using R ( R-Core Team , 2015 ) [26] .
The number of European NRLs participating in these WNV ILPTs increased from 17 in 2010 to 21 in 2013 . With such a condensed network of laboratories dedicated to WNV detection in animals , European countries should rapidly evidence WNV outbreaks . All the participants used commercial competition kits to detect anti-WNV antibodies in both ILPTs . Only 6% of NRLs ( 1/17 ) in 2010 and 5% ( 1/21 ) in 2013 used an in-house IgG indirect ELISA ( with inactivated virions ) in addition to the commercial competition kit . Following the commercialisation of horse WNV MAC-ELISAs , the number of NRLs detecting anti-WNV IgM antibodies clearly increased between 2010 ( 18% ( 3/17 ) NRLs ) and 2013 ( 62% ( 13/21 ) ) . An in-house IgM ELISA was used by only one NRL in 2010 . In contrast , the WNV VNT—which requires the handling of infectious WNV in BSL-3 facilities—was not implemented in new NRLs between 2010 ( 8/17 ) and 2013 ( 7/21 ) ( Fig 1 and S1 File ) . ELISAs are the preferred screening tools because of their rapidity , high throughput and sensitivity . Moreover , competitive ELISA kits in veterinary laboratories offer the possibility of testing sera from various animal species . During the 2010 and 2013 WNV ILPTs , the ID screen WNV competition kit was used most often ( 90% ( 19/21 ) of assays in 2013 ) , while the Ingezim WNV Compaq kit was used by only two NRLs ( 2/21 ) . These two kits detect antibodies against the WNV Envelope ( E ) glycoprotein that are induced after natural infection or vaccination ( the WNV E structural antigen is expressed by all WNV vaccines marketed for horses ) [24] . Nevertheless , depending on the WNV conjugate used in the kit ( anti E-domain III in the Ingezim WNV Compaq kit or anti-E antibodies directed against less specific domains in the ID screen WNV competition kit ) , cross-reactions following infections with other flaviviruses were observed to a greater or lesser extent ( see “results according to the method” ) . The sensitivity of IgG detection methods was evaluated with samples containing high levels ( S5 , S6 and S13 in 2010 , as well as S3-S7 in 2013 ) or low levels of WNV-IgG ( S7 and S16 in 2010 and S8-S9 in 2013 ) ( Table 2 ) . Two participants in 2010 and only one participant in 2013 encountered sensitivity issues with commercial kits ( Fig 2 , S2 and S3 Files ) . The in-house indirect IgG ELISA ( participant E , S2 File ) clearly lacked sensitivity in 2010 , producing 4 false negatives , but an improvement in its sensitivity was observed in 2013 with only one positive sample generating a doubtful result ( participant E , S3 File ) . The specificity of IgG methods was first evaluated on negative horse sera ( S1-S4 , S10 and S11 in 2010 ) . In 2010 , no cross-reactive antibodies to heterologous flaviviruses were present in the panel . The specificity was 100% with 6 negative horse sera found negative by all the participants . In 2013 , samples with antibodies against TBEV , USUV and JEV were added to the panel . Due to the fact that the monoclonal antibody used in some commercial kits recognises flaviviruses other than WNV , the test specificity was first evaluated on horse sera negative for anti-flavivirus antibodies ( S1 , S10-S12 ) . The specificity was 100% with 4 negative horse sera found negative by all the participants using either commercial or in-house ELISAs ( Fig 2 ) . Test reproducibility and coherent dose-effect curves were successfully achieved by every NRL in 2010 and 2013 , with the exception of a single participant in 2010 ( participant H ) who obtained differing results on 2010-S5 and S6 duplicate samples . Taken together , the NRL network obtained very good qualitative results , with 14/17 ( 82% ) NRLs in 2010 , and 19/20 ( 95% ) in 2013 achieving satisfactory sensitivity , specificity and reproducibility . A statistical analysis on 2010 and 2013 results was carried out on the only kit used for these 2 ILPTs ( i . e . the ID screen WNV competition kit ) , to analyse the effect of the country and/or effect of the batch on the result . This analysis revealed that the %S/N values ( on a log scale ) varied significantly according to the NRL ( p<0 . 0001 ) and to the kit batch ( p<0 . 0001 ) ( Figs 3 and 4 ) . Both effects were observed when considering 2010 data ( country: p = 0 . 0001 , batch: p<0 . 0001 ) , whereas the batch effect was not evidenced when considering data from 2013 ( country: p<0 . 0001 , batch: p = 1 ) . Indeed , in 2010 , batch 215 was clearly less sensitive than batch 145 , the S14-S15-duplicate sample , for example , being found mostly positive with batch 145 but mostly negative with batch 215 . In contrast , in 2013 , the distributions of %S/N results obtained using batches 334 and 460 appeared similar ( Fig 3 ) . MAC-ELISAs were used to detect immune response associated with acute and recent viral infection . The secretion of anti-WNV IgM can generally be detected as early as 2 to 8 days after the onset of initial clinical signs ( fever ) . Three NRLs in 2010 and 5 in 2013 used the IDEXX IgM WNV Ab kit , while 8 other NRLs used the ID screen WNV IgM capture in 2013; only 1 participant in 2010 used an in-house MAC-ELISA . Sensitivity was assessed against 2 IgM-positive horse sera in 2010 ( S14-S15-duplicate samples generated during a recent WNV infection ) and 2 IgM-positive sera in 2013 ( S2 and S3 ) produced by experimental infection . Specificity was evaluated on the same samples as for the IgG ELISA tests ( 6 negatives in 2010 and 4 negatives in 2013 , plus 3 sera produced after experimental infection with JEV , USUV and TBEV in 2013 ) . The NRL network achieved 100% specificity and 100% sensitivity in 2010 and 2013 with commercial and in-house MAC ELISAs ( S2 and S3 Files ) . A statistical analysis on 2010 and 2013 results from the IDEXX IgM WNV Ab kit ( the only test used during both the 2010 and 2013 ILPTs ) did not show a country or batch effect ( p>0 . 05 ) . VNT is the gold standard serological tool recommended by the OIE Manual of Diagnostic Tests and Vaccines for Terrestrial Animals . This method offers higher specificity but appears less sensitive than ELISAs and is usually used as a confirmation and a titration method for WNV-specific neutralising antibodies ( from serum or cerebrospinal fluid , CSF ) [27] . In all , 8 NRLs in 2010 and 7 NRLs in 2013 used either the PRNT90 or micro-VNT on Vero cells with various WNV isolates ( depending on NRLs , strains generally belonging to lineage 1—Eg101 , Is98 or NY99—were used ) ( S3 File ) . Apart from the EURL , no other NRL used USUV , TBEV or JEV isolates to evidence antibodies against related flaviviruses . All the participants achieved specific results with no false positives for true negative sera or for sera from horses infected with related flaviviruses . A lack of sensitivity was underlined in 3/8 NRLs in 2010 and in 2/7 NRLS in 2013 , with 2 participants missing at least two positive samples in 2010 ( S14-S15 duplicates ) and 2013 ( S3 and S7 ) . These WNV ILPTs were also used to evaluate the characteristics and performance of the different serological assays used by participating laboratories in terms of early detection of anti-WNV antibodies , sensitivity and specificity ( observation of cross-reactive reactions with sera from horses infected with related flaviviruses ) . Because fewer commercial kits for the detection of anti-WNV IgG and IgM were used in 2010 than in 2013 , the results generated by competitive ELISAs , MAC ELISAs and VNTs were compared only for the 2013 ILPT . The assays used by 35 participating laboratories ( 21 NRLs and 9 other EU laboratories , 3 kit manufacturers and 2 Mediterranean and Balkans reference laboratories ) are shown in Table 1 and analysed hereafter . The comparative results obtained by the analysis of ILPT results are purely indicative and do not entirely reflect the performances of the methods .
The NRL network guarantees WNV surveillance and warning of the emergence or re-emergence of the disease . The number of laboratories has increased between the two ILPTs and indeed has been continuously increasing since 2008 . Hungary , Latvia and Bulgaria were integrated as new NRL participants in addition to the 2013 network in an ILPT held in late 2016 . A comparison of the two ILPTs organised in 2010 and 2013 demonstrated an improvement in the performances of WNV analytical assays and processes in 2013 . Moreover , at least 13 NRLs are able to diagnose acute WNV infection and many of them ( Greece , Austria , Italy , Croatia and France ) were involved in the detection of recent WNV outbreaks [29] . In 2013 , the WNV NRL network had gathered together most of the EU countries facing WNV outbreaks and EU countries with the highest equine population ( Romania , Benelux countries , the United Kingdom , Germany and France ) [30 , 31] . Other European and international initiatives ( H2020 MediLabSecure project , IAEA and FAO training initiatives ) have fostered enhanced , more widespread WNV surveillance in other European countries and regions ( such as the Balkans and the Mediterranean basin area ) . Competitive ELISA kits are widely used to detect IgG antibodies against WNV . They are well established in Europe , with 17 NRLs and 20 NRLs using competitive WNV kits in 2010 and 2013 respectively . Thanks to the development of new commercial MAC ELISAs , ELISAs ensuring the early detection of infected horses have been rapidly implemented in reference laboratories between 2010 ( 3 NRLs using MAC ELISA kits ) and 2013 ( 13 NRLs using MAC ELISA kits ) . The use of kits has allowed rapid standardisation of the 2 methods in NRLs , with more than 95% ( 20/21 ) and 100% ( 13/13 ) of relevant results in competitive and MAC ELISAs respectively in 2013 . An improvement in the sensitivity of competitive ELISA kits used by the NRL network was evidenced between the 2 ILPTs , with fewer false negative results generated during the 2013 ILPT . The specificity of competitive and MAC ELISAs on sera negative to flavivirus antibodies was 100% for the 2 ILPTs . In European laboratories seeking to detect human WNV cases , only a few perform the WNV VNT [32] . An identical profile was observed in veterinary laboratories where a low but stable number of NRLs performed VNTs . The need for a BSL-3 facility and technique constraints can account for such an observation . Moreover , NRL results underlined the difficulties in standardising WNV VNTs , with a lack of sensitivity in 3/8 NRLs in 2010 and again in 2/7 NRLs in 2013 . Indirect or competitive WNV ELISAs are commonly used for screening purposes because of their reputed higher sensitivity [6] . The analytical sensitivity of these methods was evaluated in our ILPT with different WNV-positive sera originating from lineage 1 or 2 WNV-infected ponies and containing different levels of antibodies . Their sensitivity was shown in most cases to be higher than that of VNT , depending on the method [24] implemented by the NRL and the VNT’s own analytical sensitivity . Nevertheless , the false positives generated through cross-reactions with antibodies directed against heterologous flaviviruses confound the interpretation of ELISA tests . Specificity issues and the detection of infections caused by close flaviviruses have been systematically described in international quality control assessments for the human serological detection of WNV or Dengue virus ( DENV ) infections [32–34] . For example , high WNV IgG false positives were reported for samples containing anti-DENV antibodies [32] . Specifically , the ILPT organised in this study in 2013 showed that an infection by closely related flaviviruses belonging to the JEV serocomplex ( i . e . USUV and JEV ) systematically generated false positive reactions with the two commercially available competitive ELISAs , while kit specificity was more variable for flaviviruses belonging to another serocomplex ( i . e . TBEV ) . Interestingly , the last ILPT organised in 2016 showed an improvement in the diagnostic specificity of the Ingezim WNV Compaq kit with USUV and JEV positive sera generating negative results with this kit ( 6/6 assays ) [35] . Moreover , some ELISA protocols using alternative antigens ( NS1 ) or revelation systems proved to offer higher specificity than commercial ELISA kits . Indirect or competitive ELISAs are useful for the rapid screening of animal samples , keeping in mind that positive results are not indicative of WNV infection . In the event of positive results , the diagnosis of an acute WNV infection in horses should be confirmed by IgM detection in serum or cerebrospinal fluid . MAC ELISAs are used to detect acute infections in humans and horses . Nevertheless , their usefulness in determining acute infections in humans is questionable because IgM response is sometimes detectable up to 1 year after initial virus exposure in plasma/serum [36] . In horses , IgM antibodies are short-lived [37] . In equids , WNV IgM antibodies are secreted as early as 8 dpi [27] and can be detected up to 70–90 dpi with currently developed IgM ELISAs [37] . With 42% of participants generating negative results on a WNV lineage 1 serum sampled at 35 dpi ( S4 ) , the window during which anti-WNV IgM antibodies were detected in the serum by MAC-ELISA seemed more limited than described in the literature . However , precaution should be taken when interpreting these data because the samples originated from experimental infections associated with asymptomatic or very mild infection . Moreover , the variable sensitivity of MAC ELISA commercial kits can also account for such a result , since the Ingezim WNV IgM kit generated a positive result on 35 dpi for the WNV lineage 1 sample ( S4 ) as well as for the lineage 2 sample diluted to 1/48 ( S9 ) . These data should be confirmed with additional samples and assays calibrated for the evaluation of kit performances . Finally MAC ELISAs appeared to be more specific than indirect IgG or competitive ELISAs [38] . Samples from USUV- , TBEV- and JEV-infected ponies were found by MAC ELISA to be negative on 35 dpi , while clearly positive by competitive ELISA kits ( Fig 2 ) . IgM antibodies against related flaviviruses also remained undetected in non-diluted sera sampled on 8 dpi and 21 dpi ( S1 Fig ) . The short duration of anti-flavivirus IgM response in horses , associated with the high specificity of MAC ELISAs indicate that MAC ELISAs can be used to confirm recent WNV infection in horses [38] . Nevertheless , the production of IgM antibodies following primary vaccine administrations is questionable . Indeed , some papers in the literature consider IgM reactivity towards an inactivated vaccine as minimal [39] while others support low but detectable IgM antibody production [40] . Vaccination history should always be considered prior to the interpretation of MAC ELISAs . VNT is the gold standard serological tool for confirming WNV diagnosis [21] . In the WNV 2013 ILPT , WNV VNTs demonstrated variable sensitivity depending on the laboratory , but a high level of specificity with no false positives obtained with sera from horses infected with related flaviviruses . Nevertheless , cross-neutralisation by antibodies directed against viruses within the same serocomplex can still be observed in the field [41] despite no such evidence during the ILPT . Taking into account these conclusions , a decisional algorithm for serological flavivirus diagnosis is proposed in Fig 5 . All these remarks plead in favour of the development and implementation of new technologies to provide alternatives to classical methods for serological flavivirus diagnosis [42] . Many efforts have been made to develop ELISAs with enhanced specificity by using monoclonal antibodies targeting specific epitopes on NS1 or the E glycoprotein of flaviviruses [43 , 44] . The ELISA targeting epitopes on NS1 appeared more specific for related flaviviruses in the 2013 ILPT , as also evidenced in the studies by Kitai et al . for differentiating JEV from WNV infection [45] . Moreover , this ELISA was able to distinguish infected horses from horses vaccinated with recombinant prM-E WNV vaccines ( not containing the NS1 antigen ) [46 , 47] . Such a DIVA test could be useful in areas where horses have a high vaccination coverage . The WNV E glycoprotein is folded into three structural domains: DI , DII and DIII . While DII is highly conserved among flaviviruses , DI and DIII contain virus-specific epitopes [48 , 49] . Recombinant envelope proteins bearing mutations in the conserved DII part [50] or domain III of the E glycoprotein [51] have been used as specific antigens in ELISAs [44 , 52 , 53] . The development of multiplex serological tools is also another promising approach . Microsphere-based immunoassays using domain III antigens [24] or protein microarrays with recombinant NS1 proteins of flaviviruses [40] have been shown to offer advantageous alternatives to VNTs because they can test different flaviviruses in parallel . | The European network of National Reference Laboratories ( NRLs ) for equine diseases guarantees West Nile virus ( WNV ) surveillance and warning of the emergence of the disease . The WNV NRL network has gathered together most of the European countries facing WNV outbreaks . In this context , two inter-laboratory proficiency tests ( ILPTs ) were designed in 2010 and 2013 to evaluate the network’ and methods’ performances in detecting WNV infection through serology . A comparison of these two ILPTs emphasised a substantial improvement in the analytical performance of the WNV antibody detection tools over the years within the European NRLs network . Nevertheless the serological cross-reactions among related flaviviruses , such as the Japanese encephalitis , Usutu or tick-borne encephalitis viruses through IgG detection , associated with the Virus Neutralisation Tests’ ( VNT ) lower sensitivity , long duration and need for Biosafety Level 3 ( BSL-3 ) facilities are major concerns related to indirect WNV diagnosis . All these remarks plead in favour of the development and implementation of new technologies to provide alternatives to classical methods for serological flavivirus diagnosis . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"medicine",
"and",
"health",
"sciences",
"enzyme-linked",
"immunoassays",
"immune",
"physiology",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"immunology",
"geographical",
"locations",
"microbiology",
"vertebrates",
"animals",
"mammals",
"viruses",
"rna",
"... | 2017 | Improved reliability of serological tools for the diagnosis of West Nile fever in horses within Europe |
ATM and ATR are two redundant checkpoint kinases essential for the stable maintenance of telomeres in eukaryotes . Previous studies have established that MRN ( Mre11-Rad50-Nbs1 ) and ATRIP ( ATR Interacting Protein ) interact with ATM and ATR , respectively , and recruit their partner kinases to sites of DNA damage . Here , we investigated how Tel1ATM and Rad3ATR recruitment to telomeres is regulated in fission yeast . Quantitative chromatin immunoprecipitation ( ChIP ) assays unexpectedly revealed that the MRN complex could also contribute to the recruitment of Tel1ATM to telomeres independently of the previously established Nbs1 C-terminal Tel1ATM interaction domain . Recruitment of Tel1ATM to telomeres in nbs1-c60Δ cells , which lack the C-terminal 60 amino acid Tel1ATM interaction domain of Nbs1 , was dependent on Rad3ATR-Rad26ATRIP , but the kinase domain of Rad3ATR was dispensable . Thus , our results establish that the Rad3ATR-Rad26ATRIP complex contributes to the recruitment of Tel1ATM independently of Rad3ATR kinase activity , by a mechanism redundant with the Tel1ATM interaction domain of Nbs1 . Furthermore , we found that the N-terminus of Nbs1 contributes to the recruitment of Rad3ATR-Rad26ATRIP to telomeres . In response to replication stress , mammalian ATR–ATRIP also contributes to ATM activation by a mechanism that is dependent on the MRN complex but independent of the C-terminal ATM interaction domain of Nbs1 . Since telomere protection and DNA damage response mechanisms are very well conserved between fission yeast and mammalian cells , mammalian ATR–ATRIP may also contribute to the recruitment of ATM to telomeres and to sites of DNA damage independently of ATR kinase activity .
ATM ( Ataxia Telangiectasia Mutated ) and ATR ( ATM and Rad3-related ) , members of the phosphoinositol-3-kinase like kinase ( PIKK ) family , are central players in coordinating cellular responses to various forms of DNA damage , such as DNA double-stranded breaks ( DSBs ) and problems that are encountered by DNA replication forks , in eukaryotic cells [1] , [2] . ATM and ATR both preferentially recognize and phosphorylate Serine ( S ) or Threonine ( T ) amino acid residues followed by Glutamate ( Q ) , and over 900 sites in more than 700 proteins have been identified as potential phosphorylation sites for these two kinases in mammalian cells [3] . Previous studies have identified the Mre11-Rad50-Nbs1 ( MRN ) DNA repair complex as a key player in the activation of ATM kinase in response to DSBs [1] , [4] . The MRN complex interacts with ATM through an evolutionarily conserved C-terminal motif in its Nbs1 subunit , and this interaction is critical for recruitment of ATM to DSBs and phosphorylation of downstream targets by ATM [5] , [6] . Likewise , ATRIP ( ATR-Interacting Protein ) interacts with ATR through its evolutionarily conserved extreme C-terminal motif and promotes recruitment of ATR to sites of DNA damage [6] . RPA ( Replication Protein A ) -coated single-stranded DNA ( ssDNA ) serves as a platform for recruitment of the ATR-ATRIP complex , where phosphorylation of various downstream targets can take place [7] . Besides its role in activation of ATM , the MRN complex has also been shown to contribute to ATR signaling in mammalian cells [8] . Initial studies have suggested that ATM is particularly important for recognition of DSBs , while ATR is more important for recognition of replication stress and ultraviolet radiation ( UV ) -induced DNA damage [1] , [2] . However , recent studies have uncovered an intimate crosstalk between the ATM and ATR signaling pathways in response to DNA damage . For example , ATM-MRN can act upstream of ATR-ATRIP in cellular responses to DSBs by promoting recruitment of ATR-ATRIP through its contribution to generate RPA-coated ssDNA at DSBs [9]–[11] . Conversely , ATR contributes to the activation of ATM in response to DNA replication stress by converting inactive ATM dimers into active monomers through direct phosphorylation of ATM [12] . However , it is currently unknown if ATR-ATRIP may also contribute to the recruitment of ATM to stalled replication forks . In contrast to DSB-induced ATM activation , the C-terminal ATM interaction domain of Nbs1 is not required for ATR-dependent activation of ATM in response to DNA replication stress; however , the Nbs1 N-terminus is essential for ATR-dependent activation of ATM [12] . ATM-MRN and ATR-ATRIP are also redundantly required for the maintenance of telomeres , stable DSBs at ends of chromosomes , in a wide variety of eukaryotic species [13] ( Figure 1A ) . Studies in yeasts and mammalian cells have shown that ATM-MRN and ATR-ATRIP are recruited to functional telomeres during S/G2 phases of the cell cycle [14]–[16] . In budding yeast , Tel1ATM and Mec1ATR have been shown to phosphorylate several Serine residues within the telomerase-recruitment domain of the telomere capping protein Cdc13 , and these phosphorylation events have been proposed to promote efficient telomerase recruitment to telomeres [14] , [17] . In fission yeast , we have recently shown that Tel1ATM and Rad3ATR redundantly promote interaction between the Pot1 telomere capping complex ( consisting of Pot1 , Tpz1 , Poz1 and Ccq1 subunits ) and telomerase , and thereby help to recruit telomerase to telomeres [18] . On the other hand , the telomeric GT-rich repeat DNA binding factors TRF2 and POT1 , essential for protection of telomeres against degradation and recombination , play critical roles in attenuating DNA damage checkpoint activation mediated by ATM and ATR in mammalian cells [19]–[21] . Therefore , it has been suggested that telomeres transiently become de-protected during S- and G2-phases , and can thus be recognized as DSBs by ATM/ATR to allow the timely recruitment of telomerase [22] . On the other hand , we have recently found that the arrival of lagging strand DNA polymerases ( α and δ ) to telomeres is significantly delayed compared to the arrival of the leading strand DNA polymerase ε in fission yeast , and that significant quantities of RPA and Rad26ATRIP transiently accumulate at replicating telomeres [15] . These observations thus raised the possibility that replicating telomeres may be primarily recognized as unusual/stressed replication forks by Rad3ATR-Rad26ATRIP rather than being recognized as DSBs in fission yeast [15] . In mammalian cells , recruitment of ATR and MRN to telomeres precedes recruitment of ATM to telomeres [16] , and replication of lagging-strand telomeres also appears to be significantly delayed compared to leading-strand telomeres [23] . Thus , replicating mammalian telomeres might also accumulate unusually high levels of RPA on the lagging strand and activate ATR , and ATR might subsequently contribute to the activation of ATM signaling at telomeres to promote stable telomere maintenance . Here , we investigated inter-dependencies among Tel1ATM , Nbs1 , and Rad3ATR-Rad26ATRIP for their recruitment to telomeres in fission yeast . Our studies uncovered a surprising kinase-independent role of the Rad3ATR-Rad26ATRIP complex , which is redundant with the C-terminal Tel1ATM-interacting domain of Nbs1 , for the recruitment of Tel1ATM to telomeres . We also demonstrate that the N-terminal domain of Nbs1 contributes to Rad3ATR-Rad26ATRIP function by promoting recruitment of Rad3ATR-Rad26ATRIP to telomeres . Thus , our findings provide mechanistic insights into how recruitment of Tel1ATM and Rad3ATR to telomeres is regulated , and define a novel molecular crosstalk between Rad3ATR and Tel1ATM in recognition of fission yeast telomeres , potentially resembling the ATR-ATM crosstalk in response to DNA replication stress in mammalian cells .
The C-terminus of Nbs1 is important for its interaction with Tel1ATM in fission yeast Schizosaccharomyces pombe [5] ( Figure 1B ) . This interaction is evolutionarily conserved , since the corresponding region of Nbs1 is also required for its interaction with ATM in Xenopus and humans [5] , [6] . Mutating or truncating the Nbs1 C-terminus in fission yeast cells lacking functional Rad3ATR eliminated DNA damage-induced histone H2A phosphorylation ( γ-H2A ) and foci formation of the checkpoint mediator protein Crb2/Rhp9 [5] , since Tel1ATM and Rad3ATR are redundantly required for DNA damage-induced histone H2A phosphorylation and Crb2 foci formation [24] . Tel1ATM-MRN ( Rad32Mre11-Rad50-Nbs1 ) and Rad3ATR-Rad26ATRIP pathways are also redundantly required for stable telomere maintenance , and simultaneous inactivation of these two pathways results in chromosome circularization in fission yeast [25] ( Figure 1A ) . Thus , it was previously hypothesized that disruption of Tel1ATM-Nbs1 interaction in fission yeast cells lacking functional Rad3ATR would lead to telomere dysfunction and chromosome circularization [5] . We tested this hypothesis by combining the most commonly used rad3 deletion allele ( rad3Δ::ura4+ , hereafter referred to as rad3-kdΔ ) with the Nbs1 C-terminal 60 amino acid deletion allele nbs1-c60Δ that disrupts Tel1ATM-Nbs1 interaction [5] , [26] ( Figure 1B and 1C ) . After extensive restreaking of fission yeast cells on agar plates , we examined the status of telomeres by separating NotI-digested chromosomal DNA by pulsed-field gel electrophoresis ( PFGE ) , and then performing Southern blot analysis with telomeric NotI fragment-specific probes C , I , L and M ( Figure 1D ) . In this assay , cells carrying circular chromosomes lose individual telomeric NotI fragments and show I+L and C+M bands , corresponding to the fused telomeric fragments from chromosomes I and II . We were surprised to find that the double mutant nbs1-c60Δ rad3-kdΔ cells stably maintained telomeres ( Figure 1D ) . By contrast , combining tel1Δ or nbs1Δ with rad3-kdΔ caused complete loss of telomeres and chromosome circularization , as previously shown [25] , [27] ( Figure 1D ) . Thus , while Nbs1 and Tel1ATM are both essential for telomere maintenance in rad3-kdΔ cells , the Nbs1-Tel1ATM interaction is dispensable for telomere maintenance in rad3-kdΔ cells . The mutant Nbs1-c60Δ protein retains its ability to interact with Rad32Mre11 , and both rad32Δ rad3-kdΔ and rad50Δ rad3-kdΔ cells are unable to maintain telomeres [5] , [27] . Thus , the N-terminus of Nbs1 is required for the MRN ( Rad32Mre11-Rad50-Nbs1 ) complex to fulfill its essential telomere maintenance function ( s ) in rad3-kdΔ cells . The rad3-kdΔ allele does not completely delete the rad3+ coding region and only removes a C-terminal portion of the Rad3ATR protein including its kinase domain [26] ( Figure 1C ) . We found that the truncated Rad3ATR protein ( Rad3-kdΔATR ) expressed stably , and retained its ability to interact with its regulatory subunit Rad26ATRIP in co-immunoprecipitation ( co-IP ) experiments and yeast 2-hybrid assays ( Figure 1E and 1F ) . Similarly , the N-terminal regions of mammalian ATR and budding yeast Mec1 , outside of their C-terminal kinase domains , interact with their respective regulatory subunits ATRIP and Ddc2 [28] , [29] . Furthermore , when the fission yeast rad3+ gene was originally cloned , a truncated Rad3ATR protein lacking its kinase domain was found to fully suppress UV , IR and HU hypersensitivities and a G2-checkpoint defect of rad3-136 cells [30] . In addition , over-expression of the N-terminal fragments of Xenopus ATR or budding yeast Mec1ATR lacking the kinase domain has been shown to partially suppress the DNA damage sensitivity of mec1-1 budding yeast cells by activating the spindle checkpoint [31] . Therefore , we next examined if a complete deletion allele of rad3+ ( rad3Δ::LEU2 , hereafter referred to as rad3Δ , deletes genomic DNA corresponding to amino acids 1-2363 of Rad3ATR [25] ) would show a different genetic interaction with nbs1-c60Δ with respect to DNA damage sensitivity and telomere maintenance , even though we saw no difference between rad3-kdΔ and rad3Δ cells for their average telomere lengths in nbs1+ background ( Figure S1A ) . While we detected no differences in sensitivity towards various DNA damaging agents between rad3-kdΔ and rad3Δ alleles in nbs1+ , nbs1-c60Δ or nbs1Δ backgrounds ( Figure S1 ) , we found that rad3Δ nbs1-c60Δ cells cannot maintain stable telomeres , unlike rad3-kdΔ nbs1-c60Δ cells ( Figure 1D ) . Thus , the N-terminal domain of Rad3ATR contributes to Tel1ATM-dependent telomere maintenance in nbs1-c60Δ cells . Additionally , we found that rad26Δ nbs1-c60Δ cells cannot maintain telomeres ( Figure 1D ) , establishing that Rad26ATRIP is essential for telomere maintenance in nbs1-c60Δ cells . Thus , these results indicated that the Rad3-kdΔATR-Rad26ATRIP complex has a kinase-independent function for telomere maintenance in cells lacking the C-terminal Tel1ATM interaction domain of Nbs1 . The apparent discrepancy in absolute requirement for the Rad3ATR kinase domain for DNA damage sensitivity versus telomere maintenance is plausible , since previous studies have established that the downstream checkpoint kinases Chk1 and Cds1CHK2 are not involved in telomere maintenance in fission yeast , while these kinases play critical roles in cell survival in response to DNA damage [25] , [32] . To gain insights into how the Rad3-kdΔATR-Rad26ATRIP complex contributes to telomere maintenance in nbs1-c60Δ cells , we next examined the association of the Rad3ATR-Rad26ATRIP complex with fission yeast telomeres by quantitative chromatin immunoprecipitation ( ChIP ) assays . While a significant amount of telomeric DNA was found to associate with Rad26ATRIP in rad3+ and rad3-kdΔ cells , the amount of telomeric DNA precipitated by Rad26ATRIP in rad3Δ cells was reduced to background levels ( Figure 2A ) . Conversely , both wild-type Rad3ATR and Rad3-kdΔATR proteins associate with telomeric DNA in a Rad26-dependent manner , although the precipitation efficiency for Rad3-kdΔATR was reduced compared to wild-type Rad3ATR ( Figure 2B ) . Taken together , these results are consistent with the notion that Rad26ATRIP is essential for recruitment of Rad3ATR to telomeres and that the N-terminal domain of Rad3ATR contributes to the recruitment of Rad26ATRIP to telomeres . We also examined the recruitment of wild-type Nbs1 and Nbs1-c60Δ to telomeres in rad3+ , rad3-kdΔ , and rad3Δ background by ChIP assays ( Figure 2C ) . Since we were able to still detect a robust association of Nbs1-c60Δ with telomeres even in rad3Δ cells , we could exclude the possibility that the Rad3-kdΔATR-Rad26ATRIP complex contributes to telomere maintenance by promoting recruitment of Nbs1-c60Δ to telomeres . Since expression of Rad3-kdΔATR allowed cells to maintain stable telomeres in the absence of Nbs1-Tel1ATM interaction but not in tel1Δ and nbs1Δ cells ( Figure 1D ) , we hypothesized that Nbs1-c60Δ protein and the Rad3-kdΔATR-Rad26ATRIP complex may function collaboratively to recruit Tel1ATM to telomeres and maintain telomeres . Indeed , we detected robust Tel1ATM recruitment to telomeres in nbs1-c60Δ rad3-kdΔ cells , but failed to detect Tel1ATM at telomeres in nbs1-c60Δ rad3Δ , nbs1Δ rad3-kdΔ , and nbs1Δ rad3Δ cells ( Figure 2D ) . The ability to recruit Tel1ATM to telomeres in rad3 mutant backgrounds ( either rad3-kdΔ or rad3Δ ) correlated perfectly with the telomere maintenance phenotypes of these mutant cells ( Figure 1D and Figure 2D ) . Thus , our ChIP data establish that the N-terminal domain of Nbs1 and the telomere-bound Rad3-kdΔATR-Rad26ATRIP complex must both be present , in order for Tel1ATM to be recruited independently of the C-terminal Tel1ATM-interaction domain of Nbs1 . It should be noted that for those mutant strains that are defective in telomere maintenance and ultimately circularize their chromosomes , ChIP assays were performed with early generation cells prior to chromosome circularization by utilizing our recently described Rad3 plasmid loss system [18] . This system allows us to harvest cells immediately after the loss of Rad3ATR , when the average telomere length is comparable or slightly longer than in rad3Δ or rad3-kdΔ cells . Furthermore , by monitoring real-time PCR amplification cycle numbers for all ChIP input samples , we have ensured that mutant cells utilized in our experiments have not yet circularized their chromosomes , since the PCR primer-annealing sites would be completely lost upon chromosome circularization , resulting in a significant delay in PCR amplification . In contrast to nbs1-c60Δ cells , nbs1+ cells were able to recruit Tel1ATM to telomeres in both rad3-kdΔ and rad3Δ backgrounds ( Figure 2D ) . Thus , our ChIP data also support the notion that the C-terminal Tel1ATM interaction domain of Nbs1 and the Rad3-kdΔATR-Rad26ATRIP complex represent two pathways that are redundantly required for recruitment of Tel1ATM to telomeres . Since we detect Tel1ATM at telomeres only in cells carrying very short telomere-repeats ( ∼180 bp in rad3-kdΔ and rad3Δ cells ) , but not in cells with wild-type telomere-repeat length ( ∼280 bp in rad3+ cells ) ( Figure 2D , Figure S1A ) , fission yeast Tel1ATM may be preferentially recruited to short telomeres . Alternatively , active Rad3ATR kinase might prevent association of Tel1ATM with telomeres . In any case , our results establish that the C-terminal 60 amino acids of fission yeast Nbs1 are not essential for the recruitment of Tel1ATM to critically short telomeres , due to a redundant non-kinase contribution of Rad3ATR-Rad26ATRIP to Tel1ATM recruitment . In order to better understand how the N-terminal domain of Nbs1 contributes to the Rad3-kdΔATR-dependent maintenance of telomeres in nbs1-c60Δ cells , we next examined the recruitment of Rad3ATR or Rad3-kdΔATR to telomeres in nbs1+ , nbs1-c60Δ or nbs1Δ backgrounds by ChIP assays . As shown in Figure 3A , we found that nbs1Δ cells recruit significantly less wild-type Rad3 or Rad3-kdΔATR proteins to telomeres compared to nbs1+ or nbs1-c60Δ cells , suggesting that Nbs1-c60Δ protein contributes to the efficient recruitment of the Rad3ATR-Rad26ATRIP complex to telomeres . Accordingly , the loss of the Rad3-kdΔATR-Rad26ATRIP complex from telomeres in nbs1Δ cells might explain why rad3-kdΔ nbs1Δ cells fail to recruit Tel1ATM to telomeres and circularize chromosomes ( Figure 1D , Figure 2D , and Figure 3A ) . It should be noted that a substantial amount of wild-type Rad3ATR could still be recruited to telomeres in nbs1Δ cells . In fact , since nbs1Δ , rad50Δ , rad32Δmre11 and tel1Δ cells all maintain essentially wild-type length or only slightly shorter telomeres in fission yeast [27] , [33] , it appears that the MRN complex and Tel1ATM do not make significant contributions to telomere maintenance as long as cells express the wild-type Rad3ATR-Rad26ATRIP complex . The ATR-ATRIP complex preferentially binds to RPA-coated ssDNA , and the MRN complex promotes formation of 3′ ssDNA at DSBs and telomeres [10] , [34]–[36] . Thus , we considered the possibility that the fission yeast MRN complex , independently of its Nbs1 C-terminal Tel1ATM interaction domain , might promote recruitment of Rad3ATR-Rad26ATRIP to telomeres by promoting accumulation of RPA-coated ssDNA at telomeres . However , we found comparable level of RPA ( Rad11 ) associated with telomeric DNA in rad3-kdΔ nbs1-c60Δ and rad3-kdΔ nbs1Δ cells ( Figure 3B ) . We also considered the possibility that the N-terminus of Nbs1 might contribute to the recruitment of Rad3ATR-Rad26ATRIP to telomeres by promoting association between the MRN complex and the Rad3ATR-Rad26ATRIP complex , since it was previously shown that mammalian MRN interacts with ATR-ATRIP [37] . However , we could not detect a direct interaction between Nbs1 and Rad3ATR or between Nbs1 and Rad26ATRIP by yeast 2-hybrid assays , and failed to detect an interaction between Nbs1 and Rad26ATRIP by co-IP experiments ( data not shown ) . Thus , further investigations are necessary to fully understand how the N-terminal domain of Nbs1 contributes to efficient recruitment of the Rad3ATR-Rad26ATRIP complex to telomeres . Over-expression of Tel1ATM results in MRX ( Mre11-Rad50-Xrs2 ) complex-dependent elongation of telomeres in budding yeast [14] , [38] , [39] , consistent with the notion that recruitment of Tel1ATM to telomeres represents the critical rate limiting step in regulating telomere length in budding yeast . By contrast , over-expression of Tel1ATM was unable to increase telomere length in wild-type fission yeast cells ( Figure 4A ) , even though over-expressed Tel1ATM can be detected at telomeres by ChIP assays ( Figure 4D ) . However , we did observe a partial suppression of telomere shortening in rad3-kdΔ and rad3Δ cells upon Tel1ATM over-expression ( Figure 4A ) , suggesting that the level of Tel1ATM recruitment to short telomeres becomes a critical limiting factor in telomere length determination in the absence of the functional Rad3ATR-Rad26ATRIP complex . Interestingly , we also found that the Nbs1 C-terminal Tel1ATM interaction domain is essential for over-expressed Tel1ATM to associate with telomeres in rad3+ cells , but dispensable in rad3-kdΔ and rad3Δ cells ( Figure 4D ) . Thus , it appears that the presence of the kinase active Rad3ATR-Rad26ATRIP complex can prevent recruitment of over-expressed Tel1ATM to telomeres if cells are missing the C-terminal Tel1ATM interaction domain of Nbs1 . We reasoned that over-expression of Tel1ATM might be able to suppress chromosome circularization caused by simultaneous mutations in Nbs1 and Rad3ATR-Rad26ATRIP , if the essential telomere maintenance function , redundantly provided by the Nbs1 C-terminal Tel1ATM interaction domain and the kinase-inactive Rad3ATR-Rad26ATRIP complex , were to promote efficient recruitment of Tel1ATM to telomeres . Indeed , over-expression of Tel1ATM allowed recruitment of Tel1ATM in nbs1-c60Δ rad3Δ cells ( Figure 4D ) and suppressed chromosome circularization in nbs1-c60Δ rad3Δ and nbs1-c60Δ rad26Δ cells ( Figure 4B and 4C ) . However , over-expression of Tel1ATM was unable to suppress chromosome circularization in nbs1Δ rad3-kdΔ , nbs1Δ rad3Δ and nbs1Δ rad26Δ cells , indicating that the expression of Nbs1 ( and likely other subunits of the MRN complex ) is essential for telomere maintenance in the absence of kinase-active Rad3ATR-Rad26ATRIP even when Tel1ATM is over-expressed . Taken together , our results indicate that the Nbs1 C-terminal Tel1ATM interaction domain and the kinase-independent function of the Rad3-kdΔATR-Rad26ATRIP complex redundantly contribute to efficient recruitment of Tel1ATM to telomeres , when Tel1ATM is expressed at endogenous level . Our results further demonstrate that when Tel1ATM is over-expressed , the MRN complex can contribute to telomere maintenance even in the absence of both the C-terminal Tel1ATM interaction domain of Nbs1 and the Rad3ATR-Rad26ATRIP complex . Our ChIP data indicated that both the Rad26ATRIP regulatory subunit and the N-terminal domain of Nbs1 play critical roles in promoting recruitment of Rad3ATR to telomeres ( Figure 2B , Figure 3A ) . Therefore , we next investigated if over-expression of Rad3ATR might be able to bypass the requirement for Rad26ATR and/or Nbs1 in telomere maintenance . Based on ChIP analysis , over-expression of Rad3ATR bypassed the requirement for Rad26ATRIP in recruitment of Rad3ATR to telomeres ( Figure 5C ) . However , over-expression of Rad3ATR was not able to suppress telomere shortening observed in rad26Δ cells ( Figure 5A ) . Thus , in the presence of wild-type Nbs1 ( and thus intact Nbs1-Tel1ATM interaction ) , forced recruitment of Rad3ATR to telomeres by over-expression is not sufficient to elongate telomeres in rad26Δ cells . On the other hand , we found that Rad3ATR over-expression completely suppressed chromosome circularization in rad26Δ nbs1-c60Δ cells ( Figure 5B ) , consistent with the notion that the essential telomere maintenance function contributed by Rad26ATRIP in nbs1-c60Δ is to promote recruitment of Rad3ATR to telomeres . Even when Rad3ATR is over-expressed , the N-terminus of Nbs1 still appears to contribute to telomere maintenance in rad26Δ cells , since suppression of chromosome circularization by over-expression of Rad3ATR was much more complete in rad26Δ nbs1-c60Δ cells than in rad26Δ nbs1Δ cells ( Figure 5B ) . Since we observed a mixed telomere phenotype among survivor cells after extensive restreaking of nbs1Δ rad26Δ cells over-expressing Rad3ATR on agar plates , we concluded that over-expression of Rad3ATR can partially bypass the essential telomere function of Nbs1 in the absence of Rad26ATRIP ( Figure 5B ) . The observed partial suppression of chromosome circularization in nbs1Δ rad26Δ by over-expression of Rad3ATR is consistent with the notion that the N-terminal domain of Nbs1 contributes to telomere maintenance by promoting recruitment of Rad3ATR to telomeres . By contrast , Rad3ATR over-expression was unable to suppress chromosome circularization observed in rad26Δ tel1Δ cells ( Figure 5B ) . Thus , it appears that over-expression of Rad3ATR allows the Tel1ATM-dependent mechanism to maintain telomeres in rad26Δ nbs1-c60Δ or rad26Δ nbs1Δ cells . One possible mechanism by which Rad3ATR over-expression might promote Tel1ATM-dependent telomere maintenance in rad26Δ nbs1-c60Δ or rad26Δ nbs1Δ cells is to promote recruitment of Tel1ATM to telomeres . Therefore , we examined changes in Tel1ATM recruitment to telomeres in nbs1-c60Δ rad26Δ cells with or without over-expression of Rad3ATR by ChIP analyses . However , we were unable to detect Rad3ATR over-expression dependent recruitment of Tel1ATM to telomeres ( Figure 5D ) . Thus , further investigations are necessary to fully understand how over-expression of Rad3ATR contributes to the Tel1ATM-dependent suppression of chromosome circularization in fission yeast cells simultaneously lacking functional Rad26ATRIP and Nbs1 .
In the current study , we investigated how two PIKK-containing complexes , Tel1ATM-MRN and Rad3ATR-Rad26ATRIP , contribute to telomere length regulation in fission yeast ( summarized in Figure 6 ) . We have demonstrated that fission yeast Tel1ATM can be recruited to telomeres by two alternative and redundant mechanisms , which are either dependent or independent of the Nbs1 C-terminal Tel1ATM interaction domain ( Figure 6A and 6B ) . Our analyses indicated that the Nbs1 C-terminus dependent mode of Tel1ATM recruitment to telomeres does not require the Rad3ATR-Rad26ATRIP complex , and in fact , it might be inhibited by the presence of wild-type Rad3ATR-Rad26ATRIP ( Figure 2D ) . On the other hand , the Nbs1 C-terminus independent mode of Tel1ATM recruitment requires the presence of telomere bound kinase-inactive Rad3-kdΔATR-Rad26ATRIP complex ( Figure 2A , 2B , and 2D ) . Since the N-terminal domain of Nbs1 is essential for recruitment of the Rad3-kdΔATR-Rad26ATRIP complex to telomeres ( Figure 3A ) , all our results are consistent with the notion that association of Rad3-kdΔATR-Rad26ATRIP to telomeres is the crucial determinant that allows recruitment of Tel1ATM to telomeres in the absence of the Nbs1 C-terminal Tel1ATM interaction domain . The notion that the C-terminus of Nbs1 and the Rad3-kdΔATR-Rad26ATRIP complex redundantly contribute to the recruitment of Tel1ATM is also supported by our finding that over-expression of Tel1ATM can entirely bypass the requirement for Rad3ATR-Rad26ATRIP for telomere maintenance and Tel1ATM recruitment to telomeres in nbs1-c60Δ cells ( Figure 4 , Figure 6C ) . Moreover , the finding that over-expression of Rad3ATR was able to at least partially suppress the loss of Nbs1 protein ( Figure 5B , Figure 6D ) provided further support for the notion that the N-terminal domain of Nbs1 also contributes to the recruitment of Rad3ATR to telomeres . Thus , our findings are generally consistent with the model that at least one PIKK activity needs to be localized at telomeres , in order to stably maintain telomeres in fission yeast , although we have not yet tested if Tel1ATM kinase activity is indeed required for telomere maintenance in nbs1-c60Δ rad3-kdΔ cells . It is currently unknown which telomere-associated protein ( s ) represent critical substrate ( s ) of Tel1/Rad3 that are essential for telomere maintenance . However , we were able to demonstrate recently that Tel1ATM and Rad3ATR are redundantly required to prevent accumulation of homologous recombination DNA repair factors and RPA at telomeres , to promote efficient telomere recruitment of Tpz1 and Ccq1 ( subunits of the Pot1 telomere capping complex ) , and to recruit telomerase to telomeres [18] . Our results also indicate that fission yeast Tel1ATM is recruited preferentially to short telomeres ( Figure 2D ) . Due to the kinase-independent function of the Rad3ATR-Rad26ATRIP complex , the Nbs1 C-terminal 60 amino acid Tel1ATM interaction domain is dispensable for recruitment of Tel1ATM to short telomeres . Tel1ATM is also recruited preferentially to short telomeres in budding yeast [14] , [38] , [40] . Based on studies utilizing xrs2-664 mutant cells , which express the truncated Xrs2Nbs1 protein lacking the C-terminal 190 amino acids ( full length 884 amino acids ) , it was suggested that the Xrs2Nbs1 C-terminal Tel1ATM interaction domain is essential for recruitment of Tel1ATM to critically short telomeres [14] . However , a recent study has shown that , while deleting as little as 20 amino acids from C-terminus of Xrs2Nbs1 is sufficient to disrupt the interaction between Tel1ATM and Xrs2Nbs1 , xrs2 mutant cells lacking the C-terminal 20 amino acids maintain significantly longer telomeres than xrs2-664 or xrs2Δ cells [41] . Therefore , budding yeast Xrs2Nbs1 also appears to contribute to telomere maintenance independently of its C-terminal Tel1ATM interaction domain . Thus , it would be interesting to test if the N-terminal domain of Xrs2Nbs1 might also collaborate with the Mec1ATR-Ddc2ATRIP complex , perhaps independently of Mec1ATR kinase activity , in regulating Tel1ATM recruitment to telomeres in budding yeast . Our ChIP data indicate that very little or no Tel1ATM is recruited to telomeres in wild-type fission yeast cells . On the other hand , Rad3ATR-Rad26ATRIP is specifically recruited to replicating telomeres and appears to act as the primary sensor of transiently “open” telomeres during S-phase [15] . Therefore , it makes sense that rad3Δ or rad26Δ cause a more severe shortening of telomeres than tel1Δ in fission yeast [25] , [27] , [42] . Thus , Tel1ATM is likely to function as a back-up mechanism to extend critically short telomeres in fission yeast . In contrast , budding yeast mec1Δ cells show very little shortening of telomeres [43] , while tel1Δ or xrs2Δ cells carry extremely short telomeres [44] , [45] . It has also been reported that budding yeast Mec1ATR cannot be detected at telomeres by ChIP in wild-type cells [13] . However , simultaneous loss of Mec1ATR-Ddc2ATRIP and Tel1ATM-MRX pathways results in a severe defect in telomere maintenance [43] , [46] , much like in fission yeast cells . Thus , available data suggest that budding yeast telomeres are primarily regulated by Tel1ATM-MRX , and Mec1ATR-Ddc2ATRIP fulfills a back-up role for maintaining telomeres . There are intriguing similarities between our current findings and findings in mammalian cells , where ATR-ATRIP was found to act upstream of ATM . In response to replication stress or UV irradiation during S-phase , the Nbs1 C-terminal ATM interaction domain is dispensable , but the N-terminus of Nbs1 is essential , for ATR to activate ATM in mammalian cells [12] . However , it is currently unknown whether ATR-ATRIP might also contribute to the recruitment of ATM to sites of stalled or distressed DNA replication forks in a kinase-independent manner , besides the previously established kinase-dependent role of ATR in converting inactive ATM dimers into active monomers [12] . However , the similar requirement for the N-terminus , but not the C-terminus , of Nbs1 for the ATR-ATM crosstalk in response to DNA replication stress in mammalian cells and telomere maintenance in fission yeast might suggest that fission yeast telomeres may be recognized primarily as stressed and/or abnormal replication forks by ATM/ATR kinases . Studies in mammalian cells have also established the existence of an ATM-ATR crosstalk in response to DNA DSBs , where ATM and the MRN complex contribute to the recruitment of ATR-ATRIP to DSBs by promoting the generation of RPA-coated ssDNA at DSB sites [9]–[11] , [47] , [48] . The Nbs1 C-terminal ATM interaction domain is required to recruit ATM to DSBs induced by ionizing radiation ( IR ) and to promote ATM-dependent phosphorylation events [5] , [6] , [49] , and it is thus critical for the ATM-ATR crosstalk in response to DSBs . While the critical role of Tel1ATM in preferentially extending short telomeres in budding yeast has been well established [14] , [38] , mouse ATM was found not to be critical for extending short telomeres [50] . This difference might indicate that mouse ATR-ATRIP and ATM-MRN pathways work redundantly in extending critically short telomeres . Alternatively , since recruitment of human ATR to telomeres was found to occur earlier in S-phase than recruitment of ATM to telomeres [16] , mammalian ATR-ATRIP may function upstream of ATM in telomere length maintenance , and perhaps also possess kinase-independent functions in recruitment of ATM to telomeres . While further analyses are clearly needed to test these speculations , our current findings highlight a complex molecular crosstalk between ATM-MRN and ATR-ATRIP pathways in recognizing an “open” configuration of telomeres to allow their stable maintenance .
Fission yeast strains used in this study were constructed by standard techniques [51] and are listed in Table S1 . Primers listed in Table S2 were used to construct new strains . For nbs1-myc , nmt-HA-rad3 , myc-rad26 and YFP-rad26 , original strains were described previously [15] , [24] , [27] , [52] . For myc-rad3 , nbs1-c60Δ-myc , myc-tel1 and nmt-HA-tel1 , PCR-based methods [53] , [54] were used to generate tagged strains . For nbs1Δ::kanMX , nbs1-c60Δ::kanMX6 , rad3-kdΔ::ura4+ , rad3Δ::LEU2 and rad26Δ::ura4+ , original strains were described previously [5] , [25]–[27] , [55] . For nbs1Δ::natMX , PCR-based methods [53] , [54] were used to generate deletion strains . Budding yeast strains used in yeast two-hybrid assays are also listed in Table S1 . Plasmids used in this study are listed in Table S3 . Yeast two hybrid assays were performed by mating S . cerevisiae MATa strains harboring GAL4-DBD plasmids with MATα strains harboring GAL4-AD plasmids , as described in the MATCHMAKER system manual ( Clonetech ) . Positive two-hybrid interactions were identified by spotting mated cells onto SD-HTL plates . Sensitivities of fission yeast cells to IR , UV , HU ( hydroxyurea ) and CPT ( camptothecin ) were assayed as previously described [24] . For most strains , cell extracts were prepared in lysis buffer 1 [50mM Tris pH8 . 0 , 150mM NaCl , 10% glycerol , 5mM EDTA , 0 . 5% NP40 , 50mM NaF , 1mM DTT , 1mM PMSF , 1mM Na3VO4 , ‘Complete’ protease inhibitor cocktail ( Roche ) ] , either by glass bead disruption using FastPrep homogenizer ( MP Biomedical ) or by cryogenic disruption using MM301 Ball Mill ( Retsch ) . For strains expressing myc-Rad3 , myc-Rad3-kdΔ or myc-Tel1 , lysis buffer 2 [25mM Tris pH7 . 5 , 100mM NaCl , 10% glycerol , 15mM EDTA , 0 . 1% NP40 , 1% Triton , 15mM MgCl2 , 0 . 1mM NaF , 0 . 5mM DTT , 1mM PMSF , 1mM Na3VO4 , ‘Complete’ protease inhibitor cocktail] was used . For co-immunoprecipitation analyses , proteins were immunoprecipitated using either monoclonal anti-myc antibody ( 9B11 , Cell Signaling ) or monoclonal anti-HA antibody ( 12CA5 , Roche ) , and protein G Dynabeads ( Invitrogen ) or protein G sepharose beads ( GE ) respectively . Proteins in whole cell extract or from immunoprecipitations were analyzed by western blots using monoclonal anti-HA antibody ( 12CA5 ) , monoclonal anti-myc antibody ( 9B11 ) , monoclonal anti-FLAG antibody ( M2 , F1804 , Sigma ) or monoclonal anti-GFP antibody ( 7 . 1/B . 1 , Roche ) . Anti-Cdc2 antibody ( y100 . 4 , Abcam ) was used for loading control . Chromosomal DNA samples were prepared in agarose plugs , digested with NotI restriction enzyme , and fractionated in 1% agarose gels using the CHEF-DR III system ( Bio-Rad ) as previously described [25] . C , I , L , and M probes specific for telomeric NotI fragments were prepared as previously described [56] . Except for Tel1 or Rad3 over-expression experiments , cells were extensively restreaked on YES agar plates to achieve terminal telomere states prior to harvesting . For over-expression experiments , minimal media was used for nmt1+ promoter-controlled over-expression . S . pombe genomic DNA samples were digested with EcoRI or ApaI , separated on 1 . 2% ( EcoRI ) or 2% ( ApaI ) agarose , transferred to Hybond-XL membrane ( GE ) , and hybridized to telomere probe [57] in Church Buffer [0 . 25M sodium phosphate buffer pH7 . 2 , 1mM EDTA , 1% BSA , 7% SDS] at 65°C overnight to monitor telomere length . Cells were processed for ChIP and analyzed as previously described [15] , using either monoclonal anti-myc ( 9B11; Cell Signaling ) , anti-FLAG ( M2 , F1804 , Sigma ) , or anti-HA ( 12CA5 ) antibodies . Percent precipitated DNA values ( % ppt DNA ) were calculated based on ΔCt between Input and IP samples after performing several independent triplicate SYBR Green-based real-time PCR ( Bio-Rad ) using telomere primers jk380 and jk381 [15] . For genetic backgrounds that cause eventual circularization of chromosomes due to a telomere maintenance defect , a Rad3 plasmid ( pREP41H-rad3 ) was utilized to maintain linear chromosomes during strain construction [18] . Prior to ChIP experiments , single colonies that had lost the Rad3 plasmid were selected based on lack of growth on media lacking histidine and sensitivity to HU , and immediately utilized in ChIP experiments . Based on Southern blot analysis , the early generation strains that have just lost the Rad3 plasmid carry comparable or slightly longer telomeres than rad3Δ cells [18] ( data not shown ) . In order to determine statistical significance of our data , two-tailed Student's t-tests were performed , and P values ≤0 . 05 were considered as statistically significant differences . | ATM and ATR kinases are two evolutionarily conserved sensors of DNA damage , responsible for maintaining stable genomes in all eukaryotic cells . These two kinases safeguard eukaryotic genomes against undesired double-stranded DNA breaks ( DSBs ) and errors during duplication of genomic DNA . Furthermore , ATM and ATR are redundantly required for stable maintenance of telomeres , protective structures at ends of linear eukaryotic chromosomes . Our current study in fission yeast demonstrates that the previously defined C-terminal Tel1ATM interaction domain of the DNA repair protein Nbs1 , which contributes to recruitment of Tel1ATM to DSBs , is dispensable for recruitment of Tel1ATM to telomeres , due to a previously unrecognized kinase-independent role of ATR in recruitment of Tel1ATM to telomeres . Furthermore , the N-terminus of Nbs1 was found to be critical for recruitment of both ATR and ATM to telomeres . Regulators of telomere maintenance have recently emerged as potentially important therapeutic targets against tumorigenesis and aging in mammalian cells . Since proteins responsible for proper maintenance of telomeres and cellular responses to DNA damage are highly conserved between fission yeast and mammalian cells , a newly uncovered molecular crosstalk between ATM and ATR might also play critical roles in telomere maintenance and DNA damage responses in mammalian cells . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"biochemistry/replication",
"and",
"repair",
"genetics",
"and",
"genomics/chromosome",
"biology"
] | 2010 | A Kinase-Independent Role for the Rad3ATR-Rad26ATRIP Complex in Recruitment of Tel1ATM to Telomeres in Fission Yeast |
How learned experiences persist as memory for a long time is an important question . In Drosophila the persistence of memory is dependent upon amyloid-like oligomers of the Orb2 protein . However , it is not clear how the conversion of Orb2 to the amyloid-like oligomeric state is regulated . The Orb2 has two protein isoforms , and the rare Orb2A isoform is critical for oligomerization of the ubiquitous Orb2B isoform . Here , we report the discovery of a protein network comprised of protein phosphatase 2A ( PP2A ) , Transducer of Erb-B2 ( Tob ) , and Lim Kinase ( LimK ) that controls the abundance of Orb2A . PP2A maintains Orb2A in an unphosphorylated and unstable state , whereas Tob-LimK phosphorylates and stabilizes Orb2A . Mutation of LimK abolishes activity-dependent Orb2 oligomerization in the adult brain . Moreover , Tob-Orb2 association is modulated by neuronal activity and Tob activity in the mushroom body is required for stable memory formation . These observations suggest that the interplay between PP2A and Tob-LimK activity may dynamically regulate Orb2 amyloid-like oligomer formation and the stabilization of memories .
Synthesis of new protein is important for the formation of stable memory [1] . The Cytoplasmic Polyadenylation Element Binding ( CPEB ) proteins are a family of RNA binding proteins that regulate the translation and subcellular distribution of a specific set of cellular mRNAs in various cell types including neurons [2] . Previous studies found that some CPEB family members play a causal role in long-term change of synaptic activity and in stabilization of memory [3]–[9] . For example , in marine snail Aplysia , in the absence of a neuron-specific ApCPEB , serotonin mediated enhancement of synaptic transmission fails to persist beyond 24 h [7] , [10] . Likewise , the Drosophila CPEB , Orb2 , is required specifically for long-term memory but not for learning or short-term memory [3]–[5] . In humans , a particular CPEB family member , CPEB3 , has been linked to episodic memory formation , suggesting a conserved role of CPEB in synaptic plasticity and memory [11] . Interestingly , ApCPEB and Orb2 form self-sustaining amyloidogenic oligomers ( prion-like ) in response to the neurotransmitters serotonin in Aplysia and octopamine or tyramine in Drosophila [5] , [6] , [12] , [13] . More importantly , the oligomeric CPEB is required for the persistence of synaptic facilitation in Aplysia [6] and for the stabilization of memory in Drosophila [5] . These observations led us to propose that the persistent form of memory recruits an amyloidogenic oligomeric form of neuronal CPEB to the activated synapse , which in turn maintains memory through the sustained , regulated synthesis of a specific set of synaptic proteins [5] . However , considering the dominant and stable nature of amyloids , a central question is how the conversion of neuronal CPEB to the amyloidogenic state is regulated to confer activity dependence and restrict it to the relevant neuron/synapse . The Drosophila Orb2 gene has two protein isoforms , Orb2A and Orb2B , and the oligomers are composed of both Orb2A and Orb2B . In the adult brain , in comparison to the Orb2B protein , the Orb2A protein is expressed at an extremely low level [4] , [5] . In spite of its low abundance , the Orb2A protein is critical for Orb2 oligomerization , and Orb2A forms oligomers more readily than Orb2B . More importantly , a mutation that impedes Orb2A oligomerization selectively affects persistence of memory [5] , and the Orb2A prion-like domain is sufficient for long-term memory formation [4] . These observations suggested a model in which the rare Orb2A protein either acts directly as a seed to induce activity-dependent amyloid-like oligomerization of the constitutive Orb2B protein or Orb2A oligomerization indirectly affects oligomerization of Orb2B [5] . In either case the amount and localization of Orb2A protein would therefore be a key determinant of when and where amyloid-like conversion would occur . Here we find that Orb2A has a very short half-life and the Orb2 interacting protein Transducer of Erb2 ( Tob ) , a known regulator of cellular growth , stabilizes Orb2A and induces Orb2 oligomerization . Expression of dsRNA against Tob in the mushroom body neurons does not affect learning , but impairs long-term memory formation . Tob recruits the neuronal protein kinase Lim Kinase ( LimK ) to the Tob-Orb2 complex to induce Orb2 phosphorylation . Phosphorylation regulates Tob-Orb2 association as well as the stability of both proteins , and Protein Phosphatase 2A ( PP2A ) is a key regulator of the phosphorylation status of Tob and Orb2 . Intriguingly , inhibition of PP2A stabilizes Orb2A , but destabilizes Orb2-associated Tob , providing a mechanism for temporal restriction on Orb2A stabilization . Since PP2A and LimK activity can be regulated in a synapse-specific manner [14] , [15] , the phosphorylation-dephosphorylation of Orb2 and Tob provides a putative mechanism of restricting the Orb2 oligomerization to the activated synapse . Tob is also known to regulate the function of CPEB family members [16] . Therefore , the Tob-Orb2 association-dissociation may also regulate Orb2 function in the nervous system .
A regulator of Orb2 oligomerization could potentially fall into at least two distinct categories: an activator that associates with Orb2 and facilitates conversion to the oligomeric state or a repressor that binds to Orb2 and prevents its oligomerization . To identify both types of regulators we used a proteomics approach to perform a comprehensive search for Orb2 interacting proteins in the adult Drosophila brain . The Orb2 proteins were expressed pan-neuronally as C-terminal HA-tagged proteins ( ElavGal4: UAS-Orb2AHA or Orb2BHA ) , and the Orb2 complex was immunopurified using anti-HA antibodies from RNaseA-treated adult head extract ( Figure 1A and B ) . Previously we observed that the C-terminal tags are inaccessible in the Orb2 oligomeric state [5]; thus , the anti-HA antibody preferentially immunopurified the Orb2 monomers . Therefore , to identify proteins that interact with oligomeric Orb2 , we also immunopurified Orb2AHA with an anti-Orb2 antibody ( Figure 1B and C ) . The factors associated with Orb2 were identified using Multidimensional Protein Identification technology ( MudPIT ) ( Table S1 ) [17] . We found 61 proteins that were significantly enriched ( p<0 . 05 ) in the Orb2 immunoprecipitates compared to eight independent control immunoprecipitates ( Figures 1D and S1A and Table S1 ) . Eleven proteins were overrepresented in Orb2 IP compared to the controls , albeit not to statistical significance ( Table S1 ) . To determine the validity of the proteomic approach , we randomly sampled 20 candidate proteins ( out of 72 proteins ) by pair-wise interaction in S2 cells ( Figure 1E and Figure S1B ) . Approximately 50% ( 11 out of 20 proteins ) of the proteins thus tested formed a complex with at least one of the Orb2 proteins in an RNA-independent manner ( Figure 1E and Figure S1B ) . Therefore , the proteomics approach indeed identified specific components of an Orb2 protein complex in the adult Drosophila brain . The candidate proteins either interact directly with Orb2 or indirectly as part of a larger Orb2 protein complex . A gene ontology ( GO ) analysis revealed that the Orb2 proteome is significantly enriched for proteins involved in translation initiation , mRNA binding , and synaptic activity ( Figure 1F ) . The enrichment of these protein complexes supports the idea that Orb2 is involved in regulation of synaptic protein synthesis . The Orb2A protein is undetectable by Western analysis , and a genomic construct expressing Orb2A-EGFP suggests it is ∼100 times less abundant than Orb2B protein in the adult brain [5] . Moreover , monomeric Orb2A has a very short half-life compared to Orb2B ( Figure 2A and Table S2 ) . Taken together , these observations suggest that availability of the Orb2A protein could be an important determinant of efficient Orb2A oligomerization and/or function . In the course of our interaction studies in S2 cells , we noticed one of the candidate proteins , Tob , may influence the Orb2A protein level ( Figure 1E ) . To determine Orb2A and Orb2B stability independent of each other , we used Drosophila S2 cells , in which Orb2 is normally not expressed and Tob is expressed at low levels . S2 cells were transfected with only HA-tagged Orb2 or coexpressed with Flag-tagged Tob . To determine half-life , total Orb2 or Tob protein levels were measured at several time points following treatment with cycloheximide ( CHX ) , which blocks new protein synthesis . The coexpression of Tob nearly doubled the half-life of monomeric Orb2A ( Figure 2A ) . However , Tob had no significant effect on Orb2B ( Figure 2B ) , indicating that association with Tob does not automatically enhance half-life . Likewise , incubation with dsRNA against Tob reduced the level of Orb2A protein but not Orb2B ( Figure S2A ) . Earlier studies with Tob family members have suggested that the stability of Tob itself can be regulated [18] , [19] . We found a fourfold increase in Tob half-life in the presence of either Orb2A or Orb2B compared to Tob alone ( Figure 2C and Table S2 ) . These results suggest that not only does Tob stabilize Orb2A , but Orb2 proteins have stabilizing effects on Tob . The recombinant Drosophila Tob interacts with in vitro transcribed and translated Orb2 proteins , suggesting direct interaction between these proteins ( Figure S2B ) . In mammals , the Tob family consists of six members , with Drosophila Tob most closely related to the mammalian Tob1 and Tob2 proteins [20] . We found both Aplysia CPEB and mouse CPEB3 interact with the closely related Tob1 and Tob2 , and Tob2 increases the steady-state level of ApCPEB and CPEB3 ( Figure S2C ) . Recently , others have reported a direct interaction between mouse CPEB3 and Tob1 [16] , suggesting that Tob is an evolutionarily conserved interactor of CPEB proteins . Tob is required for long-term potentiation of hippocampal CA1–CA3 synapses , a cellular correlate of long-term memory in mammals [21] , and Tob activity is modulated by bone morphogenetic proteins or BMPs [22]–[24] . These observations suggest Tob could function as an extracellular signal-dependent regulator of Orb2 in the nervous system . Does Tob influence Orb2 oligomerization in the adult fly brain ? To answer this , we increased Tob level in the fly brain using the Gal4-UAS system and assessed Orb2 oligomerization by immunopurification . Overexpression of Tob-TdTomato ( Elav-Gal4: UAS-TobTdTom ) , but not the fluorophore alone ( Elav-Gal4: UAS-TdTom ) , increased the levels of 10% SDS and boiling-resistant oligomeric Orb2 in the fly brain ( Figure 2D ) . The amount of Orb2 oligomers in Tob-expressing flies increased nearly 2-fold compared to control flies ( fold increase in oligomers normalized to monomer ± SEM , 1 . 95±0 . 27 , p<0 . 05 , t test ) . Tob has been implicated in a number of cellular processes , including transcriptional regulation and RNA metabolism [23] , [25]–[28] , raising the possibility that the increase in Orb2 oligomerization is a secondary effect of Tob overexpression . We generated a series of deletion mutants of Tob and found that deletion of a conserved 28 amino acid motif , TobΔ28 ( Figure S3A ) , decreased the interaction between Tob and Orb2 in both S2 cells ( Figure S3B ) and the adult fly brain ( Figure S3C ) . However , it had no effect on the association between Tob and the deadenylase Pop2 ( Figure S3D ) or with the transcriptional repressor , Mad ( Figure S3E ) . Overexpression of TobΔ28 ( Elav-Gal4: UAS-TobΔ28TdTom ) in the adult brain did not enhance Orb2 oligomerization ( fold increase in oligomers normalized to monomer ± SEM , 0 . 5±0 . 14 ) ( Figure 2D ) . Does Tob enhance oligomerization of Orb2A , Orb2B , or both ? EGFP-tagged Orb2A and Orb2B formed stable oligomers in the adult fly brain and the oligomers associated with Tob ( Figure S3F ) . To determine the effect of Tob on Orb2A and Orb2B , we coexpressed TdTomato-tagged Tob with EGFP-tagged Orb2A or Orb2B in the adult fly brain . To distinguish from the endogenous Orb2 , we quantified changes in the number of fluorescent puncta , since the abundance of fluorescence puncta co-relate with extent of oligomerization ( Figure 2E ) [5] . The number of Orb2A-EGFP puncta increased ∼2-fold in the presence of Tob ( number of puncta/100 µm2 ± SEM , Orb2A: 4 . 41±2 . 88 , N = 12; Orb2A+Tob: 8 . 06±3 . 69 , N = 15 , t test , p = 0 . 012 ) but not in the presence of TobΔ28 ( 3 . 58±1 . 62 , N = 11 , t test , p>0 . 5 ) ( Figure 2F ) . Unlike Orb2A , Orb2B∶EGFP by itself remained mostly diffused and Tob overexpression had no significant effect on the rare Orb2B puncta ( number of puncta/100 µ∶m2 ± SEM: Orb2B , 2 . 40±2 . 11 , N = 12 , Orb2B+Tob , 1 . 07±1 . 30 , N = 9 , t test , p = 0 . 155 ) ( Figure 2F and Figure S3G ) . In addition to being more numerous , the size of Orb2A puncta also increased significantly when Tob was overexpressed ( size of puncta ± SEM; Orb2A , 0 . 39±0 . 08 µm2 , Orb2A+Tob , 0 . 51±0 . 07 µm2 , p = 0 . 0003 ) , an effect not seen with the rare Orb2B puncta ( Orb2B , 0 . 43±0 . 05 µm2 , Orb2B+Tob , 0 . 38±0 . 06 µm2 , p = 0 . 20 ) ( Figure 2G ) . Taken together , these observations suggest that Tob-Orb2 association promotes Orb2 oligomer formation either by increasing the Orb2A protein levels and/or enhancing oligomerization . Is Tob involved in activity-dependent oligomerization of Orb2 ? Previously we and others have observed that a neurotransmitter such as tyramine or dopamine regulates Orb2 oligomerization [4] , [5] . Therefore , we checked whether tyramine modulates Orb2-Tob interaction . To this end , we fed-starved flies 10 mM tyramine and after 4 h immunopurified the Tob-Orb2 complex from tyramine-stimulated or -unstimulated adult fly brain using a Drosophila Tob-specific antibody ( Figure S4A ) . Tyramine stimulation increased the Tob-bound oligomeric Orb2 nearly 4-fold ( fold increase in oligomers normalized to monomer ± SEM , 3 . 82±0 . 88 , n = 5 , t test , p<0 . 05 ) ( Figure 3A ) , and the oligomers are resistant to boiling in the presence of 10% SDS and 2 M urea , consistent with it being amyloid-like ( Figure 3B ) . The neurotransmitter serotonin ( 5-HT ) had less effect on Tob-Orb2 association ( Figure S4B ) , consistent with our earlier observation that Orb2 oligomerization is influenced by tyramine and not by 5-HT [5] . Use of Orb2B-specific antibody ( Figure 3A , right panel ) indicated Tob-Orb2B association is enhanced by tyramine stimulation . To determine whether Tob-Orb2A association is also modulated by neuronal activity , we used a genomic construct that encompasses only Orb2A coding region and carries EGFP at the C-terminal end ( pCasperOrb2AEGFP ) [5] . In Tob immunoprecipitate from tyramine-treated samples , we see EGFP reacting bands that correspond to the size of the monomeric- ( ∼87 KDa ) and oligomeric-Orb2AEGFP ( Figure 3C ) . Since it is difficult to determine which neuronal populations are activated by tyramine feeding , we also directly activated the mushroom body neurons ( c747-Gal4 , MB247-Gal4 ) with the temperature-sensitive dTrpA1 channel [29] . The flies were transiently exposed to 30°C ( dTrpA1 active ) for 25 min and then returned to 22°C ( dTrpA1 inactive ) . Compared to flies carrying only dTrpA1 or Gal4 , flies carrying both transgenes ( C747Gal4::UAS-dTrpA1 or MB247Gal4:UAS-dTrpA1 ) , there was enhanced Tob-Orb2 association ( Figure 3D ) . Taken together these observations suggest that neuronal activity that enhances Orb2 oligomerization also enhances Tob-Orb2 association . Because Tob was initially identified as a transcriptional regulator [23] , [24] , we asked whether Tob is restricted to the cell body or distributed throughout the neuron , including the synaptic region . Immunostaining of the adult fly brain revealed that , as expected , Tob is present mostly in the cell body ( Figure S4C ) . However , at low levels Tob staining was also detected in the synaptic-neuropil regions ( Figure 3E , mushroom body lobes ) . Previously we established a method to purify synaptosomes from adult Drosophila head [5] . In Western blotting of synaptosome fractions ( Figure S4D , left panel ) Tob was found in the synaptic membrane fraction , similar to Orb2 ( Figure S4D ) . In Δ80QOrb2 flies , which has significantly less Orb2 protein compared to wild-type flies [5] , the distribution of Tob was unaffected , suggesting synaptic localization of Tob is independent of Orb2 ( Figure S4D ) . Similar to the fly brain , Tob was also detected in the synaptic membrane fraction prepared from the mouse brain ( Figure S4E ) . Activity-dependent association with Orb2 and presence in the synaptic region suggest that Tob may act to regulate Orb2 function and/or oligomerization in the synapse . Because Tob is constitutively present in the adult fly brain , we wondered how Tob-mediated oligomerization of Orb2 could be temporally regulated by neuronal activity . Phosphorylation is known to regulate the activity of both Btg/Tob [30]–[32] as well as the CPEB family members [33]–[35] . Consistent with these observations , protein phosphatase 1 ( PP1-87B ) and protein phosphatase 2A ( PP2A ) regulatory subunit twins were found in the Orb2 protein complex ( Table S1 ) , suggesting that Orb2 may also be regulated via phosphorylation and/or that Orb2 recruits these phosphatases to regulate phosphorylation of other proteins ( such as Tob ) in the complex . Blotting of Orb2 immunoprecipitates from the adult brain with phospho-tag™ [36] , a biotin-tagged dinuclear metal complex that selectively binds to phospho-proteins , detected a small amount of phosphorylated monomeric Orb2B protein ( Figure 4A ) . Similar to the fly brain , when expressed ectopically in S2 cells , both Orb2A and Orb2B are phosphorylated , albeit at very low levels ( Figure 4B ) , suggesting Orb2 proteins are transiently phosphorylated in a regulated manner or kept primarily in an unphosphorylated state by the phosphatase . We observed that Tob is also phosphorylated in the adult fly brain ( Figure 4C ) . To avoid a secondary consequence of prolonged inhibition or activation of phosphatases or kinases in the nervous system , we took advantage of the phosphorylation of Orb2 and Tob in S2 cells to determine the acute role of phosphorylation . To determine if phosphorylation has any effect on Tob-Orb2 association , we blocked dephosphorylation using calyculin ( CY ) , a cell-permeable serine-phosphatase inhibitor that blocks protein phosphatase 2A ( PP2A ) at 0 . 5–1 . 0 nM concentration and protein phosphatase1 ( PP1 ) at ≥2 nM concentration [37] . We observed that an hour after treatment with 1 nM CY , the amount of Orb2A associated with Tob was reduced ( Figure 4D ) . The reduction in association was not due to reduction in Tob or Orb2A protein level an hour after treatment with CY ( Figure 4D ) . Because phosphatases influence a large number of proteins in the cell , the reduction of Tob-Orb2 association could be a secondary consequence of phosphatase inhibition . To test more directly the effect of phosphorylation , in a reciprocal experiment , we first treated cell lysates expressing Tob and Orb2A with calf intestinal phosphatase ( CIP ) and then isolated the Tob-Orb2 complex ( Figure 4E ) . We observed that prior dephosphorylation enhanced the association of Tob with Orb2A ( Figure 4E ) . Likewise , when the Orb2-Tob complex was immunopurified with anti-Orb2 antibody and probed with phospho-tag™ , only phosphorylated Orb2 , but not the hyperphosphorylated Tob , was detected in the immunoprecipitate ( Figure 4F ) . Taken together , these results indicate phosphorylation regulates Tob-Orb2 association . Hypophosphorylation promotes Tob-Orb2A association , and hyperphosphorylation reduces it . Because the Tob-Orb2 association alters the half-life of both proteins and phosphorylation affects their association , we examined the effect of phosphatase inhibition on the half-life of both proteins . When Tob was expressed by itself there was modest change in stability in the presence of CY ( Table S2 ) compared to the untreated samples ( Figure 5A ) . Interestingly , the increase in Tob stability that occurred when co-expressed with either Orb2A ( Figure 5B ) or Orb2B ( Figure 5C ) was ∼50% reduced when the phosphatases were inhibited ( Table S2 ) . The destabilization of Tob was observed only in the presence of the PP2A/PP1 inhibitor CY or okadaic acid ( 1 nM ) but not the PP1 selective inhibitor tautomycin ( 10 nM ) ( Figure S5A ) [37] , [38] . Moreover , the extent of Tob phosphorylation appears to be specifically linked to Orb2 complex formation ( Figure 5D ) . The Orb2 proteins , but not the other homologue of CPEB in Drosophila , Orb1 , enhance phosphorylation of Tob , although Tob interacts with both Orb2 and Orb1 ( Figure S5B and C ) . These results suggest un- or hypophosphorylated Tob binds Orb2 . Association of Tob with Orb2 and PP2A inactivation leads to additional phosphorylation of Tob-Orb2 , which results in dissociation and eventual destabilization of Tob . How does phosphorylation affect Orb2 ? Treatment of S2 cells with PP2A/PP1 inhibitors CY ( 1 nM ) and okadaic acid but not PP1-specific inhibitor tautomycin ( 10 nM ) enhanced phosphorylation of both Orb2A and Orb2B ( Figure 5E and Figure S5D ) . Treatment with alkaline phosphatase , which removes phosphate from serine/threonine , and λ phosphatase , which removes phosphate from serine/threonine as well as tyrosine residues [39] , revealed that upon inhibition of PP2A , Orb2 proteins are phosphorylated at multiple sites ( Figure 5F ) . One of the outcomes of these multiple phosphorylations is a significant increase in Orb2A half-life , from 1 h to >24 h , t ( 1/2 ) Orb2A , 1 . 13±0 . 08 , Orb2A+CY , 35 . 5±17 . 5 h; p = 0 . 010 , and doubling of the Orb2B half-life , t ( 1/2 ) Orb2B , 4 . 32±0 . 53 , Orb2B+CY , 8 . 09±2 . 95 h , p = 0 . 05 ( Figure 5G ) . As decreases in PP2A activity increased Orb2 level , likewise increases in PP2A activity by overexpression of PP2A catalytic subunit microtubule star ( Mts ) that associates with Orb2 ( Figure S5E ) resulted in a ∼4-fold decrease in Orb2A ( 0 . 23±0 . 01 , n = 5 ) and a ∼2-fold decrease in Orb2B ( 0 . 51±0 . 02 , n = 3 ) protein level ( Figure 5H ) . Increases or decreases in protein phosphatase 1 87B ( PP1 ) activity had no effect on Orb2A or Orb2B abundance ( Figure 5E and Figure S5F ) . These results suggest like Tob , Orb2 phosphorylation is regulated by PP2A . However , unlike Tob , inhibition of PP2A stabilizes Orb2 , particularly Orb2A . How does Tob promote Orb2A stabilization and/or enhanced Orb2 oligomerization ? Because phosphorylation enhances Orb2 stability , one possibility is that Tob prevents PP2A from accessing Orb2A . However , the association of PP2A catalytic subunit Mts or regulatory subunit Tws with Orb2 was not affected by increased levels of Tob , and the effect of PP2A on Orb2A half-life was not dependent on Tob level ( Orb2A , 25 . 6±14 . 7 h , p = 0 . 02 , and Orb2B , 19 . 5±8 . 3 , p = 0 . 01 ) ( Figure S6A ) . However , we found Tob promotes Orb2 phosphorylation by recruiting LimK to Tob-Orb2 complex . In our effort to identify kinases that phosphorylate Tob , we initially focused on MapK , as in mammals and in C . elegans Tob is phosphorylated by Map Kinase ( MapK ) [19] , [30] , [31] and MapK sites are conserved in Drosophila Tob ( Figure S6B ) . However , in an in vitro kinase assay , MapK did not phosphorylate recombinant Drosophila Tob , although as expected mammalian Tob1 and Tob2 were phosphorylated ( Figure S6C ) . We searched for other kinases and focused on the neuronal kinase LimK for several reasons . First , Tob activity is regulated by BMPs , and in the nervous system LimK is a key mediator of BMP signaling [40]–[44] . Second , neuronal activity regulates the synaptic concentration of LimK [15] . Finally , LimK is required for synapse formation [40] , [45] , [46] , which is reminiscent of the function of ApCPEB [10] and Orb2 ( our unpublished observation ) . In an in vitro kinase assay , we found LimK efficiently phosphorylates recombinant Drosophila Tob as well as the mammalian Tob1 and Tob2 ( Figure 6A ) but weakly phosphorylates maltose binding protein or Tob family member Btg . Tob is a LimK substrate because in the adult fly head ( Figure 6B ) as wells as in S2 cells ( Figure S6D ) LimK associates with Tob . Next we sought to determine whether Tob phosphorylation by LimK is influenced by Orb2 . We performed in vitro LimK assays on immunopurified Tob-Orb2 complex or on Tob alone ( Figure 6C ) . To our surprise , we observed that Orb2 is phosphorylated by exogenously added LimK in the presence of Tob ( Figure 6C ) . The Tob-Orb2 immunoprecipitate from cells contains other proteins in addition to Tob and Orb2 , and therefore Orb2 may be phosphorylated by other kinases in the presence of LimK . To test such a possibility , we incubated recombinant-soluble Orb2B protein and LimK in the presence or absence of recombinant MBP-tagged Tob . We observed phosphorylation of Orb2B by LimK in the presence of Tob ( Figure 6D ) . Furthermore , LimK copurified with both Orb2A and B only in the presence of Tob . However , in the presence of TobΔ28 , which binds efficiently to LimK ( Figure S6D ) but not to Orb2 , there was a marked reduction in the LimK-Orb2 complex ( Figure 6E ) . Together , these data suggest that Tob is a substrate for LimK and that Orb2 proteins become a substrate of LimK when associated with Tob . Does LimK affect Orb2 oligomerization ? To determine whether LimK regulates activity-dependent oligomerization of Orb2 in the adult brain , we examined Orb2 oligomer formation in a LimK hypomorphic mutant LIMK1EY08757 [40] . In the LIMK1EY08757 adult brain , the level of monomeric Orb2B protein level was similar to that of wild-type flies ( Figure 7A ) . We exposed wild-type and LimK mutant flies to 10 mM tyramine and immunopurified either the Orb2 oligomers ( Figure 7B ) or the Orb2 oligomers associated with Tob ( Figure 7C ) . In the unstimulated brain extract , little or no oligomeric Orb2 was observed in the LimK mutant flies ( Figure 7B and C ) . More importantly , unlike wild-type flies , LimK mutant flies did not undergo a tyramine-dependent increase in Orb2 oligomerization ( Figure 7B and C ) . To determine whether an increase in LimK activity enhances Orb2 oligomerization , we analyzed Orb2 puncta formation in the larval neuron , where unlike the adult brain , ectopic expression of LimK did not cause any observable developmental problem . We found that Orb2A-EGFP coexpressed with active LimK ( ElavGal4::UAS-Orb2A-EGFP; UAS-LimK ) has twice the number of puncta ( 16 . 10±1 . 36 , N = 24 ) compared with flies coexpressing a kinase dead version of LimK , LimKKD ( ELAV::UAS-Orb2A-EGFP; UAS-LimKKD ) ( 8 . 71±1 . 74 , N = 6 , p<0 . 05 ) or flies expressing only Orb2A-EGFP ( 6 . 79±1 . 01 , N = 12 , p<0 . 001 ) ( Figure 7D ) . From these several results , we conclude that Tob serves two functions for Orb2A . First , it binds and stabilizes unphosphorylated Orb2A , and second , it allows Orb2A to be phosphorylated by LimK . Each of these events results in an increase in the effective concentration of Orb2A , which induces Orb2A and/or Orb2A-Orb2B oligomerization . Because Orb2 oligomerization is important for long-term memory and Tob affects Orb2 oligomerization , we wondered whether Tob activity is important for long-term memory . To this end , we used the male courtship suppression paradigm in which a virgin male fly learns to suppress its courtship behavior upon repeated exposure to an unreceptive female ( Figure 7E ) [47] . Previously we and others have found male courtship suppression memory is dependent on Orb2 activity [4] , [5] . The TobRNAi was expressed under mushroom-body-specific driver 201Y Gal4 , which drives expression primarily in the γ-lobe neurons [48] . Expression of Orb2 in γ-lobe in an otherwise orb2 null background is sufficient to rescue the long-term memory defect [3] , [4] . We found that male flies expressing TobRNAi ( 201Y:Gal4-UAS-TobRNAi ) in the γ-lobe showed courtship suppression after training in the short term ( 5 min ) , but the courtship suppression was lost when measured at 24 h or 48 h after training ( Figure 7E ) . In contrast , flies harboring just the RNAi ( UAS-Tob RNAi ) or Gal4 ( 201Y:Gal4 ) had no impairment in courtship suppression 5 min or 24 to 48 h after training . These results are consistent with the idea that Tob activity is important for long-term courtship suppression memory .
Our previous work suggested that conversion of neuronal CPEB to an amyloid-like oligomeric state provides a molecular mechanism for the persistence of memory [5] , [6] . However , it is not known how Orb2 oligomerization is regulated so that it will occur in a neuron/synapse-specific and activity-dependent manner . Here we report that factors that influence Orb2A stability and thereby abundance regulate Orb2 oligomerization . We find that Tob , a previously known regulator of SMAD-dependent transcription [23] , [24] and CPEB-mediated translation [16] , associates with both forms of Orb2 , but increases the half-life of only Orb2A . Stimulation with tyramine or activation of mushroom body neurons enhances the association of Tob with Orb2 , and overexpression of Tob enhances Orb2 oligomerization . Both Orb2 and Tob are phosphoproteins . Phosphorylation destabilizes Orb2-associated Tob , whereas it stabilizes Orb2A . Tob promotes Orb2 phosphorylation by recruiting LimK , and PP2A controls the phosphorylation status of Orb2A and Orb2B . PP2A , an autocatalytic phosphatase , is known to act as a bidirectional switch in activity-dependent changes in synaptic activity [14] , [49]–[51] . PP2A activity is down-regulated upon induction of long-term potentiation of hippocampal CA1 synapses ( LTP ) and up-regulated during long-term depression ( LTD ) [14] . Similarly , Lim Kinase , which is synthesized locally at the synapse [15] in response to synaptic activation , is also critical for long-term changes in synaptic activity and synaptic growth [46] . Based on these observations we propose a model for activity-dependent and synapse-specific regulation of amyloid-like oligomerization of Orb2 ( Figure 8 ) . We postulate that in the basal state synaptic PP2A keeps the available Orb2A in an unphosphorylated and thereby unstable state . Neuronal stimulation results in synthesis of Orb2A by a yet unknown mechanism . The Tob protein that is constitutively present at the synapse binds to and stabilizes the unphosphorylated Orb2A and recruits the activated LimK to the Tob-Orb2 complex , allowing Orb2 phosphorylation . Concomitant decreases in PP2A activity and phosphorylation by other kinases enhances and increases Orb2A half-life . The increase in Orb2A level as well as phosphorylation may induce conformational change in Orb2A , which allows Orb2A to act as a seed . Alternatively , accumulation and oligomerization of Orb2A may create an environment that is conducive to overall Orb2 oligomerization . In the absence of an in vitro Orb2A-Orb2B oligomerization assay , we could not distinguish between these two possibilities . For Tob , initial Orb2 association stabilizes Tob . However , association with Orb2 as well as suppression of PP2A activity leads to additional phosphorylation , which results in dissociation of Tob from the Orb2-Tob complex and destabilization . The destabilization of Orb2-associated Tob provides a temporal restriction to the Orb2 oligomerization process . The coincident inactivation of PP2A and activation of LimK may also provide a mechanism for stimulus specificity and synaptic restriction . We find that Orb2A and Orb2B are phosphorylated at multiple sites , including serine/threonine and presumably tyrosine residues . These phosphorylation events are likely mediated by multiple kinases because overexpression of LimK did not affect Orb2 phosphorylation to the extent observed with the inhibition or activation of PP2A , raising several interesting questions . In what order do these phosphorylations occur ? What function do they serve individually and in combination ? What kinases are involved ? Moreover , similar to mammalian CPEB family members , in addition to changing stability , phosphorylation may also influence the function of Orb2A and Orb2B . Does Tob regulate Orb2 function ? In mammals Tob has been shown to recruit Caf1 to CPEB3 target mRNA , resulting in deadenylation [16] , and CPEB3 is known to act as a translation repressor when ectopically expressed . We find Drosophila Tob also interacts with Pop2/Caf1 ( Figure S3E ) [25] and Orb2A and Orb2B can repress translation of some mRNA [52] . Orb2 has also been identified as a modifier of Fragile-X Mental Retardation Protein ( FMRP ) –dependent translation , and Fragile-X is believed to act in translation repression [53] . Therefore , the Tob-Orb2 association may contribute to Orb2-dependent translation repression , and the degradation of Orb2-associated Tob may relieve translation repression . Additionally , if the oligomeric Orb2 has an altered affinity for either mRNA or other translation regulators , Tob can affect Orb2 function by inducing oligomerization . However , the relationship between Tob phosphorylation and its function is unclear at this point . Does involvement of Tob both in transcription and translation serve a specific purpose in the nervous system ? Tob inhibits BMP-mediated activation of the Smad-family transcription activators ( Smad 1/5/8 ) by promoting association of inhibitory Smads ( Smad 6/7 ) with the activated receptor [18] , [24] , [54]–[56] . In Drosophila BMP induces synaptic growth via activation of the Smad-family of transcriptional activators , and subsequent stabilization of these newly formed synapses via activation of LimK [57]–[60] . Our studies suggest Tob and LimK also regulate Orb2-dependent translation , raising the possibility Tob may coordinate transcriptional activation in the cell body to translational regulation in the synapse .
Please see Text S1 for a detail description of the proteomic analysis . The Orb2 lines have been previously described [5] , [52] . The following Drosophila strains were obtained from Bloomington Stock Center: mtsXE-2258 ( Stock 5684 ) , Pp2A-29BEP2332 ( Stock 17044 ) , P{EPgy2}LIMK1EY08757 ( Stock 17491 ) , UAS-LimK1HA ( Stock 9116 ) , and UAS-LimK1 Kinase dead ( Stock 9118 ) . The TobRNAi ( Stock 38299 ) on the second chromosome was obtained from Bloomington TRiP collection . The Gal4 lines were generously provided by Douglas Armstrong ( c547-Gal4 , c747-Gal4 ) [61] , Troy Zars ( MB247 , 201Y ) [48] , and Haig Keshishian ( elav-GeneSwitch ) [62] . The c547 drives expression primarily in the ellipsoid body , c747 , MB247 in all lobes of the mushroom body and 201Y primarily in the γ-lobe of the mushroom body . The elav-GeneSwitch drives expression pan-neuronally in an inducible manner . The UAS-dTrpA1 line was generously provided by Paul Garrity [29] . For expression using the GeneSwitch system , the flies were starved for 16–18 h and then transferred to 2% sucrose containing 200 µM RU486 ( mifepristone , SigmaM8046 ) for 12 h . Various genetic combinations were made by standard genetic crosses . Orb2AHA and Orb2BHA constructs were previously described [5] . The untagged Orb2 and Orb2-interacting protein constructs were made by cloning the full-length PCR products into TopoDonor vector ( Invitrogen ) and were subsequently transferred to p AWF using the Gateway cloning system ( Invitrogen ) . The Drosophila Tob cDNA was amplified by RT-PCR and cloned with Topo-TA ( Invitrogen ) . Flag-tagged Tob was created by the subsequent transfer to the mammalian expression vector , pCMV24 ( Invitrogen ) . Standard molecular techniques were then used to subclone into pMT ( Invitrogen ) for S2 cell expression and pUAST ( DGRC ) for use as a Drosophila transgene . To create TobΔ28 , containing an internal deletion of 28 amino acids ( AA235–262 ) , the amino terminal region and C-terminal regions were amplified separately and engineered to contain an internal NotI site . The two fragments ( EcoRI/NotI and NotI/SalI ) were cloned into pCMV24C . Standard techniques were then used to subclone into pMT and pUAST . For the imaging studies , the tdTomato cDNA was inserted in frame to the C-terminal to create pUAST-TobTdTom and pUAST-TobΔ28TdTom . For antigen production , the cDNA encoding Tob AA 267–564 were amplified by PCR and cloned into pRSETA ( Invitrogen ) in frame with the 6XHis tag . The mammalian cDNAs for Tob1 , Tob2 , Ana , and Btg were amplified by RT-PCR from mouse RNA and cloned with Topo-TA , which was subsequently used to produce pCMV24 . For production of recombinant proteins in E . coli , Tob , Tob1 , Tob2 , and Btg were reamplified using primers designed to produce an in-frame 6XHis tag at the C-terminus and then subcloned into pMal-c2X . A full-length cDNA encoding LimK , LD15137 was obtained from DGRC and amplified by PCR for Topo TA cloning . The insert was subsequently transferred to pAcV5 for S2 cell expression . LimKMT was engineered to mutate D500K by site-directed mutagenesis ( Stratagene ) . All sequences were confirmed against the NCBI sequence prior to use . The pCasperOrb2AEGFP construct is comprised of a genomic fragment 1446 nucleotides 3′ of the last Orb2B-specific exon and 1338 bp 5′ of the exonic sequence of the neighboring gene and therefore does not contain coding region of any of the Orb2 isoforms except Orb2A . The ∼8 . 27 Kb genomic fragment was cloned into the SpeI/XhoI site of pCasper4 , and EGFP was introduced at the C-terminal end by creating an in-frame SgrA1 site . Mammalian HEK293 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% FBS . Transfections were performed using Lipofectin reagent ( Invitrogen ) . Drosophila S2 cells were maintained in Schneider's medium supplemented with 10% FBS with transfections performed using Effectene reagent ( Qiagen ) . The constructs used are as indicated in the figures . When examining quantitative changes , an empty vector was used to ensure equal quantity of DNA in each transfection . Imagequant software was used to determine densiometric changes , which were subsequently analyzed using Graphpad Prizm software . For immunoprecipitations from cell culture , 3×105 transfected cells were used for each immunoprecipitation . The expression constructs used are as indicated in the figures . Following transfection ( 36–48 h ) , the cells were washed in PBS and lysed in 500 µl of 1% Igepal buffer ( 50 mM Tris-Cl , 7 . 5 , 150 mM NaCl , 1% NP-40 [Igpal] , 1 mM DTT , EDTA free protease inhibitor ) and clarified by centrifugation at 14 , 000 rpm for 10 min . For immunoprecipitations from flies , adult heads were collected following flash freezing and vortexing , lysed in 1% Igepal buffer , and clarified by two rounds of centrifugation at 14 , 000 rpm for 10 min . Protein concentration was determined using a BCA kit ( Pierce Biotechnology ) , and between 1–4 mg of head lysate were used for each immunoprecipitation . The following antibodies were used for immunoprecipitation: anti-HA agarose ( Sigma ) , anti-Flag agarose ( Sigma ) , anti-Tob antibody ( raised in guinea pig 2163 ) , and anti-Orb2 ( raised in guinea pig-2233 and rabbit-273 , 402 ) in conjunction with Protein-A agarose ( Repligen ) . The anti-Tob antibody was raised in guinea pig against the C-terminal end of Tob ( Pocono Rabbit Farm ) , purified using Melon resin ( Pierce Biotechnology ) , and used at 1∶100 dilution . Immunoprecipitations performed using S2 cells were incubated for 2 h at 4°C with continuous rocking , and immunoprecipitations performed using head lysates were incubated for 2 h , and then the ProteinA agarose beads were added with additional 2 h incubation . Following four washes , samples were boiled for 5 min in SDS-PAGE gel loading buffer containing 10% SDS and 2 mM freshly prepared DTT . For immunoprecipitation of Orb2 ∼1 mg of total protein and for Orb2AEGFP ∼3 mg of total protein were used . Western analysis was performed following standard protocols . The following antibodies were used for Western analysis: anti-Flag-HRP ( Sigma , 1∶1 , 000 ) , anti-HA-HRP ( Roche , 1∶500 ) , anti-Tob ( guinea pig , 1∶1 , 000 ) , anti-Orb2 ( rabbit , 1∶2 , 000 ) , anti-Orb2 ( guinea pig , 1∶1 , 000 ) , anti-Orb2B ( rabbit , 1∶1 , 000 ) , and anti-EGFP ( MBL , 1∶1 , 000 ) . To examine endogenous Tob expression in wild-type CantonS flies and c547-Gal4::UAS-Orb2AEGFP and c547-Gal4::UAS-Orb2BEGFP flies , the proboscis was removed and the flies were decapitated . The heads were fixed for 2 h at 4°C in 4% paraformaldehyde ( PFA ) /PBS , incubated overnight in 20% sucrose/PBS , followed by 2 h in a 30∶70 mixture of 20% sucrose/PBS and OCT embedding media ( Tissue-Tek ) . The heads were then embedded in 100% OCT , and frontal cryosections were made of 12 µm . The sections were permeabilized in 1% TritonX containing PBS for 5 min followed by 10 min in 0 . 1% TritonX containing PBS ( PBST ) . The slides were blocked in 10% goat serum containing PBST for 1 h , followed by overnight incubation in 1∶50 dilution of melon-purified ( Pierce Biotechnology ) anti-Tob ( 2163 ) antibody . For the CantonS flies , 1∶50 dilution of nc82 ( Developmental Studies Hybridoma Bank ) was also added to mark the synaptic regions . Anti–guinea pig Alexa-Fluor 633 ( Invitrogen ) secondary antibody was used for Tob detection , and anti-mouse Alexa Fluor 488 ( Invitrogen ) was used for nc-82 detection . Images were acquired at 512×512 pixels with a Zeiss LSM 5 . 0 confocal microscope as 1 µm Z-stacks . Images shown are projections of 10 slices . To examine changes in aggregate number in the adult Orb2EGFP flies , the whole brain was dissected to remove the exoskeleton and air sacs . The brain was fixed in 4% PFA/PBS for 30 min at room temperature , washed three times with PBST for 10 min , and then the whole brain was mounted . Expression of Orb2EGFP and TobTdTom was driven using the ellipsoid body-specific driver , c547 . Images were acquired as above . To quantitate the changes in aggregate number , projections of 20 slices were made for each image centering on the central structure of the ellipsoid body . To examine changes in aggregate number in Lim kinase and Orb2EGFP-expressing animals , third instar larvae were filleted and fixed in 4% PFA/PBS for 10 min at room temperature , washed three times with PBST for 10 min , and mounted . Images of the neurites extending from the ventral ganglia were acquired as described . Projections of 10 slices were made . Axiovision software ( Zeiss , v . 4 . 7 . 1 ) was used to quantitate total area , aggregate number , and aggregate size . A commander script was written to identify the region of interest and the puncta within the region . All measurement parameters were kept constant for each image . pMT∶FlagTob by itself or in conjunction with pMT∶Orb2AHA or pMT∶Orb2BHA was transfected into S2 cells . Expression Tob and Orb2 were induced by adding 700 µM CuSO4 . Following 16 h , the cells were washed and incubated with 50 µg/ml cycloheximide . At the indicated times , samples were collected and later analyzed by Western blot using either anti-Flag or anti-HA antibodies . Densitometric measurements were carried out using ImageQuant and plotted ( percent remaining of time zero versus time ) using Prism Graphpad 5 . The decay curve was fitted using first-order kinetics . To determine the half-life of hyperphosphorylated Tob , a similar analysis was performed with the cells being treated with both cycloheximide and calyculin . To examine Tob phosphorylation , amylose-bound MBP-tagged proteins were incubated with 5 ng of recombinant LimK ( Upstate Biotechnology ) and 10 µCi of [γ–32P]ATP for 20 min at 30°C with semiconstant shaking . Control reactions were performed identically but in the absence of LimK . Kinase dilution buffer and reaction buffer were prepared according to the manufacturer's specifications . Following phosphorylation , the proteins were washed four times in PBS with 0 . 1% TritonX and once with PBS prior to loading an 8% SDS/PAGE . Following electrophoresis , the gel was dried and exposed from 4 h to overnight . To examine phosphorylation of recombinant Orb2B , His-tagged Orb2B was expressed in E . coli BL21 ( DE3 ) using a slow induction protocol , and a low amount of soluble protein was purified in Ni+2 column . Approximately 10 ng of Orb2B , MBP-tagged Tob was used in the kinase reaction . To examine phosphorylation of the Orb2-Tob complex , 6×105 S2 cells were transfected with pAct∶Orb2AHA or pAct∶Orb2B individually and in combination with pMT∶Tob . The cells were lysed in 1% Igepal buffer ( 50 mM Tris-Cl , 7 . 5 , 150 mM NaCl , 1% NP-40 [Igpal] , 1 mM DTT , EDTA free protease inhibitor ) and incubated for 15 min with 50 U/ml CIP . Following centrifugation at 14 , 000 rpm for 10 min , the supernatant was incubated for 2 h at 4°C with anti-HA agarose . The immunoprecipitates were washed twice with 1% Igepal buffer and once with a modified RIPA buffer ( 50 mM Tris , 300 mM NaCl , 0 . 1% SDS , 1% Igepal ) . The sample was then split into thirds , with one-third examined by Western blot to ensure equality in protein levels and the other two-thirds used for the in vitro kinase assay described above . For the Tob alone samples , 12×105 S2 cells were transfected with pAcOrb2AHA and pMTTob , and the complex was purified as above and dissociated in 1% Igepal buffer containing 1 M NaCl for 15 min at room temperature . The eluate was then normalized to 150 M NaCl and Tob purified by precipitation with anti-Flag agarose ( Sigma ) . Complete dissociation was ensured by Western analysis . The protein abundance studies were carried out in 4%–12% Bis-Tris SDS-PAGE ( Invitrogen ) and run in MES-SDS ( 50 mM MES , 50 mM Tris Base , 0 . 1% SDS , 1 mM EDTA , pH 7 . 3 ) buffer . In these buffer conditions and in the gradient gel , the phosphorylated bands migrate close to each other , which simplifies the quantification of band intensity . Also , in protein abundance studies , the total cell lysates were prepared , unless mentioned , in the absence of phosphatase inhibitors , again to ensure quantification of the total protein accurately . The 4%–12% gels were also used for the detection oligomeric Orb2 and phospho-tag™ blotting of Orb2- or Tob immunoprecipitate from the adult fly head . To measure phosphorylation status via mobility shift , we found that an 8% SDS-PAGE run in conventional Tris-Glycine buffer ( 25 mM Tris , 192 mM glycine , 0 . 1% SDS , pH 8 . 6 ) is more effective , and in 8% gel the different phosphorylated forms of Orb2 and Tob proteins were better separated . For detection of the phosphoproteins via phospho-tag™ , both 8% and 4%–12% SDS-PAGE were used . Flies were maintained using standard fly husbandry methods . For behavioral analysis , flies were maintained on standard cornmeal food at 25°C and 60% relative humidity on a 12 h/12 h light-dark cycle . Virgin males and females were collected at eclosion under CO2 anesthesia . Males were isolated and placed in individual food vials . All flies were aged for 5 d before behavioral training and testing . To increase the efficiency of RNAi , flies were shifted to 30°C for 3 d before training . The control flies were treated similarly . Canton S females ( 4 d old ) were mated the night before they were used in training . Males were assayed for courtship conditioning using a modified version of the spaced training described by McBride et al . ( 1999 ) [63] . For spaced training , individual males were placed in individual small food tubes ( 16×100 mm culture tubes , VWR ) with a mated female for 2 h . The female was removed , and the male was left alone for 30 min . A different mated female was placed in the tube with the male for another 2 h . The female was removed and the male again rested for another 30 min . A third mated female was introduced in the tube for 2 h and removed at the end of the trial . Control males were treated exactly the same way , except no mated females were introduced into the tube . Memory was tested 5 min , 24 h , and 48 h after training . All tests were performed in a 1 cm courtship chamber . Fresh mated females were used for all time points . All memory tests were recorded ( for 10 min ) and analyzed using a customized software . The courtship index of each male was obtained by manual and/or automatic analysis of the movies by an experimenter blind to the genotype and experimental conditions . | The formation of stable long-term memories involves the synthesis of new protein , however the biochemical basis of this process is unclear . A family of RNA binding proteins , Cytoplasmic Polyadenylation Element Binding ( CPEB ) proteins , are known to regulate synaptic activity and stabilization of memory . The Drosophila CPEB is called Orb2 , and its amyloid-like oligomers are critical for the persistence of long-lasting memories . Amyloid formation is often unregulated and stochastic in nature , and the amyloid state is usually dominant and self-sustaining . However , to serve as a substrate for long-lasting memory , the amyloid-like oligomerization of Orb2 must be regulated in a space- , time- , and stimulus-specific manner . Orb2 has two protein isoforms: Orb2A , which is present only in low abundance , and Orb2B , which is the abundant form . Orb2A is important for oligomerization as well as memory persistence . Previous studies suggested that Orb2A may act as a seed to induce oligomerization of the constitutive Orb2B isoform . Therefore , the availability of Orb2A protein would be an important determinant of Orb2 oligomerization . Here we have analyzed how Orb2 conversion to the oligomeric state is regulated . We find that Orb2A is a very unstable protein and that phosphorylation-dephosphorylation of this isoform via canonical neuronal signaling modules can regulate Orb2A stability , and thereby its abundance . We also show that Tob , a known regulator of CPEB-mediated translation , acts as a stabilizer for Orb2A and triggers Orb2 oligomerization . These observations suggest that amyloid formation can be regulated in a dynamic manner by controlling the availability of the seeding Orb2A protein . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"neuroscience",
"learning",
"and",
"memory",
"molecular",
"neuroscience",
"biology"
] | 2014 | Contribution of Orb2A Stability in Regulated Amyloid-Like Oligomerization of Drosophila Orb2 |
Trypanosoma vivax is the main species involved in trypanosomosis , but very little is known about the immunobiology of the infective process caused by this parasite . Recently we undertook to further characterize the main parasitological , haematological and pathological characteristics of mouse models of T . vivax infection and noted severe anemia and thrombocytopenia coincident with rising parasitemia . To gain more insight into the organism's immunobiology , we studied lymphocyte populations in central ( bone marrow ) and peripherical ( spleen and blood ) tissues following mouse infection with T . vivax and showed that the immune system apparatus is affected both quantitatively and qualitatively . More precisely , after an initial increase that primarily involves CD4+ T cells and macrophages , the number of splenic B cells decreases in a step-wise manner . Our results show that while infection triggers the activation and proliferation of Hematopoietic Stem Cells , Granulocyte-Monocyte , Common Myeloid and Megacaryocyte Erythrocyte progenitors decrease in number in the course of the infection . An in-depth analysis of B-cell progenitors also indicated that maturation of pro-B into pre-B precursors seems to be compromised . This interferes with the mature B cell dynamics and renewal in the periphery . Altogether , our results show that T . vivax induces profound immunological alterations in myeloid and lymphoid progenitors which may prevent adequate control of T . vivax trypanosomosis .
African trypanosomes are extracellular parasites that cause sleeping sickness in humans and Nagana in animals . They include T . brucei species which infect both humans and ruminants , but also T . congolense and particularly T . vivax which are responsible for the vast majority of animal trypanosomosis in sub-Saharan Africa , South America and South Asia [1]–[3] . Due mainly to technical constraints such as a lack of reproducible in vitro culture conditions and relatively poor accessibility to natural hosts , our understanding of the biology and fate of T . vivax in its vertebrate hosts largely stems from the extrapolation of data obtained from the experimental murine infection with T . brucei , T . congolense and T . ewansi , but just a few studies using T . vivax infected mice [4]–[8] . Recently , in a move to gain further insight into the host - T . vivax interaction , we further developed reproducible and reliable in vivo models of T . vivax infection using three different mouse strains and the IL 1392 West African isolate ( see accompanying paper ) . Briefly , our studies showed that all the mouse strain infected with bloodstream forms of T . vivax developed the characteristic anemia and systemic alterations that include acute necrosis of the liver and spleen which are the hallmarks of animal trypanosomosis [9]–[12] . Previous immunobiological studies of trypanosomosis focused mainly on the interaction between trypanosome surface coat antigens ( Variant Surface Glycoproteins , VSGs ) and host cells [13]–[15] . The triggering of polyclonal B cell activation by trypanosomes and the ensuing hypergammaglobulinemia mainly composed of antibodies ( Ab ) that do not recognize parasite antigens or VSGs are also typical of the infection [16]–[19] . The mechanisms underlying this process are still largely unknown . Moreover , the involvement of VSGs in protecting the parasites against host specific immunoresponses provided until recently one of the most exquisite models for the study of antigenic variation . It therefore followed , for many years , that our understanding of the interaction between African trypanosomes and the immune system was limited to this “parasite-driven” view where the host's immune response was restricted to the production of specific Abs against VSGs . Whereas anti-VSG Ab doubtless contribute to early control of the infection , resistance to late phases is not only dependent on specific ( parasite-directed ) immunoglobulins but also seems to rely on T-independent processes since athymic mice and also complement-deficient mice infected with T . rhodesiense are able to mount anti parasitic responses that are sufficient to increase mouse survival and healing after an infectious challenge [7] , [20] . Interestingly , the severity of the disease correlates with the control exerted by T . brucei- and T . congolense- specific Abs over the frequency and duration of parasitemia waves but not the level of circulating parasites . This contrasts with T . vivax infections where the efficiency of the host's Ab response and the parasite-induced negative feedback of Abs raised against the parasite are responsible for regulating both the level and duration of parasitemia waves , thus determining disease severity [21] . In an attempt to throw light on the early events induced by T . vivax in mouse B cell compartments that may contribute to explaining later disturbances in the peripheral B cell pool , we studied the impact of parasite infection both on bone marrow and peripheral lymphoid tissues . Our results using an outbred strain showed that T . vivax mouse infection readily results in B cell differentiation accompanied by massive production of polyclonal immunoglobulins that are mostly nonspecific of parasite antigens . The infection profoundly disorganized the follicular structure of the spleen and similarly to T . brucei infection [6] appeared to destroy the B cell marginal zone , certainly contributing to the substantial fall in B cell population numbers both in the spleen and blood . Numbers of both marginal zone and follicular B cells decreased in the organ in the course of the infection concomitantly with a rise in plasmocytes . Bone marrow analysis showed a sustained and significant increase in the long–term , self-renewal of stem cells on infection . However , it is the fate of B cells , and more precisely of B-cell precursors , that seems to be particularly affected , and this impacts on B cell output . The long-lasting destruction of the spleen marginal zone by T . brucei was previously shown to constitute a barrier to the development of efficient B cell memory and thus to a vaccine-induced B cell response [6] . Our results presented here further indicate that T . vivax infection disturbs the development of B cells in bone marrow which could be an obstacle to immunotherapies against trypanosomosis and even impair unrelated vaccination strategies in trypanosome-exposed populations [18] .
Trypanosoma vivax isolate ILRAD 1392 was kindly provided by R . Brun ( Swiss Tropical Institute , Basel , Switzerland ) . The phenotypic and molecular identification of this isolate has previously been described ( see accompanying paper ) . Bloodstream forms of T . vivax were maintained by weekly passages in 7- to 8-week-old outbred RjOrl:Swiss mice ( CD-1 , Janvier , France ) by intra-peritoneal ( i . p . ) injection of 103 bloodstream forms . Outbred mice were chosen as the experimental model since they present significantly higher survival rates than BALB/c mice and , as described before , a detailed histopathological study of the organs committed by the infection showed similar pathognomonic signs of the disease to those observed in the infected livestock ( see accompanying paper ) . Furthermore , parasitemia in outbred mice reached a plateau by day 10 of the infection and this persisted over time contrasting with the more tolerant C57BL/6 mice that showed recurrent waves of parasitemia . Thus , in our model , 7- to 10-week-old male outbred CD-1 mice were infected i . p . with 102 bloodstream parasites . Parasitemia was determined every 2 to 3 days using a counting chamber and a light microscope as described ( see accompanying paper ) . The experiments described here were performed every 3–5 days throughout the infection , but for clarity and without any loss of important information , only days 10 and 20 post infection are portrayed and considered in the present work since they correspond respectively to peak parasitemia in outbred mice and to the day that generally precedes death , respectively . All animal work was conducted in accordance with relevant national and international guidelines ( see here below ) . All mice were housed in our animal care facilities in compliance with European animal welfare regulations . The Institut Pasteur is member of the Committee #1 of the Comité Régional d′Ethique pour l′Expérimentation Animale ( CREEA ) , Ile de France . The Animal housing conditions and protocols used in the present work were previously approved by the “Direction des Transports et de la Protection du Public , Sous-Direction de la Protection Sanitaire et de l'Environnement , Police Sanitaire des Animaux” under the number #B 75-15-28 accordingly to the Ethics Chart of animal experimentation which includes appropriate procedures to minimize pain and animal suffering . PM has permission to perform experiments on vertebrate animals #75-846 issued by the Department of Veterinary Services of Paris , DDSV and is responsible for all the experiments and protocols carried out personally or under her direction in the framework of laws and regulations relating to the protection of animals . Bloodstream forms of T . vivax were collected by cardiac punction then diluted in buffer A ( polysome buffer ) before being centrifuged for 5 min at 1000 rpm to separate the parasites from the red blood cells . The upper phase was collected and the pellet washed twice . Supernatants were recovered and pooled . Parasites were counted , pelleted by centrifugation ( 15 min at 3500 rpm ) then resuspended at a cell density of 1 to 5×108 cells/ml in buffer A . NP40 was then added ( 1 . 2 % ) , the cells were dounced 30 times with a manual douncer and the suspension centrifuged for 4 min at 14 , 000 rpm . The soluble fraction ( S14 ) was recovered and dialysed over night ( O . N . ) against phosphate buffered saline ( PBS ) . Total immunoglobulins in the sera , and specific Abs against the parasite , were determined by ELISA , as described elsewhere [22] , using flat-bottomed plates pre-coated with goat anti-mouse immunoglobulins or 10 µg/ml of T . vivax S14 extract [23] . Total IgG or IgM concentrations were deduced from standard curves constructed using purified mouse immunoglobulins or presented as titres . Spleen and peripheral blood cell ( PBL ) suspensions ( 107 cels/ml ) were stained with monoclonal Abs diluted in balanced salt solution containing 1% fetal calf serum and 0 . 01% azide . PBL cells were previously treated with ammonium/chloride/potassium buffer ( ACK ) , pH 7 . 2 , to lyse the red cells . Cells ( 106 cells ) were preincubated with anti-CD16/CD32 ( clone 2 . 4G2 ) Abs in order to block immunoglobulin nonspecific binding through Fc receptors . Cells were stained with directly-labeled ( FITC , PE ou Alexa - Fluor® 647 ) or biotinylated Abs for 30 minutes on ice . The following Abs were used: CD3 ( clone 145 . 2C11 ) , CD4 ( clone GK 1 . 5 ) , CD5 ( clone 53–7 . 3 ) , CD8 ( clone 53–6 . 7 ) , B220/CD45R ( clone RA3-6B2 ) , μchain ( clone R6-60 . 2 ) , Mac 1 ( clone M1/70 ) , IgD ( clone 11–26 ) , CD19 ( clone 1D3 ) , Vβ5 . 1+5 . 2 ( clone MR9-4 ) , Vβ6 ( clone 44 . 22 . 1 ) , Vβ8 . 1+8 . 2+8 . 3 ( clone F23 . 1 ) and Vβ14 ( clone 14–2 ) . After washings , biotinylated Abs were further incubated with fluorescein isothiocyanate-streptavidin or phycoerythrin-streptavidin conjugates . Two-color acquisition was carried out with a FACS Scan cytofluorometer ( Becton Dickinson ) or with a FACScanto ( BD biosciences ) . Dead cells were excluded from the analysis by gating out propidium iodide-stained cells . Splenic and PBL lymphocytes were gated on forward-light scatter/side-light scatter combined gate , and 20000–100000 events were acquired . Number of cycling lymphocytes ( S+G2 and M ) in the spleen was estimated by measuring individual cell ploidy by dual parameter FSC/SSC combined FACS analysis of cells stained with propidium iodide . Bone marrow-derived cells from two femurs/mouse were recovered in HBSS/2% FCS , counted , rinsed and resuspended at 1×108 cells/ml . The following Abs from PharMingen ( coupled with FITC , PE , allophycocyanin ( APC ) , or PE-Cy7 ) were used: CD3 ( clone 145-2C11 ) , Gr1 ( clone RB6-8C5 ) , TER-119 ( Ly76 ) , CD45R/B220 ( clone RA3-6B2 ) , CD19 ( clone 1D3 ) , CD11c ( clone HL3 ) , NK1 . 1 ( PK136 ) , Mac 1 ( clone M1/70 ) , CD117/cKit ( 2B8 ) , Sca1 ( clone E13-161 . 7 ) , CD34 ( RAM34 ) , and CD16/32 ( clone 2 . 4G2 ) . The lineage mix consisted of CD3 , Gr1 , TER-119 , B220/CD45R , CD19 , CD11c , NK1 . 1 and Mac 1 Abs coupled to PE . Staining was performed on 5×106 cells for 30 minutes at 4°C . The cells were then rinsed in HBSS/FCS 2% , fixed for 10 minutes in 2% paraformaldehyde , rinsed in HBSS/2% FCS and resuspended in 400 µl of HBSS/2% FCS . Flow cytometry acquisition of at least 100000–150000 events was performed in a FACSCanto . Results were analyzed by FlowJo software ( Tree Star , Inc ) . Prism software ( GraphPad , San Diego , CA ) was used for statistical analyses . Intergroup comparisons were made by an unpaired t test . Spleens were removed from control and infected mice 20 days post-infection ( d . p . i . ) . Mice were initially anesthetized by an intraperitoneal injection of 0 . 1 ml per 10 g mouse body weight of a solution containing 1 mg/ml xylazine ( Rompun 2% , Bayer , Leverkusen , Germany ) and 10 mg/ml ketamine ( Imalgène 1000 , Merial , Lyon , France ) and were then sacrificed by cervical dislocation . After a complete post-mortem examination , the spleen , liver , kidneys , lung , heart and specimens of the central nervous system were removed and immediately fixed in RCL2-CS100 ( 38% ) , a non-toxic , formalin-free fixative ( Alphelys , Plaisir , France ) . Tissue samples from these organs were embedded in paraffin and five-micrometer sections were cut and stained with hematoxylin and eosin ( HE ) . The phenotypic profile of the inflammatory infiltrates was determined by immunohistochemical analysis using the following primary Ab , diluted in sterile PBS ( VWR , Strasbourg , France ) and incubated O . N . at 4°C: rabbit polyclonal anti-human CD3 Ab to detect T lymphocytes ( Dako , Glostrup , Denmark ) , rat monoclonal anti-murine B220/CD45R mAb ( clone RA3-6B2 ) to detect B lymphocytes ( Caltag , Burlingame , CA , USA ) , rat anti-murine F4/80 mAb to detect macrophages ( clone BM8 , Caltag ) , and rat anti-murine Ly-49G2 mAb to detect NK cells ( clone 4D11 , BD Pharmingen , San Diego , CA , USA ) . After removing the paraffin ( xylene followed by ethanol ) , the slides were treated for 20 min with a blocking solution containing 2% bovine serum albumin diluted in PBS , prior to primary Ab incubation . Primary Ab against B220/CD45R , F4/80 and Ly49G2 were visualized using a Histofine Simple Stain MAX-PO kit ( Histofine Biosciences inc , Cambridge , UK ) and primary Ab against CD3 were visualized using peroxidase-labeled polymer for rabbit polyclonal Ab ( EnVision , Dako , Carpinteria , CA , USA ) , according to the manufacturer's protocol . Color was developed with 3-Amino-9-EthylCarbazole ( AEC chromogen; BD Pharmingen ) . The sections were then counterstained with Meyer's hematoxylin , and cover-slipped for microscopic examination . Red areas were considered to be positive , according to manufacturer's indications . All the experiments were performed three or four times using at least 4–5 mice per time point and per experimental group . Mice were analyzed individually and the differences between the groups used in this study were tested for statistical significance using Student's t test whenever appropriate ( Prism software , GraphPad , San Diego , CA ) . The data are expressed individually or as arithmetic means +/− the standard deviation ( SD ) of the means .
Classic features of T . vivax trypanosomosis such as severe acute anemia and the remodeled secondary organs were observed in our recently developed experimental murine models of T . vivax infection ( see accompanying paper ) . In the present study , whereas lymphocytes and white blood cells decreased significantly in the first 10 days of infection ( Figure 1A , left panel ) , monocytes and granulocytes numbers were not significantly altered by the infection ( Figure 1B , right panel ) . In the same manner as with other trypanosome infections in mice and cattle , T . vivax triggers marked lymphocyte blastogenesis that swells cell numbers in secondary lymphoid organs . In fact , as can be seen in Figure 1B , the number of lymphocytes increased throughout the infection ( left panel ) , reflecting the stimulation of the immunological apparatus . We noted that the spleen enlargement observed was not due only to an increase in cell numbers ( 1 . 5 fold ) but also to the elevated number of cycling cells ( Figure 1B , right panel ) in the organ . Also , as compared to spleens from uninfected mice , those from infected individuals included a large number of dead cells ( 2–3 fold the number of live cells ) that were appropriately excluded from the present analyses . As a result , despite considerable individual differences in cell numbers expressed by the mice used in those studies , spleen enlargement is evident as early as 10 d . p . i . and is massive at 20 d . p . i . , constituting splenomegaly ( Figure 1C ) . To investigate whether the increase in cell numbers described above involves all lymphocyte populations or is restricted to certain subsets , spleens were isolated at different time points during the infection and cell suspensions prepared for characterization by flow cytometry . Archetypal CD3+CD4+ , CD3+CD8+ , B220+IgM+ and CD5+IgM+ B1 populations were analyzed ( Table 1 ) . Only days 10 d . p . i . and 20 d . p . i . , which correspond respectively to peak of parasitemia in outbred mice and to the day that generally precedes death were shown as compared to normal age-matched uninfected controls ( see Material and Methods ) . Firstly , while the total number of spleen cells increased up to 1 . 5 fold , the total number of CD3+CD5+ lymphocytes increased 1 . 8–2 fold at 10 d . p . i . . More precisely , while the CD4/CD8 ratio appeared to be unaffected in the first 10 days of infection ( 2 . 4∶1 ) , a fall in CD8+ cells and an expansion of the CD3+CD4+ subset was observed from 10 d . p . i . , significantly increasing the CD4/CD8 ratio ( 6∶1 ) . Furthermore , and similarly to what has been described elsewhere for other trypanosoma infections double negative CD4−CD8− ( possibly gamma-delta T cells ) [24]–[25] cells increased 5 fold at 10 d . p . i . and accounted for almost 10% of the increase in CD3+CD5+ cell numbers ( Table 1 ) . When the expression of major TCR-Vβ chains by CD3+CD5+ cells was investigated , this failed to show any preferential growth of a particular T-cell population but rather reflected the polyclonal nature of the expansion in the T cell repertoire . At the same time , a progressive significant decrease in the number of B220+IgM+ B cells was pointed out during the course of infection . Similarly to previous data obtained in cattle and sheep infected with other trypanosomes [26] , [27] , this experimental T . vivax infection stimulated the expansion of CD5+/loIgMhi B cells ( B1 cells ) in the first 10 d . p . i . but a return to normal levels thereafter . Nevertheless , together with the increases observed in the CD3+CD4+ cell count , this commonly substantiates the B/T cell ratio inversion observed as early as 10 d . p . i . . The frequency of B220+CD19+ cells in the spleen was determined during the infection using additional experimental groups of mice and corroborated the consistent progressive fall in the frequencies of splenic B cells ( see Figure 2A for one exemple ) . However , since lymphocyte cell numbers in the spleen increased throughout infection ( ∼1 . 5 fold ) total counts of B cells were not systematically and significantly decreased if evaluated by the corresponding relative numbers of B220+CD19+ ( Figure 2B ) . This seems to be the consequence of both the individual mouse differences in CD19 cell frequencies and the variability of total spleen cell numbers disclosed by the mice in different experimental groups . Nonetheless , close examination within CD19+ gated population revealed dynamic and reliable alterations of B cell subpopulations in the spleen , as can be evalluated by the separate representation of mouse data ( see Figure S1 for the gating strategy ) . Therefore , newly arrived immature B ( NAI B ) cells increase in the organ ( Figure 2C ) . In contrast , a gradual and significant decrease in marginal zone B ( MZB ) cells and to a lesser extent in follicular B cell counts is observed during the infection ( Figures 2D and 2E ) , as similarly described for T . brucei infection [6] . As expected , the number of plasma/memory cells ( IgM−IgD− ) considerably increased over time and did not return to normal levels even in the late stages of the infection ( Figure 2F ) , contrasting with data obtained in T . brucei-infected mice [6] . In order to gain further insight into spleen cell dynamics , we set about analyzing the functional architecture of the spleen by in situ immunohistochemistry . As can be seen in Figure 3 , B lymphocytes were found exclusively in the spleen white pulp of control mice , more specifically in the lymphoid follicles ( Figure 3A ) . By contrast , the spleens of infected mice ( Figure 3B ) showed a markedly disorganized white pulp associated with severely depleted B cells in the follicles . Major differences in T lymphocyte distribution were also observed between the control and infected animals ( Figures 3C and 3D ) , most probably corresponding to the intense tissue disorder resulting from substantial cellular infiltration , conspicuously noted for macrophages in the red pulp of infected spleens ( Figures 3E and 3F ) . The effect of T . vivax infection on the number of bone marrow cell precursors was studied in an attempt to identify potential hematopoietic abnormalities that could explain the substantial decrease in the peripheral counts of B cell populations . Mice were infected and bone marrow cells recovered and analyzed by flow cytometry 10 and 20 days post infection ( see Materials and Methods and Figure S2 for gating strategies ) . The frequency of hematopoietic stem cells ( HSC ) and early progenitors was then determined within the gated lineage−/lo ( lin−/lo ) bone marrow cell fraction that did not express ( or expressed only low levels of ) markers for mature cells . Thus , Figure 4A shows that while untreated controls possessed a quiescent number of HSCs , the bone marrow was strikingly enriched with pluripotent HSCs ( lin−/lo cKithi Sca1+ ) in infected mice . While the total number of bone marrow cells obtained from two femurs did not increase significantly , HSC numbers increased twice as early as 7 d . p . i . ( not shown ) and remained high throughout the infection ( up to 5 fold increase by 20 d . p . i . ) . Most of these HSCs had only a short-term capacity for reconstitution since more than 80% expressed CD34 reflecting a marked switch from the G0 state to an active cell cycle ( data not shown ) . It is worth noting that HSC numbers deduced from the lin−/lo cKit+Sca1+ gate include a third of cells expressing low levels both cKit and Sca-1 . This lin−/locKitloSca1lo fraction comprises Commun Lymphoid Progenitors ( CLP ) , gives rise to all lymphoid lineages and differentiates from HSC after up-regulation of IL-7R ( not determined here ) . As a result , the marked increase in HSC/CLP combined population , reflects the intense hematopoiesis triggered by the infection . It is important to note the marked decreases observed in lin−ckit+Sca1− marrow cell populations over the first 10 days of the infection that persisted over time ( Figure 4B ) . The distribution of these cells in relation to the expression of CD16/32 and CD34 cell markers ( Figure 4B , right panels ) , showed that the frequencies of Granulocyte-Monocyte Precursors ( GMP , CD16/32+CD34hi ) and Common Myeloid Progenitors ( CMP , CD16/32+CD34lo ) were significantly altered by the infection as compared to uninfected controls . It is interesting to note that although Megakaryocyte Erythrocyte Precursor numbers ( MEP , CD16/32−CD34− ) decreased upon infection ( see Figure 1B , right panel ) , this decrease only became statistically significant on day 21 of the infection , just before death ( not shown ) corroborating the previously observed thrombocytopenia ( see accompanying paper and [10] , [28] ) . Interestingly , the highly reconstituting cells presenting the lin−cKit−Sca1+/hi phenotype [29] increased substantially in number during the infection ( Figure 4C ) and peaked on day 20 . We next analyzed bone marrow cell populations committed to the B cell lineages . As can be seen in Figure 5A CD19+ cell counts decreased significantly throughout the course of infection . Similarly , the number of Pre B + Pro B ( CD19+IgM− ) cell progenitors in the CD19+ gated population decreased more than 10 fold during the course of infection ( Figure 5B; see Figure S3 for gating strategy ) . It is noteworthy that marked individual differences , but statistically significant , were observed for CD19+IgM− at late stages of the infection ( i . e . day 20 ) , when generally only 40% of the infected individuals are still alive ( see accompanying paper ) . Some fluctuations were observed in the number of immature/mature B cells ( CD19+IgM+ ) from day 7 of infection ( not shown ) . This cell population then significantly decreased 10 d . p . i . at the peak of parasitemia and thereafter ( Figure 5C , and accompanying paper ) . These findings are consistent with the emigration of these immature/mature B cells to reconstitute the peripheral B cell pool . However , after 10 days of infection , while more than 70% of B cells expressed IgM , immature pro-B progenitors ( CD19+/loIgM−CD43+/hi ) were seen to grow in number as the infection progressed , but their differentiation into pre-B cells appeared to be delayed since no proportional increase was observed in CD19+IgM−CD43−/lo ( Figure 5B , right panel ) . Altogether , these results suggest that the process of B cell hematopoiesis is disturbed in the bone marrow and may be instrumental in the subsequently diminishing production of mature B cells ( CD19+IgM+CD43− ) in the periphery . A study made of CD19+ PBL B cells in relation to their expression of IgM and IgD detected several sub-populations of circulating B cells ( see Figure S4 for gating strategy ) . Thus , late in the infection , the number of cells recently immigrated from the bone marrow ( Transitional B ) increases in parallel with the fall in numbers of B cells in lymphoid organs ( Figure 6A ) . It was also noted that the number of naïve B lymphocytes decreased considerably throughout the study period analyzed , contrasting with the relative rise of “switched” plasma/memory cells ( IgM−IgD− ) . In order to determine the impact of the decreased B lymphocyte repertoire on the production of immunoglobulins , total IgM and IgG serum levels were quantitated during the infection and their ability to recognize parasite antigens was determined . Sera obtained from the infected mice were tested individually for their ability to react with total T . vivax extracts prepared at the peak of parasitemia ( 10 d . p . i . ) given that outbred mice showed a parasitemia plateau after 10 d . p . i . that seems coherent with a parasite population expressing thereafter the same variable antigen ( VAT ) . While IgM and especially IgG increased 5–10 fold on infection ( Figure 6B ) , Ab production only partially correlate with a capability to specifically recognize T . vivax antigens ( Fig . 6C ) . As expected , IgM specific antibody titers directed against the parasite were regularly maintained throughout the infection as a result of increasing parasite load or to ( newly ) produced IgM responses directed to ( possibly new ) VSGs ( 1/100 to 1/500 dilution out of 1–5 mg/ml of seric IgM ) . This observation contrasted with the high levels of non specific polyclonal IgG recorded at the same time period ( only 1/25 to 1/100 serum dilution out of 10–20 mg IgG/ml reacts with T . vivax extracts ) . This high production of IgG Abs correlates with the rise in peripheral “switched/memory” ( IgM−IgD− ) B cells , possibly engaged with the production of ‘non-IgM’ classes or isotypes of immunoglobulins . These data more than substantiating the polyclonal nature of the B cell response induced by the infection , indicate that T . vivax induces a B cell response that is for the most part not directed to the parasite contributing to its evasion strategies .
The work described herein provides evidence that experimental Trypanosoma vivax infection induces rapid and marked bone marrow hematopoiesis , and a significant delay in the maturation of B-cell progenitors . Maintenance of the peripheral B cell pool requires the continuous input of newly formed B cells [30] . Although our findings show that mature B cells are recruited into the periphery to replenish the pool of terminally differentiated Ig-secreting cells responsible for the massive hypergammaglobulinemia triggered by the infection , an imbalance in B cell development and supply nevertheless persists . Furthermore , early infection decreases the frequency of downstream B-cell progenitors , consequently reducing the flow of mature B cells into the periphery . It was previously shown that parasite-derived B cell directed apoptotic signals cause severe destruction of the available B cell pool in the spleen [6] . Our present findings show in addition that the infection also disturbs bone marrow maturation dynamics , thereby preventing homeostasis . Moreover , a critical drop of Granulocyte-Monocyte and Common Myeloid follows the infection . In addition , the significant decrease of Megakaryocyte Erythrocyte Precursors supports the thrombocytopenia observed with the onset of parasitemia ( accompanying paper ) . As expected , the infection triggers rapid and persistent bone marrow activity leading to the production of high levels of long-term reconstituting stem cells . HSC activation is accompanied by significant bone marrow dynamics but not by the appropriate differentiation of progenitors committed to both myeloid and lymphoid lineages . Rather , the infection induces the expression of Sca-1 in a considerable number of ckit− hematopoietic progenitors , possibly resulting from the release of pro-inflammatory cytokines , as postulated [29] , [31] . It is also possible that CpG-DNA motifs released from the parasite may trigger the production of IFN type I , lead to the intense activation of B cells and equally contribute to the expression of Sca-1 by bone marrow cells [32]–[34] . These recently described ckit−Sca-1+ bone marrow cells [29] possess the potential to more rapidly generate B and T cells than CLPs . Therefore , it is possible that a time lag in the maturation of lymphoid and myeloid precursors is observed as a consequence of differential ontogenesis . Although speculative , one hypothesis would consider that parasite-derived mitogenic ( or superantigenic ) effect on B cells at the periphery could bring about a compensatory effect in the bone marrow to reestablish homeostasis . However , our results indicate a deficiency ( or developmental arrest ) in B-cell precursors further impacting on the availability of mature B cells at the periphery and altering the B: T cell ratio in lymphoid organs . Mouse infection with T . brucei results in profound spleen remodeling with a dramatic drop in IgM+ marginal zone B cells ( MZB ) due to specific apoptosis [6] . Our studies provide further evidence that the mouse infection with T . vivax causes a marked reduction in the spleen of both B220+CD19+IgM+/hiIgDlo ( MZB ) and B220+CD19+IgMlo/hiIgDhi follicular B cells . It would thus appear that the splenomegaly and the increase in cellularity observed result from intense , compensatory T-cell blastogenesis and a significant infiltrative process , mostly composed of macrophages . These findings are supported by the fact that GMP bone marrow numbers decrease by the second week of infection , compatible with the further mobilization of macrophage/monocyte cells to committed organs , as confirmed by histopathological studies . Since no bone marrow atrophy is seen during T . vivax infection , several different hypotheses may be advanced to explain the recurring waves of activation/repression that affect some specific bone marrow-derived progenitors and the consequent delay in B cell development which impacts on the release of B cells to the periphery . One hypothesis considers the possible down regulation of c-kit expression by marrow cells that are unable to respond to stromal cell factors , thus compromising the generation of new lineage progenitors [35] . However , this is unlikely given that lin−ckithi increase in number in the course of the infection . Instead , a deficiency in IL-7 ( or IL-7R ) and IL-12 , which greatly commits HSC into lymphopoiesis , or an increase in IL-15 which interferes more specifically with pre-B cell proliferation , may impair the development and maturation of bone marrow-derived progenitors [36]–[39] . For instance , it was previously reported that increases in TGFβcause down regulation of both IL-7 [40] and VCAM-1 gene expression [41] by stromal cells , interfering with the self-renewal and differentiation of B-cell precursors . However , only a few studies concerning the differential regulation of cytokine expression in mice that are more susceptible or more tolerant to T . brucei and T . congolense infections are currently available to further substantiate these hypotheses [42]–[45] . Interestingly , recent reports have attributed the disappearance of peripheral blood leukocytes to a parasite released factor that , through CD45 , manipulates the host's cytokinic and adaptive responses , inducing lymphotoxicity [7] , [46] . Major dysfunctions in the development of bone marrow cell lineages and greatly impacting on the availability of peripheral lymphoid repertoires have previously been described in other infection models , especially those using MCMV and LCMV viruses [47]–[49] . These abnormalities were shown to be associated with the inability of B- and T-cell subsets to respond to homologous and heterologous antigens , characterizing nonspecific polyclonal lymphocyte responses and the immunosuppression that invariably follows infectious processes [18] . In agreement with the results obtained in livestock and in experimental trypanosome infections , and more specifically with T . vivax-infected cattle [18] , [50]–[52] , we show here that the infection in the mouse experimental model also induces non-specific ( but microorganism-triggered ) polyclonal B cell responses . This is worsened by the presence in the spleen of numerous abnormal plasmocytes ( Mott cells ) that are defective in the production of immunoglobulins ( see accompanying paper ) . The presence of Mott cells in the plasma and their relationship with the failure of individuals infected with African trypanosomes to mount efficient B cell responses , has previously been described [53] . It is possible that MZB cells emigrating from the spleen red pulp differentiate into short-lived plasma cells that would mainly produce T-independent , non parasite-directed , B cells responses . However , the rapid decay of MZB cells is rather compatible with these cells playing a role in the capture of new VSG expressed by the parasite during the infection and their subsequent transport into the splenic follicules for antigen presentation to follicular B cells . This hypothesis seems consistent with our present and other previous data obtained with T . brucei infected mice [6] , where similar decreases of follicular B cells as the infection progresses parallels the increase of IgM-producing ( parasite-specific ) plasma cells at the periphery . The inability of infected individuals to produce long-lasting amounts of antigen-specific IgG Abs and the levels of antigenic variation displayed by the parasites impose major difficulties that prevent immune intervention against trypanosomosis . The experimental model described here may stimulate further studies on B cell development and fate following T . vivax infection and contribute to unraveling pathways of the host-parasite interaction that could help in the design of new therapies for disease control . | Trypanosoma vivax is responsible for animal trypanosomosis , or Nagana , in cattle and small ruminants . Under experimental conditions , the outbred mouse model infected with a well studied West African T . vivax isolate reproduces the main characteristics of the infection and pathology observed in livestock . Anemia and non-specific ( parasite-directed ) polyclonal hypergammaglobulinemia are the most common disorders coincident with the rise in parasitemia . Our results presented here show that the decrease in peripheral B cell populations does not seem to be compensated by newly arriving B cells from the bone marrow . The infection nevertheless prompts intense production of stem cells that mature into myeloid and lymphoid precursors . In spite of this , B cell numbers are specifically reduced in the periphery as the infection progresses . Thus , negative feedback seems to be set in motion by the infection in the bone marrow , more precisely affecting the maturation of B precursors and consequently the output of mature B cells . The origin of these phenomena is unclear but this doubtless creates a homeostatic imbalance that contributes to the inefficient immune response against T . vivax infection . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"infectious",
"diseases/neglected",
"tropical",
"diseases",
"immunology/leukocyte",
"development",
"immunology/immune",
"response"
] | 2010 | Trypanosoma vivax Infections: Pushing Ahead with Mouse Models for the Study of Nagana. II. Immunobiological Dysfunctions |
A key observation about the human immune response to repeated exposure to influenza A is that the first strain infecting an individual apparently produces the strongest adaptive immune response . Although antibody titers measure that response , the interpretation of titers to multiple strains – from the same sera – in terms of infection history is clouded by age effects , cross reactivity and immune waning . From July to September 2009 , we collected serum samples from 151 residents of Guangdong Province , China , 7 to 81 years of age . Neutralization tests were performed against strains representing six antigenic clusters of H3N2 influenza circulating between 1968 and 2008 , and three recent locally circulating strains . Patterns of neutralization titers were compared based on age at time of testing and age at time of the first isolation of each virus . Neutralization titers were highest for H3N2 strains that circulated in an individual's first decade of life ( peaking at 7 years ) . Further , across strains and ages at testing , statistical models strongly supported a pattern of titers declining smoothly with age at the time a strain was first isolated . Those born 10 or more years after a strain emerged generally had undetectable neutralization titers to that strain ( <1∶10 ) . Among those over 60 at time of testing , titers tended to increase with age . The observed pattern in H3N2 neutralization titers can be characterized as one of antigenic seniority: repeated exposure and the immune response combine to produce antibody titers that are higher to more ‘senior’ strains encountered earlier in life .
It has long been know that humans have a higher serologic response to stains of influenza strains early in their lives , even after vaccination or exposure to more recent strains [1]–[3] . Consistent with this phenomenon , some experimental studies in animals and humans have shown that a second vaccine ( or infection ) boosts the serological response to earlier infections and may result in a less robust serological response itself [4]–[6] . However , there is some question as to whether this apparent primacy of initial antibodies in a first infection represents greater protection against similar strains and reduced protection against later strains [7] . Little is known about how the relationship between the antibody response to earlier and later infections plays out in the complex patterns of influenza infection and vaccination experienced by real populations . Understanding these patterns may aid in the interpretation of serological evidence ( i . e . , seroepidemiology ) , and provide insight into how our immune system interacts with an ever changing pathogen . The concentration of antibodies associated with different influenza strains is most often determined using the hemagglutination inhibition ( HI ) or viral neutralization ( NT ) assay [8] . However , the picture of historic influenza infections offered by these assays is imperfect . Both HI and NT assays only measure the ability of a person's serum to interfere with the processes necessary for viral replication , and do not distinguish between highly specific and cross reactive antibodies [8] . Accurately characterizing how antibody levels change over a lifetime of influenza exposure can aid in the interpretation of serological assays and expand our understanding of how the immune system responds to a complex and ever changing pathogen . Since they emerged in 1968 , human influenza A H3N2 virus strains have been in continual global circulation . During this time , H3N2 strains have undergone continual genetic drift , with genetically similar viruses predominating for one or two seasons before receding [9] . Antigenic drift of these strains is thought to be faster than genetic drift , characterized by clustering of strains within antigenic space and occasional longer jumps to form new clusters [10] . Seasonal H1N1 strains re-emerged in 1977 , developing their own sequential lineage , and continue to co-circulate with H3N2 strains to the present [11] . Nonetheless , H3N2 strains represent a sustained lineage with rapid and regular turn-over of genetically and antigenically distinct strains . As such , they present an opportunity to explore the relationship between the birth-year of individuals and their antibody response to key strains , each of which represents a possible exposure or infection at a different time in an individual's life . While it may not be possible to know exactly which strains each individual was infected with , the combination of titer and age may give us some insight into each individual's history of infection . Across an entire population , the relationship between age of potential infection and titer may reveal patterns that increase our understanding of influenza biology and our ability to interpret serological surveys . Here we characterize the serologic profiles to historic strains of H3N2 influenza in a population from Guangdong province , China . We develop a statistical model characterizing the relationship between age and neutralization titers to strains of H3N2 influenza circulating from 1968–2008 . We propose a refinement of the original antigenic sin hypothesis , antigenic seniority , which may better explain the patterns of immune response seen in this population .
Of 273 participants interviewed , 151 provided serum and were tested for H3N2 antibodies . Samples were more often from adults than children ( Table 1 ) . Age at time of testing ranged from 7 to 81 years of age . Age at time of strain isolation ranged from 34 years before birth ( for A/Hong Kong/1968 ( H3N2 ) ) to 80 years of age ( for A/Shantou/2008 ( H3N2 ) ) . A total of 1 , 359 ( 9 strains151 individuals ) neutralization tests were performed . While peak neutralization titer varied by strain , age-specific mean log-neutralization titers ( estimated from a smoothing spline ) were consistently highest among those who were in the first decade of life at the time when a given strain was isolated ( Figure 1 , S1 ) . When we compare the mean log-neutralization titer of all those in a given birth cohort to a given strain , we see that the highest titers for a given strain occur in among the youngest birth cohort alive when that strain was isolated and declines for progressively older cohorts ( Figure 2A ) , and that a birth cohort's titer is highest relative to other birth cohorts for the strains isolated when they were youngest , and declines smoothly for strains isolated later ( Figure 2B ) . In all strains we observe smoothly declining mean titers with increasing age at time of circulation until we reach that cohort of individuals who were 60 or older at the time of sample collection . For those aged 60 and older at the time of sample collection we observe a smooth increase in mean titers with age , seemingly regardless of strain ( Figure 1 ) . Those not yet born at the time of strain isolation show the lowest titers to every strain , with most born 10 years or more after strain isolation having neutralization titers below the detectable threshold ( <1∶10 ) . For example , against A/Beijing/1989 ( H3N2 ) all those born 10 or more years after 1989 have undetectable titers , those born 0–9 years after 1989 mean log titer of 3 . 1 ( 95% CI: 2 . 3 , 3 . 9 ) , those aged 0–9 in 1989 have mean log titers of 4 . 5 ( 95% CI: 3 . 9 , 5 . 0 ) , those aged 10–19 in 1989 have mean log titers of 3 . 4 ( 95% CI: 2 . 9 , 3 . 8 ) , those 20 or older in 1989 but under 70 in 2009 have mean log titers of 2 . 4 ( 95% CI: 2 . 1 , 2 . 6 ) and those 70 or older in 2009 have mean log titers of 2 . 7 ( 95% CI: 2 . 2 , 3 . 3 ) ( Figure S1 ) . We find evidence supporting a strain independent relationship between log neutralization titers , age at time of strain isolation and age at time of testing . To test the hypothesis that there is a common relationship between titer and age , we compared generalized additive models where ( A ) the relationship between neutralization titer and age is dependent only on the age at time of strain isolation and age at time of testing ( relative BIC 0 . 0 ) and ( B ) the relationship between neutralization titer and age ( at time of testing ) is unique to each strain ( relative BIC 206 . 4 ) . Based on BIC we found model A to be the best model of neutralization titers , and the performance of models A and B was roughly equivalent on other metrics ( AICc , performance on held out data , performance of bootstrapped models ) . We considered two additional generalized additive models that capture the possible role of inherent inter-individual variation in antibody response: ( C ) a model where each individual has a random intercept and there is a common effect of age at time of strain isolation ( relative BIC 542 . 9 ) and ( D ) a model with individual random intercepts and a strain specific effect of age ( relative BIC 654 . 3 ) . While these models are inferior to model A if compared by BIC , they are superior in other measures of model fit ( AICc and mean square error from cross validation ) . However , unlike models A and B , these models cannot be used to predict the titers of individuals outside of the training set . For all models the residuals for log-titers were normally distributed with a standard deviation of approximately one ( 1 . 15 for model A , 0 . 91 for model D ) ( Figure S3 ) . Based on its superior BIC and its otherwise equivalent performance to model B , we take model A as the primary model for the remaining analysis . We will refer back to the individual intercept models ( C and D ) when appropriate . Decomposing the three components of model A ( titer by age at time of testing , by age at time of strain isolation and the strain specific intercept ) illustrates the effect each of these components has on mean log neutralization titer ( Figure 3 ) . Age at testing has little effect on neutralization titer until around age 60 , at which point neutralization titers increase smoothly with age ( Figure 3A ) . Age at time of strain isolation causes the largest variation in neutralization titers , with titers peaking at 7 years of age ( increase in log titer of 3 . 4 over baseline , 95% CI , 2 . 4–4 . 5 ) and declining smoothly thereafter ( Figure 3B ) . Those born 10 or more years after a strain was first isolated had the lowest titers to that strain , with increasing titers in those born shortly after or shortly before strain isolation until the 7 year peak . Even after adjusting for the effect of age at time of testing and age at time of isolation there is still variation in titer between tested strains of H3N2 , with the highest neutralization titers being seen against A/Fujian/2002 ( H3N2 ) and the lowest against A/Victoria/1975 ( H3N2 ) ( Figure 3C ) . Visual comparison of predictions from this model with log neutralization titer by age at time of isolation show substantial agreement , and confirm the strain independent relationship between age and titer ( Figure 1 ) . Examination of each individual's neutralization titer against each tested strain gives further evidence of age-specific patterns in neutralization titers and strain-to-strain variation ( Figure 4 ) . Some strains have a very low rate of detectable titers in some age groups . For instance , only 41% of those 50–69 years show neutralization titers 1∶10 ( 18/44 ) or higher to A/Beijing/1989 . In contrast , 82% ( 36/44 ) had titers of 1∶10 or higher to A/Bangkok/1979 ( the previous chronological strain among those tested ) , and 82% ( 36/44 ) had titers of 1∶10 or higher to A/Wuhan/1995 ( the next chronological strain among those tested ) . Despite the importance of age in predicting neutralization titers , individual deviations were common , perhaps attributable to differences in exposure history ( Figure 4 ) . For instance , those 60–69 years old generally have undetectable titers ( <1∶10 ) to H3N2 strains isolated 2003 or later , but two individuals ( from separate study sites ) have high titers to these strains . In model C we account for individuals with generally high or low neutralization titers , strain to strain variation and the age at time of strain isolation; still , 7 . 7% of measured titers are at least four times higher than predicted and 7 . 1% are at least four times lower . Models with individuals intercepts outperform other models on metric that do not penalize extra parameters as aggressively as BIC , indicating a possible role for inherent individual variation in antibody response or frequency of influenza exposure not captured by other covariates . However , the maximally complex model has similar performance to the far simpler model A throughout much of the data ( Figure S4 ) . We compared the performance of model A with models capturing the hypothesis of original antigenic sin . In these models titer depends on a strain intercept and antigenic distance from the first strain in our data that circulated in each participant's lifetime ( as measured in by Smith et al . , 2004 [12] ) . Of the models considered ( see Supplemental Text S1 ) the best model of original antigenic sin treated antigenic distance as a linear term and included a terms for whether the participant was alive when the strain circulated . Despite the relative simplicity of this model , this model was out performed by model A in terms of BIC ( relative BIC = 34 . 3 ) . Models of original antigenic sin fit titers for strains circulating before birth well , suggesting that log-neutralization titer decreases by approximately 0 . 1 log for every unit of antigenic distance from the first possible infecting strain . Because of missing low vaccination rates and the frequency of missing vaccination status ( Table S2 ) , we did not consider vaccination as a covariate in the main analysis . Only 32 of the 151 individuals in our study reported ever having received an influenza vaccination and , of these , only 13 reported receiving vaccination within the last 5 years ( Table S2 ) . We performed additional analyses to assess the possibility that vaccination confounded our results . First , we refitted the models using data only from those who reported having never received a vaccination ( n = 100 ) ( Figure S5 ) . Second , we refitted the models with an additional binary term for the subset of participants who reported whether or not they had ever received a vaccination ( n = 132 ) . Third , on the same subset , we included an additional categorical term capturing the full range of reported vaccination histories ( n = 132 ) . None of these analyses produced qualitatively different results . We performed several tests of model generalizability . First we performed cross validation leaving each titer out from the training data in turn , and then predicting that value using a model fit to the remaining data . Model A performed similarly on cross validated data as to models fit to the full data , while models B , C and D showed small increase in mean squared error ( MSE ) ( Table S1 ) . Second , we fit 500 models to separate bootstrap datasets and used these models to predict on the original data ( a technique that is not valid for the individual intercept models , C and D ) ; MSE for the bootstrapped models was similar to that seen when fitting and predicting on the full data ( Table S1 ) . Third , we fit model A with titers for each strain left out in turn . We found that the relationship between titer and age was qualitatively similar regardless of which strain was held out , with the exception of the model fit without A/Hong Kong/1968 ( Figure S6 ) . In addition to A/Hong Kong/1968's unique position in our data ( it is the first pandemic strain ) , the model fit excluding this strain does not generalize well to predicting A/Hong Kong/1968 titers and does not increase model fit to other strains ( Table S3 ) . For all other strains , the reduced models perform well in predicting the titers of the held out strain , resulting in an increase in mean squared error of 10% or less ( Table S3 ) . Finally , we fit model A with observations for each of the five locations left out in turn . We found that the relationship between titer and age was qualitatively similar regardless of which location was held out , and that the model predicted well on the held out locations ( Table S4 ) .
In this study we have demonstrated a clear relationship between the age at first potential exposure to a strain of influenza A ( H3N2 ) and an individual's neutralization titer to that strain . Using robust statistical techniques , we have demonstrated that this age dependence is consistent across strains . Titers are highest against strains circulating when individuals are 5–10 years of age , and then decline steadily thereafter . Independent of strain and age at first potential exposure , titers start to rise after age 60 at time of testing . The clinical and epidemiological implications of this phenomenon depend on the mechanism leading to these differences . Several plausible options present themselves: immune boosting and interference , age dependent patterns of exposure and changes in the immune system as we age . Immunologic boosting and interference , the tendency for later infections to boost antibody levels to earlier infecting strains and for antibodies to earlier infecting strains to mitigate the later immune response , has historically received the most attention and is supported by experimental evidence [4] , [5] , [12] , [13] . However , it seems that age-specific patterns of influenza infection must also play a role . We observe numerous deviations from the age-specific patterns of neutralization titers that are most easily explained by differences in exposure; and the relationship between age at time of circulation and neutralization titer observed is remarkably similar to the pattern of infection predicted in studies of social contact and mixing patterns [14] . Regardless of the mechanism , it is evident that there remains substantial individual variation in neutralization titer not explained by any of the models considered . This is not unexpected as influenza infection and immune response are influenced by stochastic events which will not be captured by any model . The peak in neutralization titers to strains circulating when a child is around seven years of age is consistent with recent work showing that children in the Netherlands are infected with at least one strain of influenza A by age seven [15] , particularly if ( as suggested by the hypothesis of original antigenic sin ) the antibody response elicited by this first infection is greater than that elicited by later infections . However , it seems that the patterns observed here are not merely the primacy of the first infecting strain ( i . e . , original antigenic sin ) plus cross-reactivity , as we would then expect that the relationship between age at time of strain isolation and antibody titer would be symmetric around the peak ( such an interpretation would also be inconsistent with the experimental results of St Groth et al . [4] ) . However , there is some reactivity to particular strains among those who were not yet born when the strain circulated . For instance , some individuals born 10 years or more after A/Hong Kong/1968 circulated have titers 1∶20 or greater to that strain . The extent to which those who were not yet born respond to an earlier strain must represent the antigenic similarity between that strain and the ones they were exposed to . While some studies of immune response post-vaccination suggest that the inhibiting effects of earlier exposures on the production of vaccine strain specific antibodies may have been overstated , [7] the results of numerous population based and experimental studies ( including our own ) consistently show evidence of an elevated response to the first strains ( potentially ) encountered . [2]–[4] , [16] Additional laboratory experiment and observational work is needed to resolve these discrepancies . The mechanism behind the apparent increase in antibody titer with age among those over 60 years old at time of testing is unclear . Particularly interesting is the fact that this phenomena is evident in response to strains circulating when these individuals were young , middle aged and old; hence it is unlikely to be explained solely by increased exposure in older individuals . Increased longevity among those with high antibody titers ( survivor bias ) , or the effect of having lived through two influenza pandemics prior to 2009 ( 60 year-olds would have been school aged in 1957 ) are both plausible explanations . This former hypothesis is not without precedent , a strong association between higher antibody response and increased lifespan ( with death due to causes other than infection ) has been observed in experimental mouse models [17] , [18] . There are several possible reasons for the strain-to-strain variation that remained even after age at time of testing and age at first circulation is taken into account . These include differences in the extent to which each strain circulated in the region , the intrinsic ability of a specific strain to elicit an immune response , and differences in the neutralizing ability of viral stocks generated for the assays . While residual confounding of the relationship between strain and age is possible , this relationship should be captured by the spline term for age at time of strain isolation . This study was conducted in five communities in one southern province of China , where exposure histories are likely correlated . Patterns of immune response seen here may be unique to the region , though apparent similarities to historical work suggest that this is not the case [1] , [2] . The youngest and oldest age groups are poorly represented: hence , our results may not be generalizable to young children and those over 80 . At the extremes of the range of ages seen in the data , the predicted relationship between titer and age will be more sensitive to outliers and may be biased; however , cross-validation results indicate this is likely not the case . Because this was a cross sectional study , it is difficult to identify potential mechanisms behind the pattern of neutralization response to historic H3N2 strains . Knowing an individual's history of influenza infection would aid greatly in the interpretation of our results , but such data requires long running longitudinal observations and is not available in the current cross sectional study . While there may be some differences in the persistence of antibodies by strain , robustness of a model where titer patterns are shared across strains ( model A ) suggest this is not the case . However , there is some indication of inter-individual variation that may be due to different rates of antibody decay between individuals . Understanding how serological presentation varies by age has important implications for studies relying on sero-epidemiology . If we hope to measure variations in incidence between populations , it is important that we understand how differences in the age composition of different populations affect observed titers . In vaccine trials , where serological response is used as an immune correlate , understanding the background patterns in influenza serologies can improve the interpretation of results . To the extent that neutralization titers are correlates of influenza immunity , they may have clinical and public health implications . Age-specific patterns of protection may indicate those groups that would benefit most from vaccination or the use of a high antigen vaccine . Correct estimates of age-specific patterns of protection can improve simulation studies aimed at predicting the impact of influenza infection . The patterns observed here are similar to those observed by Francis in the mid-20th century [1] , [2] . These earlier studies were primarily focused on H1N1 and , to a lesser extent , H2N2 subtypes . Hence , the patterns seen here are not unique to H3N2 influenza . However , earlier authors did not have modern statistical tools and were unable to characterize the phenomena in the same detail as the present work . In addition , Francis and others were primarily focused on the primacy of the first strain to which an individual was exposed ( true original antigenic sin ) , and did not identify the importance of age at exposure to later strains . We propose that the age dependence observed in this study is more properly called “antigenic seniority” rather than “original antigenic sin” , as it is not only the first strain circulating in an individual's lifetime for which there is an elevated response . We find evidence that the earlier in life that someone is potentially exposed to a strain the higher their antibody titers are likely to be . In the strict interpretation of original antigenic sin , the first childhood influenza infection gains a privileged spot in the immune response , muting the immune response to later viruses and being boosted by later infections . [1] We hypothesize that antigenic seniority may work in a similar manner: viruses to which an individual is exposed early in life can be thought of as taking on senior positions in the hierarchy of immune response , each subsequent infection taking on the next most senior position . Later infections both boost the antibody response to the more senior virus and may have a lessened antibody response themselves . This hypothesis is consistent with the patterns observed in the present study and experimental evidence [3] , [12] , [13] , though more recent work has shown that multiple immunizations can produce a broadly protective immune response [18]–[20] . Even if immune boosting and inhibition are the predominant drivers of the patterns seen , factors such as difference in influenza exposure by age likely still play a role .
All study protocols and instruments were approved by the following institutional review boards: Johns Hopkins Bloomberg School of Public Health , University of Liverpool , University of Hong Kong , Peoples Number 12 Hospital Guangzhou , and Shantou University . Written informed consent was obtained from all participants over 12 years of age . Verbal assent was obtained for participants of 12 years of age or younger . Written permission of a legally authorized representative was obtained for all participants under the age of 18 . Participants were enrolled from 100 randomly selected households from five study locations ( 20 per location ) in a transect extending to the northeast from Guangzhou , China , as described in Lessler et al [21] . All household members over two years of age were eligible to participate . Household members agreeing to participate were administered informed consent and offered two levels of participation: ( 1 ) completing a questionnaire and ( 2 ) completing the questionnaire and providing a blood sample . Enrollment ran from July 8 , 2009 to September 21 , 2009 . Nine strains of influenza A ( H3N2 ) spanning the history of the virus from its emergence in 1968 until the present were selected for serological testing ( Figure S7 ) . We chose six vaccine strains from every second antigenic cluster , starting with a Hong Kong 1968 [10]: A/Hong Kong/1/1968 ( H3N2 ) , A/Victoria/3/1975 ( H3N2 ) , A/Bangkok/1/1979 ( H3N2 ) , A/Beijing/353/1989 ( H3N2 ) , A/Wuhan/359/1995 ( H3N2 ) and A/Fujian/411/2002 ( H3N2 ) . In addition , three recently circulating H3N2 viruses isolated in southern China were selected for testing: A/Shantou/90/2003 , A/Shantou/806/2005 and A/Shantou/904/2008 . Shantou strains are genetically similar to contemporaneous vaccine strains , and may be presumed to be in same antigenic cluster as these viruses ( see Figure S7 ) . Hemagglutination inhibition ( HI ) and virus neutralization ( NT ) assays were performed for each of the nine selected strains of influenza A ( H3N2 ) as described in Lessler et al . [21] Antibody titers were determined by testing serial two-fold dilutions from 1∶10 to 1∶1280 in duplicate ( uncertain results were resolved by repeated testing in quadruplicate ) . Positive and negative control sera were also tested . The highest dilution resulting in complete protection of the cell monolayer in more than two of the quadruplicate wells ( or both duplicate wells ) was regarded as the antibody titer . The effects of both participant age at time of testing and participant age in the year of strain isolation on NT titers were considered . In all cases serological results were assumed to be exact and participants with undetectable titers ( <1∶10 ) were assumed to have a titer of 1∶5 . Models capturing two hypotheses were compared: ( A ) the age dependency of serological response is common across strains and based only on the age at time of testing and age at time of strain isolation , and ( B ) the age dependency of serological response is strain specific . Generalized additive models representing each hypothesis were fit to log-neutralization data and compared using Bayesian information criteria ( BIC ) [22] , [23] . Generalized additive models provide a flexible and integrated framework for fitting non-linear relationships between data ( i . e . , models with a spline term ) . [23] BIC heavily favors more parsimonious models , and we selected it as the primary comparison criteria to avoid over-fitting . However , we also compare models on the basis of other information based and cross-validation based criteria ( e . g . , cross-validated MSE and AICc ) . Confidence intervals in figures were created from standard errors calculated using the mgcv package in R , which are based upon the Bayesian posterior covariance matrix [23] . All analyses were repeated using HI titers yielding qualitatively identical results ( Figure S8 ) . Details of statistical models are available in supplemental Text S1 . Data used in this analysis is available in Dataset S1 . All statistical analyses were performed using the R statistical package ( R 2 . 11 , www . cran . org ) . | The human immune response to an influenza infection is not the same for every infection . It has often been observed that we tend to have the highest antibody titer ( and presumably our strongest immune response ) against strains of influenza that we were exposed to early in life . In this study , we obtained blood samples from 151 people between 7 and 81 years of age and tested the samples for the concentration of antibodies to many different ( H3N2 ) strains . We chose strains according to when they first circulated , starting with a strain isolated just after the 1968 pandemic and going all the way through to very recent strains . We found that a participant's age at the time a strain first circulated was very predictive of the strength of their antibody against that strain . Not just for the first strain they were likely to have seen , but also for the second , third and all subsequent strains circulating during their lifetime . This suggests to us that antibody titers to influenza A H3N2 follow a pattern of antigenic seniority , suggesting that we produce progressively fewer specific antibodies to each subsequent infection as we age . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"disease",
"dynamics",
"population",
"dynamics",
"biology",
"population",
"biology"
] | 2012 | Evidence for Antigenic Seniority in Influenza A (H3N2) Antibody Responses in Southern China |
Despite evidence of the clustering of metabolic syndrome components , current approaches for identifying unifying genetic mechanisms typically evaluate clinical categories that do not provide adequate etiological information . Here , we used data from 19 , 486 European American and 6 , 287 African American Candidate Gene Association Resource Consortium participants to identify loci associated with the clustering of metabolic phenotypes . Six phenotype domains ( atherogenic dyslipidemia , vascular dysfunction , vascular inflammation , pro-thrombotic state , central obesity , and elevated plasma glucose ) encompassing 19 quantitative traits were examined . Principal components analysis was used to reduce the dimension of each domain such that >55% of the trait variance was represented within each domain . We then applied a statistically efficient and computational feasible multivariate approach that related eight principal components from the six domains to 250 , 000 imputed SNPs using an additive genetic model and including demographic covariates . In European Americans , we identified 606 genome-wide significant SNPs representing 19 loci . Many of these loci were associated with only one trait domain , were consistent with results in African Americans , and overlapped with published findings , for instance central obesity and FTO . However , our approach , which is applicable to any set of interval scale traits that is heritable and exhibits evidence of phenotypic clustering , identified three new loci in or near APOC1 , BRAP , and PLCG1 , which were associated with multiple phenotype domains . These pleiotropic loci may help characterize metabolic dysregulation and identify targets for intervention .
The metabolic syndrome represents metabolic dysregulation expressed as the clustering of several physiologic risk factors and is associated with an increased risk of atherosclerosis and type 2 diabetes [1] . The core metabolic syndrome domains are abdominal obesity , atherogenic dyslipidemia , elevated blood pressure , elevated plasma glucose , a pro-thrombotic state , and a pro-inflammatory state [2] , which are represented to varying degrees in commonly used metabolic syndrome scoring systems [3]–[7] . Several lines of evidence support a genetic basis underlying the core metabolic syndrome domains . Measures of metabolic domains cluster in families [8] and heritability estimates range from 16% for systolic blood pressure to 60% for high-density lipoprotein ( HDL ) cholesterol [9] . Genome-wide association ( GWA ) studies have also identified common variants in CETP , LPL , APOA5 , and GCKR that influence the co-occurrence of metabolic domain phenotypes [10] , [11] . Despite evidence of the clustering of metabolic domain phenotypes , current approaches for identifying unifying genetic mechanisms ( i . e . pleiotropy ) remain largely focused on clinical categories that do not provide adequate etiological information [12] . As an alternative , a phenomics approach that assembles coherent sets of phenotypic features that extend across individual measurements and diagnostic boundaries creates the opportunity for novel genetic investigations of established biological pathways and complements the traditional GWA study or candidate gene-based strategy focused on individual phenotypes [13]–[15] . In addition to making use of existing knowledge on process-related information or pathways , a multi-phenotype phenomics approach also may provide greater statistical power than analyses of individual phenotypes [16] and improve the ability to detect effects of small magnitude [17] . Although several authors have advocated the use of such strategies [15] , [18] , [19] , the approach is implemented infrequently . This study evaluated evidence of pleiotropy in clustered metabolic domains using data from five well characterized population-based studies composed of approximately 20 , 000 European American and 6 , 200 African American participants: the Atherosclerosis Risk in Communities ( ARIC ) study , the Coronary Artery Risk Development in Young Adults ( CARDIA ) study , the Cardiovascular Health Study ( CHS ) , the Framingham Heart Study ( FHS ) , and the Multi-Ethnic Study of Atherosclerosis ( MESA ) . Six phenotype domains ( atherogenic dyslipidemia , vascular dysfunction , vascular inflammation , pro-thrombotic state , central obesity and elevated plasma glucose ) encompassing 19 quantitative traits were examined . After dimension reduction , we applied a statistically efficient and computationally feasible multivariate approach that related the phenotype domains to 250 , 000 imputed SNPs . Our approach , which is applicable to studies of heritable , clustered interval scale outcomes , identified several genome-wide significant loci associated with multiple phenotype domains , which may help characterize metabolic dysregulation and identify targets for intervention .
The strongest signal for both European American and African American participants was located on chromosome 9 in the ABO gene ( P<1 . 0×10−300 and P = 6 . 1×10−75 , respectively ) . These signals overlap earlier findings between factor VIII and von Willebrand factor with ABO [20] . Nine additional loci in European Americans and eight loci in African Americans demonstrated effects limited to one metabolic syndrome trait domain that have already been reported in the GWA literature and are therefore not considered further: ABCA1 , APOB , CD36 , CELSR2 , CETP , CRP , F7 , LDLR , LIPC , PVRL2 , TRIB1 , VWF , and ZNF259 . Six loci were associated with at least two trait domains in European Americans: GCKR , ABCB11 , LPL , HNF1A , FTO , and SUGP1 , results which overlap published associations identified through GWA studies for individual trait components . For example , several GWA studies have identified associations between GCKR and elevated plasma glucose [21] , atherogenic dyslipidemia [22] , and vascular inflammation [23]–[25] . GCKR is a plausible unifying mechanism for the clustering of metabolic domains , as the protein inhibits glucokinase , the predominant glucose phosphorylating enzyme [26] . HNF1A , which encodes the transcription factor hepatocyte nuclear factor ( HNF ) -1a , also suggests a common pathogenic background , as previous GWA studies have identified associations with atherogenic dyslipidemia [27] , vascular inflammation [28] , and type 2 diabetes [29] . Of note , FTO was the only previously identified and consistently replicated obesity locus we identified . The strongest new pleiotropic signal in European Americans was for rs4420638 ( P 1 . 7×10−57 ) , located approximately 0 . 32 kilobases ( kb ) downstream of APOC1 and associated with elevated plasma glucose ( P = 8 . 7×10−4 ) , atherogenic dyslipidemia ( 1×10−31 ) , vascular inflammation ( P = 5×10−12 ) , and central obesity ( P = 1 . 2×10−6 ) . Although associations between APOC1 with atherogenic dyslipidemia [22] , [30] , [31] and vascular inflammation [32] , [33] have been reported and replicated in the GWA study literature , we consider it a novel locus due to the strong and previously unreported associations with elevated plasma glucose and central obesity . Localizing this signal is challenging , as the region contains a 48-kb gene cluster that also includes the APOE and pseudo-APOC’ genes [34] . However , the modest levels of linkage disequilibrium ( Figure 4 ) , the presence of a second signal ( Table S11 ) , studies which demonstrate that mice overexpressing human APOC1 show a marked reduction in the update of fatty acids into adipocytes [35] , and the fact the physiological role of APOC1 is less well established than APOE , APOB , and APOA1 [36] all support further evaluation and fine mapping of APOC1 . The second new locus was rs11065987 ( P = 2 . 9×10−10 ) , located approximately 9 . 9 kb upstream of BRAP and associated with atherogenic dyslipidemia ( 3 . 1×10−3 ) , vascular dysfunction ( 2 . 2×10−4 ) , and central obesity ( 9 . 7×10−3 ) . Initial reports suggested that the BRAP protein binds the breast cancer suppressor protein BRCA1 [37] . BRAP is also known to modulate mitogen activated protein kinase signaling [38] , an established cell survival , growth , differentiation , transformation , and proinflammatory pathway [39] . The GWA study literature provides few clues that link BRAP with metabolic trait domains , as associations have only been identified for alanine aminotransferase [24] and esophageal cancer [40] , both in populations of Japanese descent . However , the recombination rate ( cM/Mb ) is low from approximately 110 . 3 Mb to 111 . 5 Mb ( Figure 4 ) and this extended region includes loci associated with type 1 diabetes [41] , [42] , vascular dysfunction [43] , and waist-hip ratio [44] . The ATXN2 gene , located 27 kb from the index SNP , is an intriguing candidate gene . Expansion of a CAG repeat in the ataxin-2 protein causes the neurodegenerative disease spinocerebellar ataxia type 2 . However , instead of a neurodegenerative phenotype , ATXN2-deficient rodents exhibited phenotypes characterized by abdominal obesity , insulin resistance , and marked hepatosteatosis ( i . e . lipid accumulation in the liver ) [45] . Linkage studies of obesity in humans have also associated this region with BMI and total fat percentage [46] . A third genome-wide significant signal was identified for rs753381 ( P = 4 . 3×10−8 ) , a missense mutation in PLCG1 that results in a change from an isoleucine to a threonine . PLCG1 encodes a protein that catalyzes the formation of inositol 1 , 4 , 5-trisphosphate and diacylglycerol from phosphatidylinositol 4 , 5-bisphosphate and plays an important role in the intracellular transduction of receptor-mediated tyrosine kinase activators [47] . Few epidemiologic studies of PLCG1 or neighboring genes have been published . However , mice nullizygous for PLCG1 stop growing mid-gestation and show no evidence of vasculogenesis [48] . Vasculogenesis has been associated with insulin resistance [49] , plasminogen activator inhibitor-1 ( PAI-1 ) concentration [50] , hyperglycemia , and adiponectin levels [51] . This suggests that PLCG1 may contribute to the clustering of metabolic domains in a more subtle manner , such as through small alterations in the structure of the PLCG1 protein . Thus , the missense mutation we identified would serve as a highly intriguing candidate SNP for further study .
In this study composed of approximately 20 , 000 European American and 6 , 200 African American participants , we identified three new loci associated with multiple metabolic trait domains: APOC1 , BRAP , and PLCG1 . These loci were in or near genes previously associated with atherogenic dyslipidemia , vascular inflammation , type I diabetes , vascular dysfunction , and central adiposity . No previous genome-wide or gene-centric studies examining evidence for pleiotropy in metabolic domains has detected these loci at genome-wide significant levels . The pathogenesis of the clustering of metabolic phenotypes remains poorly understood , although it is likely that a sedentary lifestyle , combined with dietary patterns and genetic susceptibility factors , contribute . Candidate genes associated with metabolic syndrome phenotypes largely reflect current knowledge of established pathways regulating obesity , free fatty acid metabolism , insulin sensitivity , lipid metabolism , and inflammation . Although candidate gene and GWA studies have successfully identified loci influencing variation in these pathways , studies examining genetic factors influencing the co-occurrence of metabolic phenotypes are limited . Additionally , those that examine the clustering of syndromic components using the pre-defined clinical cutpoints are largely inconsistent or inconclusive . This general lack of success may reflect ongoing controversy over metabolic syndrome definitions , leading to phenotypic heterogeneity and inconsistent genetic findings across studies [52] . The utility of studying the syndrome as a binary entity as opposed to a series of component traits is also debated [12] , especially since the dichotomization of interval scale traits will discard information . Methods for examining evidence of pleiotropy remain uncommon in the GWA literature and most likely reflect the lack of methodologies and software that are scalable to GWA studies . In this paper , we present a statistically efficient and computational feasible approach to testing for pleiotropy on a genome-wide scale . Our method is applicable to population-based and family studies and identified several associations that would not have been identified through typical univariate analyses . The approach presented herein is also not limited to metabolic phenotypes . Instead , our method could be applied to any set of interval scale traits that are heritable and exhibit evidence of phenotypic clustering . Although alternative analytic approaches were available , for example estimating principal components for all traits simultaneously , we focused on the phenotype clusters presented in Figure 1 . First , evaluating the nineteen phenotypes of interest as six domains of interest is biologically plausible given evidence of phenotypic clustering . It was also easier to interpret principal components that were derived in separate phenotype domains rather than components estimated simultaneously . Additionally , estimating principal components within each phenotype domains ensured that each domain was sufficiently represented in the analysis . Challenges to the approach presented herein include careful phenotype curation , made more difficult by the inclusion of 19 traits across multiple cohorts that were not measured with a common protocol . Only the ARIC and CARDIA studies had full phenotype information on all 19 traits and CHS was the only study with all traits measured during a single visit . The use of a multivariate phenotype comprised of 19 variables also limited the number of contributing cohorts and the identification of replication cohorts , as few studies have such comprehensive phenotypic data . Nonetheless , we were able to identify approximately 25 , 000 participants from studies that used standardized , comparable protocols and many of the associations were consistent across cohorts . Further challenges that are not unique to large scale genetic studies incorporating a phenomics approach include the consistency of results across populations defined by age , race , sex , or other demographic characteristics . For example , the three new loci identified in the European American population were not detected in the African American population . Given a modest sample size of 6 , 287 participants it is difficult to determine whether an inability to generalize results to the African American population reflects different patterns of LD , varying environmental contexts , or limited statistical power . Variation in mean age between contributing cohorts , which ranged from 25 years in the CARDIA study to 72 years in the CHS , could introduce additional heterogeneity , as associations between metabolic phenotypes have been shown to diminish with age [53] . Finally , marked variation in the prevalence of the metabolic syndrome by gender , regardless of clinical definition , suggest the possibility of sex-specific metabolic syndrome effects [54] . Analyses that examine modification by sex , age , and other important clinical covariates are therefore warranted . Our use of the IBC array , which is composed of variants implicated in cardiovascular , inflammatory , hemostasis/coagulation , and metabolic pathways , was beneficial in that it allowed us to leverage the wealth of information on pathways implicated in metabolic disturbances while reducing multiple testing penalties . Admittedly this approach was limited in that it potentially excludes novel pathways not captured by the IBC chip . Although imputation allowed us to increase the number of variants , genome-wide approaches might identify additional pleiotropic loci . In summary , our results support phenomics as a complementary approach that leverages phenotypic variation for the evaluation of pleiotropy , a clear limitation of existing studies examining the metabolic syndrome using clinical definitions . Our approach , which is applicable to studies of heritable , clustered interval scale outcomes , also takes advantage of the wealth of phenotype data available in longitudinal cohort studies as well as emerging analytical and bioinformatics approaches . Ultimately , these results support the presence of genetic variants with pleiotropic effects on adiposity , inflammation , glucose regulation , dyslipidemia , vascular dysfunction and thrombosis . Such loci may help characterize metabolic dysregulation and identify targets for intervention .
This study arose from a collaboration between investigators from two National Institute of Health funded consortia examining the genetic basis of common complex diseases: the Population Architecture using Genomics and Epidemiology ( PAGE ) study , a National Human Genome Research Institute funded effort examining the epidemiologic architecture of common genetic variation that have been reproducibly associated with human diseases and traits [55] and the CARe Consortium [56] , a National Heart , Lung , and Blood Institute-supported resource for genetic analyses examining cardiovascular phenotypes . Briefly , PAGE investigators participating in the phenomics working group wanted to extend existing efforts examining evidence for pleiotropy in approximately 300 replicated genetic variants [57] to include a more comprehensive evaluation of common SNPs . A collaboration between PAGE and CARe investigators was therefore initiated , and used data from five CARe studies of European American and African American with adequate phenotype data: ARIC , CARDIA , CHS , FHS , and MESA . All participating institutions and CARe sites obtained Institutional Review Board approval for this study . Additional information on the participating CARe studies is provided in Text S1 . The Institute for the Translational Medicine and Therapeutics ( ITMAT ) -Broad-CARe ( IBC ) genotyping array [58] was used to evaluate approximately 2 , 100 genes related to cardiovascular , inflammatory , hemostasis/coagulation , and metabolic phenotypes and pathways . The IBC array tagging approach was designed to capture maximal genetic information for both common and lower frequency SNPs ( <5% minor allele frequency ( MAF ) ) in HapMap as well as European American and African American populations . The array included 49 , 320 SNPs , 15 , 000 of which were gene variants not present in HapMap . Additional details of the SNP selection and tagging approach are given in Text S1 . Imputation of untyped and missing SNP genotypes was performed using MACH 1 . 0 . 16 . [59] For the European samples , phased haplotypes from the CEU founders of HapMap 2 were used as reference . For African American populations , a combined CEU+YRI reference panel was created that includes SNPs segregating in both CEU and YRI , as well as SNPs segregating in one panel and monomorphic and non-missing in the other . Imputation for the IBC array was performed in two steps . First , individuals with pedigree relatedness or cryptic relatedness were filtered . A subset of individuals was randomly extracted from each panel and used to generate recombination and error rate estimates for the corresponding sample . Second , these rates were used to impute all sample individuals across the entire reference panel . Before cleaning , there were an average of 246 , 740 ( range: 245 , 816 , 247 , 505 ) and 227 , 224 ( range: 225 , 111 , 229 , 061 ) imputed SNPs in the European American and African American study populations , respectively . Imputation results were then filtered at an imputation quality limit of 0 . 30 and a MAF threshold of 0 . 01 , yielding 235 , 077 ( 95 . 3% of total ) and 227 , 222 ( 96 . 2% of total ) SNPs for analysis in European American and African American participants , respectively . The clustered risk factors of interest were characterized as a six-domain phenotype: atherogenic dyslipidemia , vascular dysfunction , vascular inflammation , pro-thrombotic state , elevated plasma glucose , and central obesity ( Figure 1 ) . These domains were constructed a priori based on a review of literature examining clustering in metabolic phenotypes , placing specific emphasis on the National Cholesterol Education Program’s Adult Treatment Panel III report [4] , [60] . Nineteen variables were then selected to represent one of the six domains with preference for variables measured in at least four of the contributing cohort studies or variables that were highly correlated with available measures . Measurement protocols for each variable by study are provided in Table S21 . We assessed normality , and transformations were used when variables exhibited excessive skewness or kurtosis as determined by numerical summary information and visual inspection of histograms and normal probability plots . Dimension reduction using principal components analysis was then performed for each phenotype domain separately in each race/ethnic and study population . For example , principal components for the vascular inflammation domain were calculated using the following traits: albumin , C reactive protein , fibrinogen , uric acid , and white blood cell count . Principal components were chosen so that>55% of the variance for each domain was explained ( Tables S12 , S13 , S14 , S15 , S16 , S17 , S18 , S19 , S20 ) . This threshold was chosen because all of the first ( waist circumference , pro-thrombotic state , elevated plasma glucose , and vascular dysfunction ) and the sum of first and second ( vascular inflammation and atherogenic dyslipidemia ) principal components exceeded 55% across all studies and racial/ethnic groups . For each phenotype , we fit a linear regression model relating the phenotype to the SNP genotype under the additive mode of inheritance; the model includes environmental variables ( i . e . , age , sex and study center ) as well as the first ten principal components from EIGENSTRAT to adjust for population substructure [61] . Ten population substructure components were included because each component was associated with at least one of the eight phenotypes of interest in at least one study . If the SNP genotype is not associated with any phenotype domain , then the regression coefficients for the SNP genotype are zero in all eight linear models . We tested this global null hypothesis by constructing a multivariate test statistic based on the joint distribution of the score statistics from the eight linear models , which accounted for the correlation between the eight phenotypes . We chose the score statistic because it is computationally efficient and numerically stable . The test statistic is referred to the chi-squared distribution with eight degrees of freedom . The genome-wide significance level was set as P<2 . 13×10−7 ( i . e . 0 . 05/235 , 077 ) . Q-Q plots by race are not presented , as our use of a gene-centric array highly enriched for metabolic loci complicated the identification of markers with low prior probabilities of association ( i . e . “null markers” ) for all phenotypes of interest . The data from each cohort were analyzed separately and the results were combined via meta-analysis as described in Text S2 . All analyses were stratified by race and were performed in SAS 9 . 1 and C++ . Further details are given in the Text S2 . | The metabolic syndrome represents a clustering of metabolic phenotypes ( e . g . elevated blood pressure , cholesterol levels , and plasma glucose , as well as abdominal obesity ) and is associated with an increased risk of atherosclerosis and type 2 diabetes . Although multiple genes influencing the specific metabolic syndrome components have been reported , few studies have evaluated the genetic underpinnings of the syndrome as a whole . Here , we describe an approach to evaluate multiple clustered traits , which allows us to test whether common genetic variants influence the co-occurrence of one or more metabolic phenotypes . By examining approximately 20 , 000 European American and 6 , 200 African American participants from five studies , we show that three regions on chromosomes 12 , 19 , and 20 are associated with multiple metabolic phenotypes . These genetic variants are highly intriguing candidates that may increase our understanding of the biologic basis of the clustering of metabolic phenotypes and help identify targets for early intervention . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"public",
"health",
"and",
"epidemiology",
"cardiovascular",
"disease",
"epidemiology",
"epidemiology",
"biology",
"population",
"biology",
"genetic",
"epidemiology"
] | 2011 | A Phenomics-Based Strategy Identifies Loci on APOC1, BRAP, and PLCG1 Associated with Metabolic Syndrome Phenotype Domains |
Investigating spatial patterns of loci under selection can give insight into how populations evolved in response to selective pressures and can provide monitoring tools for detecting the impact of environmental changes on populations . Drosophila is a particularly good model to study adaptation to environmental heterogeneity since it is a tropical species that originated in sub-Saharan Africa and has only recently colonized the rest of the world . There is strong evidence for the adaptive role of Transposable Elements ( TEs ) in the evolution of Drosophila , and TEs might play an important role specifically in adaptation to temperate climates . In this work , we analyzed the frequency of a set of putatively adaptive and putatively neutral TEs in populations with contrasting climates that were collected near the endpoints of two known latitudinal clines in Australia and North America . The contrasting results obtained for putatively adaptive and putatively neutral TEs and the consistency of the patterns between continents strongly suggest that putatively adaptive TEs are involved in adaptation to temperate climates . We integrated information on population behavior , possible environmental selective agents , and both molecular and functional information of the TEs and their nearby genes to infer the plausible phenotypic consequences of these insertions . We conclude that adaptation to temperate environments is widespread in Drosophila and that TEs play a significant role in this adaptation . It is remarkable that such a diverse set of TEs located next to a diverse set of genes are consistently adaptive to temperate climate-related factors . We argue that reverse population genomic analyses , as the one described in this work , are necessary to arrive at a comprehensive picture of adaptation .
The availability of genome sequences for an increasing number of organisms makes it possible to search for evidence of positive selection on an unprecedented scale . Several studies on different organisms such as bacteria , fruit flies , maize and humans suggest that positive selection is an important force shaping the genome [1]–[5] . However , how different forms of positive selection affect genome evolution and variation is still unclear . Particularly , we do not know the importance of directional selection , which promotes fixations of advantageous alleles , compared to that of spatially varying selection , which promotes maintenance of functional polymorphisms in populations . While most studies focused on the signatures of directional selection , new insights suggest that spatially varying selection might also be an important force [6]–[8] . Understanding the mechanisms and dynamics of spatially varying selection is important both from basic and applied perspectives since adaptive polymorphisms can be used as monitoring tools to detect the impact of climate change on populations [9]–[11] . Clines have long been used to infer the action of natural selection on particular genes and traits across environmental gradients [12] . Drosophila melanogaster is a good model to study adaptation in general and to environmental heterogeneity in particular because it is a tropical species that originated in sub-Saharan Africa and has only recently colonized the rest of the world [13]–[14] . Some of the adaptations that occurred in the populations that migrated out of Africa may specifically be related to temperate environments [15]–[18] . In this species , primarily populations collected along the Australian and North American East coast have been used to investigate the genetic variation associated with climatic adaptation [19]–[20] . Both geographical regions have proven to be ideal settings for this type of study because they span populations from tropical to temperate environments and flies can be easily collected at low altitudes thus avoiding the confounding effects of altitude on climate-associated patterns . Moreover , there are several lines of evidence suggesting that gene flow among populations along each one of these clines is high [19] , [21]–[22] . This evidence favors the interpretation of significant genetic differentiation as a result of natural selection rather than being a by-product of non-adaptive processes related to population structure and history [23] . Adaptation to temperate environments in D . melanogaster has been related to a variety of genes [23]–[28] , life-history traits [19] , [29] , stress resistance [30]–[31] , thermotolerance [32] and morphological traits [20] , [33]–[35] . However , most of these studies are based on a priori candidates , giving a biased picture of the genes and traits involved in adaptation to temperate environments . Studies that analyzed clines in allele frequencies also often lack an understanding of the selective agent responsible for producing the cline while in many cases the genes underlying the clines in phenotypic traits are unknown . Genome-wide analyses that not only identify candidate loci but also investigate the plausible selective agents and their phenotypic consequences are necessary in order to obtain a more comprehensive picture of adaptation to temperate environments . In this study , we investigate adaptations to temperate environments in D . melanogaster associated with a specific class of mutation , the insertion of transposable elements ( TEs ) . Potentially adaptive TEs are assessed on a genome-wide scale and based solely on their population behavior . There is strong evidence for the adaptive role of TEs in the evolution of the Drosophila genome [18] , [36]–[38] and preliminary results based on the analysis of two populations collected at the endpoints of the Australian cline suggest that TEs may indeed play an important role in adaptation to temperate climates [18] . Our analysis is based on a genome-wide screen for TEs likely to be adaptive to the out-of-Africa environments . We first identified TEs likely to have increased in frequency during or after the spread of D . melanogaster out of Africa and therefore likely to be involved in adaptation to temperate climate . However , not all the identified TEs are equally likely to be adaptive . We used information about their family identity and about the patterns of nucleotide variability in the regions flanking these insertions to classify them in two groups: putatively adaptive and putatively neutral TEs . We then looked for evidence of population differentiation for both sets of TEs in three pairs of populations with contrasting climates . By themselves , such patterns may simply reflect genetic structure along the environmental gradients [19] , [23] . However , while drift or historical processes predict similar population patterns for the neutral and adaptive TEs , selection predicts population differentiation patterns only for the adaptive TEs [23] . Therefore , we used the contrast between the two sets of TEs as evidence for the action of natural selection . We further evaluated the inference of the action of natural selection by testing for the consistency of the patterns on different continents . Once we identified the most likely TE candidates , we analyzed the association between their population frequencies and a number of climatic variables to gain insight into the environmental factors that might be contributing to selection . We ended our analysis by integrating all the information available for these TEs and their nearby genes to infer plausible phenotypic consequences of these mutations and the underlying mechanisms . We conclude that adaptation to temperate environments is widespread in Drosophila with TEs playing a significant role in this adaptation . We argue that without population genetics data of the kind described in this paper it is not possible to predict that such a diverse set of TEs located in such a diverse set of genes would be adaptive to temperate climate-related environmental variables . We believe that reverse population genomics studies as the one described in this work are necessary for a comprehensive understanding of adaptation .
We started our search from a set of 763 TEs annotated in the Release 5 of the D . melanogaster genome ( Petrov , D . A . , Fiston-Lavier , A . -S . , Lipatov , M . , Lenkov , K . and González , J . , unpublished data ) . We identified TEs present at low frequencies in the ancestral African population and at high frequencies in the derived North American populations ( see Materials and Methods ) . These TEs are likely to be involved in adaptation during or after the spread of D . melanogaster out of Africa . We further focused on TEs present in regions with a recombination rate larger than zero [39] as they are less likely to have reached high frequencies neutrally compared with regions of very low recombination where the efficacy of selection is substantially reduced [40]–[43] . Thirty-two TEs are likely to have increased in frequency during or after the migration of D . melanogaster out of Africa ( Table S1 ) . However , these 32 TEs are not equally likely to be adaptive . Some TEs belong to families in which the majority of TEs are present at high population frequencies . These families are more likely to be subject to relaxed purifying selection as a whole and therefore TEs in these families are more likely to have increased in frequency neutrally [42] . On the other hand , families in which the majority of TEs are present at low frequencies are likely to be subject to purifying selection . The few TEs present at high frequencies in these families are therefore likely to be adaptive . We used a maximum likelihood approach to estimate the selection coefficient of the 14 families represented in our dataset ( see Materials and Methods; Table S2 ) . We classified the families as putatively neutral when the estimated selection coefficients were not significantly different from zero and putatively adaptive when selection coefficients were significantly negative . TEs from families for which the selection coefficient could not be inferred due to the low number of elements in each family were also considered as possibly adaptive since we do not have clear evidence of their neutrality ( Table S2 ) . The above classification into adaptive and neutral families is strongly supported by previous analyses [18] , [44] . González et al . 2008 [18] analyzed the flanking regions of five putatively adaptive elements , including elements that belong to families with negative selection coefficients and elements from families for which selection coefficients could not be estimated . All five flanking regions showed evidence of selective sweeps suggesting that these TEs had increased in frequency due to positive selection [18] , [44] . Four elements belonging to neutral families were also sequenced and all four appeared to have increased in frequency neutrally [44] . Because D . melanogaster is a tropical species by origin [13]–[14] , some of the adaptations that happened in the populations that migrated out of Africa may be related to adaptation to temperate environments [15]–[18] . If some of the TEs in our dataset are involved in adaptation to temperate habitats , we expect them to be present at higher frequencies in the out-of-Africa populations located in the more temperate regions compared to the ones located in the more tropical regions . Preliminary results support this hypothesis . We previously analyzed the frequency of 21 of these 32 TEs in populations collected in 2007 close to the endpoints of a latitudinal cline on the East coast of Australia . We found that eight of them were present at higher frequencies in the temperate compared to the tropical population [18] . To further test this hypothesis we estimated the frequency of the 32 TEs in populations collected close to the ends of latitudinal clines in Australia in 2007 ( Innisfail and Yering Station ) and in 2008 ( Innisfail and Melbourne ) and on the East coast of North America ( Rocky Ridge and Watch Me Grow; Figure 1; see Materials and Methods ) . We expect TEs involved in adaptation to temperate environments to be present at higher frequencies in the Southern compared to the Northern populations in Australia while we expect the opposite pattern in North America: higher frequency in the Northern compared to the Southern populations . On the other hand , we do not have a priori reasons to expect directionality in the increase in frequency of putatively neutral TEs if population differentiation is present for these TEs . We looked for evidence of population differentiation using a maximum likelihood approach ( see Materials and Methods ) . However , some of the TEs that showed population differentiation are located inside one of the four cosmopolitan chromosomal inversions previously reported to show latitudinal patterns ( Table S1 ) [45] . To avoid the confounding effects of the inversions on the population frequencies of these TEs , we scored by PCR the presence of three of the four inversions in all the strains analyzed ( see Materials and Methods ) . Only a few strains showed presence of inversions In ( 3L ) Payne and In ( 2L ) t and these strains were removed from the analysis . Inversion In ( 3R ) Payne showed a strong clinal pattern as previously described [46]–[47] . Because the confounding effects of this inversion could not be discarded , TEs located inside inversion In ( 3R ) Payne ( FBti0019415 , FBti0019410 and FBti0019418 ) and inside inversion In ( 2R ) NS ( FBti0019012 ) , which was not scored , were excluded from our analysis . Table 1 shows the frequency of the 28 TEs for which we could discard the confounding effects of inversions in the three pairs of populations analyzed . The first 18 TEs in the list belong to putatively adaptive families and the last 10 belong to putatively neutral families . We plotted the frequency of each of the 28 TEs in the Northern vs the Southern populations for the Australian ( Figure 2A ) and the North American data ( Figure 2B ) both for putatively adaptive and for putatively neutral elements . Adaptive TEs are present at higher frequencies in the Southern compared to the Northern Australian populations ( G-test , P-value = 0 . 0001 ) while neutral TEs are not ( G-test , P-value = 0 . 3 ) . The same pattern was found when the populations collected in different years were considered independently ( G-test , P-value = 0 . 045 for adaptive TEs for both years and P-value = 0 . 3 and 0 . 4 for neutral TEs in 2007 and 2008 respectively ) . The difference between putatively adaptive and putatively neutral TEs is significant ( G-test , P-value = 0 . 0072 ) . As predicted , in North America putatively adaptive TEs tend to be more frequent in the Northern compared to the Southern populations although this pattern is marginally nonsignificant ( G-test , P-value = 0 . 0845 ) . However , compared to neutral TEs , adaptive TEs are more frequent in the Northern population ( G-test , P-value = 0 . 0235 ) . The observed patterns are consistent between continents ( G-test , P-value = 0 . 066 and 0 . 17 for adaptive and neutral TEs respectively ) . In summary , the contrasting results between putatively adaptive and putatively neutral TEs and the consistency of patterns between continents strongly suggest that putatively adaptive TEs play a role in adaptation to temperate environments . The majority of putatively adaptive TEs show population differentiation patterns consistent with adaptation to temperate climates when considered one by one in at least one of the three pairs of populations analyzed ( Table 1 ) . It is possible that most of these TEs are involved in adaptation to temperate environments: the different levels of significance could simply reflect the differences in the selective advantage these TEs confer to the organism . However , because we are interested in identifying the strongest candidates for further analysis , we focused on those TEs that are significant after correction for multiple testing [48] . There are 12 TEs that show significant patterns of population differentiation after correction for multiple testing , ten of which show the pattern expected if they are involved in adaptation to temperate environments ( Table 1 ) . Nine of these 12 TEs are putatively adaptive ( Table 1 ) and eight of them show a pattern consistent with adaptation to temperate environments . Furthermore , FBti0019386 and FBti0020119 show consistent population differentiation patterns in the two hemispheres ( Table 1 ) . As mentioned above , replicate observations of differentiation on two continents are considered to be strong evidence for selection [25] , [49] . The absence of replicate observations , however , does not preclude selection ( see Discussion ) . Although when considered together neutral TEs were not present at higher frequencies in temperate compared to tropical populations ( Figure 2 ) , there were three individual TEs that showed significant population differentiation patterns after correction for multiple testing . In all three cases the population differentiation patterns were only found in the North American populations and only two of the three showed patterns consistent with adaptation to temperate environments ( Table 1 ) . Finally , putatively adaptive TEs that did not show patterns of population differentiation could be adaptive to conditions that are common to both temperate and tropical out-of-Africa populations ( Table 1 ) . For example , FBti0019430 has been shown to confer resistance to pesticides [37] although a recent analysis suggests that this was not the selective reason for its spread ( Aminetzach Y . T . , Karasov T . L . and Petrov D . A . , unpublished data ) . We performed additional latitudinal and climatic analysis to investigate potential environmental agents that might be contributing to selection . These analyses were restricted to the six TEs that showed significant population differentiation patterns in the Australian 2008 populations . In order to establish clinal patterns , it is important to analyze flies that have been collected over a narrow period of time because clines can vary seasonally [50] . We only have such populations for the 2008 Australian cline . The location of the four Australian populations analyzed , Innisfail , Redland Bay , Coffs Harbour and Melbourne , is shown in Figure 1 . FBti0020119 was not included in the final analysis since we could not discard the confounding effect of inversion In ( 3L ) P on its frequency ( Figure 3 and Table S3 ) . We used regression analysis to test the association between population frequencies and latitude ( see Materials and Methods ) . We also analyzed the association between population frequencies and climatic variables that have previously been shown to be related to geographic variation [24]–[28]: mean maximum temperature ( Tmax ) , mean minimum temperature ( Tmin ) and mean rainfall ( see Materials and Methods ) . Three TEs show associations with latitude , Tmax or both: FBti0019386 shows significant association with latitude and Tmax , FBti0019056 shows a marginally non-significant association with latitude and a marginally significant association with Tmax and FBti0019164 shows a marginally non-significant association with Tmax ( Table 2 ) . FBti0019443 shows a significant association with both Tmin and rainfall ( Table 2 ) . Considering a 5% false discovery rate level , we would expect only one of the 20 tests performed to be significant . However , we obtained five significant tests suggesting that the majority of these tests are indeed significant . Overall , we found that four of the five TEs analyzed show weak clinal patterns . We believe that the weakness of these patterns is at least partly due to a lack of power given the small number of populations analyzed . Our results are therefore suggestive of clinal patterns but they are not conclusive . More populations need to be analyzed in order to get a better insight into which factors are relevant and to analyze the possible interactions among those factors . The change in TE population frequency associated with temperature may be reflecting temperature-dependent selection . However , this association could also be due to the selective action of other ecological variables that correlate with temperature such as food resources , the presence of competitors , predators , or pathogens or other unknown variables . In the case of association of population frequencies with rainfall , it is difficult to explain it in terms of direct selective effects of this climatic variable . Many indirect effects are however possible because rainfall influences many aspects of the physical , chemical and biological environment [25] . To learn more about the genes and traits likely to be under selection , we analyzed the available functional information for the neighboring genes of the 10 most likely TE candidates ( Table 3 ) . We used Fatigo to look for Gene Ontology terms under- or over-represented in this set of genes compared to the rest of genes in the genome [51] . We did not find significant under- or over-represented functional terms , which may be explained by the small number of genes together with the sparse functional annotation available for Drosophila genes . Some of the identified TEs are inserted into or close to genes involved in processes that have previously been shown to be under positive selection such as metabolism , defense response , or cell cycle [7] , [20] , [52] . This gives us confidence that our procedure is identifying promising cases , since these TEs were identified based exclusively on their population behavior without taking into account the description of their neighboring genes . To try to understand how TEs might affect the expression of their nearby genes , we analyzed several molecular and functional characteristics of the insertions . We restricted this analysis to the eight putatively adaptive TEs as they are likely to be the causative mutations . These eight TEs are distributed across all three major chromosomes and are unlinked with each other suggesting that they constitute independent cases of adaptive differentiation between temperate and tropical populations ( Table 3 ) . The three main classes of TEs , LTR , LINE-like and TIR , are represented in this dataset [53] . Six of the TEs are located in introns , one in a UTR and one in an intergenic region suggesting that they are involved in regulatory changes ( Table 3 ) . Indeed , for three of these TEs there is experimental evidence suggesting that they are affecting the expression of nearby genes ( FBti0019372 , FBti0020019 and FBti0020046 ) [18] . We compared the regions where they are inserted between D . melanogaster and D . simulans . We used the VISTA browser default parameters to examine the pairwise alignment of these regions [54] . For two of them , FBti0020046 and FBti0020119 , the sequence conservation drops in the region immediately adjacent to the insertion . This suggests that rather than disrupting existing regulatory elements these TEs might affect expression by adding regulatory elements themselves . The other six TEs are inserted in conserved regions suggesting that they might disrupt existing regulatory elements and/or add regulatory elements . Other than the mechanisms mentioned above , TEs inserted into introns may be affecting gene expression by driving antisense transcription , by interfering with normal splicing patterns of the mRNA or by being incorporated as exons . We tested the latter mechanism by searching for chimeric gene-TE ESTs using the modENCODE genome browser available at http://flybase . org . We found several ESTs containing these TEs but none of them contained genic sequences as well . Only one TE , FBti0019627 is inserted into the 3′ UTR of a gene , Kmn1 . 3′ UTRs regulate several aspects of gene expression such as mRNA decay and the spatial and temporal patterns of expression [55]–[57] . There are several ways by which a TE inserted in a 3′UTR can affect gene expression . We first looked for the presence of alternative poly ( A ) signals in the sequence of FBti0019627 and we did not find any . However , we identified a U-rich sequence in this TE that could be acting as a downstream element ( DSE ) . Since FBti0019627 is inserted 42 bp downstream of the poly ( A ) signal , the presence of a DSE in its sequence could be altering the place where the cleavage of the mRNA takes place [58] . We found an EST that is consistent with the use of this new DSE: it contained only a fragment of the TE ( 113 bp ) and it did not include the DSE . We also found an EST containing the whole TE ( 186 bp ) and an EST that ends 3 bp after the poly ( A ) signal . These results suggest that flies with the insertion produce three mRNAs that differ in the length of the 3′ UTR and therefore potentially differ for example in binding sites for miRNA or for RNA-binding proteins [55] , [59] . Another possibility is that these eight TEs affect gene expression by co-mobilizing DNA when they transposed . There is evidence for co-mobilization of DNA for non-LTR elements [60]–[62] and for transposons [63] . DNA mobilization by non-LTR elements is normally due to read-through transcripts that lead to co-mobilization of their 3′ flanking DNA [60] . However , mobilization of DNA in the 5′ end of the element has also been described [62] . We analyzed the TE sequences and their flanking regions and found no evidence for co-mobilization of DNA ( Table 3 ) . Two of the three LINE-like elements are flanked by target site duplications ( TSDs ) and the sequence between the TSDs only show homology to other TEs in the same family . Although the other LINE-like element , FBti0020046 , is apparently not flanked by TSDs , the analysis of its flanking regions revealed that the 14 bp 5′ to the annotated TE show homology with a INE-1 element . These 14 bp are followed by a TSD . Again the region between the TSDs only shows homology with other TEs in the same family . The four TIR elements are either full length or show internal deletions compared to their canonical element , and the only LTR element is a solo LTR suggesting that none of these TEs co-mobilized DNA . Further analyses such as looking for evidence of antisense transcription driven by these TEs or the existence of different splicing variants in flies with and without the insertions are needed in order to elucidate the mechanisms by which these TEs may be affecting gene expression .
In this work , we identified a set of TEs that are likely to be involved in adaptation during or after the spread of D . melanogaster out of Africa ( Table 1 ) . Because this species is tropical by origin , some of these adaptations may specifically be related to adaptation to temperate climates [15]–[18] . To test this prediction , we estimated the population frequencies of these TEs in populations with contrasting climates that were collected near the endpoints of two known latitudinal clines in Australia and North America ( Figure 1 ) . If some of these TEs are involved in adaptation to temperate climates we expect them to be present at higher frequencies in temperate compared to tropical populations . However , other than being caused by selection , patterns of population differentiation may simply be the by-product of non-adaptive processes related to population structure and history [19] , [23] . To distinguish between these two possibilities , we looked for patterns of population differentiation not only in the set of TEs identified as putatively adaptive but also in a comparable set of neutral TEs . While drift , isolation by distance or historical processes should affect the patterns of variation across the entire genome and therefore affect both neutral and adaptive TEs , we expect selection to affect only the adaptive loci . As mentioned above , the two sets of TEs , adaptive and neutral , are comparable: they are all present at low frequencies or absent in AF populations and are present in all the NA populations analyzed ( Table S1 ) . However , neutral TEs come from families with selection coefficients not significantly different from zero while adaptive TEs belong to families with selection coefficients significantly negative or from families for which we do not have a clear evidence of neutrality ( Table S2 ) . In addition , while patterns of polymorphism around the adaptive TEs show signatures of selection , the regions flanking neutral TEs suggest that these TEs have increased in frequency neutrally [18] , [44] . As predicted if selection is the cause of population differentiation patterns , TEs classified as putatively adaptive are present at higher frequencies in temperate compared to tropical populations while putatively neutral TEs are not ( Figure 2 ) . Furthermore , these patterns of population differentiation are consistent between years and between continents ( Figure 2 ) . If population differentiation patterns were entirely random , they would be unlikely to occur in both hemispheres [25] , [49] . Therefore , the contrasting results for adaptive and neutral TEs and the repeatability of population differentiation patterns across continents strongly suggest that selection is responsible for the observed population patterns . We used a maximum likelihood approach to identify the most likely TE candidates to be involved in adaptation to temperate environments . After correcting for multiple testing , we found 12 TEs with significant patterns of population differentiation . Ten of them were present at higher frequencies in temperate populations as expected if they are involved in adaptation to temperate environments ( Table 3 ) . Two of these 10 TEs show parallel patterns on the two continents . However , the absence of replicate observations for the other TEs does not preclude selection . It could simply reflect a lack of statistical power due to the limited number of strains analyzed . The difference between continents may also be due to differences in the climatic gradients . Indeed , the latitudinal range spanned by Australian populations is 17° to 37° while North American populations range from 27° to 44° . The absence of parallel clines on the two continents could also be related to differences in the genetic background of Australian and North American populations . These two continents had very different histories of colonization: D . melanogaster spread into North America in the past few centuries while it spread into Australia only in the last 100 years [13] , [64] . Finally , another possibility is that these differences are due at least in part to different patterns of isolation by distance in the two continents . However , there are several lines of evidence that suggest that gene flow among populations along each one of these clines is high [19] , [21]–[22] . Our results , based on the analysis of the TEs annotated in the sequenced D . melanogaster strain , suggest that adaptation to temperate environments is widespread in Drosophila . Although this strain has been described as a “typical” D . melanogaster strain [53] , it would be interesting to analyze the population dynamics of TEs annotated in other strains . The sequencing of 192 D . melanogaster strains currently in progress will facilitate this analysis ( http://www . hgsc . bcm . tmc . edu ) . Putatively adaptive TEs are likely to be the actual causative adaptive mutations , and not just linked to a nearby adaptive mutation . If the population differentiation of a putatively adaptive TE were due to linkage to a nearby adaptive mutation , then the direction of the differentiation should only depend on whether the adaptive mutation emerged on a haplotype with the TE , or one without it . Both directions should therefore be possible . If , however , the TE is the adaptive mutation itself , then the TE's frequency is always expected to be higher in the more temperate population . Indeed , eight out of nine putatively adaptive TEs that showed significant population differentiation are present at higher frequencies in temperate populations ( Table 1 ) . On the other hand , among the three putatively neutral TEs two were more frequent in the temperate population and one was more frequent in the tropical population . Although the numbers are small , this pattern is consistent with the population differentiation of the neutral TEs being due to linkage to a causative mutation . Additional evidence comes from the previously mentioned studies by González et al . 2008 [18] and Macpherson et al . 2008 [44] , where patterns of nucleotide variability around TEs were analyzed . In every investigated instance , the high frequency of putatively adaptive TEs was found to be consistent with positive selection while putatively neutral TEs were always confirmed to have increased in frequency neutrally . It is also possible that some of the TEs in the neutral families are actually adaptive . To elucidate whether TEs in neutral families are linked to an adaptive mutation or adaptive themselves , and to completely discard the existence of linked adaptive mutations in the vicinity of adaptive TEs , the flanking regions of each one of these 10 TEs should be analyzed . The future availability of the whole genome sequences for 192 D . melanogaster strains should facilitate this analysis ( http://www . hgsc . bcm . tmc . edu/ ) . In any case , the genomic regions where TEs showing significant population differentiation are inserted represent strong candidate regions to be involved in adaptation to temperate climates and deserve further study . They also add significantly to the set of candidate loci already available to study and monitor the impact of climate change on populations [65] . The adaptive TEs reported here span the range of TE diversity in D . melanogaster ( Table 3 ) . The putatively affected genes are also highly diverse in terms of their molecular and cellular functions ( Table 3 ) . It would be natural to assume that the resulting adaptive effects are diverse as well and have evolved in response to multiple unrelated selective pressures associated with the migration out of Africa . In contrast , our results suggest that the selective behavior of these adaptive TEs can be largely explained by latitude . These results might thus be revealing cryptic simplicity of the adaptive process in D . melanogaster – much of it might be about latitude – and challenge us to understand how diverse genes and processes can all generate adaptive effects in response to a related set of selective pressures . Taking into account both the functional information of the genes located nearby ( Table 3 ) and the information on the possible selective factors reported in this work ( Table 2 ) , we can construct plausible hypotheses about the phenotypic consequences of the adaptive insertions . For example FBti0019386 , which shows population differentiation in both continents , is inserted into a conserved region in the first intron of sra ( Table 3 ) . Variation in the population frequency of this TE is associated with latitude and Tmax . Changes in the expression level of this gene critically affect ovulation and female courtship [66] . We therefore speculate that the insertion of FBti0019386 into sra might affect fecundity specifically at low temperatures . FBti0019443 is inserted in a circadian-regulated gene , CG34353 [67] . The population frequency of this TE is associated with Tmin and rainfall . Some authors have suggested that differences in latitude challenge the circadian clock because of the associated changes in temperature and photoperiod [68] . There is no information about the biological process or the molecular function of CG34353 , but we suspect that temperature and photoperiod may play a role in its evolution . Components of fitness such as male and female fertility , survival rates throughout development or stress resistance should be analyzed under different temperatures and photoperiods in order to link this insertion to its phenotypic consequence . Another example is FBti0020046 , which is inserted in the intergenic region between Jon65Aiv and Jon65Aiii . Both genes have been associated with odor-guided behavior [69] . One important factor involved in the ability to colonize new habitats is the capacity of using different food resources [70] . The shift from a natural source to a domesticated fruit in the fly Rhagoletis pomonella is associated with , and perhaps causally related to , a shift in olfactory preferences [71] . We can therefore speculate that the insertion of this TE in the intergenic region of genes involved in olfactory-guided behaviour played a role in the ability of D . melanogaster to use different food resources . However , olfactory behavior is also involved in other processes such as avoidance of environmental toxins and predators , mate selection or reproduction [72] . Any of these processes could therefore have been affected by changes in the expression of these genes . Pleiotropic effects of adaptive mutations , as the ones just described , can severely complicate the identification of the phenotypic trait on which selection is acting even when we have clues about the potential interesting phenotypes . This is exemplified by the analysis of the Bari-Jheh insertion previously carried out in our lab [38] . This TE is inserted between genes involved in Juvenile Hormone metabolism . Juvenile Hormone has major effects on various aspects of development and life history traits [73] . Although we were able to find subtle consequences of this insertion on life history traits that were consistent with the reduced expression of the nearby genes , we could not pinpoint which of the phenotypic effects of the insertion was adaptive . Another factor that can severely complicate the detection of selection in experimental populations is that the observed changes in TE frequency may be explained not by a single environmental variable but by a combination of them [74]–[75] . Finally , although having identified both the phenotypic trait and the relevant environmental conditions , the fitness differences between the flies with and without the insertion might be too small to be experimentally detected [76] . Because adaptive mutations might be difficult to study at the phenotypic level , reverse population genomics analyses as the one described in this work which allow detection of a consistent response of a set of adaptive mutations to the environment are necessary to obtain a comprehensive picture of adaptation . We found patterns of population differentiation associated with TE insertions that are consistent with the model of an ancestral African species adapting to temperate climates . Our results suggest that adaptation to temperate climates in Drosophila is widespread with TEs playing a significant role in this adaptation . Considering the variety of TEs in our set , it is remarkable that their adaptive effects seem to be consistently associated with climate-related selective pressures , potentially revealing cryptic simplicity of the adaptive process in D . melanogaster . We identified the most likely TE candidates and integrated information on population behavior , possible environmental selective agents and both molecular and functional information of the nearby genes to infer the plausible phenotypic consequences of these insertions in the environment in which they evolved . Our long term objective is to experimentally measure the phenotypic differences between flies with and without these insertions which will help us to understand their adaptive effects . Both reverse population genomic analyses of the kind described in this work and functional analysis that link the identified mutations to their adaptive phenotypes are necessary to arrive at a comprehensive picture of adaptation .
In a previous work , we used the Release 3 annotation of TEs in the D . melanogaster genome to design primers to check the population frequency of individual TEs in the genome [18] . Using a pooled-PCR approach , we obtained frequency data for a total of 902 TEs in five North American populations ( 64 strains combined in six pools ) and one sub-Saharan ( Malawi ) African population ( 11 strains combined in one pool; for details see González et al . ( 2008 ) [18] ) . Release 5 corrected the annotations for a large number of TEs relative to Release 3 and the PCR results obtained previously were updated accordingly ( Petrov , D . A . , Fiston-Lavier , A . -S . , Lipatov , M . , Lenkov , K . and González , J . , unpublished data ) . In this work , we used the updated version of our database , containing information for 763 TEs , to re-run the query designed to search for TEs that may have contributed to adaptation during and/or after the migration of D . melanogaster out of Africa . Specifically , we looked for insertions that ( 1 ) were present in all six North American pools ( 199 TEs ) , ( 2 ) were not fixed in the African pool ( 85 TEs ) and ( 3 ) were present in regions of the genome with a recombination rate larger than zero [39] ( 45 TEs , Table S1 ) . Our results are not very sensitive to the exact value of these cutoffs . If we consider the TEs present in five North American pools instead of six pools , the number of TEs varies from 199 to 226 . However , the estimated population frequency for TEs present in 5 pools is only 5 . 1%–35% [77] . Since these TEs might be at low population frequencies we decided to focus on those ones present in all six North American pools ( estimated population frequency 11%–100%; [77] ) . Varying the recombination rate cutoff between 0 cM/Mb and 1 cM/Mb only changes the number of TEs from 45 to 42 ( Table S1 ) . A total of 45 TEs matched the above criteria ( Table S1 ) . We then estimated the frequency of those TEs in the Malawi population using PCR with individual strains ( see below; Table S4 ) . We filtered out TEs present in ≥30% of the strains analyzed because those TEs are less likely to be involved in adaptation to the out-of-Africa environments and we ended with a dataset of 32 TEs ( Table S1 ) . Again , varying the cutoff value for example from 30% to 15% only changes the results marginally ( from 32 TEs to 30 TEs; Table S1 ) . Although all the selected TEs were found to be present at low frequencies in the Malawi population , it is possible that they are present at high frequencies in other sub-Saharan African populations because only 11 strains were sampled and because there might be substantial substructure in the D . melanogaster population in sub-Saharan Africa [78] . In previous works in our laboratory , we extended the analysis of TEs found to be absent or present at low frequencies in Malawi to three other sub-Saharan populations , two from Zimbabwe and one from Kenya ( Table S4; [18] , [79] ) . We found that putatively adaptive TEs that were absent or present at low frequencies in Malawi were also absent or present at low frequencies in the three additional African populations analyzed ( Table S5 ) . Similar results were obtained for putatively neutral TEs ( Table S5 ) . Overall , the analyzed TEs seem to have increased in frequency either during or after the expansion out of Africa . In addition to the African D . melanogaster stocks mentioned above , the following populations collected along the Australia and North America East coast clines were analyzed in this study ( Figure 1; Table S4 ) : two Australian populations collected in 2007 in Innisfail in North Queensland and Yering Station in South Victoria . Four Australian populations collected in 2008 in Innisfail in North Queensland , Redland Bay in Queensland , Coffs Harbour in New South Wales and Melbourne in South Victoria . Two North American populations collected in Rocky Ridge in Bowdoinham , Maine , USA , and Watch Me Grow in Ft . Pierce , Florida , USA . The presence and/or absence of the TEs analyzed in this work was determined using PCR . Two different sets of primers were used: one set was intended to assay for the presence of the TE and consisted of a “Left” primer which lay within the TE sequence and a “Right” primer that lay in the flanking region to the right of the insertion . We expect this PCR to give a band only when the element is present . The other set of primers was intended to assay for the absence of the element and consisted of a “Flank” primer which lay in the flanking region to the left of the element and the “Right” primer mentioned above . In this case , the absence of the TE should give a shorter “absence” band and the presence of the TE should give a longer “presence” band . We assumed that the “presence” band is unlikely to be amplified if the TE is longer than 800bp . Populations were sampled only a few generations after they were collected in the field . This is important since TE frequencies may change due to laboratory selection or laboratory bottlenecks and therefore laboratory frequencies might not be representative of the field frequencies [50] . Moreover , we did not find significant differences in the population frequencies of the analyzed TEs between the two years in the Northern or Southern Australian populations ( Table S6 ) . In any case , we expect the changes due to lab conditions to affect the Northern and Southern populations similarly ( and the adaptive and neutral TEs similarly as well ) since these populations were maintained under the same laboratory conditions . We estimated the frequency of each one of the 32 TEs in each population using PCR with individual strains . For each population ( Figure 1 ) , we sampled one female per isofemale line for a total of 22–24 lines ( Table S4 ) . We then evaluated the heterogeneity of the frequencies between the Northern and the Southern populations using a maximum likelihood procedure . Strains are not fully isogenized as evidenced by the heterozygosity of many TEs for presence and absence in many strains ( data not shown ) . We assumed that each tested strain effectively contains two different haploid genomes and that different strains within a tested set come from a panmictic population . The data for each TE in each population come in the form { , , } where is the number of strains homozygous for the presence of the TE , is the number of strains heterozygous for the presence of the TE , and is the number of strains that are homozygous for the absence of the TE . The log-likelihood of observing such data conditional on the frequency p is: ( 1 ) The is maximized at the value : ( 2 ) To determine whether the frequencies in the Northern and Southern populations are different from each other we compare the log-likelihoods of two models . Under H1 we assumed that the frequencies in the two populations are different and estimate them using equation 2 using the data that come from each population separately . We also calculated the two corresponding maximum log-likelihoods . Under H2 we assumed that the frequency of the TE is the same in both populations and estimate this frequency using equation 2 with the combined data from the two populations . We also estimate the maximum log-likelihood under H2 . The heterogeneity is detected when the difference between the sums of the two maximum log-likelihood values under H1 and the maximum log-likelihood value under H2 ( denoted by ΔL ) is greater than 3 . 84 corresponding to the 5% critical value of the χ2 test with one degree of freedom . Three of the four cosmopolitan inversions described in D . melanogaster have been characterized at the molecular level . We checked for the presence of these inversions in all the strains analyzed using the following primers: for inversion In ( 2L ) t we used the primers described in Andolfatto et al . 1999 [80] . For inversion In ( 3R ) Payne we used the primers described in Matzkin et al . 2005 [81] . Finally , for inversion In ( 3L ) Payne we used the distal breakpoint sequences described in Wesley and Eanes 1994 [82] to design primers to check for the presence and the absence of the inversion . Primer pair 5′-CCGGATGGACCACATAGAAC-3′ and 5′-CATTCTGGGCCTTATCATCT- 3′ amplifies the standard , but not the inverted chromosome and primer pair 5′-CCGCAAACGAACACTTA-3′ and 5′- GATTATGGACCTAATGAAAGC-3′ amplifies the inverted , but not the standard chromosome . Associations between TE frequencies and latitude were examined using a regression analysis . Only the results of linear regressions are presented because nonlinear patterns were not detected when latitude was treated as a quadratic term . Each of the four Australian populations analyzed ( Innisfail , Redland Bay , Coffs Harbour and Melbourne ) was treated as a single datapoint . The frequencies of the five analyzed TEs in these four populations are given in Table S3 . All frequency data were angular transformed before performing the regression analyses . Climatic data for weather stations adjacent to collection sites were obtained from the Australian Bureau of Meteorology ( www . bom . gov . au; Table S7 ) . The two temperature variables used in the analysis were maximum and minimum temperature . In addition , we also considered rainfall . For all three climatic variables , 20-year averages were used because selection coefficients are small , and frequencies therefore are affected more by long-term climatic patterns than by short term trends [25] . The association between TE frequencies and the different climatic variables was also analyzed using regression . Maximum likelihood estimates for selection coefficients of TE families were derived by comparing observed TE frequencies to those expected under mutation-selection balance , using a simplified version of the approach presented in González et al . 2008 [18] . We model TEs to have codominant fitness: individuals homozygous for a TE have fitness 1+s and heterozygotes have fitness 1+s/2 . The expected population-frequency distribution of a TE family for a panmictic , constant-sized population is then given by ( 1 ) The normalization factor c is defined via the condition ∑xρ ( x|s , N ) = 1 , where the sum is taken over the set {1/ ( 2N ) , … , ( 2N−1 ) / ( 2N ) } of possible population frequencies x in a diploid population of size N . All TEs in our analysis were originally ascertained in a single sequenced strain [53] . The probability to observe a TE at population-frequency x is therefore ( 2 ) Our TE frequency data comes in the form Mi = {m1 , m2 , m3 , m4 , m5 , m6} , where m1 is the number of North American strain pools where the element is absent , m2 is the number of pools where it is polymorphic , and m3 is the number of pools where it is fixed . Counts m4 and m5 give the numbers of pools with partial information - those where the element is either absent or polymorphic , and those where the element is either polymorphic or fixed . The numbers of strains vary between 8 and 12 for different pools and in our analysis we adopted an intermediate value of 11 strains per pool , close to the pool average . The error rates of a pool appearing to have the element as either absent or fixed , while it is actually polymorphic , were estimated to be e1≈0 . 042 and e2≈0 . 010 , respectively ( Petrov , D . A . , Fiston-Lavier , A . -S . , Lipatov , M . , Lenkov , K . and González , J . , unpublished data ) . For those pools where we can distinguish perfectly between the three classifications , the probability of m1 pools classified as absent , m2 as polymorphic , and m3 as fixed , is ( 3 ) The probability of finding that the element is absent or polymorphic in m4 pools is ( 4 ) The probability of finding that an element is polymorphic or fixed in m5 pools is ( 5 ) Summing the product of the probabilities ( 2 ) – ( 5 ) over all possible frequencies x yields the probability of a particular observation Mi for an individual TE , ( 6 ) The likelihood-function for the selection coefficient s of a TE family is then defined by multiplying the probabilities ( 6 ) of all its elements ( and thus assuming independence of individual TEs ) , ( 7 ) We note that in the regime N≫1 and |s|≪1 , ( 7 ) becomes effectively a function of the product Ns and we therefore used a fixed value of N = 104 for our analysis . Maximum likelihood estimates and their confidence intervals were obtained numerically by simulated-annealing . The 95% confidence intervals around maximum likelihood estimates Ns* were thereby calculated by solving for Ns such that log[L ( Ns* ) /L ( Ns ) ] = 2 . 512 , where we assumed that log-likelihood ratios in our analysis follow a χ2 distribution with one degree of freedom . | The potential of geographic studies of genetic variation for the understanding of adaptation has been recognized for some time . In Drosophila , most of the available studies are based on a priori candidates giving a biased picture of the genes and traits under spatially varying selection . In this work , we performed a genome-wide scan of adaptations to temperate climates associated with Transposable Element ( TE ) insertions . We integrated the available information of the identified TEs and their nearby genes to provide plausible hypotheses about the phenotypic consequences of these insertions . Considering the diversity of these TEs and the variety of genes into which they are inserted , it is surprising that their adaptive effects are consistently related to temperate climate-related factors . The TEs identified in this work add substantially to the markers available to monitor the impact of climate change on populations . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"evolutionary",
"biology",
"genetics",
"and",
"genomics"
] | 2010 | Genome-Wide Patterns of Adaptation to Temperate Environments Associated with Transposable Elements in Drosophila |
Oncolytic adenoviruses , such as ONYX-015 , have been tested in clinical trials for currently untreatable tumors , but have yet to demonstrate adequate therapeutic efficacy . The extent to which viruses infect targeted cells determines the efficacy of this approach but many tumors down-regulate the Coxsackievirus and Adenovirus Receptor ( CAR ) , rendering them less susceptible to infection . Disrupting MAPK pathway signaling by pharmacological inhibition of MEK up-regulates CAR expression , offering possible enhanced adenovirus infection . MEK inhibition , however , interferes with adenovirus replication due to resulting G1-phase cell cycle arrest . Therefore , enhanced efficacy will depend on treatment protocols that productively balance these competing effects . Predictive understanding of how to attain and enhance therapeutic efficacy of combinatorial treatment is difficult since the effects of MEK inhibitors , in conjunction with adenovirus/cell interactions , are complex nonlinear dynamic processes . We investigated combinatorial treatment strategies using a mathematical model that predicts the impact of MEK inhibition on tumor cell proliferation , ONYX-015 infection , and oncolysis . Specifically , we fit a nonlinear differential equation system to dedicated experimental data and analyzed the resulting simulations for favorable treatment strategies . Simulations predicted enhanced combinatorial therapy when both treatments were applied simultaneously; we successfully validated these predictions in an ensuing explicit test study . Further analysis revealed that a CAR-independent mechanism may be responsible for amplified virus production and cell death . We conclude that integrated computational and experimental analysis of combinatorial therapy provides a useful means to identify treatment/infection protocols that yield clinically significant oncolysis . Enhanced oncolytic therapy has the potential to dramatically improve non-surgical cancer treatment , especially in locally advanced or metastatic cases where treatment options remain limited .
Therapeutic options for most patients with locally advanced or metastatic cancer are limited . Surgery is often not an option for these patients because the cancer has diffusely spread , and currently available non-surgical treatments for most solid malignancies have insufficient impact on survival rates . Therefore , novel treatment strategies that incorporate the molecular composition of individual tumors are urgently needed . Conditionally replicating oncolytic adenoviruses are designed to target and lyse cells with specific aberrations , showing promise as a new non-surgical treatment strategy [1] , [2] . The selective replication of viruses in cancer cells leads to destruction of infected cells by virus-mediated lysis . Consequently , the released viral progenies spread through the tumor mass by infecting neighboring cancer cells , resulting in self-perpetuating cycles of infection , replication , and oncolysis [3] , [4] . As this approach relies on viral replication , the virus can , theoretically , self-amplify and spread in the tumor from an initial infection of only a few cells . ONYX-015 is an oncolytic adenovirus that lacks the E1B-55K gene product required for p53 degradation and therefore was predicted to selectively replicate in tumor cells with inactive p53 pathways [5] . Later studies revealed that p53-independent effects may function as regulators of virus replication supporting the therapeutic application of ONYX-015 not only in p53-defficient tumors , but also in tumors with wild-type p53 [6] , [7] . ONYX-015 has been tested extensively; evidence for specific oncolysis was found in several clinical trials and in various tumors types [8]–[11] , including recurrent head and neck [12] , colorectal [13] , ovarian [14] , and hepatobiliary [11] cancers . Although clear antitumor activity was demonstrated using ONYX-015 in murine models of cancer , both in vitro and in vivo , its clinical efficacy in human trials has failed to fulfill the high expectations that were based on animal model studies . A potential explanation for limited activity is reduced expression of the main receptor for adenoviruses , CAR , which is required for efficient virus entry into target cells . Reduced expression of the CAR protein on the cancer cell surface is possibly a result of the epithelial-mesenchymal transition [15] . In previously published work , we explored the possibility of pharmacologically up-regulating CAR in colon cancer cell lines through inhibition of signal transduction pathways involved in its repression . We were able to demonstrate that inhibition of MEK , as well as TGFβ , up-regulates CAR expression in vitro and results in enhanced adenovirus entry into the cells [15] , [16] . Although disruption of signaling through the RAF-MEK-ERK pathway restores CAR expression , it potentially interferes with the replication of ONYX-015 due to G1-phase cell cycle arrest , since the virus has demonstrated sensitivity to the cell cycle phase of infected cells [17] , [18] . Thus , optimization of this combination treatment strategy is difficult since the effects of MEK inhibitors , as well as the interaction of adenoviruses with target cells , are highly complex , dynamic , and non-linear processes . Through mechanistic modeling of cancer cells subject to MEK-inhibition and ONYX-015 infection , we seek to characterize and predict system dynamics in order to improve the efficacy of oncolytic adenovirus cancer treatment by manipulating the timing of MEK-inhibitor treatment and oncolytic adenovirus infection . Through successful test of model predictions , our goal is to elucidate in vitro strategies that could offer practical and effective means for minimizing cancer growth in vivo . Our studies provide a paradigm for the development of optimized combination therapies for cancer through experimentally validated mathematical modeling of non-intuitive behavior of cancer cells .
In order to generate sufficient experimental data quantifying the mechanistic behavior critical to predicting nonlinear dynamics , we systematically assessed CAR expression , cell proliferation , infection , cell viability , and viral replication in the presence and absence of MEK inhibitors ( namely , CI1040 ) . In agreement with our previously published work [16] , we found that disrupting the MAPK signaling pathway through pharmacological inhibition of MEK nearly doubles the number of CAR molecules per cell relative to the control ( DMSO-treated ) cells . The largest increase in receptor levels occurred 2 days after CI1040 treatment initiation ( Figure 1a ) . Such restoration also presented a tradeoff: MEK-inhibition caused G1-phase cell cycle arrest ( Figure 1b and 1c ) . Previous studies indicated that cell cycle arrest inhibits production of new virus particles and virus replication [17] . We therefore hypothesized that effective oncolytic adenovirus infection requires pre-treatment of cells with MEK inhibitor for a sufficient amount of time , providing increased receptor expression at the cell surface . To allow the cell cycle to proceed , treatment should be followed by removal of the inhibitor at the time of infection . Thus , we pre-treated cells with either CI1040 or DMSO for 2 days prior to infecting cells at multiplicities of infection ( MOIs ) of 0 . 1 , 1 , 2 , 5 , and 10 . We observed increased infection ( Figure 1d ) and found that viability of HCT116 cells pre-treated with CI1040 decreased by 60% or 65% six days post infection following MOIs of 0 . 1 and 1 ( respectively ) when compared to the DMSO control ( Figure 1e ) . At higher MOI , pre-treatment with MEK-inhibitor accelerated cell killing by as much as 3 days . In agreement with the observed enhanced cell death upon CI1040 pre-treatment , virus production also improved: virus titer increased 20-fold at MOI of 0 . 1 and 5-fold at MOI of 1 when measured five days post-infection ( Figure 1f ) . Our findings suggest that treating cells with MEK-inhibitor prior to infection increases CAR expression , arrests cells in G1 cell cycle phase , and sensitizes cells to infection such that we observe reduced viability and improved virus replication . We sought to build a model that captures the key phenotypic behavior of tumor cells responding to combinatorial therapy . We fit an ordinary differential equation ( ODE ) model to measurements of proliferation , infection , and relative cell viability , characterizing how an in vitro cancer cell population responds to MEK inhibition and ONYX-015 infection . Quantitative time course data supported development of a 4-state nonlinear ODE system with treatment- and infection-dependent parameter values ( Figure 1 ) . In this context , state variables represent observable tumor cell conditions in response to MEK-inhibition and/or adenovirus infection . System states are depicted in black bold capital font ( Figure 2 ) and reflect the nonlinear dynamic behavior of ( i ) uninfected cancer cell density , C [cells/cm2] , ( ii ) MEK-inhibition induced G1-phase arrest cell density , CG1 [cells/cm2] , ( iii ) untreated and infected cell density , IC [cells/cm2] , and ( iv ) MEK-inhibitor treated and infected cell density , ICT [cells/cm2] , where P reflects the total cancer cell population [cells/cm2] . Corresponding parameter values govern the rate at which state variables proliferate , arrest in ( release from ) the G1 cell cycle phase as a result of MEK-inhibitor treatment ( removal ) , infect , and lyse ( Figure 2 ) . Specifically , parameter σ governs the rate at which cells proliferate until they reach 100% confluence at the threshold , sat [cells/cm2] . Parameters βn govern the rate at which cells irreversibly undergo ONYX-015 infection , where n denotes whether these cells infect from a treated ( n = T or n = T·G1 ) or untreated ( n = ‘blank’ ) state . Since pre-treatment with MEK-inhibitor enhanced infection and cell killing , we presumed that cells in the arrested CG1 state were more susceptible to infection due to increased CAR expression , motivating the transition directed by parameter βT·G1 ( Figure 2 ) . Experimental data suggests , however , that a fraction of cells responding to MEK-inhibitor do not undergo cell cycle arrest ( Figures 1b and 1c ) . Despite continued proliferation , this subpopulation might exhibit increased CAR expression . For this reason , we incorporated parameter βT , allowing a subpopulation of cells ( assumed to have increased CAR expression ) to transition from the proliferating state , C , to the infected treated state , ICT , upon MEK-inhibition and ONYX-015 infection . Parameters δn govern the rate at which cells undergo oncolysis . We interpolated intermediate parameter values and used the model to predict the extent of cell death as a function of the time of CI1040 treatment initiation , the time of ONYX-015 infection , and the MOI . We employed an exhaustive search algorithm to simulate the effect of various treatment and infection protocols . This algorithm systematically evaluated every possible sequence combination of drug treatment and infection conditions ( within a defined interval ) , with the exception of media change , tw , which was set to occur 2 days after treatment . We varied CI1040 treatment initiation between days 0–3 and infection between days 0–7 . The MOI was also varied between 0 . 1 and 10 . We evaluated percent cell death on day 8 irrespective of the sequence protocol . In this context , percent cell death is defined as the complement of cell viability ( the ratio of total cell density in a simulation consisting of treatment and infection , relative to total cell density in an independent simulation omitting infection ) . In Figure 3 , we highlight drug treatment and infection protocols that yielded over 50% cell death on simulated day 8 . ( Please refer to Text S1 for additional MOI ) . Model simulations suggested that , at low MOI , the greatest efficacy of virus-mediated cell death results from MEK-inhibition that coincides with the time of infection . At higher MOI , our model predicted maximal cell killing when inhibitor treatment occurs at the time of , or soon after , infection . To experimentally validate the predictive capabilities of the model , we simulated ( Figures 4a and 4b ) and experimentally quantified ( Figures 4c and 4d ) cell viability for three distinct drug treatment and infection protocols that employed MOIs not included in the original training data: MOIs of 0 . 5 and 7 . Specifically , we compared ( i ) pre-treatment with the MEK inhibitor on day 0 followed by media change and immediate infection on day 2 , ( ii ) simultaneous drug treatment and infection on day 0 followed by media change on day 2 , and ( iii ) infection on day 0 followed by post-treatment initiation on day 2 and media change on day 4 . Cell viability was quantified daily post-infection until day 7 . The mean error between simulations and time course measurements were promising: pre-treatment simulations were within 19% of validation data for both MOIs; simultaneous treatment simulations were within 8% and 12% of validation data for MOIs of 0 . 5 and 7 , respectively; and post-treatment simulations were within 14% and 19% of validation data for MOIs of 0 . 5 and 7 , respectively . One cause of disparity between predictions and experimental data relates to modeling constraints . The ODEs are formulated such that resulting simulations cannot yield more than 100% viability post-infection . Some experimental measurements , however , reflected an initial increase in cell viability that was observed in replicate measurements . This increase may result from viral proteins that activate cellular factors and force cells to enter S-phase and replicate . Despite this lack of correspondence , experiments and simulations share similar qualitative and quantitative time course dynamics , confirming that our model is indeed predictive , and that simultaneous treatment and infection significantly improves oncolysis . To further investigate conditions that give rise to increased therapeutic efficacy , we correlated simulated cell death profiles with ( i ) cell confluency at the time of treatment , ( ii ) the proportion of cells in CI1040 mediated G1 cell cycle arrest at the time of infection , and ( iii ) cell confluency at the time of infection . Little correlation between cell death and cell confluency at the time of drug treatment was found by Pearson correlation analysis ( R = . 2; p-value≪0 . 001 ) . Additionally , little correlation was found between cell death and the proportion of cells in G1-phase arrest at the time of infection , CG1/P ( R = . 3; p-value≪0 . 001 ) . Despite these weak correlation coefficients , simulations suggest that cell killing is inversely correlated with cell density at the time of infection ( R = 0 . 6; p-value≪0 . 001 ) . Thus , greater cell density at the time of infection may decrease the efficacy of infection despite consistent treatment/infection strategies . This result is reasonable since standard protocols suggest that the confluency for infection be approximately 70–80% [19] . We experimentally validated this finding , in vitro , by measuring cell viability after infection of cells that were seeded at low ( ∼70% ) and high ( ∼100% ) confluencies ( Figure 5 ) . We found that a lower cell density at the time of infection resulted in greater cell killing , supporting the notion that cell confluency might regulate the efficacy of infection and oncolysis . Model simulations and experimental validation confirm that simultaneous treatment with MEK-inhibitor and infection is most advantageous , suggesting that alternate ( CAR-independent ) regulatory mechanisms may be responsible for enhanced oncolysis . Given the disproportionate increase in virus replication relative to infection ( Figure 1e ) , we hypothesized that enhanced CAR may not be the key factor involved in amplifying virus replication and cell death; MEK-inhibition might provide an alternate mechanism responsible for greater efficacy of infection . Since MEK inhibitor treatment leads to G1-phase cell cycle arrest in HCT116 cells , we tested the impact of cell cycle distribution on oncolysis and virus production in cells infected with various oncolytic adenoviruses . HCT116 cells were arrested in the G1-phase of the cell cycle by contact inhibition and released by re-seeding at sub-confluent densities . We then quantified the change in cell cycle distribution as cells transitioned from the G1-phase . At 7-hours after re-seeding , 80% of the cell population remained in G1-phase . At 16-hours , 80% of cells reached S-phase ( Figure 6a ) . Meanwhile , CAR expression remained unchanged throughout these cell cycle phase transitions ( Figure 6b ) . Despite constant CAR , we observed significant differences in cell killing when infection occurred at 7-hours , 16-hours , and 24-hours after re-seeding . The greatest lytic effect occurred when infection took place 7-hours after re-seeding ( Figure 6c ) , which coincided with greatest virus production ( Figure 6d ) . The marked increase in cell death and adenovirus replication suggests that the G1-S phase transition mediates adenovirus replication . These observations were confirmed with a variety of oncolytic adenoviruses , including Delta-24RGD , which is characterized by an RGD motif on the fiber knob of the adenovirus allowing for CAR-independent infection [20] . Therefore , the expression of CAR molecules on the cell surface at the time of infection does not appear to be the sole regulatory mechanism governing the efficacy of oncolytic adenovirus therapy . The detailed molecular mechanisms by which cell cycle distribution influences viral replication are currently under investigation .
A growing number of studies make use of mathematical modeling techniques to better analyze and predict increasingly complex , dynamic data . While several groups have employed computational approaches to optimize oncolytic virotherapy [21]–[25] , only two other groups have investigated the combinatorial dynamics that govern MEK-inhibitor mediated oncolytic adenovirus therapy [26] , [27] . Here , we report our findings on improved treatment strategies for oncolytic adenovirus therapy , being the first to fully integrate modeling and experiment in the same study . We performed time course measurements that confirmed previously observed CI1040-mediated CAR up-regulation and G1 cell cycle arrest [28] . Based on these findings , we postulated that treating cells with CI1040 prior to infection , followed by its removal at the time of ONYX-015 infection , would ( i ) maximize virus uptake due to increased up-regulation of CAR , and ( ii ) maximize cell death ( and consequently viral replication ) due to the release of cells from G1-phase arrest . To explore this hypothesis , we developed an ODE model that characterized the proliferation , infection , and relative cell viability of a population of cancer cells subjected to MEK inhibition and ONYX-015 infection . We simulated combinations of different timings of MEK-inhibitor treatment initiation , timings of infection , and multiplicities of infection to ascertain their combinatorial effect on oncolysis . Surprisingly , our simulations suggested that , at low MOI , the greatest efficacy of virus-mediated cell death results from MEK-inhibition that coincides with the time of infection . This scenario is particularly relevant from a clinical perspective , since exposure to low MOIs is a likely limiting factor of treatment efficacy in vivo , particularly following systemic virus administration . At higher MOI , our model predicts maximal cell killing when inhibitor treatment occurs at the time of , or immediately after , infection . We confirmed our predictions experimentally , showing that sensitizing cells via MEK-inhibition prior to infection was less effective than treatment protocols that maintained CI1040 treatment during and following ONYX-015 infection . Consistent with our findings , simulations from an independent partial differential equation free boundary problem model presented by Tao and Guo [26] suggested that greater tumor treatment is achieved when oncolytic adenovirus infection and MEK inhibitor treatment occur at the same time . Experimental validation was not carried out in that study . The accuracy between simulated time courses and validation measurements ( Figure 4 ) were striking given that validation conditions were beyond the scope of training data . More specifically , we predicted cell dynamics associated with MOIs 0 . 5 and 7; neither condition was considered in model development . Data reflecting MOIs 0 . 1 , 1 , 2 , 5 , and 10 were used for parameter estimation and subsequently interpolated to predict intermediate values . Furthermore , model fitting was based on time course measurements quantifying the effect of pre-treatment with MEK inhibitors ( or DMSO ) followed by infection , yet we are able to extrapolate cell viability dynamics for mechanistically unique protocols . Specifically , the simultaneous and post-treatment protocols involve experimental procedures that were unaccounted for in model development . Simulations also assumed that the MOI of the system remained unaffected upon removal of MEK-inhibitor by media change . This theory is accurate when treatment occurs prior to infection ( as was the case in our training data ) . When treatment occurs at the time of or after infection , it is reasonable to imagine removal of MEK-inhibitor affecting virus titer , and hence the MOI . Despite these differences , model simulations accurately predict and extrapolate the nonlinear cellular response to combinatorial treatment strategies . Further investigations of simulated predictions identified critical virus-host mechanisms responsible for enhanced combinatorial therapy . In particular , we explored how cell cycle phase affected oncolysis and virus production . Shepard and Ornelles [29] demonstrated that ONYX-015 replicates more effectively in HeLa cells when infection occurs during S-phase rather than G1-phase . Later , Zheng et al . [30] found that adenovirus E1B55K is required to enhance cyclin E expression; the failure to induce cyclin E expression due to E1B55K mutation in ONYX-015 prevents viral DNA from undergoing efficient replication in HeLa ( and other non-permissive ) cells when infected during G0-phase . In contrast , cyclin E induction is less dependent on the function encoded in the E1B55K of HCT116 cells whether the cells are in S- or G0-phase . This finding is consistent in other cancer cells that are permissive for replication of ONYX-015 . As a result , we expanded our analysis to two additional oncolytic adenoviruses: Delta-24 , which carries a deletion in the E1A region; and Delta-24RGD , which has an RGD-4C peptide motif inserted into the adenoviral fiber [20] . The latter virus is able to anchor directly to integrins , providing CAR-independent mechanisms for infection . We found greater cell killing and virus production with Delta-24RGD when cells were infected during G1 cell cycle arrest . This result verifies the existence of a regulatory pathway that governs virus production in a CAR-independent manner . The mechanism underlying virus replication in G1-arrested cells remains unclear and warrants further investigation . Model development is an ongoing process that needs to be tightly coupled with experiments in order to maximize mechanistic relevance and reflect the nonlinear complex dynamics critical to understanding and predicting biological function . However , it is important to note that our current model does not fully encompass the physiological complexities of malignant tumors in humans . It is clear that factors influencing drug distribution and elimination play a major role in this context . For example , the extent of vascular leakiness observed in tumors will impact viral extravasation [31] . The immune responses directed against oncolytic viruses or tumor cells will also impact viral anti-tumor effects [25] . Interestingly , recent in vivo experiments demonstrate that the efficacy and specificity of virus replication in tumors modulate the immune response , highlighting yet another layer of complexity [32] . Our aim is to develop multi-scale models that account for the greater dimensionality of oncolytic virus replication . Nevertheless , our current study demonstrates that dynamical mathematical models of oncolytic virus replication , tightly coupled with experimental studies , have the potential to optimize central aspects of this therapeutic approach .
The colon cancer cell line , HCT116 , was kindly provided by Dr . B . Vogelstein ( Johns Hopkins Cancer Center , Baltimore , MD ) . HCT116 cells were cultured in McCoy's 5A medium ( UCSF Cell Culture Facility , San Francisco , CA ) supplemented with 10% fetal bovine serum ( Valley Biomedical Products , Winchester , VA ) . Viruses included a wild-type adenovirus , WtD; an E1B-55K-deficient adenovirus mutant , ONYX-015; an E1A-deficient adenovirus , Delta-24 [33]; and a modified version of Delta-24 containing an RGD-4C peptide motif inserted into the adenoviral fiber knob which allows the adenovirus to anchor directly to integrins , Delta-24RGD [20] . Delta-24 and Delta-24RGD were kindly provided by Dr . J . Fueyo ( University of Texas MD Anderson Cancer Center , Houston , Texas ) . Adenoviruses were amplified in HEK-293 cells , purified using the Adenopure Purification Kit ( Puresyn , Malvern , PA ) and their titers determined using the Adeno-X Rapid Titer Kit ( Clontech , Mountain View , CA ) . ONYX/GFP , a green fluorescence protein expressing ONYX-015 was also used . For inhibition of RAF-MEK-ERK signaling , the MEK inhibitor CI1040 ( Pfizer , Ann Arbor , MI ) was used at a final concentration of 5 µM . As a control , cells were treated with DMSO ( 0 . 1% ) . For CAR staining , cells were treated with CI1040 , DMSO , or cell culture medium alone ( as stated previously ) . Over the course of 4 days , the cells were harvested daily using 0 . 05% trypsin ( UCSF , Cell Culture Facility , San Francisco , CA ) . After media change in PBS ( UCSF ) , cells were incubated for 45 minutes at 4°C with the mouse monoclonal anti-CAR antibody RmcB ( 1∶50 ) [34] . After washing , cells were incubated for 30 minutes at 4°C with a secondary antibody conjugated to Alexa 488 ( 1∶100 , Alexa Fluor 488 F ( ab ) 2 fragment of goat anti-mouse IgG , Invitrogen , Molecular Probes , Eugene , OR ) . Propidium iodide ( PI , Sigma-Aldrich Co , St Louis , MO ) was added to a final concentration of 1 µg/mL just prior to acquisition to exclude dead cells from flow cytometric analysis . Stained cells were analyzed on a FACSCalibur cytometer ( Becton Dickinson , Franklin Lakes , NJ ) . To monitor virus replication in living cells , HCT116 cells were treated with CI1040 or DMSO for 2 days and later infected with ONYX-015/GFP at multiplicities of infection ( MOI ) of 0 . 1 , 1 , 2 , 5 and 10 in infection medium; McCoy's5A medium ( UCSF ) was supplemented with 2% fetal bovine serum ( Valley Biomedical Products , Winchester , VA ) . The medium was replaced two hours later . Cells were harvested 1 to 6 days post-infection using 0 . 05% trypsin ( UCSF ) and washed once with PBS ( UCSF ) supplemented with 5% fetal bovine serum ( Valley Biomedical Products ) . GFP expression was analyzed by a C6 Flow Cytometer ( Accuri Cytometer ) . Text S1 describes methodological details for the flow cytometry experiments , including controls , in accordance with the Minimum Information About a Flow Cytometry Experiment ( MIFlowCyt ) protocol established by Lee et al . [35] . For cell proliferation , HCT116 cells were seeded in 6-well plates and immediately treated with CI1040 or DMSO ( as stated previously ) for 1 , 2 , or 3 days and harvested 1–7 days after treatment . Cells were counted using a C6 Flow Cytometer ( Accuri Cytometer ) . Treated cells were also analyzed for cell cycle distribution ( please refer to Text S1 for details concerning flow cytometry experiments ) . After treatment , the cells were collected by trypsinization , fixed in 70% ethanol , washed in PBSTB ( PBS+0 . 5% Tween 20 ( National Diagnostics , Atlanta , GA ) +0 . 1% BSA ( Sigma ) ) , and re-suspended in 350µl of PBSTB containing 0 . 6 mg/mL RNase and 30 µg/mL PI . Cells were incubated in the dark for 30 min at room temperature and then analyzed by a FACSCalibur cytometer ( Becton Dickinson ) . The data were analyzed using ModFit LT ( Verity Software House ) . Please refer to Text S1 for additional details . HCT116 cells were seeded in 96-well plates overnight and infected with WtD , ONYX-015 , Delta-24 , or Delta-24RGD at different MOI . Cell viability was measured by the CellTiter 96 Aqueous One Solution Cell Proliferation Assay ( MTS ) ( Promega , Madison , WI ) 1 to 7 days post-infection . Cell viability was expressed as percentage of the uninfected medium control ( i . e . MOI = 0 ) . Therefore , any relevant toxic effects are normalized from the relative viability measurements . HCT116 cells were density-arrested by plating at 5e5 cells/cm2 for 2 days . Cells were released from arrest by re-plating at low density , 1e5 cells/cm2 . Cell cycle was analyzed by Propidium Iodide ( Sigma ) staining as described above . CAR expression was analyzed at 7 , 16 , and 24 hours after release from arrest using RmcB antibody as described above . Synchronized cells were infected with WtD , ONYX-015 , Delta-24 , or Delta-24RGD at an MOI of 1 and subsequently measured for viability by adding Propidium iodide ( Sigma ) to a final concentration of 1 µg/mL just prior to acquisition to exclude dead cells and counting cell numbers using a C6 Flow Cytometer ( Accuri Cytometer ) . Cell viability was expressed as percentage of the uninfected medium control ( i . e . MOI = 0 ) . Viral titers of harvested cells were determined by the Adeno-X Rapid Titer Kit ( Clontech ) as described by Shiina et al . [36] . Parameters were fit to experimental measurements by minimizing the sum of squares error ( SSE ) between the simulation and the data using the genetic algorithm function in MATLAB . When multiple data replicates were available , the SSE was weighted by the inverse standard deviation of experimental measurements . A gradient search algorithm , fmincon , was used post-estimation to ensure convergence to a local minimum . Please refer to Text S1 for details concerning parameter estimation , convergence , model fitness , and interpolation methods . | Novel cancer treatment strategies are urgently needed since currently available non-surgical methods for most solid malignancies have limited impact on survival rates . We used conditionally replicating adenoviruses as cancer-fighting agents since they are designed to target and lyse cells with specific aberrations , leaving healthy cells undamaged . Highly malignant cells , however , down-regulate the adenovirus receptor , impairing infection and subsequent cell death . We demonstrated that disruption of the MEK pathway ( which is frequently activated in cancer ) up-regulated this receptor , resulting in enhanced adenovirus entry . Although receptor expression was restored , disruption of signaling interfered with adenovirus replication due to cell cycle arrest , presenting an opposing trade-off . We developed a dynamical systems model to characterize the response of cancer cells to oncolytic adenovirus infection and drug treatment , providing a means to enhance therapeutic efficacy of combination treatment strategies . Our simulations predicted improved therapeutic efficacy when drug treatment and infection occurred simultaneously . We successfully validated predictions and found that a CAR-independent mechanism may be responsible for regulating adenovirus production and cell death . This work demonstrates the utility of modeling for accurate prediction and optimization of combinatorial treatment strategies , serving as a paradigm for improved design of anti-cancer combination therapies . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"oncology",
"virology/viruses",
"and",
"cancer",
"computational",
"biology/molecular",
"dynamics",
"molecular",
"biology",
"computational",
"biology/systems",
"biology"
] | 2011 | A Dynamical Systems Model for Combinatorial Cancer Therapy Enhances Oncolytic Adenovirus Efficacy by MEK-Inhibition |
Leishmania species are sand fly-transmitted protozoan parasites that cause leishmaniasis , neglected tropical diseases that affect millions of people . Leishmania amastigotes must overcome a variety of host defenses , including reactive oxygen species ( ROS ) produced by the NADPH oxidase . Leishmania species encode three superoxide dismutases ( SODs ) : the mitochondrial SODA and two glycosomal SODs ( SODB1 and SODB2 ) . SODs are metalloenzymes that function in antioxidant defense by converting superoxide to oxygen and hydrogen peroxide . Here , we investigated a role for SODB1 in Leishmania infection of macrophages and virulence in mice . We found that a single allele deletion of SODB1 ( SODB1/Δsodb1 ) had minimal effects on the replication of axenically-grown L . major promastigotes or differentiation to infective metacyclic promastigotes . Disruption of a single SODB1 allele also did not affect L . donovani differentiation to amastigotes induced axenically , or the replication of axenically-grown L . donovani promastigotes and amastigotes . In contrast , the persistence of SODB1/Δsodb1 L . major in WT macrophages was impaired , and the development of cutaneous lesions in SODB1/Δsodb1 L . major-infected C57BL/6 and BALB/c mice was strongly reduced . The reduced disease severity in mice was associated with reduced burdens of SODB1/Δsodb1 L . major parasites in the foot at late , but not early times post-inoculation , as well as an impaired capacity to disseminate from the site of inoculation . Collectively , these data suggest that SODB1 is critical for L . major persistence in macrophages and virulence in mice .
Leishmania parasites are sand fly-transmitted protozoan parasites that cause leishmaniasis in humans and other animals . Depending on the infecting Leishmania species , a spectrum of clinical manifestations can occur , ranging from self-limiting cutaneous lesions to invasive disease involving multiple organs . Visceral leishmaniasis ( VL ) , the most severe form of the disease , is a progressive and chronic illness that , if left untreated , is nearly always fatal [1] . More than 350 million people are at risk of contracting leishmaniasis [2] , and the disease disproportionately affects the poor . It is estimated that 20 , 000 to 40 , 000 deaths occur from leishmaniasis each year , with the Global Burden of Disease Study 2015 estimating more than 20 , 000 deaths from VL [3 , 4] . There are no approved vaccines against Leishmania infection . Although existing treatments can be effective , drug toxicity , poor access to health care services , and the development of resistance are significant complications in the treatment of leishmaniasis . In sand flies , Leishmania parasites replicate extracellularly in the digestive tract as flagellated promastigotes [5] . During blood-feeding by an infected sand fly , promastigotes are deposited into the skin where they are taken up by phagocytic cells such as neutrophils and macrophages [6] . In macrophages , Leishmania parasites differentiate into amastigotes , a morphologically and metabolically distinct infective form , and replicate within parasitophorous vacuoles [7–9] . A better understanding of the mechanisms by which Leishmania parasites replicate and persist in macrophages may identify new therapeutic targets for the treatment of leishmaniasis . Superoxide dismutases ( SODs ) are metalloenzymes that function in antioxidant defense by converting superoxide to oxygen and hydrogen peroxide [10] . SODs are a large family of proteins that use the metals manganese ( Mn ) , iron ( Fe ) , nickel ( Ni ) , or copper and zinc ( Cu/Zn ) as cofactors [11] . FeSODs , which are structurally similar to MnSODs typically found in mitochondria , are dimers with each active site containing a single iron bonded to three histidines , one aspartate , and one water molecule [11] . Multiple FeSOD genes are encoded by Leishmania parasites , including SODA , SODB1 , and SODB2 . The SODB1 and SODB2 genes are arranged in tandem on the same chromosome , and sequence comparisons revealed greater than 92% identity across multiple Leishmania species ( L . chagasi , L . donovani , and L . major ) [12] . In addition , the amino acid sequences of SODB1 and SODB2 are ~90% identical , with the major difference occurring at the 3′ end that is extended by ~13 amino acids in SODB2 [12] . In L . chagasi , SODB1 and SODB2 proteins are targeted by three amino acids at the C-terminus to the glycosome [12] , a unique membrane-bound organelle found in kinetoplastid parasites that functions in glycolysis , β-oxidation of fatty acids , and other metabolic activities [13 , 14] . A single allele knockout of SODB1 resulted in decreased survival of L . chagasi in differentiated U937 cells and decreased survival of parasites exposed to paraquat , which increases intracellular superoxide levels [12] . In addition , expression of an antisense transcript against FeSOD in L . tropica and L . donovani , which may have effected expression of SODA , SODB1 , and/or SODB2 , resulted in enhanced sensitivity of parasites to the superoxide generator menadione and H2O2 in axenic cultures as well as reduced parasite survival in murine macrophages [15] . These findings suggest that FeSODs are important for the replication and/or survival of Leishmania parasites in macrophages and for protection against oxidative stress . In this study , we investigated the role of SODB1 in promoting L . major infection in murine macrophages and pathogenesis in a mouse model of cutaneous leishmaniasis . Using conventional gene disruption techniques , we generated L . major and L . donovani parasites harboring a single allele deletion of the SODB1 gene ( SODB1/Δsodb1 ) . We show that a single allele deletion of SODB1 does not affect promastigote or amastigote growth in cell culture , promastigote metacyclogenesis , or amastigote differentiation . In vitro infection assays performed in bone marrow-derived macrophages ( BMDMs ) from WT mice revealed that SODB1/Δsodb1 L . major parasites were unable to persist at levels of WT L . major . Furthermore , complementation of SODB1/Δsodb1 L . major with exogenous expression of the SODB1 protein rescued attenuation in BMDMs . Experiments in a mouse model of Leishmania pathogenesis showed that WT mice infected with SODB1/Δsodb1 L . major developed less severe lesions compared with mice infected with WT L . major . The attenuated virulence of SODB1/Δsodb1 L . major in mice was associated with reduced parasite burdens in the inoculated foot and impaired dissemination of parasites to the draining lymph node . Collectively , these findings suggest that SODB1 is essential for L . major infection and virulence and further support SODB1 as a potential therapeutic target for the treatment of Leishmania infection and disease .
This study was conducted in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals and the American Veterinary Medical Association ( AVMA ) Guidelines for the Euthanasia of Animals . All animal experiments conducted at the University of Colorado Anschutz Medical Campus were performed with the approval of the Institutional Animal Care and Use Committee ( IACUC ) at the University of Colorado School of Medicine ( Assurance Number: A3269-01 ) under protocol 00222 . Adenine ( A2786 ) , hemin ( H9039 ) , d-biotin ( B4639 ) , folic acid ( F8758 ) , 100x RPMI vitamin mix ( R7256 ) , 100x RPMI amino acid mix ( R7131 ) , hygromycin B ( H3274 ) , phleomycin ( P9564 ) , Wright-Giemsa Stain ( WG16 ) and proteinase K ( P4250 ) were acquired from Sigma-Aldrich . M199 with Hanks’ salts ( M2852 ) was acquired from US Biological Life Sciences . L-biopterin ( 11 . 203 ) was acquired from Schircks Laboratories ( Switzerland ) . G418 ( 04 727 878 001 ) was acquired from Roche . Peanut Agglutinin ( L107025 ) was acquired from Vector Laboratories . AquaVi-421 LIVE/DEAD fixable cell dye ( L34955 ) and SuperScript IV First Strand Synthesis System ( 18091050 ) were acquired from Life Technologies . Ambion PureLink RNA mini kit ( 12183025 ) , random primers ( 48190011 ) , TRIzol Reagent ( 15596018 ) , and PureLink DNase ( 12185010 ) were acquired from Life Technologies . L . major strain NIH Friedlin V ( MHOM/IL/80/FN ) was obtained from BEI Resources ( NR-48815 ) . L . donovani 1S2D sub-strain LdBob was generated as previously described [16] . L . major promastigotes were maintained at 26°C in complete M199 culture media ( 1x M199 , 40 mM HEPES pH 7 . 4 , 100 μM adenine , 4 μM biotin , 7 . 6 μM hemin , 50 μg/ml penicillin , 50 U/ml strep , 8 μM L-biopterin and 10% FBS ) . LdBob promastigotes were maintained at 26°C in complete proM199:RPMI culture media ( 1x M199 , 25 mM HEPES pH 6 . 9 , 12 mM NaHCO3 , 1x L-glutamine , 0 . 1 mM adenine , 23 μM folic acid , 7 . 6 μM hemin , 50 U/ml penicillin , 50 μg/ml streptomycin , 1x RPMI vitamin mix and 10% FBS ) . To induce axenic amastigote differentiation , LdBob promastigotes were seeded into amaM199:RPMI culture media ( 1x M199 , 27 . 5 mM MES , 25 mM NaHCO3 , 1x L-glutamine , 0 . 1 mM adenine , 23 μM folic acid , 7 . 6 μM hemin , 50 U/ml penicillin , 50 μg/ml streptomycin , 1x RPMI vitamin mix , and 13% FBS ) , adjusted to pH 5 . 5 with HCl and maintained at 37°C with 5% CO2 as previously described [16] . To enumerate parasite cell numbers in culture , cells were fixed in 2% paraformaldehyde ( PFA ) and counted via hemocytometer . To generate SODB1/Δsodb1 L . major and L . donovani parasites , gene deletion constructs were developed . L . major and L . donovani sequences were acquired from the Kyoto Encyclopedia of Genes and Genomes ( KEGG ) ( http://www . genome . jp/kegg-bin/show_organism ? org=lma and http://www . genome . jp/kegg-bin/show_organism ? org=ldo ) , including coding regions and flanking intergenic regions ( IGRs ) for L . major SODB1 ( XM_001685449 ) and SODB2 ( XM_001685450 ) , as well as L . donovani SODB1 ( XM_003863559 ) and SODB2 ( XM_003863560 ) . The IGRs upstream and downstream of the SODB1 coding sequence were amplified from L . major and L . donovani genomic DNA ( gDNA ) by PCR ( S1 Table ) . The 5′ arm for L . major and L . donovani constructs contained the IGR immediately upstream ( 1 , 428 bp or 1 , 064 bp , respectively ) of the SODB1 start codon . The IGR for the 3′ arm of the L . major and L . donovani constructs contained the IGR immediately downstream of the SODB1 stop codon ( 1 , 064 bp or 1 , 303 bp , respectively ) . The 5′ and 3′ arm amplicons were digested and sequentially cloned into the pBluescript SK ( + ) vector ( pSK ( + ) -sodb1Lm or pSK ( + ) -sodb1Ld ) . G418 or phleomycin drug resistance genes were subcloned between the 5′ and 3′ arms of the pSK ( + ) -sodb1Lm and pSK ( + ) -sodb1Ld targeting constructs . Prior to transfection , pSK ( + ) -sodb1Lm and pSK ( + ) -sodb1Ld targeting constructs were linearized by PvuI digest and column purified , yielding a 1 , 050 bp fragment of pSK ( + ) plasmid backbone , and a 5 , 350 bp fragment containing the SODB1 targeting construct DNA . Transfection was performed as previously described [17] . Briefly , mid-log phase promastigotes were resuspended at 2 x 108 cells/ml in cytomix electroporation buffer ( 120 mM KCl2 , 150 μM CaCl2 , 10 mM KH2PO4 , 25 mM HEPES pH 7 . 6 , 2 mM EDTA and 5 mM MgCl2 ) . Ten μg of linearized pSK ( + ) -sodb1Lm or pSK ( + ) -sodb1Ld was added to 500 μl of cell suspension , parasite-DNA mixtures were electroporated in 4 mm cuvettes ( two pulses of 1500V/25 μF ) , and transferred to 10 ml culture for 24 h in the absence of selection . Transfectants were selected on M199:Lmaj or proM199:RPMI agar culture plates containing respective drug for L . major or L . donovani transfectants , respectively . For drug selection , phleomycin ( 10 μg/ml ) , G418 ( 10 μg/ml ) , and/or hygromycin ( 50 μg/ml ) were added to selection plates , and all subsequent cultures . Drug-resistant clones were genotyped for gene deletion by PCR reactions designed to amplify fragments that span the endogenous genomic locus and plasmid construct boundaries ( S1 Fig ) . All PCR fragments were sequenced to confirm precise construct integration within the SODB1 genomic locus . All parasite parent strains and clones were subsequently passaged through BALB/cJ mice to maintain virulence as previously described [18] . Log-phase promastigotes ( L . major and L . donovani ) or amastigotes ( L . donovani ) were homogenized in TRIzol reagent ( Life Technologies ) and total RNA was isolated using a PureLink RNA mini kit ( Life Technologies ) . To limit DNA contamination of the RNA samples , an on-column DNase digestion step was performed . Full-length cDNA was generated from purified RNA using SuperScript IV reverse transcriptase and random primers . Quantitative PCR was performed on cDNA using SODB1- or SODB2-specific forward and reverse primers and TaqMan probes ( S1 Table ) . For quantification , standard curves were generated from plasmids encoding SODB1 or SODB2 coding sequences . To ensure amplification specificity , no template and no reverse transcription controls for all samples were run in parallel . Log-phase parasite cultures were collected and fixed in 2% PFA . Cell suspensions were washed in H2O , pipetted onto glass coverslips , and air-dried . Dried coverslips were subjected to Wright-Giemsa stain for 2 min , washed with H2O for 5 min , and air-dried . Coverslips were mounted with Permount Mounting Media ( Fisher SP15 ) and visualized on a Zeiss AxioPlan II microscope at 100x magnification . The Leishmania gene overexpression construct pXG-HYG ( pXG ) was used for episomal gene overexpression analysis as previously described [19–21] . The L . major SODB1 coding sequence was amplified by PCR using a 5′ primer containing the human influenza hemagglutinin ( HA ) sequence ( S1 Table ) . Amplified ( HA ) -SODB1 was then cloned into pXG . SODB1/Δsodb1 L . major parasites were transfected with pXG or pXG- ( HA ) -SODB1 plasmids by electroporation and drug resistant clones were isolated . Overexpression of ( HA ) -SODB1 was confirmed in whole cell lysates of stationary phase parasites by SDS-PAGE and Western blot analysis using primary anti-HA ( clone HA-7 ) and secondary goat-αmouse IgG-HRP ( Fisher 31430 ) . Equal protein loading was confirmed by western blot analysis using anti-α-Tubulin ( clone B-5-1-2 ) . To generate BMDMs , bone marrow cells were aseptically flushed from mouse femurs and tibias , subjected to red blood cell lysis , and plated in DMEM ( Gibco; 11995–065 ) supplemented with 10% FBS , 5% horse serum , and 20% L-cell conditioned media for 7 days . BMDMs were plated in 24-well plates at 5 x 105 cells per well and cultured for 24 h . Cell cultures were inoculated with stationary phase L . major promastigotes or PNA- metacyclic promastigotes at an MOI of 10 parasites/BMDM or 1 parasite/BMDM , respectively . Plates were centrifuged for 5 min at 200 x g to synchronize infection and incubated at 37°C for 4 h . After the 4 h infection , free parasites were removed by washing three times with PBS , and cultured in DMEM supplemented with 10% FBS and 100 U/mL penicillin/streptomycin . Isolation of gDNA from axenic parasite cultures or from infected BMDMs was carried out by proteinase K digestion and isopropanol precipitation . Briefly , cells were resuspended in 400 μl DNA lysis buffer ( 0 . 1 M Tris pH 8 . 0 , 0 . 2 M NaCl , 5 mM EDTA , 0 . 4% SDS , and 0 . 2 mg/ml proteinase K ) and incubated at 56°C overnight . Following digestion , gDNA was isolated by isopropanol precipitation and resuspended in H2O . To quantify total parasite numbers by qPCR , forward and reverse primers and a TaqMan probe directed against the L . major ubiquitin hydrolase ( UbHyd; XM_003722265 ) gene were used ( S1 Table ) . Total parasite genomes were extrapolated from a standard curve generated from L . major gDNA . To quantify total macrophage genomes , forward and reverse primers and a TaqMan probe directed against the C57BL/6J gapdh gene ( BC023196 ) were used ( S1 Table ) . Total genomes were extrapolated from a standard curve generated from gDNA isolated from BMDMs . Eight week-old male and female WT C57BL/6J ( 000664 ) and eight week-old female WT BALB/cJ ( 000651 ) mice were purchased from Jackson Laboratories . Eight week-old Nos2-/- C57BL/6J ( 002609 ) mice were obtained from Jackson Laboratories and bred in house , and combinations of male and female Nos2-/- mice were included in all studies . Infectious metacyclic L . major promastigotes were isolated by agglutination assay as previously described [22 , 23] . Briefly , stationary phase promastigotes were washed twice in PBS , resuspended in 50 μg/ml peanut agglutinin ( PNA ) in PBS , and incubated at RT for 30 min . Agglutinated parasites were pelleted by centrifugation at 40 x g for 5 min , and infectious metacyclic promastigotes remaining in the supernatant were retrieved and washed twice in PBS . 104 infectious metacyclic promastigotes in 40 μl of PBS were inoculated into the dorsal side of the right hind foot . Lesion development was assessed by measuring foot height and width using digital calipers . The area represents the product of the height and width , and the % of initial area represents the quotient ( area of dpi/area of initial ) . To quantify parasite burden in mouse tissues , limiting dilution assay of tissue homogenates was performed as previously described [24–26] . Briefly , the foot ( including toes and skin ) and the draining popliteal lymph node ( popLN ) were homogenized in M199:Lmaj media with a Roche MagNA Lyser . Tissue homogenates were seeded in duplicate in 96-well plates in a 4-fold dilution series and incubated at 26°C . At 7–14 days , the last dilution that contained motile promastigotes was determined . Inoculated foot tissue was harvested from mice infected with WT or SODB1/Δsodb1 L . major into 10% neutral buffered formalin . Tissue was processed , paraffin-embedded , and longitudinal tissue sections were stained with hematoxylin and eosin ( H&E ) . H&E stained sections were scored for inflammation . The “extent of inflammation” was determined by the number of 100x fields required to cross the area of inflammation in a tilling fashion ( each 100x field was assigned a value of 1 ) . The “relative quantity” of inflammation was determined by examining the dermis and hypodermis tissues and estimating whether inflammatory cells filled 0% , 5% , 10% , 25% , or ≥50% ( and assigned values of 0 , 1 , 2 , 3 , 4 , 5 ) of the space between the normal connective tissue cells . The total inflammation was determined by multiplying the “extent of inflammation” by the “relative quantity” of inflammation . Histologic images were captured at the tissue site showing the highest degree of inflammation using an Olympus BX51 microscope equipped with a 4MP Macrofire digital camera ( Optronics ) and using the PictureFrame Application 2 . 3 ( Optronics ) . All images were processed identically by Photoshop ( Adobe Systems Inc . , Mountain View , CA ) . All data handling , graphical representation , and statistical analysis were conducted using GraphPad Prism 6 . 0 . Graphical representations of all data are expressed as the mean +/- the standard error of the mean ( SEM ) , and statistical significance was evaluated using a two-tailed unpaired t-test , one-way ANOVA with Tukey’s multiple comparison test , or two-way ANOVA with Bonferroni’s multiple comparison test as indicated in the figure legends . Where applicable , variances in data were deemed significant with a P value < 0 . 05 , and asterisks indicates the degree of significance ( *P < 0 . 05 , **P < 0 . 01 , ***P < 0 . 001 ) .
To investigate the role of SODB1 in Leishmania infection and pathogenesis , we generated Leishmania parasites with a disrupted SODB1 allele . DNA constructs encoding a G418 resistance gene ( NEO ) flanked by IGR sequences upstream and downstream of the SODB1 coding region were used to replace the complete endogenous SODB1 coding region via homologous recombination ( S1A Fig ) . Selection of L . major parasites carrying single allele disruption of the SODB1 gene was readily achieved ( S1B Fig ) . Precise integration of the targeting construct was confirmed using PCR primer combinations that specifically amplify products from either the SODB1 WT genomic locus , or products that are unique to specific construct integration within the SODB1 genomic locus on chromosome 32 ( S1B and S1C Fig ) . Consistent with previous studies with L . chagasi parasites [12] , repeated attempts to delete the second SODB1 allele by homologous recombination were unsuccessful . To confirm these findings , we successfully replaced a single SODB1 allele with a G418 drug resistance cassette in the L . donovani 1S2D sub-strain LdBob using similar gene disruption techniques as in L . major ( S1C Fig ) . Similar to our observations using L . major , repeated attempts to delete the second SODB1 allele in L . donovani parasites by homologous recombination also were unsuccessful . These data , together with previous reports , suggest that SODB1 is essential for the viability of promastigotes of multiple species of Leishmania parasites . We next compared the properties of single allele SODB1 knockout parasites ( SODB1/Δsodb1 ) with WT parasites in culture . SODB1/Δsodb1 L . major and L . donovani promastigotes showed no gross morphological abnormalities compared with WT promastigotes ( Fig 1A ) , and both SODB1/Δsodb1 L . major and L . donovani promastigotes grew at similar rates as WT in culture ( Fig 1B and 1C ) . In addition , SODB1/Δsodb1 and WT L . major parasites were capable of differentiating into infective metacyclic promastigotes . Nonetheless , SODB1/Δsodb1 parasites showed a decrease ( 1 . 9-fold; P < 0 . 05 ) in the overall percentage of metacyclic promastigotes in stationary phase cultures ( Fig 1D ) . These findings are consistent with previous analysis of SODB1/Δsodb1 L . chagasi promastigotes in which growth in culture did not deviate from WT promastigotes [12] . Next , SODB1/Δsodb1 L . donovani differentiation to amastigotes in response to low pH and increased temperature was evaluated [16] . Differentiation of promastigote forms to amastigotes and general morphology did not differ between WT and SODB1/Δsodb1 L . donovani ( Fig 1E ) , and growth of parasites under axenic amastigote growth conditions was similar for WT and SODB1/Δsodb1 L . donovani ( Fig 1F ) . To investigate if a single allele deficiency of SODB1 enhanced Leishmania susceptibility to different oxidative stresses , we exposed parasites to increasing concentrations of H2O2 and menadione , which induces superoxide generation in mitochondria . Using a fixable dead cell stain-based viability assay and quantification by flow cytometry ( S2A and S2B Fig ) , we found that WT and SODB1/Δsodb1 L . major or L . donovani promastigotes displayed similar sensitivities to both treatments ( S2C and S2D Fig ) , suggesting that SODB1 is not essential for protection of late-log phase promastigotes against these oxidative stresses . Finally , to confirm that disruption of a single allele of SODB1 reduced SODB1 expression levels , we quantified SODB1 RNA in log-phase promastigotes and in amastigotes . As shown in Fig 1G and 1H , we detected a 50–60% decrease of SODB1 RNA in SODB1/Δsodb1 L . major and L . donovani promastigotes and SODB1/Δsodb1 L . donovani amastigotes compared with WT parasites . Importantly , these reductions were specific to RNA levels of SODB1 , as the levels of SODB2 RNA in SODB1/Δsodb1 and WT parasites were similar ( Fig 1G and 1H ) . Collectively , although the generation of SODB1 null L . major or L . donovani parasites was not achieved by these methods , suggesting that SODB1 is vital for parasite survival , a single allele deletion does not alter general parasite morphology , promastigote growth in vitro , or , in the case of L . donovani , differentiation from promastigote to amastigote forms . To evaluate the role of SODB1 for infection in macrophages , BMDMs were generated from WT C57BL/6J mice . Macrophage cultures were inoculated with stationary phase WT or SODB1/Δsodb1 L . major promastigotes at an MOI of 10 parasites/cell . Total parasite genomes per 100 macrophages were quantified in infected cultures using qPCR assays specific for the Leishmania ubiquitin hydrolase gene and the murine gapdh gene . BMDMs infected with SODB1/Δsodb1 L . major promastigotes showed reduced parasite burdens on a per cell basis at 2 , 5 , and 10 day pi , but not at 4 h pi ( Fig 2A ) . These data suggest that SODB1 is dispensable for macrophage invasion but is required for intracellular replication and/or survival of L . major parasites in macrophages . To confirm that decreased SODB1/Δsodb1 L . major persistence in infected BMDMs is due to disrupted SODB1 expression , SODB1/Δsodb1 L . major parasites were complemented with a Leishmania-specific gene expression vector ( pXG ) bearing the complete L . major SODB1 coding sequence . To enable detection of exogenous SODB1 , an influenza virus HA sequence tag was added to the 5′-end of the SODB1 gene in the pXG vector [12] . SODB1/Δsodb1 L . major promastigotes were transfected with either pXG or pXG- ( HA ) -SODB1 and clones harboring the expression vector were isolated following drug selection . To confirm overexpression of HA-SODB1 , whole cell lysates were evaluated by HA-specific western blot analysis . As shown in Fig 2B , SODB1/Δsodb1 L . major promastigotes transfected with pXG- ( HA ) -SODB1 , but not empty pXG , express HA-tagged SODB1 . BMDMs were infected with purified metacyclic promastigotes from WT , SODB1/Δsodb1+pXG and SODB1/Δsodb1+pXG- ( HA ) -SODB1 L . major promastigotes at an MOI of 1 parasite/cell . Quantification of total parasites per 100 macrophages revealed similar numbers of parasites at 2 and 5 day pi in macrophages infected with SODB1/Δsodb1 parasites expressing HA-tagged SODB1 and WT parasites , whereas the number of parasites in macrophages infected with SODB1/Δsodb1 L . major plus the empty pXG vector were reduced ( Fig 2C ) . These findings suggest that impaired persistence of SODB1/Δsodb1 parasites in BMDMs is due to impaired SODB1 expression and can be rescued through episomal complementation with SODB1 . Disruption of a single SODB1 allele in L . major parasites reduced infection levels over time in macrophages in vitro . To determine if SODB1/Δsodb1 L . major parasites exhibit altered pathogenicity in vivo , we utilized a mouse model of Leishmania pathogenesis . WT C57BL/6 mice infected with L . major parasites subcutaneously in the foot show disease signs such as lesion development over time , and contain measurable parasite burdens in the foot and the draining popLN [27 , 28] . We inoculated WT C57BL/6J mice with 104 purified metacyclic WT or SODB1/Δsodb1 L . major promastigotes and monitored lesion development of the inoculated foot up to 50 dpi ( Fig 3A ) . Mice infected with WT L . major exhibited disease progression as expected , with measurable lesion development first detectable between 14 and 21 dpi , and maximal lesion size attained by 40–42 dpi ( Fig 3A ) . In contrast , mice infected with SODB1/Δsodb1 L . major metacyclic promastigotes showed significantly reduced lesion development as early as 24 dpi and failed to develop a temporal rise in lesion size up to 50 dpi ( Fig 3A ) . To confirm these findings , we assessed the severity of inflammation within the foot tissue of mice infected with WT or SODB1/Δsodb1 L . major parasites . Examination of H&E stained foot tissue sections from WT and SODB1/Δsodb1 L . major infected mice at 14 and 28 dpi revealed an abundant inflammatory infiltrate composed of neutrophils , macrophages , and lesser numbers of lymphocytes throughout the full thickness of the dermis and hypodermis compared with mock-infected control mice ( Fig 3B and S3 Fig ) . At 28 dpi , inflammation extended through the underlying fascia and muscle layers to and around the underlying periosteum that surrounds the metatarsals in all mice infected with WT L . major , but only a fraction of mice infected with SODB1/Δsodb1 L . major parasites ( Fig 3B ) . Although mice infected with SODB1/Δsodb1 L . major showed similar infiltrate composition , both the size of the inflammatory lesion ( extent of inflammation ) and the density of inflammatory cells ( quantity of inflammation ) was reduced at 28 dpi ( Fig 3C and 3D ) . Quantification of total inflammation ( the product of “extent” and “quantity” of inflammation ) revealed that SODB1/Δsodb1 L . major infected mice had reduced inflammation at 28 dpi when compared with mice infected with WT L . major ( Fig 3E ) . Lastly , the degree of inflammation in foot tissue of mice infected with WT L . major as measured by all criteria shows an increase between 14 and 28 dpi , whereas the degree of inflammation in mice infected with SODB1/Δsodb1 L . major remained similar between these time intervals ( Fig 3C–3E ) . Collectively , these findings suggest that a normal level of SODB1 is essential for L . major virulence in WT C57BL/6 mice . To determine if the decreased lesion development and severity observed in mice infected with SODB1/Δsodb1 L . major parasites was due to impaired infectivity or diminished parasite persistence , we performed a kinetic analysis of parasite burdens in foot tissue and the popLN , a primary site of parasite dissemination . At 14 and 28 dpi , parasite burdens in the foot were comparable in mice inoculated with WT and SODB1/Δsodb1 L . major metacyclic promastigotes , yet a lack of dissemination of SODB1/Δsodb1 parasites to the draining popLN was evident at both time points post-infection ( Fig 4 ) . At 50 dpi , we detected a significant decrease in parasite burdens in the foot tissue of mice inoculated with SODB1/Δsodb1 parasites compared with WT parasites , and a near absence of parasite infection in the popLN ( Fig 4 ) . Thus , the less severe inflammation and swelling observed in mice infected with SODB1/Δsodb1 L . major parasites is associated with a diminished capacity for SODB1/Δsodb1 L . major parasites to propagate or persist over time and to disseminate to tissues distal to the site of inoculation . C57BL/6 mice deficient in inducible nitric oxide synthase ( B6 . Nos2-/- ) and WT BALB/cJ mice are highly susceptible to L . major infection [29 , 30] . To determine if disruption of a single allele of SODB1 in L . major is sufficient for attenuation in these highly susceptible mouse strains , we evaluated disease progression and parasite burdens in Nos2-/- C57BL/6 mice and WT BALB/cJ mice infected with either 104 purified WT or SODB1/Δsodb1 L . major metacyclic promastigotes . Similar to observations in WT C57BL/6 mice , Nos2-/- C57BL/6 and WT BALB/cJ mice infected with WT L . major parasites developed severe lesions , with steady growth of lesion size out to 50 dpi . In contrast , Nos2-/- C57BL/6 ( B6 . Nos2-/- ) and WT BALB/cJ mice infected with SODB1/Δsodb1 L . major parasites showed little to no evidence of lesion development over time ( Fig 5A and 5C ) . In addition to reduced lesion development , parasite burdens in the foot tissue and popLN of both mouse strains were reduced in mice infected with SODB1/Δsodb1 L . major compared with mice infected with WT L . major ( Fig 5B and 5D ) . These findings provide further evidence that a normal level of SODB1 is required for L . major virulence in mice . Additionally , these data suggest that attenuation of SODB1/Δsodb1 L . major parasites is not due to heightened susceptibility to Nos2- or mouse strain-dependent effects .
An improved understanding of mechanisms by which Leishmania parasites replicate and persist in , and disperse from macrophages , particularly in vivo , may identify new therapeutic targets for the treatment of leishmaniasis . Pathogen-encoded SODs have known roles in detoxifying environmental superoxide as a countermeasure against host oxidative defenses . For example , in bacterial pathogens such as Salmonella , Streptococcus and Mycobacterium , periplasmic SODs confer protection against the oxidative burst generated by host neutrophils and macrophages , and in many cases is critical for promoting bacterial pathogenesis [31–35] . The genomes of Leishmania parasites are known to encode three superoxide dismutase genes: SODA , SODB1 , and SODB2 [12 , 36] . Notably , each of these SODs are targeted to membrane-bound organelles , with SODA targeted to mitochondria and SODB1 and SODB2 targeted to the glycosome . Recent studies implicated SODA in the regulation of L . amazonensis differentiation to infective forms , including both metacyclic promastigotes and amastigotes , by protecting against mitochondrial oxidative stress and activating ROS-dependent signaling . Moreover , disruption of a single SODA allele in L . amazonensis was sufficient to cause severe parasite attenuation in mice [21] . In contrast , no studies to date have evaluated the role of SODB1 or SODB2 in Leishmania pathogenesis in vivo . Consistent with previous studies using other Leishmania species , we were unable to generate SODB1 null L . major parasites using standard gene targeting approaches based on replacement of SODB1 coding regions with a drug resistance gene by homologous recombination , suggesting that SODB1 is essential for L . major viability [12] . Nevertheless , as heterozygous gene disruption of many Leishmania genes , including SODA [21] , has revealed important insight into parasite biology , we evaluated the impact of disrupting a single SODB1 allele on L . major and L . donovani differentiation to infective forms , infection of macrophages , and pathogenesis in mice . During standard promastigote culture conditions , the morphology and growth rates of SODB1/Δsodb1 and WT L . major and L . donovani promastigotes were comparable . In addition , SODB1/Δsodb1 L . major parasites differentiated into infective metacyclic promastigotes with nearly the same efficiency as WT parasites . Furthermore , the differentiation and axenic growth of SODB1/Δsodb1 L . donovani amastigotes also was comparable with WT parasites . Thus , in contrast to SODA , disruption of a single allele of SODB1 had minimal impact on the growth and differentiation of Leishmania parasites in culture , suggesting that Leishmania SODA and SODB1 genes have distinct functions [21] . This notion is further supported by the differential localization of SODA and SODB1 in the cell ( mitochondria versus glycosome , respectively ) [12 , 21 , 37 , 38] . However , as different Leishmania species were used in these studies , the impact of single allele deletions of SODA or SODB1 on L . major parasite biology in culture remains to be confirmed . Our data indicate that WT and SODB1/Δsodb1 L . major promastigotes have a similar capacity to establish infection in murine bone marrow macrophages . However , in contrast with WT L . major parasites , the number of SODB1/Δsodb1 L . major parasites in macrophages diminished over time , suggesting that a normal level of SODB1 is required for L . major persistence in macrophages . Complementation of SODB1/Δsodb1 L . major parasites with episomal expression of SODB1 restored parasite numbers in macrophages to levels equivalent with WT L . major , further suggesting that SODB1 promotes the survival of L . major parasites in macrophages . Due to the impaired capacity of SODB1/Δsodb1 L . major parasites to persist in murine macrophages , we evaluated the effects of a single allele deletion of SODB1 on virulence and persistence in mice . WT C57BL/6 mice infected with SODB1/Δsodb1 L . major purified metacyclic promastigotes developed mild cutaneous lesions compared with mice infected with WT L . major metacyclic promastigotes , which displayed normal lesion progression . Similar outcomes were observed in more susceptible Nos2-/- C57BL/6 and WT BALB/c mice , suggesting that SODB1/Δsodb1 L . major parasites are not more susceptible to nitric oxide produced from inducible NOS ( iNOS; Nos2 ) commonly expressed in macrophages and neutrophils . Although additional functional verification studies are ongoing , these findings suggest that SODB1 is essential for L . major virulence in mice . In WT C57BL/6 mice , the burdens of WT and SODB1/Δsodb1 L . major parasites in the inoculated foot were comparable at both 14 and 28 dpi . In contrast , by 50 dpi , parasite burdens in the foot tissue of WT and Nos2-/- C57BL/6 mice , as well as WT BALB/c mice , infected with SODB1/Δsodb1 L . major were strongly reduced compared with WT L . major burdens . Similar to our observations in bone marrow macrophages , these data suggest that a normal level of SODB1 is dispensable for L . major to establish infection , but essential for long-term L . major persistence . We also found that SODB1/Δsodb1 L . major parasite dissemination to the draining popLN was virtually absent throughout the time course evaluated . In mice , activation of innate immune responses during Leishmania infection are critical for curtailing parasite spread to organs distal to the site of inoculation [39] . Thus , the diminished capacity of SODB1/Δsodb1 L . major parasites to disseminate to the draining popLN suggests that SODB1/Δsodb1 L . major parasites are more susceptible to host innate defenses . The reduced capacity for SODB1/Δsodb1 L . major parasites to persist in macrophages may be due to critical roles for SODB1 in detoxification of superoxide produced in the glycosome , to enhanced susceptibility to host oxidative defenses , or both . The glycosome is a membrane bound organelle uniquely found in trypanosomatid parasites that compartmentalizes enzymes involved in glycolysis [14] . Thus , the localization of SODB1 to this internal organelle , the proper function of which is required for parasite survival , the attenuation of SODB1/Δsodb1 L . major parasites in Nos2-/- mice , and the decreased survival of SODB1/Δsodb1 L . chagasi parasites exposed to paraquat [12] , suggests important roles for SODB1 in detoxification of endogenous superoxide . However , it remains possible that SODB1 also influences Leishmania susceptibility to exogenous reactive oxygen species . Experiments are ongoing to evaluate both of these possibilities . Our findings expand upon previous studies suggesting that Leishmania glycosomal SODB1 protects from exogenous reactive oxygen species and promotes parasite survival in macrophages in vitro [12 , 15] . Here , we show that a single allele deletion in the L . major SODB1 gene leads to significant attenuation of disease severity , as well as parasite dissemination and persistence in mice . Therefore , therapeutic disruption of SODB1 function may provide an efficacious means for mitigating leishmaniasis . | Leishmania protozoan parasites are the causative agents of leishmaniasis , neglected tropical diseases with clinical manifestations ranging from self-limiting cutaneous lesions to invasive disease involving multiple organs . During blood-feeding by an infected sand fly , Leishmania parasites are transmitted to human hosts where they establish long-term infection in cells such as macrophages . An improved understanding of the mechanisms by which Leishmania parasites replicate and persist in macrophages may identify new therapeutic targets for the treatment of leishmaniasis . Using genetically-altered Leishmania parasites , cell-culture approaches , and a mouse model of Leishmania infection and disease , we found that a normal level of Leishmania superoxide dismutase B1 ( SODB1 ) is necessary for Leishmania major persistence in macrophages and for virulence in mice . This work contributes new knowledge about the pathogenic mechanisms of Leishmania major infection and suggests that SODB1 may be an efficacious therapeutic target . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [
"blood",
"cells",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"pathology",
"and",
"laboratory",
"medicine",
"enzymes",
"immunology",
"microbiology",
"enzymology",
"dismutases",
"parasitic",
"diseases",
"parasitic",
"protozoans",
"protozoan",
"life",
"cyc... | 2018 | SODB1 is essential for Leishmania major infection of macrophages and pathogenesis in mice |
Genome-wide studies of circadian transcription or mRNA translation have been hindered by the presence of heterogeneous cell populations in complex tissues such as the nervous system . We describe here the use of a Drosophila cell-specific translational profiling approach to document the rhythmic “translatome” of neural clock cells for the first time in any organism . Unexpectedly , translation of most clock-regulated transcripts—as assayed by mRNA ribosome association—occurs at one of two predominant circadian phases , midday or mid-night , times of behavioral quiescence; mRNAs encoding similar cellular functions are translated at the same time of day . Our analysis also indicates that fundamental cellular processes—metabolism , energy production , redox state ( e . g . , the thioredoxin system ) , cell growth , signaling and others—are rhythmically modulated within clock cells via synchronized protein synthesis . Our approach is validated by the identification of mRNAs known to exhibit circadian changes in abundance and the discovery of hundreds of novel mRNAs that show translational rhythms . This includes Tdc2 , encoding a neurotransmitter synthetic enzyme , which we demonstrate is required within clock neurons for normal circadian locomotor activity .
Genetic studies carried out in several model systems have provided seminal knowledge about the biochemistry of the circadian molecular oscillator and the neural circuitry regulating circadian behavior . The best characterized circadian oscillators consist of transcriptional/translational feedback loops ( TTFLs ) [1] , although nontranscriptional oscillators ( NTOs ) exist in organisms ranging from unicellulars to Drosophila and humans [2]–[4] . In Drosophila and mammals , a well-characterized TTFL oscillator consisting of several canonical clock genes regulates circadian behavioral rhythms ( reviewed in [1] ) . Similarly , transcription of many ( perhaps most ) genes is orchestrated by the circadian clock , based on gene profiling studies carried out in Drosophila , mammals and plants . Only a few studies , however , have documented cell-type–specific transcriptional rhythms [5]–[7] , due to methodological limitations . Most of those studies utilized Fluorescence-Activated Cell Sorting ( FACS ) , the manual isolation of identified cells , or cell-specific transcriptional profiling techniques , but such methods are either not applicable to all cell populations or lack the sensitivity to detect the entire transcriptome; nor do they distinguish between ribosome-bound ( i . e . , translating ) and soluble mRNA without the use of polyribosome isolation . Drosophila is an excellent model for cell-type–specific profiling of clock cells because of its outstanding genetics and well-characterized circadian system . Studies have described the fly circadian molecular oscillator [8] and the circadian neuronal circuitry [9] , revealing molecular and functional differences among groups of pacemaker neurons that mediate morning and evening bouts of activity or responses of the clock to environmental cues [5] , [10]–[18] . To date , no study has documented genome-wide expression profiles for all clock cells of the fly head or the complete translatome of such cells . In this study , we describe use of the Translating Ribosome Affinity Purification ( TRAP ) method [19] to define the rhythmic translatome of circadian clock cells . Our results reveal a daily synchronization of protein synthesis and identify novel cycling mRNAs within clock cells that are required for diverse physiological processes .
Previous studies have shown that TRAP reflects the translational status of mRNAs in a manner similar to that of conventional polyribosomal analysis [19] . In addition , a recent study in Drosophila indicates that an EGFP-L10a fusion incorporates into polysomes and can be employed for cell-specific translational profiling [20] . To employ TRAP in our studies , we generated Drosophila strains carrying a UAS-EGFP-L10a transgene insertion ( see Materials and Methods ) . Using a pan-neuronal driver ( elav-Gal4 ) , we found that the EGFP-L10a fusion has a cytoplasmic/nucleolar pattern of localization in neurons ( Figure 1A–C ) , consistent with incorporation into ribosomes . Indeed , the ring shape pattern in nucleoli ( seen in the nucleus of Figure 1A ) likely results from expression in the Granular Component ( GC , Figure 1D ) , a structure within which ribosomal proteins assemble into functional ribosomes . As expected , EGFP-L10a was localized in all neurons of the adult nervous system ( Figure 1E ) . In contrast , the tim-uas-Gal4 driver results in expression within the cytoplasm of clock neurons and glia of the nervous system ( Figure 1F ) or only clock neurons when combined with repo-Gal80 ( Figure 1G ) , which inhibits expression in all glial cells . A different GFP–Drosophila ribosomal protein fusion ( L11 ) has an identical intracellular localization pattern [21] . In addition , it has recently been shown that our EGFP-L10a fusion localizes to branch points of neuronal dendrites , consistent with incorporation into ribosomes that mediate local protein synthesis [22] . Collectively , these pieces of evidence indicate that EGFP-L10a incorporates into functional ribosomes . We examined circadian locomotor activity of flies overexpressing the UAS-EGFP-L10a transgene in clock cells ( Pigment Dispersing Factor , PDF , or Timeless , TIM ) to determine whether there were adverse effects on behavior . As shown in Figure 1H , these files have normal behavioral rhythmicity , indicating that EGFP-L10a does not act in a dominant negative manner even at high levels [Average periods ( P ) and Rhythmicity Indices ( RI ) were 23 . 7±0 . 08/0 . 57±0 . 02 , 24 . 0±0 . 03/0 . 55±0 . 01 , and 24 . 3±0 . 14/0 . 50±0 . 03 for control , pdf-Gal4>UAS-EGFP-L10a , and tim-uas-Gal4>UAS-EGFP-L10a flies; n = 17–30] . Thus , the presence of GFP-tagged ribosomes in clock cells does not affect their function . We optimized TRAP methods for use with Drosophila and demonstrated that significant amounts of RNA could be immunoprecipitated from head tissues of flies expressing UAS-EGFP-L10a under control of the pan-neural elav-Gal4 or clock cell tim-uas-Gal4 driver ( see Materials and Methods ) . Prior to pursuing genome-wide studies , we wished to determine if our Drosophila TRAP methods could detect bona fide changes in translational status . To ask this question , we employed overexpression of Iron Regulatory Protein ( IRP ) , which is known to repress translation of an unspliced form of ferritin ( fer ) mRNA by inhibiting binding of the small ribosomal subunit to the message . We generated act5C-Gal4/tub-Gal80ts , UAS-EGFP-L10a/UAS-IRP flies in order to be able to activate expression of the TRAP and IRP transgenes conditionally during larval development ( by raising temperature to inactivate Gal80ts , an inhibitor of Gal4 ) . Larvae of this genotype and controls ( act5C-Gal4/tub-Gal80ts; UAS-EGFP-L10a ) were exposed to 30°C to activate expression of UAS-EGFP-L10a in both genotypes and additionally UAS-IRP in the experimental class . Early pupae were collected for both genotypes and subjected to TRAP coupled with Q-RT-PCR to quantify ribosome-associated fer mRNA ( relative to control Rp49 mRNA ) . Similar to previous studies in Drosophila that employed polysome gradient analysis [23] , we observed IRP-induced translational repression of an unspliced but not a spliced form of fer ( Figure 1I ) . Indeed , translation of spliced fer was enhanced slightly by IRP overexpression , similar to that observed from the analysis of a high molecular weight polysome fraction in the previous study [23] . This result shows feasibility for the use of TRAP in Drosophila to detect changes in translational status . To determine if our methods were able to detect rhythmic changes in the ribosomal association of cycling transcripts , we examined the period ( per ) and timeless ( tim ) clock mRNAs . TRAP methods were employed to immunopurify RNA from head tissues of elav-Gal4/UAS-EGFP-L10a flies two times of day ( ZT11 and ZT23 , the times of high and low per/tim RNA abundance , respectively ) . Extracted RNA was then subjected to Q-RT-PCR , using gene-specific primers , to detect the clock mRNAs . Figure 1J shows that the abundances of ribosome-bound tim and per clock mRNAs are significantly higher at ZT11 than at ZT23 . This result is consistent with the known rhythmic profile of tim and per RNA abundances at the two time points ( higher at ZT11 ) and the expected translational status of the mRNA at the two times of day . We emphasize that Figure 1J shows differences in ribosome association of the clock RNAs , not simply the previously documented RNA abundance for per and tim . In addition , we note that the temporal resolution of our measurements does not exclude translational regulation of per mRNA , which has been suggested in certain studies [24]–[26] . Nonetheless , these results demonstrate that TRAP methods are capable of detecting diurnal changes in the translational status of specific mRNAs . Using the newly developed methods , we performed TRAP on head tissue lysates of tim-uas-Gal4; UAS-EGFP-L10a flies that were collected at 4-h intervals during the first two days of constant darkness ( DD ) following entrainment to LD 12∶12 . Such flies express the EGFP-L10a fusion in all clock cells of the head , including the ∼150 pacemaker neurons , photoreceptors , and glia . RNA was extracted from affinity-purified samples and used to generate libraries representing all ribosome-associated transcripts ( see Materials and Methods ) . TRAP libraries corresponding to six different times of the circadian cycle ( CT0 , 4 , 8 , 12 , 16 , and 20 ) were independently constructed for DD1 and DD2 ( see details in Materials and Methods ) . Libraries were sequenced , using a multiplexing strategy , to produce single end , 100 base sequencing reads; these were mapped to the Drosophila reference genome and analyzed as described in Materials and Methods . We employed two recently developed programs , JTK_CYCLE and ARSER [27] , [28] , to compare their usefulness for detecting circadian rhythms in the ribosome association of mRNAs . Using criteria and statistical cutoffs described in the Materials and Methods section , 1 , 195 and 263 translationally cycling mRNAs were detected by the ARSER and JTK_CYCLE programs , respectively . Interestingly , the majority of the cycling mRNAs ( 203 out of 263 ) detected by JTK_CYCLE were also detected by the ARSER program ( Figure 2A ) , indicating consistency of the two analyses . Figure S1 shows robust cycling for eight mRNAs out of the 60 identified by JTK_CYCLE but not ARSER . Thus , JTK_CYCLE may identify cycling mRNAs not detected by ARSER . Table S4 lists the 1 , 255 mRNAs that were identified as exhibiting significant translational cycling by either program ( mRNAs identified by both programs are indicated in bold ) . The False Discovery Rate ( FDR ) calculated by the ARSER program at the relevant p value was 0 . 148 , indicating that approximately 186 mRNAs are false positives . This FDR is quite low relative to other recent genome-wide studies of cycling mRNAs [29]–[31] . We did not compute an FDR for the JTK_CYCLE program , because 203/263 mRNAs identified by JTK_CYCLE are included in the ARSER dataset , and therefore the latter dataset represents a good approximation of FDR for our analyses . Based on the ARSER analysis , we estimate that approximately 1 , 069 of these mRNAs show circadian changes in translation in clock cells of the adult head , representing about 10% of all analyzed genes in the genome . This large number of cycling mRNAs is consistent with recent studies utilizing manual dissection approaches to perform cell-specific transcriptional profiling of the Drosophila PDF clock neurons [5] , [10] . Cell-specific profiling methods may identify a larger number of cycling Drosophila mRNAs , relative to previous studies , due to a more homogeneous starting cell population ( i . e . , clock cells ) . We examined a number of mRNAs in our datasets that had previously been shown to exhibit abundance rhythms to assess the quality of our datasets . These include both clock and clock-regulated mRNAs ( per , tim , vri , clk , to , fer2 , slob , ugt35b , 5-HT1A , bw , Ir , and WupA ) . All showed translational rhythmicity ( Figure 2B ) with an expected phase , although the tim rhythm damped on DD2 . Figure 2C , for example , shows robust rhythmicity in the sequence reads for per and lack of rhythmicity for a nearby gene . Our analysis also revealed translational cycling for many other genes that express rhythmic mRNAs . For example , our list of mRNAs includes 49 of 420 mRNAs showing circadian abundance rhythms identified in five previous microarray-based studies ( see Introduction ) . This comparison does not include a recent study that identified 2 , 751 cycling mRNAs in hand-dissected PDF neurons [10]; our results include 172 of those mRNAs ( see Table S4 ) . Interestingly , Ugt35b mRNA , one of several encoding fly glucuronosyltransferase activity , was previously shown to exhibit transcriptional cycling in head tissues but not in PDF neurons [10] . Given that we employed a clock cell tim-Gal4 driver in our TRAP studies , we suggest that Ugt35b cycles in other clock cells of the head . We conducted TRAP combined with quantitative PCR for Ugt35b , tim , and 18 novel cycling mRNAs ( not previously found to show abundance rhythms in head tissues ) to verify results obtained by RNA-seq . As expected , Ugt35b and tim exhibited rhythmicity , presumably a consequence of their mRNA abundance rhythms . Of the novel mRNAs , 15/18 showed rhythmic changes in translation , with a profile very similar to that observed with RNA-seq analysis ( Figure S2 ) . We further analyzed cycling of a number of these mRNAs in the per0 mutant , which lacks a functional clock , during the first day of constant darkness ( DD1 ) . We found that rhythmic expression of these mRNAs was abolished in the per0 mutant , confirming their circadian clock regulation ( Figure S3 ) . Previous genome-wide studies showed that peaks of mRNA abundance occur at many different circadian phases ( see Figure S4 ) . In contrast , our profiling of the clock cell translatome revealed a striking feature of circadianly regulated protein synthesis . We found that peak translation for most of the 1 , 255 mRNAs identified in our study occurs predominantly during two circadian phases: midday or mid-night ( Figure 3A–C; Figure S4 ) . These are times of relative behavioral quiescence and just prior to initiation of locomotor activity bouts ( Figure 3A , lower panel ) . Thus , protein synthesis may be confined to times of day that require reduced metabolic expenditure and/or are just prior to initiation of behavioral activities . A further analysis revealed surprisingly synchronized translation of mRNAs required for the same cellular process: translation is predominantly unimodal ( with a peak during the day or night ) or biomodal , depending on the process ( Figure 3C ) . This bias in the timing of translation was true of many other cellular processes ( Figure 3D ) . For example , of the 10 enzymes involved in glucose metabolism in our list of cycling RNAs , nine are translated during the day . In contrast , all 10 GPCRs in our list are translated during the night ( Figure 3E ) . Of note , mRNAs encoding a number of translational initiation factors ( eIF4E isoforms ) exhibit cycling with a phase that corresponds to the daytime peak of circadian translation ( Figure S5 ) . Thus , circadian translation of these eIFs may contribute to a broad clock regulation of protein synthesis . In contrast , the major initiation factor , eIF4E-1 , does not exhibit translational cycling , suggesting that it does not participate in circadian regulation ( Figure S5 ) . Consistent with previous results indicating that ribosome biogenesis is regulated in a circadian manner [32] , 20 mRNAs encoding ribosomal proteins , translation initiation factors , or other translational regulatory components show translational rhythmicity ( Table S4 ) . The synchronized rhythmic expression profiles identified by our cell-specific profiling approach may result from a clock regulation of translation or mRNA abundance . To ask whether changes in translational status contribute to the synchronization of gene expression in clock cells , we carried out additional studies , using TRAP/RNA-seq methods . We reasoned that total RNA isolated from whole heads contains mRNAs from both clock and nonclock cells . Thus , if a gene is robustly expressed in nonclock cells , the abundance profile obtained from whole head total RNA will not represent its expression profile in clock cells . However , for mRNAs predominantly expressed in clock cells ( such as per , tim , and others ) , assays of total head RNA will reflect clock cell expression . Such an mRNA ought to show enrichment in a TRAP sample from tim-uas-gal4>EGFP-L10a heads relative to total RNA from the starting lysate , and the circadian expression profile , when assayed from total RNA , should approximate the profile in clock cells . Thus , if such an mRNA shows a rhythm by TRAP but not in total RNA , then it is likely to be regulated at the translational level . To identify mRNAs enriched in clock cells , we created new genome-wide libraries for TRAP and total RNA samples from head tissues of tim-uas-gal4>EGFP-L10a–expressing flies . These were sequenced to identify mRNAs that are substantially enriched by TRAP relative to total RNA—that is , enriched in clock cells . We identified many that show an enrichment within clock cells similar to or greater than that observed for tim mRNA . Forty-nine of them are present in our previous list of cycling mRNAs . We chose 12 cycling mRNAs from the enriched list and examined their expression profiles in total RNA versus TRAP RNA samples using Q-PCR methods . Of the 12 mRNAs tested , three did not show cycling similar to that detected by RNA-seq analysis ( 25% , and the same as we reported for another set of RNAs; Figure S2 ) ; thus , these three were not examined further . Of the remaining nine mRNAs , which showed cycling by Q-PCR similar to that detected by RNA-seq , three of them exhibited constant abundance in total RNA but showed circadian cycling in ribosome association , indicating that they are likely regulated at the translational level . Figure 4 shows cycling profiles for these three mRNAs and a fourth mRNA showing both abundance and ribosome-association rhythms ( Figure 4D ) . Thus , for certain mRNAs , there is good evidence for a clock regulation of translation . We manually annotated the proteins encoded by the 1 , 255 cycling RNAs using information obtained from Flybase and classified them by biological process ( Figure 5A ) . Of the annotated genes , the most represented functional class is metabolism/energy production , including NAD-dependent processes and oxidation-reduction reactions . This class includes 85 genes involved in intermediary metabolism , 14 genes with mitochondrial functions , and 46 genes that regulate oxidation-reduction processes . These results are consistent with Drosophila and mouse circadian transcriptional profiling studies that identified a large subset of metabolic genes [33] , [34] . Another overrepresented group is signaling ( including both intracellular pathways and intercellular signaling mechanisms ) . Interestingly , 44 members of the signaling class belong to the G Protein signaling family , represented by many G Protein Coupled Receptors ( GPCRs ) and GTPases . Several particularly interesting cycling mRNAs encode proteins that potentially modulate the NADP+/NADPH ratio or are known components of the cellular redox ( thioredoxin ) system . Examples include the CG3483 and CG7755 genes , both predicted to encode isocitrate dehydrogenase-like proteins . While at least one isocitrate dehydrogenase ( IDH ) is a component of the mitochondrial citric acid cycle , others have a cytoplasmic localization , producing αketoglutarate with a conversion of NADP+ to NADPH [35] . We also found that the mRNA encoding Glutathione Transferase E10 ( GstE10 ) , which utilizes the redox regulator glutathione in detoxification reactions , exhibits a translational rhythm ( Table S4 ) . Interestingly , it was recently shown that glutathione and a different Gst mRNA ( GstD1 ) show circadian changes in abundance in Drosophila head tissues [36] , suggesting a complex regulation of redox state . Components of the thioredoxin ( TRX ) system , a general regulator of cellular redox state , are also under circadian control . Thioredoxin T ( TrxT ) and Thioredoxin reductase ( Trxr-2 ) mRNAs show robust circadian changes in translation , with peaks in the late subjective day ( Figure 5B ) . This circadian translation may reflect an underlying transcriptional control as both TrxT and Trxr-2 show mRNA abundance rhythms in Drosophila head tissues ( Figure S6 ) . Of interest , it was previously suggested that TrxT showed an mRNA abundance rhythm within the Drosophila PDF clock neurons , but this was based only on examination of two circadian times in a screen for cycling mRNAs [10] . Thioredoxin reductases are known to catalyze reduction of thioredoxin , in the process converting NADPH to NADP+ [37] , an important regulator of cellular redox . In addition to these TRX system genes , Grx-1 , a glutaredoxin also involved with cell redox state homeostasis , shows circadian translational cycling ( Table S4 ) . Rhythmicity in cellular redox state is significant as it regulates many biochemical processes including circadian transcription factors ( see Discussion ) . Previous studies have indicated that synaptic vesicle cycling mechanisms are important within clock neurons [38] and glial cells [39] for circadian oscillator or output functions . Similarly , there are reciprocal interactions between the oscillator and neuronal membrane events , including ion channel activity , that are critical for timekeeping in Drosophila and mammals [6] , [40] , [41] . It is of interest that we identified mRNAs encoding at least 20 ion channels or channel regulatory proteins that exhibit rhythms in ribosome association . These include cac ( Ca2+ channel ) , Ir ( K+ channel ) , SK ( K+ channel ) , Slob ( K+ channel regulator ) , and inaF-B ( Trp channel regulator ) , although Ir showed significant rhythmicity only during day 1 of DD . Interestingly , however , Ir was identified in a previous study as a rhythmic mRNA within PDF neurons that is important for oscillator function [6] . Likewise , a number of mRNAs encoding vesicle trafficking or release proteins , including exo70 , syn , and unc-104 , exhibited rhythmicity in our experiments . We note that at least two potential brain glial mRNAs were revealed in our study: CG9977 and CG6218 . CG9977 encodes adenosylhomocysteinase activity , whereas CG6218 encodes an ATPase . Both were identified in a previous microarray-based screen for Drosophila mRNAs enriched in glial cells [42] , and are known to be expressed in the adult brain according to FlyAtlas [43] . The CG9977 enzymatic activity converts S-adenosyl-L-homocysteine to L-homocysteine and adenosine , the latter a known mammalian gliotransmitter [44] . As the tim-uas-gal4 driver is expressed in neurons and glia ( including astrocytes ) with PER-based oscillators , CG9977 and CG6218 may be expressed in the latter cell type . Tdc2 encodes the neurally expressed isoform of tyrosine decarboxylase , which converts tyrosine to tyramine , the latter compound acting as a substrate for octopamine synthesis . In Drosophila , both tyramine and octopamine serve as neurotransmitters , regulating diverse functions including adult locomotion , male aggression , male courtship , drug sensitivity , ovulation , circadian activity rhythms , and appetitive memory formation [45]–[49] . Therefore , it is of interest that Tdc2 mRNA exhibits circadian translational rhythms in clock cells ( Figure 6A; Table S4 ) . We verified the circadian translation of Tdc2 using the TRAP technique coupled with Q-RT-PCR ( Figure 6B ) , and showed that this rhythm is abolished in per0 flies ( Figure S3 ) . Using an anti-TDC2 antibody , we further verified that expression of the TDC2 protein exhibits the predicted circadian changes in two major groups of clock neurons , the l-LNvs and the LNds ( Figure 6C and D ) . We used two different strategies to characterize the expression pattern of Tdc2 in the adult brain , in particular in various groups of clock cells . First , we characterized the expression pattern of a Tdc2-Gal4 transgene [50] and its co-localization with PERIOD protein . We found that a UAS-GFP reporter , driven by Tdc2-Gal4 , was expressed in multiple regions of the fly brain , as expected . However , the only clock cells showing GFP fluorescence ( identifiable by PER expression ) were the ventral lateral ( LNv ) PDF neurons of the brain ( Figure S7A ) , which are critical for circadian behavior [9] . Next , using anti-TDC2 antibody , we localized the TDC2 protein in flies expressing a membrane-bound GFP ( mCD8-GFP ) in all clock cells ( driven by tim-uas-gal4 ) . As expected , we found that there is a strong immunoreactive signal for TDC2 in many nonclock neurons . Within the clock neuronal population , we detected TDC2 immunoreactivity in all l-LNvs ( Figure S7B , a–d ) , s-LNvs ( a–d ) , and LNds ( i–k ) , as well as a few cells in the DN1 region ( l–n ) . Finally , a comparison of TDC2 immunoreactivity and Tdc2-gal4 driven mCD8-GFP expression found that Tdc2-gal4 does not express in all TDC2 immunoreactive cells ( unpublished data ) , indicating that the Tdc2-gal4 transgene does not reflect the complete expression pattern of the Tdc2 gene . These results suggest that rhythmic production of TDC2 in various clock neurons , and a consequent rhythm in release of tyramine and/or octopamine from these cells , may be required for normal circadian behavior . To assess the role of Tdc2 in circadian behavior , we analyzed locomotor activity of the Tdc2RO54 mutant , which carries a point mutation that abolishes the enzymatic activity of TDC2 [50] . Consistent with previous reports [45] , we found that Tdc2RO54 mutants displayed decreased activity ( Figure 7A ) . In addition , however , the mutant population exhibited decreased rhythmicity . The average Rhythmicity Index ( RI ) for Tdc2RO54 was 0 . 18±0 . 02 compared to 0 . 56±0 . 02 for control flies , and indeed only 20±3% of the mutant population displayed significant free running rhythms , whereas the control population was 100% rhythmic ( Figure 7A ) . We note that decreased activity does not result in arrhythmicity , as there was no correlation between activity level and rhythmic locomotor activity ( Figure 7B ) . These observations indicate that octopamine and/or tyramine are required for normal circadian behavior ( see Discussion ) . To ask whether the observed arrythmicity of the Tdc2 null mutant results from loss of Tdc2 function in clock cells , we examine circadian behavior in flies with a Tdc2 knockdown specifically in clock cells . We found that populations of flies expressing Tdc2 RNAi , driven by either pdf-gal4 or tim-uas-gal4 , were 75% arrhythmic and had low average Rhythmicity Indices ( Figure 8A , B , F ) , whereas control flies were normally rhythmic ( Figure 8C–F ) Thus , tdc2 is required within clock neurons for normal locomotor activity rhythms .
This study is the first to profile the circadian translatome of a defined cell population in a complex tissue . In contrast to previous studies showing that mRNA abundance rhythms peak at multiple circadian phases ( Figure S4 ) , our results indicate that translation of most rhythmic transcripts within clock cells is restricted to two major phases—midday and mid-night . Furthermore , we provide evidence that circadian regulation of either mRNA abundance or protein synthesis ( depending on the mRNA ) contributes to this synchronization . We speculate that protein synthesis may occur predominantly at circadian phases that are associated with reduced metabolic expenditure . In Drosophila , such times coincide with behavioral quiescence , just prior to initiation of locomotor activity bouts ( Figure 3 ) . The synchronized translation of functionally related mRNAs ( Figure 3B–D ) suggests a clock-orchestrated sequential activation of biological processes; these results reinforce the concept that fundamental cellular processes are under circadian control within clock cells . Two significant technical improvements enabled the discovery of synchronized translation . First , our analysis was restricted to circadian clock cells , circumventing the problem of profiling a mixed population , in which some cells express a rhythmic mRNA , whereas others express the same mRNA constitutively ( thus masking rhythmicity ) . In addition , different cell types may express out-of-phase rhythmic mRNAs , also masking a rhythm in a mixed cell population . Second , our technique analyzes ribosome association rather than steady-state mRNA abundance , representing a more direct assessment of protein expression . Although it is not currently technically feasible to directly compare transcriptional and translational rhythms in the same cell types , our results indicate that translational regulatory mechanisms contribute to synchronized protein synthesis . Consistent with this idea , we have shown that mRNAs encoding relevant translational regulatory factors are rhythmically expressed . These include translation initiation factors , ribosomal proteins , and enzymes involved in rRNA and tRNA synthesis ( Table S4 ) . In mammals , ribosome biogenesis is known to be regulated by the circadian clock [32] . Thus , it is possible that the circadian clock regulates translation of many mRNAs , including those relevant for clock function [25] , [26] by controlling availability of the translational apparatus . We document rhythmic translation of mRNAs within clock cells that is relevant for diverse biochemical and behavioral functions . A particularly interesting class includes factors important for cell redox homeostasis ( CG3483 , CG7755 , TrxT , and Trxr-2 ) , as it has been demonstrated that a clock control of redox state drives rhythms in the excitability of suprachiasmatic nuclei ( SCN ) neurons [51] . Furthermore , there is redox control of many cellular factors , including enzymes , receptors , cytokines , growth factors , and transcription factors . Thioredoxin , for example , regulates NFkB activity [52] , which is known to be under circadian control [53] . NADP ( H ) and NAD ( H ) , the reduced forms of these cofactors , stimulate DNA binding of the CLOCK/BMAL1 and NPAS2/BMAL1 transcriptional heterodimers , which are critical components of the mammalian circadian clock [54] . Circadian translational regulation of cellular redox may be important for rhythmicity of clock components and clock outputs as well as metabolic feedback to the clock [33] , [55] . The rhythm in TrxT translation may also function in another important circadian output . It has recently been demonstrated that there is circadian control of peroxiredoxin ( PRX ) protein oxidation state in organisms ranging from unicellulars to humans , and that this rhythm is regulated by an uncharacterized NTO ( reviewed in [1] ) . Significantly , oxidized PRX multimers serve as cellular chaperones and cell cycle modulators . Thioredoxin ( TRX ) mediates reduction of oxidized PRX molecules to complete the PRX catalytic cycle [1] , and thus rhythmic TrxT may contribute to circadian changes in PRX oxidation state . Of relevance , mRNAs encoding other chaperones are also rhythmically translated ( Table S4 ) . Rhythmic factors important for neurotransmission were also identified by our analysis . Among them , Tdc2—encoding the synthetic enzyme for tyramine and octopamine—is rhythmically expressed in clock neurons and localized to the PDF cell population . Rhythmic release of these transmitters from PDF or other clock neurons may contribute to the temporal coordination of the clock cell circuitry , similar to the role of PDF [56] . Alternatively , rhythmic release of octopamine and/or tyramine may regulate downstream neurons that drive locomotor activity rhythms . Of note , previous studies have suggested a clock control of tyramine synthesis , showing that there was decreased tyrosine decarboxylase activity in the brains of per clock mutants [57] . In addition , mutants of several clock genes , including per , clock , cycle , and doubletime , were found to be required for normal cocaine sensitization , a process depending on induction of tyrosine decarboxylase activity and production of tyramine [58] . Expression of Tdc2 in clock neurons is consistent with a role for tyramine , and perhaps octopamine , in this process . Tdc2 was not described as a cycling mRNA in several previous genome-wide circadian expression studies [59]–[63] that utilized whole fly heads as a starting material . Based on the expression pattern of Tdc2—that is , broad and strong expression in a large number of neurons including only certain clock neurons —it seems likely that the cycling of Tdc2 eluded detection in previous studies because of the presence of other TDC2-containing neurons in which the transcript does not exhibit rhythms in abundance . Indeed , Tdc2 was included in a long list of mRNAs ( 2 , 751 ) showing enrichment in the large LNv clock neurons at one time of day ( ZT12 ) in a recent study that utilized manual dissection procedures to profile PDF neurons [10] . Our detection of rhythmic Tdc2 translation and confirmation of its role in maintaining circadian locomotor activity rhythms clearly demonstrates the advantage of a cell-type–specific approach in genome-wide studies of gene expression .
For translational profiling of flies with a normal circadian clock , males of a homozygous w1118; tim-uas-gal4 stock were crossed to virgin females of a homozygous w1118; UAS-EGFP-mL10a stock . F1 progeny carrying both the UAS and Gal4 transgenes ( and expressing EGFP-tagged ribosomes in all clock cells ) were collected and used for TRAP experiments . To profile a clock mutant , females from a homozygous per0 w1118; UAS-EGFP-L10a stock were crossed to a Gal4 strain and only male flies from the F1 progeny were used in the TRAP experiments . All flies were reared in a lighting schedule consisting of 12 h of light and 12 h of dark ( LD 12∶12 ) at 25°C and 60% humidity . The tdc2RO54 strain and its isogenic parental strain were gifts from Dr . Jay Hirsh of University of Virginia . UAS-Tdc2 RNAi flies were obtained from the VDRC stock center ( stock numbers 10687R-1 and 10687R-3 ) . The UAS-EGFP-L10a transgene was generated by cloning the coding sequence of the EGFP-L10a fusion protein ( provided by Nat Heintz ) into the pUAST vector . We chose to use the mouse L10a ribosomal protein ( mL10a ) because it is virtually the same as fly L10a ( identical in size and ∼90% similar ) —not surprising for a ribosomal subunit—and it has been shown to work well for the TRAP method . The cloning service was provided by Entelechon ( Regensburg , Germany ) , and the resulting UAS-EGFP-L10a plasmid was verified by sequencing . The UAS-EGFP-L10a plasmid was purified using a Qiagen Maxi-prep kit and then used to generate transgenic flies ( Genetic Services , Cambridge , MA ) . Genomic insertions were mapped to chromosomes using standard segregation analysis procedures . Adult flies were collected in 50 ml conical tubes at desired time points and flash frozen in liquid nitrogen . Fly heads were collected by vigorously shaking frozen flies and passing them through geological sieves according to standard procedures . Approximately 200 heads were employed for each affinity purification experiment . Frozen heads were homogenized in a buffer containing 20 mM HEPES-KOH ( pH 7 . 4 ) , 150 mM KCl , 5 mM MgCl2 , 0 . 5 mM DTT , 100 µg/ml Cycloheximide , and 2 U/ml SUPERase ( Life Technologies ) and centrifuged at 20 , 000× g for 15 min to obtain cleared lysate . After adding DHPC and Igepal CA-630 to a final concentration of 30 mM and 1% , respectively , the lysates were incubated on ice for 5 min and centrifuged at 20 , 000× g again for 15 min . After centrifugation , the supernatant was applied to magnetic beads coated with a purified high-affinity anti-EGFP antibody ( prepared using the Dyabeads Antibody Couple Kit from Invitrogen ) and incubated at 4°C with end-to-end rotation for 1 h to allow binding of EGFP-tagged ribosome to the antibodies . Following incubation , samples were washed with a buffer containing 20 mM HEPES-KOH ( pH 7 . 4 ) , 150 mM KCl , 5 mM MgCl2 , 0 . 5 mM DTT , 100 µg/ml Cycloheximide , and 1% Igepal CA-630 for five times at room temperature . RNA was extracted from the beads using the TRIzol reagent ( Life Technologies ) . Quality and quantity of the isolated RNAs were analyzed using a Bioanalyzer ( Agilent ) . Using these methods , we affinity purified RNA-containing ribosomes from head tissues of adult flies expressing UAS-EGFP-L10a in all neurons or clock cells . Similar to published studies [19] , we optimized homogenization procedures for Drosophila head tissues , included magnesium and cycloheximide in the lysis buffer to preserve polysomes , inhibited RNAase activity , and employed a purified , high-affinity anti-GFP antibody for ribosome precipitation . In those experiments , the UAS-EGFP-L10a transgene was expressed in all neurons or all clock cells using , respectively , the elav-Gal4 or tim-uas-Gal4 drivers . In three pilot experiments—two using elav-Gal4 and one using tim-uas-Gal4—we obtained a total of 305–544 ng RNA from head tissues of 200 UAS-EGFP-L10a–expressing flies , whereas there were negligible amounts ( 50–100-fold less ) of precipitated RNA in control samples ( elav-Gal4 or tim-uas-Gal4 alone ) ( Figure S8 ) . Nearly as much RNA was precipitated using the tim-uas-Gal4 driver as with elav-Gal , and we attribute this result to the strength of the tim-uas-Gal4 driver and the observation that it is expressed in all clock neurons including photoreceptors and thousands of glial cells . With expression of UAS-EGFP-L10a in only the clock neuron population ( ∼150 neurons , some of which can be seen in Figure 1G ) , we were able to immunopurify 44 ng of RNA from 200 fly heads—10-fold more than control precipitations—indicating good sensitivity for our methods . Expression of a different ribosomal protein fusion , GFP-Drosophila L11 [21] , can also be employed for TRAP analysis; we immunopurified 118 ng of ribosome-bound RNA from elav-Gal4/UAS-GFP-L11 head tissues starting with 150 flies ( unpublished data ) . We employed standard Illumina protocols and reagents ( the TruSeq RNA sample preparation kit ) for RNA-seq library construction . RNAs extracted from the immunoprecipitation contain a mixture of mRNAs , ribosomal RNAs , and other small RNAs that are involved in translation , such as tRNAs . Using the TruSeq RNA kit , mRNAs were isolated using poly-dT coupled magnetic beads and fragmented by addition of divalent cations at 94°C . Cleaved mRNAs were then reverse transcribed into cDNA using random primers , and cDNA was subjected to second strand synthesis using DNA polymerase I and RNaseH . DNAs were end repaired , “A” tailed , and then ligated to Illumina sequencing adaptors prior to enrichment by PCR to create a library . Sequencing of libraries was accomplished using an Illumina HiSeq 2000 in the Tufts Medical School Molecular Core Facility . Sequence reads were obtained and their quality analyzed using the quality control metrics provided by the FastQC pipeline ( http://www . bioinformatics . babraham . ac . uk/projects/fastqc/ ) . We obtained , on average , 21 million high-quality 100-b reads for each of the 24 samples ( after removing low-quality reads ) , and an average of 82% of the high-quality reads could be mapped to the Drosophila 5 . 22 reference genome ( Table S1 ) using Tophat ( v 2 . 0 . 0 ) and Bowtie2 ( v 2 . 0 . 0 . 5 ) [64] , [65] . The reads represent approximately ∼12 , 000 genes that are expressed in clock cells of the Drosophila head . After mapping with Tophat and Bowtie , we counted the number of reads aligning to individual annotated genes in the Drosophila genome using HTseq-count ( EMBL ) . Using these methods , there was good agreement between the two biological replicates for each time point in the DD1 and DD2 datasets . The correlation coefficient ( r ) of the two replicates was greater than 0 . 9 for all time points ( Table S2 , representative scatter plots of two replicates are shown in Figure S9 ) . Next , we conducted a preliminary assessment of each of the four individual datasets ( DD1 , replicate 1 and 2 and DD2 , replicate 1 and 2 ) by calculating the “best cosine correlation” for all genes including 10 genes in our datasets that are known to show transcriptional cycling from previous studies ( Table S3 ) . The “best cosine correlation” is obtained by calculating the correlation coefficient ( r ) between read counts of the six time points and corresponding six values on one of 48 cosine curves each with 0 . 5 h difference in phase , and selecting the highest r value from the 48 comparisons . We found that 10 “control” RNAs had high r values in the two DD1 datasets and DD2 dataset 1 . However , poor correlation coefficients were observed for DD2 dataset 2 , and thus this dataset was not used in our subsequent analyses . Given the good correlation between DD1 datasets 1 and 2 , we pooled reads from these two replicates to generate one set of combined expression values . For DD2 , we employed only dataset 1 in the analysis . As a consequence of improved sequencing technology , samples of the DD2 dataset 2 contained roughly the same number of reads as the combined DD1 datasets 1 and 2 . Thus , the total number of reads analyzed for each sample was similar across all time points of DD1 and DD2—on average ∼32 million reads per sample . The resulting datasets ( six time points for both DD1 and DD2 ) were quantile normalized to control for variation among experiments . Relative sequence read coverage at different circadian time points , quantified using HTseq-count and quantile normalized , were used to construct a time-lapse expression series and analyzed using two different programs , ARSER and JTK_CYCLE [27] , [28] , to identify the presence of circadian periodicity . ARSER was developed by Yang and Su [27] , and it analyzes circadian expression data by harmonic regression based on autoregressive spectral estimation; JTK_CYCLE was developed by Hughes et al . [28] , and uses a nonparametric algorithm to detect rhythmic components in genome scale datasets . Results obtained from the two different analyses were filtered in several ways to obtain the final set of cycling genes: ( 1 ) we required the average raw read counts across the 12 time points to be at least 20; ( 2 ) we required a “cycling amplitude , ” defined as ½ ( maximum expression value – minimum expression value ) /median expression value , of at least 0 . 5; and ( 3 ) for results with the ARSER program , p<0 . 021 was considered statistically significant , whereas for the JTK_CYCLE program , p<0 . 015 was used as a cutoff . As the two programs appear to have different sensitivities in detecting circadian genes , different cutoff p values were chosen for them in order to include the majority of known clock genes . We think the use of this biological criterion to determine a statistical cutoff is reasonable for this type of analysis . One-week-old adult flies expressing EGFP-mL10a in all clock cells—that is , carrying one copy each of tim-uas-gal4 and UAS-EGFP-L10a—were entrained to a LD 12∶12 cycle for 3 d at 25°C and flash frozen in liquid nitrogen at ZT8 on the 4th day . Three sets of samples , each containing about 200 flies , were collected . Head collection , homogenization , TRAP , and RNA isolation were carried out as described above in “Affinity Purification of Ribosomes and Isolation of Ribosome-Bound mRNAs . ” Before the immunoprecipitation step , 1/10 of the tissue lysate was set aside for extraction of total RNA . RNAs were isolated from the TRAP immunoprecipitates ( referred to as “TRAP RNA” ) as well as from the input whole head lysates ( referred to as “total RNA” ) . Equal amounts ( 300 ng ) of TRAP RNA and total RNA were used to construct RNA-seq libraries . For each of the three sets of fly heads , one TRAP RNA library and one total RNA library were constructed . RNA-seq library construction , sequencing , and mapping were conducted as described above . Sequence read counts were obtained using HTSeq ( EMBL ) with BDGP5 , Ensembl release 68 for gene coordinates . Normalized sequence read counts were used to test for differential expression between the TRAP RNA samples and whole head total RNA samples . Differential expression was determined using the DESeq package for R [66] . Genes that showed significantly increased abundance in the TRAP RNA samples were considered to be enriched in clock cells . Adult or larval brain and ventral ganglion were dissected in PBS ( 137 mM NaCl , 2 . 7 mM KCl , 8 mM Na2HPO4 • 2 H2O , 2 mM KH2PO4 , pH 7 . 4 ) under a dissecting microscope and fixed in 4% paraformaldehyde . After fixation , tissues were washed three times with PBST ( PBS with 0 . 1% Triton-X-100 ) , blocked with 5% Normal Goat Serum ( NGS ) in PBST for 3 h , and incubated with primary antibody solution in PBST with 2% NGS at 4°C overnight . Anti-PER , anti-LARK , anti-TDC2 , and anti-PDF primary antibodies were diluted 1∶15 , 000 , 1∶2 , 000 , 1∶300 , and 1∶20 , respectively . The primary antibody solution was removed the next day and tissues were washed five times in PBST and incubated with fluorescence-conjugated secondary antibody for 3 h ( Cy3 conjugated goat anti-rabbit secondary antibody from Jackson ImmunoResearch for PER , LARK , and TDC2; Alexa Fluor 488 or Alexa Fluor 647 conjugated goat anti-mouse secondary antibody from Invitrogen for PDF ) . Following incubation with secondary antibody , tissues was washed five times in PBST and mounted on slides in VECTASHIELD mounting media ( Vector Lab ) . Florescence microscopy of brains was conducted using either a Leica SP2 confocal microscope at the Tufts Center for Neuroscience Research ( CNR ) Imaging Core or a Leica SP8 confocal microscope at the Enhanced Neuroimaging Core of the Harvard NeuroDiscovery Center . GFP , Cy3 , and Alexa Fluor 647 were excited using laser light of 488 nm , 561 nm , and 647 nm , respectively . Fluorescence excitation and image acquisition in the three different channels were performed in a sequential manner to avoid signal bleed-through between channels . One-micron optical sections were acquired in the vicinity of the LNv and LNd neurons using a 63× oil objective . Brain specimens collected from ZT1 and ZT9 were imaged in an alternating order so that every ZT1 image was paired with a ZT9 image and paired t tests were used in the final statistical analyses of image quantification . Such analyses minimize random variation due to fluctuation in laser power . To quantify TDC2 immunoreactivity in l-LNv and LNd , Regions of Interest ( ROIs ) were manually selected to include all l-LNv cells or all LNd cells based on PDF immunoreactivity ( for l-LNv ) or expression of tim-uas-gal4–driven mCD8-GFP in the appropriate region ( for LNd ) . A custom Image J program was used to calculate the average pixel intensity across the entire stack within the ROI for all pixels that had an intensity value greater than that of a manually selected background region . Quantitative real-time PCR was conducted on a Stratagene Mx3000P or Mx4000 QPCR system using SYBR Green Real-time PCR Master Mix ( Applied Biosystems ) . Primer sequences are listed in Table S5 . Primers were tested to be sure a single product was amplified with the expected melting temperature . A primer pair for an abundant noncycling gene , Rp49 , was used in all samples to serve as an internal control for the amount of starting material . The relative abundance of a gene of interest was calculated based on the difference between the Ct value of the specific primer pair and that of the Rp49 primer pair . | The circadian clock controls daily rhythms in physiology and behavior via mechanisms that regulate gene expression . While numerous studies have examined the clock regulation of gene transcription and documented rhythms in mRNA abundance , less is known about how circadian changes in protein synthesis contribute to the orchestration of physiological and behavioral programs . Here we have monitored mRNA ribosomal association ( as a proxy for translation ) to globally examine the circadian timing of protein synthesis specifically within clock cells of Drosophila . The results reveal , for the first time in any organism , the complete circadian program of protein synthesis ( the “circadian translatome” ) within these cells . A novel finding is that most mRNAs within clock cells are translated at one of two predominant circadian phases—midday or mid-night—times of low energy expenditure . Our work also finds that many clock cell processes , including metabolism , redox state , signaling , neurotransmission , and even protein synthesis itself , are coordinately regulated such that mRNAs required for similar cellular functions are translated in synchrony at the same time of day . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [] | 2013 | Translational Profiling of Clock Cells Reveals Circadianly Synchronized Protein Synthesis |
Advances in high-throughput , single cell gene expression are allowing interrogation of cell heterogeneity . However , there is concern that the cell cycle phase of a cell might bias characterizations of gene expression at the single-cell level . We assess the effect of cell cycle phase on gene expression in single cells by measuring 333 genes in 930 cells across three phases and three cell lines . We determine each cell's phase non-invasively without chemical arrest and use it as a covariate in tests of differential expression . We observe bi-modal gene expression , a previously-described phenomenon , wherein the expression of otherwise abundant genes is either strongly positive , or undetectable within individual cells . This bi-modality is likely both biologically and technically driven . Irrespective of its source , we show that it should be modeled to draw accurate inferences from single cell expression experiments . To this end , we propose a semi-continuous modeling framework based on the generalized linear model , and use it to characterize genes with consistent cell cycle effects across three cell lines . Our new computational framework improves the detection of previously characterized cell-cycle genes compared to approaches that do not account for the bi-modality of single-cell data . We use our semi-continuous modelling framework to estimate single cell gene co-expression networks . These networks suggest that in addition to having phase-dependent shifts in expression ( when averaged over many cells ) , some , but not all , canonical cell cycle genes tend to be co-expressed in groups in single cells . We estimate the amount of single cell expression variability attributable to the cell cycle . We find that the cell cycle explains only 5%–17% of expression variability , suggesting that the cell cycle will not tend to be a large nuisance factor in analysis of the single cell transcriptome .
With the advent of single cell expression profiling [1]–[4] , the assessment of cell population heterogeneity and identification of cell subpopulations from mRNA expression is achievable [5]–[7] . However , at the single cell level , there is concern that cell cycle might interfere with the characterization of gene expression variability [8] . As many biological samples are prepared from asynchronous cell populations , where each cell is in an unknown phase of the cell cycle , it is imperative to understand the impact of cell cycle in order to account for its effect on observed expression patterns and downstream data analysis . Here , we have measured mRNA expression and cell cycle from 930 single cells derived from three cell lines in order to explore this hypothesis . A distinctive feature of single-cell gene expression data is the bimodality of expression values . Genes can be on ( and a positive expression measure is recorded ) or off ( and the recorded expression is zero or negligible ) [9] , [10] . This dichotomous characteristic of the data prevents use of the typical tools of designed experiments such as linear modeling and analysis of variance ( ANOVA ) . We develop a novel computational framework to overcome this problem . First , a probabilistic mixture model-based framework allows the separation of positive expression values from background noise using gene-specific thresholds . After signal separation by thresholding , we model separately the frequency of expression ( the fraction of cells expressing a gene ) and the continuous , positive expression values . Our semi-continuous framework combines evidence from the two salient parameters of single cell expression in a statistically appropriate manner , an approach dubbed the Hurdle model [11] , [12] . Extending our previous proposal of a two-sample semi-continuous test akin to the two-sample t-test , our new framework allows for testing arbitrary contrasts and allows the use of variance components/mixed models , thus bringing to bear the full power of the general linear model . The Hurdle model allows us to identify many genes with an archetypal cell cycle expression pattern despite a frequently bimodal distribution of expression . It also suggests that stochastic variation in single cell gene expression is relatively large compared to the effect of cell cycle . We find that even in the most tightly regulated gene , cell cycle explains only 27% of the variability , while in the median gene in our data set , cell cycle explains 5%–18% of the variability , depending on the assumptions we make regarding latent technical variability . The semi-continuous model also provides a framework for estimating co-expression networks– in which edges connect genes whose partial correlations remain after removing the effect of all other genes – while adjusting for population-level nuisance factors that could bias network inference . Applying this framework to our data , we show that only a subset of canonical cell cycle genes are highly co-expressed in single cells .
In order to assess differential expression associated with actively cycling cells , expression of 333 genes was interrogated in 930 cells , across three cell lines: H9 ( HTB-176 ) , MDA-MB-231 ( HTB-26 ) , and PC3 ( CRL-1435 ) ( Figure 1A ) . Single cell expression was measured from flow-sorted cells and compared between cell cycle phases and cell lines via nCounter single cell profiling , a multiplexed hybridization-based detection technology that utilizes fluorescent barcodes to count individual target nucleic acid molecules [13] . This platform has been recently adapted to enable expression profiling from single cells via hybridization after a multiplexed target enrichment ( MTE ) in which mRNA is first converted to cDNA and then amplified [14] . Each cell was categorized as being in G0/G1 , S or G2/M phase by measuring DNA content via flow-cytometry based on retention of Hoechst dye ( Figure 1B and S1 ) [15] . Probes were selected for cell cycle associated genes ( n = 119 ) . These genes provided coverage of the entire cell cycle ( Data Set S1 ) based on peak expression and periodicity information obtained from Cyclebase , an integrated database of bulk cell cycle expression profiling experiments that scores and ranks genes based on strength of evidence for a cell cycle associated expression pattern [16] . Probes were also included for non-cell-cycle associated genes with primary roles in the inflammatory response , and housekeeping controls without a Cyclebase ranking ( n = 214 ) . We denote probes with a Cyclebase rank ( i . e . genes with the strongest evidence for cell cycle associated periodic expression ) as the ranked set . 253 genes were expressed and passed quality control ( see Methods ) . Genes showed a bimodal expression pattern in log-transformed mRNA levels ( Figure 2 ) , consistent with a burst-model of “on/off” transcription at the single cell level [17] and consistent with the kinetics of PCR amplification with low starting template concentrations , described by us and other authors [9] , [10] . Expression levels for each gene were most different between cell lines ( Figure 2 ) . Many genes , including those in the ranked set showed cell line-specific expression patterns . For example , expression of TOP2A in G0/G1 varied from 70% of cells in MB-231 and PC3 to nearly universal in H9 . This cell line effect was a nuisance factor we needed to adjust for in differential expression tests on cell cycle . Nonetheless , many genes from the ranked set , such as KIF23 , TOP2A , HJURP , NUSAP1 , and TPX2 exhibited expression patterns consistent with cell cycle regulation ( Figure 2 ) . Figure 2 also reveals that changes in both the positive expression mean ( i . e . the mean over the cells expressing that gene; PEM ) , and changes in the frequency of cells expressing a gene , occur throughout the cell cycle . The frequency and PEM in these genes also vary widely between cell lines , so it was important to adjust for cell line effects for accurate assessment of differential expression . In order to test for significant differences in expression between cell cycle phases that were consistent across cell lines , we developed an ANOVA-like model ( Hurdle model , see Methods ) that permits adjustment for additive effects due to cell line . The Hurdle model improves the power to detect changes in single-cell expression by testing both the frequency of expression ( corresponding to the relative distribution of cells between the two modes ) , and the PEM . Combining evidence from the discrete and continuous components of the data provides better sensitivity to changes in expression compared to test statistics based on frequencies of expression ( discrete ) or on the PEM ( continuous ) alone; or a union test ( see Materials and Methods ) while remaining competitive in specificity ( Figures S3 , S4 ) Within the three cell lines tested here , significant differential expression ( Bonferroni-adjusted for 253 tests at P<0 . 05 ) was observed for 78 genes in the ranked set and 28 genes in the unranked set ( Figure 3A ) . Genes showing the strongest cell cycle associated expression patterns in bulk measurements were more likely to be identified as significant in the single-cell populations ( Figure 3A–B ) . For each gene , peak time was determined based on the phase ( G0/G1 , S or G2/M ) with maximum average expression across all cell lines . Despite large cell-line-specific expression variability , peak times were broadly consistent with Cyclebase annotations ( Figure 3C ) , and especially so within the subset of genes with strongest evidence of cycle regulation in our data ( e . g . Bonferroni significant at P<0 . 05 ) . The majority of genes in the unranked set ( 115/143 or 80% ) did not exhibit significant cell cycle effects , in concordance with their primary roles in functions unrelated to the cell cycle . Of the 28 unranked genes that exhibited a significant cell cycle phase association , we noted genes involved in cytoskeletal organization ( PLAT ) , proliferation ( PDGFA ) , and signaling pathways ( IFNA1 , IFNB1 ) that have been previously demonstrated to modulate progression through the cell cycle [18] . It has been argued that a substantial portion of the stochastic variability observed in single cell gene expression experiments may be caused by global changes in transcription due to cell cycling [19] . We explore this idea by examining the proportional change in the Hurdle model fit associated with inclusion and omission of cell cycle as an explanatory variable . Because the Hurdle model accounts for both the dichotomous ( on/off ) and continuous nature of single cell data , the change in deviance ( generalized linear model log-likelihood ) between nested models can be used to calculate the amount of variability explained by cell cycle . The total deviance can be partitioned into components corresponding to cell cycle effects , nuisance effects described below , and residual effects . The ratio of cell cycle deviance to the sum of cell cycle plus residual deviance can then be interpreted as the analog to the coefficient of determination in linear least squares . We consider expression changes due to main effects and interactions of cell cycle by cell line and account for amplification efficiency and average cell line effect ( see Materials and Methods ) . Only modest amounts of the single cell expression variability can be explained by cell cycle ( Figure 4 ) . Within the ranked gene set , cell cycle phase explains 8% of the deviance in the median gene and 27% of the deviance in the top gene ( TOP2A ) . In unranked genes , phase explains only 5% of the deviance in the median gene . To derive these estimates , it is important to be able to account for the nuisance factors by using the Hurdle model . If cell-to-cell variation in amplification efficiency is not removed , we underestimate the explanatory power of cell cycle on in the median ranked gene by 26% since the unmodeled deviance would include this large additional component . Similarly , other unmeasured factors may inflate the residual deviance and attenuate the apparent role of cell cycle . These factors could include errors in inferring the cell cycle phase via FACS or imperfect modeling of changes in amplification or detection efficiency between samples . To guard against this attenuation , we set an upper bound on cell-cycle-dependent variation as follows: We suppose that transcription of the gene with the most deviance attributable to cell cycle ( TOP2A , 27% ) would be entirely regulated in a phase-dependent manner , and we characterize other genes' cell-cycle-dependent deviance relative to this maximum . For example , a gene with 13 . 5% cell-cycle-dependent deviance has half as strong a cell cycle effect as TOP2A , leading to the conclusion that at most 50% of this gene's deviance could be attributable to cell cycle . Even under these generous upper bounds , cell cycle phase explains only 18% ( eg , . 05/ . 27 ) and 29% ( eg , . 08/ . 27 ) of the deviance in the median gene in the unranked and ranked sets , respectively , suggesting that even when allowing for cell line-specific cell cycle effects , cycle is generally a small factor , compared to residual variability , in gene expression variability in the human transcriptome . Single-cell gene expression data sets have the resolution to reveal not only differential expression in response to biological variables like cell cycle phase , but also to provide insight into co-expression between genes at the cellular level ( e . g . the influence of one gene on another's expression or the sharing of upstream regulatory elements ) . In bulk-gene expression data ( e . g . microarrays ) , apparent co-expression arises from tissue-level factors inducing shared marginal changes in genes . For example , different radiation doses in samples will induce correlation amongst all the genes affected by radiation , regardless of whether these genes interact or even participate in the same biological processes . In contrast , single cell data allow isolation of co-expression arising from cellular-level factors , giving access to more fundamental biological relationships . If two genes are correlated across cells drawn from the same environment , then the two genes are likely to share an intimate biological relationship: they may be regulated by the same transcription factor , or one gene may directly regulate the other . The distinction between cellular and marginal co-expression follows from a probabilistic identity on conditional covariances ( see Materials and Methods ) . When cell cycle is not adjusted for ( Figure 5 D–F ) , known cell cycle genes with strong evidence of marginal regulation comprise the majority of the network . These genes generally peak in phase G2/M , suggesting that the co-expression is mostly driven by the coincident peak in average expression . The networks adjusted for cell cycle at least partially remove marginal effects ( Figure 5 A–C ) . In some cell cycle genes , substantial evidence for co-expression remains , but now additional co-expression is detected in genes without a previously described cell cycle role . In the unadjusted estimates , marginal shifts in expression in canonical cell cycle genes overwhelm subtler co-expression in unranked genes . Even though cell cycle variability is modest compared to residual variability , cell cycle is a substantial source of biological variability in the ranked genes and is in a sense confounded with the co-expression patterns . In an attempt to quantify the performance of the Hurdle model and the effect of cell-cycle adjustments , we examined network properties when varying the number of edges . We call an edge peaktime concordant if it connects nodes that have the same peaktime annotated in cycle base ( eg , G0/G1-G0/G1 or S-S ) . Over a range of network densities ( 30–240 edges ) the unadjusted Hurdle or Raw networks contain between 45%–80% peaktime concordant edges , while the adjusted Hurdle contains only 32%–38% peaktime concordant edges . Cell cycle adjustment in networks estimated on the raw data is not very effective compared to the unadjusted , raw networks ( Figure S6 ) . This is unsurprising , as this would occur when the model for the mean of the response is mis-specified , as is true when ignoring the bi-modality that the data exhibit ( eg , Figures 2 and S2 ) . If the Hurdle model is correct and cell cycle is additive , then the identity link cannot recover this additivity . On the other hand , the Hurdle model can still recover an additive mean model under a linear link by taking the discrete coefficient estimates to be null . Overall , the adjusted and unadjusted Hurdle networks in Figure 5 are rather different , sharing 39% of nodes ( Jaccard similarity ) and 51% of edges ( Hamming Distance/#edges ) . Combining both discrete and continuous networks ( with the top 30 edges from discrete and continuous networks ) allows a richer set of genes to be characterized . When discrete expression is used alone , networks primarily consist of G2/M peaking genes and unranked genes ( Figure 5A ) . When positive , continuous expression is also used , S and G0/G1 peaking genes enter the networks ( Figure 5B–C ) . The adjusted , semi-continuous network depicted in Figure 5C consists of two primary sub-networks , one consisting entirely of ranked genes , and another largely consisting of weakly ranked and unranked genes . While we cannot rule out that measurement error of the inferred cycle is not partially responsible for the persistence of a subset of ranked genes , previously described mutual regulation in RNA-interference experiments [20] of some of these genes suggests that this subset is co-expressed at the single cell level as opposed to being co-expressed on average at the population level . The sub-network of ranked genes contains the central node of NUF2 , a highly-conserved protein required for stable kinetochore localization of centromere-associated protein E ( CENP-E ) [21] . NUF2 is connected to other actors in mitotic organization such as ANLN , KIF23 , and CENPF , as well as the check-point genes CCNA2 and BUB1 , reflecting the central role of these genes in mitosis . The sub-network of primarily unranked genes contains two key nodes: TUBB and CCR3 . The predominance of genes associated with cell growth , like TUBB , and transmembrane proteins , like CCR3 , in the unranked cluster is likely related to the actively dividing nature of the profiled cells , i . e . dividing cells must generate new scaffolding and membrane-related materials to support growth . This relatively large sub-network of unranked and weakly ranked genes is largely missed by the unadjusted analysis that is biased by the population level cell-cycle effect .
Stochastic , bimodal expression is a hallmark of single cell data [22]–[24] . Within a population of cells , detectable expression for any given gene typically resides in one of two modes , corresponding to an “on” or “off” state . Both technical and biological factors likely contribute to this bimodality . Quantities of some species of cDNA may be minute after reverse-transcription , and in this case random variation in the number of template-primer-enzyme complexes that form during each annealing phase may dominate the kinetics of the PCR [25] . But regardless of its origin , modeling bimodality improves the power of differential expression tests . Here , we show how the Hurdle model can be adapted to complex study designs , extending our previous results describing its use for two-sample comparisons . We demonstrate the model's ability to identify many genes with a periodic expression pattern from asynchronously cultured cells utilizing a combination of FACS sorting and these new analytical techniques , including genes with little previous evidence of cell cycle associated periodic expression like MEF2D [26] and FAM189B . The Hurdle model is able to identify phase-dependent patterns of expression despite the fact that G2 and M phases are indistinguishable by DNA content . The similar rank ordering of differentially expressed genes in our single cell experiment as compared to bulk experiments and concordance in the phase of peak expression demonstrates the power of the Hurdle model . While we have applied the Hurdle model to our specific problem , the approach is general and can be applied to test any effect of interest in a single-cell gene expression dataset . We offer this modeling framework as an R package for other interested users at github . com/RGLab/SingleCellAssay . Although we recommend the Hurdle model in general for testing for differential expression , it should be noted that its desirability is contingent on the frequency of the gene under consideration . For example , if a gene is highly expressed ( eg , >90% expression ) , then the information to be derived from the 10% of cells that do not express a gene may not be worth the cost of an extra degree of freedom in the chi-square null distribution of the test statistic . However , even when this is the case , the Hurdle model might be preferred for methodological simplicity , since it is powered—although perhaps not always optimally—regardless of expression frequency , and does not require extensive pre-test simulations of power to yield acceptable performance . The data set considered here offers a relatively stringent test of the relative sensitivity of the Hurdle model , owing to the high expression frequency of the genes in this experiment ( interquartile range ranked genes: . 7– . 9; unranked genes: . 56– . 88 ) . Single cell data also allows unparalleled resolution of genes' co-expression patterns . While bulk expression data can reveal correlation induced by varying biological conditions , single-cell data has the possibility to reveal co-expression driven by shared regulatory elements within the cell . However , when inferring gene expression networks , it is important to adjust for population level covariates that could confound the network estimation , especially for genes that are marginally affected by such a population level covariate ( like known cell cycle genes in our experiment . ) By measuring a limited set of cell cycle associated genes , we are able to identify a network of co-expressed genes with known roles in cell cycle regulation even after adjusting for cell cycle phase . It should be noted that the unadjusted network estimate would be appropriate in some circumstances , for example when a summary of the co-expression occurring on average in the population of cells is desired , as opposed to inference of co-expression occurring conditionally within defined subsets . Work remains to derive network estimators that optimally combine information from discrete and continuous portions . Our current approach is likely theoretically naïve , since it is essentially a union test of the discrete and continuous portions , rather than a summation of signal from the two domains . We also have left unresolved the asymptotic consistency of our proposed network procedure under dimensional scaling . It is crucial to understand the relationship between cell cycle and the stochastic nature of single cell expression as it determines the magnitude of the cell cycle's distorting effect on single cell analyses . In contrast to earlier estimates of Zopf et al . [19] we find little evidence of periodic regulation of expression among non-cell cycle associated genes . Our results are consistent with genome-wide mRNA profiling efforts utilizing bulk expression methodologies in mammalian cells where genes with cycle-dependent periodic expression patterns are limited and well-characterized [16] , [27] , [28] . Disparity between our findings and those of Zopf et al . may arise from differences between yeast and mammalian cells . Moreover , Zopf et al . primarily focus on a single , synthetic promoter while we sample hundreds of transcripts presumably driven by many different promoters . Whether the substantial remaining variability is inherent to the human single cell , or due to thus far latent , unmeasured biological variables remains to be explored .
Three human cell lines H9 ( HTB-176 ) , MDA-MB-231 ( HTB-26 ) and PC3 ( CRL-1435 ) were commercially obtained and cultured as recommended by the supplier ( ATCC ) . Cultured cells were re-suspended in culture media containing Hoescht 33342 ( Sigma ) and incubated at 37°C for 60 minutes prior to sorting . Cultured cells were flow-sorted to isolate individual cells from each of the cell lines according to phase ( G0/G1 , M/G2 and S ) . Cells were isolated and sorted using the FACSJazz ( Becton Dickinson ) at 500 events per second using a 100 micron nozzle . Single cells were defined by gating on forward and side scatter area/width . Phase was inferred from Hoescht 3342 DNA-fluorescent dye , then cells were individually deposited and lysed in wells of a 96-well PCR plate containing 3 uL of Cells-to-Ct lysis buffer ( Life Technologies ) . The proportion of cells in G0/G1 phases varied from 54% of PC-3 cells to 73% of H9 cells ( Supplementary Figure S1 ) . A set of 333 probes was designed . It contained cell cycle associated genes and provided coverage of the entire cell cycle based on peak expression and periodicity information derived from an integrated database of cell cycle expression profiling experiments [16] . Non-cell cycle associated genes had primary roles in the inflammatory response and included housekeeping controls without a Cyclebase ranking . Genes with a Cyclebase ranking <1000 were placed in the ranked set ( n = 119 ) and all other probes were considered part of the unranked set ( n = 214 ) . After lysis , RNA was converted to cDNA with SuperScript VILO ( Life Technologies ) . Primers for 333 genes were pooled and cDNA was enriched in a multiplexed amplification ( MTE ) reaction according to the nCounter Single Cell Expression protocol ( NanoString ) . The MTE samples were hybridized overnight at 65°C with an nCounter CodeSet containing probes for all enriched targets ( cell cycle related , unrelated genes and controls ) and internal controls as recommended by the manufacturer . In single cell gene expression , we have previously found that accounting for both changes in the frequency of expression and shifts in the PEM produces more sensitive measures of differential expression compared to using either the frequency or the positive values alone , or compared to t-tests on the zero-inflated values [9] , [33] . We sought to extend this framework to any model that permits a likelihood ratio test on parameters , e . g . , generalized linear or generalized linear mixed models , in order to account for additive cell line effects . Let denote the expression threshold in the kth cell ( so thus suppressing the gene index ) . Then we model ( 1 ) where , are cell line effects , , are cell cycle effects and , are interaction effects between cell line and cell cycle , and is an independent , normally distributed error . The indices and give the cell line and cell cycle of the k th cell . The cell line effects , and cell cycle effects , are vectors in , although with the linear constraint that the sum of them is zero , eg , , while , is a matrix in with the constraints that for and for . The term accounts for cell-to-cell technical variability resulting from variation in reverse transcription and PCR amplification efficiency ( see previous section ) . Jointly modeling the PCR efficiency along with the biological effects of interest is important as one factor can affect the other . Our modeling framework can be extended to regression-type models when the right hand side is replaced with a general term for each component , and even to generalized linear mixed models . In general , let be a vector of parameters for the distribution of and let be a vector of parameters for . Then when the distribution of is divided in this fashion , inference about proceeds conditional on . The log likelihood is then additive in the and parameters . Classical hypothesis tests with chi-square asymptotic null distribution , such as Wald or likelihood ratio tests on specific components of and are null can be conducted separately . Then the test statistics are added together , combining and summarizing the evidence from the two processes , with the degrees of freedom in the null distribution doubled for the purpose of assigning significance . This approach is dubbed the “Hurdle” model and has been used in economics for several decades [34] , [35] . We extend the conditional , neighborhood-based algorithm of Meinshausen-Bulmann [36] to estimate co-expression networks using the Hurdle model . The standard Meinshausen-Bulmann algorithm uses L1-penalized regressions to estimate partial correlations between vertices ( genes ) by treating each vertex as a dependent variable in a regression that includes all other vertices as independent variables . If the vertices are jointly Gaussian , non-zero coefficients correspond to statistical dependences between vertices , conditional on all other factors and so reflect a Gauss-Markov Random Field . Here , since the distribution of expression in single cells is not multivariate Gaussian , edges in our network correspond to conditional correlations ( after possible application of the logit link ) . Although we do not attempt to show consistency of our proposed approach here , we note that Meinshausen-Bulmann-like methods have been shown to be consistent in estimating non-Gaussian graphical models under fairly general conditions [37] , [38] . Then for the k th cell , following equation ( 1 ) , we divide expression into discrete and continuous components , so fit regressions of the form ( 3 ) where is the expression of the gth gene in the kth cell , and is the expression vector of all except the gth gene in the kth cell , and is a vector of cellular covariates ( eg pre-amplification effect , cell line , cell cycle , and their interaction ) . We estimate and separately , with distinct L1 penalties and for and using the R package glmnet [39] . Unpenalized vector parameters and adjust for pre-amplification effect ; cell line and cell cycle . | Recent technological advances have enabled the measurement of gene expression in individual cells , revealing that there is substantial variability in expression , even within a homogeneous cell population . In this paper , we develop new analytical methods that account for the intrinsic , stochastic nature of single cell expression in order to characterize the effect of cell cycle on gene expression at the single-cell level . Applying these methods to populations of asynchronously cycling cells , we are able to identify large numbers of genes with cell cycle-associated expression patterns . By measuring and adjusting for cellular-level factors , we are able to derive estimates of co-expressing gene networks that more closely reflect cellular-level processes as opposed to sample-level processes . We find that cell cycle phase only accounts for a modest amount of the overall variability of gene expression within an individual cell . The analytical methods demonstrated in this paper are universally applicable to single cell expression data and represent a promising tool to the scientific community . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"mathematics",
"cell",
"biology",
"statistics",
"(mathematics)",
"gene",
"expression",
"genetics",
"biology",
"and",
"life",
"sciences",
"molecular",
"genetics",
"physical",
"sciences",
"computational",
"biology",
"molecular",
"cell",
"biology",
"biostatistics",
"statisti... | 2014 | Modeling Bi-modality Improves Characterization of Cell Cycle on Gene Expression in Single Cells |
The current article suggests that deterministic chaos self-organized in cortical dynamics could be responsible for the generation of spontaneous action sequences . Recently , various psychological observations have suggested that humans and primates can learn to extract statistical structures hidden in perceptual sequences experienced during active environmental interactions . Although it has been suggested that such statistical structures involve chunking or compositional primitives , their neuronal implementations in brains have not yet been clarified . Therefore , to reconstruct the phenomena , synthetic neuro-robotics experiments were conducted by using a neural network model , which is characterized by a generative model with intentional states and its multiple timescales dynamics . The experimental results showed that the robot successfully learned to imitate tutored behavioral sequence patterns by extracting the underlying transition probability among primitive actions . An analysis revealed that a set of primitive action patterns was embedded in the fast dynamics part , and the chaotic dynamics of spontaneously sequencing these action primitive patterns was structured in the slow dynamics part , provided that the timescale was adequately set for each part . It was also shown that self-organization of this type of functional hierarchy ensured robust action generation by the robot in its interactions with a noisy environment . This article discusses the correspondence of the synthetic experiments with the known hierarchy of the prefrontal cortex , the supplementary motor area , and the primary motor cortex for action generation . We speculate that deterministic dynamical structures organized in the prefrontal cortex could be essential because they can account for the generation of both intentional behaviors of fixed action sequences and spontaneous behaviors of pseudo-stochastic action sequences by the same mechanism .
Our everyday actions are full of spontaneity . For example , imagine that a man makes a cup of instant coffee every morning . After he pours hot water into his mug , which is already filled with a spoonful of coffee crystals , he may either add sugar first and next add milk , or add milk first and then add sugar . Or , sometimes he may even forget about adding sugar and notice it later when he first tastes the cup of coffee . Some parts of these action sequences are definite , but other parts are varied because we see spontaneity in the action generation . Similar actions can be seen in improvisations in playing jazz or in contemporary dance , where musical phrases or body movement patterns are inspired freely from one to another in an unpredicted manner . An essential question is from where the spontaneity for generating voluntary actions or images originates . The current article presents a model prediction for the underlying neural mechanism . We speculate that the necessary neural structures for generating such spontaneous actions are acquired as the results of learning from everyday experiences and practices while interacting with the world . Gibson and Pick [1] once wrote that infants are active learners who perceptually engage their environments and extract information from them . In their ecological approach , learning an action is not just about learning a motor command sequence . Rather , it involves learning the possible perceptual structures extracted during intentional interactions with the environment . Other researchers [2]–[4] have proposed that perceptual structures experienced during environmental interactions can be acquired by using forward models that are assumed to be located in the cerebellum . A forward model outputs the prediction of the next sensation by receiving the inputs of the current sensation and motor commands . Although this idea is generic and theoretical in the sense that a forward model can predict the sensory outcomes of arbitrary motor commands at every time step , it is impossible in practice due to the combinatorial explosion problem if the motor dimension becomes large . This problem is analogous to the frame problem [5] that discusses that there are no rational means to stop inferences about the outcomes of infinite action possibilities . In humans , learning by environmental interaction does not proceed with all possible combinations in motor command sequences , but with more purposeful or intentional stances [1] . Therefore , it could be sufficient for humans to predict sensory outcomes according only to purposeful behavior trajectories . With regard to this theory , Tani [6] proposed an alternative idea that an agent should learn mapping from a set of particular intentional states with the consequent sensory sequences that are expected to be experienced in corresponding purposeful environmental interactions . Ay et al . [7] also proposed a prediction based model to control an autonomous robot , but the model lacked internal states . In the connectionist implementation by Tani and his colleagues [6] , [8] , [9] , the proposed neuro-dynamic model is operated in three modes: generation , recognition and learning . In the generation mode , the corresponding visuo-proprioceptive sequence is predicted for a given intentional state . The scheme can generate mental simulation for future behaviors or motor imagery ( in terms of visuo-proprioceptive sequences ) [10] , and can also generate the corresponding physical movement by sending the next predicted proprioceptive state to the motor controller as the next target . In the recognition mode , a given visuo-proprioceptive sequence can be recognized by identifying the corresponding intentional state through an iterative search to minimize the prediction error . In the learning mode , the learning is formulated as a process to search for the optimal values for both a common synaptic weight matrix as well as a set of individual intentional states that can regenerate all visuo-proprioceptive sequences of an experience for training under minimum error criteria . This idea is formally related to “active inference” [11] , which can be regarded as a form of predictive coding [12] . Friston [13] showed that the three aspects of our neurodynamic model ( generation , recognition and learning ) can be unified in terms of minimising prediction error . Moreover , it is presumed that the neural structures acquired through intentional interactions with the environment should support “compositionality” [14] or chunk knowledge for generating and recognizing the variety of complex actions [15]–[17] . Diverse intentional actions can be generated by combining a set of reusable behavior primitives or chunks adaptively by following the acquired rules . For example , an attempt at drinking a cup of water can be decomposed into multiple behavior primitives , such as reaching for a cup , grasping the cup and moving the cup toward one's mouth . Each behavior primitive can be re-used as a component for other intentional actions , e . g . , reaching for a cup to take it away . Psychological observation on infant development as well as adult learning have suggested that chunking structures in perceptual streams can be extracted by statistical learning with sufficient amounts of passive perceptual experiences [18] , [19] as well as for active behavioral interactions [20] . Here , chunking structures are represented by repeatable sensory sequences within chunks and the probabilistic state transition among those chunks . From his observation of skill acquisitions for food processing by mountain gorillas , Byrne [21] proposed that actions can be acquired with statistical structures through imitation . It is said that juvenile gorillas take a few years to effectively imitate behaviors by observing the mothers' food processing behaviors , which are characterized by nondeterministic transition sequences of behavior primitives or chunks . After the skill acquisition by extracting the underlying statistical structures , the primitive sequences of the juveniles resemble those of their mothers . In another example , improvisers of jazz music make substantial efforts into developing vocabularies of musical patterns or phrases , which they then freely combine and vary in a manner that is sensitive to the on-going musical context [22] . In recent years , considerable evidence has been assembled in support of statistical learning for both musical pitch sequences [23] and rhythm [24] by organizing chunking structures . In searching for the neuronal mechanisms for chunking , a monkey electrophysiological study [25] showed that some neurons in the presupplementary motor area ( preSMA ) fire only at the beginning of each chunk in the regeneration of trained sequences . Also , a human behavior study [26] showed that inactivation of the preSMA by transcranial magnetic stimulation ( TMS ) affects the performance of regenerating sequences only when the TMS is applied between the chunks after extensive learning of the sequences with chunking structures . These studies suggest that the preSMA might play a crucial role in segmenting sequences into chunks . Sakai et al . [27] proposed a hierarchical function in a cortical network , in which the prefrontal cortex and the preSMA were responsible for the cognitive control of segmenting sequences and selecting the next chunk , whereas more motor-related areas , including the premotor cortex and the primary motor cortex , are responsible for processing within each chunk . This proposal agrees with the results of a monkey electrophysiological recording [28] , [29] , which showed that firing some cells in the preSMA and the SMA encode specific sequences of joystick movement patterns or specific transitions from one movement pattern to another . Model studies by the authors have shown that chunking by segmenting continuous sensory flow can be achieved by applying the criteria of prediction error minimization to connectionist models with local [17] , [30] and distributed [6] , [31] representation schemes . Those model studies have also shown that a hierarchy is indispensable in acquiring chunking structures . In the hierarchy , the lower level learns to acquire a set of behavior primitives and the higher level puts those primitives into sequences of intentional actions . The idea of intentional actions is analogous to that by Sakai et al . [32] . One essential question so far is how the next behavior primitives or actions can be decided by following the learned statistical expectation . For example , if we suppose that someone has learned that the next behavior primitive to use is either primitive-A or primitive-B , given an even chance from past experiences , it is plausible to consider that either the primitives can be decided by her/his conscious will or alternatively they can be determined automatically without consciousness . Philosophers have discussed for a long time whether humans have a “free will” to determine a next action arbitrarily , and , if the free will exists , how it is implemented in our minds [33]–[35] . Interestingly , recent neuropsychological studies on free decision [36] , [37] have suggested that the conscious will of initiating or selecting actions arbitrarily follows certain neural activity that does not accompany awareness . Libet et al . [36] showed that a conscious decision to press a button was preceded for a few hundred milliseconds by a negative brain potential , referred to as the readiness potential , which originates from the SMA . In a functional magnetic resonance imaging ( fMRI ) study , Soon et al . [37] demonstrated that brain activity is initiated in the prefrontal cortex ( PFC ) and the parietal cortex up to seven seconds before a conscious decision . It was further found that the observed brain activity can predict the outcome of a motor decision , which the subject did not consciously make , such as pushing the left button by the left index finger or the right button by the right index finger . This result implies the possibility that even though someone can believe that he had consciously decided a particular action among multiple choices , such as primitive-A and primitive-B , the conscious decision was not the direct cause of the action selection , but was the preceded neural activity without awareness . It is also plausible to assume a purely mechanist model in which an itinerant trajectory of neural dynamics [38]–[41] , instead of a conscious will , determines the next behavior primitive to be performed . The current article introduces a neuro-robotics experiment that examines how statistical structures hidden in skilled behaviors can be extracted via imitation learning and how the behaviors can be regenerated by following the statistical structures . The experiment uses a dynamic neural network model based on two essential ideas from our previous proposals . First , in the dynamic neural network model , a mapping from the initial states of the internal neural dynamics to the expected visuo-proprioceptive trajectories is self-organized , and the initial states encode various intentional states for the resultant behavioral trajectories [9] . Second , the model network employs multiple timescale dynamics [31] , [42] that allow self-organization of the functional hierarchy , which is analogous to the known cortical hierarchical network consisting of the PFC , the preSMA , and the M1 [43] , [44] for generating goal-directed actions . It is demonstrated that itinerant behaviors with accompanying spontaneous transitions among behavior primitives can be generated by reflecting the observed statistical structure and by assuring the robustness of the behavior primitives against possible noise during physical interactions , when deterministic chaos is self-organized at the higher level of the functional hierarchy ( the slow dynamics part of the network ) . Based on the results , the current article discusses the importance of deterministic neural dynamics in action generation because they account for both itinerant behaviors with accompanying spontaneous transitions of behavior primitives and intentional fixed behaviors ( repeatedly executable ) by the same dynamic mechanism . The article also discusses how the model prediction presented can be applied in future neurophysiological studies .
The proposed hierarchical dynamic neural network can be regarded as a generative model of visuo-proprioceptive inputs [6] , [11] , [31] , [45] . ( Precise mathematical descriptions are provided in the Methods section . ) The network was divided into three levels based on the value of the time constant . Time constant for each unit primarily determined the timescale of the activation dynamics of the unit ( see Figure 1 ) . The higher level consisted of slow neural units with a larger time constant ( ) , the middle level with a moderate time constant ( ) , and the lower level with a small time constant ( ) . The lower level was assembled with a set of gated modular networks [46] that interacted directly with the visuo-proprioceptive sequences . The higher level was mutually connected to the middle level but was not directly connected to the lower level . The middle level interacted with the lower level by sending the gate opening signals and receiving the sensory inputs . The parameters of this model were optimized to minimize sensory prediction errors ( i . e . , maximizing the probability of the predictions given the sensory data ) . In this sense , our model was concerned with , and only with , perception . Action per se , was a result of movements that conformed to the proprioceptive predictions of the joint angles . This means that perception and action were both trying to minimize prediction errors throughout the hierarchy , where movement minimized the prediction errors at the level of proprioceptive sensations . With this perspective , the high-level network provided predictions of the middle-level network that , in turn , provided predictions about which expert will be engaged for prediction at the lowest ( sensory ) level . This engagement depended upon the gating variables , which selected the most appropriate expert in the lowest level of the hierarchy . The network was trained to predict a set of given visuo-proprioceptive sequences by optimizing the following two types of parameters in order to minimize the prediction error: the synaptic weights and the initial state of the internal units in the whole network for each sequence [9] . This intuitively means that the learning involves determining the dynamic function of the network by changing the synaptic weights and also by inferring the intention or goal of each action sequence . The learning scheme was implemented by using the error back-propagation through time algorithm [47] . Although the biological plausibility of error back-propagation in neuronal circuits is a matter of debate , some supportive evidence [48] , [49] and related discussions [50] exist . We speculate that the retrograde axonal signal [50] conveying the error information might propagate from the sensory periphery area to the higher-order cortical area by passing through multiple stages of synapses and neurons for modulating the intentional states . After minimizing the error , each visuo-proprioceptive sequence of the training can be regenerated by setting the initial state of the internal units with the optimized value . Because the initial state specified the expected visuo-proprioceptive sequence , the initial states are considered to represent the intentions of generating specific actions . The forward dynamics of the trained network can generate motor imagery in terms of visuo-proprioceptive sequences by feeding back the sensory prediction computed at the previous time step into the current sensory inputs without the accompanying physical movements ( closed-loop operation ) . However , the physical movements can be generated by sending the next time step prediction of the proprioceptive states ( joint angles ) to the motor controller as the target ( open-loop operation ) . As one aspect of our work , we examine how the dynamic function of each level can be differentiated depending on the timescale differences . To study the timescale characteristics in more detail , we investigated cases applying various values of the time constant set in the higher-level network . Experiments on imitation learning of actions were conducted on a small humanoid robot platform ( Sony Corporation ) , and the movie of a demonstration is available ( http://www . bdc . brain . riken . jp/tani/mov/PLoS11 . html ) . The robot experiments on the aforementioned dynamic neural network model involved imitative training of the sequences of primitive actions and autonomous generation of those imitated behaviors . The target primitive action sequences to be imitated were designed with a statistical structure and with transitioning of the primitive actions , and the sequences were directly tutored to the robot , i . e . , a human assistant directly guided the movements of both hands of the robot by grasping them . In the beginning of each training sequence , the assistant guided both hands of the robot , which was positioned in front of a workbench ( see Figure 2 ) . A cubic object was placed on the workbench at one of three positions ( center , cm left of center , and cm right of center ) , and the assistance repeated primitive actions of grasping the object , moving it to one of the three positions and releasing it by guiding the hands of the robot while deciding the next object position randomly with equal probability ( ) . Although the hands of the robot were guided by the assistant , the visuo-proprioceptive sequences were recorded for later training . The neural network was trained in an off-line manner , since all training sequences gathered at each teaching session were used for the subsequent consolidation learning . Thus , the neural network learned to predict visuo-proprioceptive sequences on the basis of the experiences obtained during the imitative training session . Note that no explicit cues for segmenting the sequences into primitive actions were prepared . The related chunking structures were acquired via iterative experiences of the continuous visuo-proprioceptive sequences . After the imitative training , each training sequence was regenerated by setting the initial state with the optimized value . For the subsequent analysis , primitive action sequences were generated and recorded for far longer than the training sequences . For the detailed analysis on the possible dynamic structures acquired in the network model , the itinerant trajectories by motor imagery for longer steps were also generated . The network was trained by a learning scheme in which the higher-level time constant was , , and [45] ( see the Methods section for details ) . For each condition of the higher-level time constant , training trials were conducted for sample networks with different initial parameters and training data . Descriptions of the learning parameters are provided in the Methods section . The trained networks were tested for pseudo-generation by motor imagery , and it was shown that the networks regenerated primitive action sequences pseudo-stochastically in their deterministic itinerant trajectories . Here , pseudo-stochasticity denotes stochasticity observed through the discretization of deterministic continuous value sensory sequences into symbolically labeled primitive action sequences . The action generation test by motor imagery , also described in the Methods section , revealed that the trained neural networks were able to create novel sequential combinations of the primitive actions that were not contained in all teaching sequences . This implies that primitive actions can be generated pseudo-stochastically , as taught in all conditions . We also tested cases of training primitive action sequences having different probabilities of selected primitive actions in a specific object position , as detailed in Text S1 . It was observed that these probabilities were reconstructed in the generation of the primitive action sequences by motor imagery . The results suggest that the network is capable of extracting underlying statistical structures in the imitated primitive action sequences . Next , we tested the generation of actual actions of interacting with the physical environment by the robot . First , the behavior of the robot was generated from each acquired initial state with the higher-level time constant , set at . Although it was observed that the trained primitive action sequences can be regenerated exactly during the initial period ( 7 . 3 primitive action transitions on average ) , the sequences gradually became different from the trained ones . It was considered that this result was due to the initial sensitivity characteristics organized in the trained network . Then , we tested actual action generations for cases of different values of time constant . Although no particular difference was found between the cases with a different in the action generation test by the motor imagery mode , the stability in actual action generation in the physical environment was different for each . The success rate for moving the object without dropping it was evaluated for each trained network with a different time constant . Figure 3 shows the frequencies of the networks classified according to the success rate , where populations of individual networks were trained for each time constant . In all cases we found a network with a success rate , but the average success rate was different for each . The success rate for was higher than the success rates for the other values of . This result indicates that motor patterns can be generated stably in the physical environment when is set larger than . As mentioned in the previous section , the stability in actual action generation is dependent on the timescale characteristics . This fact implies that the developed dynamic structure is different for each timescale condition . In the following we discuss the characteristics of the self-organized functional hierarchy in terms of the timescale differences . As an example , Figure 4 illustrates the sensory motor sequences and neural states . It can be seen that an action primitive of moving the object to the left , to the right , or to the center consisted of a few different gate openings , which generated sequential switching of the stored reusable motor patterns , such as reaching for the object , picking up the object , and moving back its hand position to the starting posture . It is considered that the middle- and higher-level network dynamics encoded the combinations of these reusable motor patterns into primitive actions and further into their stochastically switching sequences . The auto-correlation for each is shown in Figure 5 to clarify the timescale characteristics . The auto-correlation measures the correlation between values at different points in time and is sometimes used to find repeating patterns . A high auto-correlation at time difference means that similar values appear repeatedly with a period of . We found a periodic pattern of auto-correlation common to both the visuo-proprioception and the middle-level network units in the cases . This periodicity is considered to occur because the periodicities of all primitive actions were approximately the same . Conversely , such a periodic pattern was not found in the higher-level network if the higher-level network had a relatively large time constant . The characteristics of auto-correlation seem to correspond to the functionality obtained by each subnetwork . To examine the functionality developed at each subnetwork , we investigated the effect of artificial lesions in the higher-level network . To model the artificial lesions , we removed all the neurons in the higher-level network . Figure 6 is a comparison of the trajectory of the object position captured by the vision sensor in the motor imagery mode for the normal case and that with the artificial lesions . In this figure , we used networks having a success rate in actual action generation . If the trajectory generated by the network traced the test data ( see Figure 2 ) , the network moved the object by correctly using the hands . When the time constant of the higher level was set as , the network was able to generate each single primitive action correctly , even if the higher level was removed . However , in this case , the capability for combining diverse primitive actions significantly deteriorated . For set at or , the removal of the higher level significantly affected the generation of each primitive action . This implies that the functions for generating each primitive action and for generating stochastic combinations of these actions were self-organized and became segregated between the higher and the middle/low levels if was set significantly larger than and . Otherwise , both functions were self-organized but not segregated . That is , the functions were distributed throughout the entire network . In this situation , a lesion in the higher-level network could affect the lower sensory-motor control level . The abovementioned functional segregation between levels could contribute to the stability in the action performance evaluated in Figure 3 . In addition , the maximum Lyapunov exponents for the subnetworks were computed while varying as , , and ( see Methods for details ) . Here , and were fixed at and , respectively . The computation was repeated for neural networks trained using different initial synaptic weights but under the same learning condition . If the Lyapunov exponent is found to be positive for specific subnetworks or for the entire network , then the intrinsic dynamics of the subnetworks or network are identified as chaotic . In chaotic dynamics , almost any minute change in an internal neural state brings about a drastic change in subsequent network outputs because of the dependence on the initial conditions . Therefore , it can be inferred that subnetworks having positive maximum Lyapunov exponents act to combine action primitives with pseudo-stochasticity . The results are summarized in Table 1 . The maximum Lyapunov exponent of the entire network was positive for all values of . This was expected because the network was able to generate pseudo-stochastic primitive action sequences regardless of the value of . However , the values of the maximum Lyapunov exponent were negative in the higher level and the middle level , when the higher-level network had a small time constant . These results indicate that the function to generate chaos was globally distributed over the entire network . In contrast , when the higher-level network was set with a large time constant , i . e . , , the maximum Lyapunov exponent of the higher-level network became positive in most cases ( 94 out of 100 network learning cases ) , whereas that of the middle-level network became negative . The abovementioned results demonstrate that if the time constant of the higher level is sufficiently larger than the time constants of the other regions , then chaotic dynamics are formed primarily in the higher-level network , separate from the other regions . These results agree with the results in the case having lesions . Furthermore , the abovementioned analysis on the artificial lesion cases indicates that the segregation of chaos from the middle and lower levels provides more stable motor generation . In summary , if we regard our neurodynamic model as a generative model of visuo-proprioceptive sequences , the anatomical segregation between sequential dynamics and motor primitives ( e . g . , within premotor and motor cortex , respectively ) emerges only when we accommodate the separation of temporal scales implicit in the hierarchical composition of those sequences . Finally , we examined how the generation of chaos that allows pseudo-stochastic transition among action primitives depends on the characteristics of the training sets . For this purpose , the networks with set at were trained by changing the length ( i . e . , number of transitions of primitive actions ) of the training sequences , as detailed in Text S1 . It was observed that the possibility of generating chaos is reduced as the length of the training sequences is reduced . If no chaos was generated , it was observed that neural activity in the higher-level network often converged to a fixed point some steps after the tutored sequences were regenerated . This result implies that the mechanism for spontaneous transition by chaos can be acquired only through training of long sequences that contain statistically enough probabilistic transitions for generalization .
The current experimental results revealed that the chaos self-organized in the higher-level network with slow-timescale dynamics facilitates spontaneous transitions among primitive actions by following statistical structures extracted from the set of visuo-proprioceptive sequences imitated , whereas primitive actions were generated in the faster-timescale networks in the lower level . The finding was repeatable in the robotic experiments , provided that the timescale differences were set adequately among the different levels of subnetworks . The results appear to be consistent with human fMRI recordings , which indicate that free-decision-related activity without consciousness is slowly built up in the PFC seconds before the conscious decision [37] . This buildup of activity in the PFC could initiate a sharp response in the SMA just a few hundred milliseconds before the decision [36] . Activation in the SMA leads to immediate motor activity [28] , and buildup of action-related cell activity in the PFC in the monkey brain takes a few seconds , whereas that in the primary motor area takes only a fraction of a second [51] . These observations support recent arguments concerning the possible hierarchy along the rostro-caudal axis of the frontal lobe . Badre and D'Esposito [43] proposed that levels of abstraction might decrease gradually from the prefrontal cortex ( PFC ) through the premotor cortex ( PMC ) to the primary motor cortex ( M1 ) along the rostro-caudal axis in the frontal cortex in both the monkey brain and the human brain [44] . Here , the rostral part is considered to be more integral in processing information than the caudal part in terms of its slower timescale dynamics . By considering the possible roles of the M1 , such as encoding the posture or direction of limbs [52] , [53] , it is speculated that this hierarchy in the frontal cortex contributes to the predictive coding of proprioceptive sequences through the M1 in one direction , and to that of visual sequences in the other direction via the possible connection between the inferior parietal cortex and the SMA , known as the parieto-frontal circuits [54] , [55] . Additionally , our experiments showed that the behavior generation of the robot in the real environment becomes substantially unstable when the timescale of the higher-level network is set smaller , that is , when the timescale is similar to the values in the middle-level network . Our analysis in such cases revealed that the two functions of generating primitive motor patterns and sequencing them cannot be segregated in the whole network if the chaos dynamics tend to be distributed . Gros [56] , who referred to higher-level as the reservoir and to the middle/low levels as attractor networks , also discussed the generation and stabilization of transient state dynamics , in terms of attractor ruins . The discussion supports one's opinion that the slower timescale part exhibits robustness of influence of external stimuli . Therefore , it is concluded that the hierarchical timescale differences , which are assumed to be in the human frontal cortex and to be responsible for the generation of voluntary actions , are essential for achieving the two functions of freely combining actions in a compositional manner and generating them stably in a physical environment . The uniqueness in the presented model is that deterministic chaos is self-organized in the process of imitating stochastic sequences , provided that sufficient training sequences are utilized to support the generalization in learning ( it was observed that the reduction of length in the training sequences can hinder the self-organization of chaos ) . Therefore , it might be asked why deterministic dynamical system models are considered more crucial than stochastic process models such as the Markov Chain [57] or Langevin equation [58] , [59] . A fundamental reason for focussing on deterministic ( as opposed to stochastic ) dynamics is that they allow for mean field approximations to neuronal dynamics and motor kinetics . This is important because although individual neuronal dynamics may be stochastic , Fokker Planck formulations and related mean field treatments render the dynamics deterministic again; and these deterministic treatments predominate in the theoretical and modelling literature . Furthermore , we speculate that deterministic neuronal dynamic systems are indispensable for generating both spontaneous behaviors and intentional behaviors under the same dynamic mechanism , especially by applying the initial sensitivity characteristics . When the system is initiated from unspecified initial states of the internal units , the resultant itinerant behavioral trajectories exhibit spontaneous transitions of primitive actions by reflecting the statistical structures extracted through the generalization in learning . However , it is also true that a particular sequence of shorter length can be regenerated by resetting with the corresponding initial state values by the deterministic nature of the model . Our robotics experiments showed that the robot can regenerate trained sequences up to several transitions of primitive actions under a noisy physical environment . In addition , our preliminary experiments suggested that longer primitive action sequences can be stably regenerated if those intentional sequences are trained more frequently than other non-intentional ones . This implies that frequently activated fixed sequences can be remembered by cash memories of their initial states . If the Markov chain model is employed for reconstructing the same feature , the model must handle the dualistic representation , namely the probabilistic state transition graph for generating itinerant behaviors and the deterministic linear sequence for generating each intentional behavior . Empirical support for the idea of encoding action sequences by initial states can been found . Tanji and Shima [28] found that some cells in the SMA and the preSMA fire during the preparatory period immediately before generating specific primitive action sequences in the electrophysiological recording of monkeys . The results can be interpreted such that those cell firings during the preparatory period may represent the initial states that determine which primitive action sequences to be generated subsequently . In the future , if this study can be extended to simultaneous observations of the populations of animal cells during spontaneous action sequence generation by employing the recent developments of multiple-electrode recording techniques , our model prediction can be further evaluated .
The evolution of a continuous-time-rate coding model is defined as ( 1 ) where is a time constant , is a weight matrix , is a bias vector , and is the activation function of a unit ( typically the sigmoid function or ) . When this differential equation is put into the form of an approximate difference equation with step size , we obtain ( 2 ) In the present paper , we assume without the loss of generality . As the hierarchical neural network for controlling a humanoid robot , we used a mixture of the recurrent neural network ( RNN ) experts model [45] , in which the gating network is a multiple-timescale RNN [31] . The mixture of RNN experts consists of expert networks together with the gating network . All experts receive the same input and have the same number of output units . The gating network receives the previous gate opening values and the input , and then controls the gate opening . The role of each expert is to compute a specific input-output function , and the role of the gating network is to decide which single expert is the winner on each occasion . In the present paper , the set of expert RNNs is referred to as the lower-level network . The middle- and higher-level networks differentiated by time constants are contained in the gating network . The dynamic states of the lowest-level neural networks ( the mixture of experts ) at time are updated according to ( 3 ) ( 4 ) ( 5 ) where is an input vector representing the current visuo-proprioception , and is an output vector representing the predicted visuo-proprioception . Here , denotes a component-wise application of , is the number of experts , and is the feedback time delay of the controlled robot ( in the experiment ) . In the present paper , denotes the concatenation of vectors and . For each , the terms , , and denote the gate opening value , the internal neural state , and the output state of the expert network , respectively . The gate opening vector represents the winner-take-all competition among experts to determine the output . The gate opening vector is the output of the gating network , defined by ( 6 ) ( 7 ) ( 8 ) ( 9 ) where and denote the internal neural states of the middle-level and higher-level networks , respectively . To satisfy and , Equation ( 9 ) is given by the soft-max function . Using the sigmoid function denoted by , Equation ( 9 ) can be expressed as ( 10 ) This equation indicates that the output of the gating network is given by the sigmoid function with global suppression . Note that the visuo-proprioceptive input enters the state Equations ( 3 ) and ( 6 ) . During imagery , this is replaced by its prediction so that these equations become ( in discrete form ) nonlinear autoregression equations that embody the learned dynamics . During inference and action , optimising the parameters of these equations can be thought of as making them into generalised Bayesian filters such that the predictions become maximum a posterior predictions under the ( formal ) priors on the dynamics specified by the form of these equations . The procedure for training the hierarchical neural network model is organized into the following two phases: The training procedure progresses to phase ( 2 ) after the convergence of phase ( 1 ) . For each phase , the learning involves choosing the best parameter based on the maximum a posteriori estimation . A learning algorithm with a likelihood function and a prior distribution was proposed in [45] . In the following , we describe a learning algorithm corresponding to the above description of the hierarchical neural network model . In the training processes , we used training sequences , including action primitives of moving an object to the left , to the center , and to the right . The time constants for the lower-level and middle-level networks were set to and , respectively . The time constant for the higher level was chosen to be . The number of experts was , and the number of internal units for each expert was . There were internal units for the middle-level network and for the higher-level network , i . e . , the total number of internal units in the gating network was . Each element of the matrices and biases of either an expert or a gating network was initialized by random selection from the uniform distribution on the interval , where is the number of internal units . The initial states for the experts and the gating network were also initialized randomly from the interval . The parameters and were initialized such that for each and , respectively . Since the maximum value of depends on the total length of the training sequences and the dimension of the output units , the learning rate was scaled by a parameter that satisfies . The parameter settings were , , , and . For each training trial , we conducted learning for the experts up to steps and learning for the gating network up to steps . We performed training of the experts for samples having different initial parameters and training data . In addition , for each set of trained experts , we performed training of the gating network while varying the higher-level time constant at , , and . As a result , samples of a mixture of RNN experts were provided for each condition of the higher-level time constant . The behaviors of the robot were described by a -dimensional time series , which consists of proprioception ( an eight-dimensional vector representing the angles of the arm joints ) and vision sense ( a two-dimensional vector representing the object position ) . On the basis of the visuo-proprioception , the neural network generated predictions of the proprioception and vision sense for the next time step . This prediction of the proprioception was sent to the robot in the form of target joint angles , which acted as motor commands for the robot to generate movements and interact with the physical environment . This process , in which values for the motor torque were computed from the desired state , was considered at a computational level to correspond to the inverse model . This inverse computation process was preprogrammed in the robot control system . Changes in the environment , including changes in the object position and changes in the actual positions of the joints , were sent back to the system as sensory feedback . We demonstrate how a trained network learns to generate combinations of primitive actions . Let us consider a sequence of symbols labeled according to the object position , e . g . , “CLRLRCL” , where C , L , and R are the center , left , and right positions , respectively . Let be a block if is a finite sequence of symbols . An -block is simply a block of length . To evaluate the performance in creating novel sequential combinations , we counted the number of -blocks that appeared in the visuo-proprioceptive time series generated by the network . We computed the ratio of the number of -blocks generated by the network to the total number of possible -blocks ( note that the total number of possible -blocks is ) . Figure 7 shows the ratios averaged over sample networks with higher-level time constants of , , and in the motor imagery mode . Note that the teaching sequences do not include all acceptable combinations , because the teaching data are finite . The maximum Lyapunov exponent of a dynamic system is a quantity that characterizes the rate of exponential divergence from the perturbed initial conditions . Consider two points , and , in a state space , each of which generates an orbit in the space by the dynamic system . The maximum Lyapunov exponent can be defined as ( 23 ) where is the initial separation vector of two trajectories , and is the separation vector at time . To evaluate the maximum Lyapunov exponent for each neural network , we computed sample sequences of time steps with a random initial state and an initial separation vector by a numerical simulation . In the numerical simulation , a network received predictions of the visuo-proprioception generated by the network itself as input for the next step . When the maximum Lyapunov exponent of a middle- or higher-level network was measured , we computed the dynamics of the entire network , but evaluated a separation vector containing only the component of a subnetwork as a middle- or higher-level component . This method measures the contribution of the subnetwork to the initial sensitivity of the dynamics . Note that if the subnetwork has a positive Lyapunov exponent , as measured in the abovementioned manner , then the entire network also has a positive Lyapunov exponent . | Various psychological observations have suggested that the spontaneously generated behaviors of humans reflect statistical structures extracted via perceptual learning of everyday practices and experiences while interacting with the world . Although those studies have further suggested that such acquired statistical structures use chunking , which generates a variety of complex actions recognized in compositional manner , the underlying neural mechanism has not been clarified . The current neuro-robotics study presents a model prediction for the mechanism and an evaluation of the model through physically grounded experiments on action imitation learning . The model features learning of a mapping from intentional states to action sequences based on multiple timescales dynamics characteristics . The experimental results suggest that deterministic chaos self-organized in the slower timescale part of the network dynamics is responsible for generating spontaneous transitions among primitive actions by reflecting the extracted statistical structures . The robustness of action generation in a noisy physical environment is preserved . These results agree with other neuroscience evidence of the hierarchical organization in the cortex for voluntary actions . Finally , as presented in a discussion of the results , the deterministic cortical dynamics are presumed crucial in generating not only more intentional fixed action sequences but also less intentional spontaneously transitive action sequences . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"learning",
"cognitive",
"neuroscience",
"psychology",
"social",
"and",
"behavioral",
"sciences",
"mathematics",
"cognitive",
"psychology",
"decision",
"making",
"biology",
"neuroscience",
"nonlinear",
"dynamics"
] | 2011 | A Neurodynamic Account of Spontaneous Behaviour |
Dihydrodipicolinate synthase ( DHDPS ) is an essential enzyme involved in the lysine biosynthesis pathway . DHDPS from E . coli is a homotetramer consisting of a ‘dimer of dimers’ , with the catalytic residues found at the tight-dimer interface . Crystallographic and biophysical evidence suggest that the dimers associate to stabilise the active site configuration , and mutation of a central dimer-dimer interface residue destabilises the tetramer , thus increasing the flexibility and reducing catalytic efficiency and substrate specificity . This has led to the hypothesis that the tetramer evolved to optimise the dynamics within the tight-dimer . In order to gain insights into DHDPS flexibility and its relationship to quaternary structure and function , we performed comparative Molecular Dynamics simulation studies of native tetrameric and dimeric forms of DHDPS from E . coli and also the native dimeric form from methicillin-resistant Staphylococcus aureus ( MRSA ) . These reveal a striking contrast between the dynamics of tetrameric and dimeric forms . Whereas the E . coli DHDPS tetramer is relatively rigid , both the E . coli and MRSA DHDPS dimers display high flexibility , resulting in monomer reorientation within the dimer and increased flexibility at the tight-dimer interface . The mutant E . coli DHDPS dimer exhibits disorder within its active site with deformation of critical catalytic residues and removal of key hydrogen bonds that render it inactive , whereas the similarly flexible MRSA DHDPS dimer maintains its catalytic geometry and is thus fully functional . Our data support the hypothesis that in both bacterial species optimal activity is achieved by fine tuning protein dynamics in different ways: E . coli DHDPS buttresses together two dimers , whereas MRSA dampens the motion using an extended tight-dimer interface .
Dihydrodipicolinate synthase ( DHDPS ) is an essential enzyme involved in the lysine biosynthesis pathway [1] . It is expressed in plants and microorganisms , but not in animals , which makes it a potential target for herbicides and antibiotics [2] . DHDPS from E . coli is a homotetramer consisting of a ‘dimer of dimers’ ( Figure 1A ) . The catalytic residues T44 , Y107 and Y133 are found at the tight-dimer interface ( Figure 1D ) , with each tight-dimer containing two complete active sites within the barrel of the monomeric ( β/α ) 8-fold and an allosteric site within a deep cleft between the subunits that binds two ( S ) -lysine molecules to mediate feedback inhibition [3] . A tyrosine residue ( Y107 ) from one subunit of the tight-dimer protrudes into the active site of the adjacent subunit and forms part of a catalytic triad that is essential for activity [4] , [5] . Although this suggests that the tight-dimer contains the minimum requirements for catalysis , mutation of a central residue in the dimer–dimer interface ( L197 ) produced dimeric variants having severely reduced catalytic function ( Figure 1B ) [6] , [7] . Crystallographic , biophysical and Small Angle X-ray Scattering ( SAXS ) evidence suggest that the dimers associate to stabilise the active site configuration , and removal of this central interface residue destabilises the dimer , thus increasing the flexibility and reducing both catalytic efficiency and substrate specificity . This has led to the hypothesis that the tetramer has evolved to optimise the dynamics within the tight-dimer unit [6] . Interestingly , DHDPS from methicillin resistant Staphylococcus aureus ( MRSA ) occurs naturally as a dimer [8] and contains a significantly more extensive tight-dimer interface compared to DHDPS from other species ( Figure 1C ) . It has been suggested [8] that this serves to restrict flexibility at the interface , and represents an alternate evolutionary solution to optimising dynamics across this interface and thus enzyme activity . Although the crystal structures for DHDPS from over 20 species have been determined to date , and together with biophysical and biochemical data have provided insight into the role of quaternary structure in regulating DHDPS activity , a detailed molecular understanding of the conformational properties of dimeric and tetrameric forms of DHDPS has not yet emerged . While X-ray crystallography is a powerful technique for understanding protein structure at atomic resolution , the final model represents a space and time average of all molecules in the crystal lattice . Therefore information about the flexibility of the molecule is limited and can only be gained from structural comparisons of the molecule in different crystal lattices or the atomic temperature ( B ) factors; although these values must be interpreted with caution . Insights into flexibility and motion can be obtained using the X-ray crystal structure combined with molecular dynamics ( MD ) simulations . This offers the ability to study the time-dependent behaviour of a molecular system , extending the information gained from crystallographic and other data . In this study , we take a unique opportunity to probe the role of quaternary structure in enzyme catalysis using three well-characterised forms of DHDPS . We perform comparative MD simulation studies of native tetrameric , mutant dimeric forms of DHDPS from E . coli and the native dimeric structure from MRSA , with the aim of understanding the importance of quaternary structure to the dynamics and function of this essential enzyme .
To probe the dynamic features of both tetrameric and dimeric forms of E . coli DHDPS , we performed comparative MD simulations of the wild-type E . coli tetramer ( referred to as tet-1 and tet-2; simulated for 0 . 48 µs each ) and E . coli dimer ( dim-A = L197Y mutant dimer; dim-B = dimer taken from the wild-type tetramer; 0 . 5 µs each ) in the absence of substrate . Both tetramer simulations consistently exhibited steady dynamics and reached an RMSD plateau from 80 ns until the end of the simulations with an RMSD = 1 . 5 Å , only slightly deviating from the crystal structure conformation ( Figure 2A , grey lines; Video S1 ) . In comparison the L197Y mutant dimer simulation ( dim-A ) showed a strikingly different behaviour ( Figure 2A , blue; Video S2 ) . While the Cα−RMSD curve remained close to the tetramer simulations for the first 150 ns , it increased to reach a RMSD plateau at ∼3 . 1 Å for the last 200 ns of simulation . Closer examination revealed that the increase in RMSD is largely a result of the 15 degrees relative re-orientation of monomers within the dimer ( Figure 2B ) . RMSDs of Cα atoms within individual monomers in dim-A remained low throughout the simulations ( mean RMSD ∼1 . 5 Å , Figure 2C ) , comparable to the steady RMSDs observed in all monomers simulations of tet-1 and tet-2 ( mean RMSD = 1 . 1 Å ) . This indicates that the monomers experience relatively little structural deviation from their crystal conformation individually in dim-A , but undergo significant rigid-body motion , relative to each other , within the dimer . The angle of rotation of the monomers for the dim-A simulation is represented in Figure 2D ( blue ) . Consistent with the dim-A simulation , dim-B Cα-RMSDs remained close to those from the tetramer for the first 130 ns , then increased to ∼2 . 1 Å for 220 ns to reach a final plateau for the last 100 ns of the simulation at 3 . 3 Å ( Figure 2A , light blue ) , only slightly above the value reached by dim-A and well above the RMSDs of the tetramer simulations . Again , the increase in the RMSDs can be explained by monomer-monomer rotation ( Figure 2D , light blue ) , with the Cα-RMSDs within each monomer remaining low throughout the simulation ( 1 to 1 . 8 Å; Figure 2C ) . Taken together , these simulations indicate that the dimer produced by disrupting the dimer-dimer interface of the native E . coli DHDPS tetramer , either as a result of the L197Y mutation or by artificially splitting the wild-type tetramer in half , loses the stabilising contribution of its adjacent dimer . Similar results have recently been obtained from MD simulations for DHDPS from the plant species , Vitis Vinifera [9] , which forms a ‘back-to-back’ dimer of dimers compared to the head-to-head arrangement of E . coli DHDPS ( Figure 1A ) . Despite the different quaternary architecture , the loss of dimer-dimer packing in the plant or bacterial tetramers also results in monomers moving more freely within the dimer . Further , SAXS studies of the E . coli mutant dimer [6] used in this work have suggested rigid-body motion of the monomers within the dimer and are thus consistent with our observations . As this motion revolves around the tight-dimer interface that also comprises some of the important active site residues , we next focused on comparing the nature and extent of active site flexibility in E . coli DHDPS tetramers and dimers . To estimate the extent of the active site deformation we calculated the RMSD values ( heavy-atoms only ) over all the simulations for the eight active residues ( T44 , Y106 , Y133 , R138 , K161 , G186 , I203 , and Y107 contributed by the adjacent monomer; Figure 1D ) . Active site residues in the tetramer simulations fluctuate within an RMSD range of 0 . 8–1 . 8 Å , with a mean of 1 . 0 Å , and are relatively stable in their conformation throughout the last 400 ns of the simulations ( Figure 3A , grey lines; Figure 4A; Video S3 ) . Conversely , the positions of active site residues in the dimer deviate from the crystal conformation to a much larger degree , with RMSD values varying from an initial 1 . 0 Å up to 2 . 8 Å ( dim-A ) and 3 . 5 Å ( dim-B ) towards the end of the simulations ( Figure 3A , blue lines; Figure 4B; Video S4 ) . Even though the residues in the dim-A and dim-B active sites show differences in their conformations , they both consistently deviate from the wild-type positions with RMSD values greater than 2 Å over the last 150 ns of the simulations . Our simulations demonstrate that the active sites show more deformation in dimers than in tetramers , where residues show relatively small deviations from their crystal conformation ( Figure 4A , B ) . To estimate potential flexibility in the 8 amino acids composing the active site we calculated the root mean square fluctuations ( RMSFs ) for the tetramer and dimer simulations ( Figure 3B ) . The results clearly show a general flexibility increase in the dimer active site compared to the tetramer . While the tetramer active site residues display individually low flexibility ( RMSF range = 0 . 4–0 . 9 Å; Figure 3B and 4A; Video S3 ) , dimer active site residues appear considerably more flexible ( RMSF range = 0 . 6–2 . 4 Å; Figures 3B and 4B; Video S4 ) . Interestingly , the catalytic residues T44 and Y107 as well as Y106 and R138 contribute most to the increased flexibility within the dimer active site . The remaining residues ( Y133 , K161 , G186 , I203 ) are also more flexible in the dimer compared to the tetramer , although they fluctuate somewhat less ( RMSF values<1 . 0 Å ) . The increase in T44 RMSF is due to flipping of its side chain , inverting the positions of the methyl and hydroxyl groups , and results in the transient loss of a hydrogen bond with the hydroxyl group of Y107 ( Figure 4A , B ) . This interaction is known to be essential for activity of the enzyme as it forms part of the catalytic triad [4] , [10] . The fluctuations of the hydrophobic patch formed by Y106 and Y107 ( both embedded in the tight-dimer interface ) contribute the most to the increase in RMSF . The catalytic residue Y107 is of particular interest , since this residue exhibits backbone Φφ dihedral angles lying in a “disallowed” region of the Ramachandran plot in E . coli DHDPS ( wild type and mutants ) , as well as in other organisms [4] , [11]–[13] corresponding to a γ-turn backbone geometry . This suggests that conformational strain is maintained in its backbone , possibly due in part to the backbone carbonyl oxygen bond formed with the guanidino group of R138 [14] . Ramachandran plots for Y107 over the course of the E . coli simulations are shown in Figure 3C ( tet-1 and tet-2 ) and Figure 3D ( dim-A and dim-B ) . Fluctuations in the simulations allow the backbone of Y107 in both tetramers and dimers to explore the Lα geometry; dimers however adopt this geometry for more than half the simulation time . A clear distinctive feature of the dimer simulations is the ability of the Y107 backbone to adopt one “favoured” region ( the α region ) of the Ramachandran plot that is not populated in the tetramer simulations . This is associated with the loss of the hydrogen bond formed with the R138 guanidino group , resulting in increased movements of the arginine side-chain ( Figure 3B , 4B and Video S4 ) . Taken together , these observations provide an explanation for the RMSF increase for this residue , and most likely induce the strain in the backbone of Y107 . This is in stark contrast to the tetramer simulations , where the backbone angles of Y107 explore the favoured Lα region of the Ramachandran plot for only 10 . 9% of the time ( Figure 3C ) . Recently Pearce et al . ( 2011 ) [15] have engineered and characterized a monomeric form of DHDPS from the bacterium T . maritama with impaired catalytic function compared to the tetrameric form . The 2 . 0 Å X-ray structure revealed a well-preserved overall fold and active site geometry compared to its tetrameric form , with the exception of the residues equivalent in E . coli to R138 and Y107 and its surrounding loop [15] . Additionally we find that our dimer simulations reproduce to some extent the backbone conformation of the latter loop of this unique monomeric form , with all Φφ angles falling in a favoured region of the Ramachandran plot . The side chains of Y106 and Y107 are also subject to large fluctuations in the dimer simulations . The well-packed hydrophobic stacking formed by the aromatic groups of Y106 and Y107 of both monomers ( four tyrosines in total ) at the tight-dimer interface in the crystal structures undergoes a dramatic rearrangement resulting in the loss of aromatic stacking in the last 200 ns of simulation . Whereas in the tetramer simulations the Y106 side chain oscillates between conformations that are relatively close to the original crystal structure ( Figure 4A and Video S3 ) , the Y107 side chain exhibits largely different conformations towards the end of the dim-A and dim-B simulations ( Figure 4B and Supporting Video S4 ) . The latter movements are associated with positional changes of the Y107 hydroxyl group 15 Å away from the two other residues of the catalytic triad ( T44 , Y133 ) , incompatible with catalysis . We therefore observe in the dimer simulations a critical disruption of the catalytic triad network of hydrogen bonds with the large conformational change of a key residue . As a result , the overall geometry of the catalytic motif is dramatically altered . In two independent MD simulations , totalling nearly 1 µs , the dynamics of the wild-type E . coli tetramer in the absence of substrate are characterised by ‘near crystal structure’ fluctuations ( Figure 3A , B; Figure 4A and Video S3 ) . The overall conformations of the individual monomers , their supra-molecular assembly and the active site only slightly deviate from the structure observed by X-ray crystallography . The dimer simulations show a radically different behaviour: alterations of the monomer arrangement and most importantly critical deformations of the catalytic triad , in particular Y107 , potentially rendering the enzyme inactive ( Figure 3A , B; Figure 4B and Video S4 ) . If “crystal-like” rigidity is a requirement for a functional enzyme at wild-type levels as shown by the tetramer simulations , the amount of plasticity observed in the isolated dimer , triggered by the change in quaternary structure , provides a straightforward explanation for the dramatic decrease in activity measured [6] . Our simulation data for E . coli DHDPS suggest that conformational fluctuations and flexibility at the active site is a primary cause of the dramatic decrease in enzymatic activity of dimers . The existence of a naturally occurring dimer from the bacterial pathogen MRSA that exhibits comparable activity to the E . coli tetramer is therefore intriguing [8] . Whereas the overall tertiary structures of MRSA and E . coli DHDPS are highly similar ( RMSD = 0 . 9 Å; Figure 1B , C ) , with only minor reorientations of active site side-chains ( Figure 1D ) , the nature of their tight-dimer interfaces differs ( Figure 5 ) . MRSA DHDPS possesses a relatively high number of hydrogen bonds at the tight-dimer interface and two electrostatic interactions that are absent in the E . coli structure , suggesting that it is perhaps less flexible that its E . coli counterpart [8] . We therefore performed two MD simulations of the MRSA DHDPS dimer in the absence of substrate and compared the results to the E . coli DHDPS simulations . The 1 . 45 Å resolution crystal structure of MRSA DHDPS [8] was used as the starting structure for two independent MD simulations of 0 . 5 µs each in length ( denoted mrsa-1 and mrsa-2 ) . Both simulations show a gradual increase in RMSD , which stabilise and reach a plateau at ∼3 Å at ∼300 ns ( Figure 6A ) . The latter corresponds to a rotation of one monomer with respect to the other ( Video S6 ) , similar to the E . coli DHDPS dimer ( Figure 2B ) . Active site residues deviate moderately from their crystal conformation over the course of the simulations ( Figure 6B and Video S6 ) , reaching a plateau for the last 200 ns , yet somewhat less deviant than the corresponding residues in the E . coli DHDPS dimer ( RMSD values of 1 . 6–3 . 0 Å compared to 2 . 2–3 . 5 Å; Figure 6B ) . RMSF values of the active site residues ( Figure 4B ) are higher than the E . coli tetramer simulations and mostly comparable ( within standard deviation ) to the E . coli dimer simulations , except for the relatively immobile Y109 ( equivalent to Y107 in the E . coli structure ) . In the mrsa-1/2 simulations the backbone dihedral angles of Y109 populate the same regions as in the dim-A/B simulations ( Figure 6C ) . The simulation time spent in the Φφ region is similar to dim-A/B , but the proportions are reversed for the γ-turn and Lα regions , consistent with this residue remaining close to the crystal geometry for more than half of the simulation . Furthermore , the extent of the Y109 side chain dynamics is reduced , in contrast to the dim-A/B simulations , and fluctuates near the crystallographic conformation . In addition the aromatic stacking formed with Y108 ( equivalent to Y106 in the E . coli structure ) as part of the dimer interface remains intact . To gain more insight into the potential changes occurring in the active sites we focused on the conserved network of hydrogen bonds present in the catalytic site ( Figure 7A ) . This network is formed by 2 hydrogen bonds between the hydroxyl groups of T44 and Y133 ( E . coli numbering ) , and between the hydroxyl groups of T44 and Y107 . Point mutation of any of these 3 residues that constitute the catalytic triad results in severely reduced activity [3] . Distances between donor and acceptor atoms were monitored throughout simulations ( Figure 7 ) . We find that atoms T44-Oγ/Y133-Oç ( Figure 7B ) remain in reasonably close contact at a similar average distance of 5 . 4±1 . 3 Å and 5 . 6±1 . 3 Å in the E . coli and the MRSA dimers respectively . The hydrogen bond is only transiently formed regardless of the species and broken upon flipping of the T44 side chain . In contrast the distance between T44-Oγ/Y107-Oç shows a marked difference ( Figure 7C ) following the repositioning of Y107 in the E . coli dimer associated with monomer re-arrangement and shown here with a large increase . The relative positions of both side chains seem affected to a smaller extent by rotation in the MRSA dimer ( average distance is 5 . 7±1 . 3 Å ) with a small distance increase suggesting weak electrostatic interaction between the hydroxyl groups . Finally the hydroxyl and ammonium groups of residues Y133 and K161 respectively ( Figure 7A , 7D ) were monitored . They form an electrostatic interaction in the crystal conformations with a distance of 2 . 9 ( E . coli tetramer ) , 3 . 4 ( MRSA ) and 3 . 7 Å ( L197Y E . coli ) . Point mutation of substrate binding K161 has been shown to result in largely impaired activity [16] . We find no discernible difference between the dimers with average distances of 4 . 2±1 . 0 ( E . coli ) and 4 . 5±1 . 1 Å ( MRSA ) . Additionally , in the E . coli tetramer simulations all distances were found comparatively shorter and compatible with a tighter and more rigid active site: 4 . 7±0 . 9 Å ( T44/Y133 ) , 5 . 1±1 . 2 Å ( T44/Y107 ) and 3 . 6±0 . 6 Å ( Y133/K161 ) . We conclude that except for the position of the E . coli dimer Y107 the overall active sites architecture and the relative positions of essential side chains remain close ( E . coli tetramer ) or reasonably close ( MRSA , E . coli dimer ) to the crystalline state , and are only to a minor extent affected by monomer re-arrangement . Although the functional MRSA DHDPS dimer displays monomer-monomer rotation as well as active site flexibility , unlike the E . coli dimer it does not undergo a similar active site deformation focused around Y109 . In contrast , its fluctuations are more distributed amongst the active site residues . Whereas the E . coli DHDPS dimer interface consists of seven hydrogen bonds and three hydrophobic contacts , the larger MRSA DHDPS dimer interface consists of 17 hydrogen bonds and two salt-bridges [8] . We therefore compared and contrasted the nature of the tight-dimer interfaces for E . coli . and MRSA enzymes . The size of the interfacial area in the E . coli tetramer is stable throughout the simulations . We find that in the MRSA dimer the rotation of the monomers is associated with a reduction in the buried interfacial area , similar in size ( ∼2700 Å2 for two monomers , Figure 8A ) to the initial E . coli interface . This does not lead to a decrease in the number of hydrogen bonds ( Figure 8B ) or salt-bridges , which remains constant . We find however that in the mutant E . coli dimer , while the interfacial buried area is constant , the number of hydrogen bonds contributing to the tight-dimer interface increases with re-orientation of the monomers . In addition we observed the formation of a new salt-bridge per monomer between residues R109 and E246 in dim-A and dim-B , permitted by the new orientation of the monomers . In mrsa-1 and mrsa-2 the equivalent salt-bridge is formed at positions K111 and D247 . This suggests that this re-organization of the monomers is more stable than the arrangement found in the crystal state but only compatible with loss of the quaternary structure . Dimer binding energies calculated by the MM-PBSA approach lend support to this hypothesis ( Text S1 ) . Disruption of the supra-molecular assembly is associated in E . coli DHDPS with dramatic conformational changes in the active site . Our simulations show that the MRSA DHDPS enzyme , in the absence of substrate , experiences relatively high flexibility . This is perhaps not unexpected for an enzyme that exists in a monomer-dimer equilibrium in solution [8] . In addition , in contrast to the E . coli dimer , it does not exhibit a localised deformation . We propose that the flexibility observed , without conformational change of critical interface residues such as Y109 , preserves the active site geometry and hence enzyme activity . The mutant dimer L197Y was crystallized in the absence of the substrate pyruvate , with a molecule of α-ketoglutarate trapped in its active site [6] . The latter was not added in the crystallization conditions but rather captured from the expression system . The repositioning of Y107 side chain observed in the L197Y E . coli DHDPS dimer is associated with an enlargement of the active site pocket ( Figure 9A and 9B ) . We propose that the widening of the pocket in the mutant dimer is responsible for allowing the substrate analogue α-ketoglutarate , which is larger than the natural substrate pyruvate , to bind K161 and form a Schiff base before cyclisation , as observed in the crystalline state [6] . This newly formed covalent species acts as a stable inhibitory adducts towards pyruvate , thus explaining the loss of specificity and affinity measured [6] . Following this hypothesis originally formulated by Griffin et al . ( 2008 ) [6] , in MRSA DHDPS the relatively stable positions of all active site residues would prohibit binding and perhaps entry of α-ketoglutarate in the active site . This is reflected by similar affinity for pyruvate and enzymatic activity in both MRSA and wild-type E . coli DHDPS [8] . Our simulations provide atomistic details of the role of high-level molecular assembly in maintaining optimal activity in the E . coli enzyme . In the mutant E . coli dimer we have identified monomer reorientation within the dimer as a major influence on activity , consistent with SAXS data [6] . With the buttressing provided by formation of the dimer of dimers active site geometry is preserved in the tetramer , while in the dimer the enzyme is stripped of a productive catalytic arrangement . Further , simulations of the E . coli mutant dimer reveal a large conformational change of Y107 , a key catalytic residue . The wild-type MRSA dimer enzyme is also subject to relatively high flexibility , but in contrast , is counter-balanced by an extended tight-dimer interface , which results in a reasonably well-preserved active site . Our results suggest that in these two different pathogenic bacterial species , DHDPS optimal activity is achieved by opposing the excess inherent dimer flexibility with two different strategies: in E . coli a higher level quaternary structure buttresses two dimers together while in MRSA an enhanced tight-dimer interface allows preservation of activity . In conclusion , this work supports the hypothesis that a driving force of DHDPS evolution is to optimize intrinsic protein fluctuations to a level compatible with its activity and function [6] , [8] , [9] , [15] . This work also adds to a growing body of evidence linking quaternary structure , protein dynamics and function [17] , [18]
The 1 . 9 Å resolution X-ray structure of the wild-type E . coli DHDPS tetramer [5] ( PDB ID 1YXC ) was used for the two independent MD simulations of tetramers ( termed tet-1 and tet-2 ) . The dimer simulations employed two different starting structures . In the first case the single mutant enzyme , DHDPS-L197Y , which was solved to 1 . 7 Å resolution [6] ( PDB ID 2OJP ) , was used ( termed dim-A ) . The coordinates of the bound tetrahedral adduct of its substrate analogue were discarded . Since this may adversely affect the simulation , the second simulation used the dimer structure contained in the asymmetric unit of the native tetramer structure ( termed dim-B ) . Finally , the 1 . 45 Å resolution crystal structure of DHDPS from MRSA [8] for two independent simulations ( mrsa-1 , mrsa-2 ) . In total , we performed 6 independent MD simulations of 3 different DHDPS molecules: two simulations of the native E . coli tetramer ( tet-1 and tet-2 ) , two simulations of an E . coli dimer ( dim-A and dim-B ) and two simulations of the native MRSA dimer ( mrsa-1 and mrsa-2 ) . In all simulations , typically 2 to 4 ns were discarded prior to analysis . All simulations employed the same protocol . Structural analysis and measurements were done with the VMD software [19] , figures and videos with VMD and PyMol [23] . Cavities were detected with MDpocket [24]; the cavities presented in Figure 9 are the grid points with frequency isovalue 0 . 3 . Ramachandran plots were produced following Lovell et al . [25] . Monomers Cα-RMSDs were calculated with the corresponding minimized crystal structure as a reference . Active sites RMSDs were calculated employing non-hydrogen atoms of the eight residues composing the active site ( see text ) with the minimized crystal structure as a reference . Active sites residues RMSDs employing the whole monomer as the reference structure displayed an identical trend . Active sites RMSF calculations employed non-hydrogen atoms of the active site as a reference , after removal of the rotation-translation motions by aligning on the first snapshot of the corresponding trajectory . Removal of rotation-translation motions by aligning on the whole monomer yielded an identical trend . | Enzyme function requires the specific placement of residues in the active site so that the correct chemistry is available for efficient catalysis . However , the inherent flexibility of proteins can present challenges in fulfilling these stringent requirements . We have investigated the role of flexibility in the enzyme Dihydrodipicolinate synthase ( DHDPS ) , which in E . coli is a homotetramer consisting of a ‘dimer of dimers’ , with the catalytic residues found at the tight-dimer interface . It is hypothesized that the tetramer arrangement has evolved to restrict the flexibility at the active site by buttressing together a pair of dimers . In contrast , DHDPS from methicillin resistant Staphylococcus aureus ( MRSA ) occurs naturally as a dimer yet retains full activity . Using molecular dynamics simulations we have investigated the flexibility of dimeric and tetrameric forms of the E . coli and MRSA enzymes , and reveal that optimal activity is achieved by minimizing the inherent dimer flexibility using two different strategies – by either buttressing two dimers together in the case of the E . coli tetrameric enzyme or strengthening and extending the dimer interface in the dimeric MRSA . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Methods"
] | [
"biochemistry",
"enzyme",
"structure",
"biochemistry",
"simulations",
"proteins",
"biophysic",
"al",
"simulations",
"enzymes",
"enzyme",
"regulation",
"biology",
"computational",
"biology"
] | 2012 | Structural and Dynamic Requirements for Optimal Activity of the Essential Bacterial Enzyme Dihydrodipicolinate Synthase |
Lactobacillus rhamnosus is a lactic acid bacterium that is found in a large variety of ecological habitats , including artisanal and industrial dairy products , the oral cavity , intestinal tract or vagina . To gain insights into the genetic complexity and ecological versatility of the species L . rhamnosus , we examined the genomes and phenotypes of 100 L . rhamnosus strains isolated from diverse sources . The genomes of 100 L . rhamnosus strains were mapped onto the L . rhamnosus GG reference genome . These strains were phenotypically characterized for a wide range of metabolic , antagonistic , signalling and functional properties . Phylogenomic analysis showed multiple groupings of the species that could partly be associated with their ecological niches . We identified 17 highly variable regions that encode functions related to lifestyle , i . e . carbohydrate transport and metabolism , production of mucus-binding pili , bile salt resistance , prophages and CRISPR adaptive immunity . Integration of the phenotypic and genomic data revealed that some L . rhamnosus strains possibly resided in multiple niches , illustrating the dynamics of bacterial habitats . The present study showed two distinctive geno-phenotypes in the L . rhamnosus species . The geno-phenotype A suggests an adaptation to stable nutrient-rich niches , i . e . milk-derivative products , reflected by the alteration or loss of biological functions associated with antimicrobial activity spectrum , stress resistance , adaptability and fitness to a distinctive range of habitats . In contrast , the geno-phenotype B displays adequate traits to a variable environment , such as the intestinal tract , in terms of nutrient resources , bacterial population density and host effects .
The current development and application of high-throughput sequencing technologies allow to intensively investigate complex microbial ecosystems , such as the human gastro-intestinal ( GI ) microbiota , consisting of over 3 million genes from mainly Gram-positive bacteria [1]–[4] . This and other metagenomic approaches obviate the necessity to culture bacterial isolates to comprehend the richness and the diversity of such ecosystem . However , detailed analysis at the strain level still requires isolation and growth of bacterial residents . Gram-positive lactobacilli are naturally found among ∼1000 phylotypes identified in the human intestinal tract [2] , but only a fraction is represented in the present metagenomic sequences that derive from faecal samples . Lactobacilli mainly reside in the intestinal mucosa and were detected in the ileum metagenome [5] , [6] . The limitations of metagenomic approach , i . e . sequencing depth , are well described and can be obviated by 16S rRNA sequencing and phylogenetic microarrays [7] , [8] . Thus , Heilig et al . provided clear sequence-based evidence for the presence of L . rhamnosus related species in the human intestinal tract [9] . As a consequence of their interactions and ecological role in the human intestinal tract [10]–[13] , lactobacilli are increasingly used in food production , food preservation and nutritional complement formulation [14]–[18] . One of the most used and documented lactobacilli marketed as a probiotic is Lactobacillus rhamnosus GG , which has been isolated from the human intestine and characterized extensively [19]–[21] . L . rhamnosus contains a 3 . 0-Mbp genome , among the largest of the lactic acid bacteria , and has the ability to persist in the human intestinal mucosa , as it produces pili that are decorated with the mucus-binding protein SpaC [22]–[26] . This significantly impacts the intestinal microbiota , via the displacement of pathogenic bacteria [27] , modulation of epithelial barrier functions [28] and potential stimulation of the host immune system via bacteria-host surface molecule crosstalk [16] , [29]–[31] . Since the interaction between host and bacteria has a pivotal role in the impact on the host , much research efforts are presently focused on characterizing the different interaction mechanisms , including the metabolic properties and host-signalling components of L . rhamnosus [30] . However , no studies have actually addressed the genomic diversity of the species L . rhamnosus , in spite of its extensive use in a variety of food products . While some Lactobacillus species have been found in only one dedicated niche , such as the milk-adapted L . helveticus [32] , other lactobacilli such as L . rhamnosus , L . casei or L . plantarum have the capacity to colonize multiple habitats [15] , [33]–[35] . More specifically , L . rhamnosus has been isolated from a large variety of ecological niches , e . g . human intestinal tract , vaginal cavity , oral cavity and cheese , exemplifying its remarkable ecological adaptability as a generalist [19] , [36]–[39] . Genome sequence analysis of a number of lactobacilli revealed that their adaptation to diverse ecological niches is promoted by the acquisition of new genes by horizontal gene transfer and the decay or loss of non-essential genes [33] , [35] , [40] , [41] . The domestication of some lactobacilli to the dairy environment is a typical example of a niche specialization , where milk-adapted strains have unusually high number of pseudogenes , reflected by the loss of metabolic pathways and transport systems that are non-essential in dairy niches rich in nutrients [40] , [42] . In contrast , bacteria from the intestinal tract , a very dynamic habitat in terms of nutrient availability and bacterial population density , have broad metabolic capacities and lifestyle traits essential for survival , persistence and colonization in this niche , e . g . bile resistance [25] , [43] , anti-microbial activity [44] , and mucus-binding pili expression [19] . In some cases , gene sets could even be specifically linked to a particular ecological niche , i . e . intestine vs . dairy environment , as reported for the related L . acidophilus and L . helveticus [40] . In L . reuteri , Frese and colleagues also demonstrated a host specialization between L . reuteri strains isolated from different vertebrates [45] . The present study of the species L . rhamnosus aimed at: ( a ) investigating the genomic diversity of the species and , ( b ) examining variable chromosomal regions associated with phenotypic and/or lifestyle traits found in L . rhamnosus isolates . Four complete L . rhamnosus genomes have been fully sequenced and assembled allowing us to have a glance at the diversity within the species [19] , [46] , [47] . In an effort to further comprehend the diversity and versatility of L . rhamnosus species , we sequenced and compared the genomes of 100 Lactobacillus rhamnosus strains that were isolated from different ecological niches and analyzed their phenotypes . This study represents the first large-scale genomic and functional analysis of L . rhamnosus , providing new insight in the genetics and lifestyle of this species that has a long history associated with human lifestyle and health .
To comprehensively depict the phenotypic and genomic diversity of the L . rhamnosus species , 100 L . rhamnosus strains were isolated from a broad spectrum of ecological niches , e . g . 77 strains of various sites of the human body ( oral cavity , vaginal cavity , blood and intestinal tract ) and 23 strains of dairy origins , including artisanal cheeses and products marketed as probiotics ( Table S1 ) . The genomes of all strains were sequenced using the SOLiD sequencing technology and reads were mapped onto the L . rhamnosus GG chromosome [48] . This allowed detailed comparative genomic analysis and data mining as described in the Materials and Methods section . The number of shared genes between the 100 L . rhamnosus isolates and L . rhamnosus GG ranged from 2622/3016 ( 86 . 9% ) to 3016/3016 ( 100% ) genes with a median number of 2918/3016 ( 96 . 7% ) genes . In terms of relative gene content , the dairy isolates significantly showed the most diversity with L . rhamnosus GG ( average of 92 . 4% ) than the human isolates ( average of 96 . 04% , excluding clinical isolates ) , indicating that the dairy isolates are genetically most distant ( p<0 . 001 between the two groups ) . It is noteworthy that 11 strains of human origin , 3 strains isolated from products marketed as probiotics , and only 1 strain isolated from artisanal cheese shared the complete set of 3016 genes present in L . rhamnosus GG . However , it has to be kept in mind that orthologous genes present in these isolates may carry mutations , i . e . single nucleotide polymorphisms , insertion and deletions that were not addressed in detail in this study . Therefore , the presence of a gene may not necessarily reflect its functionality , as observed within these 11 human strains , which showed significant phenotypic variations , i . e . sugar metabolism , indicating that these strains are not L . rhamnosus GG ( see below ) . Moreover , strain-specific genes are likely to be present in these isolates , conferring additional phenotypic traits not present in L . rhamnosus GG . Based on comparative gene content , the hierarchical clustering of the L . rhamnosus species resulted in four distinct clusters ( Figure 1 ) . Remarkably , most dairy strains were found to belong to the cluster 1 and show marked differences with other clusters . In contrast , intestinal isolates , including L . rhamnosus strains marketed as probiotics shared similarities with other human isolates ( Figures 1 and 2 ) . This is in line with the hypothesis that the genomes of probiotic-marketed strains still reflect their adaptation to their original isolation source , i . e . the human intestinal tract [19] . The distribution of the clinical isolates all across the clustering rather reflects their original ecological niche than their isolation source , since infections are extremely rare events and evolutionary dead ends . The clusters 3 and 4 consist predominantly of L . rhamnosus strains closely related to L . rhamnosus GG ( Figure 1 ) . In Figure 2 , comparison of hierarchical clustering and phylogenetic tree shows some degree of conservation in the grouping of the strains . The phylogenetic tree reflects slow evolution within the genome , i . e . point mutations , whereas the genomic tree ( or hierarchical clustering ) describes major genetic re-arrangement events , i . e . insertions or deletions . Hierarchical clustering therefore shows more recent chromosomal changes , where recombination events contribute to the diversity of the species . Similar differences have been observed in other species , such as L . casei [41] . Based on the 100 mapped genomes , we defined a set of all orthologous genes that are shared by all L . rhamnosus strains . We observed that the shared gene set ( core ) of the L . rhamnosus species consists of 2419 genes , which represents 80 . 2% of L . rhamnosus GG genome . The larger the set of strains used , the smaller the core genome becomes , a trend observed in other genomes as well , such as the core-genome of Streptococcus agalactiae and other bacterial species [49] , [50] . However , the size of the core genome remained stable above ∼20 genomes ( data not shown ) . The full comparative genomic results are shown in Tables S2 and S3 . Although the characterization of L . rhamnosus pan-genome would bring further insights into the species , we did not address it in the present study , as this would require complementary sequencing techniques . Further deep and full-coverage sequence analysis of a selected subset of heterogeneous L . rhamnosus strains is now on-going to report the pan-genome of the species ( data not shown ) . The initial read mapping to the reference genome L . rhamnosus strain GG clearly give a GG-centric view of the genome diversity within the species . However , the additional read mapping to the dairy strain LC705 of a selected set of L . rhamnosus strains revealed a similar clustering as in Figures 1 and 2 ( data not shown ) . This suggests that the use of one strain or another as a reference does not impact on the hierarchical clustering of the isolates and also supports the validity of the experimental design approach chosen in the present study . The distribution of Clusters of Orthologous Groups of proteins ( COG ) was determined for L . rhamnosus GG genome , the L . rhamnosus core-genome and the non-core gene set ( Figure S1 ) . Although no major differences in the relative COG distribution between the different subsets were found , it is noteworthy that 87 L . rhamnosus GG genes ( 30 . 2% ) out of 288 genes assigned to the COG ‘Carbohydrate transport and metabolism’ are not in the estimated core genome and are predicted to encode mostly phosphotransferase system ( PTS ) and other sugar transport systems , possibly essential for the persistence in the intestinal tract . These genes were located in highly variable regions of the L . rhamnosus genome , reflecting the metabolic diversity of this species ( Figure 3 ) . The 17 most variable chromosomal regions include all genomic islands ( GIs ) , typically rich in transposases and other mobile genetic elements ( Figure 1 and Table 1 ) . In L . rhamnosus GG , 5 GIs had previously been identified [19] . The presence of these genomic islands greatly varies among strains of the species L . rhamnosus , as observed previously for the strains LC705 and GG [19] . This suggests that horizontal gene transfer events have contributed significantly to the diversity of the L . rhamnosus species . The GIs identified here were associated with specific biological functions , including interaction and signalling with the host , optimal use of available nutrients and protection against autochthonous phages and mobile genetic elements . Hence they may be considered as lifestyle islands , as their predicted function may specifically contribute to the persistence and colonization in intestinal and other habitats . Other variable regions consisted mostly of transposases and conserved proteins with no clear function and were not further addressed ( Figure S2 ) . Comparative genomic analysis of the 100 strains revealed the loss of genes encoding various carbohydrate PTS system and metabolism-associated proteins compared to L . rhamnosus GG . To study the impact of these genomic characteristics , the metabolic capability to utilize different carbon sources was investigated . Carbohydrate utilization profiling showed that most L . rhamnosus strains use a large range of simple and complex carbohydrates ( Figure 3 ) . However , some differences may reflect their genomic diversity and also at some extent how they evolved in different ecological niches , by the acquisition or the loss of metabolic-associated genes . The ability to utilize carbohydrates mostly relies on the presence of functional transporter machinery and intact metabolic pathways . The clustering of L . rhamnosus strains ( Figure 3 ) revealed strong associations between genome diversity , carbohydrate metabolism and their origins . Typically , strains belonging to cluster 4 utilize D-arabinose , dulcitol and L-fucose , whereas other strains lost these functions but possess the ability to ferment L-sorbose , D-maltose , D-lactose , D-turanose , methyl-α-D-glucopyranoside , L-rhamnose and D-saccharose ( Figure 3 ) . Hence , we detail the differences in carbohydrate utilization within the L . rhamnosus species below . The genome of L . rhamnosus GG harbors a tagatose-6-phosphate pathway ( lacABCD ) and a lactose PTS ( lacFEG ) but the antiterminator lacT and the phospho-β-galactosidase encoding lacG genes are altered and non-functional , preventing GG from metabolizing D-lactose [19] . Strains belonging to the cluster 4 also show a poor ability or incapacity to use D-lactose , whereas other isolates , including most dairy ones utilize this disaccharide , which is found in milk and milk-derived products . We propose that the lacT and lacG genes have been kept intact in these strains , as lactose represents an important carbon source and provides a real benefit for L . rhamnosus strains residing in dairy niches . The maltose locus was predicted to be non-functional in L . rhamnosus GG due to the insertion of a conserved gene ( LGG_00950 ) between genes encoding the maltose-specific malEFGK transporter and the hydrolase ( LGG_0954-LGG_0951 and LGG_00949 , respectively ) [19] . Similarly , we found that most L . rhamnosus strains unable to use maltose also contained a maltose locus disrupted by LGG_00950 . In contrast , the majority of strains belonging to other sublineage contained an intact maltose locus and were able to utilize maltose , indicating that the insertional inactivation by LGG_00950 may have played a significant role in L . rhamnosus species ecology . Comparative genome sequencing of L . rhamnosus GG also showed that the rhamnose locus is altered: the galactitol-specific gatABCD PTS and a DeoR transcriptional regulator are missing while the rhaB gene is duplicated , possibly explaining the inability to use rhamnose compared to some other L . rhamnosus strains , such as LC705 [19] . Combination of the genomic and metabolic data indicates that most strains of the cluster 4 similarly contain a defective rhamnose locus . It is noteworthy that 74% of all isolates can partially or fully utilized L-rhamnose , a carbohydrate from which the species name derives . In contrast , fucosylated compounds such as human mucin and other glycoproteins play an important role in the human gut ecology , as a carbon source for intestinal bacterial species [30] . Close inspection of the L-fucose metabolism revealed that a large number of dairy-associated strains are unable to use L-fucose due to the lack of one or multiple genes required to transport and to metabolize L-fucose: the fucU and fucI isomerases , fcsR fucose operon repressor and α-L-fucosidase ( LGG_02652 ) . Most strains closely related to L . rhamnosus GG retained the capacity to use L-fucose , whereas dairy strains lost this ability , since L-fucose is not as abundant in bovine milk . Dulcitol , a polyol also known as galactitol , is used by the cluster 4 ( Figure 3 ) . In some strains unable to use dulcitol , the function loss was associated with the lack of an intact gatABC PTS system . Other carbohydrates such as turanose and sorbose were not metabolized by strains related to GG ( Figure 3 ) . In L . rhamnosus LC705 , an intact sorbose sorABCDEFGR locus is present , explaining its ability to utilize sorbose , whereas L . rhamnosus GG lacks such machinery [19] . L . rhamnosus strains with similar capabilities may therefore possess an intact sorbose locus . Remarkably , the strains from the cluster 1 present a similar metabolic profile as the industrial dairy strain L . rhamnosus LC705 [19] . This suggests that dairy-related strains characterized in the present study underwent similar niche adaptation as LC705 in terms of acquisition , decay or loss of genes . CRISPR ( clustered regularly interspaced short palindromic repeats ) loci are present in a large number of prokaryote genomes [51] , playing an important role in controlling horizontal gene transfer . It has been well established that some bacteria acquired the CRISPR-Cas system as a protection/immunization system against plasmid conjugation and phage predation [52]–[55] . The CRISPR-Cas system usually consists of a leader sequence , an array of CRISPRs interspaced by spacers and a cas gene cluster encoding the Cas protein complex ( Figure 4A ) [56] . The role and mechanistic of the CRISPR-Cas system in bacterial species have been extensively studied and indicate that the spacer sequences can be considered as a signature of past exposure to exogenous DNA [57] . L . rhamnosus GG has a single Type II-A CRISPR-Cas locus , consisting of 4 cas genes and one CRISPR array containing 24 spacers [19] . To determine whether the CRISPR sequences could be used as an indicator of a specific niche , we determined their diversity and the presence of the cas genes . CRISPR genotyping has been previously developed for epidemiological purposes and strain differentiation for Mycobacterium tuberculosis [58] , enterohemorrhagic Escherichia coli [59] and Salmonella enterica [60] . We were able to generate a CRISPR profile ( based on spacer oligotyping ) for each strain and this revealed a high degree of diversity among the various strains ( Figure 4B , C ) . Remarkably , all strains from cluster 4 were sharing a comparable CRISPR spacer set , whereas the genetically more distant L . rhamnosus strains were only harbouring few of the spacers found in L . rhamnosus GG and a poor conservation of the cas genes . The overall CRISPR-Cas typing analysis showed that strains from the same sublineage mostly shared identical CRISPR-Cas loci . Interestingly , strains H1093 and H4692 did not have any of L . rhamnosus GG spacers but some of the cas genes remained present , whilst strain H1275 lacked the entire CRISPR-Cas locus . It has to be kept in mind that only sequences homologous to the CRISPR-Cas locus from strain L . rhamnosus were identified , allowing the possibility that additional spacers , cas genes or even additional CRISPR loci may be present . To determine the function of the CRISPR-Cas system in protecting L . rhamnosus from exogenous DNA , blastn searches on all 24 spacers were performed against virus and plasmid database at GenBank . Out of 24 spacers , 11 spacer sequences showed substantial sequence identity with plasmid or phage sequences ( Table S4 ) . Eight spacer sequences fully or partially matched known bacteriophages genomes: L . rhamnosus phage Lc-Nu , L . casei phage φ AT3 , L . casei phage Lrm1 , L . casei phage A2 and L . casei phage PL-1 . The identified CRISPR spacers thus belonged to phages from L . rhamnosus strains or closely related bacterial species , i . e . L . casei , highlighting the role of the CRISPR-Cas system as an immunity system against phage predation . Some spacers ( 4 , 12 , 18 , 21 and 22 ) have multiple phage hits , showing that the corresponding phage genomes share the same region , preventing us to predict from which bacteriophage these particular spacers were acquired . One match for plasmids was also found: the conjugative plasmid pSB102 . The data also indicates that the CRISPR-Cas system may play a role in the L . rhamnosus species diversity by controlling horizontal gene transfer and providing phage resistance , thereby contributing to diversification of the species . Our data also showed that the degree of CRISPR diversity correlated with the genomic clustering of the 100 isolates and at some extent with their ecological niche ( Figure 5 ) . Most dairy isolates shared only 6–7 spacers with L . rhamnosus GG , indicating that the variety and the exposure to phages and other mobile genetic elements differ in each habitat , i . e . the intestinal tract and cheese . We anticipate that some of the dairy strains may have an entirely different set of CRISPR sequences , representative of their own habitat and possibly additional CRISPR-Cas Types , as seen across the lactic acid bacteria [61] . All 100 L . rhamnosus isolates were tested for resistance to bile salts , a property that is usually associated with the intestinal tract environment ( Figure 5 ) . A majority of L . rhamnosus strains were bile resistant ( 45% resistant and 30% moderately resistant ) and different bile resistance profiles were observed in each niche ( Figure 5 ) . No clear association could be seen when combining the bile salt resistance data with the hierarchical clustering . A similar distribution was observed in strains isolated from clinical specimens and cheeses , even though a slightly higher proportion of bile salt-sensitive strains could be observed in the cheese isolate group . As expected , all strains from the human intestinal tract were resistant to bile salts , illustrating that such trait is essential for persisting in the intestinal tract . All vaginal isolates also showed bile resistance , suggesting that L . rhamnosus strains of the colonic microbiota may possibly have colonized the vaginal cavity as previously reported [62] . The low number of isolates from oral cavities ( n = 3 ) did not allow us to draw any conclusions , but revealed a different profile in terms of bile sensitivity . One of the hyper-variable regions in GG contained genes encoding the taurine transport system tauABC , potentially involved in the bile salt conjugation . Nine out of 25 bile-sensitive strains had a defective tauABC locus , suggesting that the tauABC locus may affect the bile sensitivity of these strains although most likely additional genes are involved . Pili in L . rhamnosus strains play a significant role in terms of interaction , colonization , persistence and potential signalling in the human intestinal tract [19]–[21] . The spaCBA pili gene cluster is flanked by numerous IS elements , suggesting that L . rhamnosus might have acquired the spaCBA pili gene cluster by horizontal gene transfer [41] , [63] , where the integration of the iso-IS30 element had constituted a promoter that allowed the expression of the pili genes in L . rhamnosus GG [63] . It also indicates that this IS element-rich chromosomal region may be subject to important genetic recombination events within the species [19] , [64] . Hence , we examined the pili diversity among all 100 isolates , providing a detailed picture on the conservation of the pili genes in each strain , since as little as one mutation is potentially sufficient to prevent the pili production or to affect the mucus binding abilities ( Figure 6 ) . Moreover , to support the genomic data , we investigated the mucus adhesion abilities of all L . rhamnosus isolates and also verified the presence of pili in a number of these strains by immunoblotting analysis ( n = 64 ) , transmission electron microscopy ( n = 10 ) and in vitro blocking mucus binding assays ( n = 22 ) ( Figures 6 , S3 and S4 ) . The mucus binding capacity ranged from 0 . 05% to 29 . 9% in all tested strains and was clearly correlated with the presence of a functional SpaCBA pili gene cluster , as shown at both genomic and phenotypic levels ( Figure 6 ) . To further demonstrate that the mucus binding capacity of these strains was mediated by SpaCBA pili , we performed in vitro blocking mucus binding assays on 22 SpaCBA-positive isolates using SpaC anti-serum as previously described ( Figures 6 and S3 ) [19] . In all 22 strains tested , the addition of SpaC anti-serum significantly reduced mucus binding , indicating that the SpaCBA pili has a major role in the interaction between L . rhamnosus and the human intestinal mucus . Remarkably , some strains displayed significant mucus-binding capacity but lacked the canonical SpaCBA pili structures , suggesting that alternative interaction players are involved . The genes encoding the SpaCBA pili of some strains such as H1242 , H1304 , F1178 and H6110 are highly conserved but , however , with some subtle sequence differences . We propose that the sequence polymorphism of the pili genes in these strains might modulate mucus binding capacity or affinity . Alternatively , we cannot rule out that additional strain-specific traits might be involved in the mucus binding , especially in strain F1178 where the residual binding in the presence of SpaC anti-serum still remained high ( Figure S3 ) . In contrast , strains with poor mucus-binding abilities appeared to have some remnants of pili genes in a more or less decayed form ( Figure 6 ) . In strains H1275 , H4689 and H1100 , the spaCBA pili gene cluster is highly conserved ( >98% ) , but show a very poor binding , indicating that the pili production may be impaired by critical mutation ( s ) or a defective promoter . The L . rhamnosus strains were further classified according to two main criteria , i . e . their ecological niche and their pilosotype , defined as the presence of pili genes that encode functional pili ( Table 2 ) . The results indicate that the production of functional SpaCBA pili was significantly more prevalent in human isolates ( 40 . 2% or 31/77 ) than in dairy isolates ( 13% or 3/23 ) . This suggests that the lack of the SpaCBA pili gene cluster in most dairy strains reflects a possible niche specialization to a habitat where pili structures are not essential and do not bring any benefit for persistence and colonization . Among all niche groups , the intestinal strains are the most prevalent group to produce SpaCBA pili , which would confer the ability to efficiently colonize and persist in the intestinal tract . In contrast , none of the strains originated from the oral and vaginal cavities produces functional pili , indicating that such trait may not be required in these two ecological niches . Our observations support the hypothesis that the human-mucus binding properties of pili may constitute an advantage to the lactobacilli to persist in the intestinal tract , but may be lost in strains evolving in other ecological niches , such as dairy products , through the decay or loss of the non-essential SpaCBA pili gene cluster . Due to the intimate interaction between L . rhamnosus and the intestinal mucosa [30] , we studied the potential signalling pathways that could be triggered by the L . rhamnosus strains . This was realized by determining the signal transduction in intestinal epithelial cells via Toll-like Receptors ( TLRs ) TLR-2 , TLR-4 and TLR-5 . All 100 isolates were tested for signallings via TLR-4 and TLR-5 receptors , but no significant responses were observed , which is in agreement with the identified ligands for these two TLRs , i . e . lipopolysaccharides and flagellins respectively ( data not shown ) . Clearly , L . rhamnosus-host signallings are mediated through different receptors . Signalling via the TLR-2 receptor in L . rhamnosus species was observed and greatly varied among isolates ( Figure S5 ) . More than half of the isolates mediated a TLR-2 response very similar to the level observed for strain GG after 1 h ( fold-induction of ∼1 . 5 ) . Six strains ( H6111 , H0009 , H4692 , H1311 , H1226 and H1131 ) triggered a stronger TLR-2 response in this assay system . We did not determine the nature of the ligand recognized by TLR2 but assume in analogy with what has been found in L . rhamnosus GG that the signalling might be mediated by the lipoteichoic acids [65] . The levels of TLR2 signalling could not be correlated with any other traits , such as EPS production , pili production or the presence of other membrane-associated proteins . No links between the TLR2 response , hierarchical clustering and ecological niches of the various strains were either identified . This suggests that the TLR-2 response triggered by L . rhamnosus does not reflect its adaptation to one particular niche , but is rather a trait acquired , maintained , altered or exacerbated by other factors that remains yet to be identified . L . rhamnosus isolates have been isolated from various ecological habitats , showing its large ecological versatility . Niche-specialized strains have developed distinctive metabolic traits , phage resistance system , stress-resistance mechanisms and colonization traits to efficiently persist in an ecological habitat . However , the microbiota of habitats such as the human intestinal tract or the vaginal cavity are rich and complex , consisting of many phylotypes [2] , [66] . L . rhamnosus strains may therefore compete with other bacterial species by producing bacteriocins that prevent growth of other bacterial populations . In contrast , the diversity and richness of the microbiota in dairy products is much lower , suggesting less competition [67] . When testing the anti-microbial activity of 92 L . rhamnosus strains , we found that most strains displayed anti-microbial activity against pathogens E . coli , Yersinia enterocolitica and Listeria monocytogenes at different pH ( Figure S6 ) . This is in line with previous studies on L . rhamnosus anti-microbial activity [44] , [68] , [69] . Remarkably , most dairy isolates shared comparable anti-microbial capabilities and clustered together , e . g . poor anti-microbial activity against E . coli and , to a lesser degree , against L . monocytogenes . The human strains displayed a differential spectrum and level of antimicrobial activity against the three human pathogens tested than most dairy strains . This illustrates the fitness of human isolates to compete with other bacteria potentially present in the human body cavities . In contrast , a high proportion of dairy isolates seems to have lost the ability to produce antimicrobial compounds against these three human pathogens , suggesting that such trait might not be essential in an environment with a lower and different microbiota diversity than in human body cavities . It , however , does not imply that those isolates do not produce antimicrobial compounds active against other pathogens more prevalent in their respective niche . The species L . rhamnosus has been isolated in various dairy products and human body cavities , highlighting the close association and frequent interactions between L . rhamnosus and the human body . The analysis of the genomes and phenotypes of 100 strains of the species L . rhamnosus provides then a wealth of information with respect to the traits that are beneficial or essential in different ecological niches and , also allowed us to depict in details the species from an anthropocentric perspective . As expected , close inspection of the hierarchical clustering of the 100 L . rhamnosus strains showed that this can be paralleled to some extent by clustering of phenotypic data , i . e . carbohydrate metabolism , antagonistic activity , CRISPR oligotyping , bile salt resistance or pilosotype . Interestingly , the integration of both phenotypic and genomic data of each strain revealed the presence of two prevailing geno-phenotypes called A & B in the L . rhamnosus species ( Figure 7 ) . The strains belonging to the geno-phenotype A are characterized by a lack of SpaCBA pili , a different carbohydrate metabolism ( D-lactose , D-maltose and L-rhamnose ) and a distinct CRISPR system profile , indicative a possible adaptation to dairy-like environment . In contrast , the geno-phenotype B depicts strains with a specific set of lifestyle traits that would confer them adequate fitness to the intestinal tract , such as bile resistance , pili production and L-fucose utilization . The geno-phenotype B showed a high similarity with L . rhamnosus GG in terms of genomes and phenotypes . The geno-phenotype A is prevalent in the cheese group , indicating their adaptation to the dairy environment . The PTS and metabolic-related genes non-essential in dairy products were lost or decayed , i . e . loss of L-fucose utilization . In parallel , we hypothesized that additional functions were acquired possibly through horizontal gene transfers , genetic mobile elements or plasmids , i . e . the ability to use lactose , a major carbon source in milk-derivative products . The loss of pili in these dairy strains is another characteristic example of a trait lost during niche-adaptation , where the absence of mucosa surfaces is reflected by the decay or complete loss of non-essential pili . In dairy niches , phage predation is ubiquitous as showed in many studies of lactic acid bacteria [70] , [71] and the CRISPR system might evolve by the acquisition of spacers representative of phages or plasmids of a particular niche . This is the case as the CRISPR locus profile between both geno-phenotypes differs considerably . Interestingly , L . rhamnosus from the vaginal cavity and urethra have a geno-phenotype A , which is in agreement with previous studies showing that the rectal microbiota is a potential reservoir of bacteria that may colonize the vaginal cavity [62] . This also suggests that the intestinal isolates ( geno-phenotype A ) may be more adapted to the vaginal environment , possibly due to their distinct metabolic abilities . This remains speculative , as at individual level , we do not know which L . rhamnosus strains these women possibly have in the intestinal tract . Interestingly , the oral isolates also possess a geno-phenotype A . Due to the low number of strains , it is difficult to draw any definitive conclusions for the oral group . However , the prevalence of the geno-phenotype A in these three niches highlights a close link between them , indicating that the geno-phenotype A strains may likely originate from either dairy products but also oral or vaginal cavities . Both geno-phenotypes A and B were found among the intestinal isolates ( Figure 7 ) . We proposed that the geno-phenotype A strains were likely introduced in the intestinal tract via consumption of foods . Bile resistant and with different metabolic capabilities , they are able to survive in the intestinal tract but may not be able to compete with other autochthonous intestinal bacteria to colonize the intestinal tract , i . e . lack of mucus-binding pili . This would indicate the most of these isolates are transient in the intestinal tract and further eliminated along with the faecal material . Other L . rhamnosus dairy isolates that are bile sensitive may also be introduced in the gastro-intestinal tract through the diet but cannot survive the intestinal conditions . On the other hand , geno-phenotype B strains are likely to be autochonous , as they possess phenotypic traits , promoting resistance and persistence in the human intestinal gut . Still , we cannot exclude the hypothesis that geno-phenotype B strains may also be transient in the gut . But this would then indicate alternative functions for the SpaCBA pili , such as binding to other mucosa . This brings us to raise one question: is L . rhamnosus only specific to the human host or are there any other potential animal reservoirs ? Addressing the host specificity of L . rhamnosus would potentially lead to identifying novel host-specific strains and remarkable adaptation patterns as reported in the species L . reuteri [45] . The clinical isolates constitute a very eclectic pool of strains , whose genotype and phenotype do not reflect adaptation patterns of their source of isolation , i . e . blood or pus , but rather of their original ecological niche . Thirty-three clinical strains could be assigned to one distinct geno-phenotype , i . e . 10 isolates with geno-phenotype A , 9 isolates with geno-phenotype B and 14 isolates with geno-phenotype BΔspaCBA . The geno-phenotypes B and BΔspaCBA differ by the presence or lack of the genomic island containing the spaCBA pili gene cluster , which is located in an unstable genomic region [64] . A number of other strains ( n = 17 ) were not be assigned any geno-phenotype , as they possess transitory geno-phenotypes or may have atypical history . It is noteworthy , that some of the clinical isolates have similar gene content to L . rhamnosus GG . However , differences in phenotypes clearly show that these strains are not identical to GG . This indicates that they may have additional genes and/or nucleotide variations in their respective genomes and share close ancestor to L . rhamnosus GG . This is in line with a previous study that showed that the widespread and increasing use of probiotic strain L . rhamnosus GG was not associated with the augmentation of Lactobacillus bacteremia [37] . To conclude , this work represents the first extensive genomic and functional analysis of the species L . rhamnosus and provides further insights into the genetics and lifestyle of this species . The data and model presented here may serve as a basis to understand the ecology of novel L . rhamnosus isolates , to identify novel probiotic candidates and also to examine the functional properties of current commercial L . rhamnosus strains .
All 100 Lactobacillus rhamnosus strains used in this study were obtained from various institutions , universities and hospitals ( Table S1 ) . Well-characterized , L . rhamnosus GG was used as reference strain throughout the study [15] , [19] , [43] . Strains VIFIT , IDOF , AKRO , CORO and NEO were isolated from probiotic-marketed products ( Table S1 ) , whereas a number of strains were made available from strain collections or institutions . All isolates were routinely propagated in anaerobic conditions at 37°C in MRS medium ( Difco BD , NJ , USA ) . Chromosomal DNA from each isolate was extracted using Wizard Genomic DNA Purification Kit ( Promega , WI , USA ) following the manufacturer's instructions . Initial bacterial identification at the species level was performed by amplification of tuf gene as described by Ventura et al . [72] , [73] using standard PCR amplification conditions and multiplex PCR amplification ( data not shown ) . Sugar metabolism and other catabolic properties of the L . rhamnosus strains were investigated using API CH 50 kit ( bioMerieux , Marcy L'Etoile , France ) . All strains were grown until logarithmic phase and then inoculated in API galleries following the manufacturer's instructions . API galleries were further incubated at 37°C in anaerobic conditions for 48 h prior to colorimetric analysis . Genomes of all L . rhamnosus isolates were sequenced on a SOLiD sequencer platform ( Life Technologies ) at the Institute of Biotechnology ( Helsinki , Finland ) . Sequence alignments and consensus sequences were generated by mapping color-space reads to the L . rhamnosus GG reference genome , using the SOLiD BioScope software ( Life Technologies ) and the SAM tools [74] . In order to transfer annotation from a reference genome ( L . rhamnosus GG ) to each un-annotated mapped genome , sequences were compared with ‘nucmer’ to identify regions that share synteny [75] . Those regions were extracted as base range in the mapped genome and in the reference genome ( L . rhamnosus GG ) . In-house custom-made scripts were then used to transfer annotation . Synteny blocks had a nucleotide sequence identity more than or equal to 40% . For each query genome , a set of shared L . rhamnosus GG orthologous genes was obtained and further analyzed . Similarly , for a number of strains , we mapped the SOLiD reads onto the LC705 genome sequence and obtained an additional set of shared L . rhamnosus LC705 orthologous genes . The L . rhamnosus GG genome was assigned to COGs using Reverse Position Specific blast and Conserved Domain Database from NCBI . Mapped genome sequences are available upon request . Human intestinal mucus was kindly collected and provided by S . Vesterlund ( University of Turku , Finland ) and H . Huhtinen ( Turku University Central Hospital , Turku , Finland ) as previously described [27] , [76] . L . rhamnosus strains were propagated and radiolabeled overnight in MRS broth supplemented with 10 µl . ml−1 [5′-3H] thymidine ( 16 . 7 Ci . mmol−1 ) . MaxiSorp microtiter plates ( Nunc , Denmark ) were coated with 100 µL of human mucus solution prepared in PBS at a final concentration of 0 . 5 mg/mL and further incubated overnight at 4°C . The wells were then washed with PBS to remove unbound mucus and 100 µL of 3H-radiolabeled bacterial suspensions at optical density ( OD600 ) 0 . 25±0 . 01 were added to the wells . The microtiter plate was further incubated at 37°C for 1 h and then wells were washed with PBS in order to remove unbound bacteria . Bacteria adhering to mucus were incubated at 60°C for 1 h in 1% SDS-0 . 1 M NaOH solution and the radioactivity level of lyzed bacterial suspensions was measured by liquid scintillation counting in a Wallac 1414 liquid scintillation counter ( PerkinElmer ) . The percentage ratio between radioactivity values of lysed L . rhamnosus suspension ( mucus-bound fraction ) and L . rhamnosus suspension ( unbound fraction ) reflects the adhesion ability to human intestinal mucus . For each isolate the experiment was performed in quadruplicate . Human mucus binding assay was performed for L . rhamnosus isolates in the presence of polyclonal SpaC antibody as described above . 3H radio radiolabeled bacteria were co-incubated with the immobilized mucus in the presence of a 1∶100 dilution of anti-SpaC serum . For each isolate , bacterial suspension adjusted to an optical density ( OD600 ) of 1 . 0 was used to extract cell wall-associated proteins . Cell pellets were washed once with PBS and disrupted mechanically by bead-beating using sterile quartz beads ( Merck KGaA , Germany ) . Cell wall material was resuspended in 500 µL of PBS and further pelleted by centrifugation at high speed for 30 min . Next , the samples were digested for 3 h at 37°C in a 50 µL enzymatic mixture containing 50 mM Tris-HCl , 5 mM MgCl2 , 5 mM CaCl2 , 10 mg/mL lysozyme and 150 U/mL mutanolysin . Samples were mixed with 12 . 5 µL of 4× Laemmli loading buffer ( BioRad , CA , USA ) and heated at 99°C for 10 min . Cell wall proteins were resolved on 10% acrylamide gel and electroblotted onto 0 . 2 µm nitrocellulose membrane ( BioRad , CA , USA ) . Polyclonal rabbit SpaA antiserum ( 1∶10 , 000 ) and peroxidase-conjugated goat anti-rabbit IgG ( Jackson ImmunoResearch , USA ) ( 1∶10 , 000 ) were respectively used as a primary and secondary antibody in 2% ( w/v ) ECL Prime Blocking Reagent ( GE Healthcare Life Science , UK ) . Membranes were blocked with 2% ( w/v ) ECL Prime Blocking Reagent , and washed with 0 . 1% Tween 20 – PBS solution in-between incubations . Membranes were analyzed using Amersham ECL Prime Western Blotting Detection Reagent ( GE Healthcare Life Science , UK ) . HEK-Blue hTLR2/4/5 cell lines ( Invivogen , CA , USA ) were used in this assay . All cell lines were grown and subcultured up to 70–80% of confluency using as a maintenance medium Dulbecco's Modified Eagle Medium ( DMEM ) supplemented with 4 . 5 g/L D-glucose , 50 U/mL penicillin , 50 µg/mL streptomycin , 100 µg/mL Normocin , 2 mM L-glutamine , and 10% ( v/v ) of heat-inactivated fetal bovine serum . For each cell line , the immune response assay was carried out by splitting HEK-Blue cells in flat-bottom 96-well plates and stimulating them by addition of 20 µl bacterial suspensions adjusted to OD600 1 , 1∶10 , 1∶100 . The 96-well plates were incubated for 20–24 h at 37°C in a 5% CO2 incubator . Receptor ligands as Pam3CSK4 ( 100 ng/mL for hTLR2 ) , LPS-EB ( 100 ng/mL for hTLR4 ) and RecFLA-ST ( 10 ng/mL for hTLR5 ) were used as positive control while maintenance medium without any selective antibiotics was used as negative control . SEAP secretion was detected by measuring the OD600 at 15 min , 1 h , 2 h , and 3 h after addition of 180 µL of QUANTI-Blue ( Invivogen , CA , USA ) to 20 µL of induced HEK-Blue hTLR2/4/5 supernatant . All cell lines were stimulated in triplicate for each isolate . Selected L . rhamnosus isolates were analyzed by transmission electron microscopy ( TEM ) as previously described by Reunanen et al . [20] . Briefly , 20 µL of overnight bacterial cultures were added to Formvar-carbon-coated copper grids for 30 min at room temperature . Grids were then washed three times with 0 . 02 M glycine solution and further incubated for 15 min in a blocking solution containing 1% ( w/v ) of bovin serum albumin ( BSA ) . Next , a 1∶100 dilution of SpaA antibody was prepared in 1% ( w/v ) BSA solution and added to the grids for 1 h , washed with 0 . 1% ( w/v ) BSA and incubated for 20 min with protein A conjugated to 10 nm gold particles . Grids were washed several times in PBS , fixed for 5 min using 1% glutaraldehyde , washed again with deionized water and stained with a solution containing 1 . 8% methycellulose and 0 . 4% uranyl acetate . Grids were visualized using JEOL JEM-1400 transmission electron microscope ( JEOL Ltd . , Japan ) . L . rhamnosus strains were cultured in MRS broth at 37°C in anaerobic conditions . The OD600 of the bacterial culture suspensions were equalized to 1 . 5 and 3 µl of cell suspensions were spotted onto MRS agar plates containing 0 . 5% ( w/v ) Ox gall ( Sigma , MO , USA ) . Plates were incubated anaerobically at 37°C for two days and visually examined . L . rhamnosus strains were grown until stationary phase as described above . Next , the cell suspensions were thoroughly homogenized and the OD600 was equalized . Cell mixtures were then centrifuged for 20 min at 650×g at +5°C and the supernatants were pH-adjusted at 5 . 0 and 6 . 20 by addition of NaOH and HCl solutions , filtered ( 0 . 22 µm filter ) and stored at −20°C for further analysis . Antagonistic assays were performed in microtiter well plates as previously described [77] . E . coli O157:H7 ( ATCC 43894 ) , L . monocytogenes R14-2-2 and Y . enterocolitica R5-9-1 were incubated for 15 h at 37°C in the presence of 30 µl of L . rhamnosus pH-adjusted supernatant . As positive controls , 30 µl of sterile MRS broth at pH 6 . 20 or pH 5 . 0 was added the dedicated medium ( TSB or LB ) inoculated with one of the pathogenic strains . As a negative control , 300 µl of medium ( TSB or LB ) was used . The OD600 values were measured in an automatic reader ( Bioscreen C , Oy Growth Curves Ab Ltd , Finland ) every 30 min . The bacterial growth was quantified using growth curves and the area under curve ( AUC ) values , automatically processed by the BioLink software ( Oy Growth Curves Ab ) . Inhibition was expressed as an area reduction percentage ( ARP ) compared to control samples grown without the addition of supernatant . | Some bacterial species are specialists and adapted to a single niche , while others are generalists and able to grow in various environmental conditions . Lactobacillus rhamnosus is a generalist and its members can often be found in different human cavities but also in various artisanal and industrial dairy products . To gain insights into the genetic complexity and ecological versatility of this species , we collected 100 L . rhamnosus strains from different niches . Genomic and functional analysis of these revealed a dichotomy within the species that reflected its adaptation to particular niches . The variable regions identified in the L . rhamnosus genome encode lifestyle traits that allowed us to demonstrate that some L . rhamnosus isolates possibly resided in multiple habitats . Our work brings valuable data on the ecological dynamics and adaptability of the species and provides a basis for a model explaining the ecology of L . rhamnosus in an anthropocentric perspective . Finally , we observed that a set of pheno-genomic markers , i . e . CRISPR oligotyping or carbohydrate metabolism , would be sufficient and among the best ways to differentiate the L . rhamnosus strains , providing a general approach to select the highest diversity in these and other bacterial species . | [
"Abstract",
"Introduction",
"Results",
"and",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"microbial",
"evolution",
"biology",
"microbiology"
] | 2013 | Comparative Genomic and Functional Analysis of 100 Lactobacillus rhamnosus Strains and Their Comparison with Strain GG |
Among the rare colonizers of heavy-metal rich toxic soils , Arabidopsis halleri is a compelling model extremophile , physiologically distinct from its sister species A . lyrata , and A . thaliana . Naturally selected metal hypertolerance and extraordinarily high leaf metal accumulation in A . halleri both require Heavy Metal ATPase4 ( HMA4 ) encoding a PIB-type ATPase that pumps Zn2+ and Cd2+ out of specific cell types . Strongly enhanced HMA4 expression results from a combination of gene copy number expansion and cis-regulatory modifications , when compared to A . thaliana . These findings were based on a single accession of A . halleri . Few studies have addressed nucleotide sequence polymorphism at loci known to govern adaptations . We thus sequenced 13 DNA segments across the HMA4 genomic region of multiple A . halleri individuals from diverse habitats . Compared to control loci flanking the three tandem HMA4 gene copies , a gradual depletion of nucleotide sequence diversity and an excess of low-frequency polymorphisms are hallmarks of positive selection in HMA4 promoter regions , culminating at HMA4-3 . The accompanying hard selective sweep is segmentally eclipsed as a consequence of recurrent ectopic gene conversion among HMA4 protein-coding sequences , resulting in their concerted evolution . Thus , HMA4 coding sequences exhibit a network-like genealogy and locally enhanced nucleotide sequence diversity within each copy , accompanied by lowered sequence divergence between paralogs in any given individual . Quantitative PCR corroborated that , across A . halleri , three genomic HMA4 copies generate overall 20- to 130-fold higher transcript levels than in A . thaliana . Together , our observations constitute an unexpectedly complex profile of polymorphism resulting from natural selection for increased gene product dosage . We propose that these findings are paradigmatic of a category of multi-copy genes from a broad range of organisms . Our results emphasize that enhanced gene product dosage , in addition to neo- and sub-functionalization , can account for the genomic maintenance of gene duplicates underlying environmental adaptation .
Analyses of nucleotide sequence variation bear great promise for advancing our understanding of evolutionary processes . However , such analyses have so far rarely targeted loci of experimentally established roles in naturally selected adaptive traits , and , instead , have mostly been conducted on candidate loci or even anonymous sequences [1]–[3] . Among the highest selection pressures known in ecology are those encountered by plants on metalliferous soils , which contain high , toxic levels of heavy metals from geological anomalies or anthropogenic contamination [4] . Examples of metalliferous soils are the widespread ultramafic ( serpentine ) soils rich in Ni , Co and Cr , and calamine soils containing high levels of Zn , Cd , and Pb . The extremophile species Arabidopsis halleri is one of the few plant taxa capable of colonizing calamine metalliferous soils [5] . In addition to its hypertolerance to Zn , Cd and likely Pb , A . halleri groups among approximately 500 known taxa of so-called hyperaccumulators of metals such as Ni , Co , Zn or Cd [6] , [7] . Hyperaccumulators are characterized by leaf metal concentrations exceeding those of ordinary non-accumulator plants by more than two orders of magnitude . Metal hyperaccumulation contributes to metal hypertolerance and has been proposed to act as an elemental defense against biotic stress [8] , [9] . A . halleri is closely related to Arabidopsis lyrata and to the genetic model plant Arabidopsis thaliana , both of which are non-hyperaccumulators and exhibit only basal metal tolerance common to all vascular plants [10] . Different from A . thaliana , A . halleri is an outcrossing , stoloniferous perennial , with a nuclear genome of 2 n = 16 chromosomes [6] . In an attempt to address the molecular basis of Zn and Cd hyperaccumulation and associated hypertolerance in A . halleri , cross-species transcriptomics approaches employing the accession Langelsheim ( Germany ) established dozens of candidate genes with potential functions in metal homeostasis , of which transcript levels were elevated in A . halleri when compared to A . thaliana [11]–[13] . Functional characterization through various molecular approaches supported a role for several of these genes including HEAVY METAL ATPASE4 ( HMA4 ) [8] , [12] , HMA3 [11] , METAL TRANSPORT PROTEIN1 ( MTP1 ) [11] , [14] , NICOTIANAMINE SYNTHASE2 ( NAS2 ) [13] , [15] , and IRON-REGULATED TRANSPORTER3 ( IRT3 ) [6] , [7] , [16] . Transcript abundance of HMA4 was highest of all identified candidate genes , with more than 100-fold higher transcript levels in both roots and shoots of A . halleri than in A . thaliana or A . lyrata [12] , [17] . The HMA4 protein is a plasma membrane transport protein acting in ATP-driven cellular export-mediated detoxification of Zn2+ and Cd2+ , as well as root-to-shoot translocation of both metals [8] , [18] . The strongly enhanced HMA4 transcript levels present in A . halleri were shown to be necessary not only for metal hypertolerance but also for metal hyperaccumulation , by employing RNA interference-mediated silencing in the A . halleri accession Langelsheim . The introduction into A . thaliana of an AhHMA4 promoter fused to an AhHMA4 cDNA suggested that AhHMA4 alone , however , is not sufficient to generate either metal hypertolerance or hyperaccumulation [8] . In agreement with these findings , genetic studies identified HMA4 and MTP1 to be located within rather large QTL regions for metal hypertolerance in a segregating back-cross 1 population of an inter-specific hybrid cross between A . halleri ( accession Auby , France ) and A . lyrata [17] , [19] . Moreover , HMA4 co-localized with one out of several major QTL for leaf Zn and Cd hyperaccumulation , respectively , in a segregating F2 population [20] , [21] . Among the candidate genes of A . halleri characterized in detail to date , HMA4 thus makes the largest contribution to both metal hyperaccumulation and metal hypertolerance . High HMA4 transcript levels were shown to be attributable to a combination of tandem gene triplication and cis-activation in the Langelsheim accession of A . halleri [8] . Promoter-reporter fusions suggested approximately equivalent quantities and localizations of promoter activity for all three A . halleri HMA4 gene copies , in agreement with copy-specific transcript quantification through quantitative real-time RT-PCR [8] . Because of almost identical protein-coding sequences , the functions of the three HMA4 protein isoforms of A . halleri have not been individually characterized . All these findings supported a critical role of enhanced HMA4 gene product dosage in naturally selected metal hyperaccumulation and hypertolerance of A . halleri [8] . Interestingly , high HMA4 transcript levels , copy number expansion and cis-activation were also reported in Noccaea caerulescens [22] , [23] , another Zn/Cd hyperaccumulator in the Brassicaceae family , in which metal hyperaccumulation and associated hypertolerance must have evolved independently . Moreover , copy number expansion appears to be common among additional highly expressed metal hyperaccumulation/hypertolerance candidate genes of A . halleri , for example the ZINC-REGULATED TRANSPORTER , IRON-REGULATED TRANSPORTER-RELATED PROTEIN ( ZIP ) genes ZIP3 , ZIP6 and ZIP9 [12] , MTP1 [14] , [24] and PLANT DEFENSIN ( PDF ) genes [25] . Gene duplication is known as a major driver of genome evolution over long timescales [26] . In eukaryotic genomes , gene duplications occur spontaneously at rates that are between 100 and 10 , 000 times higher per locus than those of base substitutions per site [27] , [28] , thus explaining the presence of substantial gene copy number variation polymorphism in genomes . For example , per haploid genome and generation , S . cerevisiae was estimated to spontaneously acquire about 0 . 002 non-synonymous base substitutions within coding regions and 0 . 02 gene duplications [28] . A number of genetic diseases of humans are caused by gene duplication events [29] , [30] . Current theory predicts the rapid loss of recent duplicates unless they undergo neo- or sub-functionalization , with few exceptions [26] , [31] , [32] . However , the factual contribution of gene duplication to evolutionary adaptation as an outcome of natural selection remains poorly understood . Functional diversity and evolutionary dynamics of multigene families are of particular importance in plant and animal immunity , as exemplified by plant Resistance ( R ) and human Major Histocompatibility Complex ( MHC ) genes [33] . Natural selection for increased gene product dosage was implied to account for copy number expansion of the BOT1 boron tolerance locus of barley [34] , the MATE1 aluminum tolerance locus of maize [35] and the human salivary amylase gene ( AMY1 ) [36] . However , these reports were based merely on functional data encompassing genotype-phenotype relationships , without evidence for selection from an analysis of sequence polymorphism . Here , we address two gaps in present knowledge , namely whether a signature of selection can indeed be identified at a locus known to functionally govern an adaptive trait and , more specifically , whether positive selection for increased gene product dosage can result in the fixation of gene duplications [37] . We detect positive selection at the copy-number expanded HMA4 metal hypertolerance locus of Arabidopsis halleri . Moreover , we show that the profile of polymorphism is unexpectedly complex as a result of ectopic gene conversion . This work can act as a guide for related studies on other duplicated genes , and warrants caution in targeted analyses as well as genome-wide scans of polymorphism when dealing with presently or historically copy-number expanded loci .
For an analysis of intra-specific nucleotide sequence diversity across the triple HMA4 genes of A . halleri , we sequenced from multiple individuals ( Table 1 ) series of 13 genomic DNA segments positioned consecutively along the 150-kb HMA4 region and in flanking regions ( Figure 1 , Figure S1A and Table S1 ) . In more detail , amplicons of between 492 and 2245 bp in length ( see Table S1 ) were designed based on published sequence data , and sequenced from between 15 and 20 individuals ( see Table S2; http://www . ebi . ac . uk , accession nos . HE995813 to HE996227 ) . The number of alleles observed per genotype never exceeded expectations of a maximum of two for any of the amplicons ( see Table S2 , lower section; see Materials and Methods section ‘Sequencing , Sequence Assembly and Assignment of Consensi’ ) . We confirmed leaf metal accumulation in these same individuals by Inductively-Coupled Atomic Emission Spectrometry analysis of field-collected leaves . Maximal concentrations exceeded 10 , 000 µg Zn g−1 leaf dry biomass and 100 µg Cd g−1 leaf dry biomass in individuals from both non-metalliferous and metalliferous sites that are characterized by toxic levels of metals in the soil and a specialist vegetation ( Table 1 ) . For comparison , we also obtained nucleotide sequence data from single individuals of the Zn/Cd-hypertolerant and -hyperaccumulating subspecies A . halleri ssp . gemmifera [38] from East Asia and the closely related Zn/Cd-sensitive , non-hyperaccumulating Arabidopsis lyrata . The genome of A . lyrata contains a single functional HMA4 gene ( Figure S1B ) in a region that is overall syntenic to A . halleri ( Figure 1 and S1A ) and A . thaliana ( Figure S1C ) [39] . In addition , the A . lyrata genome uniquely contains a second , 5′-truncated HMA4-like pseudogene in a non-syntenic position . If a novel mutation confers a strong selective advantage , the corresponding haplotype is likely to sweep through a population . This reduces or even eliminates pre-existing nucleotide sequence diversity at the affected locus and – proportionately to the extent of genetic linkage – at flanking loci through genetic hitchhiking [40] . In order to test for evidence of a selective sweep in the HMA4 genomic region of A . halleri , we calculated statistics of genetic diversity . At distant control loci ( S1 and S13 ) and loci flanking the ∼150-kb HMA4 genomic region ( S2 and S12 ) , average pairwise nucleotide sequence diversity ( π ) was between 4 . 9 and 9 . 1‰ ( Figure 1A and Table S2 ) , and thus within the published range for random neutral loci in A . halleri [41]–[43] . Comparable studies on A . halleri ssp . halleri have reported a median π of 3 . 9‰ ( between 0 . 3 and 37 . 7‰ ) for 24 randomly chosen loci [42] and a median π of 4 . 3‰ ( between 1 . 8 and 32 . 7‰ ) for a total of 8 loci [41] ( Figure S2 ) . Indeed , this was in sharp contrast with much lower values for π of between 0 . 1 and 1 . 8‰ for segments comprising sequences in the promoter regions of the three paralogous HMA4 gene copies ( S4 , S6 , S9; Figure 1A , Figure S1D and Table S2 ) . Compared to the distant and flanking control loci , π decreased gradually towards and within the HMA4 region and reached a minimum of 0 . 1‰ at the HMA4-3 promoter ( S9 ) , yielding a profile as expected upon a hard selective sweep . This characteristic profile of nucleotide sequence diversity was found to be interrupted , however , by elevated π values of between 3 . 2 and 5 . 2‰ for segments positioned within the coding sequences of the three HMA4 gene copies ( S5 , S7 , S10 ) and also for the additional segment S8 , all comprising sequences that are present in two or more , almost identical copies in the HMA4 genomic region ( Figure 1A , Tables S2 to S4 ) . The overall profile of nucleotide sequence diversity across the HMA4 region was robust against error in sequence assignment to S5 , S7 and S10 ( see Materials and Methods , Table S4A ) , as well as towards a regionally separate analysis of individuals from the Harz Mountains and the Thuringian Forest ( Table S4B ) . To further substantiate the evidence for positive selection in the genomic HMA4 region of A . halleri , we conducted statistical tests of molecular population genetics by calculating Tajima's D , Fu and Li's D* , and Fu and Li's F* [44] , [45] . For segments in the promoter regions of HMA4-1 ( S4 ) , HMA4-2 ( S6 ) and HMA4-3 ( S9 ) , these three tests unanimously indicated an excess of rare polymorphisms resulting from a depletion of higher-frequency , ancestral polymorphisms . A statistically significant deviation from expectations under neutral evolution was detected at the promoter of HMA4-3 ( S9; Figure 1B and Table S2 ) , diagnostic of positive selection . Indeed , diversity statistics indicated a unique combination of a very low value for π with a highly negative Tajima's D for S9 ( see Figure S2 ) . In agreement with these results , there were fewer long and intermediate-length branches in the topologies of maximum likelihood phylogenetic trees for HMA4 ( S4 , S6 , S9 ) than observed for control loci on either side of the HMA4 region ( S1 , S2 , S12 , S13; Figure 2 , Figure S3 ) [44] . In the region of extremely low sequence diversity in the promoter region of HMA4-3 ( S9 ) of A . halleri ssp . halleri ( see Figure 1A ) , for example , all polymorphisms were unique to single observations ( e . g . , Figure 2B , 1 . 1-2 , 1 . 3-2 , 5 . 1-1 , 7 . 2-1; see also Figure 1B ) . Taken together , statistical tests of sequence diversity , molecular population genetics and sequence phylogenies concordantly support a hard selective sweep centered on the promoter of HMA4-3 , with genetic hitchhiking [40] covering a total of 250 kb . This is comparable to previously reported selective sweeps , which affect chromosomal regions of between 60 and 600 kb in length linked to domestication loci of crop plants [46] . As demonstrated in a single individual of A . halleri [8] , the combination of gene copy number expansion and cis-regulatory divergence results in strongly enhanced steady-state HMA4 transcript levels that are necessary for metal hyperaccumulation and hypertolerance . If this was selected for in the entire species A . halleri , as indicated by the diversity statistics ( see Figure 1 ) , then we would expect high HMA4 transcript levels in all A . halleri individuals . Indeed , we observed between 20- and 130-fold higher HMA4 transcript levels across individuals of A . halleri from different collection sites , when compared to A . thaliana ( Figure 3 ) . This result supports a substantial increase in HMA4 gene product dosage in all A . halleri ssp . halleri and ssp . gemmifera individuals analyzed here , by comparison to A . lyrata and A . thaliana . For segments located within coding sequences of the three HMA4 gene copies ( S5 , S7 , S10 ) , relationships among haplotypes differed from those for segments located in HMA4 promoters ( S4 , S6 , S9 ) . Phylogenetic reconstructions of S5 , S7 and S10 did not recover three distinct groups of haplotypes as expected for three independently evolving paralogs ( Figure S4 ) . Instead , the genealogy resembled a network-like structure , with complex relationships between HMA4 haplotypes at different loci ( Figure 4A ) . For example , out of a total of 25 haplotypes , three were found at two or more of the paralogous HMA4 genes ( h13 , h20 , h25; Figure 4B ) . These results demonstrate a recurrent transfer of genetic information between the coding sequences of different HMA4 gene copies of A . halleri . Segmental transfer of genetic information between paralogous sequences can arise in somatic cells during homologous recombination-based repair of double-strand breaks , addressed here as ectopic gene conversion ( EGC , also termed interlocus or non-allelic gene conversion ) , or alternatively result from unequal crossing-over events during meiosis [29] , [47] , [48] . Quantitative PCR analysis of genomic DNA of A . halleri individuals from different collection sites was consistent with the species-wide presence of three HMA4 gene copies per haploid genome ( Figure 5 ) . Average gene copy number was estimated at 3 . 2±0 . 2 for A . halleri , compared to 1 . 8±0 . 2 and 1 . 0±0 . 1 for A . lyrata and A . thaliana , respectively ( arithmetic means ± SD ) , whereby one of the two gene copies detected in A . lyrata is a truncated pseudogene in a non-syntenic position ( see Figure S1B ) . A total of three HMA4 gene copies is in agreement with our observations of a maximum of six alleles observed per individual upon joint PCR amplification of all 3′ HMA4 coding sequences ( S5/S7/S10; Table S2 ) , and a maximum of two alleles observed in the promoter region of each HMA4 gene copy ( S4 , S6 , S9; Table S2 ) . The lack of evidence for HMA4 copy number variation among A . halleri individuals suggests that recurrent EGC events account for the segmental transfer of genetic information between paralogous HMA4 coding regions . EGC is known to be common among some genes , for example rRNA genes [29] , [47] , [49]–[51] . Paralogous genes of eukaryotes have been reported to exchange sequence information at per-locus frequencies even higher than those of spontaneous gene duplications [52] , [53] , thus contributing significantly to human disease [54] . The contribution of EGC to adaptation , however , is poorly understood . EGC is predicted to transfer a newly arisen mutation from the site of its origin in one HMA4 paralog to the corresponding sites in the other two paralogs , thus cumulatively enriching species-wide sequence diversity in each individual HMA4 gene copy [55] . This explains the higher levels of nucleotide sequence diversity detected at S5 , S7 , S8 and S10 , when compared to S4 , S6 and S9 ( see Figure 1 ) [29] , [32] . Simultaneously , EGC suppresses between-copy sequence divergence and thus results in the concerted evolution of the affected loci [29] . Our findings imply that EGC accounts for the high extent of 99 to 99 . 3% inter-copy sequence identity among A . halleri HMA4-1 to -3 coding sequences ( Table S3 ) [8] , consistent with the prevalence of EGC among duplicates of >95% sequence identity known in other organisms [29] , [47] . Hallmarks of EGC were also detected in the multi-copy portion of segment S8 outside the HMA4 coding sequence ( Figure S5A , Table S3 ) , again with a network-like genealogy ( Figure S5B–D ) and a comparably high π of 9 . 5‰ ( as opposed to π of 1 . 7‰ for the single-copy 3′-portion of S8 ) . As in A . halleri ssp . halleri , EGC was also evident among the coding sequences of HMA4 gene copies of A . halleri ssp . gemmifera ( Figure S5D and S6 ) , with an apparent additional EGC event between the promoters of HMA4-2 and -3 that was uniquely observed in this individual ( S6 , S9; see Figure S1D and compare Figure 2B and Figure S3C and S3D ) . Population genetics theory and simulations have been developed for small multigene families undergoing concerted evolution [56]–[59] . Nucleotide substitution rates were predicted to be strongly enhanced with increasing gene copy number for beneficial mutations , whereas gene copy number had no effect on substitution rates for selectively neutral mutations [57] . Indeed , the AhHMA4 protein-coding sequences represented in S5/S7/S10 , which correspond to the cytoplasmic C-terminal regulatory domain of the HMA4 protein [60] , show an over-proportionately high nucleotide sequence divergence of 22% from A . thaliana [8] . By comparison , within coding regions in general , average divergence of both A . halleri and A . lyrata from A . thaliana is around 6% . In the corresponding region of HMA4 , A . lyrata is 9% divergent from A . thaliana and 22% divergent from A . halleri . This suggests an enhanced rate of fixed sequence alterations in 3′ AhHMA4 coding sequence of S5/S7/S10 , which – according to theoretical considerations – is likely to constitute evidence for positive selection [57] . Different from predictions , however , there is no prevalence of non-synonymous over synonymous nucleotide substitutions in this region , but a prevalence of indel polymorphisms instead . Nonetheless , these considerations suggest that HMA4 gene copy number expansion is not only a result of selection for enhanced gene product dosage , but – in combination with EGC – accommodates an enhanced evolutionary rate of HMA4 under positive directional selection . Regions of the human genome hosting multigene families that undergo segmental exchange of sequence information have been addressed as hypermutable [8] , [30] . Similarly , sequence exchange was proposed to contribute to the unusually high levels of sequence diversity among plant disease Resistance ( R ) genes , which typically belong to multigene families and are often present in the genome as tandem arrays of multiple paralogous genes [61] . Alongside unequal crossing over and illegitimate inter-allelic recombination , EGC was implicated in the generation of novel pathogen recognition specificities [33] , [61]–[64] . The pervasiveness of sequence differences between paralogous R genes , despite sequence exchange , was attributed to small exchanged tracts of sequence of mostly <100 bp among multiple paralogs [61] , [63] , [64] , to the suppression of unequal crossing over within R gene clusters of homozygotes [61] , to the occurrence of inter-allelic rather than inter-locus gene conversion [33] , or to the past discontinuation of sequence exchange [65] . By comparison to the high sequence diversification among paralogous R genes , the concerted evolution of A . halleri HMA4 paralogs is in stark contrast . This could be interpreted to indicate a prominent role for selection in determining the outcome of inter-locus sequence exchange , a process that appears to be common at least in some classes of multigene families [61] , [62] , [64] . The evolutionary events reflected in the profile of nucleotide sequence diversity across the HMA4 region of A . halleri occurred concurrent with or after the divergence from the A . lyrata lineage . Whereas nucleotide sequence diversity within A . halleri ssp . halleri was not positively correlated with the genetic divergence from A . lyrata across the HMA4 region ( Figure S7A ) , we detected shared ancestry of nucleotide sequence diversity profiles in the two subspecies of A . halleri , ssp . halleri and ssp . gemmifera . This is supported by a positive correlation between inter-subspecies sequence divergence and sequence diversity π within ssp . halleri ( Figure S7B ) , by the grouping of ssp . gemmifera alleles among ssp . halleri alleles in genealogies ( Figures 2 , S1D , S3 , S4 ) , and by shared polymorphisms among the two A . halleri subspecies in the coding sequences of HMA4 genes ( Figure S6 ) as well as downstream of HMA4-2 ( S8 ) ( Figure S5D ) . These findings also indicate that our sampling captured a large proportion of sequence diversity within A . halleri , which was further confirmed by a larger genetic diversity of A . halleri within collection sites than between collection sites or between regional subgroups of collection sites according to analyses of molecular variance ( AMOVA ) ( Table S5 ) . Our results support two consecutive duplications of HMA4 with or after the split of the lyrata and halleri lineages , which was estimated at between 2 mio . years ago according to sequence divergence [66] and around 0 . 34 mio . years ago according to approximate Bayesian computation [43] . Previous estimates of the timing of HMA4 duplication events 0 . 36 and 0 . 25 mio . years ago , respectively , are likely to require downward adjustment as they were based on single A . halleri sequences for each of the three gene copies and did not take into account EGC [43] . Enhanced HMA4 gene product dosage is known to functionally underlie the environmental adaptations of heavy metal hyperaccumulation and hypertolerance in the wild plant A . halleri [8] . Here , we detect positive selection in HMA4 promoter regions of A . halleri , incurred by either activating cis-regulatory mutations or gene copy number expansion of HMA4 , and likely by both . Furthermore , we identify ectopic gene conversion to effect the concerted evolution of paralogous HMA4 coding sequences , a finding that adds unexpected complexity to the profile of sequence polymorphism . We expect that , together , our results coin a class of multi-copy genes associated not only with instances of environmental adaptation in plants [6] , [51] , [67] , but also more generally with eukaryotic adaptation [29] , [32] , [36] , [37] , [68] . Thus , this work will stimulate the development of crop breeding strategies based on gene copy number variation [34] , [69] . In the future , complex profiles of nucleotide sequence polymorphism , as exemplified by the HMA4 region of A . halleri , will deserve designated attention in advanced targeted studies as well as in large-scale genome scanning approaches [2] , [3] , [70] . Subsequent to gene duplication events [27] , alongside neo- and sub-functionalization , selection for more of the same gene product is of higher evolutionary relevance than previously appreciated [26] , [31] , [32] , [71] .
Leaf tissues and soil samples were collected in the field from 18 randomly selected A . halleri ssp . halleri individuals at 7 European sites ( Table 1 ) . A minimum distance of 2 m was kept between sampled individuals to avoid sampling clones because A . halleri is stoloniferous . From a subset of collected genotypes , clones were propagated vegetatively and maintained in a greenhouse . For element analysis by Inductively-Coupled Plasma Atomic Emission Spectrometry ( ICP-AES ) , leaf material was washed with ultrapure water and dried at room temperature ( RT ) for >1 week , followed by processing of samples and measurements as described [11] , [12] . For the determination of extractable soil metal concentrations , soil cores were taken down to 0 . 05 m depth within 0 . 1 m distance from each individual . Three g of air-dried , sieved soil ( 2 mm particle size ) were extracted in 25 ml of 0 . 1 M HCl with rotary shaking at 150 rpm at RT for 0 . 5 h . For DNA extraction , leaf tissues were frozen in liquid nitrogen immediately after harvest , kept on dry ice for up to 20 h , and stored at −80°C until further processing . Additionally , previously characterized greenhouse-cultivated , clonally propagated genotypes were included in some experiments: the BC1 parent individual from Auby ( individual 8 . 1 ) [17] , [72] , individuals 1 . 1/Lan 3 . 1 [8] , [12] and 1 . 4/W504 [13] from Langelsheim , and an individual ( 9 . 1 ) of A . halleri ssp . gemmifera [38] ( Table 1 ) . Genomic DNA was extracted using the DNeasy Plant Mini Kit ( Qiagen , Venlo , The Netherlands ) from 100 mg of frozen leaf material of each genotype . The thirteen amplicons designed to analyze sequence diversity ( S1 to S13 ) comprised either non-coding ( i . e . , promoter , UTR and intron ) or both non-coding and coding sequences , and were positioned within all three of the HMA4 gene copies and at loci of increasing distances upstream and downstream of HMA4 ( Figure 1 , Figure S1A , Tables S1 and S2 ) . No additional amplicons could be designed in the repeat- and transposon-rich genomic regions between HMA4 genes [8] . Primer sequences for amplicons S2 to S12 were designed based on available A . halleri BAC sequences ( Genbank accession numbers EU382073 . 1 and EU382072 . 1 ) ( Table S1 ) [8] . Primer design for S1 and S13 was based on the Arabidopsis thaliana and Arabidopsis lyrata ssp . lyrata genome sequences [39] , [73] . In A . thaliana and A . lyrata , S1 is located 116 and 198 kb upstream of S2 , and S13 is located 113 kb and 2 . 47 Mbp downstream of S12 , respectively . Amplicons comprising the 3′-portions of AhHMA4-1 ( S5 ) , AhHMA4-2 ( S7 ) and AhHMA4-3 ( S10 ) were simultaneously amplified in each of three independent PCRs using primer pairs that were not copy-specific ( Table S1 ) . In contrast , primers for S8 amplified only the 3′-end of AhHMA4-2 and additional downstream intergenic sequence , taking advantage of copy-specific sequence polymorphisms in the design of the reverse primer ( see Figure S5A ) . For PCR amplification , 2 µl of genomic DNA were used with GoTaq DNA polymerase ( S1 , S2 , S11 and S13 , Promega , Leiden , The Netherlands ) , Bio-X-Act Long DNA polymerase ( S4 , S6 , Bioline/Gentaur , Brussels , Belgium ) or a mix of both enzymes ( S3 , S9 , S12 and S5/S7/S10 ) , the respective primer pairs ( 0 . 5 µM each ) ( Table S1 ) and dNTPs ( 200 µM each ) ( Fermentas , St . Leon-Rot , Germany ) in a final volume of 25 µl , the latter enzyme allowing more efficient amplification . PCR reactions were carried out as follows: 3 min at 95°C , followed by 30–32 cycles of 30 s at 95°C , 30 s at 58°C , 1 min per kb at 70–72°C , and a final extension step of 7 min at 70–72°C . PCR products were gel-purified and cloned into the pGEM-T easy vector ( Promega , Leiden , The Netherlands ) before transformation of E . coli DH5α . In order to ensure with high probability that both alleles were sampled in heterozygous individuals through DNA sequencing , plasmid DNA was isolated from overnight cultures of at least eight independent bacterial colonies per amplicon and genotype , 20 clones for S6 and a total of 56 clones for S5/S7/S10 , respectively , before sequencing of inserts by the Sanger method on an ABI 3730xl automated sequencer ( Applied Biosystems , Darmstadt , Germany ) using vector-specific and additional locus-specific primers when required ( Table S1 ) . For two individuals , 48 additional clones from two further independent PCRs were sequenced for S5/S7/S10 to resolve remaining sequence ambiguities . For the S6 amplicon ( corresponding to the promoter region of HMA4-2 ) , a set of substantially divergent sequences was initially obtained , and , including these , a total of more than the expected maximum of two types of S6 sequences , corresponding to two alleles expected per individual at this single locus , were found in several A . halleri individuals . Using a combination of PCR , BAC end sequencing and DNA gel blot analyses of previously isolated A . halleri BACs harboring HMA4 and related sequences [8] , the divergent set of sequences was unequivocally attributed to the promoter of AhHMA2 , which was found to occur in tandem with AhHMA3 on a BAC clone , but this BAC did not contain any AhHMA4 coding sequence . AhHMA2 and AhHMA3 are orthologs of AtHMA2 and AtHMA3 that are located in tandem on chromosome 4 of A . thaliana whereas AtHMA4 is on chromosome 2 . HMA4 , HMA2 and HMA3 genes all encode divalent transition metal cation-transporting P1B-type ATPases [18] , [74] . Sequence assembly was conducted with DNASTAR ( DNASTAR Inc . , Madison , USA ) . First , a consensus sequence was generated for each clone . Then , each consensus was compared to all other consensi from the same amplicon in a given individual and to all consensi of the same amplicon from all other individuals to i ) correct Taq polymerase errors , ii ) identify recombinant chimeras that resulted from template switches during PCR amplification [75] and iii ) distinguish heterozygous from homozygous loci . For the 3′-regions of the three HMA4 gene copies ( S5 , S7 , S10 , S8 ) more than 800 sequences were obtained in total . Among these , sequences were considered to be authentic when the same sequence was observed at least three times from one PCR reaction or in at least two independent PCRs of the same genotype . After removal of chimeras ( which accounted for ca . 5% of the sequences ) , a total of 25 consensus sequences were retained for the 3′-regions of HMA4 gene copies . These consensi were assigned to the three HMA4 loci taking advantage of i ) the copy-specific sequence information for AhHMA4-2 via the overlap between S7 and S8 for each individual ( see Figure S5A ) , ii ) position information available from two completely sequenced BACs [8] , and iii ) step-wise inference using a strictly parsimonious approach , similar to the strategy used to solve a SUDOKU in two times three double-blind independent replicates to ensure reproducibility . After sequence assembly and alignment , DnaSP v5 [76] was used to calculate sequence diversity ( π ) , Tajima's D [45] , Fu and Li's D* and F* [44] , and to conduct other statistical tests of molecular population genetics . MEGA v5 was used for phylogenetic analyses [77] . The ML trees shown throughout were constructed using a general time-reversible model . Rates among sites were assumed to be gamma-distributed with invariant sites , and 5 discrete categories of gamma were used . All sites were used . To estimate bootstrap support for the nodes , 1000 replicates were calculated . Neighbor joining methods yielded essentially the same results for tree branching orders . Genome sequence information from A . lyrata ssp . lyrata was used as a reference [39] . Network analyses for HMA4 genes and for S8 were conducted with TCS v1 . 21 using a connection limit of 95% [78] . Alignment gaps were re-coded with nucleotides to reflect the exact number of mutational steps between sequences in the respective sequence portion . AMOVA ( Analysis of Molecular Variance ) was carried out with Arlequin 3 . 5 [79] to compare the contribution of three hierarchical levels to genetic variance: among the geographic regions of the Thuringian Forest ( A . halleri ssp . halleri ) , the Harz Mountains ( A . halleri ssp . halleri ) , and Japan ( A . halleri ssp . gemmifera ) , among geographic collection sites in each of these three regions , and within single geographic collection sites . A total of 1000 permutations were carried out for each locus , with equal weights of 1 for transitions and transversions , and a deletion weight of 0 . Quantitative PCR reactions were performed on 5 ng of genomic DNA in 384-well plates with an ABI Prism 7900HT system ( Applied Biosystems , Brussels , Belgium ) using MESA GREEN qPCR MasterMix ( Eurogentec , Liège , Belgium ) . Mean reaction efficiencies were determined from all reactions for each amplicon ( >270 reactions , Table S6 ) [80] and used to calculate relative gene copy number by normalization with the qBase software [81] using ( i ) multiple single-copy reference amplicons and ( ii ) A . thaliana genomic DNA ( Col-0 ) as a calibrator [82] . Three single-copy reference amplicons were selected and designed at the 5′- and 3′-ends of the AhFRD3 gene [12] and in the S13 amplicon ( this study ) , respectively . Their adequacy to normalize gene copy number in our experimental conditions was validated using the geNorm module in qBase ( gene stability measure M = 0 . 309 , pairwise variation CV = 0 . 121 ) [83] . Fresh cuttings of greenhouse-grown A . halleri and A . lyrata genotypes were cultivated hydroponically in 0 . 1× Hoagland solution for about 2 weeks [13] . After rooting , plants were transferred to pots with soil and further grown in a greenhouse with temperature settings of 22°C ( day ) /20°C ( night ) and a photoperiod of 16 h light and 8 h dark . Leaf material was harvested twice independently from the same individuals at an interval of eight weeks , immediately frozen in liquid nitrogen and stored at −80°C . A . thaliana and A . lyrata plants were grown from seeds as described , with harvest of leaves from 6-week-old plants , alongside harvest of A . halleri tissues [12] . Total RNA was extracted with TRIzol Reagent ( Invitrogen , Karlsruhe , Germany ) , cDNA was synthesized from 1 µg of DNaseI-treated ( Invitrogen ) total RNA using oligo-dT and the SuperScript First-Strand Synthesis System ( Invitrogen ) . Quantitative PCR was conducted in 96-well plates with a MyiQ Single Color Real-Time PCR Detection System ( Bio-Rad , Munich , Germany ) using SYBR Green qPCR Master Mix ( Eurogentec , Cologne , Germany ) . A total of three technical repeats were run per cDNA and primer pair combination . Data were analyzed using iQ5 Optical System Software version 2 . 0 ( Bio-Rad ) . Relative transcript levels of HMA4 were calculated by normalization to EF1α as a constitutively expressed reference gene [12] . Primers were as follows: AhHMA4 primers ( 5′- GCTGCAGCGATGAAAAACAAAC-3′ and 5′-TCCATACAACATCCCGAGGAAC-3′; amplification efficiency: 1 . 88 ) ; AlHMA4 primers ( 5′- TGAAGGTGGTGGTGATTGCA-3′ and 5′-CTCTCCACATTGACCAACTTTG-3′; amplification efficiency: 1 . 90 ) . AtHMA4 and EF1α primers were described earlier [12] . Sequence data are available through EBI ( http://www . ebi . ac . uk ) , accession nos . HE995813 to HE996227 . | Existing genetic diversity reflects evolutionary history , but it has rarely been possible to probe for footprints of selection at loci known to functionally govern adaptive traits . Both naturally selected metal hypertolerance and extraordinary leaf metal accumulation of the extremophile Arabidopsis halleri require strongly enhanced transcript levels of Heavy Metal ATPase4 ( HMA4 ) encoding a PIB-type ATPase that pumps Zn2+ and Cd2+ out of specific cells . By comparison to the metal-sensitive A . thaliana , highly elevated HMA4 expression results from a combination of gene copy number expansion and cis-regulatory modifications . But how do these findings , which were based on a single accession , relate to species-wide HMA4 sequence diversity in A . halleri ? Addressing this question , we detect positive selection in the promoter regions of three tandem A . halleri HMA4 paralogs , which are uniformly cis-activated . The accompanying hard selective sweep , however , is segmentally eclipsed as a consequence of recurrent ectopic gene conversion among HMA4 protein-coding sequences , which undergo concerted evolution . Together , this constitutes an unexpectedly complex profile of polymorphism as a result of natural selection . Our observations can serve as a blueprint for future analyses of duplicated genes that have undergone selection for more of the same gene product . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] | [
"evolutionary",
"ecology",
"genetic",
"mutation",
"mutation",
"types",
"ecology",
"natural",
"selection",
"genetics",
"population",
"genetics",
"molecular",
"genetics",
"plant",
"ecology",
"biology",
"evolutionary",
"biology",
"gene",
"duplication",
"plant-environment",
"... | 2013 | Hard Selective Sweep and Ectopic Gene Conversion in a Gene Cluster Affording Environmental Adaptation |
Plants respond to herbivory with the emission of induced plant volatiles . These volatiles may attract parasitic wasps ( parasitoids ) that attack the herbivores . Although in this sense the emission of volatiles has been hypothesized to be beneficial to the plant , it is still debated whether this is also the case under natural conditions because other organisms such as herbivores also respond to the emitted volatiles . One important group of organisms , the enemies of parasitoids , hyperparasitoids , has not been included in this debate because little is known about their foraging behaviour . Here , we address whether hyperparasitoids use herbivore-induced plant volatiles to locate their host . We show that hyperparasitoids find their victims through herbivore-induced plant volatiles emitted in response to attack by caterpillars that in turn had been parasitized by primary parasitoids . Moreover , only one of two species of parasitoids affected herbivore-induced plant volatiles resulting in the attraction of more hyperparasitoids than volatiles from plants damaged by healthy caterpillars . This resulted in higher levels of hyperparasitism of the parasitoid that indirectly gave away its presence through its effect on plant odours induced by its caterpillar host . Here , we provide evidence for a role of compounds in the oral secretion of parasitized caterpillars that induce these changes in plant volatile emission . Our results demonstrate that the effects of herbivore-induced plant volatiles should be placed in a community-wide perspective that includes species in the fourth trophic level to improve our understanding of the ecological functions of volatile release by plants . Furthermore , these findings suggest that the impact of species in the fourth trophic level should also be considered when developing Integrated Pest Management strategies aimed at optimizing the control of insect pests using parasitoids .
Plant volatiles play a profoundly important role in the structure and function of ecological communities [1]–[6] . Volatiles make a plant and its condition apparent to community members at different trophic levels [6] , [7] and may , thereby , mediate interactions between organisms at higher trophic levels [8] , [9] . Nowhere has this been better investigated than for interactions between insect herbivores and their natural enemies , such as primary parasitic wasps ( or “parasitoids” ) at the third trophic level . Many parasitoids have evolved finely tuned responses to volatiles emitted by plants that are attacked by their otherwise inconspicuous herbivorous hosts . By responding to volatiles and parasitizing the herbivores , parasitoids may reduce the amount of herbivory that plants are exposed to [1]–[3] and are , therefore , hypothesized to benefit plant fitness [10]–[12] . However , besides attracting beneficial parasitoids , the volatiles affect interactions between plants and other community members that may affect the fitness benefit of volatile release . Food webs generally include four or more trophic levels [13] , [14] . Thus far , little is known about foraging behaviour of the enemies of parasitoids ( i . e . , hyperparasitoids ) that are an important group of fourth-trophic-level organisms , because hyperparasitoids have not been included in the debate on the fitness benefit of volatile release by plants [15] . Hyperparasitoids are parasitic wasps that attack the larvae and pupae of primary parasitoids , and they comprise a major component of the fourth trophic level in insect communities [15] . Thus far , little is known about the cues that hyperparasitoids use to find their primary parasitoid hosts [15] , [16] . For secondary hyperparasitoids ( i . e . , hyperparasitoids that attack the fully cocooned pupae of primary parasitoids ) , their hosts are likely to be inconspicuous because the pupae do not feed and , therefore , do not indirectly reveal their presence through induced volatiles of the food plant . Furthermore , the time window for successful hyperparasitism of pupae is often narrow and restricted to the first few days after the pupae are formed [17] . However , plants have been shown to respond differently to feeding damage inflicted by parasitized or unparasitized herbivores [18]–[20] . As a result , plant-derived volatile cues may provide hyperparasitoids with reliable information on the presence of their host [15] . Many hyperparasitoid species parasitize the pupae of a range of primary parasitoid species , including both solitary species , which lay a single egg within an herbivore , and gregarious species , which lay multiple eggs within a single herbivore [21] . Single or multiple parasitoid larvae developing in a caterpillar differentially affect the physiology and feeding behaviour of the herbivore , thereby inducing different plant volatile blends [19] , [20] . Consequently , hyperparasitoids may be better able to locate one parasitoid host than the other , and thus variation in plant volatiles induced by parasitized herbivores may cause variation in the level of hyperparasitoid attack on different species of primary parasitoids . Here , we test whether plants can mediate interactions between third- and fourth-trophic-level organisms by providing cues on the presence of hosts for hyperparasitoids and whether hyperparasitoids respond similarly to herbivores that contain different parasitoids . To study these questions we used the hyperparasitoid Lysibia nana ( Ichneumonidae ) that attacks pupae of primary parasitoids in the genus Cotesia ( Hymenoptera: Braconidae ) . The solitary parasitoid C . rubecula ( CR ) and the gregarious C . glomerata ( CG ) are primary parasitoids that both attack caterpillars of the Small Cabbage White butterfly , Pieris rapae ( PR ) , that feed on brassicaceous plants ( Figure 1 ) [22] . When fully developed , the parasitoid larvae leave their host to spin a silk cocoon in which they pupate . Individual C . glomerata cocoons are approximately 40% smaller ( in terms of mass ) than individual C . rubecula cocoons . In terms of the per capita fitness potential of hyperparasitoid offspring , L . nana may benefit when developing in pupae of the larger C . rubecula [23] . However , in terms of cumulative maternal fitness , L . nana hyperparasitoids benefit more when finding a caterpillar parasitized by the gregarious C . glomerata . In the field , caterpillars parasitized by C . glomerata produced an average of 39 C . glomerata cocoons . Upon finding clusters of their primary parasitoid host , hyperparasitoids generally parasitize most or even all pupae in the brood , whereas in the case of finding a solitary primary parasitoid they have to disperse after each parasitization . The egg load of L . nana females closely approximates the average brood size ( e . g . , 20–40 ) of C . glomerata [17] , suggesting that it co-evolved with gregarious host species such as C . glomerata . We hypothesized that L . nana uses volatile plant cues induced by C . glomerata–parasitized caterpillars to locate aggregated pupae and that they prefer those volatiles over volatile plant cues induced by the feeding of solitarily parasitized caterpillars .
Hyperparasitoids did respond to herbivore-induced plant volatiles . L . nana females preferred volatiles from plants damaged by either caterpillars parasitized by primary parasitoids ( both C . glomerata [PR-CG] and C . rubecula [PR-CR] ) or unparasitized caterpillars ( PR ) over volatiles from undamaged plants in a Y-tube olfactometer ( Figure 4; binomial tests , p<0 . 001 ) . The hyperparasitoids did not discriminate between volatile blends from plants damaged by unparasitized caterpillars and plants damaged by caterpillars parasitized by the solitary parasitoid C . rubecula ( Figure 4; binomial test , p = 0 . 480 ) . However , plant volatiles induced by C . glomerata–parasitized caterpillars were more attractive to L . nana than volatiles from plants damaged by unparasitized caterpillars or those parasitized by the solitary parasitoid C . rubecula ( binomial tests , p = 0 . 021 and p = 0 . 007 , respectively ) ( Figure 4 ) . Unparasitized and C . glomerata–parasitized caterpillars inflicted more damage to plants than caterpillars parasitized by a solitary C . rubecula parasitoid—that is , respectively , 2 , 485±1 , 183 ( mean ± SD ) , 1 , 855±810 , and 705±313 mm2 of leaf tissue consumed per caterpillar in 24 h ( Kruskal-Wallis Test , p<0 . 001 ) . Because parasitoid species differentially affect caterpillar feeding rate by regulating the growth of their host [19] and the rate of feeding damage per se may mediate the attraction of L . nana , we controlled for the amount of damage in a subsequent test . We damaged plants with a pattern wheel and applied 25 µl of oral secretion from either unparasitized or parasitized caterpillars to the damaged sites . Parasitoid species are known to alter the composition of the oral secretions of their host and thereby strongly affect the response of the plant to a parasitized caterpillar [20] . Compounds in the oral secretions of Pieris caterpillars play a key role in inducing volatile release by their food plant [24] . L . nana preferred the volatiles from plants that were treated with oral secretions obtained from C . glomerata–parasitized caterpillars over volatiles from plants treated with oral secretions from unparasitized caterpillars ( Figure 4 ) . Oral secretions of parasitized caterpillars alone ( i . e . , without application to wounded sites ) did not attract the hyperparasitoids ( Figure 4 ) . Analysis of the volatile blends of plants induced by C . glomerata–parasitized , C . rubecula–parasitized , or unparasitized caterpillars revealed that these three herbivore treatments induce volatile blends that differ from undamaged control plants . A total of 33 compounds that were present in all samples of at least one of the four plant treatments were tentatively identified and included in further analysis ( Table 2 ) . In the PLS-DA , undamaged control plants grouped separately from the three treatments with caterpillar feeding ( Figure 5 ) . Amongst the caterpillar-damage treatments , plants damaged by feeding of unparasitized caterpillars and caterpillars parasitized by C . rubecula overlapped largely in their volatile headspace as shown by PLS-DA . Plants damaged by C . glomerata–parasitized caterpillars were only 40% similar in their volatile headspace to plant headspaces induced by the two other caterpillar treatments and were most distinctly different from undamaged control plants . Nine compounds most strongly contributed to the differences among treatments are indicated by VIP scores higher than 1 . These compounds included terpenoids , a ketone , a nitrile , and two unknown compounds ( Table 2 ) . The concentrations of two compounds differed significantly among the caterpillar treatments . Plants damaged by C . glomerata–parasitized caterpillars produced higher concentrations of ( E ) -4 , 8-dimethylnona-1 , 3 , 7-triene [ ( E ) -DMNT] , a known attractant for parasitoids [25] , and of an unknown compound compared to plants damaged by C . rubecula–parasitized or unparasitized caterpillars . The similarity of the volatile blends from plants damaged by unparasitized and C . rubecula–parasitized P . rapae matches the observation that hyperparasitoids did not discriminate the two treatments in choice assays . The hyperparasitoids did prefer plants damaged by C . glomerata–parasitized caterpillars over other damage treatments , which is supported by the difference in the composition of the volatile blends emitted by the plants submitted to these treatments . In the field , we confirmed that plant volatiles play an important role in the location of parasitoid pupae by hyperparasitoids . In an experimental field , B . oleracea plants were subjected to four induction treatments: no damage ( UD ) , feeding by healthy P . rapae caterpillars ( PR ) , or feeding by P . rapae caterpillars parasitized by either C . rubecula ( PR-CR ) or C . glomerata ( PR-CG ) . After the caterpillars had fed on the plants for 10 d , which was approximately the total development period of the Cotesia larvae , the caterpillars were removed . On half of the plants per treatment , we then attached C . glomerata cocoons and on the other half C . rubecula cocoons . The cocoons were exposed to the natural population of hyperparasitoids and recollected to assess the number of cocoons that was hyperparasitized . C . glomerata pupae that were attached to plants damaged by C . glomerata–parasitized caterpillars were more frequently hyperparasitized than pupae attached to plants damaged by unparasitized or C . rubecula–parasitized caterpillars ( Figure 6 , Table 3 ) . However , when C . rubecula cocoons were used to assess hyperparasitism rates , we found no induction treatment effect . The preference of L . nana for volatiles derived from plants damaged by C . glomerata–parasitized caterpillars has profound consequences for the primary parasitoid C . glomerata in the field . During the growing season of cabbage plants in 3 consecutive years in the vicinity of Wageningen , the Netherlands , we collected 1 , 256 cocoon clusters of the gregarious primary parasitoid C . glomerata and 1 , 668 cocoons of the solitary primary parasitoid C . rubecula and assessed the rate of natural hyperparasitoid attack . Clusters of C . glomerata cocoons more often contained at least a single hyperparasitoid than did solitary cocoons of C . rubecula ( Generalized Linear Model , deviance = 496 . 62 , p<0 . 001; Table 4 ) . From 17 . 4% of the C . glomerata clusters , more than one ( and occasionally even four ) hyperparasitoid species emerged . Hyperparasitoid communities associated with the gregarious primary parasitoid also consisted of more species than were found on the solitary parasitoid ( Figure 7; Table 5 ) . Within clusters of C . glomerata cocoons that were attacked by hyperparasitoids , 65%–81% of the pupae in the cluster yielded hyperparasitoid wasps . The combined attack rate of clusters and the fraction of pupae hyperparasitized in a cluster resulted in a total hyperparasitism rate of individual C . glomerata pupae of 20%–55% over the 3 y ( Figure 7 ) . By contrast , only 5%–15% of individual C . rubecula cocoons were hyperparasitized over the 3 y of our field experiments . Gregarious C . glomerata are , therefore , not only more easily found by hyperparasitoids , but once found , the hyperparasitoid wasps are able to parasitize large numbers of parasitoid pupae within broods , revealing that they exhibit an aggregative response to this clustered resource .
Secondary hyperparasitoids that are attracted to volatiles emitted by plants that are damaged by parasitized caterpillars containing fully grown parasitoid larvae may suffer fitness costs by arriving too early on a plant ( i . e . , before the larvae of the their primary parasitoid hosts have emerged and constructed cocoons ) . However , pupae themselves do not interact with plants , and plants on which no active feeding takes place are likely to have reduced volatile emission and are , therefore , more difficult to detect [1] . In addition , hyperparasitoids are even further constrained in host searching as they can only successfully parasitize young pupae of primary parasitoids ( i . e . , within several days after they are formed ) [30] . Therefore , hyperparasitoids are likely to have evolved to respond to highly detectable and reliable cues that predict the presence of available pupae in the near future and require hyperparasitoids to wait , rather than responding to cues that may result in arriving too late . Similar waiting strategies have been observed for pupal parasitoids that are similarly constrained in terms of host suitability ( i . e . , they can only parasitize the pupae of herbivores shortly after these are formed ) . We have observed on numerous occasions that the pupal parasitoid Pteromalus puparum sits on or next to final instar Pieris caterpillars awaiting their pupation ( E . H . Poelman and J . A . Harvey , unpublished observations ) . For both hyperparasitoids and pupal parasitoids , natural selection may favour such a waiting strategy even further , because healthy and parasitized caterpillars are well known to leave the host plant during the wandering phase and to climb onto a neighbouring plant and are thus not detectable through volatiles emitted by the plant on which they were previously feeding . The strategy of herbivores or parasitized herbivores to wander off the plant on which they have been feeding suggests that selection is being imposed on such behaviour , which is likely to be mediated by responses of their own natural enemies such as hyperparasitoids . Despite the potential delay of arriving too early and having to wait on the plant to be able to parasitize the primary parasitoid pupae , the plant volatile cues derived from feeding by parasitized herbivores are the most detectable cues predicting host presence . L . nana females generally carry up to 40 mature eggs after several days that are ready for oviposition and thus can exploit an entire brood of C . glomerata within several hours [30] . Moreover , a female can mature an additional 20–30 eggs over the course of 24 h [30] . This suggests that L . nana has probably co-evolved with host species such as C . glomerata because of the strong synchrony between egg load dynamics in the hyperparasitoid and average cocoon cluster size in C . glomerata [31] . Foraging decisions of the hyperparasitoids on the food plant clearly underlie the contrast between C . glomerata and C . rubecula hyperparasitism levels when cocoons of C . glomerata and C . rubecula were exposed to the natural hyperparasitoid community on plants damaged by parasitized or healthy caterpillars . The field study in 2011 showed a preference of hyperparasitoids for plants damaged by gregariously parasitized caterpillars , and this was reflected in hyperparasitism rates on gregarious C . glomerata pupae ( Figure 6 ) , supporting our findings in the laboratory choice assays . However , the effects of the herbivory treatments in the field assay did not prevail on solitary pupae of C . rubecula . Moreover , we also found a higher hyperparasitism rate of L . nana in the solitary pupae in the field season of 2011 compared to the field seasons of 2005 to 2007 . Several factors , which are not necessarily mutually exclusive , may account for this: first , our method of offering pupae on paper may expose solitary cocoons more to hyperparasitoids than occurs in nature . Second , when considering the total number of pupae in a brood , we offered more C . glomerata than C . rubecula pupae . Although more hyperparasitoids were recovered from gregarious pupae , the rates of individual clutches of C . glomerata that contained any hyperparasitoid and solitary pupae of C . rubecula that were hyperparasitized were similar . The hyperparasitism rates on C . glomerata underestimate the actual rates at which clutches were found by more than one hyperparasitoid . Third , due to the setup of this field study , we excluded hyperparasitism by primary hyperparasitoids ( that oviposit in the parasitoid larvae when these develop within the caterpillars ) . Therefore , L . nana might encounter less competition from primary hyperparasitoids and may therefore alter its oviposition strategies . Fourth , as described above , female hyperparasitoids , such as L . nana , may exploit a large proportion of their host pupae once they locate a host clutch [31] . The hyperparasitoids locating a gregarious brood spend more time on the brood and are egg limited when exploiting the whole brood , whereas they are time limited when exploiting solitary pupae . Therefore , hyperparasitism rates may have been elevated on solitary C . rubecula pupae despite the larger number of eggs laid in gregarious broods . Our study shows that enemies of those natural enemies that benefit plant fitness may also use plant-produced odours to find their hosts or prey . In this way , the plant may be caught between a “rock and a hard place , ” in that two out of three trophic levels of consumers that are detrimental to the plant ( either directly , through herbivory , or indirectly , through a reduction in the abundance of beneficial carnivores at the third trophic level , caused by organisms at the fourth trophic level ) benefit from using herbivore-induced plant volatiles . The beneficial effect on plant fitness of attracting parasitic wasps , to indirectly defend itself against their herbivore attackers , has been intensively discussed [2] , [3] , [6] . Although it has been recognized that volatiles released by plants that are under attack by herbivores provide parasitoids and predators with a cue that can be used in host location , the presence of active “signalling” and associated selection on plants that are stronger signallers has thus far received less attention . Although some studies have reported a fitness benefit of plants on which herbivores were attacked by parasitoids [10]–[12] , other studies have reported negative fitness consequences of plants emitting volatiles by becoming more apparent to herbivores [8] . It is important to emphasize that volatile cues may provide many community members with information and thereby may not necessarily result in a fitness benefit to plants [6]: although plant volatiles may function as a “cue” to parasitoids , they may not be a specific “signal” released by the plant ( implying a selective benefit ) . Although short-term negative consequences of attracting hyperparasitoids for plants may be absent , as the hyperparasitoid is not affecting the direct benefit of reduced herbivory by parasitized caterpillars , the plant may be presented with a cost of reduced population size of its beneficial natural enemies when a next generation of herbivores arrives . Our results show that hyperparasitoids may parasitize up to 55% of the parasitoid offspring , therefore potentially playing a major role in parasitoid population dynamics . Furthermore , the parasitoid species studied here have been found to parasitize over 90% of the herbivores when parasitoids are at their peak abundance during the season [32] . The effect of parasitoid–hyperparasitoid interactions therefore may have significant consequences for herbivore populations , and thereby indirectly hyperparasitoids may significantly contribute to selection on plant traits such as volatile release . The fitness consequences of the emission of herbivore-induced plant volatiles are dependent on the quantitative composition of the plant-associated community at several trophic levels . The fitness benefit of volatile release should , therefore , be evaluated in the natural context of the plant-associated insect community including fourth-trophic-level organisms [6] . This will help to improve our understanding of the function of herbivore-induced volatiles in plants and of how the ecological effects of volatiles can shape the life histories of species interacting in insect communities associated with these plants [33] , [34] . Furthermore , these findings are important in the context of developing Integrated Pest Management strategies in which herbivore-induced volatiles of crops are manipulated to optimize the control of insect pests by using parasitoids . Overexpression of herbivore-induced plant volatiles in crops or field application of synthetic parasitoid attractants may not benefit pest control in conditions where the responses of hyperparasitoids to HIPVs cause major mortality to parasitoids [35] .
Brassica oleracea var gemmifera cv . Cyrus plants used for olfactometer experiments were grown in 1 . 45-l pots containing peat soil ( Lentse potgrond , no . 4 , Lent , the Netherlands ) and provided with SON-T light ( 500 µmol/m2/s; L16:D8 ) in addition to natural daylight in a glasshouse compartment ( 18–26°C , 50%–70% r . h . ) . When plants were 4 wk old , they were fertilized weekly by applying 100 ml nutrient solution of 2 . 5 mg/l Kristalon Blauw ( Hydro Agri Rotterdam , the Netherlands ( N-P-K-Mg ) 19-6-20-3 ) to the soil and used in experiments when they were 7 wk old . To prepare parasitized caterpillars for the induction treatments , individual first instar P . rapae larvae were exposed to a single female C . glomerata or C . rubecula , which were allowed to parasitize the caterpillar in a glass vial . For C . glomerata , caterpillars were considered to be parasitized when the wasp had inserted her ovipositor in the caterpillar for at least 5 s . For C . rubecula , because of herbivore immune responses to parasitoid eggs [34] , the wasp was allowed to oviposit 3 times in the same caterpillar , to increase the success rate of parasitism . Due to larval cannibalism among the parasitoids , only a single C . rubecula larva would develop eventually [36] . The hyperparasitoid L . nana was reared on C . glomerata cocoons in the absence of plant and herbivore-derived cues . All Brassica oleracea var gemmifera cv . Cyrus plants for the olfactometer assays were treated 24 h before the tests . First , plants were infested with either two unparasitized fourth instar Pieris rapae caterpillars or two fourth instar caterpillars that contained fully grown parasitoid larvae of either C . glomerata or C . rubecula as a result of parasitization of the caterpillar in their first instar . In a second experiment with oral secretions of caterpillars , plants were artificially damaged with a pattern wheel by drawing three lines of 3 cm long on each of the four youngest fully expanded leaves and treated with 25 µl of caterpillar oral secretions onto the damaged sites . Oral secretions were collected from healthy and C . glomerata–parasitized fourth instar P . rapae caterpillars , using 5 µl capillaries . Single caterpillars regurgitated 2–8 µl that we pooled to be used in the induction treatments . We decided not to test the relative attractiveness of plants induced with oral secretions of C . rubecula–parasitized caterpillars , because it would lack biological relevance as the data from our choice assays with actual feeding damage indicate that the quantity of damage by gregariously parasitized caterpillars is likely to explain the preference of hyperparasitoids for treatments with higher amounts of leaf damage ( Figure 4 ) . We have restricted to testing the effect of qualitative differences in the oral secretion of parasitized and unparasitized caterpillars only to treatment combinations where we did not identify a statistical difference in the amount of damage between the treatments . To test whether volatiles derived from oral secretion itself may be attractive to hyperparasitoids , we applied 25 µl oral secretion of C . glomerata–parasitized P . rapae caterpillars onto undamaged plants with a fine brush . We tested the relative attractiveness of the oral-secretion-treated plants to undamaged plants treated with 25 µl of water . Shortly before L . nana females were tested for their behavioural response to plant volatiles in Y-tube olfactometer bio-assays , we removed caterpillars and their feces from the plants and placed the plants in one of two glass jars ( 30 l each ) that were connected to the two olfactometer arms . A charcoal-filtered airflow ( 4 l/min ) was led through each arm of the Y-tube olfactometer system , and a single wasp was released at the base of the stem section ( 3 . 5 cm diameter , 22 cm length ) in each test [37] . Wasps that passed a set line at the end of one of the olfactometer arms within 10 min and stayed there for at least 15 s were considered to have chosen for the odour source connected to that olfactometer arm . To compensate for unforeseen asymmetry in the setup , we swapped the jars containing the plants after testing five wasps and replaced the set of plants by a new set of plants after testing 10 wasps . The Y-tube olfactometer setup was placed in a climatized room , and in addition to daylight it was illuminated with four fluorescent tube lights ( FTD 32 W/84 HF , Pope , the Netherlands ) . Eighty-four-week-old plants were transplanted into the field with 1×1 m spacing between plants and allowed to adjust to field conditions for 1 wk . Thereafter , the plants were subjected to one of four induction treatments: ( 1 ) not treated with herbivory ( i . e . , undamaged controls , UD ) , ( 2 ) infested individually with either two unparasitized first instar P . rapae caterpillars ( PR ) , ( 3 ) two C . glomerata–parasitized P . rapae caterpillars ( PR-CG ) , or ( 4 ) two C . rubecula–parasitized P . rapae caterpillars ( PR-CR ) . Unparasitized and parasitized caterpillars were allowed to feed on plants for 10 d , which was approximately the whole development period of the Cotesia larvae . Each plant was covered with a fine-mesh net when planted to avoid other herbivore infestations on the foliage and to prevent the herbivores used for induction to wander off the plant . To test the effects of plant induction with different types of herbivory on hyperparasitism , we attached parasitoid pupae onto the plants in the field . Individual pupae of C . rubecula , or clutches of C . glomerata , were first attached to a paper disc ( 3×3 cm ) with a small droplet of glue . We removed nets and caterpillars just before attaching the paper discs carrying the pupae with a pin needle . Half of the plants for each treatment received five C . glomerata clutches , and the other half received five C . rubecula pupae . The pupae were exposed to the natural community of hyperparasitoids and recollected after 5 d . They were kept separately in 2 . 2 ml Eppendorf tubes that were closed with cotton wool . The Eppendorf tubes were checked daily for emerging primary parasitoids and hyperparasitoids . All wasps were identified to species level . A completely randomized design was applied to the field studies . Five replications , each with 80 plants ( 10 replicates of each treatment ) , were carried out from June until October 2011 . To assess hyperparasitism rates and species communities on the solitary and gregarious primary parasitoid , we established plots of 6×6 m containing 49 plants of B . oleracea cultivars in an experimental field in the vicinity of Wageningen , the Netherlands , during 3 consecutive years ( 2005–2007 ) . Within plots , plants were planted in a square of 7×7 plants with a spacing of 75 cm between plants . Plots were isolated by strips of 6 m wide that were sown with a grass mixture of Lolium and Poa species . During the growth season of cabbage plants , from early May until the end of September , we conducted weekly surveys on the plants for C . glomerata and C . rubecula pupae by investigating both sides of all leaves of all plants in the plots . The pupae were collected and placed individually in 2 . 2 ml Eppendorf tubes and closed with cotton wool . The pupae with their external silk cocoon were weighed , and for the gregarious C . glomerata , the brood size was determined . The Eppendorf tubes were checked daily for emerging parasitoids , which were individually transferred to another Eppendorf tube and stored at −20°C . All wasps were identified to species level . L . nana preferences for herbivore-induced plant volatiles , as tested in two-choice Y-tube olfactometer assays , were analysed using two-tailed binomial tests . Hyperparasitoid preferences for plant volatiles induced by unparasitized P . rapae caterpillars and caterpillars parasitized by gregarious or solitary primary parasitoids under field conditions were analysed using two Generalized Linear Models ( GLMs ) . To analyse the effects of plant inductions with different types of herbivory on hyperparasitism at plant level , we modelled the dependent variable as a binomial occurrence of hyperparasitism per plant ( 400 plants equally divided over five replicates ) and scored presence of hyperparasitoids in pupae as 1 and absence as 0 . Additionally , to test the effects of the plant inductions on hyperparasitism at cocoon level , we modelled the dependent variable as the number of pupae or clutches giving any hyperparasitoid out of the fixed totals of five pupae attached to the plant . Into the two models we included the fixed factors caterpillar induction ( undamaged , unparasitized P . rapae , P . rapae parasitized by C . glomerata , and P . rapae parasitized by C . rubecula ) , replicate ( five replications ) , types of pupae ( gregarious or solitary ) , and the interactions between the three terms . For the field collections of solitary and gregarious pupae ( n = 1 , 668 and 1 , 256 , respectively ) , we analysed whether the gregarious broods of primary parasitoids were more frequently found by hyperparasitoids than solitary pupae and whether this results in differences in the total fraction of hyperparasitism of primary parasitoid offspring from gregarious and solitary species . First , we tested whether solitary and gregarious pupae of the primary parasitoids differed in the proportion of occasions that these were found by a hyperparasitoid . For each solitary parasitoid , we scored a 1 when there was a hyperparasitoid emerging and a 0 when the primary parasitoid emerged . Because gregarious broods could be hyperparasitized to different degrees ( percentage pupae parasitized ) , which may be a result of a single hyperparasitoid finding the gregarious brood and parasitizing several pupae or a result from multiple occasions on which hyperparasitoids found the gregarious brood , we scored a 1 when any hyperparasitoid emerged from the gregarious Cotesia brood . We used a Generalized Linear Model ( GLM ) to test for the effect of Cotesia species and year as well as their interaction on the binomially distributed occurrence of hyperparasitism . Second , we tested whether hyperparasitoids exert different levels of parasitism of gregarious and solitary primary parasitoids . Within each of the years , for each Cotesia species we counted the total number of emerging wasps for each of the primary parasitoid and hyperparasitoid species . The data on species composition and their abundance per Cotesia species were subjected to Chi-square tests to assess parasitoid community differences for primary parasitoid brood size ( solitary or gregarious ) . One of the hyperparasitoid species ( i . e . , Baryscapus galactopus ) parasitizes the larvae of Cotesia when these are still inside a caterpillar and lay several eggs within a single Cotesia larva . The B . galactopus brood develops when the Cotesia larvae spin their cocoon outside the caterpillar and B . galactopus wasps emerge from the Cotesia pupa . The brood size of B . galactopus on Cotesia is on average eight B . galactopus per Cotesia pupa , and therefore , we recalculated the total incidence of parasitization of Cotesia pupae by dividing the B . galactopus numbers by 8 and rounding off to the nearest whole number ( numbers of each parasitoid species are presented in Table 5 ) . All statistical tests were performed with the statistical software package Gen Stat ( 10th edition ) . We used Partial Least Squares Projection to Latent Structures-Discriminant Analysis ( PLS-DA ) to analyse which of the compounds contributed most to describing the difference among plant treatments . The compounds that scored >1 in their Variable Importance in the Projection ( VIP ) scores were subjected to Mann–Whitney U tests among treatment pairs to test for significant differences among treatments . | In nature , plants often release volatiles in response to damage by herbivores ( e . g . , by caterpillars ) , and these can indirectly help defend the plants . Indeed , it is well documented that volatiles can recruit the natural enemies of herbivores , such as predators and parasitoid wasps , whose offspring feed on and develop within their caterpillar hosts . However , such induced plant odours can also be detected by other organisms . One important group of organisms , hyperparasitoids , the enemies of the parasitoids that indirectly benefit the plants , have not been included in this trophic web because so little is known about their foraging behaviour . Here , using a combination of laboratory and field experiments , we demonstrate that hyperparasitoid wasps also take advantage of the odours that plants produce in response to the feeding by caterpillars . The larvae of parasitic wasps developing inside the caterpillar alter the composition of the oral secretions of their herbivorous host and thereby affect the cocktail of volatiles the plant produces . The hyperparasitoids on the lookout for their parasitoid prey can preferentially detect infected caterpillars , although not all parasitoid wasps gave away their presence through this host–plant interaction . We conclude that herbivore-induced plant volatiles can affect the interaction among parasitoids and their enemies and thereby may reduce the indirect defence accrued for the plant . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"community",
"ecology",
"chemical",
"ecology",
"ecology",
"entomology",
"biology",
"plant",
"ecology",
"zoology",
"terrestrial",
"ecology",
"behavioral",
"ecology"
] | 2012 | Hyperparasitoids Use Herbivore-Induced Plant Volatiles to Locate Their Parasitoid Host |
One goal of human genetics is to understand how the information for precise and dynamic gene expression programs is encoded in the genome . The interactions of transcription factors ( TFs ) with DNA regulatory elements clearly play an important role in determining gene expression outputs , yet the regulatory logic underlying functional transcription factor binding is poorly understood . Many studies have focused on characterizing the genomic locations of TF binding , yet it is unclear to what extent TF binding at any specific locus has functional consequences with respect to gene expression output . To evaluate the context of functional TF binding we knocked down 59 TFs and chromatin modifiers in one HapMap lymphoblastoid cell line . We then identified genes whose expression was affected by the knockdowns . We intersected the gene expression data with transcription factor binding data ( based on ChIP-seq and DNase-seq ) within 10 kb of the transcription start sites of expressed genes . This combination of data allowed us to infer functional TF binding . Using this approach , we found that only a small subset of genes bound by a factor were differentially expressed following the knockdown of that factor , suggesting that most interactions between TF and chromatin do not result in measurable changes in gene expression levels of putative target genes . We found that functional TF binding is enriched in regulatory elements that harbor a large number of TF binding sites , at sites with predicted higher binding affinity , and at sites that are enriched in genomic regions annotated as “active enhancers . ”
Understanding the regulatory logic of the genome is critical to understanding human biology . Ultimately , we aim to be able to predict the expression pattern of a gene based on its regulatory sequence alone . However , the regulatory code of the human genome is much more complicated than the triplet code of protein coding sequences , and is highly context-specific , depending on cell-type and other factors [1] . In addition , regulatory regions are not necessarily organized into discrete , easily identifiable regions of the genome and may exert their influence on genes over large genomic distances [2] . Consequently , the rules governing the sequence specificity as well as the functional output of even the most common regulatory interactions , such as interactions between transcription factors and the genome , are not yet fully understood . To date , genomic studies addressing questions of the regulatory logic of the human genome have largely taken one of two approaches . On the one hand are studies aimed at collecting transcription factor binding maps using techniques such as ChIP-seq and DNase-seq [3]–[6] . These studies are mainly focused on identifying the specific genomic locations and DNA sequences associated with transcription factor binding and histone modifications . On the other hand are studies aimed at mapping various quantitative trait loci ( QTL ) , such as gene expression levels ( eQTLs ) [7] , DNA methylation ( meQTLs ) [8] and chromatin accessibility ( dsQTLs ) [9] . These studies are mainly focused on identifying specific genetic variants that functionally impact gene regulation . Cumulatively , binding map studies and QTL map studies have led to many insights into the principles and mechanisms of gene regulation [7] , [10]–[12] . However , there are questions that neither mapping approach on its own is well equipped to address . One outstanding issue is the fraction of factor binding in the genome that is “functional” , which we define here to mean that disturbing the protein-DNA interaction leads to a measurable downstream effect on gene regulation . ( Note that we do not concern ourselves with the question of whether the regulatory outcome and/or the interaction are evolving under natural selection ) . An experimental technique that could help address this issue is transcription factor knockdown . In knockdown experiments , the RNA interference pathway is employed to greatly reduce the expression level of a specific target gene by using small interfering RNAs ( siRNAs ) . The cellular or organismal response to the knockdown can then be measured ( e . g . [13] ) . Instead of measuring a cellular phenotype , one can collect RNA after the knockdown and measure global changes in gene expression patterns after specifically attenuating the expression level of a given factor . Combining a TF knockdown approach with TF binding data can help us to distinguish functional binding from non-functional binding . This approach has previously been applied to the study of human TFs ( e . g . [14]–[16] ) , although for the most part studies have only focused on the regulatory relationship of a single factor with its downstream targets . The FANTOM consortium previously knocked down 52 different transcription factors in the THP-1 cell line [17] , an acute monocytic leukemia-derived cell line , and used a subset of these knockdowns to validate certain regulatory predictions based on binding motif enrichments [18] . However , the amount of transcription factor binding information available for the THP-1 cell line is limited ( it is not a part of the ENCODE reference lines ) . Many groups , including our own , have previously studied the regulatory architecture of gene expression in the model system of HapMap lymphoblastoid cell lines ( LCLs ) using both binding map strategies [3] , [19] , [20] and QTL mapping strategies [7] , [9] . As a complement to that work , we sought to use knockdown experiments targeting transcription factors in a HapMap LCL to refine our understanding of the gene regulatory circuitry of the human genome . We integrated the results of the knockdown experiments with previous data on transcription factor binding to better characterize the regulatory targets of 59 different factors and to learn when a disruption in transcription factor binding is most likely to be associated with variation in the expression level of a nearby gene .
Following normalization and quality control of the arrays , we identified genes that were differentially expressed between the three knockdown replicates of each factor and the six controls . Depending on the factor targeted , the knockdowns resulted in between 39 and 3 , 892 differentially expressed genes at an FDR of 5% ( Figure 1B; see Table S3 for a summary of the results ) . The knockdown efficiency for the 59 factors ranged from 50% to 90% ( based on qPCR; Table S1 ) . The qPCR measurements of the knockdown level were significantly correlated with estimates of the TF expression levels based on the microarray data ( P = 0 . 001; Figure 1C ) . Reassuringly , we did not observe a significant correlation between the knockdown efficiency of a given factor and the number of genes classified as differentially expressed following the knockdown experiment ( this was true whether we estimated the knockdown effect based on qPCR ( P = 0 . 10; Figure 1D ) or microarray ( P = 0 . 99; not shown ) data . Nor did we observe a correlation between variance in qPCR-estimated knockdown efficiency ( between replicates ) and the number of genes differentially expressed ( P = 0 . 94; Figure 1E ) . We noticed that the large variation in the number of differentially expressed genes extended even to knockdowns of factors from the same gene family . For example , knocking down IRF4 ( with a knockdown efficiency of 86% ) resulted in 3 , 892 differentially expressed genes ( including IRF4 ) , while knocking down IRF3 ( with a knockdown efficiency of 91% ) , a paralog of IRF4 [23] , only significantly affected the expression of 113 genes ( including IRF3 ) . Because we knocked down 59 different factors in this experiment we were able to assess general patterns associated with the perturbation of transcription factors beyond merely the number of affected target genes . Globally , despite the range in the number of genes we identified as differentially expressed in each knockdown , the effect sizes of the differences in expression were relatively modest and consistent in magnitude across all knockdowns . The median effect size for genes classified as differentially expressed at an FDR of 5% in any knockdown was a 9 . 2% difference in expression level between the controls and the knockdown ( Figure 2 ) , while the median effect size for any individual knockdown experiment ranged between 8 . 1% and 11 . 0% . To further evaluate the biological implications of our observations , we used the Gene Ontology ( GO ) [24] annotations to identify functional categories enriched among genes that were classified as differentially expressed following the knockdown experiments . In general , the differentially expressed genes tend to be annotated within pathways that fit well with what is already known about the biology of each of the 59 factors ( Table S4 ) . For example , differentially expressed genes following the knockdowns of both IRF4 ( 3 , 892 genes differentially expressed ) and IRF9 ( 243 genes differentially expressed ) are enriched for many immune response annotations . However , differentially expressed genes in the IRF4 knockdown are enriched for both type I and II interferon signaling pathways , among other pathways , consistent with the known role of IRF4 in immune responses [25] . Genes differentially expressed in the IRF9 knockdown are enriched for type I interferon responses ( among other pathways ) but not type II responses , which is again consistent with the known biology [26] . As another example , knocking down SREBF2 ( 1 , 286 genes differentially expressed ) , a key regulator of cholesterol homeostasis [27] , results in changes in the expression of genes that are significantly enriched for cholesterol and sterol biosynthesis annotations . While not all factors exhibited striking enrichments for relevant functional categories and pathways , the overall picture is that perturbations of many of the factors primarily affected pathways consistent with their known biology . In order to assess functional TF binding , we next incorporated binding maps together with the knockdown expression data . In particular , we combined binding data based on DNase-seq footprints in 70 HapMap LCLs , reported by Degner et al . [9] ( Table S5 ) and from ChIP-seq experiments in LCL GM12878 , published by ENCODE [3] . We were thus able to obtain genome-wide binding maps for a total of 131 factors that were either directly targeted by an siRNA in our experiment ( 29 factors ) or were differentially expressed in one of the knockdown experiments ( see Methods for more details ) . We classified a gene as a bound target of a particular factor when binding of that factor was inferred within 10kb of the transcription start site ( TSS ) of the target gene . Using this approach , we found that the 131 TFs were bound in proximity to a median of 1 , 922 genes per factor ( range 11 to 7 , 053 target genes; Figure S7A; only the 8 , 872 genes expressed in at least one knockdown experiment were considered for this analysis ) . Target genes were bound by a median of 34 different factors ( range 0 to 96; Figure S7B; only 288 genes expressed in our experiments were not classified as bound targets of any of the 131 factors considered ) . We considered binding of a factor to be functional if the target gene was differentially expressed after perturbing the expression level the bound transcription factor . We then asked about the concordance between the transcription factor binding data and the knockdown expression data . Specifically , we studied the extent to which differences in gene expression levels following the knockdowns might be predicted by binding of the transcription factors within the putative regulatory regions of the responsive genes . Likewise , we asked what proportion of putative target ( bound ) genes of a given TF were also differentially expressed following the knockdown of the factor . We performed this analysis in two stages . First , we only considered binding data for the specific TF that was knocked down in each experiment ( binding data was available for 29 different factors ) . In general , we found that the number of differentially expressed genes following the knockdowns was positively correlated with the number of bound target genes by these 29 factors ( Spearman's ρ = 0 . 45; permutation P = 0 . 015 ) . For 12 of the 29 knockdowns , we observed significant overlaps between binding and differential expression ( Fisher's exact test; P<0 . 05 ) . We also found that between 3 . 4–75 . 9% ( median = 32 . 3% ) of differentially expressed genes were bound by the TF in a given knockdown ( mean relative enrichment = 1 . 08 ) . Perhaps somewhat less expected , we found that between 46 . 4% and 99 . 1% ( median = 88 . 9% ) of the binding was apparently not functional , namely it was not associated with changes in gene expression levels . This observation is robust with respect to the size of the window we used to classify genes as bound by a factor ( range 1–20 kb from the TSS; Table S6 ) . It is also consistent with our previous findings that most DNase-I sensitive QTLs are not also classified as eQTLs [9] . We next considered the expression data in the context of the binding data for both the knocked down TFs and any other TF whose expression level was indirectly affected by the knockdown . We again examined the overlap between binding and differential expression ( Figure 3 ) . Considering only the expressed genes in each experiment , the fraction of genes bound ( by any TF whose expression has changed ) ranged between 16 . 2% and 95 . 3% with a median of 85 . 4% ( Figure 3B ) . However , the fraction of bound genes that were also differentially expressed in a given experiment was generally quite low ( median = 7 . 9%; Figure 3C; mean relative enrichment = 1 . 02 ) , with significant ( P<0 . 05 ) overlap between bound and differentially expressed genes seen for only 13 of the 59 knockdown experiments . Even if we relaxed the statistical threshold with which we classify genes as differentially expressed four-fold ( to an FDR of 20% ) , a majority of bound genes still failed to show significant evidence of differential expression ( median = 68 . 8%; Figure 3C; mean relative enrichment = 1 . 01 ) . The discrepancy in the number of genes bound by a particular factor ( or the TFs it regulates ) and the number of differentially expressed genes in a knockdown experiment begged the question of whether any characteristics of factor binding might distinguish functionally bound target genes . In order to address this question , we examined a variety of features . First , focusing only on the binding sites classified using the DNase-seq data ( which were assigned to a specific instance of the binding motif , unlike the ChIP data ) , we examined sequence features that might distinguish functional binding . In particular , we considered whether binding at conserved sites was more likely to be functional ( estimating conservation by using PhastCons 46 way placental scores [28] ) and we also considered whether binding sites that better matched the known PWM for the factor were more likely to be functional . Interestingly , we did not observe a significant shift in the conservation of functional binding sites ( Wilcoxon rank sum P = 0 . 34 ) , but we did observe that binding around differentially expressed genes occurred at sites that were significantly better matches to the canonical binding motif ( P<10−8 ) , although the absolute difference in PWM score was very small . Next , considering bound targets determined from either the ChIP-seq or DNase-seq data , we observed that differentially expressed genes were associated with both a higher number of binding events for the relevant factors within 10 kb of the TSS ( P<10−16; Figure 4A ) as well as with a larger number of different binding factors ( considering the siRNA-targeted factor and any TFs that were DE in the knockdown; P<10−16; Figure 4B ) . We hoped to distinguish between coordinated co-regulation of the factors and generally higher levels of binding nearby differentially expressed genes . To do so , we asked whether the genes differentially expressed in common between any two knockdown experiments were more likely to be co-occupied by the same transcription factors ( considering only transcription factors whose expression was affected by the knockdown ) . Binning all pairwise comparisons between knockdown experiments based on the fraction of differentially expressed transcription factors in common , we observed that enrichment for functional co-occupancy increased proportionally to the fraction of TFs in common ( Figure 4C ) . This suggests that co-regulation is at least partially responsible for the increased numbers of factors binding near differentially expressed genes . We proceeded by examining the distribution of binding about the TSS . Most factor binding was concentrated near the TSS whether or not the genes were classified as differentially expressed ( Figure 5A ) . However , surprisingly , the distance from the TSS to the binding sites was significantly longer for differentially expressed genes ( P<10−16; Fig . 5B ) . We then investigated the distribution of factor binding across various chromatin states , as defined by Ernst et al . [11] . This dataset lists regions of the genome that have been assigned to different activity states based on ChIP-seq data for various histone modifications and CTCF binding . For each knockdown , we separated binding events by the genomic state in which they occurred and then tested whether binding in that state was enriched around differentially expressed genes . After correcting for multiple testing , 19 knockdowns showed significant enrichment for binding in “strong enhancers” around genes that were differentially expressed and four knockdowns had significant enrichments for “weak enhancers” . Further , eight knockdowns showed significant depletion of binding in “active promoters” of genes that were differentially expressed and six knockdowns had significant depletions for “transcription elongation” . Finally , we asked whether the factors tended to have a consistent effect ( either up- or down-regulation ) on the expression levels of genes they purportedly regulated . Perhaps surprisingly , all factors we tested are associated with both up- and down-regulation of downstream targets ( Figure 6 ) . A slight majority of downstream target genes were expressed at higher levels following the knockdown for 15 of the 29 factors for which we had binding information ( Figure 6B ) . The factor that is associated with the largest fraction ( 68 . 8% ) of up-regulated target genes following the knockdown is EZH2 , the enzymatic component of the Polycomb group complex . On the other end of the spectrum was JUND , a member of the AP-1 complex , for which 66 . 7% of differentially expressed targets were down-regulated following the knockdown . The remaining 27 factors ( with a median of 170 direct targets ) all show a more even balance between up- and down-regulated targets . These trends are consistent when we considered all genes that were differentially expressed following a knockdown ( not just the genes that were also bound by the knocked down factor ) . We observed that an average of 51 . 9% of downstream differentially expressed genes had elevated expression levels following knockdown of the transcription factors . Furthermore , for 39 of the experiments a slight majority of differentially expressed genes following knockdown of the factor were up-regulated , while in only 17 of the experiments were the majority of differentially expressed genes down-regulated following the knockdown .
Nearly all expressed genes in the LCL we worked with are bound within 10 kb of their TSS by at least one of the 131 TFs for which we were able to obtain binding data . Yet , the regulation of the vast majority of target genes is not affected by perturbations to the expression levels of the TFs . Our observations suggest that it may ultimately be possible to predict functional transcription factor binding based on the biological context , yet since bound genes are only modestly enriched among those that are differentially expressed , an effective classifier may be difficult to develop . In that context , several of the associations we observed might seem counterintuitive from a purely biochemical perspective but are consistent with our definition of functional TF binding as participating in gene regulation . In particular , from a biochemical perspective , binding at stronger motifs might be expected to be less affected by a decrease in factor concentration in the cell ( following the knockdown ) and regulatory regions with more binding sites and a larger number of bound factors might be expected to be less influenced by the perturbation of one single factor . Yet , we observed the opposite patterns: Functional binding is associated with stronger binding motifs and greater levels of factor binding near differentially expressed genes . Viewed from an evolutionary rather than purely biochemical perspective , these observations are quite in tune with expectations . In other words , genomic regions of functional importance evolve to ensure factor binding . This can be accomplished by selection for different properties , including increased affinity of the binding site and more cooperative binding . Our results also indicate that binding in the context of certain chromatin states was more likely to be functional . For 19 of the 56 factors that we knocked down and for which we were able to obtain binding data ( on either direct or indirect bound targets ) , there was a significant enrichment of binding in “strong enhancers” near differentially expressed genes , which is consistent with our observations that functional binding occurs further from the TSS ( namely , not in promoter regions ) . While further experiments are required before we can put forward a more concrete explanation , these observations suggest that binding at active promoters may be buffered against acute changes in transcription factor concentration . This may also explain why most of the effect sizes associated with differences in gene expression levels following the knockdowns were relatively modest . While there is compelling evidence for our inferences , the current chromatin functional annotations do not fully explain the regulatory effects of the knockdown experiments . For example , the enrichments for binding in “strong enhancer” regions of the genome range from 7 . 2% to 50 . 1% ( median = 19 . 2% ) , much beyond what is expected by chance alone , but far from accounting for all functional binding . Previous studies in model organisms suggested redundancy as a partial explanation for the common observation of non-functional TF binding . For example , Gitter et al . examined similar themes using gene expression data from yeast in which transcription factors had been knocked out [30] . They found that factors without identifiable paralogs were more likely to have a regulatory affect on bound targets than factors with identifiable paralogs , which can potentially have overlapping functions . We considered a similar explanation for our observations . To do so , we used Ensembl's definition of paralogs [31] to stratify the 29 factors that we knocked down , and for which we also had direct binding data , by the number of identifiable paralogs each had ( range 1–5 ) . We found no correlation between the number of paralogs and the fraction of bound targets that were differentially expressed . We also did not observe a significant correlation when we considered whether the percent identity of the closest paralog might be predicative of the fraction of bound genes that were differentially expressed following the knockdown ( Figure S8 ) . In addition to considering the distinguishing characteristics of functional binding , we also examined the direction of effect that perturbing a transcription factor had on the expression level of its direct targets . We specifically addressed whether knocking down a particular factor tended to drive expression of its putatively direct ( namely , bound ) targets up or down , which can be used to infer that the factor represses or activates the target , respectively . Transcription factors have traditionally been thought of primarily as activators [11] , [32] , [33] , and previous work from our group is consistent with that notion [9] . Surprisingly , the most straightforward inference from the present study is that many of the factors function as repressors at least as often as they function as activators . For example , we inferred that EZH2 had a negative regulatory relationship with the largest fraction of direct targets ( 68 . 8% ) , while JUND seemed to have a positive regulatory relationship with the largest fraction of direct targets ( 66 . 7% ) . These particular observations seem consistent with the known role of EZH2 as the active member of the Polycomb group complex PC2 [34] and the biochemical characterization of the AP-1 complex ( of which JUND is a component ) as a transactivator [35] . More generally , however , our results , combined with the previous work from our group and others ( e . g . [9] , [11] ) make for a complicated view of the role of transcription factors in gene regulation as it seems difficult to reconcile the inference from previous work that many transcription factors should primarily act as activators with the results presented here . One somewhat complicated hypothesis , which nevertheless can resolve the apparent discrepancy , is that the “repressive” effects we observe for known activators may be at sites in which the activator is acting as a weak enhancer of transcription and that reducing the cellular concentration of the factor releases the regulatory region to binding by an alternative , stronger activator . We believe that this may alleviate the apparently contradictory interpretations of transcription factor activity , although additional work on this topic is needed . There are important caveats that should be kept in mind when interpreting our results . Care must be taken in interpreting the lack of evidence for gene expression differences in our system as previous studies have suggested that the specific effects of knockdowns may be difficult to detect because of redundancy in gene regulatory mechanisms [36] , [37] . In addition , our definition of target genes , which is based on genomic proximity , is certain to be somewhat noisy . For some genes , we are undoubtedly assigning irrelevant binding while for others we are missing true regulatory interactions that may occur further than 10 kb from the transcription start site . Moreover , we only considered binding and expression in one set of conditions ( one cell type at one timepoint ) . Some binding events may require co-factors that are only present after specific stimuli in order to become functional or may otherwise be functional in other contexts . Because of all of this , we are likely underestimating the fraction of bound genes that are functionally regulated . We have taken measures to control for this as much as possible . First , integrating our results with binding data allows us to focus our analysis on the most likely direct targets . Second , the consistency of our results across a variety of factors ( including factors with very different numbers of differentially expressed genes following the knockdown ) , suggests that technical explanations for our observations are unlikely . To more explicitly address the effect that our proximity-based definition of target genes might have on our analyses , we reanalyzed the overlap between factor binding and differential expression following the knockdowns using an independent , empirically determined set of target genes . Specifically , Thurman et al . used correlations in DNase hypersensitivity between intergenic hypersensitive sites and promoter hypersensitive sites across diverse tissues in order to assign intergenic regulatory regions to specific genes , independently of proximity to a particular promoter [38] . We thus performed an alternative analysis in which we assigned binding events to genes based on the classification of Thurman et al . We then considered the overlap between binding and differential expression in this new dataset . The results were largely consistent with our proximity-based observations . A median of 9 . 5% of genes that were bound by a factor were also differentially expressed following the knockdown of that factor ( compared to 11 . 1% when the assignment of binding sites to genes is based on proximity ) . From the opposite perspective , a median of 28 . 0% of differentially expressed genes were bound by that factor ( compared to 32 . 3% for the proximity-based definition ) . The results of this analysis are summarized in Table S7 . Another practical limitation of our study design is the choice of microarrays instead of a more advanced technology such as RNA-seq . As a result of this choice , the expression of only the genes represented by probes on the arrays could be measured . In addition , alternative splicing events and novel transcription were not measured in our experiments . Furthermore , microarrays are known to have sensitivity issues at the extreme ranges of expression levels ( although this will only affect a small subset of transcripts ) [39] . Because of this , our results should not be considered a comprehensive census of regulatory events in the human genome . Instead , we adopted a gene-centric approach , focusing only on binding events near the genes for which we could measure expression to learn some of the principles of functional transcription factor binding . In light of our observations a reassessment of our estimates of binding may be warranted . In particular , because functional binding is skewed away from promoters ( our system is apparently not well-suited to observe functional promoter binding , perhaps because of protection by large protein complexes ) , a more conservative estimate of the fraction of binding that is indeed functional would not consider data within the promoter . Importantly , excluding the putative promoter region from our analysis ( i . e . only considering a window >1 kb from the TSS and <10 kb from the TSS ) does not change our conclusions . Considering this smaller window , a median of 67 . 0% of expressed genes are still classified as bound by either the knocked down transcription factor or a downstream factors that is differentially expressed in each experiment , yet a median of only 8 . 1% of the bound genes are also differentially expressed after the knockdowns . Further work in this field is clearly justified . Much of what distinguishes functional binding ( as we define it ) has yet to be explained . Furthermore , we are unable to explain much of the differential expression observed in our experiments by the presence of least one relevant binding event . This may not be altogether surprising , as we are only considering binding in a limited window around the transcription start site . To address these issues , more factors should be perturbed to further evaluate the robustness of our results and to add insight . Together , such studies will help us develop a more sophisticated understanding of functional transcription factor binding in particular , the gene regulatory logic more generally .
The cell line ( GM19238 ) was cultured at 5% CO2 and 37°C in RPMI 1640 medium supplemented with 2 mM L-glutamine and 15% fetal bovine serum , per Coriell's recommendations ( http://ccr . coriell . org/Sections/Support/Global/Lymphoblastoid . aspx ? PgId=213 ) . The medium was also supplemented with 100 IU/ml penicillin and 100 µg/ml streptomycin . Cells were counted and split three times a week to 350 , 000 cells/ml . Transfections were performed with the Lonza 96-well nucleofector system , using transfection solution SF and transfection program DN-100 . On-Target SmartPool siRNAs from Dharmacon were used to knockdown target genes . For each transfection , one million cells were transfected with 50 pmol siRNA . After transfection , cells were plated in 1 ml of medium in 96-well plates and incubated for 48 . At the 48 hour time point , 500 µl of cell culture were removed for RNA extraction ( for qPCR ) , 100 µl of fresh medium was added to each well and plates were incubated for an additional 24 hours . At the 72 hour time point , the remaining culture was pelleted for RNA extraction ( for array hybridization ) . RNA from all timepoints was extracted using the RNeay Plus 96 kit ( Qiagen ) . The control siRNAs consisted of a pool of four siRNAs specifically designed not to target any human or mouse gene . In addition to two replicates of the negative control pool , we also transfected each of the negative control siRNAs independently for a total of six negative control samples . We learned from an initial pilot experiment that there were strong batch effects between rounds of transfection and so we included negative control transfection in parallel with all three batches of transfection . The transfections were conducted in two phases . For the first phase , we screened siRNAs for their knockdown efficiency under our experimental conditions by transfecting cells and extracting RNA 48 hours later . qPCR was performed with SYBR Green and custom primers ( Table S1 ) . Knockdown efficiency was assessed relative to a sample transfected with the negative control siRNA pool . We used a relative quantification approach , referencing the POLR2C control gene and using the DART-PCR method [40] to determine PCR efficiency . For the second phase , those siRNA transfections resulting in ≥50% knockdown ( arbitrary cutoff ) were transfected again and RNA was extracted at both 48 hours ( to confirm the knockdown level by qPCR ) and 72 hours ( for hybridization to microarrays ) . Two genes ( NFYC and ZHX2 ) were not knocked down by 50% in the screen but were included in the final transfection where they did reach 50% ( in all three replicates ) and were therefore included in the microarray experiment . Additionally , 16 factors that passed the screen were not knocked down by 50% at 48 hours in all three replicates in the final transfection and were not included in the microarray experiment ( Table S1 ) . The quality and concentration of all RNA samples were measured using the Agilent 2100 Bioanalyzer . ARNTL2 , BATF , BCL3 , CEBPG , CEBPZ , CLOCK , CREBBP , DIP2B , E2F1 , E2F4 , E2F6 , EP300 , ESRRA , EZH2 , FOXA3 , GTF2B , HCST , HOXB7 , IKZF3 , IRF3 , IRF4 , IRF5 , IRF7 , IRF8 , IRF9 , JUND , KLF13 , LCORL , NFE2L1 , NFKB2 , NFX1 , NFYC , NR1D2 , NR2F6 , NR3C1 , PAX5 , POU2F1 , POU2F2 , RAD21 , RDBP , RELA , RELB , RXRA , SKIL , SP1 , SP3 , SREBF2 , STAT2 , STAT6 , TAF1 , TCF12 , TFDP1 , TFDP2 , TFE3 , USF1 , WHSC1 , YY1 , ZBTB38 , ZHX2 . Samples were hybridized to Illumina HT-12 v4R2 arrays in two batches . For the first batch , 69 samples ( 150 ng of total RNA ) , representing 21 knockdowns and one set of controls , were sent to the UCLA Southern California Genotyping Consortium where the RNA was converted to cRNA and then hybridized to arrays using the standard protocol . For the second batch , 132 samples ( 1 µg of total RNA ) , representing 40 knockdowns and two sets of controls , were sent to the University of Chicago Functional Genomics Core where the RNA was converted to cRNA and then hybridized to the arrays using the standard protocol . Both cores returned raw probe intensities to us . The batch aspect of the study design is clearly less than optimal , but it was a result of necessary practical considerations . As mentioned above , we included a full set of negative controls with each batch . We also accounted for batch effects explicitly in our analysis ( see below ) . Before processing the arrays , we determined the set of usable probes . To do this , we first mapped probes to the hg19 reference genome using the BWA alignment program [41] and only retained probes that mapped perfectly to the genome ( probes spanning introns were discarded ) . We then removed probes that mapped perfectly to a single site in the genome and also mapped to a second site allowing for one mismatch . As GM19238 was derived from a female donor , we excluded all probes for Y chromosome genes . We then removed probes that contained a SNP for which GM19238 was heterozygous based on 1000 Genomes data [42] ( http://ftp . 1000genomes . ebi . ac . uk/vol1/ftp/pilot_data/paper_data_sets/a_map_of_human_variation/trio/snps/ ) to avoid detecting spurious results from an interaction between allele specific expression and probe effects influencing hybridization intensities . After this series of exclusions , for each gene that was assayed by more than one remaining probe , we chose the 3′-most probe to represent the gene . We then excluded probes for genes that were not expressed ( detection P<0 . 01 ) in the knockdown triplicates or the full set of controls ( 18 arrays ) for each experiment . We log-transformed and quantile normalized the data from all arrays together using the ‘lumi’ Bioconductor package [43] , [44] in the R statistical environment . Our initial analysis indicated that this was not sufficient to correct for all of the batch effects present between different rounds of transfection ( Figure S5 ) , so we also used the RUV-2 method [45] to further adjust the data . To do so , we defined a set of genes that should not be differentially expressed under any conditions in our experiment as the basis for the RUV-2 correction . While the choice of unaffected genes might not be straightforward in all experimental settings , the fact that we had 59 different knockdown experiments ( and controls ) across three rounds of transfections provided an opportunity to define a reasonable set of control probes . Specifically , we created a list of the 2 , 000 least variable probes ( considering only probes for genes that passed our expression threshold ) for each of the batches of arrays separately . We then intersected the three lists to obtain a high confidence list of 787 invariant genes . These are the genes with the least evidence of difference in expression levels in any knockdown experiment . We also needed to specify the number of components to regress out . Again , this choice may not always be straightforward , however , we had two knockdown experiments ( IRF5 and SP1 ) that were each repeated across different batches . We thus maximized the correlation between these experiments in deciding on the number of components to remove . We found that removing eight components resulted in zero genes identified as differentially expressed ( at an FDR of 5% ) between replicate experiments ( or the control arrays by themselves , across batches ) and had the maximum number of differentially expressed genes replicated between the two experiments for both IRF5 and SP1 ( Figure S2; these models were fit with the ‘limma’ package [46] ) . Lastly , we averaged each of the negative control knockdowns across each of the time points in order to have a single set of controls with which to compare each of the knockdowns . In addition to examining heatmaps for the arrays ( Figures S1 , S3 ) , the quality of each array was assessed by relative log expression ( ‘RLE’ ) plots ( Figure S4; [47] , [48] ) and PCA ( Figure S5 , Table S2 ) . For the remainder of the analyses , we only used the IRF5 and SP1 experiment that had the greater knockdown efficiency . We note that the top principle components of the adjusted data are still correlated with the microarray chip ( Table S2 ) , but we decided that further correction was not helpful for several reasons . First , the amount of variance captured by any one component is quite low after RUV-2 adjustment ( Table S2 ) . Second , our analysis indicated that removing additional components with RUV-2 reduced the correlations between replicate knockdown experiments ( Figure S2 ) . Third , we randomized samples across chips so as to avoid the introduction of a bias in our results based on such an effect . In order to identify differentially expressed genes in each knockdown , we compared the expression level of each gene on the three knockdown arrays to its expression level on the six control arrays . We used a likelihood-ratio test within the framework of a fixed-effect linear model:Here , Yij is the expression level of gene i on array j . μi is the mean expression level for gene i . βj is the status of array j ( either “knockdown” or “control” ) and Xi is the knockdown effect for gene i . εij is the error term for gene i on array j . For each gene , we compared a model with a βjXi term to the nested model with no βjXi term using a likelihood ratio test to determine which model fit the data better . To adjust for multiple testing within each knockdown experiment , we calculated the q-value for each gene using the Storey and Tibshirani method [49] as provided in the ‘qvalues’ R package . For each knockdown , we also assessed whether there were any Gene Ontology categories ( either from the “MF” or “BP” categories ) that were overrepresented among the differentially expressed genes . To do so , we used the ‘topGO’ package and the ‘org . Hs . eg’ database in R . For each knockdown , we tested both “MF” and “BP” categories , combined the results , and filtered out any categories without a single gene differentially expressed . P-values were adjusted for multiple testing using the ‘BH’ method in the ‘p . adjust’ function in R . After identifying differentially expressed genes , we intersected the gene expression data with factor binding data from both DNase-seq experiments and ChIP-seq experiments . We considered binding data within a fixed window around the TSS of each gene . To determine a single TSS for each gene , we used the ENCODE CAGE data downloaded from UCSC [3] , [50] . For each gene , we chose the TSS with the highest CAGE score as the reference TSS for that gene unless there was a tie between multiple TSSs , in which case we used the midpoint between these TSSs as the reference TSS . The DNase data was from a previous study conducted in our lab [9] . Binding was determined using the Centipede algorithm [4] on DNase-seq data from 70 Yoruba HapMap cell lines . For each factor expressed in our experiments , we classified all sites with a Centipede posterior probability greater than 0 . 95 as bound . The binding was originally mapped to the hg18 reference genome , so we used liftOver ( http://hgdownload . cse . ucsc . edu/admin/exe/ ) to convert the coordinates to h19 . For the ChIP data , we downloaded all ENCODE ChIP-seq data for GM12878 called using the SPP peak caller , except for the POL2 datasets and an NFKB dataset that was collected following tumor necrosis factor-α stimulation [3] ( http://ftp . ebi . ac . uk/pub/databases/ensembl/encode/integration_data_jan2011/byDataType/peaks/jan2011/spp/optimal/hub/ ) . For each gene that we were able to link to a ChIP dataset or a TRANSFAC binding motif [51] , we combined all binding records and then considered the union of that set as the binding profile for the factor . After obtaining the union set , we calculated the midpoint for each discrete binding record and used the midpoints as the estimated binding location in all subsequent analyses . Using this approach , we obtained binding data for 201 factors; 138 of these factors were represented by a usable probe on the array and were expressed in at least one of the knockdown experiments . 131 of the factors were differentially expressed in at least one knockdown experiment . For DNase-based binding sites , we also evaluated the PhastCons alignment score [28] . PhastCons 46 way placental wig files were downloaded from UCSC and the average score for each DNase-seq binding site was calculated . For all of the comparisons between functional and non-functional binding , we used the Wilcoxon rank sum test to assess differences . We also downloaded the Ernst chromatin states [11] from UCSC . This file contained 15 different chromatin states , including two separate categories each for “strong enhancer” , “weak enhancer” , and “repetitive” . We combined each of the replicate states into a single category so that we ended up with 12 distinct chromatin states . To identify which state the binding occurred in , we intersected the binding record midpoints with the chromatin states . We calculated the enrichment or depletion of functional binding in specific chromatin states using a Fisher's Exact Test . To test whether the number of paralogs or the degree of similarity with the closest paralog for each transcription factor knocked down might influence the number of genes differentially expressed in our experiments , we obtained definitions of paralogy and the calculations of percent identity for 29 different factors from Ensembl's BioMart ( http://useast . ensembl . org/biomart/martview/ ) [31] . We used genome build GRCh37 . p13 . For each gene , we counted the number of paralogs classified as a “within_species_paralog” . After selecting only genes considered a “within_species_paralog” , we also assigned the maximum percent identity as the closest paralog . To evaluate the effect that an independent assignment of target genes to regulatory regions might have on our analyses , we used the definition of target genes defined by Thurman et al . ( ftp://ftp . ebi . ac . uk/pub/databases/ensembl/encode/integration_data_jan2011/byDataType/openchrom/jan2011/dhs_gene_connectivity/genomewideCorrs_above0 . 7_promoterPlusMinus500kb_withGeneNames_32celltypeCategories . bed8 . gz ) , which use correlations in DNase hypersensitivity between distal and proximal regulatory regions across different cell types to link distal elements to putative target genes [38] . We intersected the midpoints of our called binding events ( defined above ) with these regulatory elements in order to assign our binding events to specific target genes and then re-analyzed the overlap between binding and differential expression in our experiments . All analyses were performed with a combination of BedTools [52] , [53] and BEDOPS [54] commands , along with custom Python and R scripts . The knockdown data have been deposited in NCBI's Gene Expression Omnibus ( [55]; accessible through GEO Series accession number GSE50588 at http://www . ncbi . nlm . nih . gov/geo/query/acc . cgi ? acc=GSE50588 ) | An important question in genomics is to understand how a class of proteins called “transcription factors” controls the expression level of other genes in the genome in a cell-type-specific manner – a process that is essential to human development . One major approach to this problem is to study where these transcription factors bind in the genome , but this does not tell us about the effect of that binding on gene expression levels and it is generally accepted that much of the binding does not strongly influence gene expression . To address this issue , we artificially reduced the concentration of 59 different transcription factors in the cell and then examined which genes were impacted by the reduced transcription factor level . Our results implicate some attributes that might influence what binding is functional , but they also suggest that a simple model of functional vs . non-functional binding may not suffice . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"genome",
"expression",
"analysis",
"genomics",
"functional",
"genomics",
"rna",
"interference",
"gene",
"expression",
"genetics",
"gene",
"regulation",
"molecular",
"genetics",
"biology",
"human",
"genetics",
"gene",
"networks"
] | 2014 | The Functional Consequences of Variation in Transcription Factor Binding |
While many disease-associated single nucleotide polymorphisms ( SNPs ) are associated with gene expression ( expression quantitative trait loci , eQTLs ) , a large proportion of complex disease genome-wide association study ( GWAS ) variants are of unknown function . Some of these SNPs may contribute to disease by regulating gene splicing . Here , we investigate whether SNPs that are associated with alternative splicing ( splice QTL or sQTL ) can identify novel functions for existing GWAS variants or suggest new associated variants in chronic obstructive pulmonary disease ( COPD ) . RNA sequencing was performed on whole blood from 376 subjects from the COPDGene Study . Using linear models , we identified 561 , 060 unique sQTL SNPs associated with 30 , 333 splice sites corresponding to 6 , 419 unique genes . Similarly , 708 , 928 unique eQTL SNPs involving 15 , 913 genes were detected at 10% FDR . While there is overlap between sQTLs and eQTLs , 55 . 3% of sQTLs are not eQTLs . Co-localization analysis revealed that 7 out of 21 loci associated with COPD ( p<1x10−6 ) in a published GWAS have at least one shared causal variant between the GWAS and sQTL studies . Among the genes identified to have splice sites associated with top GWAS SNPs was FBXO38 , in which a novel exon was discovered to be protective against COPD . Importantly , the sQTL in this locus was validated by qPCR in both blood and lung tissue , demonstrating that splice variants relevant to lung tissue can be identified in blood . Other identified genes included CDK11A and SULT1A2 . Overall , these data indicate that analysis of alternative splicing can provide novel insights into disease mechanisms . In particular , we demonstrated that SNPs in a known COPD GWAS locus on chromosome 5q32 influence alternative splicing in the gene FBXO38 .
Chronic obstructive pulmonary disease ( COPD ) is characterized by irreversible airflow obstruction . While cigarette smoking is the leading environmental risk factor for COPD , only a subset of smokers develop the disease . Genetic factors have been shown to contribute to COPD susceptibility , with the best characterized example being SERPINA1 , the causal gene for α1-antitrypsin deficiency [1–5] . Genome-wide association studies ( GWAS ) have been used to identify common genetic variants associated with many complex diseases , including COPD . These studies have identified multiple well-replicated genome-wide significant loci , including a locus on 15q25 ( CHRNA3/CHRNA5/IREB2 ) , FAM13A , HHIP , CYP2A6 and HTR4 [6–10] . In addition , a recent large meta-analysis identified twenty-two genome-wide significant loci in COPD , of which 13 were newly genome-wide significant [11 , 12] . However , the majority of single nucleotide polymorphisms ( SNPs ) that have been identified in COPD GWAS are of unknown function . Since most GWAS variants are located in noncoding regions , it is possible that these variants could contribute to COPD susceptibility through transcriptional regulation of target genes , amongst other possible mechanisms . Expression quantitative trait locus ( eQTL ) studies have been used to identify SNPs that contribute to gene expression levels , thereby providing insight into the biological mechanism responsible , and providing additional confidence in GWAS findings by implicating putative causative genes . In addition , methods such as TWAS and PrediXcan , which utilize eQTL data to analyze genetic regulation of gene expression , have replicated GWAS findings [13] . To date , only a moderate percentage of GWAS findings have been shown to be eQTLs with strong effect size [14] . This is likely due to a variety of mechanisms including inadequate sample size , multiple testing correction , tissue type investigated , as well as the focus on total gene-level expression levels without consideration of transcript isoforms . COPD has been shown to have a disproportionate amount of alternative splicing compared to other complex diseases such as type 2 diabetes , Alzheimer’s disease , and Parkinson’s disease , suggesting that transcriptional regulation through splicing may play an important role [15] . Genetic variation could contribute by altering mRNA splicing , which in turn could result in changes in protein sequence or expression levels . Several recent studies have demonstrated that SNPs associated with alternative splicing ( sQTLs ) are enriched for GWAS variants . Evidence suggests that at least 20–30% of disease causing mutations may affect pre-mRNA splicing [16 , 17] . Furthermore , several reports have discovered that there are a proportion of GWAS SNPs that have evidence of sQTLs but not eQTLs , indicating that sQTL analysis can provide additional insight into the functional mechanisms underlying GWAS results [18] [19 , 20] . Here , we characterize sQTLs in human peripheral blood in COPD to determine whether these loci can identify novel functions for COPD GWAS variants . We hypothesize that a substantial fraction of COPD GWAS loci influence disease susceptibility through sQTLs . Previous studies to identify sQTLs have been performed in small sample sizes or have focused on exon expression instead of differential exon usage [18] . This study characterizes variants associated with differential exon usage in a large population .
This study has been approved by Partners Healthcare IRB ( Protocol # 2007P000554 ) . All subjects provided written informed consent . This study included 376 non-Hispanic white subjects from the COPDGene study . COPDGene enrolled individuals between the ages of 45 and 80 years with a minimum of 10 pack-years of lifetime smoking history from 21 centers across the United States [21] . These subjects returned for a second study visit 5 years after the initial visit at which time they completed additional questionnaires , pre-and post-bronchodilator spirometry , computed tomography of the chest , and provided blood for complete blood counts ( CBCs ) and RNA sequencing . In this study , moderate to severe COPD was defined as GOLD spirometric grades 2–4 [22] . The protocol for RNASeq data generation has been previously described [23] . Total RNA was extracted from PAXgene Blood RNA tubes using the Qiagen PreAnalytiX PAXgene Blood miRNA Kit ( Qiagen , Valencia , CA ) . Extracted samples with a RIN > 7 and concentration ≥ 25ug/uL were included in sequencing . Globin reduction , ribosomal RNA depletion and cDNA library prep was performed using the TruSeq Stranded Total RNA with Ribo-Zero Globin kit ( Illumina , Inc . , San Diego , CA ) . The Illumina HiSeq 2500 was used to generate 75 bp reads , and an average of 20 million reads were generated per sample . Reads were trimmed using skewer [24] to remove specified TruSeq adapter sequences . Quality control was performed using the FASTQC [25] and RNA-SeQC [26] programs . Trimmed reads were aligned to the GRCh37 reference genome using a two-pass alignment method with STAR 2 . 5 [27] . Following sequence alignment , mappability filtering to correct for allelic bias in read mapping was performed using WASP [28] . An average of 840 , 000 reads were removed due to read mapping bias resulting in an average of 19 . 9 million reads being available for subsequent analysis . Genotyping was performed by Illumina ( San Diego , CA ) on the HumanOmniExpress Array . Eagle v . 2 . 3 was used for phasing and HRC reference panel version 1 . 1 was used for imputation . SNPs with minor allele frequency greater than 0 . 05 and imputation R2 greater than 0 . 5 were included in analysis . Details on genotyping QC and imputation have been previously published . [9 , 11] . Gene expression counts were computed using Rsubread [29] . Quantification of splicing ratios was performed using Leafcutter [19] . This method extracts junctional reads ( or reads that span introns ) from aligned bam files and clusters them according to shared start or stop positions . Default leafcutter parameters were used in the detection of clusters , i . e . , 50 split reads across all individuals were required to support each cluster , and introns up to 500 kb were included . For sQTL analysis , intron ratios were calculated by determining how many reads support a given exon-intron junction in relation to the number of reads in that region . Introns used in less than 40% of individuals were filtered out , and the remaining intron ratios were used as input for sQTL analysis . MatrixeQTL [30] was used to test for association between genotype of all SNPs within 1000 kb of a gene ( cis- ) and quantifications of gene expression or alternative splicing using linear models , adjusting for age , gender , pack-years of smoking , current smoking status , white blood cell differential , PEER factors of expression data and five principal components of genetic ancestry . Calculation of principal components has been previously described [8] . Out of a total of 5 , 405 , 234 SNPs passing QC and filtering , 4 , 664 , 386 were within 1000kb of a gene and were tested for association with 25 , 313 genes and 97 , 365 splice sites . COPD-associated SNPs were obtained from a subset of a published GWAS [11] . We selected 920 SNPs with p<1 x 10−6 in white subjects to match the ethnicity of the RNA-Seq data . These SNPs were grouped into 21 loci based on their genomic positions ( S1 Fig; Table 1 ) . Specifically , SNPs with positions located within the window shown in Table 1 were grouped into the corresponding locus . SNPs that were associated with splicing at the 10% FDR , or with gene expression at the 10% FDR were identified . The 10% FDR threshold was selected based on published sQTL studies [20 , 31] . Co-localization analysis was performed for GWAS loci that contained sQTLs using eCAVIAR [32] . All SNP-cluster and SNP-gene pairs within these loci with FDR<0 . 1 were included in the colocalization analysis . Thirty COPDGene blood samples were selected for qPCR based on expression levels in RNASeq data . An additional ninety resected lung tissue samples from individuals undergoing thoracic surgery [33] were selected based on genotype . A total of 400 ng of RNA was reverse transcribed using the SuperScript III First-Strand Synthesis System ( ThermoFisher Scientific , Waltham , MA ) for blood RNA or the iScript cDNA Synthesis Kit ( BioRad , Hercules , CA ) for lung RNA . A Taqman assay was designed to amplify cDNA fragments from the 3’ region of the cryptic FBX038 exon , to the 5’ region of exon 10 ( S2 Fig ) . A predesigned Taqman assay ( Hs01004563_mH ) was used to amplify the alternate isoforms from the 3’ region of exon 9 to the 5’ region of exon 10 . A final amount of 30 ng of cDNA was amplified per well , and each sample was assayed in triplicate . GAPDH was amplified as a housekeeping gene to control for RNA concentration . To calculate the ratio of transcripts containing the novel exon , delta CT values for the novel isoform were divided by delta CT values for the alternate isoform . Linear regression was performed to test for an additive relationship between genotype and splicing ratio . Cells were lysed in EBC buffer ( 50 mMTris pH 7 . 5 , 120 mMNaCl , 0 . 5% NP-40 ) supplemented with protease inhibitors ( Complete Mini , Roche ) and phosphatase inhibitors ( phosphatase inhibitor cocktail set I and II , Calbiochem ) . The protein concentrations of lysates were measured by the Beckman Coulter DU-800 spectrophotometer using the Bio-Rad protein assay reagent . Same amounts of whole cell lysates were resolved by SDS-PAGE and immunoblotted with indicated antibodies . For immunoprecipitation , 1000 μg whole cell lysates were incubated with the indicated anti-Flag M2 affinity gel and monoclonal anti-HA agarose for 3–4 hr at 4 degrees Celcius ( Millipore Sigma ) . Immunoprecipitants were washed five times with NETN buffer ( 20 mMTris , pH 8 . 0 , 100 mMNaCl , 1 mM EDTA and 0 . 5% NP-40 ) before being resolved by SDS-PAGE and immunoblotted with indicated antibodies .
This analysis included 376 Non-Hispanic white COPDGene participants ( S1 Table ) . Samples were sequenced to a depth of approximately 20 million reads per sample , and splice ratios from a total of 97 , 365 splice clusters ( defined as overlapping introns that share a spice donor or acceptor site ) were included in the sQTL analysis . Gene expression counts for 25 , 312 genes were used for eQTL detection . We identified 1 , 706 , 704 cis-sQTLs at 10% FDR , comprising 561 , 060 unique SNPs ( Table 2 , S2 Table ) . These SNPs were associated with 30 , 333 splice sites which were annotated to 6742 unique genes ( S3 Table ) . Similarly , we identified 1 , 242 , 993 cis-eQTLs corresponding to 708 , 928 unique SNPs . These SNPs were associated with expression of 15 , 913 genes . We found that 44 . 6% of sQTLs were also eQTLs , but that 55 . 3% ( 310 , 361 SNPs ) were sQTLs exclusively . In addition , 2299 genes contained at least one splice site that was significantly associated with genotype of a neighboring SNP , while total gene expression of the same gene was not significantly associated with any SNP . This suggests that analysis of sQTLs can identify novel regulatory events that are not captured through whole gene expression analysis . To characterize the biological functions of genes for which alternative splicing was associated with nearby SNPs , Sigora [34] was used to identify overrepresented pathways in genes that had sQTLs but not eQLTs . This method of pathway analysis focuses on genes or gene pairs that are specific to a single pathway . In this way it utilizes the status of other genes in the experimental context to identify the most relevant pathways and minimize the identification of spurious pathways . We identified 2299 genes which had significant sQTLs but not eQTLs at the 10% FDR . These genes were enriched for 33 KEGG pathways and 70 Reactome pathways ( S4 and S5 Tables ) , mostly related to RNA processing . Both eQTLs and sQTLs were categorized on the basis of their location relative to the gene with which they were associated , and the results are shown in Table 2 . The greatest difference in distribution of sQTLs and eQTLs was at intergenic ( 38% of sQTLs and 41% of eQTLs; p<2 . 2x10−16 ) and intronic ( 50% of sQTLs and 46% of eQTLs; p<2 . 2x10−16 ) sites . Only a small number of sQTLs ( n = 68 ) and eQTLs ( n = 74 ) were located in splice sites , defined as intronic and within 2 bp of an exon/intron boundary . In addition , we identified 7 sQTLs and 3 eQTLs located within an exon and 2 bp of an exon/intron boundary . sQTLs were at least as enriched among low p-value associations with COPD case status as eQTLs ( S3 Fig ) . To identify whether sQTLs can identify functions for GWAS SNPs that could not be explained by eQTLs alone , 920 SNPs associated with COPD in white subjects with a p <1 x 10−6 were grouped into 21 genetic loci according to genomic position ( Table 1 ) . These SNPs were then interrogated in the eQTL and sQTL data sets to identify which GWAS SNPs were also associated with alternative splicing or gene expression at 10% FDR . Of these 920 SNPs , 67 SNPs were both sQTLs and eQTLs , 71 SNPs were eQTLs alone , and 156 were sQTLs alone , indicating that a greater number of GWAS SNPs are sQTLs than eQTLs ( Fisher’s Exact Test P-value = 5 . 31x10−6 ) ( S6 and S7 Tables ) . Out of the 21 genomic loci , 6 included GWAS-associated SNPs that were eQTLs , and 7 included GWAS SNPs that were sQTLs ( Table 1 ) . In addition there were three loci that contained sQTLs but not eQTLs for any gene . To investigate whether GWAS associations may be attributed to alternative splicing , colocalization analysis was performed using eCAVIAR in the seven genetic loci containing both GWAS SNPs and sQTLs ( Table 3 ) . All significant ( FDR<0 . 1 ) SNP-gene and SNP-cluster pairs were included in this analysis . All seven of the GWAS loci had at least one variant with significant colocalization to sQTL data ( colocalization posterior probability [CLPP] > 0 . 01 ) ( top results are shown in Table 3 and all CLPP scores > 0 . 01 are shown in S8 Table ) . The locus with the strongest colocalization between either GWAS & sQTL or GWAS & eQTL was selected for in depth follow up; this was the 5q32 region containing HTR4/FBXO38 . This region was identified in the COPD case control GWAS , with 151 SNPs with p-value < 1x10−6; rs3995091 SNP located in the HTR4 gene had the lowest p-value ( 2 . 59 x 10−14 ) . Additional significant variants from an independent association in this region include rs7730971 and rs4597955 ( Fig 1a ) with GWAS p-values of 5 . 51x10−12 and 4 . 78 x 10−12 , respectively . rs7730971 was significantly associated with splice sites in the FBXO38 gene ( S4 Fig ) , while this SNP was not associated with total expression of any gene . eCAVIAR analysis revealed that rs7730971 colocalized with the sQTL and GWAS data ( CLPP = 0 . 845 ) ( Fig 1b ) , while there was no colocalization with eQTLs for FBXO38 , suggesting that the GWAS association may be caused be alternative splicing . Furthermore , eCAVIAR identified rs7730971 to be the SNP with the highest degree of colocalization between sQTL and GWAS data , suggesting that this may be the causative variant ( Table 3 and S8 Table ) . Despite being the GWAS SNP with the minimal p-value in the locus , there was no colocalization between sQTL and GWAS data for rs3995091 ( CLPP = 0 . 002 ) . Characterization of the splicing cluster associated with genotype of rs7730971 revealed a previously unannotated cryptic exon located at chromosome 16: 147 , 790 , 643–147 , 790 , 801 . This 158 bp exon is present in a greater proportion of subjects with the GG genotype ( 13% ) than the CC genotype ( 8% ) ( Fig 1c ) . The CC genotype is also associated with greater risk of COPD , so the novel isoform may be protective against COPD . Since this splice site has not been previously documented , quantitative PCR was performed to independently validate the existence of the novel exon as well as replicate the effect of genotype on exon inclusion levels in RNA from whole blood ( linear regression p = 0 . 01 , Fig 1d ) . Furthermore , the novel exon was identified in RNA from homogenized lung tissue , where splicing levels were also associated with rs7730971 genotype ( linear regression p = 0 . 007 , Fig 1d ) . The cryptic exon leads to a premature stop codon in FBXO38 , which could alter or inhibit protein function . Additionally , immunoprecipitation was performed in 293T cells which indicated that FBXO38 , an F-box protein with unknown substrates , interacts with Cullin 1 , but not other Cullin members ( S5 Fig ) , indicating that it may be a component of a SKP1-Cullin-1-F-box ( SCF ) type of E3 ubiquitin ligase complex . In combination , these findings suggest that the GWAS association at 5q32 may be partly explained by the inclusion of an exon which results in a truncated FBXO38 protein .
eQTL studies can provide insight into the biological mechanisms responsible for disease associations . While many disease-associated SNPs are eQTLs , a large proportion of GWAS variants are of unknown function . SNPs that are associated with transcript isoform variation may contribute to disease risk and explain additional GWAS associations . In this study , we characterized eQTLs and sQTLs in human peripheral blood to identify novel functions for COPD GWAS associations . Among the genes identified to have splice sites associated with GWAS SNPs were CDK11A , which may be of biological relevance for COPD due to its role in apoptosis; SULT1A2 which has both eQTLs and sQTLs which strongly colocalize to the GWAS signal; and of particular interest , FBXO38 , in which a novel exon may be protective against COPD . Although we found a larger number of eQTLs than sQTLs genome wide ( 708 , 928 vs 561 , 060 unique SNPs ) , a greater number of COPD GWAS SNPs were sQTLs ( 156/920 ) than eQTLs ( 89/920 ) . This phenomenon has previously been shown in multiple sclerosis , where GWAS SNPs were more highly enriched among sQTLs than eQTLs [19] . Furthermore , several studies have shown that sQTL analyses can uncover functions for GWAS-associated polymorphisms that would not have been identified through eQTL analysis alone . Zhang et al . found that 4 . 5% of GWAS SNPs from the GWAS catalog had evidence of cis-sQTLs but not cis-eQTLs [18] . Li et al . showed that sQTLs identified in lymphoblastoid cell lines are enriched among autoimmune-disease associated variants [20] . Another study demonstrated [19] that there were similar levels of enrichment of both eQTLs and sQTLs among SNPs associated with rheumatoid arthritis . In addition , Li et al . showed that in an analysis of rheumatoid arthritis , the inclusion of intronic splicing data allowed for the identification of 18 putative disease genes , of which 13 would not have been associated on the basis of gene-expression level measurements alone [19] . Twelve to 36% of the 156 GWAS eQTLs that were identified in this study have been found in the large datasets produced by GTEX and eQTLgen ( S6 Table ) , indicating that some of our data can be applied to the general population and are not COPD specific . We identified 156 GWAS SNPs that were significant in the sQTL study , and these SNPs were grouped into 7 loci based on SNP position ( S1 Fig ) . Each of the seven loci had at least one SNP which colocalized with GWAS data . Of these loci , three could only be explained by sQTLs and not eQTLs . Of particular interest was the association at 5q32 in which SNPs in HTR4 were associated with COPD case-control status . This is a well-replicated GWAS association for lung function [35–37] , COPD [11 , 38] and airflow obstruction in smokers [10] . Despite the consistent genetic association , as well as prior studies identifying HTR4 expression in developing lung and increased airways resistance in a murine model , the mechanism by which the specific SNPs contribute to COPD risk is unknown , and additional genes in the region have not been investigated . Our analysis revealed that these SNPs may contribute to COPD by regulating splicing of a neighboring gene , FBXO38 . This is consistent with evidence that the nearest gene to an associated SNP is the causative gene in only a minority of cases [39 , 40] . Therefore , analyses such as eQTL and sQTL studies are critical to uncover a relationship between an implicated SNP and the gene responsible for the association . The characterization of splicing in F-box protein 38 ( FBXO38 ) resulted in the discovery of a cryptic exon which has not been previously annotated . Through qPCR we were able to validate the existence of the exon as well as replicate the association with genotype in both blood and lung tissue . This is of particular importance as it demonstrates that analysis of splicing in the blood can uncover sQTLs that are also relevant to the lung . Sequence analysis using ORF Finder [41] determined that this newly identified transcript encodes a premature stop codon in the cryptic exon . This will result in a truncated protein that could have altered structure and function . To date , relatively little is known about the biological processes as well as the downstream signaling pathways through which FBXO38 operates to possibly influence COPD . Furthermore , no known ubiquitin substrate has been identified for FBXO38 , which makes FBXO38 one of the orphan F-box proteins . Here we identified Cullin-1 as a binding partner for FBXO38 . Therefore , the biological role of FBXO38 may rely mostly on its biochemical feature as an E3 ligase , through the formation of a complex with Cullin-1 and Skp1 , similar to its family members Skp2 and Fbw7 [42 , 43] . FBXO38 is additionally known to co-activate the transcriptional regulator Kruppel Like Factor 7 ( KLF7 ) [43 , 44] . Members of the KLF gene family regulate cell proliferation , differentiation and survival , and have been found to play a role in airway inflammation [45] . In particular , KLF7 is involved in regulating epithelial cell differentiation and epithelial-mesenchymal transition [46 , 47] , and possibly in airway remodeling [48] . Another gene of interest is SULT1A2 , located at 16p11 . 2 . In this locus , SNPs were associated with both gene expression and splicing of SULT1A2 . Therefore , two lines of evidence implicated SULT1A2 as the causative gene . SULT1A2 is a sulfotransferase enzyme , which is a family of phase II liver enzymes that detoxify a variety of endogenous and xenobiotic compounds [49] . SULT1A2 can sulfonate hormones like estrogens and androgens , but of particular interest for COPD , sulfation through SULT1A2 is a pathway for the metabolism of cigarette smoke compounds [50] . The 16p11 . 2 region contains a large inversion spanning ~0 . 45 MB . This region encompasses the entire SULT1A2 gene as well as the CCDC101 gene in which the associated SNPs are located , and is near other genes such as TUFM . This region contains SNPs which are associated with both obesity and body mass index ( BMI ) [51 , 52] asthma [53] , and autoimmune diseases like diabetes and inflammatory bowel disease; the inversion allele itself has been shown to be protective against the joint occurrence of asthma and obesity[54] . There are 24 polymorphisms that have been shown accurately tag the inversion [54] . The rs4788084 SNP which was identified in our study to be associated with both splicing and gene expression is among these markers , and is in LD with the inversion ( R2 = 0 . 982 ) . However , since both the SNP and the associated splice sites are located within the inversion , it is unlikely that inversion alters or contributes to this regulation of splicing . Finally , we identified the CDK11A gene as a novel COPD candidate gene in the 1p36 . 32 locus . Here , SNPs were associated with the occurrence of a single exon in the CDK11A gene . CDK11A encodes a member of the serine/threonine protein kinase family . This kinase can be cleaved by caspases and may play a role in cell apoptosis [55 , 56] , which is a key dysregulated pathway in COPD [57 , 58] . As this locus did not reach genome-wide significance in the larger GWAS , additional studies will be needed to confirm the phenotypic association . Most polymorphisms that were associated with splicing were acting from a distance , and were not located within or close to the splice donor or acceptor site . The majority of sQTLs ( 88% ) were located in intronic or intergenic regions , and only 0 . 01% percent were located within splice sites . However , we found that 12% of all tested SNPs were sQTLs , while 31% of splice variant SNPs were associated with splicing , indicating that there is enrichment of sQTLs in splice sites ( Fisher exact test p-value = 6 . 1 x 10−11 ) , although only a small proportion of tested SNPs ( 241/4 , 664 , 386; or 0 . 005% ) were located in splice sites . This is consistent with what has been shown in the literature . Kurmangaliyev et al . [59] found that 2 , 259 out of 37 , 852 , 169 SNPs ( 0 . 006% ) from 1000 Genomes are located in splice sites , and Yang et al . [60] found that 1 , 576 out of 14 , 718 , 752 SNPs ( 0 . 01% ) from dbSNP build 129 are located in splice sites . Furthermore , it has been found that the SNPs within splice sites tend to have little effect on gene function [59] , suggesting that splice sites are well conserved and the majority of splice regulation is not through splice site mutations . In our study , out of the seven GWAS loci that were discovered to have sQTLs , the most likely causative SNP was located greater than 500bp from the splice site in all cases . The mechanism by which these SNPs act on splicing is likely complex , but may involve auxiliary splicing regulatory elements such as exonic splicing enhancers ( ESEs ) , intronic splicing enhancers ( ISEs ) , exonic splicing silencers ( ESSs ) and intronic splicing silencers ( ISSs ) . These regulatory elements control splicing through the recruitment of trans-acting factors that interact with other regulatory factors or core spliceosome components [61 , 62] . A potential limitation of this study is that our eQTLs and sQTLs were identified in whole blood samples . COPD is a respiratory disease , and the most relevant cell-types in which to study gene expression changes may be located in the lung . However , due to the strong inflammatory component of COPD , characterization of gene expression and splicing in immune cells also has biological relevance . In addition , there is a high proportion of sQTL sharing across tissues [63] , which has been recently supported by the finding that 75–93% of sQTLs are replicated across tissue pairs from the GTEx consortium , with the estimated level of sharing between whole blood and lung being 92% [19] . Therefore , it is likely that the majority of the sQTLs identified in peripheral blood are also sQTLs in lung tissue . Furthermore , we experimentally demonstrated that the association with FBXO38 splicing is also present in lung tissue , and therefore the mechanism discovered in whole blood is likely to also be of relevance in the lung . Another potential limitation is that peripheral blood samples contain a mixture of cell types , any of which could contribute to gene expression signals . We included white blood cell percentages as a covariate in eQTL and sQTL analyses to limit potential confounding by differences in cell proportions . An additional limitation of this study is the depth of sequencing of approximately 20 million reads per sample . This read depth was selected to maximize sequencing value within a large genetic epidemiology study . In order to be able capture the effect of genotype on gene expression and splicing , a large sample size was required , and therefore there was a tradeoff to be made between sequencing depth and sample size . While this could lead to the failure to capture rare splicing events , the goal of the study was to capture common splicing variations associated with genotype , which this design allowed us to do . Furthermore , it is important to note that the eCAVIAR calculation of CLPP is dependent on sample size as well as effect size of the eQTL/sQTL . As previously shown [32] , an increase in sample size , or an increase in effect size results in higher CLPP values , and thus there is more power to detect colocalization with a larger sample size , or with eQTLs/sQTLs of strong effect . This means that while we may have been able to detect additional colocalization with a larger sample size , this does not impact our confidence in the CLPP values we have calculated here . This study was restricted to non-Hispanic white subjects , however RNA sequencing in additional subjects including those with African American ancestry is currently underway , and thus future studies will have sufficient sample size to perform analyses in these individuals . In conclusion , we found that many SNPs were associated with alternative splicing in peripheral blood . More COPD-associated variants were sQTLs than eQTLs , and we identified variant associations with splice sites in three genes including FBXO38 , an orphan F-box protein , which have a function role in COPD through an effect as an E3 ligase on a currently unknown substrate . These data indicate that analysis of alternative splicing may provide novel insights into disease mechanisms . | While it is known that chronic obstructive pulmonary disease ( COPD ) is caused in part by genetic factors , few studies have identified specific causative genes . Genetic variants that alter the expression levels of genes have explained part of the genetic component of COPD , however , there are additional genetic variants with unknown function . In some genes the protein coding sequence can be altered by a mechanism known as RNA splicing . We hypothesized that some genetic variants that are associated with risk of COPD contribute to the disease by altering RNA splicing . In this study , we identified genetic variants that are associated both with COPD risk and RNA splicing . In particular , we found that a COPD associated variant of previously unknown function may contribute to the inclusion of a new exon in the FBXO38 gene . These finding are significant because they indicate that analysis of RNA splicing can help identify genes that contribute to disease . | [
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] | [
"genome-wide",
"association",
"studies",
"medicine",
"and",
"health",
"sciences",
"body",
"fluids",
"gene",
"regulation",
"pulmonology",
"alternative",
"splicing",
"chronic",
"obstructive",
"pulmonary",
"disease",
"genome",
"analysis",
"gene",
"expression",
"genetic",
"... | 2019 | Analysis of genetically driven alternative splicing identifies FBXO38 as a novel COPD susceptibility gene |
Cytomegalovirus ( CMV ) is frequently transmitted by solid organ transplantation and is associated with graft failure . By forming the boundary between circulation and organ parenchyma , endothelial cells ( EC ) are suited for bidirectional virus spread from and to the transplant . We applied Cre/loxP-mediated green-fluorescence-tagging of EC-derived murine CMV ( MCMV ) to quantify the role of infected EC in transplantation-associated CMV dissemination in the mouse model . Both EC- and non-EC-derived virus originating from infected Tie2-cre+ heart and kidney transplants were readily transmitted to MCMV-naïve recipients by primary viremia . In contrast , when a Tie2-cre+ transplant was infected by primary viremia in an infected recipient , the recombined EC-derived virus poorly spread to recipient tissues . Similarly , in reverse direction , EC-derived virus from infected Tie2-cre+ recipient tissues poorly spread to the transplant . These data contradict any privileged role of EC in CMV dissemination and challenge an indiscriminate applicability of the primary and secondary viremia concept of virus dissemination .
Human Cytomegalovirus ( HCMV ) , a member of the betaherpesvirus subfamily , represents an important opportunistic viral pathogen in the immune compromised host . Fetuses , AIDS patients , and recipients of both bone marrow and solid organ transplants are at high risk for the development of debilitating and potentially life-threatening CMV disease . Depending on the risk constellation and immunosuppressive regimen , CMV disease can occur in up to 60% of heart or kidney transplant recipients . Therefore , HCMV is the most important viral pathogen especially during the first six months after transplantation [1] , [2] . The large variety of symptoms results from the broad cell and organ tropism of the virus [3] , [4] . In addition , the virus is able to disseminate via blood [5] . According to Fenner ( 1949 ) a virus enters - after initial replication at the entry site ( epithelia or transplant ) - the blood stream and disseminates throughout the body to distal organs via a so-called primary viremia , which was confirmed to apply also to HCMV and MCMV [6] , [7] . It is proposed that progeny virus from such organs can re-enter the blood circulation leading to a secondary viremia [6] , [7] thus increasing the risk for widespread dissemination . Leukocyte depletion of blood products derived from seropositive donors prior to transfusion efficiently prevents transfer of CMV to seronegative recipients [8] , [9] indicating that virus present in blood is predominantly cell associated . The cell types responsible for this dissemination are of particular interest . Three kinds of cells have been suggested to be involved in virus dissemination via blood . All of them have been shown to be able to transfer infectious virus ex vivo: polymorphonuclear leukocytes ( PMNL ) , monocytes/macrophages , and detached infected vascular endothelial cells ( EC ) . Although PMNL are thought to be only abortively infected , they might still function as vehicles for infectious virus [10] . Circulating infected monocytes become permissive upon differentiation into tissue macrophages and may then release infectious progeny within target organs [11] . For example , rat CMV was transferred via in vitro infected granulocytes or monocytes [12] . Vascular EC are suggested to play an important role in CMV dissemination , and unique genetic features govern the CMV - EC interaction [13] . EC support productive infection and may detach upon infection thus serving as shuttles for the virus to other organs via the blood stream [14] , [15] , [16] . EC are permissive for HCMV in vitro [3] and are commonly found to be infected in tissue samples from both immune compromised patients [17] and mice [18] . In addition , EC support latent infection with the potential to reactivate CMV [19] and to start a new episode of infection . Notably , HCMV infection is a risk factor for restenosis after coronary atherectomy [20] and accelerates atherosclerosis following cardiac transplantation [21] . The anatomical position of EC lining blood vessels implies a bidirectional role in virus entry into and exit from the blood circulation and therefore might define the ability of viruses in general to disseminate via blood . In fact , HCMV-infected EC can protrude from the wall into the lumen of the blood vessels in patients with active cytomegalovirus infection [16] . Furthermore , circulating giant endothelial cells were found in blood samples of transplant patients [14] suggesting detachment of infected EC from the vessel wall and dissemination of HCMV via EC throughout the body . Despite the undisputed and unique potential of EC in CMV infection and pathogenesis , it is still unknown whether infected EC are responsible for systemic virus dissemination during primary infection , contribute to this process , or merely represent an epiphenomenon with no causal involvement in the pathogenesis of organ disease [22] . Quantitative aspects of the contribution of infected EC to virus dissemination in the transplant situation are scarce and the presence of infected EC in the circulating blood does not prove that infected EC or HCMV produced by EC contribute or even govern virus dissemination from one site or organ to another . To quantify and to address the fate of virus produced by specific cells , we developed a Cre/loxP mediated approach to label virus in defined cell types in vivo and then trace the viral progeny of that cell type [23] , [24] . Cre recombinase recognizes two adjacent loxP sites and deletes the intervening DNA sequence . This reaction can remove a transcriptional stop signal between promoter and coding sequence resulting in gene expression . To study the role of EC in MCMV replication an MCMV mutant was used that contains a Cre-inducible egfp expression cassette ( MCMV-flox ) . Mice expressing Cre recombinase under control of either the Tie2 or the Tek promoter , which is selectively expressed in vascular EC ( Tie2-cre and Tek-cre mice ) , were infected with MCMV-flox . In this in vivo infection model MCMV-flox is efficiently recombined resulting in MCMV-rec only during virus replication in EC . It is important to note that Cre-mediated recombination of MCMV-flox is equally efficient in Tie2- and Tek-cre mice and only mediated by EC - as shown using bone marrow chimeras - thus providing highly concordant results by both mouse strains [24] . The resulting recombination is then stably maintained in the viral genome of the virus progeny . Vascular EC are present in all organs . A way to study the role of EC in virus dissemination from one organ to another is to either introduce organs from an EC cre-negative donor mouse into an EC cre-positive host or vice versa . Here , we investigated export of EC-derived virus from heart and kidney transplants to recipients as well as import of EC-derived virus from recipients into heart transplants . This was achieved by counting and comparing the contribution of EGFP-positive EC-derived progeny to the total virus load of organs and tissues . EC-derived virus from infected heart or kidney transplants readily disseminated to organs of MCMV-naïve recipients . The bulk of virus produced in and disseminated from heart is EC-derived , whereas in kidneys infected EC only provide a minor contribution . Yet , we found no evidence for any preferential dissemination of EC-derived virus from both types of transplants to other organs . The heart transplant was also tested as a target organ of EC-derived virus produced in recipient tissues . To our surprise , in contrast to the strong dissemination of virus originating from an infected transplant there was only minimal seeding of host EC-derived virus progeny to the transplant . Interestingly , this was independent of whether transplantation was performed prior to or after systemic host infection . In summary , our data argue against a privileged role of EC in virus dissemination .
Transplantation of organs from HCMV seropositive donors to seronegative recipients ( D+/R- ) is a known situation in transplantation medicine and represents the “high risk constellation” because up to 60% of the recipients can develop CMV disease [25] . In this D+R- setting CMV disease is caused by dissemination of HCMV from the transplanted organ to the recipient causing systemic symptoms with multiple organs being involved . The cellular source of disseminated virus has not been addressed , yet virus dissemination from infected heart transplants has also been described in the mouse model [26] , [27] . To investigate whether and to which extent virus derived from EC of the transplant disseminates to organs of uninfected recipients , hearts from acutely infected Tie2-cre mice were transplanted heterotopically into non-infected syngeneic C57BL/6 mice ( Fig . 1A ) . Four days after transplantation mice were sacrificed and organs collected to determine the amounts of non-recombined ( non-EC-derived ) and recombined ( EC-derived ) virus . In the heart transplant , high virus loads ( ∼105 PFU/g organ ) of predominantly recombined virus ( ∼85% ) were observed , confirming that the transplantation procedure itself did not affect MCMV replication in general and demonstrating a very high recombination efficiency ( Fig . 1B ) . This is in accordance with high recombination efficiency observed previously for heart and lungs of Tie2-cre mice [24] . Virus titers in different organs of mice infected via the heart transplant were 10 to 10 , 000-fold lower than generally seen following systemic ( i . v . ) infection with ∼1×106 PFU [24] . The relative amounts of MCMV-rec and MCMV-flox in the recipient organs , however , essentially reflected the situation in the heart transplant , with some minor variance . Thus , EC-derived virus virtually disseminated equally well as non-EC-derived virus from the heart transplant . Next , we studied dissemination of MCMV following kidney transplantation . Kidneys represent the majority of transplanted organs in medicine . Similar to heart transplantation , the transplantation of kidneys from seropositive donors to seronegative recipients is associated with a high risk to develop CMV-related complications [28] , [29] , [30] . Four days after heterotopic transplantation of infected kidneys of Tie2-cre mice into non-infected C57BL/6 mice , recipient organs were analyzed for the presence of disseminated virus ( Fig . 2A ) . In contrast to the heart , only about 20% of virus within transplanted kidneys was recombined ( Fig . 2B ) . This low contribution of EC-derived virus to total virus load in kidney is in line with previous observations [24] . As recombination rates were similar in Tie2-cre and Tek-cre mice , we believe that the low proportion of MCMV-rec in Tie2-cre kidneys does not necessarily indicate a low recombination efficiency in renal EC but may rather result from an alternative mode of virus entry into kidney tissue bypassing the vascular endothelium for replication in other cell types , one candidate being kidney epithelial cells . The relative levels of virus titers in liver , spleen , and lungs were comparable to those observed following heart transplantation . Yet , the percentage of recombined , EC-derived virus in most organs essentially mirrored the situation in the transplanted kidney , and there was no preferential dissemination of EC-derived virus ( ∼20% ) . Collectively , the findings after transplantation of two different organs did not support the hypothesis of a predominant role of EC in virus dissemination during the first four days of infection . Only in blood a significantly higher proportion of MCMV-rec was found on day four post kidney transplantation in 3 out of 4 mice . However , as the absolute virus titers were close to the detection limit , any interpretation has to be seen with caution . In the preceding experiments , the systemic infection originated from a pre-infected transplanted organ . Next , we studied the contribution of EC of a transplanted Tie2-cre+ heart to virus dissemination during the situation of systemic infection of C57BL/6 recipients . Under these conditions , all organs , including the transplanted heart , become infected simultaneously . Thus , MCMV-rec , wherever found , must have originated from ECs of the transplant . Note that under such conditions the infection of the transplant does not have a head start . Four days after transplantation mice were systemically ( i . v . ) infected with MCMV-flox and four days later they were sacrificed and virus titers determined ( Fig . 3A ) . As expected , the majority of virus in the transgenic heart transplant was found to be recombined ( Fig . 3B ) . Despite this , we observed only very little dissemination of EGFP+ EC-derived MCMV from the transplant to infected recipient organs . In the lungs , some MCMV-rec was found at very low numbers , four orders of magnitude lower than MCMV-flox , whereas in other organs MCMV-rec was at or below detection limit . This result is in stark contrast to the dissemination from the infected transplant ( Fig . 1 ) where 70-90% of virus progeny in recipient organs were EC-derived . It is important to note that total virus titers in both the endogenous and transplanted heart were very similar ( Fig . 3B ) , indicating efficient vascularization of the heterotopic heart transplant after surgery . This excludes an impaired blood flow as a presumed reason for the observed poor dissemination of recombined virus . We thus conclude that virus dissemination from the heart plays a negligible role during systemic infection . Another striking difference between virus dissemination from both transplanted heart and kidney as compared to systemic ( i . v . ) infection was the extent of virus production in different organs . In contrast to i . v . infection , which resulted in peak titers in the lung and high titers in heart , kidney , liver , spleen and adipose tissue ( Fig . 3 ) , virus dissemination from transplanted heart and kidney ( Figs . 1 and 2 ) resulted in peak titers in spleen but significantly lower titers in liver and lung , and in almost no virus detectable in the endogenous heart , kidneys , and adipose tissue . This cannot simply be explained by organ specific differences in virus production kinetics [24] but rather indicates a qualitative difference in virus dissemination between systemic ( i . v . ) infection ( free virus ) and transplant-mediated infection . During systemic infection following transplantation of a cre-positive heart to a cre-negative mouse no significant contribution of virus dissemination from the heart transplant to other organs was observed . To study not only cardiac EC but EC in general as a source of virus dissemination , Tie2-cre or Tek-cre recipient mice received a non-transgenic heart . Recipients were then infected i . v . with MCMV-flox ( Fig . 4A ) . As expected , the host organs showed the previously described organ-specific contribution of EC to total virus load [24] . Specifically , in liver the bulk of virus is derived from hepatocytes as we recently showed using Alb-cre mice selectively expressing Cre recombinase in hepatocytes [24] , whereas the cell type producing the bulk of virus in kidney remains to be determined . In all other organs , >60% of virus proved to be EC-derived . Yet , although the transplanted heart contained a total virus load comparable to that of the endogenous heart , there was only a minute ( about 1% ) contribution of recombined virus to the amount of virus in the transplant ( Fig . 4B ) . We repeated the experiments in Tek-cre mice , another mouse line transgenic for cre in EC , and obtained essentially the same results ( Fig . 4C ) . To confirm that this small contribution of MCMV-rec to the infection of a heart transplant was truly due to virus seeding to the organ and not just reflected virus present in the circulation , organ perfusion was performed in order to flush out blood cells prior to analysis ( Fig . 4C ) . In any case , the data revealed an only minute dissemination of EC-derived virus via secondary viremia following systemic infection . The low degree of dissemination of MCMV-rec into the heart could be the result of two scenarios . We expected that the immune response induced by systemic infection actively prevented secondary import of EC-derived virus into the transplant . Alternatively , after initial virus seeding by systemic ( i . v . ) infection , local virus production might simply outnumber secondary import of EC-derived virus . To address this issue and to initiate the activation of immune functions , systemic infection was performed prior to transplantation . Specifically , Tie2-cre or Tek-cre mice were first i . v . infected with MCMV-flox and only then received heart transplants of non-infected C57BL/6 mice either 20 h or 3 days after infection ( Fig . 5A/B ) . Strikingly , systemic infection prior to transplantation increased the relative contribution of EC-derived virus in the transplant from ∼5% ( Fig . 4B ) to ∼60% independent of the time delay between infection and transplantation ( Fig . 5A/B ) . This average of about 60% MCMV-rec reflects the average contribution of MCMV-rec in the organism in general . However , total virus titers in the heart transplant were 100- to 1000-fold lower than in both the endogenous heart exposed to i . v . infection as well as the hearts transplanted prior to i . v . infection ( Fig . 4B/C ) . It is important to note that the absolute amounts of recombined virus in the heart transplants ( grey circles in Fig . 5A/B ) were on the same level with those observed following i . v . infection after heart transplantation ( grey circles in Fig . 4B/C ) . Similar results were obtained after perfusion of recipient organs thus demonstrating that the detected virus was not blood-borne but was indeed produced within the transplanted organ . Total titers in the transplant decreased when transplantation was delayed from 20 h to 3 days after infection , reflecting the situation at day 5 and 7 p . i . , respectively . In two animals transplanted three days after infection , virus titers in the heart transplant even fell below the detection limit of 10 PFU/g organ , probably reflecting enhanced control by the host immune system at day 7 . This is supported by the relatively low virus titers in spleen , kidneys and adipose tissues as well as by the lack of detectable virus in blood ( Fig . 5B ) . In conclusion , we were surprised to see that ongoing virus replication and the accompanying immune response in the transplanted heart did obviously not alter the absolute amount of EC-derived virus originating from recipients' tissues by secondary viremia . These data demonstrate that virus dissemination between organs – originating from both endothelial and non-endothelial cells – has only minor effects on organ viral load following systemic infection .
One hallmark of CMV infection is the ability of the virus to infect many cell types and tissues from which again the virus may spread . Apparently , immune control defines to which extent this potential is realized in a given scenario . Therefore , the various clinical conditions need to be considered to explain CMV pathogenesis . Blood specimens play an important role in CMV diagnostics . Proper usage of the information gained by this analysis should monitor or even predict events that happen in organs . However , it is currently unclear under which conditions CMV is spread via blood . Fenner et al . were the first to propose a two-step dissemination model for systemic virus infections . Primary viremia transports the virus from the site of entry to liver and spleen where the virus replicates . Secondary viremia then causes dissemination from liver and spleen throughout the body [31] . This model became widely accepted for many viruses to this day , including CMV [6] , [32] . Yet , the original model was developed prior to any knowledge on innate immunity control functions and did thus not consider major factors in virus host defense . Recently , we challenged this view for CMV infection in the mouse model with respect of the role of the liver . Virus produced in hepatocytes is locally disseminated to other cell types but is not distributed from the liver to other organs via secondary viremia [24] . In the present study the vascular EC were analyzed for their claimed role in contributing to the CMV load in organs , and in disseminating the virus via primary or secondary viremia . Our salient findings are as follows: EC-derived virus significantly , ∼50% of the body virus pool , contributes to total virus load during acute infection . This contribution was quantified for the first time for the major organs . Yet , there was obviously no preference for dissemination of EC-derived virus over virus produced by other cell types . In addition , and similar to hepatocyte-derived virus , EC-derived virus was poorly disseminated via secondary viremia . These data raise doubts on the indiscriminate applicability of the primary and secondary viremia concept to virus infections in general . Properties of EC have enticed scientists to consider them as key production sites for virus dissemination , as they may release free virus particles directly into the blood stream or may detach from the vessel wall and transfer virus to other organs via the blood stream [14] , [15] . Moreover , EC could transfer the virus by contact to other cell types such as monocytes or granulocytes [33] , [34] , which would then disseminate the EC-derived virus to other organs [12] , [35] . On the other hand , EC-derived virus may also spread to underlying parenchyma and leave the organs via the draining lymph nodes to eventually reach the blood circulation via the thoracic duct . As heterotopic , abdominal transplants are not connected to lymph vessels , exiting virus would enter the peritoneal cavity that is drained by the mediastinal lymph nodes . This lymphatic dissemination route was recently described after intraperitoneal MCMV infection [36] and is also generally accepted as dissemination route after local infections , including intraplantar infection with MCMV [37] . Here , we provide the first quantitative analysis of organ- and cell type-specific virus dissemination . From an infected organ EC-derived virus readily disseminated to the other , uninfected organs . In the specific cases shown here , the infected transplanted organ ( heart or kidney ) created the condition of a primary viremia initiating from a defined source . EC-derived virus remained a stable fraction in both heart ( ∼80% ) and kidney ( ∼20% ) throughout the first week of infection [24] , thereby providing a constant supply of virus . Yet , the percentage of EC-derived virus that disseminated to other organs essentially mirrored the relative contribution of EC in the transplanted organ . Thus there was no preferential seeding of EC-derived virus . Infected EC might detach from the vessel wall and circulate . In fact , HCMV-infected EC were considered as a parameter for the diagnosis of HCMV organ involvement and for the study of the pathogenesis of disseminated infection [16] . This conclusion was originally based on the finding of two symptomatic patients with a high load of infected circulating EC , but experimental evidence for EC-derived virus colonizing other organs was missing so far . In the mouse model we now provide a nuanced view on the role of EC in virus dissemination . If the infected heart transplant is the source of primary infection , then EC-derived virus is readily disseminated , but without preference . During secondary viremia , however , there is only negligible import of EC-derived virus into the transplant as well as export from the transplant , and this is apparently independent of the extent of ongoing virus replication , associated inflammation , and immune control . Do our findings formally exclude any prominent role of EC-derived virus ? The answer is both yes and no . Yes , we can exclude this role in the mouse model and for the temporal conditions of our experiments . Unfortunately , the more time passes after initial infection of the animal the definition of virus as being EC-derived virus becomes more and more indirect . EC-derived virus progeny keeps the marker independent of the cell type in which the virus replicates in further replication rounds . Thus , with our experimental setup we cannot study later phases of infection when other conditions of virus productivity and dissemination may prevail . However , according to our previous experience , second and third replication rounds contribute less and less to the viral load in the immune competent host due to the onset of immune control [24] . We have not yet studied the situation of the immune deficient host for the EC progeny . For the hepatocyte-derived progeny , however , we know that immune suppressive regimens , even if combined , do not lift the strong dissemination block [24] . Nevertheless , by comparing the virus titers in different organs following transplant-derived and i . v . infection , we observed striking differences . Systemic ( i . v . ) infection with tissue culture produced virus preparations resulted in a uniform distribution of virus to many organs , whereas transplant-derived virus appeared to preferentially colonize spleen , lung and liver but not heart , adipose tissues and kidneys . This cannot be explained by known differences in organ specific virus kinetics . Therefore , cell-free virus , which is usually used for experimental infection , is apparently able to efficiently colonize all organs , whereas virus leaving an infected organ via a natural route reveals a different kind of spread . What could be the cause of the difference between i . v . infection with a solution enriched in isolated virions and the spread of infection from an infected organ ? The most plausible explanation is that the virus leaving an infected organ during systemic infection is predominantly transported in a cell-associated manner . Yet , this difference in organ and tissue distribution shows no preference for EC-derived virus and is altogether marginal with respect to total virus load in an organ .
This study was carried out in strict accordance with the recommendations and guidelines for the care and use of laboratory animals according to Tierschutzgesetz ( TierSchG , BGBI S . 1105; 25 . 05 . 1998 ) . All animal experiments were approved by the responsible state office ( Regierung von Oberbayern ) under permit number 55 . 2-1-54-2531-19-07 . M2-10B4 ( CRL-1972; ATCC ) and BALB/c-derived mouse embryo fibroblasts ( MEF ) were grown as described previously [38] . Transgenic Tie2-cre [39] and Tek-cre [11] mice were housed at the animal facility of the Max von Pettenkofer-Institute under specified-pathogen-free ( SPF ) conditions . Cre-transgenic mouse strains were maintained on the C57BL/6J background . Experiments were performed with gender matched pairs of mice at 3 to 12 months of age . C57BL/6J mice were obtained from Janvier . Tek-cre mice were obtained from Jackson Laboratories ( nr . 4128 ) . All viruses were derived from the molecular clone pSM3-fr [40] . Mutant virus ( MCMV-flox ) was generated as described [24] . Viruses were propagated on M2-10B4 cells and purified as described [41] . Virus quantification was done by standard plaque titration assay on MEF . Mice were infected intravenously ( i . v . ; into a tail vein ) with 8×105 PFU in a volume of 300 µl . Syngeneic transplantations of hearts or kidneys were performed between C57BL/6 mice and Tie2-cre or Tek-cre mice that were maintained on the genetic background of C57BL/6 mice . Heart transplant model: Abdominal-heterotopic cardiac transplants were performed , as previously described [42] . Briefly , the ascending aorta of the graft was anastomosed to the abdominal aorta of the recipient and the pulmonary artery to the inferior vena cava while the pulmonary veins were ligated . The graft function was assessed by daily palpation . Kidney transplant model: The murine kidney transplantation was performed as described previously [43] . Briefly , the left kidney of the donor was harvested and transplanted into the recipient . The kidneys of the recipients were not removed . A bladder patch was anastomosed to the recipient's bladder . No signs of rejection due to Cre expression by EC of the transplants or by EC of the recipient were seen throughout the experiments excluding host versus graft or graft versus host reactions , respectively . Virus load in organs was determined by plaque assay as described previously [24] with the modification that blood samples were sonicated before they were added to MEF in a volume of 10 µl per well . The numbers of MCMV-rec and MCMV-flox plaque forming units ( PFU ) were determined from organ homogenates after 4 days and from blood after 5 days using a fluorescence microscope ( Olympus ) . Only plaques visible in bright field were considered for the calculation . PFU were calculated per ml of blood or g of organ . Mice were anaesthetized and the peritoneal cavity was opened . After injection of 50 µl of heparin into the inferior vena cava , abdominal aorta and vena cava were cut cranially of the transplant . After all organs were perfused with 5 ml PBS via the vena cava the heart transplant was removed and perfused separately with 3 ml PBS . The percentage of MCMV-rec compared to total virus organ load per group , mean values , and standard deviations were determined . | More than sixty years ago Frank Fenner proposed that virus dissemination during acute infection originates from organs replicating virus to high titer ( often liver or spleen ) early in infection . Although never formally proven , this model has become commonly accepted and was applied to acute virus infections in general . Recently , we challenged this model by showing that - during acute murine cytomegalovirus infection – hepatocyte-derived virus hardly disseminates to other organs . We now applied our well established model of Cre/loxP-mediated green-fluorescence-tagging of MCMV to determine and quantify the role of infected endothelial cells ( EC ) in transplantation-associated CMV dissemination . We observed an only very poor dissemination of MCMV from the transplant to recipient tissues and vice versa . Interestingly , we observed no evidence for preferential dissemination of EC-derived virus . Significant differences in virus organ titers were found when comparing intravenous infection with transplant-mediated infection . This suggests a preferential dissemination of cell-associated virus in the transplant setting . In summary , our findings argue for a preferential dissemination of cell-associated MCMV but demonstrate that the Fenner model does not apply to MCMV . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"animal",
"models",
"medicine",
"oncology",
"infectious",
"diseases",
"model",
"organisms",
"medical",
"microbiology",
"virology",
"surgery",
"biology",
"microbiology",
"host-pathogen",
"interaction"
] | 2011 | Shedding Light on the Elusive Role of Endothelial Cells in Cytomegalovirus Dissemination |
Numerous studies have suggested that hub proteins in the S . cerevisiae physical interaction network are more likely to be essential than other proteins . The proposed reasons underlying this observed relationship between topology and functioning have been subject to some controversy , with recent work suggesting that it arises due to the participation of hub proteins in essential complexes and processes . However , do these essential modules themselves have distinct network characteristics , and how do their essential proteins differ in their topological properties from their non-essential proteins ? We aimed to advance our understanding of protein essentiality by analyzing proteins , complexes and processes within their broader functional context and by considering physical interactions both within and across complexes and biological processes . In agreement with the view that essentiality is a modular property , we found that the number of intracomplex or intraprocess interactions that a protein has is a better indicator of its essentiality than its overall number of interactions . Moreover , we found that within an essential complex , its essential proteins have on average more interactions , especially intracomplex interactions , than its non-essential proteins . Finally , we built a module-level interaction network and found that essential complexes and processes tend to have higher interaction degrees in this network than non-essential complexes and processes; that is , they exhibit a larger amount of functional cross-talk than their non-essential counterparts .
High-throughput experimental approaches for determining protein interactions have resulted in large-scale cellular networks for numerous organisms . Graph-theoretic analyses of these networks have been a great aid in advancing our understanding of cellular functioning and organization ( review , [1] ) . One of the most fundamental discoveries is that there is a strong relationship between the topological characteristics of cellular networks and their underlying functioning . For example , cellular networks consist of tightly clustered groups of interacting proteins , and these proteins work together as protein complexes or biological processes to achieve specific biological functions [2]–[7] . An orthogonal decomposition reveals that there are recurring and over-represented topological and functional patterns within larger cellular networks , and these network motifs [8] , [9] and network schemas [10] can be associated with dynamic regulatory properties and shared mechanisms of functioning . Here , we revisit perhaps the most basic structure-to-function relationship that has been proposed for cellular networks—that between the number of interactions a protein has and its overall functional importance . The importance of a gene to a cell or an organism can be quantitatively measured by considering the phenotypic effects of gene deletion or disruption . Experimental studies in the baker's yeast S . cerevisiae have demonstrated that approximately 19% of its proteins are essential; that is , the deletion of these proteins results in cell death , even in optimal growth conditions [11] , [12] . Early computational analysis of the yeast S . cerevisiae protein-protein physical interaction network revealed a scale-free topology , where a few “hub” proteins have many interactions , and also showed that hub proteins are more likely to be essential than other proteins [13] . Numerous subsequent studies have confirmed this centrality-lethality relationship , not only in yeast [14]–[19] but also in other organisms [20] . On the other hand , the relationship has been observed to be weak in networks consisting of interactions determined via high-throughput yeast two-hybrid experiments while stronger in other types of networks [16] , [18] , [19] , and it has been proposed that , in yeast two-hybrid networks , the observed relationship is due to study bias favoring the determination of interactions of essential proteins [21] . Nevertheless , the positive correlation between protein interaction degree and essentiality is generally accepted , with numerous reasons proposed in the literature to explain this relationship . Initial work suggested that high-degree proteins may be essential due to their role in interaction network connectivity [13]; however , this is unlikely to be the case as it was subsequently shown that non-essential hubs are just as important as essential hubs for maintaining connectivity , and that essentiality is better correlated with local , rather than global , measures of connectivity in protein-protein interaction networks [17] , [18] . It was alternatively proposed that essentiality is a property of interactions [22] . That is , there are essential protein interactions , without which an organism cannot survive , and these are randomly distributed across the network; hubs then tend to be essential as they are more likely to participate in essential interactions . However , this model implies that the probabilities that two non-interacting proteins are essential are independent of each other , and this is not the case [18] . Instead , Zotenko et al . [18] argued that the correlation between degree and essentiality is due to the participation of essential proteins in essential functional modules consisting of groups of densely clustered and functionally related proteins . They further showed that the essentiality of hubs that are not in these computationally extracted modules are only weakly correlated with degree [18] . Indeed , it had previously been found that essential proteins tended to be densely connected to each other [15] and concentrated in complexes 23 , 24 , suggesting that essentiality is a modular property rather than a property of individual proteins . Building upon this , it has been argued that essential complexes tend to be large , and thus proteins within them have a larger number of interactions , and that this explains why hubs tend to be essential [19] . While there is substantial evidence that essentiality is a modular property in protein-protein interaction networks , it is also clear that complexes and processes do not consist entirely of essential or non-essential proteins . Do essential proteins within an essential complex or process differ from the non-essential ones ? Further , not all complexes and processes contain essential proteins . Do such essential modules have distinctive roles in cellular networks ? In this paper , we aimed to discover whether , within modules , their essential and non-essential proteins differ in their interaction properties , and at a more global scale , whether essential and non-essential modules differ in their network-level properties . To accomplish this , we developed a computational framework that incorporates information about functional modules within the context of network analysis techniques . To uncover general and robust principles , we performed our analysis on three types of S . cerevisiae protein-protein interaction networks and considered functional modules derived from protein complexes as well as Gene Ontology ( GO ) biological process annotations [25] at different levels of resolution . Further , to address the issue of study bias , we performed our analysis on additional networks which removed interactions determined in small-scale experiments . We began by re-examining the relationship between protein essentiality and network modularity . We hypothesized that if essentiality is a modular property , as has been proposed previously [18] , then a protein's intramodular physical interaction degree should be a better predictor of a protein's essentiality than its intermodular physical interaction degree . To test this , we utilized biological process functional annotations of proteins and classified physical interactions into intraprocess interactions within processes and interprocess interactions between processes . We found that essential proteins tend to have many interactions with proteins within the same functional modules and that the intraprocess interaction degree is more correlated with essentiality than overall degree . Further , we found that the relationship between overall degree and essentiality is significantly weakened when controlling for intramodular degree , but is not as affected when controlling for intermodular degree . Thus , we show in a direct and simple manner that , for many essential proteins , their essentiality is likely to be a consequence of their participation within essential modules consisting of functionally similar proteins . To further ascertain whether the modularity of essential proteins is due to their roles within essential protein complexes or more generally within essential biological processes , we repeated this analysis while first exclusively focusing on proteins within protein complexes and next focusing only on proteins that are not within known protein complexes . We found that most essential proteins with many intraprocess interactions in fact participate in essential protein complexes or in essential biological processes that include one or more protein complexes; that is , the modularity of protein essentiality appears to be a consequence of protein complexes , not more broadly of biological processes . Next , we examined complexes that contain essential proteins , and compared their essential and non-essential proteins . We reasoned that if the relationship between essentiality and interaction degree for proteins within these complexes is entirely a consequence of the complexes themselves being essential , then essential and non-essential proteins within the same complex should not differ with respect to degree . On the contrary , we found that essential proteins tend to have more interactions , particularly intracomplex interactions , than their non-essential counterparts within protein complexes . That is , while essentiality appears to be a modular property , the degree of a protein is associated with essentiality within essential complexes; we hypothesize that these essential proteins may play a more important role in maintaining the structure and/or function of complexes . Finally , we analyzed modules containing essential proteins within the context of other functional modules . We inferred significant “cross-talks” between protein complexes and biological processes and used them to build module-level networks , in which two complexes or processes are linked if they have an enriched number of physical interactions between them . Using these module-level networks , we uncovered that functional modules with essential proteins tend on average to have higher degree; that is , degree in the module-level network is positively correlated with module essentiality . Overall , by considering proteins within the functional context of the yeast interactome , we give evidence that there is a relationship between essentiality and network topology at different levels of cellular organization: at the protein level , within protein complexes , and also more globally at the module level , with complexes and processes that are essential tending to interact with more functional groups .
For a given interaction network , we labeled protein interactions as either “intramodular , ” “intermodular” or neither using two sources of functional data . In particular , we utilized yeast protein complex data compiled in [27] and Gene Ontology ( GO ) Biological Process ( BP ) annotations [25] . Thus , intramodular interactions can arise from either intracomplex or intraprocess interactions , and intermodular interactions arise as either intercomplex or interprocess interactions; we will separately consider both types of intramodular and intermodular interactions . For protein complex data , “intracomplex” interactions are between all pairs of proteins that participate in a shared complex and “intercomplex” interactions are between pairs of proteins that are each found in at least one complex but are never found in the same complex . It is more complicated to characterize interactions as intramodular or intermodular using GO BP terms , as the terms are hierarchically related and annotate different numbers of proteins , with some very general terms . To get only informative and specific terms , we considered GO BP terms that annotate at most 50 proteins in the yeast proteome . An interaction is unannotated unless both proteins are annotated with any one of these specific GO BP terms . An interaction is “intraprocess” if it is between two proteins sharing one of these specific BP terms . If two proteins with an interaction are annotated with specific GO BP terms but do not share any of them , the interaction is “interprocess . ” We note that while physical interactions are largely thought of as “within process , ” especially as compared to other types of interactions [28] , a significant fraction of physical interactions are interprocess ( Table S2 ) ; this is true even as the threshold for choosing specific terms is increased . As a first step towards relating protein essentiality to network modularity , for each protein , we computed its number of intraprocess interactions , interprocess interactions , and total annotated interactions . We then considered each of the intraprocess , interprocess and total annotated interaction degrees in turn , and ordered all proteins from high to low degrees with respect to it . As we varied a threshold for the number of proteins considered , we computed the fraction of essential proteins in the “high degree” or “hub” set . Over a large range of thresholds , the high degree proteins , as ranked by intraprocess degree , have a higher fraction of essential proteins than the high degree proteins as ranked by either total annotated degree or interprocess degree ( Figures 1 ( a ) , S1 ( a ) and S2 ( a ) ) . For the Pull-down and Full networks , the fraction of essential proteins tends to decrease as the threshold for intraprocess , interprocess , or total degree is lowered . In the Direct network ( Figure 1 ( a ) ) , this trend is only true for intraprocess interaction degree and is notably not true for total degree; this is consistent with previous work showing that the relationship between essentiality and overall interaction degree is weak in networks consisting of interactions determined by yeast two-hybrid [16] , [18] , [21] . To further quantify the correlation between essentiality and degree , we used the Spearman's rho rank correlation coefficient ( SRCC ) [29] ( Figures 1 ( b ) , S1 ( b ) and S2 ( b ) ) , and found that the SRCC is highest for intraprocess degree ( 0 . 25 for Direct , and 0 . 35 for the other two networks ) , and much lower for interprocess degree ( 0 for Direct , 0 . 22 for Pull-down and 0 . 21 for Full ) . We note that since protein essentiality is a binary value and thus there are many tied values , it is not possible for the SRCC to achieve a value of 1 . For example , the SRCC between essentiality and all annotated degree in the Direct network could at most reach a maximum value of 0 . 7680 ( i . e . , the case where the essentiality values for the proteins are swapped so that all essential proteins have higher degrees than all non-essential proteins ) . Next , to disentangle the contributions of intraprocess and interprocess degree to the observed correlations , we computed partial correlations between essentiality and all annotated interactions , when controlling for intraprocess and interprocess degree . For the three networks we found that when controlling for intraprocess degree , the SRCC between total degree and essentiality notably diminished , whereas when controlling for interprocess degree , the SRCC remained high ( Figures 1 ( b ) , S1 ( b ) and S2 ( b ) ) , and even increased for the Direct network . As another way of looking at the difference between intraprocess and interprocess interaction degree , we compared the degree distributions of essential proteins and non-essential proteins ( Figures 1 , S1 and S2 ( c ) – ( e ) ) using the Wilcoxon rank sum test . For comparing degree distributions , we included all proteins with at least one annotated interaction; these proteins may have zero intraprocess or interprocess interactions . Since the same number of proteins are considered when comparing total , intraprocess , or interprocess degree ( Figures 1 , S1 and S2 ( c ) – ( e ) ) , the -values given are comparable . The difference in the number of interactions between essential and non-essential proteins is much more significant when only intraprocess interactions are considered ( Figures 1 ( d ) , S1 ( d ) and S2 ( d ) ) , as compared with the case when all annotated interactions are considered ( Figures 1 ( c ) , S1 ( c ) and S2 ( c ) ) or when only interprocess interactions are considered ( Figures 1 ( e ) , S1 ( e ) and S2 ( e ) ) . As an alternative to categorizing all annotated interactions as either interprocess or intraprocess , we also considered the case where interactions are weighted according to the semantic similarity [30] between the functional terms annotating the two proteins . This weight is in the range of 0 and 1 with proteins sharing highly specific functional terms getting higher scores ( see Materials and Methods for more details ) . Thus , the semantic similarity between two interacting proteins is a continuous measure of the “intramodularity” of the interaction . Then , the semantic similarity degree of a protein is defined as the sum of semantic similarity of interactions . Across the Direct , Pull-down and Full networks , we find that there is a stronger correlation between essentiality and degree when all interactions are weighted with semantic similarity than when they are just counted ( Figure S3 ) . In other words , proteins having many interactions within a similar functional context are more likely to be essential than proteins having many interactions . Altogether , a range of computational analyses shows that a large portion of the observed relationship between essentiality and interaction degree can be explained when considering just intraprocess interactions . Having shown the strong correlation between intraprocess interaction degree and essentiality , we sought to characterize the contribution of intracomplex interactions . In particular , previously it had been observed that essential proteins tend to be clustered together within essential protein complexes [18] , [24] . Thus , we hypothesized that having intracomplex physical interactions for a protein is more important for predicting its essentiality than having other types of physical interactions . That is , as we have defined them , functional modules can be comprised either of protein complexes or biological processes corresponding to GO BP terms . In the previous section , we utilized modules derived from BP terms . We next focus on modules derived from protein complexes , as compiled in [27] . We begin by observing that complexes as a whole are enriched in essential proteins . In particular , whereas 18 . 60% ( or ) of proteins are essential in the yeast genome , 37 . 54% ( or ) are essential when considering proteins involved in the set of complexes we are considering . In fact , 57 . 01% ( or ) of all essential proteins are involved in protein complexes , even though only 28 . 24% ( or ) of proteins take part in our set of complexes . Thus , any conclusions arising from the analysis of protein complexes is based on the interaction properties of a significant fraction of essential proteins . For each network , we derived a subnetwork where nodes represent proteins involved in any protein complex and edges represent interactions from our original interactions between these proteins . In the Direct network , 35 . 66% ( or ) of interactions are intracomplex ( Table S2 ) . Repeating the analysis we performed for intraprocess vs . interprocess interactions , we found that intracomplex physical interactions are more correlated with protein essentiality than all annotated physical interactions ( Figures 2 ( a ) , S4 ( a ) and S5 ( a ) ) . It has been previously observed that there is a strong correlation between complex size and essentiality [19] , and argued that essential complexes tend to be large , and proteins within them tend to have more interactions , and this is a driving force in the relationship between essentiality and interaction degree . In our dataset , there is a clear positive correlation between complex size and the fraction of essential proteins within the complex ( SRCC: 0 . 24 , -value: 2e-6 ) . Moreover , there is a strong correlation between protein essentiality and the size of the largest complex to which it belongs , with SRCCs of 0 . 25 , 0 . 24 and 0 . 24 for Direct , Pull-down and Full networks , respectively ( Figures 2 ( b ) , S4 ( b ) and S5 ( b ) ) . We found , however , this relationship is not as strong as that between essentiality and intracomplex degree in our networks ( black vs . green curve in Figures 2 ( a ) , S4 ( a ) and S5 ( a ) ) . We also computed partial correlations between essentiality and all annotated interactions , when controlling for intracomplex degree , intercomplex degree , or complex size . We found that when controlling for intracomplex degree , the SRCC between total degree and essentiality notably diminished ( from 0 . 17 , 0 . 32 and 0 . 32 to −0 . 01 , 0 . 13 and 0 . 14 for the Direct , Pull-down and Full networks , respectively ) , whereas when controlling for intercomplex degree or complex size , the SRCC was not as greatly diminished ( Figures 2 ( b ) , S4 ( b ) and S5 ( b ) ) . Further , the difference in degree distribution between essential and non-essential proteins ( Figuress 2 , S4 and S5 ( c ) – ( f ) ) is most significant when considering intracomplex degree and least significant when considering intercomplex degree . We note that there is a correlation between a protein's intracomplex degree and the size of the complex to which it belongs ( SRCC: 0 . 3790 , 0 . 7319 and 0 . 7809 for the Direct , Pull-down and Full networks , respectively ) ; the much stronger correlations for the Pull-down and Full networks as compared to the Direct network are expected as the former two networks include many indirect ( i . e . , co-complex ) interactions . Thus far , we have found a stronger correlation between essentiality and intramodular degree than between essentiality and all annotated degree when we focus on either biological process or protein complex derived modules . Instead of using biological process or protein complex annotations to categorize interactions as either intramodular or intermodular , we next considered modules derived from network clustering approaches . In particular , we applied the state-of-the-art SPICi network clustering algorithm [31] , and categorized interactions within clusters as intramodular and interactions between clusters as intermodular . We note that clusters are uncovered in a purely topological manner and may correspond to either protein complexes or functional modules . On the Direct , Pull-down and Full networks , essentiality is more correlated with intramodular interaction degree , defined via network clustering , than it is with either intermodular or total degree ( Figure S7 ) . What happens if we consider intraprocess interactions when excluding those that are intracomplex ? That is , some biological processes may consist of a single protein complex or several protein complexes; in these cases some of the observed intraprocess interactions are more specifically intracomplex interactions within complexes that take part in the process . To focus on interactions that are not intracomplex , we filtered biological processes to remove these interactions ( see Materials and Methods for more details ) . Among the proteins that are annotated with any filtered biological process , 16 . 52% ( or ) proteins are essential , which is slightly less than that when considering all proteins in the genome . In a subnetwork for the set of filtered biological processes from each of three interaction networks , there is a weaker correlation between interaction degree and essentiality as compared to the correlation for complexes , and the intraprocess degree is not more correlated with essentiality than all annotated degree ( Figure 3 ( a ) , ( b ) , ( c ) ) . The correlations are especially weak in the Direct network . Moreover , in the Pull-down and Full networks , the correlation between essentiality and interprocess filtered interaction degree is somewhat higher than that between essentiality and intraprocess filtered degree . Having shown in a global analysis of proteins within complexes that essential proteins tend to have more intracomplex interactions than non-essential proteins , we next considered a per-complex analysis . We hypothesized that , for each essential protein complex , its essential proteins should be more central or have a higher intracomplex degree than its non-essential proteins . We tested this hypothesis for a subset of protein complexes with enough member proteins and intracomplex interactions . In particular , we included a complex in our test if it has at least two essential proteins and at least two non-essential proteins , each of which has intracomplex interactions . Table 1 shows that for a large fraction of complexes , essential proteins tend to have a higher average intracomplex degree than non-essential proteins . In particular , in the Direct network , for more than 76% of complexes , essential proteins have higher average intracomplex degree ( empirical -value ) . In the Pull-down or the Full network , the fraction of complexes with a higher average degree for essential proteins is lower than in the Direct network ( 58 . 8% and 61 . 5% , respectively ) ; this is as expected since these networks include indirect intracomplex interactions . In fact , in the Pull-down and the Full networks , there are seven “clique” complexes in which every protein has an intracomplex interactions with all other proteins within the complex , whereas there are no such complexes in the Direct network . Without these clique complexes , the percent of complexes with higher average intracomplex degree for essential proteins goes up to 68 . 2% and 71 . 1% for the Pull-down and the Full networks , respectively . By considering each complex individually , this analysis better handles proteins involved in multiple complexes . Although we removed highly overlapping complexes ( see Materials and Methods ) , 14% ( or ) of proteins belong to two or more complexes . Moreover , these proteins tend to be essential; among proteins in more than one complex , 53 . 81% ( or ) are essential ( as opposed to 37 . 54% for all proteins within complexes ) . Thus , it is possible that one reason that essential proteins tend to have a higher intracomplex degree ( Figure 2 ) is because essentiality is enriched in proteins belonging to multiple complexes , and the intracomplex degree of an essential protein is computed over the complexes to which it belongs to; however , looking at one complex at a time should alleviate this problem . As another way of addressing the possible bias due to proteins participating in multiple complexes , for each protein , we computed the intracomplex degree using only interactions within the largest complex to which it belongs . Next , we compared all proteins within complexes , and found that there is a significant difference in degree distribution between essential and non-essential proteins ( Figure 4 ( a ) ) , and this is also true in the other two networks ( Figures S8 ( a ) and S9 ( a ) ) . Since there is a correlation between complex size and the fraction of essential proteins within the complex [19] , and complex size is also correlated with the intracomplex degree of its member proteins , it is possible that the observed relationship between intracomplex degree and essentiality is due to the correlation between the complex size and essentiality . To address this , we next normalized interaction degree by complex size; that is , the normalized intracomplex degree of a protein is computed as the number of intracomplex interactions divided by the complex size . We found that the normalized degree of essential proteins tends to be significantly greater than that of non-essential proteins ( Figures 4 ( b ) , S8 ( b ) and S9 ( b ) ) . As we have just shown , essential proteins tend to have more intramodular interactions , and for complexes with essential proteins , its essential proteins tend to have more intracomplex interactions than its non-essential proteins . In contrast , the intermodular interaction degree of a protein has a weaker relationship with its essentiality . Nevertheless , as noted earlier , there are a significant number of intermodular physical interactions ( see Table S2 ) , and presumably these physical interactions connecting different functional modules in the network are important for the module to accomplish a task . We hypothesized that the essentiality of a protein complex or functional module may be related to its topological prominence within a module-level network . To test this , we built a “module network” where nodes are modules and edges are between modules that have an enriched number of intermodular cross-talk interactions . In particular , we constructed a module network for either protein complexes or filtered biological processes from each physical interaction network by computing the number of physical interactions between two modules and comparing this to the average number found in randomized networks ( see Materials and Methods ) . For each network , we give the number of cross-talks uncovered using modules derived either from protein complexes or filtered biological processes in Tables 2 and 3 , respectively . We note that the number of cross-talks for processes is much higher than that for complexes because a relatively higher number of interactions for processes are intermodular rather than intramodular ( 86 . 98% vs . 64 . 34% , Table S2 ) . For modules , defined by either complexes or filtered biological processes , as we decrease the threshold for the number of cross-talks required for a module to be a considered a hub module , we find that the fraction of modules that contain an essential protein tends to decline ( Figure 5 ) . Further , there is a significant positive correlation between whether a module contains an essential protein and its cross-talk degree , with SRCCs on the three networks when considering complexes and when considering filtered biological processes . Since modules that have more proteins may also have larger cross-talk degree , we also computed the partial correlation between cross-talk degree and module essentiality when controlling for the number of proteins in the module ( Table S3 ) ; this varies in the three networks from 0 . 22–0 . 29 when considering complexes and 0 . 23–0 . 28 when considering filtered biological processes . Further , we found a significantly positive correlation between the normalized cross-talk degree of a module , defined as the cross-talk degree divided by module size , and module essentiality ( Table S3 ) . We also compared the cross-talk degree distribution between essential and non-essential modules using the Wilcoxon rank sum test . In our three networks , whether considering protein complexes or biological processes , the essential modules have significantly higher cross-talk degree than non-essential modules ( Figure 6 ) . Finally , since modules with a larger number of proteins have a greater chance of containing an essential protein , we also considered the fraction of proteins within a module that are essential . We found that the cross-talk degree of a module is positively correlated with the fraction of proteins within a module that are essential ( Table S4 ) , though these values are not as high as for binary essentiality ( SRCCs on the three networks for complexes and for filtered biological processes ) . We observed that many cross-talks occur between functional modules that are functionally related ( i . e . , they both take part in a more general , shared biological process ) . These types of cross-talks can be interpreted as intraprocess interactions at a broader level of functional similarity . As one example , the Ndc80p complex has a high cross-talk degree in all networks studied . In the Direct network , we uncover seven cross-talks ( Figure 7 ) . Ndc80p is a component of the kinetochore , which is central to chromosome segregation and couples chromosomes to microtubule polymers . Two of the uncovered cross-talks are with the DASH and MIND complexes , both of which are also kinetochore associated; these cross-talks can be thus be interpreted as “intramodular” interactions at a higher level of organization . On the other hand , Ndc80 also has cross-talks with other complexes that take part in a range of distinct biological processes , including the nucleosome remodeling complex SWI/SNF , the dynactin microtubule associated complex , the MRX complex involved in DNA damage repair , the nuclear condensin complex and the nuclear cohesion complex . To see if essential functional modules have many cross-talks with functional modules representing truly different biological processes , we considered a set of expert-selected GO BP terms in yeast [32] , and focused on those that annotate at most 500 proteins . We considered a functional module to be annotated with one of these terms if % of its proteins are annotated with it . We next ignored cross-talks between two functional modules if they are both annotated with a shared term; even in this case , we found that essential functional modules are still correlated with cross-talk degree ( Table S5 ) . Thus , a range of analyses reveals that there is a relationship between the topological importance of a functional module and its tendency to contain essential proteins . Because low-throughput studies may be biased towards studying essential proteins , essential proteins may appear to have more interactions in existing network databases . Further , high-throughput studies may themselves utilize a select set of “bait” proteins that may bias the degree distribution of interaction networks . To address these potential concerns , we performed several additional network analyses . First , we repeated our analysis on the Y2H-union network [21] and the more recently built BinaryHQHT network [33] , both derived from high-throughput yeast two-hybrid data . In these networks , interactions found in experiments involving a few bait proteins were removed , and only high-quality yeast-two hybrid interactions ( found in several experiments ) were retained . The networks have notably smaller size when focusing on annotated interactions ( Tables S9 , S10 and S11 ) ; nevertheless , as we outline below , repeating the analysis yields similar results as for the Direct network . In the Y2H-union and BinaryHQHT networks , the intraprocess interaction degree of a protein has a weak but statistically significant correlation with its essentiality while the overall degree of a protein is not correlated with essentiality in the Y2H-union network and is only weakly correlated with essentiality in the BinaryHQHT network ( Figures S10 and S11 , ( a ) and ( b ) ) . That is , protein essentiality is reflected in intraprocess degree in these networks , not overall degree . Further , the intraprocess degree is found to be significantly higher for essential proteins than non-essential proteins ( Figure S10 ( d ) and S11 ( d ) ) , but this is not true for overall degree and for interprocess degree ( Figures S10 and S11 , ( c ) and ( e ) ) . For 84 . 62% and 78 . 57% of essential protein complexes in the Y2H-union and BinaryHQHT networks , essential proteins tend to have a higher average intracomplex degree than non-essential proteins ( Table S12 ) , and essential proteins have higher intracomplex degree and normalized intracomplex degree than non-essential proteins ( Figures S12 and S13 ) . Next , in our module-level analysis , we find that the cross-talk degree in a complex-level network is significantly correlated with complex essentiality in the BinaryHQHT network ( Figure S14 ( a ) ) , and the cross-talk degree in a process-level network is significantly correlated with process essentiality on both networks ( Figures S14 ( b ) and S15 ( c ) and ( d ) ) . For the Y2H-union network , only four complexes are found to have cross-talks , and the relationship between cross-talk degree and complex essentiality is weak ( Figures S14 ( a ) and S15 ( a ) ) ; this may be due to the small number of intercomplex interactions in this network ( Table S10 ) . In our second analysis , we removed interactions uncovered in low-throughput experiments ( where less than 50 interactions were determined ) from the Pull-down network , and restricted our analysis to the interaction properties of proteins labelled as bait proteins . Bait proteins have a higher fraction of essential proteins in this network ( Table S14 ) . When considering just bait proteins , we find stronger relationships between intraprocess degree and essentiality than between overall interaction degree and essentiality ( Figure S16 ( a ) and ( b ) ) . Moreover , the relationship between overall interaction degree and essentiality is no longer significant when controlling for intraprocess degree ( Figure S16 ( b ) ) . Further , the intraprocess degree of bait proteins is found to be higher for essential proteins than non-essential proteins ( Figure S16 ) , and within complexes with both essential bait and non-essential bait proteins , essential proteins have higher intracomplex degree ( Figure S17 ) . Overall , the relationships between intraprocess degree and protein essentiality , degree within complexes and essentiality , and module essentiality and cross-talk degree are largely recapitulated in the smaller networks , Y2H-union and BinaryHQHT networks , where low-throughput experiments and experiments biased towards essential proteins are specifically excluded . Further , comparisons between bait proteins , which may be enriched in interactions , also confirms a relationship between the number of ( intraprocess ) interactions a protein has and whether it is essential .
A long line of previous research has studied the relationship between network topology and protein essentiality . Recent work has argued that hubs take part in densely connected essential complexes and processes [18] , and these essential complexes tend to be large [19] . That is , it has been argued that essentiality is a modular property , and essential proteins within essential modules tend to have many interactions as these modules tend to be large . Our initial analysis , revealing that a protein's intramodular interaction degree is more predictive of essentiality than its overall degree , largely supports this argument . We also found that if we focus on proteins that do not belong in complexes , the intraprocess interaction degree does not correlate with essentiality any better than overall interaction degree; this suggests that the observed network modularity of essential proteins is largely due to complexes , and is not a more general feature of biological processes . The observed positive correlation between protein essentiality and intramodular degree cannot be attributed only to module-level complex essentiality . In particular , within essential protein complexes , we found that their essential proteins tend to have higher intracomplex degrees than their non-essential counterparts . That is , within essential complexes , the topological prominence of its constitutent proteins is related to essentiality; this may be due to the importance of these proteins in maintaining the structural integrity of these complexes . This view is consistent with the relative enrichment of essentiality amongst proteins with many structural interfaces as opposed to just one or two structural interfaces [34] . While we found that intermodular interactions were less important than intramodular interactions in explaining protein essentiality , we also observed a significant number of intermodular interactions in physical interaction networks . We considered these interactions at a modular level , and demonstrated that essential functional modules tend to have more cross-talks with other functional modules . That is , our analysis showed that there is correlation between network topology and essentiality both at the protein level as well as at the modular level . Further , we observed that functionally related modules are likely to interconnect to each other , thereby revealing the hierarchical structure of physical interaction networks . Overall , our work has advanced our understanding of the relationship between essentiality and network topology . We have shown the importance of intramodular interactions , especially intracomplex interactions , and demonstrated that essential modules tend to have a higher cross-talk degree than non-essential modules . These findings are likely to yield improvements in our ability to predict protein essentiality . Indeed , integrative machine learning approaches that use a range of network and sequence features have been previously applied to predict protein essentiality ( e . g . , see [35]–[41] ) ; based on our work , information about functional modules and protein complexes , especially with respect to intramodular and cross-talk degree , should also be incorporated within these frameworks . In the future , it would be interesting to characterize the network properties of essential proteins that are not central in protein physical interaction networks . Based on our current findings , we can speculate that some of these proteins are important for the functioning of specific essential modules , and this may be reflected in their interactions with other proteins within their modules , but these relationships may be better represented via other types of interactions ( e . g . , regulatory , metabolic or genetic ) . Our framework for incorporating functional information into network analysis is likely to be useful in establishing whether or not this is the case . Finally , while we have performed our analysis on S . cerevisiae , our approach can be applied to study essential proteins in other well-annotated organisms with large-scale interaction networks and genome-scale gene deletion or disruption data .
We performed our analysis on five physical interaction datasets . For our first network , physical interactions were gathered from BioGRID [26] , release 3 . 1 . 78 , using all evidence codes indicative of physical interactions except “Affinity Capture-RNA” and “Protein-RNA . ” For the early yeast two-hybrid paper of Ito et al . [42] , we only included the core data . To remove artifacts due to “sticky proteins” in certain experiments , if a protein has more than 30 interactions from a single experimental data source , we removed these interactions . For our second network , we extracted direct physical interactions from the initial network by utilizing only interactions that were determined from one of the following experimental systems: Biochemical activity , Co-crystal structure , Far western , FRET , Protein-peptide , Reconstituted complex , and Two-hybrid . For our third network , we extracted from the initial network those interactions that were determined either by Affinity capture-Western or Affinity capture-MS . We refer to these three networks as Full , Direct and Pull-down , respectively , and their sizes are given in Table S1 . We also considered two additional networks , comprised of interactions that were not determined in small-scale experimental assays; in this manner , we attempt to minimize the effect of study bias . The first of these networks , which we refer to as Y2H-union , was built in an earlier study [21]; it included only interactions determined in large-scale high-quality yeast two-hybrid studies , and excluded an experiment using a specific set of “bait” proteins that was enriched in essential proteins [43] . We next used the more recently built high-throughput yeast two-hybrid network of [33] , which we refer to as BinaryHQHT . Finally , we built a high-throughput network from our Pull-down network by keeping only those interactions that were found in experiments uncovering at least 50 interactions and for which there were more than 10 “bait” proteins . We refer to this second network as the Pull-downf network , and use it to compare the network properties of bait proteins with respect to each other . We used the set of 430 protein complexes compiled in [27] , which includes the SGD Macromolecular Complex GO standard [44] , the CYC2008 protein complex catalog [45] and a set of manually curated complexes . From this initial set , we removed highly overlapping complexes as follows . First , if the proteins comprising one complex are a subset of the proteins comprising another complex , the smaller complex is removed . Next , for any two complexes , if the Jaccard index of the proteins making them up ( i . e . , the number of overlapping proteins divided by the size of the union of the protein sets ) is , we removed the smaller complexes . Additionally , as in previous work [19] , we removed the four complexes corresponding to the subunits of the ribosome , as they contain a large number of proteins; that is , these four complexes can disproportionately affect the per-protein analysis . After these filters , we were left with 390 complexes . ( See the Supplement Figure S6 , Text S1 for intraprocess and interprocess results including the four ribosomal complexes ) . For our functional analysis , we worked with a subset of specific Gene Ontology ( GO ) Biological Process ( BP ) terms [25] that were derived from the entire GO ( version 1 . 1 . 2130 ) as follows . First , we extracted 1418 BP terms , each of which annotates at least 5 yeast proteins and at most 50 . Next , to hone in on the contribution of a specific biological process ( as opposed to the effects arising from proteins that are annotated with that process but are within protein complexes ) , we pruned the set of proteins that are associated with these functional terms . More specifically , if the size of the intersection between a biological process and one of our original set of 430 protein complexes is , the proteins in the intersection were no longer associated with the process . If this left fewer than 2 proteins associated with the process , or with less than half the number of proteins that it is known to annotate , then this term was removed from consideration . Finally , highly overlapping processes were removed in the same manner as described above for complexes . This procedure resulted in 391 “filtered” processes , with 2567 proteins associated with at least one of these processes . For a given network , we exhaustively determined whether pairs of functional modules are enriched in the number of interactions found between them [10] . We considered modules arising from complexes or processes in turn ( i . e . , functional modules consist of either proteins within the same complex , or that have a shared process annotation from the 391 filtered processes considered ) . We considered the proteins within the network that are associated with any of the modules that we are considering , as well as all the edges that correspond to intermodular interactions amongst these proteins . Next , for any two modules and we counted the number of “cross-talk” interactions between the proteins comprising each of these modules . Note that interactions where either of the proteins is annotated with both and were not included as these are intramodular interactions . To determine whether the number of observed cross-talk interactions for this pair is more than would be expected by chance , we randomized the intermodular interactions within the network 100 times using stub-rewiring ( as in [8] ) , thereby preserving degree distribution , module annotation , and the overall number of cross-talk interactions . Then , if is the number of cross-talk interactions between and in the real network , and is the average number of corresponding cross-talk interactions in randomized networks , the odds-score of the module pair is defined as: The addition of the pseudocount of 1 downweighs the contribution of very rare cross-talks that could otherwise obtain high scores simply due to very small ( or zero ) average counts in the randomized graphs . In order for a module pair to be considered a cross-talk , we required that there should be at least two independent ( i . e . , non-overlapping ) cross-talk interactions , and that its odds-score should be at least . The observed relationship between module essentiality and cross-talk degree persists for a range of odds-scores ( see Tables S6–S8 ) . The semantic similarity between two GO terms within the same ontology is an estimate of the functional similarity between the terms . We use the semantic similarity measure introduced by [30] . In particular , let be the fraction of proteins in yeast annotated with term among the total number of proteins . Then is a measure of how specific a term is . We compute the term semantic similarity of and , as , where is a least common ancestor of and in the GO ontology . Note that if the LCA of two terms is a root term ( e . g . , GO:0008150 ‘biological process’ ) , then . Moreover , if two terms are the same , then . This measure is naturally extended to functional relationships between proteins that have multiple annotations . For a protein , let be the set of terms with which is annotated . If a term annotates , then all its parent terms are naturally included in . Then , between proteins and , the protein semantic similarity ( pSS ) is defined as follows [7]: | Network analyses of large-scale interactomes have been a great aid in advancing our understanding of cellular functioning and organization . Here , we examine one of the most basic and intensely-studied structure-to-function relationships observed in cellular networks: that between the number of interactions a protein has and its tendency to be essential . We develop a new computational framework to systematically analyze essential proteins within their cellular context by explicitly incorporating functional information . We apply this framework to the yeast interactome and demonstrate that the previously observed positive relationship between interaction degree and essentiality is largely due to intramodular interactions . Further , essentiality appears to be a modular property of protein complexes and not more broadly of biological processes . Within an essential complex , its essential proteins tend to have more interactions , especially intra-complex interactions , than its non-essential proteins . Finally , in a computationally inferred module-level interaction network , essential complexes and processes tend to have higher interaction degrees than their non-essential counterparts . In summary , we show a relationship between connectivity and essentiality not only at the protein level , but also within modules and at the module level , with complexes and processes that are essential tending to interact with many functional groups . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"systems",
"biology",
"genomics",
"functional",
"genomics",
"biology",
"computational",
"biology",
"genetics",
"and",
"genomics"
] | 2013 | From Hub Proteins to Hub Modules: The Relationship Between Essentiality and Centrality in the Yeast Interactome at Different Scales of Organization |
As humans age , they experience a progressive loss of thymic function and a corresponding shift in the makeup of the circulating CD8+ T cell population from naïve to memory phenotype . These alterations are believed to result in impaired CD8+ T cell responses in older individuals; however , evidence that these global changes impact virus-specific CD8+ T cell immunity in the elderly is lacking . To gain further insight into the functionality of virus-specific CD8+ T cells in older individuals , we interrogated a cohort of individuals who were acutely infected with West Nile virus ( WNV ) and chronically infected with Epstein Barr virus ( EBV ) and Cytomegalovirus ( CMV ) . The cohort was stratified into young ( <40 yrs ) , middle-aged ( 41–59 yrs ) and aged ( >60 yrs ) groups . In the aged cohort , the CD8+ T cell compartment displayed a marked reduction in the frequency of naïve CD8+ T cells and increased frequencies of CD8+ T cells that expressed CD57 and lacked CD28 , as previously described . However , we did not observe an influence of age on either the frequency of virus-specific CD8+ T cells within the circulating pool nor their functionality ( based on the production of IFNγ , TNFα , IL2 , Granzyme B , Perforin and mobilization of CD107a ) . We did note that CD8+ T cells specific for WNV , CMV or EBV displayed distinct functional profiles , but these differences were unrelated to age . Collectively , these data fail to support the hypothesis that immunosenescence leads to defective CD8+ T cell immunity and suggest that it should be possible to develop CD8+ T cell vaccines to protect aged individuals from infections with novel emerging viruses .
CD8+ T cells can provide robust protection against pathogens and tumors . As a result , significant effort has been invested into developing vaccines that elicit protective CD8+ T cell memory responses . It is generally believed that immunological function decreases with advanced age , a phenomenon known as immunosenescence , rendering older individuals at higher risk of infection . While vaccination would seem to be an appropriate intervention to improve protective immunity , several reports have demonstrated that older individuals mount impaired responses to conventional vaccines , suggesting that alternate platforms or strategies may be required . Notably , antibody responses to influenza and tick-borne encephalitis vaccines were impaired in the elderly [1] , [2] , [3] . With regard to T cell immunity , recent reports from a large-scale immunization study with a live vaccine against varicella zoster demonstrated that while it is possible to boost zoster-specific CD4+ T cells to a protective level in individuals >60 years of age , vaccine responsiveness did appear to wane in individuals >75 years of age [4] , [5] . These data support the concept that immunosenescence may be an issue to overcome in the development of effective vaccines for elderly individuals; however , further research is required to truly understand the extent of immune dysfunction in older humans . Alterations in the CD8+ T cell compartment are among the most common characteristics in the elderly T cell repertoire and are thought to reflect an impaired ability to control infection [6] , [7] . The aged CD8+ T cell population is characterized by a high proportion of CD28− cells ( often co-expressing NK markers , such as CD57 ) , which are believed to reflect highly differentiated T cells that lack the capacity to proliferate [1] , [8] . In some cases , the CD8+ CD28− T cell population comprises an oligoclonal expansion of CMV-reactive cells , suggesting that chronic infections may preoccupy the immune response in the elderly , leading to a CD8+ T cell repertoire with limited diversity [9] , [10] , [11] , [12] . The sum of these observations suggests that the CD8+ T cell population in the elderly is compromised in its capacity to respond to novel infections . However , the exact relationship between the global phenotypic changes in the CD8+ T cell compartment that appear with age and the functionality of antigen-specific CD8+ T cells is poorly defined . Further , there is a paucity of data regarding the ability of the elderly to mount CD8+ T cell responses to novel infections . Although it is generally assumed that age-associated changes in the CD8+ T cell compartment may explain the heightened risk of elderly individuals to infection , experimental data are sparse . Herein we provide one of the few studies in humans that demonstrate the impact of age on CD8+ T cell immunity to pre-existing and novel viral infections . West Nile virus ( WNV ) emerged as a novel human pathogen in the Northern hemisphere in 1999 , and since then has caused numerous viral outbreaks across North America [13] , [14] , [15] . From 2003–2008 , we collected sequential blood specimens from >100 people acutely infected with WNV with an age distribution ranging from 19–85 years . Given this age range , we reasoned that our cohort would be suitable to study the relationship between age and the development of virus-specific CD8+ T cells following a novel acute infection . In our original report of this cohort , we observed that age did not influence the magnitude or breadth of the memory T cell response to WNV [16] , suggesting that age may not impair the development of CD8+ T cell immunity against acute infections . Our previous work did not address the longevity or functionality of CD8+ T cell memory that develops following WNV infection . Thus , it remained possible that the older members of our cohort failed to develop a CD8+ T cell memory pool that was functionally equivalent to the younger members . In this current report , we have examined the polyfunctionality of the WNV-reactive CD8+ T cell population at later time points post-infection . We have also examined memory responses to EBV and CMV within this cohort , as these lifelong infections may differentially impact the functionality of memory CD8+ T cells . Our results reveal that although the memory CD8+ T cells display distinct polyfunctional states that are virus-specific , we observed no impact of ageing on polyfunctionality . These studies have revealed that memory CD8+ T cell immunity in older individuals is intact and suggest that vaccine development should focus on other parameters that may be defective in the elderly .
For these studies , we have examined the CD8+ T cell memory responses from a cohort of 72 patients who were naturally infected with West Nile virus ( WNV ) . We stratified our cohort into 3 groups: young ( <40 years of age; n = 21 ) , middle-aged ( 41–59 years of age; n = 25 ) and aged ( >60 years of age; n = 26 ) . To confirm that these cohorts displayed the expected age-associated changes in the CD8+ T cell compartment , we compared the phenotype of CD8+ T cells among the three different age groups . Significantly higher frequencies of CD8+ CD28− and CD8+ CD28− CD57+ cells were observed within the aged cohort ( Figure 1A and 1B ) . Likewise , we noted that the CD45RA+ CD28+ CD8+ T cell population was significantly decreased in the naïve T cell pool in middle-aged and aged populations compared to the young population ( Figure 1C ) . We also observed a significant reduction in the presence of naïve ( CD45RA+ CCR7+ ) T cells in the aged subjects ( Figure 1D ) . These observations confirm that our aged cohort displayed the expected immunosenescent phenotype within the CD8+ T cell compartment . We first sought to confirm our previous results showing that age did not impact the magnitude of the WNV-specific CD8+ T cell response . In our original study , we employed ELISPOT to monitor WNV-specific CD8+ T cells . However , cytokine production by ELISPOT cannot be attributed solely to CD8+ T cells . Therefore , in the current study , we employed flow cytometry to specifically identify cytokine-producing CD8+ T cells and provide a more accurate assessment of the functionality of the virus-specific CD8+ T cells . For these experiments , we used specimens obtained 6–7 months following WNV infection . Since our study population consisted of individuals with diverse HLAs , virus-specific CD8+ T cells were identified based on cytokine production following stimulation with a broad collection of immunodominant peptides that span the breadth of HLAs expressed by our cohort . Briefly , for these experiments , freshly thawed PBMCs were stimulated with pools of dominant epitope peptides derived from WNV , CMV or EBV and cytokine production ( IFN-γ , TNF-α and IL-2 ) was measured on a per-cell basis using flow cytometry . While the CD8+ T cells produced varying amounts of cytokine following peptide stimulation , we did not observe any peptide-specific CD8+ T cells that could produce IL-2 or TNF-α in the absence of IFN-γ . Since all of our peptide-stimulated CD8+ T cells expressed IFN-γ , which is considered to be the cytokine that mediates the primary anti-viral response by the adaptive immune system [17] , we defined “virus-specific” CD8+ T cells as those which produced IFN-γ following stimulation with specific peptide epitopes ( see Table S1 for a complete list of measured IFN-γ frequencies ) . The frequencies of CD8+ T cells specific for either WNV , CMV or EBV was similar among all age groups ( Figure 2 ) . We did note a trend towards elevated frequencies of CMV- and EBV-specific CD8+ IFN-γ+ T cells in the middle-aged and aged cohorts relative to the young cohort , but this trend did not reach statistical significance ( Figures 2B–D ) . While our data indicate that older individuals mount CD8+ T cell responses to acute infection ( i . e . WNV ) that are equivalent in magnitude to younger individuals , it is possible that the responses display different stability . To address this question , we measured the frequencies of WNV-specific CD8+ T cells in our cohort at 2 additional time points: baseline ( early memory; average of 35 days after symptom onset ) and 2–4 years post symptom onset ( late memory ) . The magnitude of the WNV-specific CD8+ T cell response was highest at 1 month and declined thereafter ( Table 1 ) . Importantly , the magnitude of the WNV-specific CD8+ T cells was equivalent among the various age cohorts at all 3 time points , suggesting that the longevity of the memory CD8+ T cell response is not age-dependent . For these experiments , we also examined the production of TNF-α and IL-2 following peptide stimulation . As stated above , we did not observe any CD8+ T cells that produced TNF-α or IL-2 in the absence of IFN-γ following WNV peptide stimulation . We observed that 30%–50% of the WNV-specific CD8+ T cells were IFN-γ+ TNF-α+ double positive ( Table 1 ) . We also noted that only a fraction of WNV-specific CD8+ T cells could produce IL-2 and this did not increase with time . No difference was observed in the frequencies of TNF-α- or IL-2-producing WNV-specific CD8+ T cells among the 3 age groups at any time point ( Table 1 ) . In the previous paragraph , WNV-specific CD8+ T cells were crudely separated into 3 populations based on the expression of either IFN-γ , TNF-α , or IL-2 . To gain further insight into the polyfunctional nature of the virus-specific CD8+ T cells , we also measured the cytotoxic capacity of the CD8+ T cells by granzyme B expression , upregulation of perforin and mobilization of CD107a ( a measure of degranulation ) following peptide stimulation of WNV samples obtained 6–7 months post symptom onset . Similar to our observations with cytokine production , we did not observe any peptide-specific CD8+ T cells that could upregulate perforin or mobilize CD107a in the absence of IFN-γ . Therefore , all functional parameters have been defined relative to the expression of IFN-γ . Polyfunctionality of antigen-specific CD8+ T cells was defined using a newly developed computational analysis of flow cytometry data: FLOCK ( FLOw Cytometry without K ) , publicly available in the Immunology Database and Analysis Portal – ImmPort ( www . immport . org ) . FLOCK utilizes a density-based clustering approach and algorithms to define biologically relevant populations from multiparametric data sets without the bias of manual gating [18] . Using FLOCK , we identified 16 distinct functional populations from IFN-γ+ CD8+ T cells for all three antigens ( WNV , CMV and EBV ) , that were defined as negative ( neg ) , low ( lo ) , intermediate ( int ) , and high ( hi ) , based on the signal intensity of each marker ( Figure 3 ) . All populations were derived from IFN-γ+ events , thus there were no IFN-γ negative events . To address the question of whether advanced age impacted the development of polyfunctional memory CD8+ T cell responses , we analyzed the large data set comprising functional population frequencies ( FLOCK identified ) by Principal Component Analysis ( PCA ) . PCA is a linear technique that transforms data of interrelated variables into a set of uncorrelated principal components ( PCs ) while maintaining the original variation of the data set in reduced dimensionality [19] . The polyfunctional analysis of antigen-specific CD8+ T cells was separated into two parameters: cytokine functional populations ( C2–C7; Figure 4 ) and cytotoxic functional populations ( C8–C17; Figure 4 ) . Consequently , two PCA analyses were generated per antigen . PCA plots comprising cytokine functional populations for WNV , CMV and EBV were generated using the top two PCs that accounted for 78% , 86% and 78% of the overall variance , respectively . Functional populations C2 ( IFNγlo TNFαlo IL2lo ) and C3 ( IFNγlo TNFαneg IL2neg ) ; C4 ( IFNγint TNFαhi IL2neg ) and C5 ( IFNγint TNFαhi IL2lo ) ; C6 ( IFNγhi TNFαhi IL2hi ) and C7 ( IFNγint TNFαint IL2int ) tended to cluster , indicating a strong positive correlation ( Figure 4 ) . The vector clustering would suggest that these functional populations are the same or very similar . We observed no specific age clustering , suggesting no relationship between age and CD8+ T cell function based on cytokine production . PCA plots depicting the cytotoxic functional phenotypes for WNV , CMV and EBV- specific CD8+ T cells were generated using the top two PCs and account for roughly 65% of the overall variance ( Figure 4 ) . It is important to note that the interpretation of the data did not change when we examined 3-D plots of the first 3 PCs ( approximately 80% of the total variance ) and for simplicity of interpretation we used biplots to explain these data . We observed high positive correlations between populations C9 ( IFNγint GrBlo Prfneg CD107ahi ) and C11 ( IFNγhi GrBlo Prflo CD107aint ) ; C13 ( IFNγhi GrBlo Prflo CD107ahi ) and C16 ( IFNγhi GrBlo Prfneg CD107ahi ) , which suggests that they might belong to the same functional population but were segregated in into two based on automated binning by FLOCK analysis . We also find that population C10 ( IFNγhi GrBint Prflo CD107aint ) contributes very little to the overall variance of the system for WNV and CMV since its vector length is small relative to the other defined cytotoxic phenotypes . Furthermore , as observed for antigen-specific CD8+ T cell cytokine function , we found no evidence that CD8+ T cell cytotoxicity was affected by age . These results suggest that while different combinations of cytotoxic markers define virus-specific CD8+ T cell responses , they show no linear relationship with age . Preliminary analysis of polyfunctional WNV , CMV , and EBV-specific IFN-γ+ CD8+ T cells ( producing cytokines; IL2 and TNF-α , and mobilizing cytotoxic mediators; GrB , perforin and CD107a ) revealed that WNV and CMV polyfunctional responses were more similar than EBV-specific polyfunctional CD8+ T cells ( Figure 5A ) . On average , EBV-specific CD8+ T cells were better producers of IL2 but failed to upregulate Granzyme B or perforin in comparison to WNV and CMV-specific CD8+ T cells ( Figure 5A ) . We next performed the Kolmogorov-Smirnov ( KS ) test to determine whether WNV , CMV or EBV-specific CD8+ T cell functional phenotypes defined by FLOCk were drawn from the same distributions . The KS test is based on the null hypothesis that the samples are drawn from the same distribution , thus larger p-values suggest that the two sets are similar . A comparison of the evaluated KS statistics between the different viral antigens for all functional phenotypes showed that WNV and CMV were more functionally similar than EBV ( Figure 5B ) . For example , T cell phenotype identified in population C4 ( IFNγint TNFαhi IL2neg ) for WNV and CMV had a calculated p value of 0 . 889 suggesting a very similar distribution of this functional population , which was not observed for EBV . We further noted that more than 40% of the functional cells ( IFNγ+ ) simultaneously produced TNFα+ for all three antigens , but there was a difference in the ability to produce IL-2 . WNV-specific CD8+ T cells produced the least IL-2 , EBV-specific CD8+ T cells produced the highest amounts of IL-2 and CMV-specific CD8+ T cells displayed an intermediate phenotype ( Figure 5A ) . The KS analysis of the cytotoxic functional populations revealed a striking similarity between CMV- and WNV-specific T cells , where populations C8 ( IFNγloGrBloPrfloCD107ahi ) and C16 ( IFNγhiGrBloPrfnegCD107ahi ) were distributed similarly ( p = 0 . 909 and p = 0 . 882 , respectively ) . Overall , a higher frequency of CMV- and WNV-specific memory T cells were cytotoxic and polyfunctional ( GrB+ Prf+ CD107a+ ) in comparison to the EBV-specific T cells , which became CD107ahi following peptide stimulation but remained low in terms of GrB and Perforin expression ( Figure 5A ) . Using PCA biplots , we were able to discriminate antigenic stimulation ( WNV , CMV or EBV ) based on the resultant functional phenotypes ( Figure 5B ) . Corroborating the KS distribution analysis , the PCA showed that CMV and WNV are indiscriminant based on the above-mentioned functional populations , whereas EBV-specific functional phenotypes cluster separately . This effect of EBV segregating away from WNV and CMV was especially evident when cytotoxic populations were analyzed by PCA ( Figure 5B ) . Altogether , it does not appear that age has an impact on the development of memory CD8+ T cells with the capacity to elaborate multiple functions . Rather , it appears that the polyfunctional profile of virus-specific CD8+ T cells appears to be a function of the pathogen .
Contrary to the suggestion that susceptibility to new infections in the aged occurs due to insufficient CD8+ T cell immunity as a result of diminished frequencies of naïve CD8+ T cells and/or dysfunctional CD8+ T cell memory [20] , [21] , we have shown that aged individuals mount CD8+ T cell memory responses to a novel viral agent that are equivalent to young individuals . In fact , extensive analysis of CD8+ T cell functional parameters revealed no relationship between age and the capacity to produce cytokines or mobilize cytotoxic mediators in response to stimulation by peptides derived from viruses responsible for both acute ( WNV ) and chronic ( CMV , EBV ) infections despite clear evidence of an immunosenescent phenotype in the bulk CD8+ T cell pool ( elevated frequencies of CD28− CD57+ cells and decreased frequencies of CD45RA+ CCR7+ cells relative to the younger members of the cohort ) . Thus , although the members of our aged cohort displayed expected age-related changes in the composition of the CD8+ T cell compartment , these alterations did not manifest as a defect in functional virus-specific immunity , even when the primary virus infection occurred in old age , as in the case of WNV . A recent study revealed that infection of middle-aged and old macaques with Rhesus CMV ( RhCMV ) produced RhCMV-specific CD8+ T cells with comparable functionality in both age groups [22] , supporting the concept that anti-viral CD8+ T cell responses may not be dysfunctional in aged individuals . In contrast , immunization with modified vaccinia Ankara ( MVA ) elicited weaker CD8+ T cell responses in old macaques compared to young macaques [7] . While the results of the MVA experiments may seem at odds with our observations , the authors of this latter report employed live vaccinia virus to stimulate MVA-specific CD8+ T cells in vitro for their functional assays . In contrast , our current report and the report on RhCMV employed synthetic peptides that do not require additional processing for presentation to CD8+ T cells . Since stimulation of CD8+ T cells with live vaccinia virus relies upon the infection , expression and processing of antigen by the cells in the test sample , it is possible that the CD8+ T cell response was intact but antigen presentation by the cells used to present vaccinia antigens in the in vitro assay were defective in the aged monkeys . The authors argued that DCs were not affected by the age of the monkeys; however , they only investigated a limited number of parameters and they did not examine antigen processing through the classical MHC class I pathway . Therefore , we cannot discount a possible role for defective antigen presentation in their in vitro stimulation . Another possible explanation for the differences may stem from the nature of the immunogens . MVA is a variant of vaccinia virus that replicates poorly in primate cells , whereas RhCMV and WNV replicate effectively in primate cells . Therefore , effective stimulation of CD8+ T cells responses in the elderly may rely upon the nature of the infectious agent . This will be an important point to consider with regard to vaccine design . It has been proposed that chronic CMV infection may drive immune senescence due to repeated oligoclonal expansions of CMV-specific CD8+ T cells leading to overpopulation of the memory T cell pool [7] , [11] , [23] , [24] , [25] and ultimately limiting the ability of the aging individual to combat previously encountered or novel viral infections [25] . However , a recent report has suggested that the size of the CD8+ T cell compartment may increase with age to accommodate expanding memory T cell populations without depleting CD8+ T cells with other specificities [26] . Interestingly , this phenomenon was not reflected within the peripheral blood where the expanding memory populations increased in frequency at the expense of T cells with other specificities . Rather , the expansion of antigen-specific memory CD8+ T cells was accommodated by increased numbers of CD8+ T cells present within the tissues , suggesting that measures of CD8+ T cell frequencies within the peripheral blood may not accurately reflect the true composition of the CD8+ T cell pool . Although it is relatively easy to measure CD8+ T cells present in the tissues in murine studies , addressing this concept in humans is not trivial . Nevertheless , in light of this recent report , the apparent decline in available naïve CD8+ T cells in the peripheral blood of individuals with evident expansion of CMV-specific CD8+ T cells may not truly reflect a corresponding decrease in the availability of naïve T cells in the lymphoid tissues , where primary responses to viruses are initiated . Similar to previous reports , we have observed a trend towards higher frequencies of CMV- and EBV-reactive CD8+ T cells in the aged cohort . However , this trend did not achieve statistical significance and not all aged individuals displayed an expanded CMV- or EBV-specific CD8+ T cell pool . Similar results have been reported by others [27] , [28] . Importantly , in all of these reports , the functionality of the CMV-specific CD8+ T cells did not change with age ( the other reports did not investigate EBV-specific CD8+ T cells ) . It is notable that all of these reports employed functional analyses to define the CMV-specific CD8+ T cells . In contrast , when CMV- and EBV-specific CD8+ T cells were quantified using MHC multimers , it was noted that dysfunctional populations of CMV- and EBV-specific CD8+ T cells accumulate with age based on tetramer staining and IFN-γ production [29] , [30] . The implications of these dysfunctional cells are unclear as these aged individuals successfully control both CMV and EBV infections and , based on our results , are able to mount effective CD8+ T cell responses to novel infections . We noted a number of mid-aged and aged individuals with frequencies of CMV-reactive CD8+ T cells that represented more than 9% ( 9 subjects ) of the circulating CD8+ T cell pool , but we did not observe any relationship between expanded CMV-specific CD8+ T cells and impaired generation of WNV-specific CD8+ T cells , indicating that CMV expansions do not limit the ability of the host to respond to a novel infection , consistent with the report of Vezys et al . [26] . Detailed comparison of the functional CD8+ T cell response between the different viruses ( WNV , CMV and EBV ) revealed interesting differences in functional profiles , corroborating previous reports examining virus-specific CD8+ T cell immunity in humans [31] , [32] , [33] . Striking similarities in both phenotype and cytotoxic profile were observed between memory WNV- and CMV-specific CD8+ T cells , despite the fact that the former is an acute infection and the latter is a chronic infection . The majority of WNV- and CMV-specific CD8+ T cells displayed a phenotype consistent with terminally-differentiated effectors ( CD45RA+ CD28− ) whereas EBV-specific CD8+ T cells were mostly less differentiated ( CD45RA− CD28+ ) ( data not shown ) . Consistent with the phenotype and the differentiation status , CMV-specific CD8+ T cells produced high levels of GrB and perforin but failed to produce IL-2 , whereas EBV-specific CD8+ T cells failed to produce perforin and had less GrB but significantly more IL-2 ( Figure 5A ) ; a similar dichotomy in the production of perforin and IL-2 was described in our previous work with a smaller cohort of patients [33] . This is consistent with previous reports that show the expression of cytotoxic enzymes is related to cellular maturity , such that CD45RA+/− CD28− cells express high levels of cytotoxicity due to highly differentiated phenotype and CD45RA+/− CD28+ T cells express little cytotoxic attributes [33] , [34] , [35] , [36] . Collectively , we demonstrate here that aging individuals are capable of mounting polyfunctional memory CD8+ T cell responses to a novel pathogen , which has significant implications for vaccine development for the elderly . Most of our current understanding on the relationship of aging to vaccination has relied upon measurements of antibodies following vaccination and it is clear that the serological response in the elderly is attenuated [1] , [2] , [3] . In striking contrast , our results described herein and in our previous report [16] reveal that the aged can mount a robust , polyfunctional CD8+ T cell response to novel pathogens while sustaining a robust polyfunctional responses to chronic infections . Collectively , our data suggest that vaccination in older humans should focus on CD8+ T cell immunity and that live vaccines should be considered as the platform of choice . The results presented here entice our curiosity and desire to better understand the aging immune system for the purpose of developing much needed vaccines for our greatly expanding aging population .
This research was approved by the Hamilton Health Sciences/McMaster Health Sciences Research Ethics Board that operates in compliance with the ICH Good Clinical Practice Guidelines and the Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans and Division 5 Health Canada Food and Drug Regulations . All patients in this study provided informed written consent . Seventy-two patients were enrolled into the study following detection of serum WNV IgM by public health laboratories after presentation of WNV-related symptoms . Serology for WNV was assessed by plaque reduction neutralization assay as described previously [37] . Recruitment of patients occurred over a period of 5 years ( 2003–2007 ) . This trial was reviewed and approved by the Research Ethics Board at McMaster University . Patients were entered into our study within 1 month following symptom onset ( median = 30 days , ranging from 4–100 post symptom onset ) and blood was collected on the first visit ( baseline sample ) and once every month thereafter for a period of one year . Twenty-five patients were contacted 2–4 years post symptom onset and their blood was collected at convalescence of disease . The population consisted of 37 men and 35 women ranging in age from 19 to 85 years . Patients were subdivided into three cohorts for these experiments based on age; young <40 years of age , mid-aged 41–59 years , and aged >60 years of age ( Table S1 ) . Blood samples were drawn into heparanized tubes and PBMC were isolated from the blood by centrifugation on Ficoll ( Amersham Pharmacia ) . PBMC were cryopresrved in RPMI 1640 containing 12 . 5% human serum albumin ( Sigma-Aldrich ) and 10% DMSO according to the method described by Disis et al . [38] . WNV peptides used for the stimulation of PBMCs were identified previously [16] and 13 of commonly immunogenic peptides were pooled together for the purpose of having a single WNV stimulation that would encompass the vast majority of reactivities within the cohort . Peptides were either deconvoluted to a minimal epitope of 8–9 amino acids or were uses as a 15-mer . CMV stimulation consisted of 168 identified CMV-specific CD8+ T cells epitopes pooled into a single pool . Likewise , EBV stimulation consisted of 91 identified EBV-specific CD8+ T cell epitopes pooled together into a single pool . PBMC were thawed and placed immediately into 37°C pre-warmed complete RPMI ( Invitrogen ) supplemented with 10% fetal bovine serum ( FBS ) , 2 mM L-glutamine , 50 µM 2-ME , 10 µM HEPES , 100 U/ml penicillin , and 100 µg/ml streptomycin . Thawed PBMC were cultured overnight at 37°C incubator . The cells were subsequently harvested , counted , and viability was assessed by trypan blue exclusion . Cells were aliquoted ( 2–2 . 5×106 cell/well ) into round-bottom , 96-deep-well plate ( Costar ) ; peptides were added to a final concentration of 2 µg/ml and were incubated for 1 hr at 37°C . DMSO diluted in cRPMI was used as a peptide-non-specific negative control . Brefeldin A and Monensin A ( BD Biosciences ) were added to the cell/peptide mixture as per manufacturer's instructions and were incubated for an additional 4 hrs . At this point , cells were pelleted and washed in 10 µM EDTA . The cells were first stained with Near IR viability stain ( Invitrogen , Molecular Probes ) and subsequently with different antibody cocktails depending on the analysis . Cytokine analysis cocktail comprised of anti-human CD3-Qdot 605 ( Molecular Probes ) , CD8-Alexa flour 700 , CD45RA-PE Texas red ( Beckman Coulter ) , CD28-PE Cy5 , CD4-Pacific Blue , CD14-Pacific Blue , CD19-Pacific Blue surface antibodies . Cells were then permeabilized with Cytofix/Cytoperm ( BD Biosciences ) , and intracellular cytokines were identified using anti-human IFNγ-APC , TNFα-FITC , and IL-2-PE [Note: all flow cytometry reagents were obtained from BD Biosciences unless otherwise specified] . Fluorescence data were acquired using LSRII flow cytometer ( BD Biosciences ) and 500 , 000–1 000 , 000 events based on the live lymphocyte gate were collected per sample . Data were analyzed using FlowJo . A positive response was measured as the IFN-gamma frequency greater than 0 . 05 and three fold above DMSO background . PBMCs were cultured and stimulated as described above however in addition to the stimulatory peptides , anti-human CD107a-PE conjugated antibody ( BD Biosciences ) was added to the cells at the beginning of the stimulation for 1 hr . Brefeldin A and Monensin A ( BD Biosciences ) were added to the cell/peptide/CD107a mixture as per manufacturer's instructions and were incubated for an additional 4 hrs . The cells were subsequently stained with a Near IR viability stain ( Invitrogen , Molecular Probes ) as per manufactures instructions followed by the cytotoxic antibody cocktail: anti-human surface antibodies [CD8-PerCP Cy5 . 5 ( eBiosciences ) , CD4-Alexa Flour700 ( BD Biosciences ) , CD19 and CD14-Alexa Flour 700 ( eBiosciences ) ] and intracellular anti-human antibodies [IFNγ-APC , Granzyme B-FITC and Perforin-Pacific Blue ( conjugated to Pacific Blue in house using standard conjugation protocols ) ] . The perforin antibody detects de novo as well as pre-formed perforin and when used in conjuction with IFNγ following in-vitro peptide simulation we are able to determine the frequency of de novo formed perforin only . Fluorescent data was acquired using the LSR II as described above . An aliquot of thawed patient PBMCs ( 0 . 5–1×106 cell/stain ) was used for the purpose of phenotyping the cells . Cells were stained in round-bottom 96-well plates with anti-CD3-APC-H7 , CD8-Alexa Flour 700 , CD4-Pacific Blue , CD45RA-PE Texas Red ( Beckman Coulter ) , CD28-PE , CD57-FITC , and anti-CCR7-PE Cy7 [Unless otherwise stated all antibodies were purchased from BD Biosciences] . Analysis of surface marker staining was done by LSR II flow cytometer and data was analyzed using FlowJo software . FLOCK is an automated computational approach publically available at the Immunology Database and Analysis Portal – ImmPort ( www . immport . org ) , which utilizes algorithms and density-based clustering to identify cell subsets . FLOCK analysis is comprised of five components: data preprocessing , grid-based density clustering , cross-samples comparison , result visualization , and population statistics calculations . Detailed methodology for FLOCK analysis can be found in [18] . In summary , binary . fcs files specifically gated on live/singlet/CD3+/CD8+ CD4−/IFNγ+ events were converted to tab-delimited ACSII text format and exported from FlowJo ( Tree Star ) in a data matrix file . Samples ( 6–7 months post WNV symptom onset ) were considered positive if following peptide stimulation they expressed IFNγ frequency above 0 . 05 and 3 fold above DMSO background and consequently were included in the FLOCK analysis . This was the means by which our data was normalized . The exported CD8+ IFNγ+ events were than subjected to density-based grouping based on expression of IFNγ , TNFα and IL2 for determination of cytokine populations; and IFN-γ , CD107a , GrB and Perforin for determination of cytotoxic populations depending on the distances between each point and where its coordinates lie in the defined grid . FLOCK identified 16 cell populations: 6 defining cytokine populations and 10 defining cytotoxic populations ( Figure 3 ) . Population centroids ( the average of coordinates of a given set of points ) were applied to multiple samples in a cross-sample analysis to enable population comparisons between WNV , CMV , and EBV-specific CD8+ memory T cell populations . The data are presented as mean values . Simple descriptive statistics ( means , standard deviations , Students t test and regression analysis ) were calculated using GraphPad Prism version 1 . 0 . Box and whiskers plots are calculated at 95% confidenced interval and generated using GraphPad Prism version 1 . 0 . The large data set comprising functional population frequencies was analyzed by Principal Component Analysis ( PCA ) , which is a linear technique that transforms data of interrelated variation of the data set in reduced dimensionality [19] . To graphically reveal clustering , multi-collinearity and outliers of our data set following PCA we used a biplot consisting of top −2 PCs . The biplots show both the samples and features of the data set , where each sample is displayed as a point in a two-dimensional plane , and each functional population ( defined by FLOCk clustering ) is presented as a vector . The length of each vector indicates the approximate variance of the specific functional population . The distance between two points is an approximate of the Euclidean distance between their associated functional phenotype . Thus , samples that cluster together are interpreted as similar . Conversely , observing no clustering of points implies very little similarity among the data points . The correlation between any two functional phenotypes can be approximated by the angles between them , where angles of 90 or 270 degrees apart show correlations approaching zero , and angles of 0 or 180 degrees show a correlation of 1 or −1 , respectively . The Kolmogorov-Smirnov ( KS ) test was used to determine whether WNV , CMV or EBV-specific CD8+ T cell functional phenotypes were drawn from the same or different distributions . The KS statistic quantifies a distance between the empirical distribution functions of two samples and is based on the null hypothesis that the samples are drawn from the same distribution if the p value approaches 1 . | The prevalence and severity of viral infections increases with advanced age , a phenomenon associated with a defective immune system . The thymic output of naïve T cells declines as we age and it is this lack of naïve T cells that is believed to contribute to the inability of the aged to respond to novel infections and develop subsequent memory T cell responses . Here we show that individuals aged 60+ are capable of developing memory CD8+ T cells to West Nile virus , novel pathogen , indistinguishable in terms of polyfunctionality to those of subjects <60 years of age . Furthermore , we show that chronic and life-long infections with CMV and EBV result in similar polyfunctional virus-specific memory CD8+ T cell responses in subjects of all age groups . Our work demonstrates that aged individuals can elicit functional memory CD8+ T cell responses to a new pathogen while maintaining polyfunctional CD8+ T cells against recurrent chronic virus infections . Current vaccine platforms , which rely upon inactivated pathogens or recombinant subunits , are poorly effective in the aged . Our data suggest that live viruses may be more effective vaccine platforms in older humans . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"infectious",
"diseases",
"clinical",
"immunology",
"immunology",
"biology"
] | 2012 | The Polyfunctionality of Human Memory CD8+ T Cells Elicited by Acute and Chronic Virus Infections Is Not Influenced by Age |
Huge research effort has been invested over many years to determine the phenotypes of natural or artificial mutations in HIV proteins—interpretation of mutation phenotypes is an invaluable source of new knowledge . The results of this research effort are recorded in the scientific literature , but it is difficult for virologists to rapidly find it . Manually locating data on phenotypic variation within the approximately 270 , 000 available HIV-related research articles , or the further 1 , 500 articles that are published each month is a daunting task . Accordingly , the HIV research community would benefit from a resource cataloguing the available HIV mutation literature . We have applied computational text-mining techniques to parse and map mutagenesis and polymorphism information from the HIV literature , have enriched the data with ancillary information and have developed a public , web-based interface through which it can be intuitively explored: the HIV mutation browser . The current release of the HIV mutation browser describes the phenotypes of 7 , 608 unique mutations at 2 , 520 sites in the HIV proteome , resulting from the analysis of 120 , 899 papers . The mutation information for each protein is organised in a residue-centric manner and each residue is linked to the relevant experimental literature . The importance of HIV as a global health burden advocates extensive effort to maximise the efficiency of HIV research . The HIV mutation browser provides a valuable new resource for the research community . The HIV mutation browser is available at: http://hivmut . org .
Human immunodeficiency virus ( HIV ) , the causative agent of acquired immunodeficiency syndrome ( AIDS ) , infects millions of people worldwide and , to date , has been responsible for over 25 million deaths [1] . The clinical importance of the virus has prompted substantial funding of HIV/AIDS research across many diverse clinical , therapeutic ( drug design , vaccine production ) and basic research fields . This research has produced an extensive catalogue of HIV literature and consequently finding literature pertinent to a particular topic is a difficult task . Researchers are often interested in the phenotypic variation resulting from naturally occurring single nucleotide polymorphism or directed mutagenesis in the HIV genome . Traditionally , mutation data for a particular protein or region must be manually collected by trawling literature repositories such as PubMed using author names , protein/gene names , keywords or a mixture of all three . The scale of the HIV literature ( over 270 , 000 articles ) makes such an approach inadequate . Several valuable online resources have provided mutation data to researchers by manually curating polymorphism and mutagenesis data from HIV studies . These include the Stanford Drug Resistance database [2] , which curates mutations related to drug resistance , the UniProt knowledgebase [3] , which manually annotates articles describing mutagenesis of HIV proteins and the Los Alamos HIV Database , which annotates various sources of HIV data including epitope variants and escape mutations ( http://www . hiv . lanl . gov/ ) . However , these resources are limited in scope because manual curation cannot feasibly be carried out on all of the available literature . The technology exists to quickly computationally scan , annotate and organise scientific literature and these techniques should be applied to facilitate the work of HIV researchers [4] , [5] . Consequently , it is surprising that so few resources are available to access the available literature in an organised and structured way . This incongruity can partly be explained by the strict licensing agreements with scientific publishers that prohibit the bulk download and computational processing of scientific research literature . Fortunately , recent pressure from government and scientific bodies and the rise of open access publishing has softened the stance of publishers and many are now receptive to waiving these restrictions . Such advances will pave the way for many large-scale literature text-mining projects and will likely change the way we access scientific literature . Here we have applied text-mining techniques to extract data on polymorphisms and mutations from the available HIV literature . We have organised this data in a protein and residue-centric way and have made it available through an online resource , the HIV mutation browser ( http://hivmut . org ) . This publicly available resource will simplify the task of virologists attempting to identify the relevant literature for their research , thereby aiding experimental design and reducing replication of efforts .
We identified ∼270 , 000 articles containing the search term “HIV” or “Human Immunodeficiency Virus” indexed in the PubMed database ( from a total of ∼23 million publications ) . We retained 120 , 899 of these articles , published across 2 , 614 journals , representing approximately 45% of the total ( see materials and methods , Figure 1 ) . For the remaining ∼150 , 000 citations , permission for computational processing of articles was not obtained from the publisher . The 120 , 899 articles from participating publishers were text-mined for mutagenesis or polymorphism information , and the mutations were mapped to particular residues within the HIV proteome . This required the development of a method to retrieve the text of these articles , scan the articles for patterns that are widely used to describe directed mutations in mutagenesis experiments or polymorphisms , and to map these mutations to the correct position in the correct protein ( see materials and methods ) . A total of 7 , 608 distinct mutations ( a unique non-wildtype amino acid at a given residue in a given protein ) were collected . As each mutation can be described in multiple articles and each article can describe multiple mutations , the 7 , 608 distinct mutations were defined by 43 , 264 unique references to 5 , 267 articles . The identified mutations shed light on the nature of the HIV research effort of the last decades . On the one hand it has been broad in scope: 2 , 520 of the 3 , 118 residues in the HIV proteome have one or more associated references to a mutation in the repository . On the other hand it has been narrow in focus: the coverage is far from uniform and certain regions such as the catalytic sites of the protease and reverse transcriptase , as well as host interaction interfaces , are much more highly studied ( Figure 2 ) . The above analysis resulted in a database within which each reference to a mutagenesis experiment or polymorphism in a citation is indexed using three pieces of information: the protein in which the mutation is present , the position in the protein which has been mutated , and the non-wildtype amino acid to which the wildtype residue has been mutated . To make this data accessible to virologists in a simple , intuitive and informative manner , we designed the HIV Mutation Browser , as a web-interface that acts as a front end for the database . The browser presents the data in a hierarchically organised manner . The user selects a gene of interest , then a position of interest , and the citations relating to this position are presented to the user grouped by non-wildtype amino acid . The web interface is organised in three panels: the navigation panel at the top; the protein panel in the middle; and the residue panel at the bottom ( Figure 3 ) . The available mutagenesis and polymorphism data for a residue can be downloaded in both tab delimited text and Excel formats directly from the web interface .
HIV is an important therapeutic target and has been the subject of a major research effort as evidenced by the large catalogue of HIV experimental literature . Appropriate organisation and categorisation of the available HIV literature is necessary to allow efficient and intuitive access to relevant data . In this paper , we have presented the HIV Mutation Browser , a residue-centric resource of HIV mutagenesis and polymorphism literature designed for use by those carrying out basic and applied HIV research . The HIV Mutation Browser is one of the first resources to computationally text-mine mutagenesis and polymorphism data [7] , [8] , [9] , and the first to apply such methods to the extensive corpus of HIV literature . As such the HIV Mutation Browser will complement the available manually annotated and curated HIV resources such as the Stanford Drug Resistance database [2] , the UniProt knowledgebase [3] the Los Alamos HIV Database ( http://www . hiv . lanl . gov/ ) . In the coming years , we expect this method or similar methods to be applied to other viral or cellular systems . The resource will continue to evolve in the following ways . Firstly , HIV literature is produced continuously at a rate of approximately 1 , 500 articles a month and consequently the HIV Mutation Browser resource will be updated on a quarterly basis . Secondly , while the resource does contain the majority of important HIV and general interest journals , it is still incomplete , as we did not receive permission from all publishers to text-mine their HIV related articles . Journals from additional publishers will be added when possible . Thirdly , not all mutations can be correctly identified and assigned by the text-mining methods . There are various reasons for this . Many mutations are annotated in an article using non-standard patterns that are not widely used to describe directed mutations in mutagenesis experiments or polymorphisms . For example , consider the following excerpt taken from an article by Mitchell et al . , “The phenotype of the combination mutant VpuD51A-S52/56N was indistinguishable from that of Vpu-D51A and Vpu-S52/56N” [10] . The pattern “S52/56N” is a non-canonical construct for describing a mutagenesis experiment and currently will not be discovered by the text-mining method . Furthermore , the position of a mutation in a paper can be ambiguous and as a result mapping of the mutation information to the correct residue and protein can be a difficult task . For example , when multiple proteins are referenced and only a single mutation is discussed ( more than one possible mapping can be possible ) , when unconventional numbering is used ( particularly when describing mutations in Gag or Env as both are translated as polypeptide chains and subsequently cleaved ) or when unusual strains with insertions and deletions are used ( this shifts the numbering of residues ) . We will continue to improve the methods for text-mining and assignment . We request that members of the community utilise the feedback system for misannotated mutations in the resource and contact us about mutation data that should be in the resource yet is not present . This community input will improve the quality of the annotated data and will pinpoint parts of the text-mining method that require improvement . In summary , the HIV Mutation Browser is a valuable addition to the currently available HIV resources that will allow researchers to quickly and intuitively access data on mutagenesis and phenotypic variation . We expect the database to aid the process of experimental design and be a key resource for the HIV community .
A list of HIV-related articles was programmatically retrieved from PubMed using the search terms “HIV” and “Human Immunodeficiency Virus” . A list of target journals was constructed based on the number of published HIV-related articles . The licensing agreements of the majority of scientific journals prohibit a licensee from ( 1 ) downloading articles in bulk and ( 2 ) computationally processing the text of an article . Permission to waive these aspects of the licensing agreement was requested and received from the majority of virology and general interest scientific journals . The text of all HIV-related articles from the participating journals was retrieved programmatically from the publisher's websites to create a HIV literature dataset . An up to date list of the participating journals and publishers is available on the HIV Mutation Browsers website . There is no globally applied nomenclature to define directed mutations in mutagenesis experiments or polymorphisms [5] , [11] , [12] . A set of templates that define phrases and shorthand widely used to describe directed mutations in mutagenesis experiments or polymorphisms was created based on the work of Caporaso et al . [5] ( Figure 4A , see Table S1 for full list ) . Each article in the HIV literature dataset was converted to plain text and scanned using this set of templates . These templates consist of 3 pieces of information: the position in the protein which has been mutated , the amino acid present in that position in the wildtype sequence of the isolate , and the non-wildtype amino acid to which the residue has been mutated . For example , consider the sentence “As reported previously , S52A and S56A mutations of Vpu had no effect on virus release” [13] . S52A and S56A refer to the experimental mutagenesis of a serine to an alanine at position 52 and 56 in the Vpu protein . The annotation of a mutation text-mined from papers in the HIV literature dataset requires three piece of information: the sequence of the isolate used in the study; the protein containing the mutation; and the position of the mutation within the protein . This information is sufficient to map a mutation to a reference HIV-1 proteome , but cannot always be directly extracted from the text-mined paper . The nomenclature for describing isolates , genes , proteins , chains and domains have not been standardised . Therefore , mapping dictionaries for HIV isolates and HIV proteins were constructed . The isolate mapping dictionary was constructed from isolate names and their synonyms retrieved from HIV data within the UniProt [3] and Allie [14] resources ( Table S2 ) . The protein mapping dictionary was constructed from synonyms for genes , proteins , cleavage products , chains and domain names from the HIV data , also retrieved from the UniProt [3] and Allie [14] resources ( Table S3 ) . The highly-studied HIV group M subtype B HXB2 isolate was selected as the reference proteome and all HIV genes , proteins , cleavage products , chains and domain names , and their synonyms , were mapped onto the 9 proteins of the isolate ( Gag , Gag-Pol , Env , Tat , Nef , Rev , Vif , Vpr , Vpu ) . This mapping included normalised start positions to correct the inconsistent numbering schemes of cleavage products , chains and domain ( Figure 4B ) . Both dictionaries were further manually curated to improve upon the computationally retrieved mapping . Several different experimental isolates are commonly used in HIV experiments . Each paper in the HIV literature dataset was scanned using the contents of the isolate mapping dictionary to identify the experimental isolate used in the study ( Table S2 ) . If no isolate information was retrieved , the Human immunodeficiency virus type 1 group M subtype B HXB2 isolate was set as the experimental isolate for the paper . The numbering of a mutation in the HIV literature can refer to the numbering of a protein , chain , domain or cleavage product , consequently , for a defined mutation numbering a conclusive mapping may not possible . Inconclusively mapped mutations text-mined from the HIV literature dataset were mapped to the HIV proteome using a co-occurrence based approach ( Figure 4C ) . The co-occurrence based approach utilised the Reflect tool for automated tagging of biological entities to scan mutation-containing sentences for protein identifying terms from the protein mapping dictionary [4] . Each mutation's position was normalised to the protein-numbering scheme of the full-length protein based on the co-occurring protein identifying terms . If the mutation wildtype amino acid matched the amino acid at the normalised mutation position in the experimental isolate , the mutation was retained as a mapped mutation . In the cases where no information relating to the mutated protein was available , all HIV HXB2 proteins were scanned at their full-length protein , chain and domain levels . In the case of chain and domain , a displacement factor was applied to adjust the mutation's position and map the mutation to all possible positions in the proteome . A mutation mapping score ( see below ) was calculated for each putative mutation mapping and the top scoring mapping was retained as the mapped position of the mutation . In the cases where no matches to the experimental isolate proteome were found , the search was expanded to other commonly studied HIV isolates ( Table S2 ) . For each mutation mapped using the above approach , a mutation mapping score , S , is calculated . The score is the function of three parameters: the probability of a match by chance; the number of mapped mutations in the paper; and the displacement from the reference protein position numbering scheme . The score ranges from 0 to 1 , with values closer to 1 representing high confidence mapping of a mutation . The top-scoring mapping was retained as the mapped position of the mutation . The score , S , is calculated as:where M is the number of mapped mutations in the paper , N is the total number of mutations mentioned in the paper , d is the distance between the defined mutation position and the mapped position , L is the sequence length of the protein , the values of a , b , and c are constants to weight the contribution of each parameter in the equation ( a = 0 . 7 , b = 0 . 15 and c = 0 . 15 ) and P is the probability that the mapped mutation would map to the protein by chance and is calculated as:where p is 0 . 05 , the probability of matching an amino acid by chance given a 20 amino acid alphabet and assuming an equal frequency for each amino acid in the HIV proteome , and r is the number of mutations that have been mapped unambiguously to the protein . The HIV Mutation Browser interface integrates information from several resources to increase the ease of interpretation of the available HIV mutation and mutagenesis data . Conservation information is displayed using multiple sequence alignments ( aligned using the MAAFT algorithm [15] ) retrieved from the HIV Subtype Reference Protein sequences from the Los Alamos National Laboratory ( http://www . hiv . lanl . gov/ ) . Structural information is displayed using structures of HIV proteins retrieved from the RCSB Protein Data Bank ( PDB ) [6] . Intrinsic disorder predictions for the proteins are calculated using the IUPred algorithm [16] . Enzymatic active sites , sites of post-translational moiety addition , sites of proteolytic cleavage and other sites of functional importance are retrieved from the UniProt resource [3] . Short linear motif interaction interfaces are retrieved from the ELM databases [17] . An up to date list of the ancillary information used and displayed is available on the HIV Mutation Browsers website . | Naturally occurring mutations within the HIV proteome are of therapeutic interest as they can affect the virulence of the virus or result in drug resistance . Furthermore , directed mutagenesis of specific residues is a common method to investigate the function and mechanism of the viral proteins . We have developed novel computational text-mining tools to analyse over 120 , 000 HIV research articles , identify data on mutations and work out which amino-acid in which protein has been mutated . We have organised these data and made them available in an online resource—The HIV mutation browser . The resource allows HIV researchers to efficiently access previously completed research related to their region of interest in the HIV proteome . The HIV Mutation Browser complements currently available manually curated HIV resources and is a valuable tool for HIV researchers . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"text",
"mining",
"computer",
"and",
"information",
"sciences",
"virology",
"biology",
"and",
"life",
"sciences",
"microbiology",
"information",
"technology"
] | 2014 | The HIV Mutation Browser: A Resource for Human Immunodeficiency Virus Mutagenesis and Polymorphism Data |
Mycoplasma hominis is an opportunistic human mycoplasma . Two other pathogenic human species , M . genitalium and Ureaplasma parvum , reside within the same natural niche as M . hominis: the urogenital tract . These three species have overlapping , but distinct , pathogenic roles . They have minimal genomes and , thus , reduced metabolic capabilities characterized by distinct energy-generating pathways . Analysis of the M . hominis PG21 genome sequence revealed that it is the second smallest genome among self-replicating free living organisms ( 665 , 445 bp , 537 coding sequences ( CDSs ) ) . Five clusters of genes were predicted to have undergone horizontal gene transfer ( HGT ) between M . hominis and the phylogenetically distant U . parvum species . We reconstructed M . hominis metabolic pathways from the predicted genes , with particular emphasis on energy-generating pathways . The Embden–Meyerhoff–Parnas pathway was incomplete , with a single enzyme absent . We identified the three proteins constituting the arginine dihydrolase pathway . This pathway was found essential to promote growth in vivo . The predicted presence of dimethylarginine dimethylaminohydrolase suggested that arginine catabolism is more complex than initially described . This enzyme may have been acquired by HGT from non-mollicute bacteria . Comparison of the three minimal mollicute genomes showed that 247 CDSs were common to all three genomes , whereas 220 CDSs were specific to M . hominis , 172 CDSs were specific to M . genitalium , and 280 CDSs were specific to U . parvum . Within these species-specific genes , two major sets of genes could be identified: one including genes involved in various energy-generating pathways , depending on the energy source used ( glucose , urea , or arginine ) and another involved in cytadherence and virulence . Therefore , a minimal mycoplasma cell , not including cytadherence and virulence-related genes , could be envisaged containing a core genome ( 247 genes ) , plus a set of genes required for providing energy . For M . hominis , this set would include 247+9 genes , resulting in a theoretical minimal genome of 256 genes .
Mycoplasma hominis is an opportunistic human mycoplasma species which resides , as a commensal , in the lower urogenital tract . However , it can also cause pelvic inflammatory disease and postpartum or postabortion fevers , and has been associated with bacterial vaginosis [1] . In newborns , it can cause pneumonia , meningitis or abscesses . It has also been implicated in extragenital infections , especially in immunocompromised patients . Two other mollicute species , M . genitalium and Ureaplasma urealyticum , the latter having been recently separated into two species U . parvum and U . urealyticum [2] , have established pathogenic roles in humans and reside within the same natural niche , the urogenital tract , as M . hominis . However , these species have overlapping but distinct pathogenic roles [1] . M . genitalium and Ureaplasma spp . , but not M . hominis , are involved in male urethritis . M . genitalium is also the only mycoplasmal species involved in cervicitis , whereas Ureaplasma spp . are significantly associated with prematurity , low birth weight and chronic lung disease in infants [1] . These three urogenital species belong to two different phylogenetic groups within the class Mollicutes: M . genitalium and Ureaplasma spp . belong to the Pneumoniae group and M . hominis belongs to the Hominis group . They possess the smallest genomes among self-replicating free living organisms . The M . genitalium genome is the smallest of these , comprising 580 Kbp , with a capacity to encode only 482 genes [3] . The minimal nature of the M . genitalium genome triggered particular interest in this organism and several studies have addressed the concept of a minimal cell [4]–[6] . More recent studies have attempted to reconstruct its genome by chemical synthesis [7] , [8] , with a view to engineering a new living organism , referred to as Mycoplasma laboratorium [9] . The genome of M . hominis is slightly larger than that of M . genitalium , with pulsed-field gel electrophoresis revealing a chromosome size of 696 Kbp [10] . The sequenced genome of U . parvum ( previously called U . urealyticum serovar 3 ) is the largest of the three species , encompassing 751 Kbp [11] . These three species can therefore be considered as minimal bacterial cell prototypes with reduced metabolic abilities . Interestingly , in addition to their distinct pathogenic roles , they also have different energy-generating pathways . M . genitalium is a glycolytic species , whereas M . hominis and Ureaplasma spp . are both nonglycolytic species , producing energy through arginine degradation or urea hydrolysis , respectively [12] . Thus , during the evolution of mollicutes , these three human pathogens have undergone substantial genome reduction resulting in minimal , but distinct , metabolic mechanisms . Of all the mycoplasma species with a known pathogenic role in humans , M . hominis was the only one that had not been sequenced . We therefore sequenced the whole genome of the M . hominis PG21 type strain . We examined its energy-generating pathways and investigated the essential role of its arginine dihydrolase pathway in vivo . To provide further insight into the composition of hypothetical minimal gene sets and of the associated energy-generating pathways , we carried out whole genome comparisons of the three pathogenic mycoplasmas , M . hominis , M . genitalium and U . parvum , sharing the same urogenital niche and displaying near minimal genomes . This analysis resulted in the identification of a set of shared genes representing the core genome . Other genes include those that are involved in the energy-yielding pathways specific to each mollicute and those that play a role in the interaction with the host , such as cytadherence and virulence .
The general features of the M . hominis PG21 genome and their comparison with those of M . genitalium G37 and U . parvum serovar 3 are shown in Table 1 . The M . hominis PG21 genome is a single , circular chromosome of 665 , 445 bp with an overall G+C content of 27 . 1% . It contains 537 putative coding DNA sequences ( CDSs ) , representing a 89 . 8% gene density , and 14 pseudogenes were found . Function could be predicted for 345 of the CDSs , whereas 86 were conserved hypothetical proteins ( CHP ) and 106 were hypothetical proteins ( HP ) . A minimal but complete set of 33 tRNA genes was identified . The M . hominis genome contained two copies of rRNA genes , as previously described [10] . The 5S rRNA genes are not located within the 16S–23S rRNA operons , as is the case in M . arthritidis [13] , the phylogenetically closest genome-sequenced species . No insertion sequence , transposon , or endogenous plasmid was found in the genome . A set of 43 lipoproteins was predicted from the M . hominis genome , including three ABC transporter substrate-binding proteins ( MHO_3610 , MHO_3620 , MHO_1510 ) , two predicted nucleases ( MHO_0660 , MHO_0730 ) and one predicted peptidase ( MHO_4970 ) . Interestingly , the lipobox sequence , located upstream from the cleavage site , was highly conserved in five of the predicted lipoproteins , MHO_1730 , MHO_2100 , MHO_2340 , MHO_2440 , MHO_2620 , with the consensus motif PLVAAGC present in all of them . In contrast to these conserved regions , the other regions of the proteins were very different in length and sequence , suggesting their rapid and divergent evolution from a common ancestral gene . We assigned the origin of replication based on sequence homology with the M . arthritidis genome [13] and several other mycoplasmas [14] for which this oriC was experimentally demonstrated to be functional . We did not detect any significant inversion in the GC skew for M . hominis . The synteny rnpA → rpmH → dnaA → dnaN was present , as previously observed in M . arthritidis , M . capricolum subsp . capricolum and M . mycoides subsp . mycoides SC [13]–[15] . Two identical putative DnaA boxes ( TTATTAACA ) were found in the intergenic region upstream of the dnaA gene , but none between dnaA and dnaN . These boxes showed sequence identity at seven of the nine positions of the Escherichia coli consensus sequence TTATCCACA [16] . They were also identical to the unique DnaA box upstream from the dnaA gene in M . mobile [17] and M . arthritidis [13] , both of which belong to the same phylogenetic group as M . hominis . We identified genes that may have been exchanged through horizontal gene transfer ( HGT ) , by searching for M . hominis CDSs that show a best Blast hit ( BBH ) in species other than those belonging to the Hominis phylogenetic group . From the 537 CDSs predicted from the M . hominis chromosome , 59 had a BBH in a mollicute belonging to the phylogenetic groups Pneumoniae , Spiroplasma or Phytoplasma and 12 had a non-mollicute BBH . We then examined each of the candidate CDSs for potential HGT , using phylogenetic reconstructions and synteny analysis . Five CDSs seemed to be likely candidates for HGT between M . hominis and non-mollicute bacteria ( Table 2 ) . Their closest homologs were found in commensal or pathogenic bacteria previously identified in humans , suggesting a possible scenario for gene exchange . One of the five CDSs ( MHO_2540 ) was of particular interest because it encoded a putative N-dimethylarginine dimethylaminohydrolase ( DDAH ) , an enzyme linked to the arginine pathway ( see below ) . Among other mollicutes , only one clear orthologous gene was predicted in the phylogenetically closest related M . arthritidis ( 85% similarity ) , but the deduced protein was truncated in N-terminal by a third of its length . A putative DDAH was also proposed in M . penetrans ( MYPE1510 ) but it appeared very different from MHO_2540 ( 45% similarity ) and from other bacterial orthologs , suggesting a specific evolution and/or origin . Except in M . arthritidis , closest homologs of MHO_2540 were found in non-mollicute bacteria including the human pathogens Eggerthella lenta ( 76% similarity ) , Atopobium vaginae ( 72% similarity ) and Pseudomonas aeruginosa ( 71% similarity ) . Although phylogenetic reconstructions remained unclear , this suggests that MHO_2540 and its truncated ortholog in M . arthritidis might have a non-mollicute origin . Most of the 59 CDSs showing BBHs in mollicutes belonging to groups other than the Hominis phylogenetic group also had closely related homologs within the Hominis group and phylogenetic reconstructions did not allow clarification of their origin . Nevertheless , features indicative of HGT were found for six genes or groups of genes . In particular , MHO_0360 encoded a predicted type II cytosine-specific methyltransferase , for which the only homolog in mollicutes was found in M . mycoides subsp . mycoides SC ( Table 2 ) . Closely related homologs were also identified in other bacteria , such as Clostridium and Streptococcus spp . , with more than 80% amino-acid sequence similarity . This suggested that a type II restriction/modification system may have been exchanged between these bacteria . Five clusters of genes from the M . hominis genome were found to have their closest homologs in the urogenital pathogen U . parvum , which belongs to the phylogenetically distant Pneumoniae group ( Table 2 ) . The first cluster included MHO_0120 to MHO_0140 , encoding a predicted type III restriction/modification system ( Figure S1 ) . Closely related homologs were found in U . parvum , M . penetrans and several species of the Hominis group . Phylogenetic reconstruction , inferred from restriction enzyme sequences , associated the M . hominis ( MHO_0120 ) and U . parvum ( UU476 ) homologs with a bootstrap support value of 100% , suggesting that HGT occurred between the two species . In both species , the modification methylase-encoding gene was split into two by a frameshift that occurred within tracts of repeated nucleotides . In MHO_0130/MHO_0140 , a simple deletion of one nucleotide within a tract of 11 guanosines restored the complete ORF , suggesting that the expression of the encoded methylase may be switched on and off by a phase variation mechanism , similar to that previously described for another type III restriction/modification system in M . pulmonis [18] . The two gene clusters , MHO_5210 to MHO_5240 and MHO_3220 to MHO_3250 , encoded predicted type I restriction/modification systems ( Figure S2 ) . The first cluster contained full-length hsdR , hsdS ( 2 genes ) and hsdM genes and encoded the components of the system . In contrast , the second cluster only contained the two hsdS genes and an hsdM gene that was split into two by a frameshift mutation . Nucleotide sequences of the genes within the two clusters were nearly identical , indicative of a probable duplication event . Phylogenetic trees inferred from the protein sequences of the M . hominis R and M subunits showed a close relationship with their homologs in U . parvum ( UU095 to UU100 ) . A fourth cluster included CDSs MHO_2520 and MHO_2530 , two fragments of a transposase-encoding gene from an IS1138 element ( Figure S3 ) . Phylogenetic analysis demonstrated a stronger relationship between this gene and its fragmented homolog in U . parvum ( UU374/UU373/UU372 ) than with any other transposase gene . Moreover , sequence alignments revealed more than 90% similarity for the predicted proteins and extensive regions in which the nucleotide sequences were strictly identical . These data strongly suggest recent exchange of this mobile element between the two species . The fifth group of genes that may have been exchanged with U . parvum encoded seven conserved hypothetical proteins and two additional ATP synthase subunits ( MHO_3120 to MHO_3200 ) ( Figure S4 ) . As described for the other clusters , this potential gene exchange was deduced from protein-inferred phylogenetic trees and conservation of gene order . The seven CHPs were predicted to include one lipoprotein and three proteins with transmembrane segments , suggesting a location at the cell surface . It should be noted that , whereas five clusters of genes may have been exchanged with the U . parvum , no HGT was predicted with the urogenital pathogen M . genitalium , or with M . penetrans , another human mollicute reported in the urogenital tract [1] . The M . hominis genome was compared to the genomes of M . genitalium and U . parvum , the two other urogenital mollicute species that infect humans . Orthology relationships between the three genomes were computed using a bidirectional best hit method . Results are shown in Figure 1A , with the complete analysis provided in Table S1 . We found 247 CDSs common to all three genomes , forming the core genome , and identified 220 CDSs specific to M . hominis , 172 CDSs specific to M . genitalium and 280 CDSs specific to U . parvum . The M . hominis genome shared 24 CDSs only with M . genitalium and 46 CDSs only with U . parvum , whereas 41 CDSs found in both M . genitalium and U . parvum were not found in the M . hominis genome . The 247 orthologous genes shared by the three mollicutes encoded proteins involved in major cellular functions , including DNA metabolism , protein synthesis , nucleotide synthesis , transport and binding of substrates , and fatty acid and phospholipid metabolism ( Table S1 ) . Interestingly , these core genes were found to encode only a few cell envelope proteins: the putative membrane conserved hypothetical proteins MHO_3900 , MHO_4290 and MHO_4370 . Most of these 247 shared genes are likely to be essential , consistent with findings from an extensive gene inactivation study in M . genitalium [6] . Indeed , 213 of the 247 genes ( 86 . 2% ) could not be inactivated by transposon mutagenesis . Moreover , only 35 of these 213 genes could be inactivated in the two phylogenetically-related species M . arthritidis [13] or M . pulmonis [19] ( Table S2 ) . Among the 247 shared genes , 217 were also predicted in the genomes of other human and animal mycoplasmas ( Table S2 ) . Most of the 30 remaining genes were only missing in one or two species , suggesting a recent loss . Finally , a total of 160 genes that could not be inactivated were shared by all available mycoplasma genomes . Many of the species-specific genes were associated with cytadherence and virulence ( Figure 1B ) . Genes found specifically in M . hominis coded for: ( i ) the lipoprotein P120 ( MHO_3660 ) and the protein P120' ( MHO_3800 ) , which are M . hominis-specific surface-exposed proteins displaying antigenic variation [20] , [21]; ( ii ) the Vaa surface lipoprotein adhesin ( MHO_3470 ) , which is an abundant surface antigen displaying high-frequency phase and size variation and which is involved in adhesion to host cells [21] , [22]; ( iii ) the P60 and P80 proteins ( MHO_3490 , MHO_3500 ) , which form a membrane complex at the surface and are encoded by an operon [23]; ( iv ) the P75 lipoprotein ( MHO_3720 ) , which is also present at the surface of M . hominis [24] , and the P75-related lipoprotein ( MHO_3100 ) , which shows 37 . 9% similarity with the P75 protein . CDSs specific to M . hominis also included OppA ( MHO_1510 ) , an oligopeptide permease substrate-binding protein . OppA is a multifunctional lipoprotein involved in cytadherence , but is probably also the main ecto-ATPase at the surface of the M . hominis cell [25] , [26] . A recent study showed that it induced ATP release from cells , resulting in apoptosis , thus suggesting its role as a M . hominis virulence factor [27] . Lmp1 ( MHO_0530 ) and Lmp3 ( MHO_1640 ) , two related surface membrane proteins [28] , and six additional Lmp-related proteins ( MHO_0540 , MHO_3070 , MHO_3110 , MHO_3730 , MHO_4280 and MHO_4920 ) , were also found in this group of M . hominis-specific genes . The eight Lmp proteins represent a family of proteins characterized by a predicted N-terminal transmembrane helix and several repeat sequences . In particular , using the MEME method , we identified a 57 amino-acid motif , [QK][QK]L[DKQ][DN]LI[DK]S[NQE][DE][AG][KQ][DKN][VI]D[KT][SQ]K[EA][TN][QDN][IS][LF][NQ]N[TN]N[LVI][DT][AG][SKN][SD][LT][IT][KD][DQ]I[EKV][SN][KA][TI][KN][TE]I[EK][DK]A[IT][QEK][SD]L[TQ]K[KL]I[ND] , repeated between one and 15 times , in all proteins except MHO_0530 and MHO_4280 . Similarly , genes specific to M . genitalium and U . parvum genomes included several CDSs associated with cytadherence and virulence ( Table S1 , Figure 1B ) . The 172 M . genitalium specific CDSs included genes encoding the adhesin MgPa ( MG191 ) , the MgpC protein ( MG 192 ) , the adhesin P32 ( MG318 ) , the protein P200 ( MG386 ) and the cytadherence accessory proteins Hmw1 , Hmw2 and Hmw3 ( MG218 , MG312 , MG317 ) . Of the 280 CDSs specific to U . parvum , genes associated with cytadherence and virulence included those encoding the MBA protein ( UU375 ) — a major antigen recognized during infection of humans — and the five MBA N-terminal paralogs ( UU172 , UU189 , UU483 , UU487 , UU526 ) . Genes encoding eight iron transporters ( UU022 , UU023 , UU025 , UU357 , UU358 , UU400 , UU515 , UU516 ) , nine putative ABC substrate-binding protein-iron ( UU024 , UU027 , UU028 , UU069 , UU071 , UU359 , UU360 , UU401 , UU517 ) and two hemin transporters ( UU070 , UU399 ) were also found in the U . parvum genome , but not in M . hominis or M . genitalium . It should be noted that the virulence-related gene msrA , encoding the peptide methionine sulfoxide reductase , was found in both M . genitalium ( MG408 ) and U . parvum ( UU289 ) , but not in M . hominis . In M . genitalium , inactivation of the msrA gene reduces adherence and abolishes virulence [29] . Several species-specific genes were involved in the major energy-generating metabolic pathways . In M . genitalium , which is a glycolytic species , several genes encoding components of the carbohydrate , pyruvate and glycerol metabolic pathways were found among its 172 specific CDSs , including carbohydrate transporter genes ( Table S1 ) . Indeed , the 1-phosphofructokinase-encoding gene fruK ( MG063 ) , fruA ( MG062 ) and pstG ( MG069 ) genes of the fructose and glucose phosphoenolpyruvate-dependent sugar phosphotransferase transport system ( PTS ) were found specifically in M . genitalium . Moreover , the four genes encoding the pyruvate dehydrogenase complex , pdhA ( MG274 ) , pdhB ( MG273 ) , pdhC ( MG272 ) and pdhD ( MG271 ) , the glycerol kinase gene gplK ( MG038 ) and the glycerol uptake facilitator gene glpF ( MG033 ) were predicted in the M . genitalium genome only , and not in the M . hominis or U . parvum genomes . In U . parvum , which generates ATP through urea hydrolysis , the seven components of the urease complex ( UU428 to UU434 ) , the ammonium transporters Amt-1 and Amt-2 ( UU218 , UU219 ) and the atypical delta subunits of the FoF1-ATPase ( UU133 , UU134 ) probably involved in this unique energy production pathway [11] , were found among the 280 CDSs that were absent from M . hominis and M . genitalium ( Table S1 ) . In M . hominis , a nonglycolytic species , all the genes coding for all enzymes of the Embden-Meyerhoff-Parnas ( EMP ) pathway were present except 6-phosphofructokinase ( Figure 2 ) . The nonglycolytic U . parvum , however , had the 6-phosphofructokinase gene , but did not have the gene encoding glucose-6-phosphate isomerase , which catalyzes the step leading to fructose 6-phosphate production [11] ( Figure 2 ) . The set of genes coding for enzymes in the pentose phosphate pathway was incomplete in M . hominis , with the absence of genes encoding glucose-6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase . These genes were also absent from M . genitalium and U . parvum . The genes coding for other enzymes involved in carbohydrate metabolism were present in M . hominis and U . parvum , with the exception of 1-phosphofructokinase , which is involved in fructose metabolism , and the pyruvate dehydrogenase complex , as mentioned above [11] , [30] . Thus , a complete set of genes encoding components of the EMP pathway was present in M . genitalium [3] , [30] , which is able to generate ATP through glucose hydrolysis , whereas the M . hominis and U . parvum genomes each lacked a single gene encoding a component of the EMP pathway . It should also be noted that M . hominis , like U . parvum [11] , does not seem to possess a complete phophoenolpyruvate phosphotransferase system ( PTS ) , as only a putative energy-coupling protein HPr ( MHO_0590 ) and the component B of the enzyme II complex ( MHO_1070 ) were predicted . In contrast , the glycolytic M . genitalium possessed two complete glucose and fructose PTS . Finally , the two genes encoding enzymes of the acetate kinase–phosphotransacetylase pathway , which is generally fully reversible ( acetate+ATP+CoA↔acetyl-CoA+ADP+Pi ) in bacteria , were present in M . genitalium and U . parvum; but the phosphotransacetylase gene was absent from the M . hominis genome ( Figure 2 , Table S1 ) . Interestingly , glycerol 3-phosphate dehydrogenase ( gpsA , MHO_1690 ) was present in both M . hominis and U . parvum . This enzyme could lead to glyceraldehyde 3-phosphate production via the second part of the EMP pathway involving triosephosphate isomerase ( Figure 2 ) . In M . hominis PG21 , the three proteins constituting the arginine dihydrolase pathway ( Figure 2 ) — arginine deiminase ( ADI ) ( arcA , MHO_0690 ) , ornithine carbamoyltransferase ( arcB , MHO_0640 ) and carbamate kinase ( arcC , MHO_0630 ) — were found to be specific among the three species discussed here . Moreover , a putative dimethylarginine dimethylaminohydrolase ( DDAH , MHO_2540 ) was predicted in M . hominis but not in U . parvum and M . genitalium . In P . aeruginosa , DDAH ( EC 3 . 5 . 3 . 18 ) catalyzes the conversion of the Nω-alkylated arginine residues to citrulline . A more detailed overview of the M . hominis energy-generating pathway was obtained by searching for specific transporter genes in the genome . The results are graphically displayed in Figure 2 . A putative arginine transporter , MHO_5040 , containing 12 transmembrane domains was identified as a homolog of the amino-acid permease MYPE6110 , a protein which is encoded immediately downstream from the arginine dihydrolase pathway cluster , arcABC , in the M . penetrans genome [31] , another arginine-utilizing mycoplasma . However , in M . hominis , MHO_5040 was not located in close proximity to the arcABC cluster . In addition , it should be noted that no ammonia/ammonium transporters was identified in the M . hominis genome . To further determine the major energy-generating pathway in M . hominis , we studied the growth of M . hominis PG21 in the presence of arginine and arginine analogs ( Table 3 ) . In Hayflick modified medium containing 2 . 5% horse serum , the M . hominis culture reached 107 color changing units ( CCU ) /ml and the pH increased from 7 . 0 to 7 . 2 after 48 hours . In medium supplemented with 10 mM arginine , the titer after 48 hour growth was unchanged , but the pH was higher than that observed without addition of arginine , increasing from 7 . 0 to over 8 . 1 . A total loss of viability was observed after 120 hours under these conditions . The effect on growth of adding horse serum was checked by inoculating media with or without 2 . 5% horse serum . Growth was independent of serum content in medium supplemented with 10 mM arginine . However , a non-supplemented medium lacking horse serum was growth-limiting , with a 10-fold lower titer of organisms after 48 hours ( data not shown ) . Thus , the serum-containing Hayflick modified medium supplied the cells with nutrients that were necessary to reach a high cell titer after 48 hours . Furthermore , in the absence of horse serum , the exogenously added arginine provided energy for cell growth . The effect on growth of canavanine , an arginine analog that interferes with the arginine uptake and metabolism was evaluated . Canavanine is an ADI-specific competitive inhibitor [32] and its incorporation into proteins in bacterial species leads to cell death [33] . In the presence of 10 mM canavanine , the cell concentration remained constant at 105 CCU/ml over 120 hours . Addition of 23 mM arginine at 48 hours was sufficient to displace the competitive inhibitor and restore growth ( Table 3 ) . Canavanine suppressed the cell growth but did not trigger the M . hominis death within 120 hours . Thus , while a possible incorporation of canavanine into proteins cannot be totally ruled out , the bacteriostatic effect of canavanine was most probably due to its inhibitory effect on ADI . These data suggest that the ADI pathway was required for initiation of growth in M . hominis . Finally , to test whether asymmetric Nω , Nω-dimethyl-L-arginine ( ADMA ) could be used as a substrate by DDAH and could serve as an energy source for protein synthesis in M . hominis , growth was evaluated in the presence of ADMA with ADI activity suppressed by canavanine . The maximum concentration of ADMA that could be assessed was 0 . 5 mM , because of the poor solubility and dissolution rate of the arginine analog in aqueous solutions . Under these conditions , in presence of 2 . 5% horse serum in the medium ( Table 3 ) , as well in medium lacking horse serum , no growth was detected after 48 hours .
In this study , we first sequenced , annotated and analyzed the genome of M . hominis PG21 . Our analyses suggested the occurrence of HGT between M . hominis and U parvum . Indeed , five clusters of genes from the M . hominis genome were found to have their closest homolog in U . parvum , which belongs to the phylogenetically distant Pneumoniae group . These mollicutes share the same urogenital niche , providing a potential explanation for the predicted HGT . Phylogenomic studies have recently reported the occurrence of HGT among animal mycoplasmas sharing the same host . Indeed , 18% of the M . agalactiae genome has undergone HGT with other pathogenic , ruminant mycoplasmas from the phylogenetically distant Mycoides cluster [34] . In M . hominis , the potential advantages of these events remain unclear . However , the largest gene cluster predicted to have undergone HGT ( MHO_3120 to MHO_3200 ) contains four putative cell-surface proteins that may be involved in the interaction between the mycoplasma and its human host . This is the first report of HGT among human mollicutes . We did not find any evidence of HGT between M . hominis and M . genitalium , or M . penetrans . M . hominis and U . parvum are both frequent commensals of the human urogenital tract and they are commonly detected together in healthy individuals . In contrast , M . penetrans is very rarely detected in the urogenital tract [1] , and the commensal status of M . genitalium , which has a low prevalence in the general population [35] , remains unclear . Overall , these data suggest that genome reduction was not the only factor affecting the evolution of these minimal organisms . We also compared the whole genomes of the three human pathogenic mycoplasmas . Two major sets of genes could be identified among species-specific genes . The first group contained genes involved in energy-generating mechanisms , with the second group containing genes involved in cytadherence and virulence . Thus , these three mycoplasma species have their own specific genetic equipment underlying their pathogenic roles . We found several CDSs specific to M . hominis . These included the genes encoding P120 , P120' , Vaa , LMP1 and LMP3 , which have been previously identified as surface proteins displaying size , sequence , and antigenic variation [20]–[22] , [24]–[26] , [28] , and six genes annotated as LMP-related proteins . The large number of members of the LMP-related protein group and the repetitions present in their DNA sequences is likely to provide the basis for genetic variability , potentially playing a significant role in determining the response to host defense mechanisms . Parallels can be drawn with the multiple complete or partial copies of the M . pneumoniae P1 adhesin gene [36] or the M . genitalium mgpB adhesin gene [3] dispersed on the chromosome , which serve as reservoirs to generate antigenic variation and participate in host defense evasion [37] , [38] . Proteomic studies could be used to identify the LMP-related proteins expressed in M . hominis to further elucidate the role of LMPs . Size variation among these cell envelope proteins could also explain the heterogeneity in genome size within the M . hominis species . Indeed , genome size was previously estimated for 15 strains using pulsed-field gel electrophoresis , and was found to vary by more than 15% , ranging from 696 to 825 Kbp [10] . It is likely that there are a number of genes involved in M . hominis pathogenic mechanisms that are yet to be described , so the characterization of genes specific to M . hominis could help to identify unknown adhesion or virulence factors . Indeed , such factors could be searched for among the 220 M . hominis-specific CDSs , particularly among the hypothetical proteins , although virulence factors may also be shared with other mycoplasma species . M . hominis has been described as a nonglycolytic species that generates ATP through arginine hydrolysis [12] , [39] , as previously described for M . arthritidis , the mycoplasma most closely related to it for which the genome sequence is also available [12] . The question of whether arginine degradation was the sole or the major energy-generating mechanism has been asked for more than 30 years . We thus analyzed the genomes of the three mycoplasmas for their genes encoding enzymes involved in the energy generation from carbohydrate metabolism . Analysis of the M . hominis genome revealed that the gene sets of the pentose phosphate pathway and of the glucose and fructose transport PTS were incomplete , as has been observed for other sequenced glycolytic and nonglycolytic mollicutes [30] , [39] , [40] . Furthermore , previous enzymatic activity studies , reporting the absence of 6-phosphofructokinase activity , from the EMP pathway , in seven different strains of M . hominis were confirmed in silico by the lack of the corresponding gene [12] . Thus , as previously proposed for U . parvum [11] , [30] , it is unlikely that M . hominis metabolizes hexoses such as glucose or fructose . The presence of a gene encoding glycerol 3-phosphate dehydrogenase in M . hominis suggests that glycerol 3-phosphate may be oxidized into dihydroxyacetone phosphate , which could then enter glycolysis ( Figure 2 ) . This would require glycerol or glycerol 3-phosphate to be efficiently imported into the cell . This transport could be mediated by a specific ABC transporter for sn-glycerol phosphate , putatively encoded by MHO_0740–0760 . Indeed , MHO_0740–0760 shows a BBH and conserved synteny with genes encoding putative glycerol transporters in other mollicute species , such as M . synoviae and M . pneumoniae [36] , [41] . Further downstream in this pathway , energy is not generated through oxidation of pyruvate to acetyl-CoA , since the genes coding for the pyruvate dehydrogenase complex are absent from the M . hominis genome ( Figure 2 ) . The reversible acetate kinase-phosphotransacetylase pathway was predicted to be incomplete in M . hominis , suggesting that the putative conversion of acetate into acetylphosphate does not lead to generation of acetylCoA . Nevertheless , the presence of the gene encoding acetate kinase ( ackA , MHO_3840 ) is particularly notable , given the possible role of this enzyme in ATP generation in M . hominis [42] . Acetylphosphate may be produced during the conversion of xylulose 5-phosphate to glyceraldehyde 3-phosphate by phosphoketolase ( MHO_3010 ) , the gene for which is in the M . hominis genome but is absent from the U . parvum and M . genitalium genomes . The presence of several gaps in the M . hominis glycolysis and pyruvate pathways supports the notion that these energy-generating pathways have a low efficiency in this species . Consistent with this , we observed that the growth of M . hominis in Hayflick modified broth supplemented with glucose [43] led to alkalinisation of the culture medium ( data not shown ) . M . hominis species-specific genes included a complete set of genes coding for the enzymes of the arginine dihydrolase pathway , which can generate energy in the form of ATP for growth . Considering that alternative energy-yielding pathways have been demonstrated for several arginine-utilizing mollicutes [44] , [45] , it remains unclear whether the ATP produced by arginine breakdown can satisfy all M . hominis growth requirements . Arginine is considered an essential nutrient for initiation of growth of M . hominis [46] . However , the lack of correlation between CO2 production , cell growth and ADI activity , as well as the delayed induction of ADI relative to the cell cycle , led Fenske and Kenny [47] to suggest that M . hominis used the arginine pathway only as an alternative energy source . In this study , we showed arginine to be sufficient for cell growth in a growth-limiting medium . Interestingly , the pH of the medium after 48 hours of growth varied with the initial composition of the medium . Supplementation of the basal medium with arginine resulted in an increase of the pH , probably due to a higher rate of ammonium production via the arginine dihydrolase pathway . This increase in pH could have deleterious effects on cell survival in vitro , as suggested by the loss of viability after 120 hours in the presence of arginine . Moreover , canavanine , a specific competitive inhibitor of ADI from M . arthritidis [32] , suppressed growth of M . hominis , suggesting that ADI activity is essential for its growth . For its metabolism , arginine can be transported into M . hominis cells in its free-form [46] . In our study , arginine was provided in its free and peptide-bound forms . The free amino-acid may enter the cell via the putative arginine transporter ( MHO_5040 ) . Additionally , given that there are genes encoding a putative Opp ABC system ( MHO_1510–1550 ) in the M . hominis genome , arginine-containing oligopeptides should be considered as possible alternative source of arginine . Currently , a considerable limitation on in vivo metabolic studies is the absence of a defined medium for growing M . hominis . An additional in silico genome-scale metabolic reconstruction of the metabolic pathways could help to elucidate the true biochemical capacity of M . hominis , as recently described for M . genitalium [48] . We identified a gene encoding a putative DDAH , an enzyme structurally related to ADI , in the M . hominis PG21 genome . This suggests that a secondary citrulline-producing pathway , involving the catalytic conversion of ADMA , may occur in M . hominis . However , addition of a maximum soluble concentration of ADMA to a medium supplemented with canavanine was not sufficient to promote detectable growth . The ADMA concentration was probably not high enough in our experiments to allow a significant increase in bacterial cell concentration . Thus , the potential production of additional citrulline through metabolism of ADMA in M . hominis remains uncertain but cannot be excluded . ADMA is naturally present in biological fluids and inhibits nitric oxide synthase in humans [49] . As previously reported for P . aeruginosa [50] , production of DDAH by M . hominis could indirectly increase host NO production leading to tissue damage . Consistent with this hypothesis , M . hominis can stimulate nitric oxide synthase production by host macrophages [51] . It should also be noted that the DDAH gene found in M . hominis and its truncated ortholog found in the closely related M . arthritidis may have been acquired by HGT from non-mollicute bacterial species , notably from A . vaginae which belongs to the same urogenital niche than M . hominis and is also involved in bacterial vaginosis [52] . The arginine utilization pathway may have implications for the relationship between M . hominis and its host through ammonia formation . This compound is also generated during the hydrolysis of urea in U . parvum . As both species are mainly found in the acidic female genital tract , these pathways could be required to protect them from the deleterious effects of the acidic environment and may also promote their pathogenicity . Moreover , M . hominis may be able to form biofilms since it was identified in microbial biofilms involved in intraamniotic infection [53] . The ADI pathway may therefore also play a role in biofilm formation , as previously suggested for Staphylococcus aureus [54] . Indeed , in S . aureus , the expression levels of the deiminase pathway were higher in biofilm cells than in planktonic cells . To identify a minimal gene set essential for cell viability , Glass et al . used a global transposon mutagenesis strategy in M . genitalium . They found 382 of the 482 protein-coding genes to be essential under laboratory conditions [6] . However , when the 537 M . hominis protein-coding genes were compared with this set , 147 ( 38% ) were absent from the M . hominis genome ( data not shown ) . These 147 genes absent from M . hominis mainly encoded proteins involved in carbohydrate transport and metabolism ( PtsH , PtsG , PtsI , 6-phosphofructokinase , glycerol kinase , pyruvate dehydrogenase complex ) or in cytadherence and virulence ( MgPa , MgpC , P32 , P200 , cytadherence accessory proteins Hmw1 , 2 and 3 ) . It is therefore possible that at least two whole gene sets involved in cytadherence/virulence and in energy-generating carbohydrate metabolism may be replaced by two other whole gene sets facilitating similar functions in another minimal species ( Figure 1B ) . Thus , it is possible that a complete set of genes constituting an effective energetic pathway , rather than an isolated gene , should be considered essential . This notion can be extended to the adhesion/virulence gene set , as most of the M . genitalium adhesion/virulence genes have been demonstrated to be essential [6] . Consequently , in M . hominis , in addition to the 247 core CDSs shared by the three minimal species infecting humans , the energy-generating gene set would mainly consist of the three genes of the arginine deiminase pathway and the DDAH-encoding gene . However , the phosphopentomutase- ( MHO_4170 ) and the phosphoketolase- ( MHO_3010 ) genes , which are specific to M . hominis , and three genes absent from the core genome but present in M . hominis and in one other species ( encoding glucose 6-phosphate isomerase , lactate dehydrogenase and glycerol 3-phosphate dehydrogenase ) ( Figure 1B ) should be added to this set of energy-providing genes , as they may be required for optimal energy metabolism in M . hominis . Thus , a minimal mycoplasma cell , not including cytadherence and virulence-related factors , could be envisaged to have the 247 gene core genome plus a set of genes involved in energy production but not already in the core . This set would include at least nine genes in M . hominis , given that the genes encoding putative transporters of arginine or sn-glycerol 3-phosphate have not been clearly identified . Thus , in M . hominis , this would result in a theoretical minimal genome of 256 ( 247+9 ) genes . This gene list could be useful for the synthesis of artificial genomes using the recently reported complete chemical synthesis procedure [7] , [8] . In conclusion , M . hominis PG21 has the second smallest genome among self-replicating free-living organisms . This species hydrolyses arginine as its major energy-source and possesses its own set of cell-surface proteins as the basis of its urogenital pathogenicity . We have described a 247 gene core genome based on whole genome comparisons between M . hominis and the two other human urogenital pathogens with minimal genomes , M . genitalium and U . parvum . An additional set of nine genes may be required for the energy generation in M . hominis . Moreover , a set of 220 M . hominis-specific genes were identified , which may harbor unknown virulence genes .
The M . hominis type strain PG21 ( ATCC 23114 ) was chosen and grown in Hayflick modified medium supplemented with arginine [43] . Genomic DNA was isolated as previously described [55] . We constructed three libraries to determine the complete sequence of the M . hominis PG21 genome . For plasmid libraries , genomic DNA was mechanically sheared and 3 kb fragments were cloned into pNAV ( A ) and pCNS vectors ( B ) ( derived from pcDNA2 . 1 and pSU18 , respectively ) . Large inserts ( 25 kb ) , generated by HindIII partial digestion , were introduced into pBeloBac11 ( New England Biolabs , Ipswich , USA ) ( C ) . Vector DNA was purified and end-sequenced using dye-terminator chemistry on ABI3730 sequencers ( 13440 , 8109 and 2304 reads for library A , B and C , respectively ) . The Phred/Phrap/Consed software package ( www . phrap . com ) was used for the sequence assembly . A complement of 91 sequences was needed for gap closure and quality assessment . The genome annotation was performed using the CAAT-Box platform [56] , extended with different tools to facilitate the annotation process , as previously described [34] . Briefly , CDSs were detected with the trained GeneMark software [57] , implemented in CAAT-Box , which also integrated results of BLAST searches [58] in three databases — SwissProt ( http://www . ebi . ac . uk/swissprot/index . html ) , TrEMBL ( http://www . ebi . ac . uk/embl/index . html ) , and MolliGen ( http://cbi . labri . fr/outils/molligen/ ) . Sequence similarity and start codons were determined as previously described [34] . Predicted proteins with similarity lower than 40% or with only local similarities with previously characterized proteins were annotated as hypothetical proteins . Small CDSs or gene fragments were systematically searched for in intergenic sequences of more than 80 bp with BLASTX . The annotation of each CDS was manually verified by at least two annotators . The tRNA genes were identified as previously described [34] and the rRNA genes were mapped onto the chromosome using previously published sequences [10] , [55] . Lipoproteins were predicted using the PROSITE prokaryotic membrane lipoprotein lipid attachment site motif ( PROKAR_LIPOPROTEIN ) . Putative HGTs were identified as previously described [34] . Briefly , BBHs were identified for every predicted protein using a BLASTP threshold E-value of 10−8 . CDSs displaying a BBH from a species not belonging to the Hominis phylogenetic group were further investigated by phylogenetic analyses . Protein phylogenies were determined using the MEGA4 package [59] . Trees were constructed using the distance/neighbor-joining method and the gap complete deletion option; bootstrap statistical analyses were performed with 500 replicates . Bootstrap values lower than 90% were not considered significant . Incongruence between protein and species phylogenies , supported by significant bootstrap values , was considered to be a potential HGT . Genes for which only very few homologs were identified or with branches only supported by low bootstrap values , were not considered to have undergone HGT unless other independent results suggested otherwise . Independent results indicative of the occurrence of HGT were a particularly high similarity value ( >80% ) and/or conservation of gene synteny . Genome analysis and comparisons were conducted using MolliGen , a database dedicated to the mollicute genomes [60] . Orthologous genes in M . hominis PG21 , M . genitalium G37 and U . parvum serovar 3 ( ATCC 700970 ) were determined using the bi-directional best hit method followed by manual curation . M . hominis PG21 was grown in a basal medium containing heart infusion broth ( Difco ) , yeast extract ( Bio-Rad ) , and 2 . 5% horse serum ( Difco ) , with or without 10 mM arginine ( Fluka ) , 10 mM canavanine ( Sigma ) or 0 . 5 mM Nω , Nω-dimethyl-L-arginine ( ADMA ) ( Sigma ) in 4% dimethylsufoxide , individually or in combination . Identical experiments were conducted with a basal medium without horse serum to limit the arginine content . The pH of the media was adjusted to 7 before inoculation with M . hominis . The initial free arginine concentration of 2 . 5% serum-containing Hayflick media was measured at 1 . 4 mM by HPLC ( High Performance Liquid Chromatography ) after deproteinisation . Incubation time was 48 and 120 hours and growth was measured in CCU per ml [43] . At t = 0 , 48 and 120 hours , 100 µl of culture was diluted to generate a tenfold dilution series to 10−12 in Hayflick modified broth medium supplemented with 23 mM arginine . Results are the means of at least three independent experiments . The genome sequence and related features of the M . hominis PG21 strain were submitted to the EMBL/GenBank/DDBJ databases under accession number FP236530 . All data are also available from the MolliGen database ( http://cbib1 . cbib . u-bordeaux2 . fr/outils/molligen/home . php ) . | Mycoplasma hominis , M . genitalium , and Ureaplasma parvum are human pathogenic bacteria that colonize the urogenital tract . They have minimal genomes , and thus have a minimal metabolic capacity . However , they have distinct energy-generating pathways and distinct pathogenic roles . We compared the genomes of these three human pathogen minimal species , providing further insight into the composition of hypothetical minimal gene sets needed for life . To this end , we sequenced the whole M . hominis genome and reconstructed its energy-generating pathways from gene predictions . Its unusual major energy-producing pathway through arginine hydrolysis was confirmed in both genome analyses and in vivo assays . Our findings suggest that M . hominis and U . parvum underwent genetic exchange , probably while sharing a common host . We proposed a set of genes likely to represent a minimal genome . For M . hominis , this minimal genome , not including cytadherence and virulence-related genes , can be defined comprising the 247 genes shared by the three minimal genital mollicutes , combined with a set of nine genes needed for energy production for cell metabolism . This study provides insight for the synthesis of artificial genomes . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genetics",
"and",
"genomics/microbial",
"evolution",
"and",
"genomics",
"genetics",
"and",
"genomics/comparative",
"genomics",
"microbiology/microbial",
"evolution",
"and",
"genomics",
"genetics",
"and",
"genomics/genome",
"projects",
"microbiology/medical",
"microbiology"
] | 2009 | Life on Arginine for Mycoplasma hominis: Clues from Its Minimal Genome and Comparison with Other Human Urogenital Mycoplasmas |
We often learn and recall long sequences in smaller segments , such as a phone number 858 534 22 30 memorized as four segments . Behavioral experiments suggest that humans and some animals employ this strategy of breaking down cognitive or behavioral sequences into chunks in a wide variety of tasks , but the dynamical principles of how this is achieved remains unknown . Here , we study the temporal dynamics of chunking for learning cognitive sequences in a chunking representation using a dynamical model of competing modes arranged to evoke hierarchical Winnerless Competition ( WLC ) dynamics . Sequential memory is represented as trajectories along a chain of metastable fixed points at each level of the hierarchy , and bistable Hebbian dynamics enables the learning of such trajectories in an unsupervised fashion . Using computer simulations , we demonstrate the learning of a chunking representation of sequences and their robust recall . During learning , the dynamics associates a set of modes to each information-carrying item in the sequence and encodes their relative order . During recall , hierarchical WLC guarantees the robustness of the sequence order when the sequence is not too long . The resulting patterns of activities share several features observed in behavioral experiments , such as the pauses between boundaries of chunks , their size and their duration . Failures in learning chunking sequences provide new insights into the dynamical causes of neurological disorders such as Parkinson’s disease and Schizophrenia .
Sequence learning is a critical component of human intelligence . The ability to recognize and produce ordered sequences is a defining feature of the brain and a key component of many cognitive performances . Sequence learning and production is a hierarchical process , such as in speech organization , behavioral sequences , and thought processes . By segmenting a sequence of elements into blocks , or chunks , information becomes easier to retain and recall in the correct order [1] . Such chunking organization in memory has been investigated for more than half a century , when Bousfield formulated the idea that information-carrying items seem to be recalled in associated clusters [2] , and Miller pointed out that limits in our working memory capacity for processing information necessitated the organization of items into chunks [3] . A chunk is often defined as a collection of elements having strong associations with each other , but weaker associations with elements within other chunks [4] . For example , complex motor movements are represented as a chain of subordinate movements , which are concatenated in a goal-specific fashion [5] . Behavioral visuo-motor sequence learning experiments suggest that action sequences are organized as chunks of information-carrying items [6–9] . Imaging and behavioral studies further suggest that chunking learning extends to language processing [10 , 11] , visual perception [12] , habit learning [13] , and motor skills [14–17] . Several studies provided models for chunking learning that explain some behavioral observations . For example , a model of chunking learning explains why skill improves with practice according to a power law [18] . Another example is that of competitive chunking [19] , whereby a bottom-up perception process strengthens the chunks . Such computational models are informative as high-level descriptions of chunking learning , but do not incorporate temporal dynamics in a natural way . As a result , such models cannot provide principled insight into the temporal aspects of behavior . On the other hand , a dynamical systems approach naturally allows the study of temporal interactions [20] , and can provide tight connections with biophysical models of neurons . Experimental findings in imaging and behavioral studies provide the structure and dynamics of chunking in the brain at the mesoscopic level , allowing one to build theoretical models for the description of chunking in cognition and behavior [21] . These models are non-linear dynamical systems that describe the interaction of core components—or cognitive modes—participating in a specific mental function [22] . Here , we describe a dynamical model of the cognitive mechanisms for learning chunking representations of sequences . The dynamical system is based on the sequential competition between different information-carrying items that are represented as metastable states , such as saddle nodes . In the neighborhood of a saddle point , elementary volumes in the phase space are compressed along stable separatrices and stretched along an unstable separatrix . Saddle nodes can be chained such that the unstable separatrix of one node corresponds to the stable separatrix of the next node along the chain . If the compressing at the saddle node is larger than the stretching and all nodes in the chain are dissipative , the trajectories stably follow a channel [22] . Such channels are known as Stable Heteroclinic Channels ( SHCs ) , and are argued to form the basis of sequential working memory through Winnerless Competition ( WLC ) dynamics [23 , 24] . The WLC principle depicts itinerant dynamics whereby a “winning” state transiently dominates the network in a sequential fashion . Its function is to transform inputs ( e . g . a task input ) into spatiotemporal outputs based on the intrinsic switching dynamics of an ensemble of modes [23] . As a concrete model of WLC , we employ a generalization of the Lotka-Volterra evolutionary prey-predator model [25] , known as the Generalized Lotka-Volterra ( GLV ) model . GLVs represent a canonical non-linear model of non-equilibrium dissipative systems [26] , and is widely used to study local bifurcations of SHCs . Many other models can be written in the form of GLV after some recasting [27] , and its dynamical properties are consistent with a wide range of neuron models [23 , 28–30] . Extending this idea , a dynamical image of chunking processing is a two-layer model describing a heteroclinic chain of heteroclinic chains . Under these dynamics , one metastable state in a “chunking layer” is associated to a heteroclinic sequence in another “elementary layer” [31] . In such representation , the chunks—or groups of elementary items—are learned in the “chunking layer” , whereas the elementary items are learned in the “elementary layer” . For example in the phone number 8585342230 broken down in four chunks , 858-534-22-30 , each digit in a chunk is represented by a separate elementary unit , while every group of digits is represented by a chunking unit . This way , the chunking representation is a heteroclinic chain ( in the chunking layer ) of heteroclinic chains ( in the elementary layer ) . Earlier work described a similar model for the recognition of sequences of sequences [32] . Our previous work demonstrated a model of sequential spatial memory learning based on the WLC principle [33] . The dynamics was endowed with learning dynamics which led to the self-organization of WLC . To learn chunking sequences , we extend our previous model with a hierarchical neural network [21] , and augment it with bistable Hebbian plasticity dynamics [34] for unsupervised learning . Unsupervised here refers to the fact that learning is self-organized: During training , no external signal other than the perceptual information enters the dynamical system . The competitive dynamics in the cognitive network and the plasticity rules interact to learn a chunking representation of the sequence . Within each layer , the couplings in the system are initialized to a state where the network performs Winner-Take-All ( WTA ) : the node receiving the strongest input activates and all other node are silenced . When the couplings within a layer become sufficiently asymmetric , the dynamics within that layer switch from a WTA behavior [35] to a WLC behavior . At each layer the system learns chunks of information provided by the layer below it and stores syntactical information by modifying the couplings according to the directions indicated by the perceived items . After training , the system can reproduce the entire sequence by transitioning the activity of its corresponding modes in the same order .
Our dynamical model of chunking learning is composed of Perceptual Modes ( PMs ) , Elementary Modes ( EMs ) and Chunking Modes ( CMs ) . These are organized in a two-layer network plus a perceptual input layer , as shown in Fig 1 . The activity of the PMs is dictated by a pre-determined sequence of patterns , presented multiple times as a repeated loop . The PM project to NX EMs , according to a projection weight matrix P . The NY CMs receive excitatory input from the EMs according to a weight matrix Q and inhibit the EMs back through a weight matrix R . Here , we define inhibitory as couplings that result in a negative contribution to the node activity . Within the elementary and the chunking layer , the nodes have all-to-all inhibitory couplings , the weights of which are stored in competition matrices V and W , respectively . The two-layer chunking dynamics is a GLV system of the form: τ x d d t x i ( t ) = x i ( t ) ( ∑ k = 1 M P k i ( t ) s k ( t ) + b x - ∑ i ′ = 1 N X V i ′ i ( t ) x i ′ ( t ) - ∑ j = 1 N Y R j i ( t ) y j ( t ) ) + σ x η i ( t ) , τ y d d t y j ( t ) = y j ( t ) ( z j ( t ) + b y ) + σ y ξ j ( t ) , τ z d d t z j ( t ) = - z j ( t ) + ( ∑ i = 1 N X Q i j ( t ) x i ( t ) - ∑ j ′ = 1 N Y W j ′ j ( t ) y j ′ ( t ) + b z ( t ) ) , ( 1 ) where state variables xi , yj represent compositions of brain activities such as population firing rates [36] , bx , by are the respective constant growth rates and ηi ( t ) , ξj ( t ) are random ( Wiener ) processes with amplitudes σx and σy respectively . Perceptual modes sk ( e . g . visual or auditory cues ) stimulate the elementary modes xi , which in turn drive the chunking modes yj through variables zj . Variables zj convey the regulation between different brain domains or cognitive modes [22 , 37] . In our chunking model , we have used the simplest description that reminds the first order kinetic of synapses in spiking neuronal networks [38] . The τz is the characteristic time scale of zj that determines the temporal distance between different informational units ( i . e . those that would be part of different chunks ) by delaying the competition between different CMs [39] . Finally , bz ( t ) is a time-varying bias used to dynamically modulate chunking . We construct a dynamical learning model that concatenates sequence elements within one layer , and segments longer sequence portions in multiple groups ( chunks ) . Such two interacting processes are believed to be at the heart of chunking learning in the brain [5 , 7–9] . The key components of the learning model can be separated in two parts: 1 ) An asymmetric , bistable Hebbian learning rule within the WLC network learns the sequence ( order ) of the activity of the subordinate layer , by potentiating the weights corresponding to the transitions occurring in the elementary layer . The effect of this operation is to “concatenate” informational items , such that , during recall the same order is reproduced in a robust fashion . Hebbian learning within the WLC layer has been previously demonstrated in [33] , but the proposed learning rule had a single fixed point . By selecting the two fixed points of the bistable rule according to the bifurcation of the SHC ( one above the bifurcation point , one below ) , bistability renders the learning much more robust and prevents the formation of spurious channels . 2 ) The connections between two consecutive layers are learned through a symmetric , bistable Hebbian rule . This rule causes a superordinate layer to associate one ( or more ) modes to a group of modes in a subordinate layer . The WLC dynamics in a superordinate layer causes the network to transition its active mode , causing it to associate one mode to a finite number of modes of a subordinate layer . The association to a finite number of modes guarantees the chunking process in the learning . The number of modes within one chunk depends on the learning dynamics and the WLC dynamics in each layer . In particular we show that the size of the chunk is further bounded by the ratios of the potentiation vs . depotentiation magnitudes . This effect is further explained and quantified in section Learning dynamics determine chunk size . For these two learning rules , we used the bistable rule demonstrated in [34] . This rule has been demonstrated to reproduce many of the learning curves observed in experiments , and its dynamics are well understood . Similarly to [21 , 32] , we can construct a hierarchy for chunking learning by setting the time constant of a superordinate layer larger than the time constant of the subordinate layer . In addition to the learning rules above , the elementary layer learns to associate one mode to each element in the sequence through competitive learning [40 , 41] . Such learning has been extensively documented and shown to perform the Expectation-Maximization algorithm [41] , and is thus robust to the noise in sensory modes . Fig 1 illustrates chunking learning before and after training . In this example , a sequence composed of five patterns symbolized as a , b , c , d , and e , is presented multiple times during the learning phase . Distinct modes associate to each of the five patterns through weights of the projection matrix Pki . For example , in Fig 1 the weights in the directions a to b , b to c , and d to e are weakened ( arrow thickness denotes coupling strength ) , while the weights in the opposite direction are strengthened . The same learning dynamics apply to the inhibitory couplings between the chunking modes . In this illustration , three chunks are learned: ab , c and de . Fig 2 ( right ) shows a projection of the phase portrait of the chunking dynamics obtained after learning . Before learning , the network reaches stable fixed points , which appear as red “spikes” in Fig 2 ( left ) . This example illustrates how learning endows the network with a closed chunking sequence ( black ) that consists of several heteroclinic cycles that represent the chunks , which appear as red triangles in Fig 2 ( right ) . In general , the number of elementary items in each chunk are different and the chunking sequence can be open . In the three following paragraphs , we detail the learning dynamics between the sensory layer , the elementary layer , and the chunking layer . We examined the ability to learn and recall sequence of patterns of a network with the architecture described above with 3 CMs , 24 EMs and 144 PMs , as well as its ability to perform chunking . The sensory input consisted of 24 different patterns that were presented sequentially . The patterns were composed of 144 pixels that were binary for presentation simplicity . Each input pattern was composed of 6 high-intensity pixels and 138 low-intensity pixels . The high/low pixels for each pattern were selected such that there was no overlap between inputs , meaning that the position of the high-intensity pixels were different than those of the low-intensity pixels . For simplicity , we chose a stimulus that consisted of 24 , non-overlapping horizontal bars . A previous analysis of the learning rule of Pki showed that the shape of the patterns can be arbitrary , but the overlap and the relative sizes of the patterns increases the difficulty of the learning task [41] . Fig 3 shows the input patterns and the activity of the EMs and CMs during learning and sequence recall . For visualization purposes we present the activity of the PMs grouped according to their activation time . While chunks can be formed of informational items that have some clear association with each other , chunking can also occur spontaneously , i . e . in the absence of clear structure in the stimuli [7] . In this section , we show chunking in the case of spontaneous chunking . During the training phase , the sequence was repeatedly presented in a closed loop . After an initial transient in which EMs compete against each other , a given input pattern activates the same EM consistently ( Fig 3 , top ) . Similarly , the CMs always activate with the same subset of about 8 EMs . The resulting associations between PMs and EMs , and EMs and CMs are determined by the random variations present at the beginning of the learning . Therefore , each simulation run produced different association maps , similarly to the subject-specific chunking patterns during in behavioral experiments in the human [8] . After learning , the system is able to reproduce the sequence: EMs and CMs are driven with constant growth terms bx and by to reproduce the activity in a periodic and continuous cycle ( Fig 3 , bottom ) . The order of the sequences were often reproduced perfectly , but the timing depends on the dynamics of the model . Namely , we observe the appearance of pauses in the EMs between chunks reminiscent of those observed in behavioral studies [7 , 8] . The weights of the competition matrices , V and W , transition from a WTA configuration at the beginning of the learning to a WLC dynamics after learning ( see Fig 4 ) . Initially , the couplings are all-to-all inhibitory , leading to WTA . After learning , V and W become asymmetric , leading to WLC in both layers . The arrows in Fig 4 illustrate the succession of the state transitions in the resulting WLC . The matrices R and Q evolve to store the chunk association map . Fig 4 ( Bottom ) shows that weights in the matrices Q and R form three groups with similar weights which correspond to the chunks . The patterns presented to the system are stored in the synaptic weights of the projection matrix P . Successive presentations of the input pattern modify P such that the presented patterns are stored ( see Fig 5 ) . The results above used a small chunking layer ( Ny = 3 ) in order to illustrate the model . However , the dynamics of chunking during learning are much more interesting for a large chunking layer , since the number of possible state trajectories grows factorially with the size of the network [23] . For this reason , in the results below , we test the model for Ny = 30 and Nx = 30 . The training of the model consisted of multiple epochs . Each epoch consisted of a full sequence presentation phase , immediately followed by a recall phase . After the sequence had ended , the recall phase was initiated by cueing the network with the first element of the sequence and observing the ensuing sequence of patterns in the elementary layer . During the recall phase , the parameters of the network were kept fixed ( no learning ) . We quantified recall by computing the normalized Levenshtein distance between the presented sequence and the reproduced one ( see Methods—Characterizing Sequence Recall ) . Using the Levenshtein distance , we observe that overall 95% of the elements in the sequence were reproduced . The progress of chunking learning is monitored by inspecting the magnitude of the chunking and the presence of sequential activity in the chunking layer during recall . The magnitude of the chunking is monitored by computing the chunking rate during learning , defined as the number of transitions taking place in the chunking layer during the presentation of each pattern in the sequence . A chunking rate equal to 1 signifies that a different CM was active for each pattern in the sequence ( no chunking ) , while a chunking rate significantly smaller than one during training implies that chunks were formed . Note that a measure based on sequence recall only is not sufficient to characterize chunking since accurate recall is possible without the chunking layer . To further assess the robustness of the chunking in the presence of noise in the sensory layer , a fixed noise drawn from a rectified Gaussian distribution was independently added to each pixel at each presentation of a sequence element ( see also section 3 of S1 Text ) . Sequence recall accuracies ( measured using the Levenshtein distance ) and the chunking rates degraded gracefully as the noise magnitude was increased . We observe that the boundaries of the chunks can change from trial to trial during training , and that chunks can undergo substantial reconfigurations throughout the learning , including the creation of new chunking modes . The dynamical nature of chunking was already observed in behavioral experiments , where chunk boundaries could vary substantially even after a large number of trials [7 , 46] . [46] use a Bayesian algorithm combining reaction time and error rates to reveal the chunking structure in humans performing a discrete sequence production . Interestingly , the chunking structure also evolves slowly over the course of the trials . A visual inspection of our model results suggests that this slow evolution might be caused by the enrollment of new chunking modes and the disenrollment of existing ones ( see Fig 6 , right panel ) . Chunks in motor learning are often identified by the pauses between successive actions [49] . More specifically , psycholinguistic studies often focus on pauses between words and utterance-final syllable prolongations [50] , which are indicative of a hierarchical organization of the overall speech production apparatus [10] . Other experiments also show the hierarchical organization of information in chunks when performing other visuo-motor tasks [5–9] . The network activity in our model exhibits a temporal structure that is reminiscent of these studies . In the recall phase , the network activity is paused until the new chunk has been “loaded” ( Fig 3 ( c ) , dashed lines in Fig 3 ( b ) ) . The pauses in the chunking are a result of the synchronization between elementary chunking layers . The duration of the EM and the CM activations depend on the magnitude of the growth terms bx and by , but the two layers are bound to each other by the feedback connections Qij and Rji . As a consequence , the EMs are delayed until the next chunk in the sequence is activated . The function of the pause is therefore to synchronize the activity of the CM and the sequential activity of the EM belonging to this chunk , and therefore depends on the relative speed between the elementary layer and the chunking layer . The duration of the pause is variable and did not depend on the number of items in each chunk . In [7] , the pause is assumed a direct result of two interacting processes running in parallel: one segmenting long sequential structures into shorter ones , and one process concatenating these same groups of motor elements into longer sequences . In our model , the ongoing competition within the layer and the cooperation between its layers are also two interacting parallel processes as in [7] . Concatenation in our model is performed by the competitive process along a given layer , while segmentation is performed by the cooperative couplings between layers . Our model is therefore consistent with the one described in [7] . In the learned state , we find that the number of items in each chunk depends on the learning dynamics and the time constant in the synaptic dynamics z ( Fig 7 ) . The chunk size is the result of an equilibrium between competing learning processes in the dynamics . The size of the chunk is bounded by the magnitude of the Qij and Rij potentiation when xi and yj are co-active , and the magnitude of the depotentiation when other elements xi′ , i′ ≠ i belonging to the same chunk are active . This is because a coupling between a CM and EM undergoes depotentiation when other EM belonging to the same CM are active . The maximum number of elements in a chunk will therefore be limited by how much a CM and a EM potentiate when both are active versus the magnitude of the depotentiation when only the CM is active ( and other EMs belonging to that chunk are active ) . This observation suggests the important result that the neural mechanisms for acquiring the chunking sequence also play a role in determining the capacity of chunking sequential memory , and lead to new experimental predictions . For example , there is evidence that dopamine modulates the cortico-striatal plasticity chunking during motor sequence learning in humans and monkeys . In monkeys the learning of new sequences was significantly affected by injection of a dopamine receptor antagonist , but did not affect sequences that were learned prior to the injection [47] . In the context of our model , this dopamine related modulation could translate into reducing γp or increasing γd . For example , if γp were gradually reduced , our model would predict a gradual decrease in chunk sizes in a chunking task such as those conducted in [7 , 8] ( e . g . Fig 7 , left ) . Note that not all of the chunking units are used to learn and recall the presented sequence , and therefore they remain available for the learning of other sequences . Chunk size can also be modulated within the sequence , by injecting a time-varying input into the synaptic variable zk . We observe that the chunk size is proportional to the magnitude of this input S2 Fig . A neural analog of this modulation can be viewed as top-down attention [48] , where sequential attention switching between multimodal mental activities depend on internal or external cues .
The number of sequences that can be stored simultaneously in the network is the total number of elements in all the learned sequences , since one unit is required for a single element of a sequence . In the case of a closed SHC , the number of different sequences that the SHC can store is equal to the number of distinct channels than can be formed with N nodes , which is of order exp ( 1 ) ⋅ ( N − 1 ) ! [23] . We note however , that under reasonable neuro-biological perturbations of the recurrent connectivity , the capacity is reduced . In that case , the maximal sequence length that can be stably recalled is about 7 [57] . Our model raises new questions on chunking capacity and recall under such perturbations . The benefit of chunking can be studied by comparing the maximal length of sequence in the presence or absence of chunking . This study is complicated by the fact that the average chunk size in the network is strongly dependent on the parameters of the learning dynamics ( Fig 7 ) , and is the target of future work . Note that for simplicity , our current model cannot learn sequences that have recurring patterns . However this is possible in principle since other closely related work dealt with recurring patterns in sequences by retaining a memory of the past patterns in the sequence [58 , 59] or by using “template” connectivity matrices [32] . The learning in the elementary layer of our model shares many features with models of competitive learning [60 , 61] and self-organizing maps [62] . In competitive learning , each stimulus is compared with a feature vector stored at each neuron . The neuron with the highest similarity is selected as the winner , and the feature vector is updated . This mechanism is similar to the effect of learning in the projection matrix P and the competitive dynamics in the WLC in our model . Our model extends this idea further by embedding the order of the stimuli in the network as winnerless competition dynamics . Our model bears strong similarities with previous work in the recognition of sequences of sequences [32 , 63 , 64] . Kiebel et al . study the recognition of complex sequences , where the generative model is assumed a priori [32] . There , the within-layer connectivity matrix is modulated by activity in supra-ordinate levels . In contrast , feedback in our model is an additive term whose effect is to turn on or off circuits ( SHCs ) in the subordinate layers . This modeling choice comes at the cost of more nodes , but does not require the modulation of the connections . While the model presented in [64] addressed the learning of sound sequences , it did not address the learning of chunks ( i . e sequences of sequences ) . Other related methods for learning sequences in brain-inspired models are reservoir computers [65–67] , synfire chains [68–70] and chains of WTA networks [71] . The idea of exploiting asymmetrically coupled networks for sequence learning was reported in multiple works based on attractor networks [45 , 58 , 65 , 69 , 72–74] . The novelty of our approach is the learning of the hierarchical dynamics as a sequence of metastable states . Hence , our model offers a non-linear dynamical perspective on the problem of hierarchical sequence learning in neural substrates that is fundamentally different from attractor networks . Another attempt to map this type of dynamics on the cortex is the hierarchical temporal memory model [75] , although that work does not address the dynamics of biologically inspired learning of hierarchical sequences . Stability can be viewed from two related perspectives: robustness of the dynamics to noise in the nodes and in the connections ( structural stability ) ; and stability of the metastable states , i . e . their Lyapunov exponents . In either case , the study of learning stability in the general case is notoriously difficult , because the addition of new information-carrying items can destroy existing metastable states for example by creating spurious attractors [76] . In the three dimensional case , the Lotka Volterra dynamics can be thoroughly analyzed . However , many more difficulties appear in four or more dimensions , such as new metastable states in the phase space of the system , making the analysis much more difficult [36] . However , it is possible to gain some insight in the asymptotic case where the time scales in the system are well separated . In our case these are arranged such that P reaches equilibrium before V , V before Q , W before R . The overall dynamics of the elementary P associates stimulus items to neurons through a competitive learning mechanism and can be thoroughly analyzed . Because P modulates the increment to the nodes , it does not interfere with the structure of the elementary network . As long as LTP and LTD in the couplings V and W are balanced and the transitions in the network are monotonic , the weights in the network tend to a WLC configuration ( see section 1 of S1 Text ) . The dynamics of the synapses between EM and CM capture the chunking behavior , and are very similar to the P dynamics . It segments the chain of activations in the elementary layer into chunks , by detecting change points in the sequence . Its function is comparable to sequence segmentation using the sliding window algorithm commonly used for online natural language processing [77] . In this asymptotic case , the parameters can be selected manually such that learning at each time scale progresses as described above . In some cases , the model failed to recall the chunking sequences , especially when the parameters of learning dynamics were not appropriately chosen . The scenarios through which recall fails is of particular interest because these can provide insights into the dynamical causes of chunking deficits in neurodegenerative diseases , such as Parkinson’s disease . The most common cause of failing to learn was that a transition between two EM’s did not form , or was not strong enough to drive it . As a result , the state of the network remained “stuck” and is reminiscent of certain motor disorders observed in Parkinson’s patients . The recall typically resumes by providing a stimulus corresponding to an item in the cue , which is consistent with how sensory cues can improve symptoms of bradykinesia [78] . Similar behavioral observations were made on elderly who could not learn motor chunks during a sequence production task [79] . In the elderly , reduced cognitive abilities impede the learning of motor chunks , although most of the tested individuals were capable of correctly reacting to the stimuli that indicated the sequence to recall . In our model , this is equivalent to a successful learning between the perceptual layer and the elementary layer , but failing to learn the weights within the elementary layer . In other cases where learning failed , the chunking modes did not reach a WLC configuration , although the sequential structure was learned in the elementary layer . The result is that the activity in the chunking layer remained constant and did not affect the sequential structure of the EMs activations . This shortcoming was revealed in the elementary layer by the lack of pauses during the sequence recall . In this paper , we proposed a model of hierarchical chunking learning dynamics that can represent several forms of cognitive activities such as working memory and speech construction . This model is capable of learning patterns and their order as metastable states of a hierarchical SHC , and reproduces several key features observed in chunking behavior in humans . The model and the results outlined in this paper sheds new light onto the formation of sequential working memory and chunking . Complex action ( such as speech or song production ) can be viewed as a chain of subordinate movements , which need to be combined according to a syntax in order to reach a goal . Recent studies suggest that failures in reaching a functional configuration of the couplings is related to other diseases such as schizophrenia [39] , obsessive-compulsive disorder [80] , and Parkinson’s . Our model can generalize the dynamical image of these diseases by taking into account learning and chunking dynamics , in order to provide novel insights into treating them .
Our overarching hypothesis is that cognitive function in the brain is described by the non-linear interaction of brain “modes” . The number of these modes is assumed much smaller than the number of variables required to describe the state of the brain ( e . g . membrane potentials , channel states ) . Backed by recent brain imaging techniques , we follow a top-down approach for identifying the nature of these modes , and how they interact in a transient , robust and scalable fashion to process information [36 , 81] . In this context , a mode is defined as a metastable composition of elements from different brain areas that activate coherently to perform a specific cognitive task . Here , we focus on the cognitive task of recalling a sequence , which can be described by the sequential activation of brain modes . In particular , our approach is based on spatiotemporal mental modes that contain metastable states as equilibrium points since it resolves the contradiction by which the system must be robust to noise and , at the same time , sensitive to inputs [52–54] . Metastable states are semi-transient signals that can be represented as saddle nodes . These saddle nodes can be arranged to form a SHC , which consists of a sequence of successive states that are connected through their respective unstable separatrices ( Fig 8 ) . Under appropriate parametrizations , namely if the compressing of phase space around the saddle is larger than the stretching and if all saddles in the chain are dissipative , then the trajectories in the neighborhood of the metastable states that form the chain remain in the channel [22] . The GLV dynamics is a canonical model for implementing a SHC [42]: d d t x i ( t ) = x i ( t ) ( s i ( t ) - ∑ i ′ = 1 N X V i ′ i ( t ) x i ′ ( t ) + η i ) , ∀ i = 1 , ⋯ , N ( 5 ) The terms Vii′ determine the interaction between the variables xi , and ηi is an additive noise term . This asymmetry in Vii′ installs metastable nodes in the network , which results in successive and temporary winners as in WLC dynamics [23] . The simplicity of this model enables theoretical study of the transient solutions representing sequential competition [42] . The dynamical features of the system Eq ( 5 ) extend to a wide class of dynamical systems , known as Kolmogorov models [26] . The biological relevance of these models is confirmed by several previous works [28–30] . The state variables in Eq ( 5 ) are modes that represent abstract quantities that do not necessarily map directly or exactly onto individual neuron or populations activities . For instance , [29] show the existence of a SHC in a network of inhibitory Hodgkin Huxley-type ( H&H ) neurons short-term synaptic depression , despite that the differential equations there differ significantly from Eq ( 5 ) . Another example is given by [28] , which describes the conditions under which the firing rate of leaky Integrate & Fire ( I&F ) neurons approximately map onto Eq ( 5 ) . The hierarchical chunking dynamics is represented by robust transient activity modes at each scale of the hierarchy . The above Eq ( 5 ) serves as an elementary building block for each layer of the chunking dynamics . The two-layer chunking dynamics is a GLV system of the form of Eq ( 1 ) . This model has slight modifications to the one presented in [21] , which reflect the necessities for chunk formation during training . Firstly , the polarity of the couplings between the two layers is reversed ( in [21] elementary modes inhibit chunking modes ) . This modification allows the elementary modes to directly drive a CM . Secondly , the synaptic dynamics represented by the dimension z are applied to the growth terms of the chunking layer ( in contrast to [21] , where only inhibitory couplings are subject to synaptic dynamics ) . The synaptic dynamics helps a single CM to remain active over several items in the stimulus . The structure of the sequential activity is determined by the connectivity matrix among the respective modes . Within each layer , the amount of asymmetry in the couplings represents an order parameter that controls the dynamical behavior of the network . The inter-layer connections represent the association of the information-carrying items and chunks with the modes . After the presentation of the inputs , the network is run for a consolidation time , and the weights are held fixed to the values reached at the end of this time for recall . The learning can be understood as the adjustment of this order parameter and the associations in a way that the recall dynamics of the elementary and the chunking modes is consistent with the training sequences . At the end of successful training , the network is able to recall the presented sequences . Successful recall is defined when the sequence order is produced with perfect accuracy . However , it occurred that the sequence was reproduced to a reasonable extent ( e . g . missing elements , sequence reproduced correctly up to certain element ) . To take into account such events , we used a normalized Levenshtein distance to estimate the quality of the reproduction [83] . This distance computes the number of changes between two sequences ( addition , subtraction ) , normalized by the length of the longest sequence . Note that sequence recall does not characterize chunking since accurate recall can be obtained without learning in the chunking layer . | Because chunking is a hallmark of the brain’s organization , efforts to understand its dynamics can provide valuable insights into the brain and its disorders . For identifying the dynamical principles of chunking learning , we hypothesize that perceptual sequences can be learned and stored as a chain of metastable fixed points in a low-dimensional dynamical system , similar to the trajectory of a ball rolling down a pinball machine . During a learning phase , the interactions in the network evolve such that the network learns a chunking representation of the sequence , as when memorizing a phone number in segments . In the example of the pinball machine , learning can be identified with the gradual placement of the pins . After learning , the pins are placed in a way that , at each run , the ball follows the same trajectory ( recall of the same sequence ) that encodes the perceptual sequence . Simulations show that the dynamics are endowed with the hallmarks of chunking observed in behavioral experiments , such as increased delays observed before loading new chunks . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [] | 2015 | Learning of Chunking Sequences in Cognition and Behavior |
Understanding the genetic architecture of gene expression traits is key to elucidating the underlying mechanisms of complex traits . Here , for the first time , we perform a systematic survey of the heritability and the distribution of effect sizes across all representative tissues in the human body . We find that local h2 can be relatively well characterized with 59% of expressed genes showing significant h2 ( FDR < 0 . 1 ) in the DGN whole blood cohort . However , current sample sizes ( n ≤ 922 ) do not allow us to compute distal h2 . Bayesian Sparse Linear Mixed Model ( BSLMM ) analysis provides strong evidence that the genetic contribution to local expression traits is dominated by a handful of genetic variants rather than by the collective contribution of a large number of variants each of modest size . In other words , the local architecture of gene expression traits is sparse rather than polygenic across all 40 tissues ( from DGN and GTEx ) examined . This result is confirmed by the sparsity of optimal performing gene expression predictors via elastic net modeling . To further explore the tissue context specificity , we decompose the expression traits into cross-tissue and tissue-specific components using a novel Orthogonal Tissue Decomposition ( OTD ) approach . Through a series of simulations we show that the cross-tissue and tissue-specific components are identifiable via OTD . Heritability and sparsity estimates of these derived expression phenotypes show similar characteristics to the original traits . Consistent properties relative to prior GTEx multi-tissue analysis results suggest that these traits reflect the expected biology . Finally , we apply this knowledge to develop prediction models of gene expression traits for all tissues . The prediction models , heritability , and prediction performance R2 for original and decomposed expression phenotypes are made publicly available ( https://github . com/hakyimlab/PrediXcan ) .
Regulatory variation plays a key role in the genetics of complex traits [1–3] . Methods that partition the contribution of environment and genetic components are useful tools to understand the biology underlying complex traits . Partitioning heritability into different functional classes ( e . g . promoters , coding regions , DNase I hypersensitivity sites ) has been successful in quantifying the contribution of different mechanisms that drive the etiology of diseases [3–5] . Most human expression quantitative trait loci ( eQTL ) studies have focused on how local genetic variation affects gene expression in order to reduce the multiple testing burden that would be required for a global analysis [6 , 7] . Furthermore , when both local and distal eQTLs are reported [8–10] , effect sizes and replicability are much higher for local eQTLs . While many common diseases are likely polygenic [11–13] , it is unclear whether gene expression levels are also polygenic or instead have simpler genetic architectures . It is also unclear how much these expression architectures vary across genes [6] . Bayesian Sparse Linear Mixed Modeling ( BSLMM ) models complex traits as a mixture of sparse and polygenic contributions . The sparse component consists of a handful of variants of large effect sizes whereas the polygenic component allows for most variants to contribute to the trait albeit with small effect sizes . BSLMM assumes the genotypic effects come from a mixture of two normal distributions and thus is flexible to both polygenic and sparse genetic architectures as well as everything in-between [14] . The model is enforced by sparsity inducing priors on the regression coefficients . BSLMM allows us to directly estimate the sparse and polygenic components of a trait . As a somewhat independent approach to determine the sparsity and polygenicity of gene expression traits , we can look at the relative prediction performance of sparse and polygenic models . For example , if the true genetic architecture of a trait is polygenic , it is natural to expect that polygenic models will predict better ( higher predicted vs . observed R2 ) than sparse ones . We assessed the ability of various models , with different underlying assumptions , to predict gene expression in order to understand the underlying genetic architecture of gene expression . For gene expression prediction , we have shown that sparse models such as LASSO ( Least Absolute Shrinkage and Selection Operator ) perform better than a polygenic score model and that a model that uses the top eQTL variant outperformed the polygenic score but did not do as well as LASSO or elastic net ( mixing parameter α = 0 . 5 ) [15] . These results suggest that for many genes , the genetic architecture is sparse , but not regulated by a single SNP , which is consistent with previous work describing the allelic heterogeneity of gene expression [16–18] . Thus , gene expression traits with sparse architecture should be better predicted with models such as LASSO , which prefers solutions with fewer parameters , each of large effect [19] . Conversely , highly polygenic traits should be better predicted with ridge regression or similarly polygenic models that prefer solutions with many parameters , each of small effect [20–22] . Elastic net [23] is a good multi-purpose model that encompasses both LASSO and ridge regression at its extremes and has been shown to predict well across many complex traits with diverse genetic architectures [24] . Most previous human eQTL studies were performed in whole blood or lymphoblastoid cell lines due to ease of access or culturabilty [8 , 25 , 26] . Although studies with a few other tissues have been published , comprehensive coverage of human tissues was not available until the launching of the Genotype-Tissue Expression ( GTEx ) Project . GTEx aims to examine the genetics of gene expression more comprehensively and has recently published a pilot analysis of eQTL data from 1641 samples across 43 tissues from 175 individuals [27] . Here we use a much larger set of 8555 samples across 53 tissues corresponding to 544 individuals . One of the findings of the pilot analysis was that a large portion of the local regulation of expression traits is shared across multiple tissues . Corroborating this finding , our prediction model built in DGN whole blood showed robust prediction [15] across the nine tissues with the largest sample size from the GTEx Pilot Project [27] . This shared regulation across tissues implies that there is much to be learned from large sample studies of easily accessible tissues . Yet , a portion of gene regulation seems to be tissue dependent [27] . To further investigate the genetic architecture that is common across tissues or specific to each tissue , we use a mixed effects model-based approach termed orthogonal tissue decomposition ( OTD ) and sought to decouple the cross-tissue and tissue-specific mechanisms in the rich GTEx dataset . We show that the cross-tissue and tissue-specific expression phenotypes constructed with our OTD model reflect the expected biology .
We estimated the local and distal heritability of gene expression levels in 40 tissues from the GTEx consortium and whole blood from the Depression Genes and Networks ( DGN ) cohort . The sample size in GTEx varied from 72 to 361 depending on the tissue , while 922 samples were available in DGN [26] . We used linear mixed models ( see Methods ) and calculated variances using restricted maximum likelihood ( REML ) as implemented in GCTA [28] . For the local heritability component , we used common ( minor allele frequency > 0 . 05 ) variants within 1Mb of the transcription start and end of each protein coding gene , whereas for the distal component , we used common variants outside of the chromosome where the gene was located . Table 1 summarizes the local heritability estimate results across all tissues . In order to obtain an unbiased estimate of mean h2 across genes , we do not constrain the model to only output h2 estimates between 0 and 1 . Instead , as done previously [10 , 29] , we allow the h2 estimates to be negative when fitting the model and thus refer to it as the unconstrained REML . This approach reduces the standard error of the estimated mean of heritability ( law of large numbers ) . For distal heritability , the errors in the individual heritability estimates were still too large to render a significant mean distal heritability , even in DGN whole blood , the tissue with the largest sample size ( S1 Fig ) . The local component of h2 is relatively well estimated in DGN whole blood with 59% of genes ( 7474 out of 12719 ) showing FDR < 0 . 1 ( Table 1 ) . It has been shown that local-eQTLs are more likely to be distal-eQTLs of target genes [30] . However , restricting the distal h2 estimates to known eQTLs on non-gene chromosomes discovered in a separate cohort ( see Methods ) did not improve distal h2 estimates . We examined the sensitivity of our local h2 estimates to uneven linkage disequilibrium ( LD ) across the genome using LDAK [31] , a method proposed to account for LD . Overall , we find good concordance between GCTA and LDAK estimates , with slightly lower LDAK estimates ( S2 Fig ) . Given the limited sample size we will focus on local regulation for the remainder of the paper . Across all tissues 15–59% of genes had significant estimates ( FDR < 0 . 1 ) as shown in Table 1 . Gene expression local heritability estimates were consistent between whole blood tissue from the DGN and GTEx cohorts ( Spearman’s ρ = 0 . 27 [95% confidence interval ( CI ) : 0 . 25–0 . 30] ) . In addition , the DGN estimates made here were consistent with those made in Price et al . [29] using an Icelandic family cohort ( ρ = 0 . 32 [0 . 30−0 . 35] ) . We also found that more heritable genes tend to be more tolerant to loss of function mutations ( Fig 1 ) . Presumably , such genes are more tolerant to both loss of function mutations and mutations that alter gene expression regulation . This result is consistent with the finding that more eQTLs are found in more tolerant genes [32] . Next , we sought to determine whether the local common genetic contribution to gene expression is polygenic or sparse . In other words , whether many variants with small effects or a small number of large effects were contributing to expression trait variability . To do this , we used the BSLMM [14] approach , which models the genetic contribution as the sum of a sparse component and a highly polygenic component . The parameter PGE in this model represents the proportion of genetic variance explained by sparse effects . Another parameter , the total variance explained ( PVE ) by additive genetic variants , is a more flexible Bayesian equivalent of the heritability we have estimated using a linear mixed model ( LMM ) as implemented in GCTA . For genes with low heritability , our power to discern between sparse and polygenic architecture is limited . In contrast , for highly heritable genes , the credible sets are tighter and we are able to determine that the sparse component is large . For example , the median PGE was 0 . 99 [95% CI: 0 . 94–1] for genes with PVE > 0 . 50 ( Fig 2A ) . The median PGE was 0 . 95 [95% CI: 0 . 72–0 . 99] for genes with PVE > 0 . 1 . Fittingly , for most ( 96 . 3% ) of the genes with heritability > 0 . 10 , the number of SNPs included in the model was below 10 . To further confirm the local sparsity of gene expression traits , we looked at the prediction performance of a range of models with different degrees of polygenicity , such as the elastic net model with mixing parameter values ranging from 0 ( fully polygenic , ridge regression ) to 1 ( sparse , LASSO ) . We performed 10-fold cross-validation using the elastic net [23] to test the predictive performance of local SNPs for gene expression across a range of mixing parameters ( α ) . Given common folds and seeds , the mixing parameter that yields the largest cross-validation R2 informs the degree of sparsity of each gene expression trait . That is , at one extreme , if the optimal α = 0 ( equivalent to ridge regression ) , the gene expression trait is highly polygenic , whereas if the optimal α = 1 ( equivalent to LASSO ) , the trait is highly sparse . We found that for most gene expression traits , the cross-validated R2 was smaller for α = 0 and α = 0 . 05 , but nearly identical for α = 0 . 5 through α = 1 in the DGN cohort ( Fig 3 ) . An α = 0 . 05 was also clearly suboptimal for gene expression prediction in the GTEx tissues , while models with α = 0 . 5 or 1 had similar predictive power ( S3 Fig ) . Together with the BSLMM results , this suggests that given current data , the effect of local common genetic variation on gene expression is sparse rather than polygenic for most genes . In DGN , there is a strong correlation between BSLMM-estimated PVE and GCTA-estimated h2 ( Fig 2B , R = 0 . 96 ) . In contrast , when we applied BSLMM to the GTEx data , we found that many genes had measurably larger BSLMM-estimated PVE than LMM-estimated h2 ( Fig 4 ) . This is further confirmation of the predominantly sparse local architecture of gene expression traits: the underlying assumption in the LMM approach to estimate heritability is that the genetic effect sizes are normally distributed , i . e . most variants have small effect sizes . LMM is quite robust to departure from this assumption , but only when the sample size is rather large ( S4 Fig ) . For the relatively small sample sizes in GTEx ( n ≤ 361 ) , we found that directly modeling the sparse component with BSLMM ( polygenic + sparse components ) outperforms LMM ( single polygenic component ) for estimating h2 . Here , unlike in previous sections ( see Table 1 and S1 Fig ) , both the BSLMM and LMM model estimates are constrained to be between 0 and 1 . To ensure that the high values of heritability estimated by BSLMM are due to actually higher h2 and not to limitations or bias of the BSLMM , we use the fact that cross-validated prediction R2 ( observed vs predicted squared correlation when prediction parameters are estimated without including the observations to be predicted ) is bounded by the heritable component . In other words , genetically based prediction models can only explain or predict the portion that is heritable . As shown in Fig 4 , most of the genes with high BSLMM heritability but low LMM h2 have large values of prediction R2 . We further validated this behavior by showing Price et al . [29] identity-by-descent ( IBD ) heritability estimates correlate more strongly with BSLMM than LMM estimates ( S5 Fig ) . For all tissues , correlation is higher between IBD estimates and BSLMM than between IBD and LMM . These findings provide strong evidence that LMM is underestimating h2 for these genes . Since a substantial portion of local regulation was shown to be common across multiple tissues [27] , we sought to decompose the expression levels into a component that is common across all tissues and tissue-specific components . Fig 5 shows a diagram describing the decomposition for which we use a linear mixed effects model with a person-level random effect ( see Methods ) . We use the posterior mean of this random effect as an estimate of the cross-tissue component . We consider the residual component of this model as the tissue-specific component . We call this approach orthogonal tissue decomposition ( OTD ) because the cross-tissue and tissue-specific components are assumed to be independent in the model . First , we sought to demonstrate that OTD is able to identify the cross-tissue and tissue-specific components via simulations . We generated simulated traits so that the true cross-tissue and tissue-specific components are known . To preserve the correlation structure between genes and tissues , we considered the estimated cross-tissue and tissue-specific components as the true values . Then , we simulated expression traits as the sum of the cross-tissue and tissue-specific components and an error term ( expression trait = cross-tissue + tissue-specific + noise ) . We simulated expression traits using different magnitudes of the noise level and then applied our OTD approach . Simulated values were constructed using noise variance equal in magnitude to the original variance of the expression levels of each gene . This is a rather conservative choice since the variance of the noise component is unlikely to be larger than the total variance of the expression trait . In S6 Fig , we show for one simulated example the correlation between true and OTD-estimated components for one representative gene , SLTM , and the distribution of the correlation between true and OTD-estimates for all genes . The median Pearson correlation between true and OTD-estimted cross-tissue components was 0 . 54 [95% CI: 0 . 33–0 . 77] and for tissue-specific components was 0 . 75 [0 . 46–0 . 88] , demonstrating the robustness of our model ( S6 Fig ) . For the OTD derived cross-tissue and 40 tissue-specific expression phenotypes , we computed the local heritability and generated prediction models . The decomposition is applied at the expression trait level so that the downstream genetic regulation analysis is performed separately for each derived trait , cross-tissue and tissue-specific expression , which greatly reduces computational burden . Our estimates of h2 for cross-tissue expression traits are larger than the corresponding estimates for each whole tissue ( S7 Fig ) because our OTD approach increases the ratio of the genetically regulated component to noise by averaging across multiple tissues . In addition to the increased h2 , we observe reduction in standard errors of the estimated cross-tissue h2 . This is due , in part , to the larger effective sample size for cross-tissue phenotypes . There were 450 samples for which cross-tissue traits were available whereas the maximum sample size for whole tissue phenotypes was 362 . Similarly , cross-tissue BSLMM PVE estimates had lower error than whole tissue PVE ( S8 and S9 Figs ) . As for the tissue-specific components , the cross-tissue heritability estimates were also larger and the standard errors were smaller reflecting the fact that a substantial portion of regulation is common across tissues ( S10 Fig ) . The percentage of GCTA h2 estimates with FDR <0 . 1 was much larger for cross-tissue expression ( 20% ) than the tissue-specific expressions ( 6–13% , S1 Table ) . Similarly , the percentage of BSLMM PVE estimates with a lower credible set greater than 0 . 01 was 49% for cross-tissue expression , but ranged from 24–27% for tissue-specific expression ( S9 Fig ) . Cross-tissue predictive performance exceeded that of both tissue-specific and whole tissue expression as indicated by higher cross-validated R2 ( S11 Fig ) . Like whole tissue expression , cross-tissue and tissue-specific expression showed higher predictive performance when using more sparse models . In other words elastic-net models with α ≥ 0 . 5 predicted better than the ones with α = 0 . 05 ( S11 Fig ) . To verify that the cross-tissue phenotype has the properties we expect , we compared our OTD results to those from a joint multi-tissue eQTL analysis [33] , which was previously performed on a subset of the GTEx data [27] covering 9 tissues . In particular , we used the posterior probability of a gene being actively regulated ( PPA ) in a tissue . These analysis results are available on the GTEx portal ( see Methods ) . First , we reasoned that genes with high cross-tissue h2 would be actively regulated in most tissues so that the PPA of a gene would be roughly uniform across tissues . By contrast , a gene with tissue-specific regulation would have concentrated posterior probability in one or a few tissues . Thus we decided to define a measure of uniformity of the posterior probability vector across the 9 tissues using the concept of entropy . More specifically , for each gene we normalized the vector of posterior probabilities so that the sum equaled 1 . Then we applied the usual entropy definition ( negative of the sum of the log of the posterior probabilities weighted by the same probabilities , see Methods ) . In other words , we defined a uniformity statistic that combines the nine posterior probabilities into one value such that higher values mean the gene regulation is more uniform across all nine tissues , rather than in just a small subset of the nine . Thus , we expected that genes with high cross-tissue heritability would show high probability of being active in multiple tissues and have high uniformity measure . Reassuringly , this is exactly what we find . Genes with high cross-tissue heritability concentrate on the higher end of the uniformity measure ( Fig 6 , S12 Fig ) . For the original whole tissue , we expected the whole tissue expression heritability to correlate with the posterior probability of a gene being actively regulated in a tissue . This is confirmed in Fig 7A where PPA in each tissue is correlated with the BSLMM PVE of the expression in that tissue . In the off diagonal elements we observe high correlation between tissues , which was expected given that large portion of the regulation has been shown to be common across tissues . Whole blood has the lowest correlation consistent with whole blood clustering separately from other tissues [27] . In contrast , Fig 7B shows that the tissue-specific expression PVE correlates well with matching tissue PPA but the off diagonal correlations are substantially reduced consistent with these phenotypes representing tissue-specific components . Again whole blood shows a negative correlation which could be indicative of some over correction of the cross-tissue component . Overall these results indicate that the cross-tissue and tissue-specific phenotypes have properties that are consistent with the intended decomposition .
Motivated by the key role that regulatory variation plays in the genetic control of complex traits [1–3] , we performed a survey of the heritability and patterns of effect sizes of gene expression traits across a comprehensive set of human tissues . We quantified the local and distal heritability of gene expression in DGN and 40 different tissues from the GTEx consortium . Distal components estimates were too noisy to draw meaningful conclusions . Using results implied by the improved predictive performance of sparse models and by directly estimating sparsity using BSLMM ( Bayesian Sparse Linear Mixed Model ) , we show evidence that local common variant regulation is sparse across all the tissues analyzed here . Our finding that a handful of genetic variants seem to contribute to the variability in gene expression traits has important implications for the strategies used in future investigations of gene regulation and downstream effects on complex traits . For example , most fine mapping methods that attempt to find the causal SNPs that drive a GWAS locus focus on a limited number of variants . This only makes sense in cases where the underlying genetic architecture is sparse , that is , when a handful of causal variants are determining the variability of the traits . We note that any rare variants not tagged by common variants are not included in our gene expression heritability estimates and prediction models . For genes with moderate and low heritability the evidence is not as strong , due to reduced power , but results are consistent with a sparse local architecture . Sparse models capture the most variance in gene expression at current sample sizes and have been used successfully in gene expression prediction methods [15 , 34–36] . Better methods to correct for hidden confounders that do not dilute distal signals and larger sample sizes will be needed to determine the properties of distal regulation . Given that a substantial portion of local regulation is shared across tissues , we proposed here to decompose the expression traits into cross-tissue and tissue-specific components in order to facilitate biological interpretation . This approach , called orthogonal tissue decomposition ( OTD ) , aims to decouple the shared regulation from the tissue-specific regulation . The orthogonality is a feature that simplifies the interpretation of association results . Many groups have proposed integrating genotype and expression data to understand complex traits [15 , 22 , 35–38] . When we find a gene target of interest , a natural question is whether the association is driven by processes within a specific tissue or across all tissues . Through integration of our OTD expression traits with studies of complex diseases , we expect results from the cross-tissue models to relate to mechanisms that are shared across multiple tissues , whereas results from the tissue-specific models will inform us about the context specific mechanisms . We examined the genetic architecture of the derived cross-tissue and tissue-specific traits and found that they follow similar sparse patterns as the original whole tissue expression traits . The cross-tissue component benefits from an effectively larger sample size ( 16 tissues/person on average ) than any individual tissue trait , which is reflected in more accurate heritability estimates and consistently higher prediction performance . Encouragingly , we find that genes with high cross-tissue heritability tend to be regulated more uniformly across tissues . As for the tissue-specific expression traits , we found that they recapitulate correlation with the vector of probability of tissue-specific regulation . In this paper , we quantitate the genetic architecture of gene expression and develop predictors across tissues . We show that local heritability can be accurately estimated across tissues , but distal heritability cannot be reliably estimated at current sample sizes . Using two different approaches , BSLMM and the elastic net , we show that for common local gene regulation , the genetic architecture is mostly sparse rather than polygenic . Using new expression phenotypes generated in our OTD model , we show that cross-tissue predictive performance exceeded that of both tissue-specific and whole tissue expression as indicated by higher elastic net cross-validated R2 . Predictors , heritability estimates and cross-validation statistics generated in this study of gene expression architecture are freely available ( https://github . com/hakyimlab/PrediXcan ) for use in future studies of complex trait genetics .
Motivated by the observed differences in regulatory effect sizes of variants located in the vicinity of the genes and distal to the gene , we partitioned the proportion of gene expression variance explained by SNPs in the DGN cohort into two components: local ( SNPs within 1Mb of the gene ) and distal ( SNPs on non-gene chromosomes ) as defined by the GENCODE [49] version 12 gene annotation . We calculated the proportion of the variance ( narrow-sense heritability ) explained by each component using the following mixed-effects model: Y g = ∑ k ∈ l o c a l w k , g local X k + ∑ k ∈ d i s t a l w k , g distal X k + ϵ where Yg represents the expression of gene g , Xk is the allelic dosage for SNP k , local refers to the set of SNPs located within 1Mb of the gene’s transcription start and end , distal refers to SNPs in other chromosomes , and ϵ is the error term representing environmental and other unknown factors . We assume that the local and distal components are independent of each other as well as independent of the error term . We assume random effects for w k , g local ∼ N ( 0 , σ w , local 2 ) , w k , g distal ∼ N ( 0 , σ w , distal 2 ) , and ϵ ∼ N ( 0 , σ ϵ 2 I n ) , where In is the identity matrix . We calculated the total variability explained by local and distal components using restricted maximum likelihood ( REML ) as implemented in the GCTA software [28] . In an effort to determine if distal heritability estimates could be improved , we also tested a mixed-effect model restricting the distal component to known eQTLs on non-gene chromosomes discovered in the Framingham cohort at FDR < 0 . 05 . When we found the distal estimate could not be improved , we focused on estimating local heritability without the distal component in the equation above . For the purpose of estimating the mean heritability ( see Table 1 , S1 Fig and S1 Table ) , we allowed the heritability estimates to take negative values ( unconstrained model ) . Despite the lack of obvious biological interpretation of a negative heritability , it is an accepted procedure used in order to avoid bias in the estimated mean [10 , 29] . Genes with FDR < 0 . 1 ( derived from the two-sided GCTA P-value ) were considered to have significant heritability . For comparing to BSLMM PVE , we restricted the GCTA heritability estimates to be within the [0 , 1] interval ( constrained model , see Figs 2 , 4 and 6 ) . We used BSLMM [14] to model the effect of local genetic variation ( common SNPs within 1 Mb of gene ) on the genetic architecture of gene expression . BSLMM uses a linear model with a polygenic component ( small effects ) and a sparse component ( large effects ) enforced by sparsity inducing priors on the regression coefficients [14] . BSLMM assumes the genotypic effects come from a mixture of two normal distributions and thus is flexible to both polygenic and sparse genetic architectures [14] . We used the software GEMMA [50] to implement BSLMM for each gene with 100K sampling steps per gene . BSLMM estimates the PVE ( the proportion of variance in phenotype explained by the additive genetic model , analogous to the heritability estimated in GCTA ) and PGE ( the proportion of genetic variance explained by the sparse effects terms where 0 means that genetic effect is purely polygenic and 1 means that the effect is purely sparse ) . From the second half of the sampling iterations for each gene , we report the median and the 95% credible sets of the PVE , PGE , and the |γ| parameter ( the number of SNPs with non-zero coefficients ) . We used the glmnet R package to fit an elastic net model where the tuning parameter is chosen via 10-fold cross-validation to maximize prediction performance measured by Pearson’s R2 [51 , 52] . The elastic net penalty is controlled by mixing parameter α , which spans LASSO ( α = 1 , the default ) [19] at one extreme and ridge regression ( α = 0 ) [20] at the other . The ridge penalty shrinks the coefficients of correlated SNPs towards each other , while the LASSO tends to pick one of the correlated SNPs and discard the others . Thus , an optimal prediction R2 for α = 0 means the gene expression trait is highly polygenic , while an optimal prediction R2 for α = 1 means the trait is highly sparse . In the DGN cohort , we tested 21 values of the mixing parameter ( α = 0 , 0 . 05 , 0 . 1 , … , 0 . 90 , 0 . 95 , 1 ) for optimal prediction of gene expression of the 341 genes on chromosome 22 . In order to compare prediction R2 values across α values , we used common folds and seeds for each run . For the rest of the autosomes in DGN and for whole tissue , cross-tissue , and tissue-specific expression in the GTEx cohort , we tested α = 0 . 05 , 0 . 5 , 1 . We use a mixed effects model to decompose the expression level of a gene into a subject-specific component and a subject-by-tissue-specific component . We fit the model one gene at a time and to simplify notation we assume the gene index , g , is implicit and drop it from the equations below . The expression Y of a gene g for individual i in tissue t is modeled as Y i , t , g = Y i , t = Y i CT + Y i , t TS + Z i ′ β + ϵ i , t where Y i CT is the random subject level intercept , Y i , t TS is the random subject by tissue intercept , Zi represents covariates ( for overall intercept , tissue intercept , gender , and PEER factors ) , and ϵi , t is the error term . We assume Y i CT ∼ N ( 0 , σ CT 2 ) , Y i , t TS ∼ N ( 0 , σ TS 2 ) , ϵ ∼ N ( 0 , σ ϵ 2 ) , and that all three terms are independent of each other . All variances are estimated by restricted maximum likelihood ( REML ) . Yi , t , g = Yi , t is a scalar . Zi is a vector of length p , which represents the number of covariates . β is a vector of length p and represents the effects of covariates on the expression level of the gene . The prime in Z i ′ β represents the inner product between the two vectors . All variances are estimated by restricted maximum likelihood ( REML ) . These variances will be different for each gene . The tissue-specific component’s variance is common across tissues . Differences between tissues will be reflected in the posterior mean of each tissue/individual random effects . For the cross-tissue component ( Y i CT ) to be identifiable , multiple replicates of expression are needed for each subject . In the same vein , for the tissue-specific component ( Y i , t TS ) to be identifiable , multiple replicates of expression are needed for a given tissue/subject pair . GTEx [27] data consisted of expression measurement in multiple tissues for each subject , thus multiple replicates per subject were available . However , there were very few replicated measurement for a given tissue/subject pair . Thus , we fit the reduced model and use the estimates of the residual as the tissue-specific component . Y i , t = Y i CT + Z i ′ β + ϵ i , t We consider the expression level of a gene at a given tissue for individual i to be composed of a cross-tissue component which we represent as Y i CT and a tissue-specific component , which given the lack of replicates , we estimate as the difference between the expression level and the cross-tissue components ( after adjusting for covariates ) . The assumptions are the same as the full model , i . e . , Y i CT ∼ N ( 0 , σ CT 2 ) , ϵ ∼ N ( 0 , σ ϵ 2 ) , and that both terms are independent of each other . The mixed effects model parameters were estimated using the lme4 package [53] in R . Batch effects and unmeasured confounders were accounted for using 15 PEER factors computed with the PEER [54] package in R . We also included gender as a covariate . Posterior modes of the subject level random intercepts were used as estimates of the cross-tissue components whereas the residuals of the models were used as tissue-specific components . The model included whole tissue gene expression levels in 8555 GTEx tissue samples from 544 unique subjects . A total of 17 , 647 protein-coding genes ( defined by GENCODE [49] version 18 ) with a mean gene expression level across tissues greater than 0 . 1 RPKM ( reads per kilobase of transcript per million reads mapped ) and RPKM > 0 in at least 3 individuals were included in the model . To test the robustness of our method , we generated simulated expression levels so that we could compare our estimates with the ground truth . To preserve correlation structure between genes and tissues , we used the cross-tissue and tissue-specific estimates of OTD based on the observed data as a basis for the simulations . Thus , we simulated expression phenotypes as the sum of the observed OTD cross-tissue and tissue-specific components plus an error term derived by randomly sampling a normal distribution with a standard deviation equal to 1 , 2 , or 10 times the variance of each gene’s component sum . We performed OTD on the simulated phenotypes and compared the OTD-estimated cross-tissue and tissue-specific expression levels for each gene to the true levels using Pearson correlation . As expected , higher levels of noise reduced the correlation between true and simulated values , but the correlation remained significant for all simulated phenotypes . To verify that the newly derived cross-tissue and tissue-specific traits were capturing the expected properties , we used the results of the multi-tissue eQTL analysis developed by Flutre et al . [33] and performed on nine tissues from the pilot phase of the GTEx project [27] . In particular , we downloaded the posterior probabilities of a gene being actively regulated in a tissue ( PPA ) from the GTEx portal at http://www . gtexportal . org/static/datasets/gtex_analysis_pilot_v3/multi_tissue_eqtls/Multi_tissue_eQTL_GTEx_Pilot_Phase_datasets . tar . PPA can be interpreted as the probability a gene is regulated by an eQTL in tissue t given the data . In the GTEx pilot , the most significant eQTL per gene was used to compute PPA [27] . PPA is computed from a joint analysis of all tissues and takes account of sharing of eQTLs among tissues [33] . For example , consider a SNP showing modest association with expression in tissue t . If this SNP also shows strong association in the other tissues , then it will be assigned a higher probability of being an active eQTL in tissue t than if it showed no association in the other tissues [33] . We reasoned that genes with large cross-tissue component ( i . e . high cross-tissue h2 ) would have more uniform PPA across tissues . Thus we defined for each gene a measure of uniformity , Ug , across tissues based on the nine-dimensional vector of PPAs using the entropy formula . More specifically , we divided each vector of PPA by their sum across tissues and computed the measure of uniformity as follows: U g = - ∑ t p t , g log p t , g where pt , g is the normalized PPA for gene g and tissue t . | Gene regulation is known to contribute to the underlying mechanisms of complex traits . The GTEx project has generated RNA-Seq data on hundreds of individuals across more than 40 tissues providing a comprehensive atlas of gene expression traits . Here , we systematically examined the local versus distant heritability as well as the sparsity versus polygenicity of protein coding gene expression traits in tissues across the entire human body . To determine tissue context specificity , we decomposed the expression levels into cross-tissue and tissue-specific components . Regardless of tissue type , we found that local heritability , but not distal heritability , can be well characterized with current sample sizes . We found that the distribution of effect sizes is more consistent with a sparse local architecture in all tissues . We also show that the cross-tissue and tissue-specific expression phenotypes constructed with our orthogonal tissue decomposition model recapitulate complex Bayesian multi-tissue analysis results . This knowledge was applied to develop prediction models of gene expression traits for all tissues , which we make publicly available . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"body",
"fluids",
"evolutionary",
"biology",
"gene",
"regulation",
"population",
"genetics",
"mathematics",
"forecasting",
"statistics",
"(mathematics)",
"genome",
"analysis",
"population",
"biology",
"research",
"and",
"analysis",... | 2016 | Survey of the Heritability and Sparse Architecture of Gene Expression Traits across Human Tissues |
During the first meiotic division , crossovers ( COs ) between homologous chromosomes ensure their correct segregation . COs are produced by homologous recombination ( HR ) -mediated repair of programmed DNA double strand breaks ( DSBs ) . As more DSBs are induced than COs , mechanisms are required to establish a regulated number of COs and to repair remaining intermediates as non-crossovers ( NCOs ) . We show that the Caenorhabditis elegans RMI1 homolog-1 ( RMH-1 ) functions during meiosis to promote both CO and NCO HR at appropriate chromosomal sites . RMH-1 accumulates at CO sites , dependent on known pro-CO factors , and acts to promote CO designation and enforce the CO outcome of HR-intermediate resolution . RMH-1 also localizes at NCO sites and functions in parallel with SMC-5 to antagonize excess HR-based connections between chromosomes . Moreover , RMH-1 also has a major role in channeling DSBs into an NCO HR outcome near the centers of chromosomes , thereby ensuring that COs form predominantly at off-center positions .
During meiosis , the accurate segregation of chromosomes relies on the formation of crossovers ( COs ) and sister chromatid cohesion [1] . COs are produced by homologous recombination ( HR ) -mediated repair of programmed DNA double strand breaks ( DSBs ) . In Caenorhabditis elegans , the holocentric chromosome pairs usually undergo a single off-center CO , resulting in a cruciform and asymmetric bivalent [2 , 3] . This is important to define the organization of kinetochore proteins , motor proteins , protein kinases , and the domains of cohesion release , which is necessary for the accurate segregation of homologs during anaphase I [4] . During prophase I , several mechanisms guarantee the formation of COs as well as the repair of the additional recombination intermediates as non-crossovers ( NCOs ) . First , to ensure CO formation , programmed DSBs are induced in excess compared to the actual number of COs , e . g . , [5–7] . Second , a subset of the early recombination intermediates are stabilized and protected by pro-CO factors [8–10] . CO maturation is supported by the MutSγ complex ( MSH-4/5 ) , sumo/ubiquitin ligases ZHP-3/RNF212 , and HEI10 and the cyclin-related protein COSA-1/CNTD1 [11–16] . In addition to efficient CO designation , it is essential that the resolution of joint molecules ( JM ) at CO-designated sites is biased toward the CO outcome . JMs result from second-strand capture and DNA synthesis after strand invasion [17] . Depending on where the cut is made by structure-specific endonucleases ( resolvases ) , CO or NCO products can be generated [18–21] . In C . elegans , at least two parallel pathways of a redundant resolvase system have been defined ( MUS-81/SLX-1 and XPF-1/HIM-6 ) [22–24] . It is unclear how the bias in resolution toward either a CO or NCO is achieved . In yeast , the bulk of DSBs not destined to become COs are processed by synthesis-dependent strand annealing ( SDSA ) at an early stage [25] . In C . elegans , a similar activity is attributed to RTEL-1 by D-loop displacement after strand invasion [26 , 27] . NCOs can also be generated by the activity of the RTR complex , composed of a RecQ helicase ( Bloom syndrome protein [BLM] ) , topoisomerase IIIα ( TOP3 ) , and the RecQ-mediated genome instability 1 and 2 ( RMI1 , 2 ) scaffolding proteins [28–30] . This complex provides a major NCO activity during mitosis . Importantly , patients with mutated BLM helicase display genomic instability and a cancer predisposition , and they exhibit a cellular phenotype of elevated COs between sister chromatids , which potentially drives cancer by loss of heterozygosity [31 , 32] . In vitro studies have shown that the RTR complex dismantles double Holliday junctions ( dHJs ) . A branch migration step mediated by the RecQ helicase brings the two dHJs in close proximity to allow the topoisomerase to unhook the two DNA strands by decatenation , stimulated by the scaffolding proteins RMI1/2 [33] . RMI1 was discovered as a protein associated with BLM containing protein complexes [34] . It colocalizes with recombination foci containing BLM helicase [35] . RMI1 depletion leads to destabilization of BLM and topoisomerase foci . Moreover , its depletion phenocopies the BLM mutant phenotype [35] . Furthermore , RMI1 directly interacts with both BLM helicase and TOP3 , shown with pulldown experiments using recombinant protein [36] . Human RMI1 protein contains two oligonucleotide/oligosaccharide-binding ( OB ) fold domains , in which the N-terminal OB fold binds both the helicase and the topoisomerase [37] . Structural and biochemical studies established that RMI1 cooperates with the topoisomerase in the strand passage reaction during decatenation . The topoisomerase-generated nick allows strand passage of the other strand , in which RMI1 plays a role in the regulation of opening and closure of the “gate” [38 , 39] . An “anti-CO” role has been described for BLM orthologs in yeast , plants , mice , and Drosophila in meiosis [25 , 40–45] . A recent study showed that , in C . elegans , him-6 is required to eliminate persistent MutSγ-independent recombination intermediates when an excess is induced by irradiation [46] . Similar “anti-CO” functions have been attributed to other members of the complex: Rmi1 and/or Top3 in yeast and Arabidopsis [47–52] . In yeast , Top3 and Rmi1 cooperate with Sgs1 to prevent aberrant joint molecule ( JM ) accumulation , but they also act later , without Sgs1 , to allow chromosome separation [51 , 52] . Despite the widely conserved anti-CO function of the RTR complex , several studies suggest that it could also contribute a pro-CO activity; however , a direct involvement has not been demonstrated . In C . elegans , him-6 ( the BLM ortholog ) is required to ensure that CO-designated recombination intermediates mature into interhomolog COs [46 , 53 , 54] . In yeast , Sgs1 has been found colocalizing with the pro-CO factor Zip3 [40] . Nevertheless , in sgs1 mutants , the recombination rate is increased rather than decreased , as one would expect if it favored CO formation . More recently , the Sgs1/BLM helicase has been proposed to have an early role in promoting COs , acting to disassemble unprotected strand invasion intermediates to channel them back into CO pathways [41 , 44 , 45 , 55 , 56] . Furthermore , in a situation in which JM resolution is blocked ( mms4 slx4 yen1 ) , Sgs1 can promote CO formation . Here , sgs1 might generate JMs with a certain configuration that could be recognized by the major yeast CO pathway Exo1-MutLγ [55] . In this study , we present the role of C . elegans RMH-1 . We show that RMH-1 and HIM-6 localize at numerous recombination intermediates that will become either a CO or NCO . Consistent with its localization , RMH-1 plays genetically separable and antagonistic roles during meiosis . On one hand , rmh-1 ensures CO formation at two levels: by promoting CO designation and by protecting CO-designated sites and promoting their maturation into chiasmata . Remarkably , RMH-1 and HIM-6 localize as closely juxtaposed doublets , suggesting that they may flank the two junctions of a dHJ CO intermediate . In addition , rmh-1 prevents illegitimate connections between homologs redundantly with smc-5 , an activity known from yeast meiosis to prevent multi-JMs [57–59] . Moreover , RMH-1 anti-CO activity is required to inhibit CO formation at chromosome centers , which would facilitate correct segregation of chromosomes .
The C . elegans genome encodes two RMI1 homologs: M01E11 . 3 and T07C12 . 12 , named RMI1 homolog-1 ( RMH-1 ) and -2 ( RMH-2 ) , respectively . These proteins contain three domains characteristic of RMI1 orthologs: an N-terminal domain of unknown function ( DUF ) 1767 and two OB-fold domains ( Fig 1A and S1A Fig ) . Point mutations in the recessive alleles rmh-1 ( jf54 ) and rmh-1 ( tn309 ) were isolated based on defects in meiotic chromosome segregation and embryonic lethality [60–62] . The mutations lead to protein truncations after the DUF1767 or OB1 domains , respectively ( Fig 1A′ and S1B–S1D Fig ) . In rmh-1 ( jf92 ) and rmh-2 ( jf94 ) , the ORFs were disrupted right after the start codon ( see the Experimental Procedures section ) . To establish the respective requirements of rmh-1 and rmh-2 , we examined the mutants for embryonic lethality , larval arrest , and a high incidence of males ( Him ) phenotype among the progeny . An increased incidence of males in the progeny indicates X chromosome nondisjunction during meiosis , as C . elegans hermaphrodites have the genotype X/X and males X/O . The combination of increased embryonic lethality with a Him phenotype further suggests the occurrence of aneuploid eggs arising from meiotic missegregation of autosomes [63] . Increased larval arrest is indicative of somatic defects . Analysis of rmh-1 mutants revealed the cohort of phenotypes characteristic of a defect in meiotic recombination , as described above ( Fig 1G and 1H ) . Second , DAPI staining of oocytes at diakinesis , the last stage of meiotic prophase , revealed that a fraction of chromosome pairs was not connected by chiasmata in rmh-1 mutants . Whereas in the wild type , six bivalents ( chromosome pairs connected by chiasmata ) can be seen in diakinesis oocytes , rmh-1 mutant alleles showed an increase of DAPI-stained bodies , reflecting a mixture of bivalents and univalents ( achiasmate chromosomes ) ( Fig 1B–1E ) . In the -1 oocyte ( the most mature oocyte , next to the spermatheca , the oocyte prior to fertilization ) , we observed on average 8 . 9 ± 1 . 0 DAPI bodies in rmh-1 ( jf92 ) and 9 . 3 ± 1 . 4 in rmh-1 ( jf54 ) ( Fig 1F ) . As the jf54 allele behaved identically to the jf92 insertion allele with regards to the hatch rate and Him and diakinesis phenotypes , and as we were unable to detect GFP::RMH-1 containing the jf54 mutation ( S2B Fig ) , we infer that rmh-1 ( jf54 ) behaves like a null allele . Interestingly , fewer univalents were observed in rmh-1 ( tn309 ) ( 7 . 4 ± 1 . 1 DAPI bodies ) , suggesting that RMH-1 function is differently affected by this allele ( Fig 1F ) . To confirm the presence of univalents in rmh-1 mutants ( and not fragments due to absence of DNA repair ) , we measured the average volumes of DAPI bodies in diakinesis ( S3A and S3B Fig and S1 Text ) and confirmed both the presence of univalents and the absence of DNA fragments . In contrast to the rmh-1 mutants , the rmh-2 ( jf94 ) single mutant did not exhibit any meiotic phenotypes . However , all rmh mutants showed increased levels of larval arrest ( around 12% for rmh-1 mutants and 7% for rmh-2 ) ( Fig 1I ) . Furthermore , the double mutant rmh-1 ( jf54 ) ; rmh-2 ( jf94 ) was inviable , suggesting a redundant function of the two genes in somatic tissue . To better understand which meiotic processes are impaired in rmh-1 ( jf54 ) , we analyzed chromosome pairing , synapsis , and induction and repair of DSBs . No defects in pairing and synapsis were observed ( S4 Fig ) . To study the induction and processing of DSBs , we assessed the appearance and disappearance of foci of the strand exchange protein RAD-51 ( Fig 2 ) [64 , 65] . RAD-51 foci appeared with normal timing in the rmh-1 ( jf54 ) mutant , but accumulated at higher levels than in wild type and persisted at elevated levels through late pachytene . However , RAD-51 foci disappeared in diplotene , suggesting that repair of DSBs was delayed but did eventually occur . We also found that rmh-1-specific chromosomal defects depended on the induction of meiotic DSBs . In summary , meiotic recombination is aberrant in rmh-1 mutants . To assess RMH-1 localization in the gonad , we expressed GFP::RMH-1 using its endogenous regulatory elements as a functional fusion protein . Indeed , GFP::RMH-1 is capable of rescuing embryonic lethality and the Him phenotype when expressed in the null allele jf92 ( S5A and S5B Fig ) . GFP::RMH-1 was found in foci throughout pachytene ( Fig 3A–3C ) . In early pachytene , RMH-1 was first detected as a nucleoplasmic haze , and faint foci progressively appeared on chromosomes . During mid pachytene , RMH-1 foci increased in number and intensity . Mid pachytene nuclei contained , on average , 15 . 2 foci ± 3 . 6 , ranging from ten to 25 , and , in late pachytene , the number of RMH-1 foci drastically reduced to 5 . 9 ± 1 . 7 ( Fig 3D ) . In late pachytene , the number of foci ranged from one to nine foci; however , the majority of nuclei contained six or seven foci ( Fig 3E ) . ( This distribution is broader than the distribution of COSA-1 foci , visible as six signals in late pachytene . However , the mutant phenotype of cosa-1 already suggests that it acts before the protein can be observed in the six prominent foci . In addition , we believe RMH-1 marks a stage of recombination , through which all mature CO intermediates have to pass ) . Importantly , RMH-1 foci seem to increase in signal intensity ( about 1 . 8-fold ) between mid and late pachytene . This suggested that the protein might accumulate over time . RMH-1 localization in foci in pachytene is reminiscent of localization of markers of ongoing recombination events such as RAD-51 and CO-promoting factors MSH-5 , ZHP-3 , or COSA-1 [12 , 64–66] . Moreover , RMH-1 localization depends on meiotic DSBs , since foci were absent in a spo-11 mutant ( S2A Fig ) . Co-immunostaining experiments indicated that RAD-51 foci appear and disappear earlier than RMH-1 ( S6 Fig ) . RAD-51 foci appear in transition zone , peak in mid pachytene , and disappear in late pachytene . GFP::RMH-1 foci appear in early pachytene , peak in mid pachytene , and disappear in diplotene . Moreover , when RAD-51 and RMH-1 are present in the same nucleus , they do not colocalize ( only 4 . 0 +/- 4 . 0% RMH-1 foci coincide with RAD-51 ) ( Fig 3F and 3F′ ) . This suggests that RAD-51 and RMH-1 mark different recombination intermediates and that RMH-1 may act after RAD-51 removal . In contrast with the lack of colocalization with RAD-51 foci , the six bright RMH-1 foci in late pachytene nuclei do coincide with the localization of pro-CO factors MSH-5 , ZHP-3 , and COSA-1 at the CO sites ( 90 . 1 +/- 10 . 7% RMH-1 foci colocalize with COSA-1 and MSH-5 ) ( Fig 3G , 3J and 3K ) [12 , 16 , 66] . Interestingly , RMH-1 foci disappear before COSA-1 foci , which persist until diplotene ( S6 Fig ) . RMH-1 foci also partially colocalize with the earlier MSH-5 foci present during mid pachytene; the brighter foci of each protein tend to colocalize , while the fainter ones do not ( Fig 3G–3I ) . Furthermore , we found that late-pachytene RMH-1 foci are severely reduced ( to one focus on average ) in msh-5 , zhp-3 , and cosa-1 mutants , which fail to form CO-designated recombination intermediates . RMH-1 foci in mid pachytene are also less abundant and smaller in these mutants ( Fig 4A–4D , S7D and S7E Fig ) , suggesting that msh-5 , zhp-3 , and cosa-1 also contribute to stable localization of RMH-1 during earlier stages . The dynamic localization of RMH-1 through pachytene suggests that it may function at both CO and NCO repair events . The mid pachytene number of RMH-1 foci , at least twice the number of eventual COs , is consistent with RMH-1 localizing both at CO sites and at sites that will ultimately become NCOs . This is consistent with our analysis of GFP::RMH-1 localization following induction of excessive DNA breaks by ionizing radiation ( IR , 50Gy ) . We found that at 4 h post IR treatment , the number of RMH-1 foci in mid pachytene nuclei had risen significantly , suggesting that RMH-1 was loaded on many of the excess recombination intermediates produced by IR . To understand what would happen to those excess recombination intermediates positive for RMH-1 , we analyzed gonads 8 h post IR treatment ( after this time , the nuclei with the increased number of RMH-1 foci should have reached late pachytene ) . However , at this stage , the numbers were reduced to wild type numbers of foci in late pachytene nuclei ( S7A–S7C Fig ) . Thus , we conclude that RMH-1 marks recombination intermediates that can be increased transiently by irradiation but will be repaired as NCOs , as the final number of marked CO sites is not affected . The robust accumulation of RMH-1 at CO sites , the dependence of its localization on pro-CO factors , and the presence of univalents in the rmh-1 mutants together suggest that RMH-1 likely functions in promoting the CO outcome of initiated recombination events . Several lines of evidence provide further support for pro-CO role ( s ) for RMH-1 . First , we tested whether rmh-1 is required for normal localization of pro-CO factors by assessing the localization of GFP::COSA-1 in rmh-1 mutants . This analysis revealed that only 75% of nuclei in tn309 and 57% of nuclei in jf54 had six COSA-1 foci ( compared to 88% in wild-type ) ( Fig 4E and 4F ) , indicating a deficit of CO-designated sites in rmh-1 mutants . Furthermore , we observed several indications of a delay in pachytene progression in both rmh-1 mutants . First , we assessed localization of ZHP-3 , which normally localizes along the full length of the synaptonemal complex ( SC ) during mid pachytene , then retracts during late pachytene to a short stretch and , finally , to a single focus at the CO site in diplotene [66]; we found that ZHP-3 retraction was delayed in rmh-1 ( jf54 ) ( S8A–S8D Fig ) . Second , we evaluated a readout of a checkpoint-like feedback mechanism that operates during C . elegans meiosis to coordinate meiotic progression with generation of CO-eligible recombination intermediates . A single chromosome pair that lacks a CO-eligible recombination intermediate can be detected and elicit a prophase progression delay , visible by the persistence of markers such as phospho-SUN-1 S8 on the nuclear envelope [67] . SUN-1 S8Pi staining was prolonged in rmh-1 mutants , indicating a delay in prophase progression ( S8E–S8H Fig ) , which is in line with delayed and/or impaired formation of CO-specific recombination intermediates . These data in combination with the observed reduction in COSA-1-marked CO-designated sites together indicate a role for RMH-1 in ensuring efficient CO designation . Although the number of CO-designated sites is reduced in rmh-1 mutants , CO designation still occurs to a large extent , as 57% to 75% of nuclei contain six COSA-1 foci . Moreover , the reduction in number of COSA-1-marked CO sites is not sufficient to account for the number of univalents observed . To calculate the expected number of DAPI bodies in diakinesis nuclei based on our quantification of COSA-1 foci in rmh-1 mutants , we assumed that pachytene nuclei containing six COSA-1 foci would lead to diakinesis nuclei with six bivalents , pachytene nuclei containing five COSA-1 foci would lead to diakinesis nuclei with seven DAPI bodies ( five bivalents and two univalents ) , and so on . This approach yields an expectation of 6 . 7 DAPI bodies at diakinesis for rmh-1 ( jf54 ) and 6 . 6 for rmh-1 ( tn309 ) , values that are significantly lower than those observed in the -1 oocytes of these mutants ( Fig 1B and 1F ) . Thus , the number of univalents present at the end of meiotic prophase is higher than would be expected if all COSA-1-marked sites had successfully matured into COs . To understand this excess of univalents , we investigated the organization of chromosomes in diakinesis . In wild type , the presence of a CO precursor triggers domain restructuring of the ensuing bivalent such that the axial proteins HTP-1/2 and LAB-1 become concentrated along the long arm of the bivalent , and the synapsis protein SYP-1 and the Aurora B kinase ( AIR-2 ) are found on the short arm [68–70] . In spo-11 or msh-5 mutants , short and long arm markers overlapped on the univalents . In rmh-1 mutants , however , we observed long and short arm markers localized to reciprocal domains on univalents ( Fig 4G ) , suggesting that a CO site had been designated but failed to mature to actually link the homologs . Furthermore , quantification of DAPI bodies during progression through the diakinesis stage ( -3 , -2 , and -1 oocyte ) showed that they increased significantly ( Fig 4H–4J ) . We observed that in the -3 oocyte , the number of DAPI bodies in rmh-1 mutants ( 6 . 9 ± 0 . 7 DAPI bodies for jf54 and 6 . 8 ± 0 . 9 for tn309 ) was significantly increased compared to wild type . Interestingly , the observed number of DAPI bodies in -3 oocytes of the mutants corresponds to the calculated number of DAPI bodies based on COSA-1 foci analysis . However , in both mutants , the number of DAPI bodies strongly increased during oocyte maturation . In rmh-1 ( jf54 ) , we observed 9 . 3 ± 1 . 4 DAPI bodies in the -1 oocytes , while the effect was milder but still significant in rmh-1 ( tn309 ) ( 7 . 4 ± 1 . 1 in -1 oocytes ) . We conclude that , in rmh-1 mutants , connections between homologs may sometimes be lost after CO designation . Dissociation of bivalents that appeared to have undergone CO designation suggested that RMH-1 might function to enforce CO bias in the resolution of recombination intermediates at CO sites . This hypothesis prompted us to examine double mutants carrying rmh-1 ( jf54 ) in combination with mutations in xpf-1 , mus-81 , or slx-1 , which encode structure-specific endonucleases proposed to function in two separate and partially redundant meiotic resolution pathways [22–24] , and him-18 ( Slx4 ortholog ) , which encodes a nuclease scaffolding protein [71] . Dissociation of bivalents in rmh-1 ( jf54 ) was reduced by loss of xpf-1 , mus-81 , or slx-1 ( Fig 4K ) , consistent with persistence of recombination intermediates; however , it is unclear whether such putative persistent intermediates are located at CO-designated sites and/or at other sites on the chromosomes . Furthermore , we observed an elevated incidence of unresolved bivalents linked by DNA bridges ( Fig 4M and 4N ) in these double mutants compared to the xpf-1 , mus-81 , or slx-1 single mutants . Together with previous work showing that such linkages observed in nuclease-defective mutants are spo-11-dependent [22] , these results suggest that loss of RMH-1 function may result in the accumulation of excess and/or aberrant recombination intermediates that require these structure-specific endonucleases for their resolution . We also observed the occurrence of DNA bridges in the double rmh-1 ( jf54 ) ; him-18 mutant , but , those linkages appeared with the same frequency as in the him-18 single mutant . Furthermore , loss of him-18 in rmh-1 ( jf54 ) did not appreciably suppress the dissociation of homologs ( Fig 4K ) , in contrast to the outcomes with the single-nuclease mutants . We can reconcile these seemingly paradoxical observations by postulating ( 1 ) that unscaffolded nucleases may be able to act on recombination intermediates produced in rmh-1 mutants to remove excess connections between homologs , and ( 2 ) that the presence of multiple unresolved intermediates connecting homologs ( in rmh-1; xpf-1 , rmh-1 mus-81 , or rmh-1 slx-1 mutants ) might obscure detection of discrete bridges . As RMI1 was first identified as a component of a complex containing the RecQ helicase BLM and the topoisomerase TOP3 , we used the yeast-two-hybrid assay to confirm interactions between RMH-1 and HIM-6 ( the C . elegans BLM ortholog ) and between RMH-1 and TOP-3 ( S9A Fig ) . Furthermore , as several phenotypes of the rmh-1 mutants resemble phenotypes seen in him-6 mutants ( S9B and S9C Fig ) [46 , 53] , we investigated the cytological and functional inter-relationships between RMH-1 and HIM-6 . To examine colocalization of RMH-1 and HIM-6 , we used the CRISPR method to tag the him-6 locus with a C-terminal human influenza hemagglutinin ( HA ) epitope . The tag did not compromise HIM-6 function , since neither an increase in embryonic inviability nor a Him phenotype was observed ( S5 Fig ) . HIM-6 is found in few foci in early pachytene . The number of foci increases in mid pachytene and decreases again in late pachytene , reminiscent of the RMH-1 pattern . In mid pachytene , RMH-1 and HIM-6 substantially colocalized , and , in late pachytene , RMH-1 strictly colocalized with HIM-6; however , additional HIM-6 foci were also present ( 96 . 6 +/- 4 . 1% RMH-1 foci colocalize with HIM-6 in mid pachytene and 96 . 3 +/- 6 . 7% in late pachytene; Fig 5A–5C ) . To investigate the interdependence of localization , we assayed localization of HIM-6::HA in rmh-1 mutants . In tn309 , HIM-6 was completely absent in most of the gonads we imaged . In the rest , HIM-6 was barely detectable , and when foci were seen , they were not at CO sites in late pachytene ( Fig 5D ) . In jf54 , HIM-6 foci were present in all gonads but seemed smaller and fainter in mid and late pachytene , suggesting that RMH-1 is required to stabilize and enrich HIM-6 in foci ( Fig 5E ) . Although HIM-6 is sometimes found in proximity to CO site markers , these do not exhibit robust colocalization in the absence of rmh-1 ( Fig 5E ) . Assessment of the localization of GFP::RMH-1 in the him-6 ( ok412 ) mutant revealed that him-6 is required for RMH-1 localization in mid pachytene but not in late pachytene ( Fig 5H–5J ) . This finding is consistent with the idea that RMH-1 may be present in genetically distinguishable complexes in mid and late pachytene stages . However , him-6 is required for normal numbers of RMH-1 foci even in late pachytene ( Fig 5K ) . An important difference between rmh-1 and him-6 mutants is their impact on CO designation . A previous study showed that in the null allele of him-6 , the abundance of COSA-1 foci was unaffected [46]; thus , the him-6 null mutant appears to be proficient for CO designation . On the other hand , both rmh-1 alleles impair CO designation , as evidenced by an elevated frequency of nuclei containing fewer than six COSA-1 foci ( Fig 5F ) . Furthermore , the fact that CO designation is unaffected in a him-6 null mutant but is impaired in an rmh-1; him-6 double mutant implies that RMH-1 can assist in promoting CO designation in the absence of HIM-6 . A deficit of bivalent connections is a shared feature of rmh-1 and him-6 mutants . The phenotype is evident but less pronounced in him-6 or rmh-1 ( tn309 ) ( p < 0 . 05 ) , while it is more prominent in rmh-1 ( jf54 ) ( p < 0 . 01 ) . The number of DAPI bodies in rmh-1 ( tn309 ) ; him-6 ( ok412 ) and the respective single mutants was not different . Interestingly , however , we found that the deficit of connections between homologs in rmh-1 ( jf54 ) ; him-6 ( ok412 ) oocytes was reduced compared to the rmh-1 ( jf54 ) single mutant ( Fig 5G ) , despite the fact that COSA-1 foci were reduced in this double mutant . One possible explanation for this observation would be that the dissociation of bivalents observed in rmh-1 ( jf54 ) might be mediated by HIM-6 . An alternative explanation is that the simultaneous loss of HIM-6 and RMH-1 may lead to persistence of unresolved recombination intermediates that maintain connections between the homologs . We investigated the foci structure of RMH-1 and HIM-6 foci using structured illumination microscopy ( SIM ) . Both in mid pachytene and late pachytene , RMH-1 and HIM-6 signals were resolved into more complex structures ( Fig 6 ) . In mid pachytene , we observed foci that clearly exhibit a doublet structure and foci that show an elongated shape ( Fig 6A and 6B ) . In late pachytene , RMH-1 and HIM-6 appear even more clearly as doublets ( Fig 6C–6E ) . In both pachytene stages , we hypothesize that RMH-1 and HIM-6 foci mark similar recombination intermediates . The doublet structure is likely easier to observe in late foci as RMH-1 protein accumulates during pachytene progression , leading to larger foci in late pachytene . We measured the average distance between the foci peak in the three-dimensional stacks as an average of 227 +/- 46 nm . Interestingly , in Drosophila , the size of a recombination nodule has been estimated to around 100 nm [72] . It has also been shown that 1 kb of B-form DNA is 340 nm in length . We conclude that the structure we describe here is on an appropriate scale for flanking dHJs . We also observed that the late pachytene HIM-6 doublet structures flank a COSA-1 focus in the wild type ( Fig 6F ) . Interestingly , in the rmh-1 ( jf54 ) mutant , HIM-6 appears as single foci and not as elongated or doublet structures in late pachytene ( Fig 6G ) . This suggests that rmh-1 is required to concentrate HIM-6 at CO sites and also to organize a complex structure surrounding late recombination intermediates . The mid pachytene localization of RMH-1 ( and HIM-6 ) to numerous foci in excess of the eventual number of COs , together with the well-known role for the RTR complex in discouraging COs in other systems ( see Introduction ) , led us to investigate the possibility that RMH-1 also functions at NCO sites during C . elegans meiosis . We focused on the possibility that RMH-1 might function in parallel with the SMC-5/6 complex , based on several considerations . First , several studies in budding yeast showed that in the absence of Smc5 or Smc6 during meiosis , COs form but joint molecules accumulate , leading to chromosome segregation defects [57–59] . Moreover , it was shown that Smc6 was required for the resolution of the joint molecules accumulating in absence of Sgs1 , suggesting collaboration of pathways mediated through those two proteins [59] . Second , a prior study showed that the C . elegans SMC-5/6 complex is important for meiotic DSB repair under conditions in which interhomolog CO formation is abrogated and/or in mutant situations when intersister repair is the only option [73] . Moreover , the data of Bickel et al . are fully consistent with the SMC-5/6 complex participating in interhomolog NCO repair as well . Several lines of evidence support the hypothesis that RMH-1 functions in parallel with the SMC-5/6 complex to antagonize accumulation of aberrant interhomolog connections . First , we observed a significant increase in RMH-1 foci in mid pachytene nuclei in the smc-5 mutant; 29 . 1% of nuclei contained more than 25 foci , never seen in wild type ( Fig 7A and 7B ) . However , late pachytene nuclei in the smc-5 mutant had an average of six foci per nucleus , as in wild type ( Fig 7C ) , suggesting that an excess of RMH-1-associated recombination intermediates accumulate in the smc-5 mutant , but they are ultimately taken care of in agreement with the wild type viability of the single mutant ( Fig 7D ) . Second , the rmh-1 ( jf54 ) ; smc-5 ( ok2421 ) double mutant showed both a synthetic decrease in embryonic hatch rates and increase in larval arrest ( Fig 7D and 7E ) , indicating a strong genetic interaction between rmh-1 and smc-5 in the soma but also likely during meiosis ( a similar genetic interaction is reported between him-6 and smc-5 in [74] ) . Third , quantification of the number of DAPI bodies in diakinesis oocytes in the rmh-1; smc-5 double mutant revealed an average of 5 . 9 ± 0 . 3 DAPI bodies in the -1 oocyte ( Fig 7F ) , consistent with the presence of connections between the homologs . However , cytological examination of chromosome organization revealed that these interhomolog connections were aberrant . Consistent with the low viability of this double mutant , bivalent organization was dramatically defective: both long arms ( indicated by LAB-1 staining ) were always found together , precluding the wild-type cross-shaped appearance of the bivalent ( Fig 7G–7J ) . Such abnormal bivalents were already observed at a low frequency in each of the single mutants , but are now present in each diakinesis nucleus in the double mutant ( Fig 7K ) . This result strongly suggests that rmh-1 and smc-5 cooperate to eliminate and/or prevent accumulation of connections between homologs at NCO sites . Finally , we found that simultaneous loss of both RMH-1 and SMC-5 results in a high frequency of interhomolog connections even in the absence of the conserved meiotic CO factor ZHP-3 . Whereas COs and chiasmata are eliminated in the zhp-3 single mutant [16] , a small but significant reduction in DAPI bodies in -1 oocytes was observed in the zhp-3; smc-5 mutant ( Fig 7L , 7M and 7O ) . This ability to create aberrant homolog connections in a zhp-3 mutant is consistent with the finding that a mutation affecting the budding yeast Smc5/6 complex has the ability to create connections between homologous chromosomes in a mutant lacking Zip3 ( the ZHP-3 ortholog ) [59] . Moreover , ectopic ZHP-3-independent connections occurred at very high frequency in a zhp-3; rmh-1; smc-5 triple mutant , in which we observed an average of 7 . 3 DAPI bodies at diakinesis ( Fig 7N and 7O ) . Together , our data indicate that smc-5 and rmh-1 act in parallel to antagonize MutS gamma-independent inter-homolog connections , supporting the conclusion that RMH-1 also functions at NCO sites . To assess the role of RMH-1 at NCO sites in more detail , we analyzed the localization of COs along the chromosomes . We used deep sequencing to compare CO frequency and CO distribution between the wild type and rmh-1 mutants ( jf54 and tn309 ) . We took advantage of the C . elegans Hawaiian ( Hw ) strain , which contains a high frequency of SNPs compared to the Bristol strain ( wild type ) . Thus , we generated strains containing rmh-1 mutants in combination with a subset of introgressed “Hawaii chromosomes”: chromosomes II and V for rmh-1 ( jf54 ) and chromosomes X , IV , and V for rmh-1 ( tn309 ) . ( Introgression of all Hawaii chromosomes into the rmh-1 mutants led to sterility ) . We sequenced all the parental strains ( Bristol , Hw , and the “Hw introgressed rmh-1 mutants” ) and used them as references . Recombinants between the Bristol strain and Hw were generated as described in Fig 8A . We singled F2 animals and allowed them to self-fertilize for three to four generations and deep-sequenced DNA isolated from their progeny . Paired-end reads were mapped as described in the Experimental Procedures section . Bioinformatic analysis identified 166 , 928 homozygous unique SNPs in the Hw genome . To assess the chromosome composition in each recombinant , we used the homozygous unique SNPs of the Hw strain . We used SNPs located at least 400 bp apart to avoid having two or more SNPs per paired read . In the absence of recombination , two genotypes are expected in the offspring: fully homozygous ( i . e . , both homologous chromosomes come from the same parental strain ) and fully heterozygous ( each homolog comes from one parental strain ) . If recombination occurred , blocks of homozygosity and blocks of heterozygosity were detected . We identified the recombination break-points by analyzing the changes in heterozygosity along the chromosome ( for more details , see the Experimental Procedures section ) . We did not detect a significant difference between the mutants and wild-type control in the total frequencies of crossovers ( Fig 8B ) . This observation contrasts with the reduction in COSA-1 foci observed in both mutants , and could potentially reflect sample size or the fact that viable progeny were used in the recombination assay; alternatively , it may reflect crossovers occurring at sites not marked by COSA-1 foci in the mutants . However , we found that the positions of the COs were shifted significantly toward the center of chromosomes in both mutants , in contrast to the preference observed in the wild type , in which the majority of COs occurred in the gene-sparse “arm” regions of the chromosomes ( Fig 8C and 8D ) . We conclude that rmh-1 is required to discourage COs in the center of chromosomes . A shift of COs toward chromosome centers in the rmh-1 ( jf54 ) mutant was confirmed by using a PCR-based assay to genotype four SNPs along chromosome IV ( S10 Fig ) . A parallel study analyzed the recombination rate and CO position in the him-6 mutant . Here , the entire brood was used ( dead and living progeny ) to determine recombination; similarly , a shift toward the chromosome centers was observed [74] .
Although protein complexes involving RMI1 and/or BLM helicase were initially recognized based on anti-CO activities , recent work in several systems has led to a growing realization that BLM and RMI1 play multiple roles during meiotic recombination , functioning both in antagonizing and promoting COs ( for review , see [75] and Introduction ) . One current view for Saccharomyces cerevisiae is that the primary role of the RTR complex is as a “recombination intermediate chaperone” [41 , 44 , 45 , 55 , 56] . Under this model , RTR plays an indirect role in promoting COs during yeast meiosis , either by generating recombination intermediates recognized and processed by the Exo1-MutLγ pathway or by disassembling aberrant recombination intermediates , thereby helping to channel early recombination intermediates into either a MutSγ-dependent CO pathway or into the SDSA pathway for generating NCOs . Yeast Sgs1 has also been implicated in promoting resolution of CO intermediates under circumstances in which JM resolution is impaired [55] . A more prominent pro-CO role has been uncovered for HIM-6/BLM during C . elegans meiosis , in which loss of BLM function results in reduced COs [53 , 54] . It was proposed that HIM-6 functions to enforce a CO-biased outcome of resolution of CO intermediates; however , it was not clear whether this proposed pro-CO role for HIM-6 was direct or indirect [22 , 46] . Our combined genetic and cytological analysis provides a strong case that RMH-1 functions directly at both CO and NCO sites . For a summary , see Fig 9 . First , RMH-1 and HIM-6 colocalize at CO designated sites ( Figs 3G–3K , 5A–5C and 6F ) , and this late pachytene localization is dependent on conserved pro-CO factors ( Fig 4A–4D ) . Conversely , pro-CO factors can concentrate at presumptive CO sites in the absence of HIM-6 or RMH-1 ( albeit less reliably in rmh-1 mutants ) consistent with RMH-1 and HIM-6 being recruited to CO designated sites to perform a late function in CO formation ( Figs 4E , 4F and 5F ) . Moreover , SIM imaging of CO sites in late pachytene nuclei revealed that RMH-1 and HIM-6 colocalize in an elongated structure or closely juxtaposed doublets , potentially reflecting two separate populations of RTR complexes flanking the two junctions of a dHJ CO intermediate ( Fig 6 ) . The “doublet” organization leads us to suggest that RMH-1 and HIM-6 localized at CO sites may support the CO outcome of recombination intermediate resolution either by promoting an optimal geometry for endonucleases ( resolvases ) to position DNA cleavage , or by protecting the structure from unscheduled action of endonucleases . Our analysis with the truncation allele tn309 raises the possibility that HIM-6 itself may contribute to inappropriate resolution when RMH-1 is absent . Interestingly , RMH-1 and HIM-6 differ regarding the requirements for their recruitment at CO sites . Whereas RMH-1 is successfully recruited to most presumptive CO sites in him-6 mutants ( Fig 5H–5K ) , HIM-6 does not stably concentrate at the sites of pro-CO factors in rmh-1 mutants ( Fig 6G ) , raising the possibility that RMH-1 ( or TOP-3 associated with RMH-1 ) may interact more directly with pro-CO factors at such sites . The absence of HIM-6 at CO sites in rmh-1 mutants suggests that RMH-1 might provide a structural support to ensure the final processing of a mature recombination intermediate by limiting branch migration or the collapse of a putative dHJ and , thereby , influencing the outcome of resolution . We provide multiple lines of evidence that RMH-1 also localizes to NCO sites and that its function ( s ) in NCO repair events are distinct from its pro-CO role ( s ) . First , localization of RMH-1 during early and mid pachytene has different genetic requirements than its late pachytene localization at CO-associated foci: Early/mid pachytene RMH-1 foci can still form ( albeit at reduced levels ) in mutants defective for pro-CO factors , but are strictly dependent on HIM-6 , suggesting that the intermediates present at early and late foci differ in structure ( Figs 4A–4D and 5H–5K ) . Furthermore , we show that such foci can occur at elevated levels when numbers of DSB are increased or when alternative pathways for DSB repair are abrogated ( Fig 7A and S7A–S7C Fig ) . In mid pachytene , HIM-6 and RMH-1 proteins colocalize , but only a fraction colocalize with MSH-5 ( Fig 3G–3I ) . We hypothesize that foci positive for all three proteins correspond to CO-eligible recombination sites , while those positive for only HIM-6 and RMH-1 may be processed as NCOs ( perhaps by a decatenation reaction ) . The occurrence of chromosome bridges in rmh-1; resolvase double mutants could indicate either the accumulation of recombination intermediates due to defective dissolution or a more direct collaboration between RMH-1 and resolvases to produce COs and/or NCOs ( Fig 4K–4M ) . Together , our data suggest direct but distinct roles for RMH-1 in both CO and NCO repair during C . elegans meiosis . Another important concept emphasized by our work analyzing the functions of RMH-1 during C . elegans meiosis is that a successful outcome of meiosis requires proper execution of the prescribed recombination events at both CO and NCO repair sites . Whereas COs serve as the basis of chiasmata that connect homologous chromosomes to enable them to orient toward opposite poles of the meiosis I spindle , efficient homolog segregation also depends on elimination of other recombination-based interactions between homologs that might impede their timely separation . Beyond ensuring CO formation , RMH-1 has ( at least ) two roles in preparing homologous chromosome for segregation at meiosis I . First , RMH-1 functions in parallel with SMC-5 in eliminating excess associations between homologs ( Fig 7 ) . Importantly , we also demonstrate another major role for RMH-1 in ensuring that COs usually form in positions that favor reliable homolog segregation . In the absence of localized centromeres ( due to the holocentric nature of C . elegans chromosomes ) , the position of the CO dictates multiple features of bivalent organization that are crucial for correct alignment and behavior of chromosomes [69] . This defines distinct long arm and short arm domains where cohesion will be retained or released at the meiosis I division and delineates how kinetochore components , motor proteins , and cell-division protein kinases associate with the chromosomes . We found that , whereas COs usually occur at off-center positions during wild-type meiosis , COs occurred more frequently near the centers of the chromosomes in rmh-1 mutants ( Fig 8 and S10 Fig ) . We speculate that an altered distribution of COs in an rmh-1 mutant may reflect impaired execution at both CO and NCO repair sites . On the one hand , an impaired ability to impose CO-biased resolution at CO-designated sites results in intermediates at some of those sites being resolved as NCOs . On the other hand , impairment of a mechanism that would normally help dissociate intermediates to promote repair via SDSA may result in accumulation of DNA structures that require structure-specific endonucleases for their resolution , resulting in CO repair products at some sites that should have been NCOs . We have not observed a significant frequency of double COs in the mutants , suggesting that the misplaced COs are still subjected to interference . Interestingly , the SLX-1 resolvase is also required to suppress CO formation in the chromosome center [24] . It is appealing to speculate that RMH-1 might cooperate with SLX-1 in a locally defined region to promote NCO resolution . We note that while CO positioning may be especially important in organisms like C . elegans that use COs to trigger differentiation of the bivalent , CO position is also very important during Drosophila , yeast , and human meiosis . in Drosophila , centromere proximal recombination events have been correlated to metaphase II nondisjunction [76] . In yeast , centromere proximal COs have been shown to trigger premature separation of sister chromatids ( PSSC ) . Interestingly , sgs1 mutants exhibit reduced spore viability , potentially due to an excess of COs near the centromere leading to PSSC [77] . In human meiosis , COs located very near to either telomeres or centromeres are associated with elevated risk of aneuploidies , e . g . , [78–82] .
Dissection and immunostaining was performed as described in [67] with modifications when using SYP-1 , LAB-1 , and S10 phospho-histone 3: dissection and fixation was performed in 1x EBT instead of PBS 1X [66] . Adults 24 h post L4 were dissected . Images are projections of stacks encompassing whole nuclei . For details on antibodies and microscopy , refer to S1 Text . See also S1 Text for rmh-1 allele isolation and characterization , protein tagging ( RMH-1 , HIM-6::HA ) , generation of the rmh-2 knockout , irradiation conditions , yeast two hybrid assays , photoconversion experiments , and recombination assays . | During meiosis , faithful separation of chromosomes into gametes is essential for fertility and healthy progeny . During the first meiotic division , crossovers ( CO ) between parental homologs ensure their correct segregation . Programmed DNA double strand breaks ( DSBs ) and resection steps generate single-stranded overhangs that invade a sister chromatid of the homolog to initiate homologous recombination . This culminates in the generation of a DNA double Holliday junction ( dHJ ) . This can be acted upon by resolvases to produce CO and non-crossover ( NCO ) products , depending on where the resolvases cut the DNA . Alternatively , NCOs can also be produced by decatenation via the RecQ helicase–topoisomeraseIII–Rmi1 ( RTR ) complex . The mammalian RTR contains a topoisomerase , Bloom’s helicase , and RMI1/2 scaffolding components . It disassembles dHJs in vitro and contributes the major NCO activity in mitosis . Here , we provide evidence that the Caenorhabditis elegans RMH-1 functions in distinct complexes during meiosis to produce both COs and NCOs in an in vivo animal model system . Strikingly , RMH-1 spatially regulates the distribution of COs on chromosomes , demonstrating that the RTR complex can act locally within specific chromosome domains . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Experimental",
"Procedures"
] | [
"invertebrates",
"meiosis",
"medicine",
"and",
"health",
"sciences",
"reproductive",
"system",
"gonads",
"caenorhabditis",
"cell",
"cycle",
"and",
"cell",
"division",
"cell",
"processes",
"animals",
"animal",
"models",
"germ",
"cells",
"caenorhabditis",
"elegans",
"oo... | 2016 | Separable Roles for a Caenorhabditis elegans RMI1 Homolog in Promoting and Antagonizing Meiotic Crossovers Ensure Faithful Chromosome Inheritance |
Current genome-wide association studies ( GWAS ) have high power to detect intermediate frequency SNPs making modest contributions to complex disease , but they are underpowered to detect rare alleles of large effect ( RALE ) . This has led to speculation that the bulk of variation for most complex diseases is due to RALE . One concern with existing models of RALE is that they do not make explicit assumptions about the evolution of a phenotype and its molecular basis . Rather , much of the existing literature relies on arbitrary mapping of phenotypes onto genotypes obtained either from standard population-genetic simulation tools or from non-genetic models . We introduce a novel simulation of a 100-kilobase gene region , based on the standard definition of a gene , in which mutations are unconditionally deleterious , are continuously arising , have partially recessive and non-complementing effects on phenotype ( analogous to what is widely observed for most Mendelian disorders ) , and are interspersed with neutral markers that can be genotyped . Genes evolving according to this model exhibit a characteristic GWAS signature consisting of an excess of marginally significant markers . Existing tests for an excess burden of rare alleles in cases have low power while a simple new statistic has high power to identify disease genes evolving under our model . The structure of linkage disequilibrium between causative mutations and significantly associated markers under our model differs fundamentally from that seen when rare causative markers are assumed to be neutral . Rather than tagging single haplotypes bearing a large number of rare causative alleles , we find that significant SNPs in a GWAS tend to tag single causative mutations of small effect relative to other mutations in the same gene . Our results emphasize the importance of evaluating the power to detect associations under models that are genetically and evolutionarily motivated .
Genome-wide association studies ( GWAS ) genotype upwards of 500 , 000 common SNPs and test for allele frequency differences in case/control panels consisting of several thousand individuals . Such studies have identified highly significant and replicable associations , and as a result have uncovered entirely new pathways contributing to complex disease risk ( http://www . genome . gov/gwastudies/ ) . However , these associations explain only a small fraction of the known heritability of risk for the diseases examined [1] . It is well-known that the GWAS paradigm of testing for associations primarily using intermediate frequency markers has high power to identify an association only if disease causing alleles are also at intermediate frequency [2] . This “missing heritability” has led to speculation that a new round of GWAS should be designed to detect rarer variants of presumably larger effect . The potential importance of rare alleles of large effect ( RALE ) is supported empirically by studies that have carried out deep resequencing of candidate gene exons and observed an excess of rare radical amino acid polymorphisms in cases relative to controls for a variety of diseases ( HDL cholesterol levels [3] , susceptibility to colorectal adenomas [4] , LDL cholesterol levels [5] , triglyceride levels [6] , folate metabolism [7] , and hypertriglyceridemia susceptibility [8] . Collectively , these studies suggest the possibility that the same sort of genetic heterogeneity commonly observed for Mendelian disorders [9] , [10] ) may characterize complex disease . A weakness of the RALE model for complex disease variation is that it is not a population-genetic model , but rather an easy to understand verbal model . As a result it does not make quantitative predictions concerning the nature of genetic variation at the genes underlying complex disease , neither in terms of the number of causative alleles , their frequencies and effects , nor in terms of the patterns of linkage disequilibrium ( LD ) between causative alleles and linked neutral markers . Ideally , the predictions of various RALE models would come from explicit population-genetic models of disease , with concrete assumptions about the fitness effects of causative mutations and the relationship between phenotype and fitness determining the frequency of causative mutations in the population . To date , Prichard's [11] work is the best attempt to model the impact of the equilibrium between mutation and selection on the frequencies of disease-risk mutations . His model generates scenarios where the genetic basis of a complex disease consists of many rare mutations at different sites within a gene , with the frequency of causative mutations being determined by the balance of mutation and natural selection [12] . His model is a case for which the power of standard association tests is greatly reduced [2] , [13] , [14] . Pritchard's [11] work did not model intragenic recombination , nor track neutral mutations partially-linked to causative deleterious mutations . Thus , Pritchard was unable to explicitly address the power of GWA studies , which genotype both causative and linked neutral SNP markers throughout the genome . This question instead requires the use of explicit simulations of an evolving phenotype and its molecular basis . In recent work , the application of population-genetic principles to understanding the properties of GWAS have been highly heterogeneous . For example , some authors have modeled the frequencies of risk alleles in a region as independent random variables , as opposed to simulating a recombining region [15]–17 . In these studies , the genotype/phenotype relationship is based on arbitrary choices as to the number of causative mutations [15]–[17] . A second set of studies have simulated recombining regions using coalescent simulations without selection [e . g . , 18] to generate a large set of haplotypes from a model with explicit assumptions about demography ( these assumptions vary from study to study ) [19]–[21] . The authors then selected an arbitrary number of mutations from arbitrary frequency ranges to be causative mutations , and arbitrary effect sizes are assigned . Finally , some studies have used forward simulation machinery [e . g . , 22] to simulate multiple partially-linked deleterious mutations subject to natural selection in a region interspersed with neutral mutations [23]–[25] . In these forward simulations , fitness values were assigned to particular sites according to genotype , and the final fitness of a diploid is typically either the sum or product of fitness effects over deleterious mutations . The simulation output is then used to map genotype to phenotype using either an arbitrary model 24 , 25 or an explicit quantitative genetic model [23] . Although this last class of models represents the most sophisticated application of evolutionary simulations , they are still limited in that the phenotype itself is not the target of natural selection ( as is the case in [11] ) , and thus the simulated distributions of phenotypes are not the outcome of an evolutionary process ( even though the underlying mutation frequencies are ) . Here , we propose an explicit model of a quantitative trait subject to natural selection , with case/control status treated as a liability trait . Our model is similar to that of Pritchard [11] in that the frequencies of deleterious mutations are the result of the balance between mutation and natural selection [12] , and similar to other recent work [23] , [24] , [25] in employing explicit forward simulations . We depart from existing work using forward simulations in two important ways . First , the phenotype itself determines fitness and thus is the target of natural selection . Second , our model of gene action is based on the standard definition of a gene as a region in which recessive mutations fail to complement [26] , such that affected individuals will generally be a trans-heterozygote for causative mutations ( e . g . heterozygous for at least two different causative mutations at different positions in the gene region ) , as is commonly-observed for Mendelian disorders [10] , [27] . Our model of fitness is , therefore , based on the partial recessivity of haplotypes and not on the standard population-genetic assumptions of multiplicative or additive fitness across individual mutations . Under these standard assumptions ( used in simulation programs such as [22] ) , fully-recessive mutations at different positions complement one another in terms of fitness ( e . g . , the fitness of a trans-heterozygote is the same as a diploid that is homozygous for wild-type alleles at both sites ) and , therefore , the different deleterious mutations within a simulated region are , themselves , different genes ( sensu [26] ) . We develop a novel forward simulation and use it to simulate a “typical” 100 kilobase region of the human genome , tracking both causative deleterious and non-causative neutral mutations . In our simulations , disease risk is due to an underlying continuously varying liability score , with a causative disease “gene” contributing ∼5% to variation in that score . Given the size of the region considered , the idea that the vast majority of mutations are neutral , and the nature of gene action being modeled , our models are most consistent with mutations impacting both the structural product and the cis-regulatory regions controlling expression of a gene contributing to risk of a complex disease . We use our simulation machinery to explore the population-genetic signals of selection against causative sites , and to explore the power of GWAS to test the hypothesis that variation within a gene region contributes to complex disease . Our model results in gene regions evolving under the “allelic heterogeneity” model involving many non-complementing risk mutations segregating within a gene region . Since the 1990s , many human geneticists believed that this model was likely to explain complex variation [9] , [ pg 492; 10] . Under this model , complex traits are genetically analogous to Mendelian disease genes , but the mutations are simply less penetrant due to other genes and environmental variation impacting the trait [9 , Chapt . 14&15] . Our model results in weak selection against causal variants , with no detectable average effect on patterns of variation at linked neutral sites . A major finding is that statistical tests designed to detect an overall greater number of rare alleles in cases have very low power ( such tests have been argued to have high power to detect RALE ) , and less power than the standard single-marker logistic regression assuming an additive model . The frequencies of significant associations from single-marker tests applied to our simulated GWAS involving common markers are consistent with empirical observations from real GWA studies [28] , in contrast to previous results based on simulating RALE as neutral [19] , [28] . We propose a simple statistic based on the excess of marginally significant markers in a region , and find that it has higher power to detect associations than other tests considered , although the power of the SKAT package [21] , [29] , [30] can have comparable power . Finally , in our simulations , the explanation for missing heritability in current GWAS in that significantly associated common markers tend to be associated with causative mutations with relatively small effect sizes , but fail to tag rarer variants of larger effect . The observation that more subtle-effect variants can drift to higher allele frequencies is consistent with population-genetic predictions of an inverse relationship between average frequency and effect on fitness [e . g . , 23] , which we expect to be a general property of any model involving mutation-selection balance . This explanation for missing heritability differs from the hypothesis of “synthetic associations” arising when RALE are assumed to be neutral [19] . The simulated datasets represent an important resource for evaluating the power of novel test statistics under the heterogeneity model .
Under our gene-based model , the effect size of a causative mutation is exponentially-distributed with mean λ ( λ = 0 implies a mutation that does not contribute to a complex disease phenotype ) , and the effect of a maternal or paternal haplotype is additive over causative mutations . The phenotype of a diploid is the geometric mean effect of the maternal and paternal haplotype plus a random Gaussian environmental effect scaled so that the gene-region being modeled accounts for some fraction of the total disease burden . Figure 1a shows the difference between our model of gene action ( non-complementation of loss-of-function mutations ) and the standard population-genetic assumption that mutation effects are multiplicative . By setting the contribution of a gene to an individual's phenotype equal to the geometric mean of the maternal and paternal haplotypes , the haplotype closer to being mutation-free dominates the genotypic effect ( Figure 1b ) , resulting in partially recessive model of gene-action ( Figure S1a ) , which is empirically supported for mutations of moderate effect [31]–[34] . As in Pritchard's work [11] , the distribution of effect sizes at causative sites at equilibrium is not equivalent to the distribution of newly arising mutations . Rather , the frequency distribution of causative mutations at equilibrium is determined by a balance of mutation , genetic drift , and natural selection . This contrasts with other attempts to model the frequency distribution of causative mutations using an arbitrarily defined statistical distribution ( as in [17] ) or arbitrary numbers of causative mutations [19] , [21] , [29] . Our approach also differs from previous approaches in that the number of causative mutations in a region is a random variable , as opposed to a fixed an arbitrary quantity [15] , [16] , [19] , [21] , [29] , [35] . Insomuch as the assumptions of our model are correct , we are properly specifying the equilibrium distribution of the number , frequencies , and effects of causative mutations , as well as the extent of LD between causative and linked neutral sites . In our simulated populations , liabilities are close to normally distributed , except in the extreme “diseased” tail , where there is a slight excess of extremely affected individuals ( Figure 1c ) . Further , the fitness of an affected individual is generally high ( Figure S1b ) , meaning that even individuals with the most extreme liabilities are capable of approximately normal reproduction . Although there is considerable uncertainty surrounding the distribution of fitness in human populations , and the strength of purifying selection on complex diseases remains a subject of debate , our model is consistent with the idea that purifying selection on complex disease phenotypes is generally weak , as has been claimed in the literature [36] . Given the computational demands of forward simulation , we focus our attention on a set of parameters ( see Methods ) that results in the proportion of total phenotypic variation in the population attributable to the focal gene region reaching a plateau at ∼4% as λ increases to ∼0 . 075–0 . 10 ( Figure 1d ) . Although our simulations assume a uniform rate of crossing over per generation , heritability similarly plateaus at ∼4% for a region with zero recombination ( Figure S1c ) when using the same mutational parameters as in Figure 1d . Since the power to detect an association depends on the recombination rate between genotyped markers and causative mutations , we present results only for the case of no recombination ( representing extreme “cold” regions of recombination ) and for a region recombining at a uniform rate representing the genome average . Thus , while we are not explicitly modeling hotspots of recombination , we view the results as broadly-applicable on average . The value of heritability at the plateau depends on the model parameters . Plateau height is approximately linear as a function of the deleterious mutation rate ( Figure S2a ) , holding all other parameters the same as Figure 1c . Thus , holding the per-site neutral mutation rate constant , the heritability due to a gene region is a function of the proportion of sites mutable to disease alleles and the physical size of a gene . Further , we can “tune” the expected value of the heritability at the plateau by increasing or decreasing the value of ( the ratio of the variance in fitness and variance due to random effects ) in a manner broadly consistent with the house-of-cards model for the maintenance of quantitative genetic variation under mutation selection balance ( [37]; Figure S2b and S2c ) . Thus , in spite of the considerable uncertainty in the values of μd ( the deleterious mutation rate , or the product of proportion of sites mutable to a causative allele and the size of a gene for a constant per site mutation rate ) an λ ( the average effect size of newly arising exponentially distributed deleterious mutations ) , our model is able to generate a single gene of small to large effect contributing to disease risk for plausible parameters ( Figure S2a ) . It is additionally noteworthy that the stochastic variation around the expected gene-specific heritability ( conditional on μd and λ ) is quite large . This implies that even for parameter combinations that predict equilibrium heritability values of ∼2% ( e . g . Figure S2a ) , 2–9% of genes sharing these parameters will each account for >5% of the total phenotypic variation in a complex trait ( data not shown ) . Therefore , despite our focus on parameter values that result in a heritability plateau of ∼4% , parameter values predicting lower plateaus are clearly of interest . We examined the frequencies of mutations in a sample of 100 diploids drawn from each of the simulated regions . On average , causative mutations are more rare than expected in the absence of natural selection ( Figure S3a–S3d ) . The strength of the skew towards rare alleles is stronger with increasing λ , consistent with the observation that at the heritability plateau , larger λ are associated with variation being due to fewer ( Figure S3e ) , more rare ( Figure S3f ) , and larger-effect mutations compared to small λ . In contrast , the site frequency spectrum at non-causative variants is indistinguishable from neutrality irrespective of λ . As the vast majority of polymorphisms are non-causative under our model ( Figure S3f ) , it is unlikely that population-genetic methods would identify these gene regions as abnormal , despite their strong contribution to disease . Thus , under our model there is only a very slight excess of rare alleles observed in case versus controls ( Figure S3e ) , and this is not likely a fruitful signal to look for . It is important to note that this result may not extend to “exon sequencing” where there may indeed be an excess of rare non-synonymous ( NS ) SNPs in cases relative to controls . If disease is primarily due to rare NS mutations in genes making strong contributions to total risk , the fraction of NS SNPs mutable to causative alleles is likely much greater than the same fraction for a gene region in general ( including non-coding regulatory regions ) as modeled here . Figure 2 shows representative Manhattan plots ( based on a marker-by-marker logistic regression of 3000 case/control samples ) for 100 kb gene regions and different λ ( λ = 0 is a no deleterious mutation control simulation ) . Since we track every SNP in a region , we can distinguish neutral from causative markers as well as common from rare markers . For both large and small λ , it is difficult for markers from either SNP-chips or complete resequencing studies to reach a genome-wide statistical significance threshold ( Figure 2a , 2b , and 2f; we assume SNP-chips type exclusively SNPs with a MAF ≥5% and all such SNPs are genotyped ) . For intermediate λ it is possible for markers to be significant at a genome-wide threshold , and rare markers are more likely to approach significance than intermediate-frequency markers ( Figure 2c–2e ) . For intermediate λ , only a small percentage of rare sites are causative ( <10% on average for all λ ) , yet causative sites are much more likely to be among those markers reaching significance than non-causative sites ( 17 . 6% to 31 . 3% of significant rare markers are causative; Figure S4a ) . Thus , unlike current GWAS using intermediate frequency markers , under our gene-based model significant associations between rare markers and disease status are likely to identify truly causative mutations . We also observe examples of significant , common , non-causative markers ( e . g . Figure 2c ) , consistent with current GWAS hits occasionally uncovering genes evolving under our gene-based model . In general , significant common markers are only in strong linkage disequilibrium with a single causative SNP ( Figure S4b ) ; similarly , when several common non-causative markers are significant in a single region , they tend to “tag” different causative SNPs ( Figure S4c ) , which themselves tend to be surprisingly common with small effect sizes ( Figure S4d ) . Thus , an individual significant common neutral marker is typically associated with a single causative site of weak effect that has drifted to an intermediate frequency ( i . e . , an evolutionary outlier ) . This observed relationship between common SNPs significant in a GWAS and causative alleles is inconsistent with the claim of recently published work that introduced the idea of a synthetic association resulting from a common marker tagging a haplotype ( s ) harboring an excess of rare causative mutations [19] , [38] . It is likely this difference stems from the fact that the synthetic association papers assume causative alleles are neutrally evolving , yet narrowly constrained in allele frequencies , whereas here we assume causative alleles are deleterious with equilibrium frequencies and patterns LD ultimately determined by evolutionary forces . We estimated the power of the widely-used logistic regression approach to identify regions containing at least one significant marker . For the parameters simulated , power maximizes at 28% in a GWAS using common markers and at 38% in a resequencing study , when λ = 0 . 075 ( Figure 3a ) . When λ = 0 ( no deleterious risk mutations present ) , power is 0 at significance level α = 10−8 . Further , the cumulative distribution of p-values for λ = 0 is a line with a slope less than one , indicating that the logistic regression test is conservative when applied to our simulated data ( data not shown ) . For small values of λ>0 , broad-sense heritability is also lower ( Figure 1d ) , resulting in less power . As λ increases broad-sense heritability reaches a plateau ( Figure 1d ) , but after that plateau is reached power begins to decrease as causative variants become more and more rare in the general population ( Figure 3a , Figure S4 ) . Thus , depending on the value of the largely unknown parameter λ , current approaches based on common markers have limited power to identify genes harboring causative deleterious alleles , consistent with the idea that some of the “missing heritability” associated with current generation GWAS is due to RALE . Although we are not the first to point this out ( c . f . [2] ) , complete resequencing of cases and controls may only yield a modest improvement in power under a marker-by-marker GWAS ( Figure 3a ) . Finally , the power of GWAS and resequencing studies to identify gene regions is only slightly higher in a region of zero recombination ( Figure 3a ) . We applied Madsen and Browning's rank-sum test [16] , Li and Leal's multiple marker test [15] , and the software package SKAT [21] , [29] , [30] to our simulated data . The first two tests have been proposed to detect an excess of rare alleles amongst cases for gene regions ( typically genes , or a fixed physical sliding window ) . We employed a p-value threshold of 10−6 ( compared to the more conservative 10−8 for a SNP-by-SNP GWAS ) for these gene-based tests , as they integrate over markers , and thus fewer tests are carried out when doing a genome-wide scan . The Madsen and Browning test results in an excess of small p-values ( compared to the same test in λ = 0 controls ) across a wide range of λ , with the excess being greater in resequencing studies than chip-based GWAS ( Figure S5a–S5j ) , but the p-values are rarely small enough to reach genome-wide significance . As a result power maximizes at 5 . 2% for resequencing studies and intermediate λ ( Figure 3b ) . In contrast , the Li and Leal multiple marker test shows no enrichment for small p-values ( Figure S6a–S6j ) , and power <1% for all λ ( Figure 3c ) . In the absence of recombination , the Madsen and Browning test shows a greater excess of small p-values and power maximizes at 13 . 6% when λ = 0 . 025 ( Figure 3b , Figure S7 ) . The power of the Li and Leal test was unchanged in the absence of recombination ( Figure S8 and Figure 3c ) . Both the Madsen and Browning and Li and Leal tests are designed to detect an excess of rare alleles in cases versus controls . However , under our model there is only a very slight excess of rare variants in cases relative to controls at disease genes . This is because the proportion of rare variants that are disease-causing ( as opposed to neutral ) in a gene region at equilibrium is small and the sampling variance on this proportion is large under the mutation-selection balance model we consider . Although the test statistics proposed by Madsen and Browning and by Li and Leal are reasonable , the information they are exploiting , which depends on a net excess of rare alleles in cases , is generally unable to distinguish cases from controls when applied to our simulated data . We note that it is possible that the Madsen and Browing and the Li and Leal tests would be more powerful when applied to a subset or markers chosen a-priori to be potentially functional . However , the high variance in the relationship between effect size and average allele frequency of a causative deleterious marker ( see Figure 2 of [23] for the case of multiplicative fitness effects ) suggests that the signal-to-noise ratio may still remain low . The power of the SKAT software to detect associations in recombining regions is shown in Figure 3d . For GWA studies , power peaks at 27 . 2% when λ = 0 . 05 . The two weighting schemes applied to individual markers ( see Methods ) result in similar power profiles . For resequencing studies , power maximizes at 54 . 5% when λ = 0 . 05 , with the power being greatest when using Madsen and Browning's [16] weighting scheme for individual markers ( Figure 3d; Madsen Browning weights are not equivalent to the test proposed in [16] ) . When applied to non-recombining regions , we observe approximately 10% less power across all effect sizes , and the different weighting schemes give similar power profiles for both GWAS and resequencing studies ( Figure 3e ) . An interesting feature of the Manhattan plots ( Figure 2 ) is that for all but the highest values of λ we observe a “genetic signature” of a gene contributing to a complex phenotype that consists of a large number of markers with suggestive , but not globally significant , p-values ( e . g . , contrast Figure 2a with b–e ) . Upon further examination the majority of the tagged causative mutations are rare in the population and , since they tend to occur on different genetic backgrounds , are only weakly correlated with one-another . This observation suggests that the design and implementation of a statistical test that integrates over approximately independent rare markers located in a gene-region is a fruitful avenue for future research . We applied a new statistical test ( ESM , described in Methods ) to our simulated gene regions to determine if there is information not currently being exploited by published statistical tests . The ESM test statistic is the sum of the difference in the observed and expected p-values ( on a log10 scale ) of the M most significant markers in a genomic region ( see Methods for details ) . Control simulations with no causative mutations show that a permutation procedure ( see Methods ) results in the correct distribution of p-values ( Figure S10a ) and a power of zero at a significance threshold of 10−6 ( Figure 3f ) . The ESM statistic is the most powerful of all the statistics evaluated under either a common marker GWAS or resequencing experimental paradigm , with complete resequencing giving the highest power than GWAS over all values of λ . When only common markers are genotyped the marker-based logistic regression and SKAT are the second and third most-powerful approaches respectively ( Figure 3a and 3d ) , whereas SKAT is the second most-powerful approach under resequencing ( compare Figure 3d to Figure 3f ) . For λ in the range of 0 . 05–0 . 15 , the power of the ESM test can approach 77% and the power only drops below 20% for λ>0 . 35 . In the absence of recombination , power can be as high as 82% for intermediate λ ( Figure 3f ) . The statistical properties of this statistic are detailed in Figures S9 and S10 . Such a test could be implemented genome-wide using a sliding-window or a gene-centric approach . We developed this test to serve as an illustrative example of a test that attempts to integrate information over a gene-region , and we suspect that more sophisticated tests could be designed to detect the cumulative effects of rare variants in a gene region . Goldstein and colleagues [19] , [38] have proposed that common variants may be tagging haplotype ( s ) harboring several low frequency causative alleles . However , their model assumes that causative mutations may be modeled by placing them on neutral genealogies within a small window of frequencies [19] . When both the effect sizes and allele frequencies of causative polymorphisms are random outcomes of the evolutionary process , we observe that significant common variants tend to tag a single causative variant ( Figure S4c ) of small effect that has drifted to modest frequency ( Figure S4d ) . If RALE are deleterious instead of neutral , this observation casts doubt on the claim that common variants generally tag haplotypes harboring several low-frequency causative alleles . Our results suggest a different interpretation of missing heritability , one which is consistent with standard population-genetic predictions of an inverse relationship between frequency and the effect size of a deleterious mutation [11] , [12] , [23] . Conditional on observing a significant common marker near a gene experiencing recurrent deleterious mutations , that marker likely tags a single causative SNP whose effect size is small enough ( and therefore selection weak enough ) that that mutation drifted to high frequency . Thus , the missing heritability in our simulations is due to that single association tagging only one out of several causative variants segregating in a region , with the effect size of that tagged mutant being smaller than that of others segregating in the region . We note that this phenomenon is not unique to the model simulated here . Any evolutionary model with a distribution of negative selection coefficients associated with newly arising deleterious mutations will predict an inverse relationship between population frequency and selection coefficient , conditional on a variant segregating in the population ( e . g . [23] ) . Consistent with this hypothesis , the mean number of singletons on a haplotype defined either by the number of copies of the derived allele at the most significant marker in a GWAS , or by the number of copies of the derived allele at a SNP not associated with case control status , do not differ appreciably for the parameters considered here ( Figure S11 ) . Therefore , under the gene-based mutation-selection balance model considered here , significant associations are not tagging haplotypes with unusual numbers of rare alleles on average ( c . f . 19 ) . Wray et al . [28] have pointed out that the allele frequencies of strongest associations in current GWAS are nearly uniformly distributed ( see their Figure 2a ) . We sampled markers from our simulated case/control samples such that the MAF on GWAS chips are uniformly distributed ( Figure 4a ) . This distribution matches the simulated MAF distribution in [28] and that seen on SNP chips . When we use such a SNP chip to carry out GWAS under our evolutionary model of RALE , the resulting MAF distribution at significantly associated SNPs appears rather uniform , with a slight excess toward intermediate MAF for some λ ( Figure 4b–4f ) . Thus our simulations are consistent with the results of current GWAS , and inconsistent with Dickson and colleagues [19] , [38] as represented in [28] ( c . f . , their Figure 2 ) . We conclude that many currently reported associations presumably reflect bona fide intermediate frequency variants , and that the “missing heritability” problem may arise from GWAS being biased towards detecting associations with causative mutations of small effect relative to the average effect size at a causative gene .
Risch [39] , [40] presented an early and influential attempt to model the genetics of complex traits and to frame the model in terms of measurable parameters such as relative risk . For a given locus , he considered the case of a single risk allele ( the product of a single mutational event some time in the past ) with some specified effect size . Risk alleles at different genes interact multiplicatively to generate an individual's phenotype , a common assumption in multi-locus models in evolutionary biology [41] , [42] . Risch and Merikangas [13] used this model to calculate the power to detect a risk allele at an arbitrary frequency in the population . Pritchard [11] was the first to add explicit evolutionary considerations to this model , extending Risch's model to the case of a constant-size , randomly-mating population subject to recurrent mutation to risk alleles at multiple loci ( with constant effect sizes of risk alleles at a single locus , but varying across loci ) , multiplicative interaction between loci , and natural selection against risk alleles . In Pritchard's model , the equilibrium frequency distribution of the risk allele class at a single locus is known from population genetics theory ( [43] , also see equation 1 of [11] ) , and the frequency of the risk allele class is the sum of the frequencies of the individual susceptibility alleles that have arisen independently at different positions in a non-recombining region . Pritchard's model was an important conceptual advance , allowing the frequencies of the risk allele class to be the random output of the interplay between recurrent mutation , genetic drift , and natural selection . However , due to computational constraints , Pritchard did not explicitly track the frequency of each individual mutation within the risk allele class , nor did he incorporate neutral markers into the model . These two limitations , and the assumed lack of recombination within loci , prevented him from explicitly evaluating the power to detect associations in the case where risk alleles at a single gene are the result of different mutational events embedded in a genomic region consisting largely of linked neutral SNP markers . Since Pritchard's [11] paper , the application of population-genetic principles to our understanding of the properties of GWAS has been heterogeneous . Rather than employing explicit simulations of the evolution of a disease phenotype , recent studies have employed a variety of approximations ( [11] , [15]–[17] , [19]–[21] , [24]–[25] , [29] , also see Introduction ) , largely due to computational constraints , and possibly due to the lack of appropriate simulation machinery ( but see [44] ) . As a result , much of the theoretical/statistical current literature on RALE does not incorporate a notion of a gene ( e . g . [26] ) and statistical methods are rarely tested on simulated data that can be described as outcomes of a biological or evolutionary process . Therefore , to more accurately model the ability of GWAS to identify a gene-region harboring RALE , new evolutionary models of gene action are required that are based on a standard well-accepted definition of a gene [26] . Several studies that have carried out resequencing of candidate gene exons in case/control samples have observed an excess of rare non-synonymous mutations in the cases [3]–[8] . Implemented on a genome-wide scale , this “exomics” approach to the genetic dissection of complex traits would most certainly pay dividends [45] . However , it is important to note that scanning for an excess of rare variants within cases may be less fruitful when variants cannot be classified a priori as putatively causative ( e . g . focusing on amino acid variants in coding regions ) . Our model indeed suggests that tests focusing on detecting such an excess of rare mutations in cases have low power when there are no a priori weights applied to different sites within a region and when causative mutations are a small fraction of the total number of variants in a region ( Figure 3b–3c , Figure S3e–S3e ) . These assumptions are likely to be satisfied if some fraction of complex disease is due to mutations in cis-regulatory regions and , thus , intuition gained from scanning for RALE in exons may be misleading . Our model is consistent with the hypothesis that many rare variants could exist at a relatively small number of genes , and as a class those variants are likely to make a measurable contribution to the variation in complex traits . It is not unreasonable to assume that those variants are partially recessive and partially fail to complement one another when located in the same gene . An important aspect of our model is that causative mutations may be located anywhere in a large gene region that includes regulatory and splicing control regions , and causative mutations are not limited to point mutations . We show that simple extensions to current marker-by-marker tests have considerable power to detect genes harboring such variants . GWAS employing common markers have harvested the “low-hanging fruit” associated with intermediate frequency causative variants . In light of mounting evidence that common variants only explain a small fraction of the genetic variation in complex disease phenotypes , it behooves us to design experiments that have reasonable power to uncover the genetic architecture of complex traits under specific population-genetic models purporting to explain the existence of variation in these traits . Forward simulations that can track entire gene regions under intuitively appealing models of gene action and fitness allow us to assess the power of different experimental designs .
We implemented a forward-time simulation of a Wright-Fisher population with mutation following the infinitely-many sites model [46] , recombination , and selection occurring each generation . We simulated a population of N = 20 , 000 diploids with a neutral mutation rate of μ = 0 . 00125 per gamete per generation , and a recombination rate of r = 0 . 00125 per diploid per generation . These values correspond to the scaled parameters θ = 4Nμ = 100 and ρ = 4Nr = 100 , and thus correspond to a “typical” 100 kilobase region of the human genome . The mutation rate to causative mutations was μd = 0 . 1 µ per gamete per generation . In our model , causative mutations are treated as SNPs for simplicity , but should be viewed more generally as genetic events ( including copy-number variants and transposable element insertions ) that we assume to be detectable via a chip or resequencing assay . We note that there are a variety of forward-time simulation programs in the literature . However , the majority of these either simulate non-gene-based models [22] , [47] , [48] , models involving only unlinked makers [49] , or only neutral models [50] . Further , none of them simulate the explicit genotype-phenotype relationship assumed here ( see Introduction ) . An individual carries c1 and c2 causative mutations on each haplotype . The effect size of the ith mutation on the jth haplotype is , and the phenotype of an individual is P = where x is a Gaussian deviate with mean 0 and standard deviation σe , which we fix at 0 . 075 in the simulations . In words , the phenotypic effect of a single haplotype is additive over causal mutations , and the phenotype of an individual is the geometric mean of the effects of each haplotype plus Gaussian noise . Since phenotypes are continuous they represent the underlying liability of developing a disease [51 , chapter 18] . When we refer to heritability and phenotypic distributions in the population in the text , such references are in regards to these liabilities . The phenotypes are under Gaussian stabilizing selection with a standard deviation of σs = 1 , and w , the fitness of a diploid , is proportional to . In our simulations , the effect sizes of causative mutations are exponentially distributed with means of λ = 0 , 0 . 01 , 0 . 025 , 0 . 05 , 0 . 075 , 0 . 1 , 0 . 125 , 0 . 175 , 0 . 25 , 0 . 35 , or 0 . 5 . For each λ>0 , we performed 250 independent simulations . For an effect size of 0 , representing “control” simulations where there is no genetic contribution to risk , we simulated 1000 independent replicates . All simulations were run for 8N generations prior to sampling . For the parameters μ = 0 . 00125/gamete , μd = 0 . 1 µ , σs = 1 , σe = 0 . 075 , and r = 0 . 00125/diploid or 0 , we simulated both neutral and causative markers , allowing us to examine the properties of GWAS in detail . In order to reduce computational time , for all other parameter values explored , we set μ = 0 ( i . e . , no neutral mutations were simulated ) and only simulated the causative sites . By not simulating the neutral mutations , simulations run orders of magnitude faster , allowing us to look at heritability across a broader parameter space . For each simulated population , 3000 cases and 3000 controls were sampled . A case was defined as being in the upper 15% of the phenotypic distribution , and controls were within 1 standard deviation of the population mean . For each case-control panel , we define a GWAS to include all markers present in the panel with a minor allele frequency ≥5% , and a resequencing study to include all markers . For both types of study , we performed a logistic regression of case/control status onto genotype under an additive model . The significance threshold used was 10−8 , representing a typical cutoff used in current GWAS [19] , [52] . The power simulations refer to these case/control samples , with power defined as the proportion of replicate simulations with at least one marker in the gene region exceeding the genome-wide significance threshold . The forward simulations required approximately six weeks on a cluster of 96 computing cores ( AMD Opteron 6168 , 1900 Mhz ) . To facilitate the further development of tests for detecting associations in gene regions , we have made all source code , forward simulation output , and case/control files available online at http://www . molpopgen . org/ThorntonForanLongPLoSGenetics . html . In addition to the single-marker test , we also applied several existing and one new test of an association of genotype with case/control status to our simulated data . For tests applied to a set of markers within a genomic region , the significance threshold should be less conservative than the 10−8 used for the single-marker test . Our simulated data are 100 kilobase regions , from a genome of approximately 3×109 base pairs , giving 3×109/105 = 3×104 non-overlapping windows . A conservative significance threshold would thus be 0 . 05/ ( 3×104 ) = 1 . 67×10−6 . Here , we take p< = 10−6 as the significance threshold for all region-based tests ( following , for example , [21] ) . We developed a statistical test that attempts to integrate significance over marginally significant variants in a single gene . Under the gene-based model , genes harboring causative mutations tend to display such a genetic signature , and the ESM statistic is larger when there are more marginally significant mutations in a genomic region . Given a vector of Fisher's exact test p-values ( p ) comparing allele counts in cases and controls for M unique markers ( i . e . , redundant markers collapsed ) from a gene region , we define Y1 to be the largest value of negative base ten logarithm of p , Y2 the second largest , etc . Then our test statistic is:The rationale for the statistic comes from the fact that , if the data were truly drawn from the null model of no contribution of genotype to case/control status , then the expected distribution of p-values is uniform on the interval ( 0 , 1] . In other words , for a large number of independent tests applied to data from the null model , the expected fraction of tests with p< = x is x . However , when data truly come from an alternative model , and a test has power greater than the false positive rate , the expected fraction of independent tests with p< = x is greater than x . The ESM statistic is the sum of the difference between the observed and expected p-values ( on a log10 scale ) of the M most significant markers in a region . The test statistic was calculated for two different conditions . First , for GWA studies where , as above , only minor allele frequencies ( MAF ) ≥0 . 05 were included . The second condition assumed complete resequencing of individuals and included all markers . For the latter case , and for GWAS assuming a recombining region , we considered values of M = 50 , since the value of ZM was generally observed to plateau by this point ( averaging the statistic over replicates as a function of M ) . For GWAS in non-recombining regions , we considered values of M = 25 , as too few simulations had more than 25 unique markers with the requisite MAF to consider larger M . For both GWA and resequencing studies , the minimum count of a minor allele had to be 4 in order for a marker to be included in the analysis . We have also applied several other “region-based” tests designed to detect a contribution of rare alleles to disease risk within a defined genomic region . The first test is Madsen and Browning's [16] rank-sum test . This test ranks individuals using a score that is a function of how many mutations they carry ( weighting the contribution of each mutation to the score by its frequency in the control individuals ) , and the statistic is the sum of ranks in affected individuals . We calculated the test statistic under Madsen and Browning's “general genetic” model , where the score for an individual at a particular marker equals the number of copies of the minor allele carried by the individual ( 0 , 1 or 2 ) . We chose this model because , under the genetic model that we have simulated , it will be rare for affected individuals to be homozygous for a single causative mutation , implying that the per-marker recessive model of Madsen and Browning would not be appropriate . We applied the test to two treatments of the data—GWAS ( MAF>5% ) , and complete resequencing ( no MAF filter ) . The second test is Li and Leal's [15] multiple marker test , which amounts to calculating Hotelling's T statistic on a matrix of genotype scores ( aa = −1 , Aa = 0 , AA = 1 , where a is the minor allele ) . We applied this test to either the 50 , 100 , 200 , or 250 rarest variants ( by minor allele frequency ) present in a case-control panel . The Hotelling T statistic was calculated using the “pseudoinverse” function from the “corpcor” library [53] in R [54] for matrix inversion . In practice , the routines used for the matrix inversions required to calculate the test statistic were numerically unstable for larger numbers of markers , resulting in the absence of a p-value for some replicates . For Madsen and Browning's , and for Li and Leal's , test statistics , we did not first collapse redundant markers . The rationale for not collapsing is that if a “case” contains , for example , two singleton mutations ( e . g . , present only once in the entire case/control panel ) , then those two mutations would count more towards case/control differences in the permutation test than they would in the ESM test statistic . Thus , any differences in the power between ESM and the other two statistics should be viewed as conservative . Finally , we applied the SKAT software [21] , [29] , [30] ( available from http://www . hsph . harvard . edu/xlin/software . html ) to all of our simulated data . We applied the software in two different ways . First , we used default weights on individual markers and the optimal p-value approach described in [30] . Second , we applied the marker weights proposed in Madsen and Browning [16] in combination with the optimal p-value approach . Note that the latter is not equivalent to Madsen and Browning's rank-sum test [16] , but is simply a variant of the SKAT procedure using a different weighting scheme . Because the data are simulated with no complications such as population substructure , sex-specific effects of risk alleles , etc . , the only covariate needed for the assessment of significance is the case/control status of individuals . For the ESM , Madsen-Browning , and Li and Leal tests , we assessed statistical significance following the permutation procedure outlined in [16] . Case and control labels were permuted 1000 times , resulting in a permutation distribution of the statistic , . The observed value of the statistic was converted into a z-score ( z = ) , where is the standard deviation of the permuted distribution . The distribution of z-scores under the null model of no association with disease is expected to be a unit Gaussian with mean 0 ( which we confirmed using control simulations , see panel A of Figures S5 , S6 , S7 , S8 , S9 , S10 ) , which was used to calculate two-tailed p-values . The SKAT software obtains p-values by fitting a logit model to the data [21] and thus there is no need for permutation . In the analyses described above , we assume that a GWA study is conducted using perfect genotyping technology able to assay 100% of markers with minor allele frequencies >5% . However , the majority of GWAS to date have used genotyping chips that assay a subset of ascertained markers whose minor allele frequencies are uniform in the range of 0 . 05 to 0 . 5 [28] . In order to mimic these chips , we resample markers from our case/control panels ( described above ) , including a marker on the “chip” if a uniform random number on the interval ( 0 , 1] is < = the heterozygosity of the minor allele in the control population . This sampling results in sample of markers with a uniform distribution of MAF in the desired frequency interval , although some MAF may be >0 . 5 because the minor allele is defined in the general population , and the control population is a random sample of the general population . We use these imperfect chips to look at the MAF distribution of the most significant marker ( defined by a logistic regression test described above ) in a gene region ( following [28] ) . Specifically , we ask what the frequency of the most significant minor allele is in the case population . However , as the number of significant markers per simulated replicate may be quite low ( even when using a chip assaying all markers ) , the resulting distribution of MAF may be very noisy . To reduce this noise , we estimate the expected number of most-associated markers in different frequency bins by randomly sampling 1 , 000 imperfect chips from each of our 250 replicate case/control populations for each value of λ . | Current GWA studies typically only explain a small fraction of heritable variation in complex traits , resulting in speculation that a large fraction of variation in such traits may be due to rare alleles of large effect ( RALE ) . The most parsimonious evolutionary mechanism that results in an inverse relationship between the frequency and effect size of causative alleles is an equilibrium between newly arising deleterious mutations and selection eliminating those mutations , resulting in an inverse relation between effect size and average frequency . This assumption is not built into many current models of RALE and , as a result , power calculations may be misleading . We use forward population genetic simulations to explore the ability of GWAS to detect genes in which unconditionally deleterious , partially recessive mutations arise each generation . Our model is based on the standard definition of a gene as a region within which loss-of-function mutations fail to complement , consistent with the multi-allelic basis for Mendelian disorders . Our model predicts that it may not be uncommon for single genes evolving under our model to contribute upwards of 5% to variation in a complex trait , and that such genes could be routinely detected via modified GWAS approaches . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"genome-wide",
"association",
"studies",
"genetic",
"polymorphism",
"genetics",
"population",
"genetics",
"biology",
"human",
"genetics",
"genetics",
"and",
"genomics"
] | 2013 | Properties and Modeling of GWAS when Complex Disease Risk Is Due to Non-Complementing, Deleterious Mutations in Genes of Large Effect |
Globally re-emerging dengue viruses are transmitted from human-to-human by Aedes mosquitoes . While viral determinants of human pathogenicity have been defined , there is a lack of knowledge of how dengue viruses influence mosquito transmission . Identification of viral determinants of transmission can help identify isolates with high epidemiological potential . Additionally , mechanistic understanding of transmission will lead to better understanding of how dengue viruses harness evolution to cycle between the two hosts . Here , we identified viral determinants of transmission and characterized mechanisms that enhance production of infectious saliva by inhibiting immunity specifically in salivary glands . Combining oral infection of Aedes aegypti mosquitoes and reverse genetics , we identified two 3’ UTR substitutions in epidemic isolates that increased subgenomic flaviviral RNA ( sfRNA ) quantity , infectious particles in salivary glands and infection rate of saliva , which represents a measure of transmission . We also demonstrated that various 3’UTR modifications similarly affect sfRNA quantity in both whole mosquitoes and human cells , suggesting a shared determinism of sfRNA quantity . Furthermore , higher relative quantity of sfRNA in salivary glands compared to midgut and carcass pointed to sfRNA function in salivary glands . We showed that the Toll innate immune response was preferentially inhibited in salivary glands by viruses with the 3’UTR substitutions associated to high epidemiological fitness and high sfRNA quantity , pointing to a mechanism for higher saliva infection rate . By determining that sfRNA is an immune suppressor in a tissue relevant to mosquito transmission , we propose that 3’UTR/sfRNA sequence evolution shapes dengue epidemiology not only by influencing human pathogenicity but also by increasing mosquito transmission , thereby revealing a viral determinant of epidemiological fitness that is shared between the two hosts .
Dengue viruses ( DENV ) , members of the flavivirus genus , infect about 390 million people annually and threaten one third of the world population [1] making them the most important arboviruses in the world . DENV are primarily transmitted by Aedes aegypti mosquitoes [2] . Increase in transmission efficiency augments arbovirus epidemiological fitness ( EF ) [3 , 4] , defined hereafter as the capacity to generate an epidemic , and can trigger epidemics as observed in other mosquito-virus systems [5–7] . Successful transmission following an infectious blood meal occurs when the virus infects the midgut , escapes from this organ to multiply in secondary tissues , including hemocytes , fat bodies or muscles , and productively infects the salivary glands , from which it is expectorated along with saliva during subsequent blood feedings [3 , 8 , 9] . Upon infection , mosquitoes mount a powerful innate immune response [10] mediated by the Toll , IMD , Jak/STAT , RNAi , and/or TRAF/Rel2 pathways that can inhibit DENV [11–17] . Transmission efficiency is likely dictated by the balance of power between host antiviral measures and viral countermeasures [18 , 19] . Identification of the viral factors that interfere with the immune response will reveal viral strategies to enhance transmission and facilitate identification of viruses with high EF . One such factor is the subgenomic flaviviral RNA fragment ( sfRNA ) , produced from partial degradation of flavivirus genomes by the host exonuclease XRN1 that stalls at 3’UTR nuclease resistant structures , functionally labelled XRN1-resistant RNAs ( xrRNAs ) [20–23] . Different species of DENV sfRNA can be produced from the 3’UTR sequence depending on which xrRNA the degradation stalls at [24] . In mosquitoes , sfRNA from DENV-1 ( serotype 1 ) and West Nile virus ( WNV ) can suppress RNAi processing of an ectopic substrate by inhibiting Dicer-2 cleavage [25] , likely through direct binding [26] . Nonetheless , a recent in vivo study demonstrated the lack of RNAi alteration by WNV sfRNA in spite of sfRNA processing by the RNAi machinery [27] . The study further showed a transmission decrease by Culex mosquitoes ( the main vector of WNV ) for sfRNA-deficient WNV , and indicated that reduction of transmission potential was mediated by an unidentified mechanism in midgut . Phylogenetic clustering of flavivirus 3’UTR sequences according to vector species suggests functional specialization for sfRNAs in different vector-virus system [28] . In humans , sfRNA alters innate immune response by altering interferon production and the cellular response to interferons [29–32] . Evolution of the DENV-2 ( serotype 2 ) sfRNA sequence in more fit viruses increases production of this non-coding RNA and/or augments binding-specificity to innate immune regulators , both concurrently antagonizing the antiviral response [30] . The resulting increased infectivity provided an explanation for a DENV-2 lineage replacement that occurred in Puerto Rico in 1994–1995 ( PR-2B replacing PR-1 isolates ) , which resulted in an important outbreak [33] . Altogether , these studies , and several others , suggest a strong immune suppressor function for sfRNA in both humans and mosquitoes . Nevertheless , understanding of the relationship between sfRNA evolution , its anti-immune function in mosquitoes and how these impact mosquito transmission is lacking [34] . Such knowledge would serve molecular surveillance to predict emergence of viruses with high transmission capacity and improve the characterization of the evolutionary pressures that shape sfRNA , a determinant of dengue pathogenesis[30] . To address this gap in our knowledge , we tested whether or not variant 3’UTR sequences alter sfRNA quantity in mosquitoes and investigated how these influence transmission . We made use of an existing bank of DENV-2 isolates collected in Puerto Rico with different EF and variable 3’UTR sequences [30 , 33] . After oral infection of Ae . aegypti mosquitoes with high and low EF isolates , we observed that , although overall levels of viral genomes ( gRNA ) and progeny were not different between these , high EF isolates produced higher sfRNA:gRNA ratios . Using two representatives of each epidemic level , we noted significantly higher sfRNA:gRNA ratio in salivary glands for the high EF isolate . Remarkably , regardless of the isolate , the sfRNA:gRNA ratio was higher in salivary glands than in midgut or carcass , suggesting a heretofore unprecedented tissue-specific regulation of sfRNA accumulation . Using reverse genetics , we demonstrated that two 3’UTR substitutions in the high EF isolate were responsible for the difference in sfRNA:gRNA ratios in salivary glands . We then found that the isolate with higher EF and the chimeric virus with the corresponding 3’UTR resulted in higher infection rate of saliva . Finally , we determined that the high EF isolate disrupted the Toll immune response in salivary glands and incriminated the same two 3’UTR substitutions that alter sfRNA quantity .
To study whether sfRNA determines EF by influencing mosquito infection , we selected DENV-2 isolates within the Asia subtype IIIb with different EF and 3’UTR sequences . The isolates were selected from those identified in a study that examined viral phylogeny over the twenty year period from 1981–2001 in Puerto Rico [33] . This study revealed the replacement of the endemic PR-1 by an epidemic PR-2B clade over the years 1994–1995 , resulting in an important outbreak [33] . Full genome sequencing of viruses from the two clades refined the phylogenetic history and implicated changes in the 3’UTR in EF [30] . Three isolates from the clade PR-2B with high EF , which we defined as such because many highly related isolates were collected during the outbreak , were selected [30] ( Table 1 ) . Isolates with lower EF included five isolates from the replaced endemic PR-1 clade and PR315022 , which belongs to PR-2B clade but does not have homologous isolates from the outbreak , suggesting low EF . Sequence identity between our virus stocks and the original passage 0 isolates was validated using NGS for one representative of PR-1 ( PR1940 ) and two of PR-2B ( PR6452 and PR315022 ) ( raw sequences available on NCBI: SRX2617313-SRX26173151 ) . The 3’UTR sequences varied between and within the clades and all PR-2B isolates possessed the three substitutions associated with immune suppression in humans ( S1 Fig ) [30] . To test whether higher EF was associated with increased vector susceptibility , we orally infected Ae . aegypti mosquitoes with the aforementioned PR-1 and PR-2B isolates , and quantified feeding rate , mosquito survival and virus infectivity . To use an epidemiologically relevant inoculum dose , we offered mosquitoes an artificial blood meal containing 106 pfu/ml . The inoculum concentration is in the lower range of the viremia measured in hospitalized children during the first couple of days after fever onset[35] , which overlaps with the peak of transmissibility to mosquitoes[36 , 37] . Mosquito feeding rate was not affected by the isolates; except for two PR1 isolates , PR2974 and PR4056 , which exhibited significantly higher and lower rates , respectively ( S2A and S2B Fig ) . Survival of mosquitoes at 10 days post-infection ( p . i . ) was not altered after infection with the isolates , except for PR1940 , PR0013 and PR6913 that had significantly lower survival than PR315022 ( S3A Fig ) . Importantly , survival did not segregate with EF level ( S3B Fig ) . At 21 days p . i . , survival of mosquitoes infected with three viral isolates , which represented high and low EF viruses , was not altered as compared to non-infected mosquitoes ( S3C Fig ) . A decrease in the rate of blood feeding or in survival could diminish vector capacity and therefore virus fitness [38] , but our data are not consistent with either of these parameters explaining fitness of the studied DENV-2 isolates . We detected and quantified DENV genomic RNA ( gRNA ) in whole mosquitoes 10 days p . i . using RT-qPCR ( S4 Fig ) . Infection rate , which was defined as percentage of infected mosquitoes among blood-fed ones , varied between isolates ( Fig 1A ) but did not segregate with EF level ( Fig 1B; p = 0 . 27 ) . Similarly , the number of gRNA copies per infected mosquito varied between isolates ( Fig 1C ) but did not segregate with EF level ( Fig 1D; p = 0 . 98 ) . To determine whether the production of infectious particles was linked to EF level , we orally infected mosquitoes with the three representative isolates , PR1940 , PR6452 and PR315022 ( Table 1 ) , and 10 days p . i . measured plaque forming units ( PFU ) from whole mosquitoes . Virus titer was not different between the three isolates ( Fig 1E ) . Therefore , the level of viral genomes or of infectious particles in whole mosquitoes did not correlate with viral EF . We then quantified the sfRNA copies per infected mosquito at 10 days p . i . with the same nine isolates , and calculated the ratio of sfRNA to gRNA ( sfRNA:gRNA ) in whole mosquitoes . The forward primer used to quantify sfRNA with RT-qPCR annealed after the xrRNA1 structure ( S1 Fig ) and , hence , only detects the longer sfRNA species that results from XRN1 stalling at the xrRNA1 [24] . Strikingly , the sfRNA:gRNA ratio was significantly higher for the high EF isolates than for isolates with lower EF ( p = 0 . 001 ) , PR-1 isolates and the PR-2B isolate , PR315022 ( Fig 1F and 1G ) . Infection with PR315022 produced the lowest sfRNA:gRNA ratio of 3 . 6 ( ± 0 . 83 ) , whereas infection with PR6452 ( high EF PR-2B ) resulted in the highest ratio of 38 ( ± 14 . 57 ) . The 3’UTR sequences for PR6452 and PR315022 differ by two nucleotides: a C to U transition 5 nt downstream of the termination codon that is not expected to be in the mature sfRNA [39] , and a U to C transition in the distal loop of the first nuclease resistant ( xrRNA1 ) structure ( Fig 1H ) . Both substitutions are absent from the other PR-1 and PR-2B isolates ( S1 Fig ) . Substitutions in sfRNA sequences can disrupt the nuclease resistant structures and result in shorter sfRNAs [24 , 39] , to test whether or not this was the case between PR6452 and PR315022 we used Northern blots to compare sfRNA size and quantity in the whole body of Ae . aegypti collected 10 days p . i . with these isolates . The probe used was complementary to the 3’UTR sequence starting immediately downstream of xrRNA1 and ending at the 3’ end of the genome ( S1 Fig ) , and can thus detect all sfRNA species [24] . The predominant sfRNA in PR315022 infected cells appeared to be very similar in size to that observed in PR6452 infected cells ( Fig 1I ) , although high resolution Northern Blot with acrylamide gel would be required to confirm this . Further , we confirmed using Northern Blot the higher ratio of sfRNA:gRNA in PR6452-infected whole mosquitoes . Altogether , our data show an association between higher EF and higher sfRNA:gRNA ratio in mosquitoes . To compare sfRNA production for the aforementioned nine DENV-2 isolates in human cells , we infected HuH-7 cells and quantified gRNA and sfRNA . As we have reported before [30] , PR-1 isolates produced higher gRNA levels than PR-2B isolates at 24 h . p . i . ( Fig 2A ) . However , similarly to what we report now in mosquitoes ( Fig 1F and 1G ) , high EF isolates had higher sfRNA:gRNA ratios ( Fig 2B and 2C ) . PR315022 was the exception among the PR-2B isolates and had a sfRNA:gRNA ratio more typical of low EF PR-1 isolates . Interestingly , the ratios in human cells and mosquitoes were positively correlated ( Pearson correlation = 0 . 85; p = 0 . 004 ) and the best fit was an exponential regression ( R2 = 0 . 96 vs . linear regression: R2 = 0 . 71 ) ( Fig 2D ) , suggesting that shared determinants set sfRNA:gRNA ratios in humans and mosquitoes . EF for DENV depends in part on transmission efficiency . One measure of efficiency of mosquito transmission is the extrinsic incubation period , determined by the time to infect salivary glands after oral infection [3] . To test whether or not viruses with different EF have different infection kinetics , we orally infected mosquitoes with PR-2B isolates: the high EF PR6452 and the low EF PR315022 . We decided to focus on these two isolates since they are both in the same PR-2B clade and yet they appear to be very different in terms of fitness . At 3 , 7 , 10 and 14 days p . i . , we quantified DENV gRNA copies in the midgut , carcass and salivary glands . We did not pursue the experiments past 14 days p . i . as a majority of female mosquitoes do not live longer than this in wild settings [40] . Infection rate for PR6452 was significantly higher in midgut and carcass at 3 days , and in the salivary glands at 14 days p . i . Indeed , 14 days p . i . , salivary gland infection rate reached 80 and 57% for PR6452 and PR315022 , respectively ( Fig 3A ) . The number of gRNA copies per infected mosquitoes was not significantly different between the two viruses in the three tissues within each time point ( Fig 3B , S1 Table ) . The infection rate results suggest that the high EF isolate PR6452 have slightly faster kinetic of infection than the lower EF PR315022 . Quantification of sfRNA revealed two interesting observations . First , regardless of virus , sfRNA:gRNA ratio was higher in salivary glands than in midgut and carcass ( Fig 3C , S2 Table ) . The effect of tissue was significant in our ANOVA model ( S2 Table; P < 0 . 001 ) and , at 10 days p . i . which was the apparent peak of sfRNA production , there was approximately a ten-fold higher sfRNA:gRNA ratio in salivary glands than in the two other tissues . Second , the sfRNA:gRNA ratio in salivary glands was higher in PR6452 infected mosquitoes than those infected with PR315022 at 3 , 7 and 10 days p . i , . and the difference was significant at 10 days p . i . Both the higher sfRNA:gRNA ratio in salivary glands and for high EF virus suggest a role for sfRNA in salivary glands in transmission efficiency . To investigate whether or not tissue specific variations in exonuclease XRN1 expression could explain the sfRNA:gRNA ratio , we quantified XRN1 mRNA expression in the different tissues . We orally infected mosquitoes with PR6452 or PR315022 or offered them a non-infectious blood meal , and 10 days p . i . , isolated total RNA from salivary glands , midgut and carcass . XRN1 RNA normalized to Actin RNA was measured by RT-qPCR . XRN1 RNA expression was the lowest in salivary glands and was not altered by virus infection ( S5A Fig ) . To estimate efficiency of XRN1-degradation of gRNA , we normalized XRN1 expression to DENV gRNA . XRN1 expression to gRNA was lower in midgut and higher in carcass than in salivary glands , and not different in mosquitoes infected with any of the isolates ( S5B Fig ) . The results suggest that XRN1 RNA expression could not account for differences in sfRNA:gRNA ratio between the tissues and the two isolates . The high EF PR6452 and the lower EF PR315022 isolates have many nucleotide differences ( S3 Table ) that could cause the observed higher sfRNA:gRNA ratio . Since sequences in the 3’UTR are most likely to cause changes in sfRNA levels , we tested the role of the two transitions in this region on infection parameters including the sfRNA:gRNA ratio ( S3 Table , Fig 1H ) . We constructed two chimeric viruses , based on the DENV-2 NGC isolate , identical except for the 3’UTR sequences which were derived from PR6452 or PR315022; thereafter named IC6452 and IC315022 , respectively . The sequence of the two chimeric viruses were validated using NGS ( raw sequences available on NCBI: SRX2617311-SRX2617312 ) and were as expected . The chimeric viruses were used to orally infect mosquitoes , and at 3 , 7 , 10 and 14 days p . i . infection rate , levels of DENV gRNA and sfRNA:gRNA ratios were quantified in midgut , carcass and salivary glands . Infection rate was not different between the two chimeric viruses in midgut and carcass at 3 , 7 and 10 days p . i . , but was significantly higher for IC6452 than for IC315022 in salivary glands at 7 and 10 days p . i . ( Fig 4A ) . Salivary gland infection rate reached 17 . 5% and 2 . 5% at 7 days , and 35% and 15% at 10 days p . i . for IC6452 and IC315022 , respectively . The number of gRNA copies per infected mosquitoes was not significantly different between the two chimeric viruses in the three tissues within each time point ( Fig 4B , S4 Table ) . Similar to what we observed for PR isolates , infection kinetic appeared faster for IC6452 and DENV gRNA copies did not vary between the two viruses . Quantification of sfRNA copies and calculation of the sfRNA:gRNA ratio showed that the chimeric viruses recapitulated the two key observations made with the parental PR isolates . First , sfRNA:gRNA ratio was higher in salivary glands than in midgut and carcass ( Fig 4C , S5 Table ) . The effect of virus was significant in our ANOVA model ( S5 Table; P < 0 . 001 ) and , at 14 days which was the highest sfRNA level , there was at least a ten-fold higher sfRNA:gRNA ratio in salivary glands than in the two other tissues . Second , IC6452 had a higher sfRNA:gRNA ratio in salivary glands at 14 days p . i . . Additionally , we quantified gRNA , sfRNA and viral titers at 14 days p . i . in whole mosquitoes infected with IC6452 and IC315022 . While the infection rate , DENV gRNA levels and viral titer were similar for the two chimeric viruses , the sfRNA:gRNA ratio was significantly higher after infection with IC6452 ( S6 Fig ) . IC6452 and IC315022 recapitulated what we observed with PR6452 and PR315022 , demonstrating that the 3’UTR sequence from the high EF isolates was responsible for higher sfRNA accumulation in salivary glands . While the 3’UTR sequence did not influence gRNA level and the infection rate in salivary glands at the peak of sfRNA accumulation ( 14 days p . i . ) when quantified using RT-qPCR ( Fig 4A ) , we tested whether 3’UTR sequence from high or low EF altered the production of infectious particles in salivary glands . Mosquitoes were orally infected with the chimeric viruses IC6452 or IC315022 , and progeny virus in salivary glands was tittered using a plaque forming assay at 14 days p . i . , which corresponded to the peak of sfRNA production ( Fig 4C ) . The infection rates of salivary glands were not different after infection with either of the chimeric viruses ( Fig 5A ) , and were comparable to the ones calculated by detecting gRNA ( Fig 4A ) . Importantly however , the virus titer in salivary glands was seven-fold higher and significantly different ( p = 0 . 001 ) after infection with IC6452 ( Fig 5B ) . Altogether , our results suggest that 3’UTR sequence from the high EF virus increases the production of infectious particles in salivary glands . To test whether or not higher sfRNA:gRNA ratio and virus titer in salivary glands resulted in higher saliva infection rate ( defined as the proportion of saliva containing DENV gRNA ) , we quantified DENV gRNA in saliva collected 10 days p . i . with PR6452 and PR315022 . To avoid false negatives caused when a mosquito did not salivate , we collected saliva in blood and only analysed samples from infected mosquitoes with blood-containing abdomen . In our conditions , the rate of blood imbibing and the gRNA copies per infected whole mosquitoes were not different between the two isolates ( Figs 6A and S7 ) . While infected saliva contained the same quantity of DENV gRNA copies for both isolates ( Fig 6B ) , the percentage of infected saliva was 28% ( p = 0 . 001 ) higher for mosquitoes infected with PR6452 ( Fig 6C ) . These data suggest that the more epidemiologically fit PR6452 is more likely to be transmitted than its less fit relative PR315022 . To determine the role of the 3’UTR in determining rate of saliva infection , we repeated the saliva quantification experiment with IC6452 and IC315022 . Blood imbibing rate and DENV gRNA in whole mosquitoes were measured 14 days p . i . , which corresponded to the highest observed sfRNA:gRNA ratio , and neither was different between the chimeric viruses ( Figs S8 and 6D ) . As we observed above with the corresponding parental isolates , DENV gRNA in saliva was similar between chimeric viruses ( Fig 6E ) , however , saliva infection rate was 29% ( p < 0 . 001 ) higher after infection with IC6452 ( Fig 6F ) . Altogether , these results indicate that the 3’UTR from the high EF PR6452 increases the sfRNA:gRNA ratio in salivary glands and the infection rate in saliva . To test whether viruses with higher epidemiological fitness disrupt the mosquito immune response in salivary glands , we orally infected mosquitoes with PR6452 , PR315022 or fed them with non-infectious blood . At 10 days post-oral feeding , which corresponded to the sfRNA:gRNA ratio peak in the salivary glands , we quantified gene expression for activators of Toll , IMD , Jak/STAT and Rel2/TRAF pathway in the salivary glands . We also tested these in midgut and carcass to validate the specificity of the immune disruption . We chose signalling activators that are up-regulated upon DENV infection in midgut or carcass [15 , 16]; Rel1a as an activator of Toll , Rel2 of IMD , Domeless of Jak/STAT and Vago of TRAF/Rel2 [16 , 41] . We first validated that DENV gRNA copies were similar between the two isolates for each tissue ( S9 Fig ) . In salivary glands , Rel1a expression was downregulated 3 . 7 fold ( p-value = 0 . 09 ) more by PR6452 than by PR315022 , and Domeless was downregulated 2 . 6 fold ( p-value = 0 . 071 ) more by PR6452 as compared to blood feeding while PR315022 did not alter Domeless expression ( Fig 7A ) . Quantification of expression of the same immune markers in midgut and carcass showed that Rel2 was significantly 1 . 87 times ( p-value = 0 . 016 ) up-regulated in carcass infected with PR6452 as compared to PR315022 ( S10A and S10B Fig ) . To confirm the inhibition of Toll and Jak/STAT pathways in salivary glands infected with PR6452 , we quantified the expression of genes under the control of either of these pathways in the same samples . We chose to study Cecropin G ( CecG ) and Defensin C ( Def C ) , two antimicrobial peptides which are positively regulated by Toll [42] , and Thioester-containing protein 22 ( TEP22 ) , a component of the complement system in mosquitoes [43] , and vir-1 , which function is unknown , both positively regulated by Jak/STAT [11 , 14 , 15 , 44] . In salivary glands , CecG RNA was significantly downregulated after infection with PR6452 as compared to both PR315022 ( 1 . 78 fold; p-value = 0 . 043 ) and non-infectious conditions ( 2 . 04 fold; p-value = 0 . 012 ) ( Fig 7B ) . In midgut and carcass , however , only TEP22 was significantly up-regulated in carcass upon PR6452 infection as compared to both PR315022 ( 2 . 05 times; p-value = 0 . 04 ) and non-infectious conditions ( 1 . 65 times; p-value = 0 . 04 ) ( S10C and S10D Fig ) . These results strongly suggest that PR6452 partially disrupts the Toll pathway in salivary glands by inhibiting Rel1a expression . To determine whether the 3’UTR was responsible for the differential immune disruption of the DENV2 isolates , we orally infected mosquitoes with the chimeric viruses IC6452 or IC315022 , or fed them non-infectious blood . At 14 days post-oral feeding , which corresponded to the time where we observed the highest sfRNA:gRNA ratio , we quantified expression of the same immune pathway activators interrogated above . We first validated that DENV gRNA levels were similar between the two chimeric viruses in each tissue ( S11 Fig ) . In salivary glands , while Domeless expression was 1 . 19 times lower after infection with IC6452 than after non-infectious blood feeding the difference was not significant ( Fig 7C ) . Supportive of a role for the 3’UTR sequence in disrupting Toll pathway , Rel1a was downregulated 2 . 14 fold ( p-value = 0 . 001 ) more by IC6452 than by IC315022 . In midgut and carcass samples , only Domeless in carcass was 1 . 49 fold ( p-value = 0 . 041 ) downregulated upon infection with IC6452 ( S12A and S12B Fig ) . As we did for the parental viruses we quantified expression of genes under the control of Toll and Jak/STAT pathways . In salivary glands , CecG was downregulated 2 . 22 fold ( p-value = 0 . 024 ) upon infection with IC6452 as compared to IC315022 ( Fig 7D ) , whereas none of the genes were differentially expressed in midgut and carcass ( S12C and S12D Fig ) . Altogether , these data demonstrate that 3’UTR sequences determine the preferential disruption of the Toll/Rel1a antiviral pathway in salivary glands infected with the high EF isolate PR6452 .
By demonstrating a link between high EF 3’UTR DENV sequence on one hand , and increased sfRNA quantity , infectious particles in salivary glands , saliva infection rate , and immune disruption in the salivary glands on the other hand , we provide an explanation for how 3’UTR sequence evolution augments mosquito transmission . Our study revealed an unprecedented tissue-specific pattern of sfRNA accumulation in DENV2-infected mosquitoes . The much higher sfRNA:gRNA ratio in salivary glands compared to other tissues suggests an important function for sfRNA in salivary glands , a critical organ for transmission . Furthermore , in salivary glands , but not in the midgut or carcass , we observed dramatically higher sfRNA levels with the high EF isolate when compared to the low EF one . We also observed an interesting temporal variation of sfRNA in salivary glands , where sfRNA peaked after 10 days for the parental isolates and 14 days for the chimeric viruses ( Figs 3C and 4C ) . The time difference in sfRNA peaks may be attributed to the slower infection kinetic of infectious clones , as measured by the slower infection rate in the different tissues ( Figs 3A and 4A ) , and may partially be caused by the limited virus genetic variation for the infectious clones [45] . Because sfRNA production comes at the expense of gRNA , which decrease can restrict replication , translation and packaging , sfRNA production might be optimized to match the transmission window and limit its associated cost . Accordingly , we observed that sfRNA peak levels occur after the extrinsic incubation period measured at the same temperature [9 , 46] . It is intriguing to speculate that the high levels of sfRNA in salivary glands and the timing of its accumulation play important roles in transmission and perhaps infectivity of the virus in the mammalian host . The tissue-specific overexpression of sfRNA raises several questions . We suspect that other tissues or cell types we did not capture in our analysis may also produce high levels of sfRNA . In fact , it should be noted that carcass samples consist of the material remaining left after dissection of salivary glands and midgut , and are mostly devoid of hemolymph fluids and loose tissues that can be infected [9] . This technical limitation impacts sfRNA quantification in dissected tissues as compared to that obtained with whole mosquitoes . In terms of a mechanism , sfRNA quantity is determined by production and degradation kinetics . Increased production could be the results of increased XRN1 activity; however , the lack of correlation between sfRNA:gRNA ratio and XRN1 exonuclease gene expression does not support a determining role for XRN1 expression . These data do not refute possible increases in XRN1 protein or activity . In the same line of thought , increased production should concomitantly diminish gRNA levels , which we did not observe ( Figs 3 and 4 ) . Alternatively , binding of tissue-specific proteins to sfRNA could modulate susceptibility to XRN1 degradation or sfRNA stability . Salivary glands have different transcriptome as compared to carcass [47] and are thus expected to have a unique collection of protein capable of interacting with the sfRNA . Using reverse genetics , we established a link between 3’UTR sequence and sfRNA quantity in whole mosquitoes and in salivary glands . Our group has previously demonstrated that single substitutions in a replicon are sufficient to alter sfRNA:gRNA ratio in human cells [30] . Three 3’UTR substitutions previously associated with high sfRNA:gRNA ratio in humans [30] are present in all our PR-2B isolates ( S1 Fig ) . Two additional nucleotide substitutions in the PR-2B PR315022 , a C to U transition at nucleotide 10 , 278 and a U to C transition at nucleotide 10 , 326 , were sufficient to impair the gain in sfRNA:gRNA ratio in both human cells and mosquitoes . While we cannot definitively state which of the two substitution leads to low sfRNA levels for PR315022 , we propose that U to C at 10 , 326 is the critical one . Using the recently solved structure of Zika virus sfRNA [48] as a model , we propose that this U to C transition disrupts the terminal loop and would weaken the pseudoknot pk1 and the long-range Watson-Crick base-pair between the bulged A in the long stem of xrRNA1 and the U at the base of the same structure ( indicated for PR6452 in green in Fig 1H ) . These alterations are likely to make xrRNA1 much less resistant to XRN1 degradation reducing levels of full length sfRNA from PR315022 . Indeed , the more dramatic reduction in sfRNA:gRNA ratio in PR315022 observed by RT-qPCR , which only measured full length sfRNA ( Fig 1F ) , versus that seen in Northern blots , which detected all sfRNAs ( Fig 1I ) , suggests that the longer sfRNA were likely the most affected by the PR315022 substitution . This is fully consistent with a recent study showing the existence of several species of sfRNA in DENV , with relative proportions influenced by point mutations in the 3’UTR [24] . The continuous unravelling of the plasticity of the nuclease-resistant structures indicates that multi-domain interactions are involved in determining sfRNA quantity . Correlated ratios of sfRNA:gRNA in human and mosquito further suggests a shared 3’UTR-based determinism of sfRNA:gRNA ratio , which could reduce dual-host constraints on 3’UTR evolution if sfRNA fitness advantage is also shared between the two hosts . DENV infection induces an immune response in midgut through Toll and Jak/STAT signalling pathways [10 , 15] . Artificial activation or repression of the Toll and Jak/STAT decreases or increases DENV replication , respectively , by regulating antiviral effectors [11 , 12 , 14 , 15] . In DENV-infected salivary glands , Toll and IMD components are upregulated together with the antimicrobial Cecropin [13] . In another study , genes related to Toll pathway were found to be upregulated in uninfected salivary glands as compared to uninfected carcass and DENV infection was shown to induce several immune-related genes , although Rel1a was not regulated [47] . Interestingly , the different immune signalling pathways appear to co-regulate a set of identical genes as well as components from other immune pathways in mosquitoes [12 , 15] , suggesting cross-talks between signalling cascades . This is exemplified by the Rel2/TRAF pathway . Upon flavivirus infection in mosquito cells , TRAF is activated in a Dicer-2-dependent manner to induce Vago , which activates the Jak/STAT pathway and inhibits flavivirus replication [16] . In our study , salivary gland infection with the high EF isolate or the chimeric virus with the corresponding 3’UTR inhibited gene expression for the Toll component , Rel1a , and Cecropin G ( Cec G ) , which is regulated by Toll [15] . Although the same gene expression pattern was observed at 10 and 14 days p . i . with the isolates and the chimeric viruses , respectively , modification of other immune pathways may occur earlier or later during salivary gland infection . We did not monitor gene expression of the RNAi components as they are not regulated upon DENV infection [15] and , hence , cannot be studied in our system . Nonetheless , sfRNA may alter the RNAi response as previously shown [25] . Inhibition of immune gene expression in DENV-infected cells was previously reported but no mechanism was provided [19] . Here , we observed that sfRNA quantity correlates with Toll pathway inhibition in salivary glands , strongly suggesting an immune-suppressor role for sfRNA in mosquitoes additional to its effects on RNAi . That one of the Toll-induced genes we tested , Defensin C , was not differentially regulated after infection by viruses containing the high EF 3’UTR is suggestive of a fine-tuned inhibition of the immune response . It should be noted that PR6452 not only produces more sfRNA than its less fit relative PR315022 , but it also produces sfRNA ( s ) with different sequence ( Fig 1H ) . The ability of different sfRNAs to bind innate immune regulators in human cells has been demonstrated [30 , 49] and may function similarly in mosquitoes . Indeed , binding to innate immune regulators will be controlled by both sfRNA levels and sfRNA sequence variation . The evolution of sfRNA variants offers flaviviruses a flexible strategy to sample various anti-immune tactics [50] . Our study showed that the presence of a high EF 3’UTR increased the production of infectious particles but not accumulation of gRNA in salivary glands ( Figs 4 and 5 ) . Previous studies observed a decoupling of infectious particle production from gRNA replication in midgut during the course of a normal infection [9 , 51] . In midgut , gRNA copies steadily increase throughout the infection , whereas infectious particles correlate with gRNA copies early during infection and decrease at later time points . Human immune response targets the difference steps of virus production , from replication , translation , packaging to virus egress [52] . While the impact of mosquito anti-DENV response remains to be characterized , our results show that inhibition of the Toll pathway correlates with higher number of infectious particles , which explains the increased probability that excreted saliva contains viruses . In conclusion , while outbreaks of DENV are influenced by multiple ecological and host factors [53 , 54] , viral characteristics are major contributors to epidemic transmission [55] . Here , we identified the 3’UTR as a genetic determinant of virus fitness in mosquitoes and report the first in vivo characterization of how dengue virus influences transmission through the production of sfRNA . Our study provides new mechanistic insights on the role of mosquito transmission in shaping the epidemiology of dengue , and , together with previous studies in mammals [30] , emphasizes the importance of sfRNA evolution . Identification and characterization of viral determinants of mosquito transmission will help forecast emergence of DENV strains with higher epidemic potential .
The Ae . aegypti colony was established in 2010 from eggs collected in Singapore . Adult mosquitoes were held in rearing cages ( Bioquip ) supplemented with 10% sucrose and fed pig’s blood ( Primary Industries Pte Ltd ) twice weekly to maintain the colony . After collection on wet papers , eggs were hatched in mili-Q water . Larvae were fed a mix of fish food ( TetraMin fish flakes ) , yeast and liver powder ( MP Biomedicals ) . The insectary was held at 28°C with 50% humidity on a 12:12h dark:light cycle . Low-passage stocks of virus isolates were a gift from the Dengue Branch of the Centers for Disease Control and Prevention , Puerto Rico and were isolated in Eng Eong Ooi’s laboratory [30] . These isolates were generated after inoculating acute patient sera onto C6/36 cells ( ATCC CRL-1660 ) as previously described [56] . These virus stocks were propagated in C6/36 cell , harvested 5 days post inoculation , aliquoted and stored at −80°C until use . All viral isolates included in this study were passaged 5–8 times . Viral titers were determined by plaque assay as previously reported [30] . An overlap cloning strategy was used to construct the full-length cDNA clones of DENV-2 ( strain NGC ) containing the 3’UTR mutations . Primers are detailed in S6 Table . Fragment A covering “MluI-NS5partial” was amplified with the primers 9127F and NS5-UTR-R using the DENV-2 NGC infectious clone as a template [57] . Fragment B was amplified from plasmids containing the desired 3’UTR with primers NS5-UTR-F and UTR-HDVr-R . Fragment C spanning “HDVr-to-XbaI unique site in pACYC” was amplified with primers UTR-HDVr-F and pACYC-11125-R using the DENV-2 NGC infectious clone as a template . Fragments A , B , and C were fused together to create cassette “MluI-NS5partial-3’UTR-HDVr-XbaI” by overlapping PCR with primers 9127F and pACYC-11125-R . The overlap PCR product was digested by MluI and XbaI and ligated into the full-length DENV-2 NGC infectious clone . All constructs were verified by DNA sequencing . Full-length cDNA plasmids were linearized by XbaI and in vitro transcribed using a T7 mMessage mMachine kit ( Ambion ) . The RNA transcripts ( 10 μg ) were electroporated into BHK-21 cells following as described [58] . The transfected cells were seeded in a T-175 flask ( 8×106 cells in 25 ml DMEM medium supplemented with 10% FBS ) and incubated at 37°C for 24 h before culturing in DMEM medium with 2% FBS at 30°C . At every 24 h post-transfection , 200 μl culture fluids were collected and stored at -80°C . On day 5 post-transfection , all culture fluids were centrifuged at 4°C , 500 g for 5 min , aliquoted , and stored at -80°C . Viral titers were quantified by plaque assay using BHK-21 cell as described [30] . Total RNA from 500μl of the virus stocks was extracted using RNAzol RT ( MRC ) . Library preparation was done using NEDNext Ultra Directional RNA Library Prep Kit ( NEB ) . Next generation sequencing was run according to MiSeq Illumina paired-end protocol at the Duke-NUS Genome Biology Facility . Illumina sequences were trimmed using Trimmomatic [59] . Alignments were done with Geneious software v . 9 . 0 . 5 . Sequences are available on GeneBank , accession SRX2617311-SRX2617315 . Three to five day-old female mosquitoes were sugar-deprived for 24 h and subsequently offered a blood meal containing a 40% volume of washed erythrocytes from SPF pig’s blood ( PWG Genetics ) , 5% of 100 mM ATP ( Thermo Scientific ) , 5% human serum ( Sigma ) and 50% volume of virus in RPMI media ( Gibco ) . The viral titer in the blood mix was 1x106 pfu/ml and was validated by plaque assay as previously described [30] . Mosquitoes were exposed to the artificial blood meal for one hour using a Hemotek membrane feeder system ( Discovery Workshops ) with a porcine intestine membrane . Fully engorged females were selected and provided access to a 10% sugar solution in an incubation chamber with conditions similar to insect rearing . Mosquitoes were analyzed at different time points depending on the experiment . Single mosquitoes were homogenized in 350μl of TRK lysis buffer ( E . Z . N . A . Total RNA kit I ( OMEGA Bio-Tek ) ) using a bead Mill homogenizer ( FastPrep-24 , MP Biomedicals ) . Total RNA was extracted using E . Z . N . A . Total RNA kit I ( OMEGA Bio-Tek ) and eluted in 30μl of DEPC-treated water . Genomic RNA ( gRNA ) was quantified with a one-step RT-qPCR with iTaq Universal probe kit ( Bio-Rad ) and primers and probes targeting the DENV-2 Envelope [60] . Subgenomic flaviviral RNA ( sfRNA ) was quantified with a one-step RT-qPCR with iTaq Universal Sybr kit ( Bio-Rad ) using primers targeting the beginning of the sfRNA sequence [29] . The 25 μl reaction mix contained 1 μM of forward and reverse primer , 0 . 125 μM of probe for gRNA only and 4μl of RNA extract . Quantification was conducted on a CFX96 Touch Real-Time PCR Detection System ( Bio-Rad ) . Thermal profile was 50°C for 10 min , 95°C for 1 min and 40 cycles of 95°C for 10 sec and 60°C for 15 sec . An absolute standard curve was generated by amplifying fragments containing the qPCR targets ( one fragment for each target ) using forward primers tagged with a T7 promoter; for gRNA we used 5'-CAGGATAAGAGGTTCGTCTG-3' and 5'-TTGACTCTTGTTTATCCGCT-3' , resulting in a 453bp fragment; for sfRNA we used 5'-AGAAGAGGAAGAGGCAGGA-3' and 5'-CATTGTTGCTGCGATTTGT-3' , resulting in a 319bp fragment . The fragments were reverse transcribed using MegaScript T7 transcription kit ( Ambion ) and purified using E . Z . N . A . Total RNA kit I . The total amount of RNA was quantified using a Nanodrop ( ThermoScientific ) to estimate copy number . Ten times serial dilutions were made and used to generate absolute standard equations for gRNA and sfRNA ( S4 Fig ) . In each subsequent RT-qPCR plate , quantification of four standard aliquot dilutions were used to adjust for threshold variation between plates . Results from gRNA quantification were used to calculate the infection rate , which was the number of infected samples divided by the number of analyzed samples , and the average of gRNA copies per organ , which was calculated only from infected samples . Results from sfRNA and gRNA quantification were used to calculate the sfRNA:gRNA ratio by using only infected samples . Individual mosquitoes or pairs of salivary glands were homogenized in 500μl of RPMI , filtered through 0 . 22 μm filter ( Sartorius ) and titered using plaque assay with BHK-21 ( ATCC CCL-10 ) cells as previously described [30] . Total RNA was harvested from 50 mosquitoes 10 days p . i . with PR6452 or PR315022 using RNAzol RT ( MRC ) . Northern Blot was conducted using NorthernMax kit ( Ambion ) . We used the previously designed primers [29] to generate a biotin-16-dUTP ( Roche ) labelled dsDNA probe that complements 3’-UTR sequence ( S1 Fig ) . Briefly , 32μg of total RNA mixed with loading dye was separated on a 1% agarose gel . The gel was then soaked in alkaline buffer ( NaOH 0 . 01N and NaCl 3M ) for 20 min and equilibrated 5 min in transfer buffer . Transfer to a nylon membrane ( Biodyne B ) was conducted by downward transfer as detailed in the NorthernMax kit protocol , with addition of a second bridge and buffer reservoir opposite to the first one , for 4h . After UV cross-linking , the membrane was subjected to two cycles of pre-hybridization with ULTRAhyb-oligo buffer ( Ambion ) at 42°C for 30min . 800ng of dsDNA labelled probe diluted in 10mM EDTA was denatured for 10 min at 95°C , immediately mixed with ULTRAhyb-oligo buffer and incubated overnight at 42°C on a rotating oven . Two low stringency wash at room temperature for 5 min and two high stringency wash at 50°C for 15 min were conducted . The membrane was blocked with 1% SDS Odyssey blocking buffer ( TBS ) ( Li-cor ) and stained at room temperature for 60 min with IRDYE 800cw streptavidin ( Li-cor ) . Picture was taken with Odyssey CLx imaging system ( Li-Cor ) and image intensity , used to quantify blots , was measured with Image Studio Lite ( Li-cor ) . Salivary glands , midguts and carcasses from 10 orally infected mosquitoes were dissected and total RNA was extracted using E . Z . N . A . Total RNA kit I . Genomic DNA was removed using RapidOut DNA Removal kit ( Thermo Scientific ) . Reverse transcription was performed with iScript cDNA Synthesis Kit ( Bio-Rad ) and gene expression was quantified using iTaq Universal SYBR Green Supermix ( Bio-Rad ) for Rel1a , Rel2 , Domeless , Vago , CecG , CecD , DefC , vir-1 and TEP22 with primers detailed in S7 Table . Actin expression was used for normalization and quantified using primers detailed in S7 Table . Quantification was conducted on a CFX96 Touch Real-Time PCR Detection System ( Bio-Rad ) . Thermal profile was 95°C for 1 min and 40 cycles of 95°C for 10 sec and 60°C for 15 sec . Six repeats were conducted . Orally infected mosquitoes were cold-anesthetized and severed from their wings and legs . Their proboscis was inserted into a 10μl filter tip containing 10μl of SPF blood . Mosquitoes were allowed to expectorate for 30 min . Total RNA was extracted from the blood and the mosquitoes and DENV gRNA copies quantified as detailed above . The analysis was conducted only on mosquitoes that could be visually identified as having consumed blood and that were infected following RT-qPCR analysis . Infection rate of saliva was calculated by dividing the number of DENV-positive saliva over total number of saliva . The effects of virus on log-transformed DENV gRNA copies per infected mosquito , pfu per infected mosquitoes , sfRNA:gRNA ratio and gene expression in different tissues were analyzed using one-way ANOVAs . For the kinetic experiment , the influence of tissue , day of collection and virus strain on log-transformed DENV gRNA copies per infected mosquito and sfRNA:gRNA ratio were analysed using three-way ANOVA . The effects of virus and tissue on XRN1 expression was tested using two-ANOVA . Post-hoc Tukey’s tests with Bonferroni adjustment were conducted when the effect was significant in the ANOVA test . Percentage differences were tested using a Z-test . In the saliva experiment , the effect of viruses on DENV gRNA copies per infected mosquitoes and per infected saliva were analyzed using Mann-Whitney’s non-parametric test . All tests were calculated using Systat 13 . 0 software ( SYSTAT ) . | Dengue is a re-emerging global disease transmitted from human-to-human by mosquitoes . While environmental and host immune factors are important , viral determinants of mosquito transmission also shape the epidemiology of dengue . Understanding how dengue viruses influence transmission will help identify isolates with high epidemic potential and untangle the evolutionary pressures at play in the dual-host cycle . Here , we identified 2 substitutions in the 3’UTR of epidemic isolates that increase transmission through immune suppression in the salivary glands . Using oral infection of Aedes aegypti mosquitoes , we reported that epidemic isolates produced more subgenomic flaviviral RNA ( sfRNA ) in salivary glands . SfRNA is generated from the 3’UTR sequence remaining after partial genome degradation by a host nuclease . Using reverse genetics , we identified the two 3’UTR substitutions responsible for the higher sfRNA quantity in salivary glands . We further showed that these substitutions increased dengue virus titer in salivary glands and rate of saliva infection , and suppressed the Toll immune response in salivary glands . Our study identifies the substitutions that determine virus epidemiological fitness and provides a mechanism for sfRNA-mediated enhancement of transmission . Together with previous work demonstrating that sfRNA sequence modification influences dengue virus pathogenicity in human , and that shows variation in sfRNA sequence when the viruses transition from one host to vector and vice versa , our study supports that sfRNA evolution is constrained in the two hosts . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [
"invertebrates",
"dengue",
"virus",
"medicine",
"and",
"health",
"sciences",
"body",
"fluids",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"immunology",
"rna",
"extraction",
"microbiology",
"saliva",
"animals",
"viruses",
"rna",
"viruses",
"insect",
"ve... | 2017 | Dengue subgenomic flaviviral RNA disrupts immunity in mosquito salivary glands to increase virus transmission |
New experimental results on bacterial growth inspire a novel top-down approach to study cell metabolism , combining mass balance and proteomic constraints to extend and complement Flux Balance Analysis . We introduce here Constrained Allocation Flux Balance Analysis , CAFBA , in which the biosynthetic costs associated to growth are accounted for in an effective way through a single additional genome-wide constraint . Its roots lie in the experimentally observed pattern of proteome allocation for metabolic functions , allowing to bridge regulation and metabolism in a transparent way under the principle of growth-rate maximization . We provide a simple method to solve CAFBA efficiently and propose an “ensemble averaging” procedure to account for unknown protein costs . Applying this approach to modeling E . coli metabolism , we find that , as the growth rate increases , CAFBA solutions cross over from respiratory , growth-yield maximizing states ( preferred at slow growth ) to fermentative states with carbon overflow ( preferred at fast growth ) . In addition , CAFBA allows for quantitatively accurate predictions on the rate of acetate excretion and growth yield based on only 3 parameters determined by empirical growth laws .
The coupling between the physiology of cell growth and cellular composition has been actively investigated since the 1940s . In exponentially growing bacteria , whose growth state is conveniently associated to a single parameter , namely their growth rate , such interdependence is best expressed in a quantitative way by the bacterial ‘growth laws’ that directly relate the protein , DNA and RNA content of a cell to the growth rate . Many such laws have been experimentally characterized [1–4] and many more are currently being probed at increasingly high resolution [5 , 6] . The emerging scenario suggests that proteome organization in bacteria is actively regulated in response to the growth conditions . Recent experiments have in particular provided validation to the picture according to which , as the growth rate changes , bacteria adjust the relative amounts of ribosome-affiliated , nutrient scavenging and metabolic proteins ( enzymes ) , so as to optimize their growth performance and energy production strategy [6–8] . At present , several phenomenological models explain the origin of different growth laws at a coarse-grained level [5 , 7] . In contrast , genome-scale approaches probing such relationships at molecular levels are less developed . Constraint-based models ( CBMs ) are powerful in silico tools that can be used to examine metabolic networks at genome scale . Starting from a non-equilibrium steady state assumption for metabolic fluxes , CBMs define the space of feasible reaction profiles through simple physico-chemical constraints like mass-balance . Once physiologically or thermodynamically motivated bounds of variability are assigned to fluxes , the solution space is essentially determined by the stoichiometry of the network alone . On the other hand , in genome-scale models stoichiometric constraints usually generate high-dimensional solution spaces in which physiologically relevant flux patterns may be hard to isolate . In many cases , optimal flux patterns can be defined through the maximization of specific objective functions . Flux Balance Analysis ( FBA ) [9–15] allows for instance to compute optimal flux configurations by means of linear programming ( LP ) , employing biomass production as a standard objective function [16] . This approach is widely used to describe microbial growth in lab conditions . It is clear that in order to capture the phenomenology of growth laws one needs to go beyond the basic elements of CBMs , and incorporate the costs associated with gene expression and protein synthesis into models of cellular metabolism . Resource Balance Analysis ( RBA ) [17 , 18] and ME-models [19 , 20] have taken important steps in this direction . These approaches propose a data-based optimization scheme to predict the growth-maximizing metabolic flux configurations under a variety of constraints , including stoichiometric mass-balance , ‘demand functions’ characterizing how the amounts of cellular components change with the growth rate , and specific prescriptions that relate fluxes to enzyme levels . The resulting schemes are more involved than FBA ( resulting in nonlinear optimization problems ) and require a large number of parameters . It is therefore important to devise a theoretical framework with the conceptual appeal and computational simplicity of FBA , in particular one that is more resilient to the choice of parameters and in which the interplay between metabolism and regulation is expressed through a more intuitive and transparent framework . In this work we present a generalized FBA scheme , called Constrained Allocation FBA or CAFBA , in which ( optimal ) regulation is accounted for effectively through a single additional global constraint on fluxes that encodes for the relative adjustment of proteome sectors at different growth rates . In a nutshell , the CAFBA-specific constraint describes the tug-of-war in the allocation of cellular resources across ribosomal , transport and biosynthetic proteins that has been observed in experiments . By imposing that the ribosomal share of the proteome behaves in accordance with empirically established growth laws [5 , 21 , 22] , CAFBA is able to reproduce observed behaviors without requiring parameter tuning . In addition , CAFBA generates a variety of testable predictions , including about the usage of metabolic pathways , despite lacking the level of biochemical detail that characterizes ME-models or RBA . Cellular strategies for energy production are the central focus of CAFBA . It is well known that fast-growing microorganisms tend to avoid using high-yield respiratory pathways to generate ATP even in the presence of oxygen , relying instead on aerobic fermentation [23–28] . The preference for low-yield pathways is manifested in the secretion of fermentation products like acetate for E . coli or ethanol for S . cerevisiæ [23 , 25 , 26 , 29] . This phenomenon , known as ‘overflow metabolism’ , is captured by standard FBA schemes at a qualitative level when additional capacity constraints on respiratory pathways [30] or density constraints for soluble [31 , 32] or membrane-bound [33] enzymes are included . However , certain quantitative aspects of potential interest for industrial applications , like the rate of metabolic overflow and the growth rate at which it occurs , have so far eluded comprehensive mechanistic models . By effectively modeling the trade-off between growth and its biosynthetic costs , CAFBA naturally produces cellular states with suboptimal growth yields , where carbon overflow is obtained with quantitative accuracy . This paper focuses on the scenario obtained by CAFBA for carbon-limited growth of E . coli . We find in particular that acetate secretion appears in E . coli at fast growth rates , whereas yield-maximizing FBA-like solutions dominate at slow growth rates . In spite of the nominal need for a large number of uncharacterized parameters in genome-wide models , CAFBA solutions remarkably depend only on a few global parameters . In particular , overflow metabolism is obtained consistently with quantitative accuracy , while all results are robust against 10-fold changes in the values of the enzymatic efficiency parameters . From a technical viewpoint , CAFBA effectively turns out to be an LP problem even when one accounts for growth-rate dependent biomass composition . This , together with its simple conceptual framework , makes CAFBA a very convenient scheme to analyze the interplay of metabolism and gene expression at genome scale .
Phenomenological studies of bacterial growth physiology suggest that the bacterial proteome is organized into “sectors” whose mass fractions adjust linearly with the growth rate in response to specific environmental and intracellular changes , including carbon limitation , anabolic limitation and translational inhibition [5 , 6 , 8] . Proteome organization and optimal growth constitute in essence an intertwined allocation problem , with the cell trying to optimally partition its proteome so as to maximize its growth performance . Based on empirical evidence on E . coli growth in carbon-limited media , CAFBA posits a 4-sector partitioning of the proteome in The corresponding proteome fractions ( denoted by ϕX for the X-sector ) should sum up to 1 , i . e . ϕ C + ϕ E + ϕ R + ϕ Q = 1 . ( 1 ) We shall now provide an explicit characterization of the different terms in the above sum . Putting the different terms together , the sum rule Eq ( 1 ) for proteome fractions can be recast as w C v C + ∑ i w i | v i | + w R λ = ϕ max , ( 11 ) where ϕmax = 1 − ϕQ − ϕC , 0 − ϕE , 0 − ϕR , 0 denotes the proteome fraction accessible to growth-rate dependent components of the protein sectors , which was estimated to be of the order of 50% for E . coli [5] . The linear constraint ( 11 ) encodes for the tug-of-war that ultimately determines optimal growth and proteome allocation , as depicted in Fig 1A: as λ increases , so does the proteome fraction of the R-sector , and the E- and C- sectors will concomitantly have to adjust their shares so as to satisfy Eq ( 11 ) , forcing in turn a remodeling of the underlying flux and nutrient intake patterns . Formally , the proteome allocation constraint ( 11 ) resembles the molecular crowding constraint defined in [31 , 32] , which essentially enforces a global upper bound on fluxes due to finite solvent capacity and was also adopted in RBA [17 , 18] . However , the intracellular density is empirically known to be ( roughly ) constant across different growth conditions [44] , suggesting that cells can adapt their volume to accommodate additional metabolites and macromolecules when necessary . In this respect , a hard constraint on solvent capacity is not fully justified . The CAFBA constraint ( 1 ) is instead derived from the normalization of protein fractions , due ultimately to the limited translational capacity of the ribosomes [5] . Note that the growth rate λ is explicitly involved in Eq ( 11 ) . Summing up , CAFBA is defined by the following optimization problem: max v λ subject to ( i ) ∑ i S μ i v i = 0 ∀ μ ( ii ) ℓ i ≤ v i ≤ u i ∀ i ( iii ) w C v C + ∑ i w i | v i | + w R λ = ϕ max ( 12 ) where Sμi stands for elements of the metabolic network’s stoichiometric matrix ( with μ indexing metabolites and i indexing reactions ) , ℓi and ui denote lower and upper bounds for the flux vi , while the value of λ is defined by the flux of the biomass reaction [16] .
We set wE so that the extrapolated maximal growth rate in unlimited carbon supply , corresponding to wC = 0 , is close to the value 1 . 1–1 . 2/h found in [6 , 41] . In the case of glucose as the sole carbon source , as well as for a number of other glycolytic carbon sources ( see Table A in S1 Text ) , the value wE = 8 . 3 × 10−4 gh/mmol turned out to yield λmax = 1/h . ( Slightly larger growth rates are obtained with phosphorylated carbon sources due to the fact that the extra energy carried by these carbon sources allow for a reduced flux in the E-sector . ) To capture the effects of changing the glucose level , we simply increased the value of wC from zero . For the sakes of completeness , the values of wC leading to empirically observed growth rates for E . coli growth on different carbon sources are reported in Table B in S1 Text . Fig 1 reports results obtained for growth on glucose with this choice of parameters , while results for growth on other carbon sources are shown in Fig A in S1 Text . One sees in Fig 1B that the growth-dependent fraction of C-proteins ( ΔϕC , blue line ) increases almost linearly with decreasing λ as the carbon concentration is limited , in line with the experimentally observed expressions of catabolic proteins [6 , 8] and PTS activity [20] . Both the proteome fractions of the E- and R-sectors ( ΔϕE and ΔϕR , yellow and green lines , respectively ) instead decrease linearly with growth rate . CAFBA therefore confirms the findings from a coarse grained model of proteome allocation [7]: in the optimal state , the cell invests more and more of its proteomic resources in intake systems as nutrient becomes limiting , while translational machinery and biosynthetic pathways are favored at high growth rates . Fig 1C displays the main fluxes of central carbon metabolism . The rate of acetate secretion and the flux through the Entner-Doudoroff pathway ( red and orange colors , respectively ) both drop fast as the growth rate decreases . Respiration , represented by the flux through the TCA cycle ( blue color ) is the predominant energy-producing pathway at small growth rates , while at high growth rates fermentation is preferred and acetate is secreted . Note that the acetate onset point is within 10% of the one observed experimentally for NCM3722 [29] roughly independently of the specific carbon source ( see Fig A in S1 Text ) —a remarkable result given the simplicity of the homogeneity assumption for the wi’s . Translational limitation [5] is modeled by increasing wR from the value of wR , 0 while keeping all other parameters fixed , including wC . In this case ( see Fig 1D ) , the ribosomal proteome fraction ( ΔϕR ) increases as translation is increasingly inhibited , while the other growth-dependent sectors ( ΔϕC and ΔϕE ) shrink almost linearly . Acetate secretion extends to the slowest growth rates in accordance with experimental findings [29] , while the respiratory flux ( see Fig 1E ) is negligible . It is interesting to compare CAFBA results with the phenomenological proteome allocation model introduced in [5] , which describes how proteome is allocated in different environments . There , the growth rate was predicted to be a Michaelis-Menten function of the “nutritional capacity” κn and “translation capacity” κt , independent phenomenological parameters that can be estimated from empirical growth laws . CAFBA recovers this result within a genome-scale model , with 1/wC playing the role of κn and 1/wR acting as κt ( see Fig B in S1 Text and Note D in S1 Text for a detailed discussion ) . It transpires from Fig 1C and Fig A in S1 Text that the optimal flux configurations in carbon limitation vary discontinuously with the growth rate . This is due to the fact that the control parameter is not a flux ( as in standard FBA ) , but , rather , the weight of the C-sector wC , which , as discussed above , is a proxy for either the external carbon concentration or the amount of glucose intake proteins [6] . Even though wC is varied continuously , growth-rate maximization can induce large rearrangements of the active pathways in response to small changes of the control parameter . This behavior is ultimately a mathematical feature due to the way in which the optimal solution in constraint-based models like CAFBA changes as one modifies wC . For the heterogeneous case , for each value of wC we generated 1000 models , each with a random set of weights wi independently drawn from the same probability density p ( w ) ∝ 1 / w , w min ≤ w ≤ w max , ( 13 ) which corresponds to a uniform density for the logarithm of w . p ( w ) is fully determined by its average 〈w〉 and width δ ≡ log10 ( wmax/wmin ) . We set 〈w〉 so that the average value of the maximum achievable growth rate λmax ( obtained for wC , 0 = 0 ) equals 1/h . This fixes 〈w〉 = 8 . 8 × 10−4 gh/mmol , a value that is remarkably close to wE = 8 . 3 × 10−4 gh/mmol as determined in the homogeneous case . δ was instead fixed to 1 , implying that the weights are assumed to span one order of magnitude ( results obtained for different values of δ are discussed in Note E in S1 Text ) . Each set of weights {wi} leads to a corresponding optimal flux pattern , growth rate , acetate secretion rate , etc . The distribution of growth rates obtained from many realizations of the weights is shown in Fig 2A . Note that in spite of the 10-fold variability of the weights , the growth rate remains within a modest range of ±20% . The distribution for acetate secretion rates is instead conveyed in Fig 2B: it is rather heterogeneous , with a marked peak for phenotypes with very low acetate secretion . While individual fluxes can fluctuate significantly across solutions , average fluxes are strikingly well-behaved . This phenomenon is illustrated in Fig 2C where we show a set of average fluxes plotted against the average growth rate . The average acetate secretion rate ( red symbols ) has an approximate linear dependence on the growth rate starting from λac ≃ 0 . 79/h . Average fluxes through TCA and the glyoxylate shunt ( blue up- and down- triangles ) reach their respective maxima close to λac . Notice that a smooth transition from a predominantly fermentative to a predominantly respiratory mode of energy production clearly emerges , in full agreement with empirical evidence . It is especially remarkable that this scenario does not seem to depend on the specific choice of p ( w ) . For instance , a log-normal distribution gives qualitatively similar results ( see Fig C in S1 Text ) . Despite the crude approximations , CAFBA solutions appear to reproduce experimental findings with surprising accuracy . Fig 3A shows how the average acetate excretion rate compares with data from different experiments [29 , 48–51] . Secretion rates from experiments using the MG1655 strain are consistent among each other ( open triangles ) , as are results obtained with the NCM3722 and ML308 strains ( open circles ) . CAFBA predictions are shown as solid circles for the two classes of strains . Data obtained with NCM3722 and ML308 were compared with CAFBA solutions obtained by setting λmax = 1/h and hence 〈w〉 = 8 . 8 × 10−4 gh/mmol . Instead , based on experimental evidence suggesting that MG1655 cells grow about 1/3 slower than the other two strains ( see Fig D in S1 Text ) , for MG1655 we set λmax = 0 . 67/h , leading to 〈w〉 = 1 . 55 × 10−3 gh/mmol . With this choice , CAFBA quantitatively reproduces the growth-rate dependence of acetate secretion . Growth yields , instead , are less consistent across different experiments and/or strains , see Fig 3B . Without any further parameter tuning , CAFBA solutions capture the growth yields for NCM3722 and MG1655 at a quantitative level , although they fail for ML308 . It should be noted that the differences in yield among experiments done on the same strain ( MG1655 ) suggest that other factors beyond the scope of this simple model might be at play , such as differences in growth conditions and/or maintenance requirements . We have also analyzed how the flux patterns of various intracellular pathways are modulated by the growth rate . Results for the central carbon pathways are summarized in Fig E in S1 Text , with the fluxes through the TCA cycle and glyoxylate shunt consistently increasing in proportion as glucose is limited . A similar behavior has been observed in measured expression levels of the corresponding enzymes [8 , 51] . Glycolytic fluxes are heterogeneously regulated , due to the interplay between the EMP pathway , the ED pathway and the switch between glyoxylate shunt and the phosphoenolpyruvate carboxylase ( PPC ) reaction . The redox balance of the cell appears to be affected , as described in Fig F in S1 Text . Indeed we find that NADP transhydrogenase switches on at high growth rates , oxidizing NADH and reducing NADP+ , in agreement with the different roles of the two transhydrogenases , UdhA and PntA , as quantified by transcription data [52] . Moreover we observe a switch between two separate ubiquinol oxidase reactions , characterized by different abilities to generate proton-motive force , in agreement with studies focused on the crowding of the cell’s membrane [33] . We further tested CAFBA’s ability to describe E . coli growth on carbon sources other than glucose . For illustration purposes , for each carbon source studied we have varied wC from zero to high values , so as to produce result in the entire range of growth rates 0 ≤ λ ≤ λmax , even though growth rates measured on individual carbon sources are always smaller than λmax due to non-zero values of wC , 0 ( see Table B in S1 Text ) . The typical behaviour of CAFBA solutions with different glycolytic carbon sources is remarkably consistent ( see Fig 4 ) . For each of the nutrients we tested , as the carbon supply becomes limiting , acetate excretion ( Fig 4A ) decreases almost linearly with growth rate , extrapolating to zero roughly at λac ≃ 0 . 79/h ( continuous black line ) . By contrast , fluxes through TCA and glyoxylate shunt ( Fig 4B and 4C ) rise linearly with decreasing growth rate at fast growth , reaching a maximum close to λac before decreasing at slower growth . The secretion rate of CO2 ( Fig 4D ) almost always diminishes as λ is reduced . For λ < λac the decrease is linear , while it is non-linear for λ > λac . Altogether , for all carbon sources , results point to two distinct types of behaviors arising , respectively , below and above λac . The Entner-Doudoroff ( ED ) pathway , an alternative to the Embden-Meyerhoff-Parnass ( EMP ) pathway , is used in E . coli for glucose catabolism at high growth rates [53 , 54] . CAFBA solutions reproduce this feature , relying on the ED pathway from medium ( λ ≃ 0 . 3/h ) to high growth rates as shown in Fig 4E . Interestingly , average fluxes are consistent for lactose , glucose and maltose on the one hand , and for fructose , sorbitol and mannose on the other . The reason is that , in the former group of substrates , the carbon source enters glycolysis as glucose-6P , which can be processed either by upper glycolysis or by the ED and pentose phosphate pathways . In the latter group , instead , carbon is transformed into fructose-6P , which is more conveniently processed into fructose biphosphate . A similar behavior is observed for phosphatated carbon sources or other substrates of the glycolytic or pentose pathways , see Fig G in S1 Text . The ED pathway , despite having a smaller ATP yield , requires a much smaller number of enzymes than the EMP pathway . Therefore , the use of ED over EMP may be the result of a proteome-saving strategy . Our findings thus agree with the conclusions of [42 , 54 , 55] . The switch between the EMP and ED pathways sets in at a growth rate close to 0 . 3/h , well below λac , suggesting that it is independent of acetate secretion . Nonetheless , both features appear in CAFBA in order to cope with increasingly expensive proteins , in agreement with quantitative proteomics data [46] . On the other hand , CAFBA shows that a variety of strategies exist for cells growing on carbon substrates belonging to the lower part of glycolysis or to the TCA cycle , see Fig H in S1 Text . What these strategies share is an increased production of CO2 at faster growth , and a vanishing activity of the ED pathway . The latter is of course due to the intrinsic glycolytic , as opposed to gluconeogenic , nature of the ED pathway . Standard FBA optimizes the growth yield subject to constraining the carbon intake flux . It is useful to compare CAFBA solutions with solutions obtained by FBA at the same growth rate and with glucose as the sole carbon source for both models . To do so , we have first solved Parsimonius Enzyme Usage FBA ( pFBA , see [56] ) varying the bounds on glucose intake so as to obtain FBA solutions as a function of the growth rate . We shall denote them as z ( λ ) = {zi ( λ ) } . CAFBA solutions found upon varying wC lead instead to wC-dependent mean growth rates λ ¯ ( w C ) . We shall denote CAFBA solutions obtained for a value of wC such that λ ¯ ( w C ) = λ by v ( λ ) = {vi ( λ ) } . We have then computed , for a given set R of reactions of interest , the similarity index q R called “overlap” and defined as [57] q R ( λ ) = 1 N R ∑ i ∈ R 2 v i ( λ ) z i ( λ ) v i ( λ ) 2 + z i ( λ ) 2 , ( 14 ) where the sum is restricted to reactions in R and the brackets 〈⋯〉 denote an average over 1000 different CAFBA solutions v ( λ ) . If in each solution vi = zi for each i ∈ R , then q R = 1 . Conversely , the more the two flux vectors differ , the smaller q gets . In particular , if in each solution vi = −zi for each i ∈ R , one finds q R = - 1 . Fig 5A shows the behavior of q R versus λ for different choices of R . When all reactions are accounted for , q is generally very large at low growth rates and decreases slowly as λ increases . When focusing on individual pathways , one sees that the overlap for TCA fluxes ( cyan ) drops abruptly above the acetate onset point λac ≃ 0 . 79/h , where the growth yield of CAFBA solutions starts to reduce significantly compared to that of FBA solutions ( Fig 5B , shown in red and blue symbols respectively ) . The overlap of fluxes in the glycolytic pathway instead diminishes with λ in a more gradual way , corresponding to the smooth increase in the activity of the ED pathway , see Fig 4E . Thus , as the growth rate increases , CAFBA solutions cross over from flux distributions that maximize the growth yield ( slow growth ) to a regime in which low-yield fermentation , accompanied by carbon overflow and energy spilling , is favored ( fast growth ) . In CBMs , the energetic cost of anabolic pathways is accounted for by the stoichiometry of the network . By contrast , the energetic requirements of growth ( e . g . protein synthesis ) and homeostasis must be included separately as an additional ATP hydrolysis flux vATP . In metabolic models , the latter is assumed to be linearly related to the growth rate , i . e . vATP = vATPM + βATP λ [15] . The first term is a growth-rate independent maintenance flux that represents the energy required to sustain basal cellular activities . The second term , instead , accounts for λ-dependence through a coefficient βATP that fixes the moles of ATP to be hydrolyzed per gram of dry weight . The values of vATPM and βATP are usually fitted from growth yield curves [16] , and different metabolic reconstructions of E . coli use different numerical values for both of them , see [45 , 58 , 59] and Table C in S1 Text . However , as the cell’s composition ( and specifically the amounts of RNA , DNA , proteins , fatty acids , etc . ) adjusts with the growth rate , biomass coefficients , including the demand of growth-related ATP , are in general λ-dependent [39] . A natural question to ask at this point is how cellular ATP requirements impact the shift between respiration and fermentation . Results obtained by solving CAFBA with λ-dependent biomass composition are shown in Fig 6 ( open symbols ) , together with the solution obtained for constant biomass composition at the same 〈w〉 = 8 . 8 × 10−4 gh/mmol ( filled blue circles ) . We tested CAFBA predictions with three different values of βATP while keeping vATPM fixed: ( i ) the default value for iJR904 model , ( ii ) the default value for the iAF1260 model , which is 30% larger than ( i ) [58] , and ( iii ) a value 30% smaller than ( i ) . One sees that , for the same ATP hydrolysis parameter ( open and filled blue symbols ) , solutions for the two versions of CAFBA nearly overlap . On the other hand , both the slope and the onset growth rate λac for acetate secretion appear to depend on the value of βATP . Likewise , the flux through TCA increases with βATP so as to satisfy energetic requirements . The growth yield and maximum growth rate λmax obtained at wC = 0 decrease accordingly . However , if we tune 〈w〉 to fix λmax = 1/h for each value of βATP , acetate secretion starts consistently at λac ≃ 0 . 8/h ( Fig J in S1 Text ) , implying that energetic costs do not affect the ratio λac/λmax .
Under carbon limitation , solutions of CAFBA are obtained by varying the parameter wC , a proxy for the extracellular carbon level representing the proteome cost of the C-sector ( carbon intake ) . In essence , for any given substrate level , CAFBA allocates the C-sector proteins per unit flux by simultaneously optimizing the allocation of the proteome fractions required to sustain biosynthesis and translation in order to maximize growth . The use of wC as a control parameter as opposed to directly dialing the nutrient intake flux is one of the elements that distinguish CAFBA from closely related CBMs like FBA [15] , FBAwMC [31 , 32] and ME-models [19 , 20] . In fact , the CAFBA constraint effectively reduces to a finite capacity constraint similar to the one that characterizes FBAwMC when an upper bound on the glucose intake flux is used to modulate growth at fixed wC = 0 ( see Note B in S1 Text ) . ( Note however that tuning nutrient levels as opposed to nutrient influx was employed in RBA to model the substitution between low affinity and high affinity cysteine transporters in B . subtilis [18] . ) Secondly , CAFBA does not provide the detailed mechanistic description of gene expression and protein synthesis conveyed by ME-models and RBA , whose definition includes , for instance , explicit variables for macromolecular concentrations ( ribosomes , DNA , RNA , etc . ) . Rather , it relies on an effective formulation based on empirical growth laws and ( when desired ) on a growth-rate dependent biomass composition . In this light , while less comprehensive than its closely related CBMs , the CAFBA scheme highlights the key biological ingredients constraining proteome allocation . On a more technical level , both RBA and ME-models are intrinsically non-linear and handle non-linearity by approximating their underlying optimization problems through sequences of linear feasibility problems . In CAFBA , even the worst case is solved through a fast iterative algorithm involving a small number of LP problems . Finally , the optimal proteome allocation problem posed by CAFBA can be seen as an assumption of “optimal enzymatic efficiency” , close to that underlying FBA approaches based on flux minimization [65 , 66] . One of the strong points of standard FBA consists in its reliance on the stoichiometric matrix and on thermodynamic reversibility constraints alone , making kinetic parameters unnecessary . CAFBA’s proteome allocation constraint in principle introduces a large number of additional parameters related to reaction and/or transport kinetics that , for the most part , are either uncharacterized or inferred from in vitro studies performed in different biochemical conditions [67] . This raises the issue of parameter selection . Two of the constants entering the proteome constraints ( namely wR and ϕmax ) are obtained directly from empirical growth laws . With wC acting as the control variable , the only free parameters left are the weights wi characterizing intracellular reactions . While quantitative CAFBA predictions appear to be dependent on their specific values , the qualitative behaviour of the solutions is not . Furthermore , the scenario obtained by averaging CAFBA solutions over different choices of the wi quantitatively reproduces experimental findings for acetate secretion and growth yield . These results point to a considerable degree of robustness of the CAFBA framework against fluctuations in parameter values . Notice however that the CAFBA picture can be further improved upon tuning the weights of individual reactions . For example , by increasing the average weight of reactions involved in respiration one sees a shift in the onset of acetate secretion and the value of λac/λmax changes , see Fig K in S1 Text . On the other hand , parameters can also be tuned according to empirical evidence so as to allow for a more thorough comparison with experiments performed on different strains and/or growth conditions , e . g . concerning intracellular fluxes ( see Fig L in S1 Text ) . In perspective , detailed flux measurements may allow to estimate typical weights for each pathway , and possibly even for individual enzymes , opening for the possibility to better calibrate the model and obtain completely quantitative predictions . Our work here has aimed at keeping the number of parameters as small as possible . In this light , many emerging features of the interplay between metabolism and gene expression appear to be mostly determined by the topology of the metabolic network . Elucidating the origin of this simplification is a foremost theoretical challenge for future studies of metabolic systems . The need to resort to an averaging procedure in order to reproduce bulk measurements for the growth yield and the acetate excretion forces us to ask whether the CAFBA averaging may have some further meaning . We offer here two possible scenarios . The first one is based on the fact that , even in well controlled growth conditions , cells in a population are normally heterogeneous , as transcription levels , protein abundances , reaction fluxes and instantaneous growth rates may change significantly from one to the other [68–70] . This would in turn reflect in fluctuations in the values of each wi across cells . Averaging over different choices for the weights could then simply be interpreted as averaging over a population of heterogeneous cells ( as would seem appropriate in modelling batch culture or chemostats ) . The alternative scenario presupposes that , even in absence of any cell-to-cell variability , cells may not be able to perfectly adjust fluxes to the distribution maximizing the instantaneous growth rate . This may occur for different reasons . First , the regulatory machinery needed to perform protein allocation requires by itself an investment of metabolic and proteomic resources [71–73] . This burden becomes more severe as the regulatory system gets more sensitive and fine-tuned , and , clearly , CAFBA does not account for it . Secondly , environments where cells grow are always fluctuating . Any regulatory machinery implementing fast adjustments in response to small environmental changes will necessarily come at a cost that will negatively affect the growth rate . Under such constraints , regulatory programs selected over evolutionary time scales may prefer to maximize an average growth rate , the average being taken over life process history . The actual regulatory programs implemented would then balance the trade-off between the costs of not being exactly in the instantaneously optimal growth state and the costs of adjusting regulation too frequently in “natural” conditions ( not those provided in the laboratory ) . An interpretation of the CAFBA averaging prescription would then be that it is a way to implement an “average” strategy that smooths the output upon variations of the environmental conditions . In both of the above scenarios , CAFBA points to the emergence of acetate excretion as triggered by regulatory system ( s ) sensing the abundance of the carbon source and the balance of biomass synthesis and energy generation [29] . We note that carbon overflow in E . coli has been proposed to be modulated by catabolite repression mediated ACS down-regulation [49] . Discriminating between the two scenarios we have just presented could be achieved already in bulk experiments , by changing the weight of specific enzymes over time ( e . g . , by expressing useless proteins specific to certain pathways ) and monitoring whether the associated fluxes adjust dynamically in real time . Naturally , tests of cell-to-cell heterogeneities would also allow to favor one scenario over the other . Finally , we address the magnitude of fluctuations of the weights wi . The spread in the enzyme catalytic rates kcat , i , as tabulated in databases , is notoriously broad , exceeding 3 orders of magnitude [32 , 67] . In the absence of more refined information , it is reasonable to expect that the weights wi should fluctuate by about the same amount . However , we have seen that unrealistic results are generated by CAFBA if weights are allowed to fluctuate more than 10-fold . Therefore , either the true width of the distribution of the weights wi is much smaller than what is suggested by the values of kcat , i estimated in vitro , or weights are subtly distributed across pathways in such a way that strong compensatory effects occur that reduce fluctuations . With steady improvements in proteomic methods , it may soon be possible to quantitatively determine these parameters empirically and elucidate this puzzle .
Given a metabolic network encoded by a stoichiometric matrix S = {Sμi} , CAFBA is stated in the case of carbon limitation as max v λ subject to ( i ) ∑ i S μ i v i = 0 ∀ μ ( 15 ) ( ii ) ℓ i ≤ v i ≤ u i ∀ i ( 16 ) ( iii ) w C v C + ∑ i w i | v i | + w R λ = ϕ max , ( 17 ) where λ denotes the growth rate , v = {vi} is a flux vector , and ( ℓi , ui ) represent lower and upper bounds for each flux vi , respectively . Condition ( 17 ) corresponds to the proteome allocation constraint ϕC + ϕE + ϕR + ϕQ = 1 , with vC ≥ 0 being the active glucose intake flux and with the sum in ∑i wi|vi| running over all enzyme–catalyzed reactions except for transports , exchanges and carbon intake pathways . The biomass flux λ and ATP maintenance reaction are also excluded from Eq ( 17 ) . In principle , CAFBA is a MILP ( Mixed Integer-Linear Programming ) problem due to the presence of absolute values in Eq ( 17 ) . However , in CAFBA they can be disposed of by splitting each flux vi into a forward v i + and a backward v i - component , both non-negative . Note that if either v i + or v i - can be set to zero for each i , net fluxes v i = v i + - v i - are univocally determined , one has | v i | = v i + + v i - for absolute values , and CAFBA becomes equivalent to max v + , v - λ subject to ( i ) ∑ i S μ i ( v i + - v i - ) = 0 ∀ μ ( 18 ) ( ii ) 0 ≤ v i - ≤ - ℓ i 0 ≤ v i + ≤ u i ∀ i ( 19 ) ( iii ) w C v C + ∑ i w i ( v i + + v i - ) + w R λ = ϕ max , ( 20 ) which is a simple LP problem rather than a MILP . The key observation is that , as long as λ is maximized , CAFBA actually adjusts fluxes so that either the forward or the backward component vanish for each i . Indeed , a necessary condition for maximizing λ is that the quantity v i + + v i - is minimized for each i , which , at fixed v i = v i + - v i - , is achieved by setting either v i + or v i - to zero . Therefore CAFBA reduces from a MILP to a LP problem . Note that , because of the tight link between CAFBA and flux minimization , degeneracies in CAFBA solutions can only arise from the presence of ( a ) futile loops or ( b ) pathways that perform the same overall chemical conversion with the same flux at the same proteome cost , and which therefore can be used alternatively . In CAFBA with heterogeneous weights , however , the chance that two equivalent flux configurations have exactly the same total weight is negligible , since weights are i . i . d . random variables . On the other hand , futile loops only concern transports that do not involve the main carbon source and therefore are not included explicitly in the CAFBA constraint . These loops however do not affect other fluxes and are easily spotted and removed , either by manually shutting off redundant processes or by including them into the proteome allocation constraint with an arbitrarily small but non-zero weight . Therefore , each instance of the inhomogeneous CAFBA scheme has a unique solution almost surely . Because our main results are obtained in this framework , alternate optima are in practice not an issue in CAFBA . We implemented CAFBA on the E . coli iJR904 genome-scale model [45] , comprehensive of 761 metabolites and 1075 reactions , as a Matlab function , using the COBRA Toolbox [74] to load the network reconstruction with a minor modification . Specifically , we shut off the glucose dehydrogenase reaction , since it is only functional if the cofactor pyrroloquinoline quinone is supplied in the environment ( see the Ecocyc [75] entry on the enzyme ) . Both GLPK- and Gurobi-compatible CAFBA solvers for Matlab are provided as S1 Code , along with a small set of utility functions . Running times for a single CAFBA optimization of the iJR904 network with a common laptop ( single thread of an Intel Core i7–2630QM CPU @ 2 . 00GHz ) are around 0 . 12 s for the GLPK ( version 4 . 47 ) LP solver and 0 . 05 s the Gurobi Optimizer ( version 5 . 6 ) solver . For comparison , the time required to compute the standard FBA solution for the same network with the COBRA toolbox using GLPK is around 0 . 06 s . The fact that cells adapt their composition with the growth rate [2 , 4 , 5] implies that biomass composition is itself λ-dependent . Growth-rate dependent biomass coefficients ( see e . g . [39] for E . coli ) indeed reflect empirical knowledge of how the amounts of RNA , DNA , proteins , fatty acids , etc . are modulated by λ . While constraint-based models such as FBA and CAFBA with growth-dependent biomass are non-linear , approximate solutions can be obtained efficiently by simple iterated LP protocols as follows: ( a ) starting from a given biomass vector , solve the model by optimizing the growth rate; ( b ) update the biomass composition to the computed optimal growth rate using the prescribed set of λ-dependent biomass coefficients; ( c ) iterate until a solution is reached , such that further iterations do not change the optimal growth rate within a desired precision . For CAFBA , this procedure typically converges in a very small number of iterations ( see Fig I in S1 Text ) . Further details about the case of growth rate-dependent biomass composition and the iterative algorithm for computing the optimal FBA or CAFBA solutions are given in Note F in S1 Text . Besides the full study of the E . coli iJR904 model , we have tested CAFBA on the more recent reconstructions iAF1260 [58] and iJO1366 [59] , obtaining very similar results . COBRA-compatible Matlab functions [74] to run CAFBA on these models are provided as S1 Code . Note C in S1 Text describes in detail how to port CAFBA to different growth media , nutrient limitations and/or bacterial species . However , provided the input coming from empirical growth laws is available together with the network structure and the biomass composition , the CAFBA framework can in principle be extended to growth-maximizing organisms other than bacteria . | The intracellular protein levels of exponentially growing bacteria are known to vary strongly with growth conditions , as described by quantitative “growth laws” . This work introduces a computational genome-scale framework ( Constrained Allocation Flux Balance Analysis , CAFBA ) which incorporates growth laws into canonical Flux Balance Analysis . Upon introducing 3 parameters based on established growth laws for E . coli , CAFBA accurately reproduces empirical results on the growth-rate dependent rate of carbon overflow and growth yield , and generates testable predictions about cellular energetic strategies and protein expression levels . CAFBA therefore provides a simple , quantitative approach to balancing the trade-off between growth and its associated biosynthetic costs at genome-scale , without the burden of tuning many inaccessible parameters . | [
"Abstract",
"Introduction",
"Model",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"cell",
"physiology",
"medicine",
"and",
"health",
"sciences",
"protein",
"metabolism",
"chemical",
"compounds",
"enzymes",
"enzymology",
"carbohydrates",
"cell",
"metabolism",
"organic",
"compounds",
"glucose",
"physiological",
"processes",
"enzyme",
"metabolism",
"enzym... | 2016 | Constrained Allocation Flux Balance Analysis |
Aside from malaria , infectious diseases are an important cause of death in sub-Saharan Africa and continue to pose major public health problems in African countries , notably pneumonia . Streptococcus pneumoniae remains the most common bacterial cause of pneumonia in all age groups . The skin is one of the main infection sites followed by the oropharynx . The skin carriage of certain pathogenic bacteria such as S . pneumoniae is often ignored or under-diagnosed . Finally , the mode of transmission of these infections remains uncertain . Here , we hypothesized that skin could play a role in the transmission of these infections . We collected 649 cotton swabs from a healthy population in Dielmo and Ndiop , rural Senegal . The sampling was carried out on the palm of the hands . After DNA extraction and actin control , qPCR targeting eight different bacteria was performed on 614 skin samples . We detected Streptococcus pneumoniae in 33 . 06% ( 203/614 ) , Staphylococcus aureus in 18 . 08% ( 111/614 ) and Streptococcus pyogenes in 1 . 95% ( 12/614 ) of samples . A skin S . pneumoniae carriage was detected in more than a third of a rural population in rural Africa , highlighting the need to develop hand disinfection programs in order to reduce the burden of infections .
Infectious diseases are the most important cause of death in sub-Saharan Africa [1 , 2] and continue to pose major public health problems in African countries . Globally and collectively they account for 20% of the mortality in all age groups ( and 33% of the mortality in the least developed countries ) and 50% of infant mortality [3] . A study performed in Karachi , Pakistan found that 41% of deaths of children under 5 year were due to diarrhea and 15% to acute respiratory infections which include pneumonia [4] . The pathogenic role of Streptococcus pneumoniae in pneumonia , otitis media , bacteremia and meningitis is undisputed . However , its isolation on the skin is an unusual discovery with a difficult clinical interpretation [5] which can range from simple colonization in immunocompetent hosts to severe infection in patients with different underlying conditions [5–6] . In a study performed in 2014 , Fenollar et al . reported that some bacteria that cause fever in Africa such as Staphylococcus aureus , Streptococcus pyogenes and Streptococcus pneumoniae are neglected in Senegal [7] . The monitoring of the carriage of these bacteria is important for several reasons . First , colonization in healthy individuals is a prerequisite for developing invasive and non-invasive diseases , and reduced colonization has been correlated with decreased pneumococcal and staphylococcal infection rates [8–9] . Second , healthy carriers serve as reservoirs for S . aureus and S . pneumoniae transmission to others in the community and in the hospital [10–11] . S . pneumoniae is one of the major pathogens infecting humans worldwide and is the most common cause of community-acquired bacterial pneumonia and otitis media , but can also give rise to severe cases of meningitis and sepsis [12] . Approximately 1 . 6 million people die each year from pneumococcal diseases [12] . The most frequently bacteria isolated in acute respiratory infections are S . pneumoniae and Haemophilus influenzae , which can occur secondarily following primary infection due to viral pathogens . Despite causing severe diseases , the asymptomatic carriage of S . pneumoniae in the nose , nasopharynx and throat was also reported . The isolation rates of S . pneumoniae obtained by nasal and nasopharyngeal ( NP ) sampling are similar in children , but higher than that of oropharyngeal sampling [13 , 14] . Its prevalence in nasopharyngeal samples varies from 7 to 99% and depends on the age , health , and socioeconomic status of the study population [15] . In Senegal , few studies on the viral and bacterial etiology of respiratory tract infections are available in pediatric settings [16] . Most of the publications on the prevalence of S . pneumoniae in NP samples concerned young children . Indeed , in 2003 , Echave et al . reported a prevalence rate of 56% in Senegal [17] . However , Baylet et al . , in 1983 , reported higher NP carriage prevalence rates in rural ( 77% ) and urban ( 69% ) areas in Senegal and other African countries [18] . The study conducted by Mediannikov et al . in 2014 , in Senegal revealed a skin carriage of S . aureus ( 21 . 7% % ) , S . pneumoniae ( 5% ) and S . pyogenes ( 5% ) on the skin of healthy people leaving in rural areas [19] . Here , we studied the skin carriage of major pathogenic bacteria such as S . pneumoniae , S . aureus and S . pyogenes in the populations of Dielmo and Ndiop , two rural villages . In an ancillary study , we tested the skin carriage of R . felis , B . crocidurae , T . whipplei , B . quintana and C . burnetii [19–22] which have been described as causes of fever in this rural area .
Before inclusion in the study , written informed consent was required from all adult participants ( ≥ 18 years ) and from parents or legal guardians of minors ( ≤ 18 years ) . An information document that clearly explains the risks and benefits associated with the participation to the study was handed over to each patient . This sheet states the reasons and the purpose of the sampling in the presence of a parent / guardian or guarantor . The consent form stipulates that samples will not be used in the future in other studies not related with the present work without preliminary agreement of the Senegalese national Ethics Committee . Consent was obtained from each individual , and the study was approved by the national Ethics Committee of Senegal ( N° 53 / MSAS / DPRS / CNERS du 31 mars 2015 ) . Dielmo and Ndiop are two villages located about 280 km Southeast of Dakar , near the Gambian border in an area of Sudan-type savannah . About 700 inhabitants have been included in an epidemiological study of malaria since 1990 and monitored daily for fever and illness; the detection of cases is both active and passive . The geographical and epidemiological characteristics of the Dielmo village have been previously described in detail elsewhere [23 , 24] , most of the houses are built in the traditional style with mud walls and thatched roofs . The main source of drinking water of the population is underground water . The study population consisted of people residing in the two villages , participating in the epidemiological follow-up of Dielmo and Ndiop , adhering to the principle of the project and having given their consent to participate in the study and thus provide a swab . Sampling was carried out on the palm of the hands ( the right and left palm of unwashed hands ) of a healthy population , all categories of age were included . It was performed in January in order to make an inventory of the skin carriage of the targeted pathogens . In addition , the sampling was done at the same location in each study participant . Swabs are performed on the hands , after moistening the swab with sterile physiological serum . The swabs obtained are immersed in an individual tube containing 600 μl of 1X Phosphate Buffer Saline ( PBS , OXOID LIMITED , HAMPSHIRE , ENGLAND ) . Once impregnated , the swab is pressed against the edges of the tube to release the sample , then 200 μl of the swab suspension was taken for DNA extraction using the CTAB 2% method . To extract the DNA , 200μl of bacterial suspension from the swab was mixed with 180μl of 2% Cetyl Trimethyl Ammonium Bromide ( CTAB ) [25] . The mixture was incubated in a water bath at 65°C for 1 hour . Two hundred microliters of chloroform were added to the mixture , and the supernatant was recovered after centrifugation at 12 , 000 rpm for 5 min . The nucleic acids were precipitated by 200 μl of isopropanol after15 min of centrifugation at 12 , 000 rpm . The pellet was then dried in a speed vac for 3–4 min and resuspended in 200 μl RNase-free water . The DNA solution was stored in the refrigerator at 8°C until further use and the PCR was done 24 hours after DNA extraction , as DNA cannot deteriorate after only 24 hours of storage . After the PCR , the DNA was conserved at -20°C . Except for Bartonella , bacteria were detected using a first intent PCR , when a specimen was tested positive in the first assay , the result was confirmed by a second quantitative PCR . A positive sample was defined as 2 positive quantitative PCR results in assays targeting 2 different repeated DNA sequences . We performed all PCR reactions in a CFX 96 thermal cycler ( Biorad ) . Each reaction was performed at a final volume of 20 μl , containing , 10μl of polymerase TAKYON , 1μl of each primer , 1μl of probe , 2μl of RNase-free water and 5μl DNA . A positive and negative control was included in each experiment . DNA extracted from the swabbing of a healthy person in Dakar was used as negative control . The positive control consists of a suspension made from a swabbing of healthy person , in which bacterial cultures were added . The strains used as positive control are available on the « Collection de Souches de l'Unité des Rickettsies ( CSUR , WDCM 875 ) » under the accession number: S . pneumoniae CSURP5700 , S . aureus CSURP2200 and S . pyogenes CSURP6897 . For each species , about ten colonies were suspended in 200μl of PBS and DNA was extracted as mentioned above . Molecular identification by qPCR involved pathogenic bacteria such as S . pneumoniae , S . aureus , R . felis , B . crocidurae , T . whipplei , B . quintana , S . pyogenes and C . burnetii ( Table 1 ) . The quality of the extraction was measured by actin ( the b-actin gene amplification by quantitative PCR confirmed the quality of the extracted DNA ) . Any sample with a Ct number ( cycle threshold ) that did not exceed 35 was considered positive . This number corresponds to the ability to reveal 10–20 copies of bacterial DNA [20 , 21] . Data were analyzed using Open Epi , version 3 . 4 . 1 ( Centers for Disease Control and Prevention , Atlanta , GA , USA ) . Non-parametric values were compared using a X2 test . Statistical significance was defined as p < 0 . 05 .
We detected S . pneumoniae in 33 . 06% ( 203/614 ) individuals , S . aureus in 18 . 08% ( 111/614 ) individuals and S . pyogenes in 1 . 95% ( 12/614 ) individuals . We also detected C . burnetii in 13 . 35% ( 82/614 ) individuals , B . crocidurae in 3 . 42% ( 21/614 ) individuals . However , we didn’t find any case of T . whipplei , B . quintana or R . felis ( Table 2 ) . We observed that S . pneumoniae and S . aureus were the two predominantly isolated bacteria out of eight targeted bacteria , and tried to see their incidence according to age and sex . In Dielmo and Ndiop S . pneumoniae DNA was detected in all age groups with percentages greater than 20% . S . pneumoniae DNA was mostly detected in the youngest age groups of 0–5 and 5–10 years with respectively 39 . 40% ( 47/119 ) and 47 . 93% ( 58/121 ) . The lowest rates , 17 . 02% ( 8/47 ) was obtained in the age groups of 45–60 years . S . aureus was detected mostly in the 5–10 , 10–15 and 15–30 years age groups with 19 . 83% ( 24/121 ) , 22 . 89% ( 19/83 ) and 19 . 55% ( 26/133 ) , respectively , and in a minority in the +60-years groups , with 8 . 82% ( 3/34 ) . Significant difference was noticed only in the 0–5 and 5–10 years age groups ( Table 3 ) . Studies on these samples from 614 villagers in Dielmo and Ndiop showed a skin carriage in 46 . 74% ( 115 /246 ) of men and 23 . 91% ( 88/368 ) of women for S . pneumoniae , in 14 . 22% ( 35/246 ) of men and 20 . 65% ( 76/368 ) of women for S . aureus . Statistical analysis showed a significant difference between men and women for S . pneumoniae ( p <0 . 0000001 ) but for S . aureus no significant difference between men and women was observed ( p = 0 . 05485 ) ( Table 4 ) .
In our study , we attempted to identify the prevalence of skin carriage of specific bacteria in a generally healthy rural population in Senegal . We are confident of our results because the validity of the data reported in this study is based on strict experimental procedures and positive and negative controls . The sampling was correct because 95% of samples were actin positive . Molecular analysis were carried out in two villages Dielmo and Ndiop . The most common pathogens detected were S . pneumoniae and S . aureus . They represented 51 . 14% of the pathogenic bacteria identified on the skin . To our knowledge , this study is the first attempt to investigate the presence of S . pneumoniae and S . aureus in the skin of asymptomatic peoples in Africa . We are also considering whether there is a link between identified pathogens and skin infections; pneumonia and respiratory infections . It is important to note that the differences in bacterial diversity on the skin varies from one site to another at the inter and intrapersonal level , for example , the bacterial diversity of the forehead is lower than the diversity of the palm in each person , and this is also true for the forehead versus forearm communities [30] . For our study , we found it more interesting to swab the palms of the hands because we believe that it is the most likely means of bacterial transmission , through salutation , food , etc . Next to S . pneumoniae and S . aureus we detected the DNA of B . crocidurae and C . burnetii , but their known mode of transmission is not through the palms of the hands . S . pneumoniae was the most prevalent bacteria detected on the on skin . This bacterium is one of the leading causes of pneumonia in children under five years of age in Senegal [31] . S . aureus was the second most prevalent bacteria detected . The results obtained on the prevalence of this bacterium are in line with those of Mediannikov et al . , 2014 which detected 21 . 7% ( 13/60 ) of S . aureus DNA on the forearm of asymptomatic populations residing in the same area [19] . We detected also S . pyogenes in the skin swabs with a smaller proportion . This rate was lower than that previously reported by Mediannikov et al . , 2014 , which showed the presence of this bacterium on the forearm of healthy populations ( intact skin swabs ) , with a prevalence of 5% ( 3/60 ) on the same study area [19] . This difference could be explained by seasonal variation , being more common in dry than wet seasons in monsoonal climates [32] . Crowding and poor hygiene therefore increase the chance of the transmission of S . pyogenes . Also , skin infections are more frequent and are a more important cause of morbidity in overcrowded communities with poor sanitation [33] . In addition , variations in the prevalence of S . pyogenes skin infections are related to accessibility to appropriate housing and hygiene . The absence of previous work on the exploration of the skin microbiota in Africa , more specifically in sub-Saharan Africa , is a limitation for our study because these results cannot be compared to those of other countries with different climatic and environmental conditions . Most of the publications on the carriage of S . pneumoniae were done on nasopharyngeal swabs and often concerned young children [34] . The highest rates of S . pneumoniae and S . aureus were detected in the youngest age groups . S . pneumoniae was observed in the age groups 0–5 and 5–10 years . The lowest prevalence was found in age groups 45–60 years . These results could be explained by the presence of these bacteria in the environment , and those children who would be much more exposed because they take less care of their hygiene compared to adults who would be much more exposed because they take less care of their hygiene compared to adults . These results are consistent with data from a previous study that found S . pneumoniae DNA in 22% ( 8/36 ) of skin swabs from the forearm of children in the age group 0–6 years [35] . S . aureus was detected mainly in the three youngest age groups 5–10 , 10–15 and 15–30 years , and in a minority in the +60-years . Just like S . pneumoniae , S aureus carriage was affected by age ( peak prevalence at youngest age groups ) . A relationship could be made between S . pneumoniae infections in these villagers , a previous study on influenza like illnesses ( ILI ) had found that the incidence rates differed significantly between age groups , and were highest in the [6–24 month ) and [0–6 month ) age groups [36] . These results are comparable to ours . Children under five years of age have a higher incidence of S . pneumoniae . Finally , in these villages we have set up a field laboratory for the diagnosis of infectious diseases using the molecular biology method [22] . This technical device allows the rapid diagnosis and monitoring of infectious diseases for which laboratory analyses were generally too late to guide therapy [22] . The results show that S . pneumoniae infections are mainly localized in children under five years of age . S . pneumoniae causes morbidity and mortality in young children , the elderly and immunodeficient patients [37] , but asymptomatic carriage is more common in children . Most publications on S . pneumoniae vaccine research target the youngest age groups . Children are considered an important vector for the spread of this microorganism in the community , and preventing the carriage of pneumococci could therefore reduce the prevalence of infections . Pneumococcal conjugate vaccination protects young children against invasive diseases with S . pneumoniae [38] . Researchers have showed that the carriage rate of S . pneumoniae is low in adults compared with children because the prevalence rate , risk factors for carrying and factors promoting the spread of the organism are limited in adults [39] . The main result of our molecular analysis is the carriage of S . pneumoniae on the skin in asymptomatic people . S . pneumoniae is found all over the world . The incidence of infection is higher in children under 2 years of age and adults over 60 years of age [40] . It belongs to the family Streptococcaceae . It is a Gram positive bacterium . There are about 90 serotypes; the capsule surrounding the pneumococcus is the main virulence factor [41] . This bacterium generally colonizes the mucosal surfaces of the nasopharynx and upper respiratory tract , and symptoms of inflammation appear when the bacterium migrates to the sterile parts of the respiratory tract [42] . It is transmitted by infectious cells that can be spread by aerosolized microdroplets sprayed during coughing or sneezing , or by oral contact from one person to another [41] . The nasopharynx is the only documented niche for S . pneumoniae in humans , many researchers have speculated on skin colonization following reports of pneumococcal skin and soft tissue infections in adults and children in the absence of prior systemic disease [5 , 6 , 43] . However , the rather high detection rates of S . pneumoniae in skin samples in this study suggest a possible existence of a true reservoir of this pathogen on the skin . The carriage rates of S . pneumoniae reported in previous studies strongly depended on the social , demographic and medical risk factors of the study subjects , as well as the methodological variations in the methodology used . Finally , the sampling sites may vary from one study to another [38] . We have not attempted to associate our prevalence rates of carriage with specific characteristics of the population . In this study , we used qPCR testing to determine the overall prevalence of major pathogens , which had been previously isolated in other types of samples and in the same area , in febrile or healthy subjects [19] . For S . pneumoniae , the highest rate of nasopharyngeal colonization has been shown to occur at an early stage of life [11–19] . This corresponds to the low number of S . pneumonia carriers in our study , as only adult individuals were sampled . In this study , we found that 46 . 74% ( 115 /246 ) of men and 23 . 91% ( 88/368 ) of women carried S . pneumoniae in their skin samples and statistical analysis showed a significant difference between men and women for the carriage of S . pneumoniae ( p <0 . 0000001 ) . Our findings are in line with a previously published study on the carriage of S . pneumoniae among older adults in Indonesia [44] . In addition , this is the first time , to our knowledge , that S . pneumoniae has been detected in the palm of hands in Senegal . We found that men and women carried S . aureus , and statistical analysis showed no significant difference between men and women for the carriage of S . aureus ( p = 0 . 05485 ) . Our results differ from those of two previous studies that showed that S . aureus carriage varies according to the sex and is higher in men [45 , 46] . To our knowledge , only viral respiratory infections have been studied in Dielmo and Ndiop . Available information shows that respiratory infections due to influenza ( flu ) viruses are more frequent [47] . From 2012 to 2013 , the overall flu incidence density rate was 19 . 2 per 100 person-years . The flu incidence density rates were significantly different between age groups , the highest being in the [6–24 months ) age group ( 30 . 3 to 50 . 7 per 100 person years ) [36] . PCR-based techniques suffer from possible biases due to the state of the bacteria ( dead bacteria ) [43 , 48] . According to Anna Engelbrektson and al . , 2010 , molecular-based approaches do not distinguish between living bacteria and dead bacteria , so this can lead to the detection of an excessive number of pathogens by qPCR [43 , 48] . The culture of S . pneumoniae from skin swabs that we made in parallel with the PCR would confirm more accurately the real existence of this bacterium on the skin .
In our study , we tried to demonstrate the existence of the target pathogens on the skin of people in our generally healthy study population in two villages of rural Senegal and to evaluate their impact in our two study villages using qPCR . Molecular analysis in Dielmo and Ndiop showed a high prevalence of S . aureus and S . pneumoniae carriage , especially among the youngest age groups . Our results suggest that random samples of skin swabs may contain S . aureus and S . pneumoniae . In addition , in asymptomatic subjects , we can detect the presence of certain pathogens by qPCR . In rural areas , the economic context and daily activities make this part of the population particularly vulnerable to infectious diseases . These populations live in poverty , the majority of whom are farmers and livestock breeders . As a result , they are less involved in their personal hygiene and food and household hygiene , which would lead to a considerable increase in the prevalence of infectious diseases . Fortunately , previous studies and WHO recommendations showed that body hygiene and more specifically hand hygiene could lead to a significant reduction in the prevalence of these diseases [44 , 49] . It appears necessary to undertake a so-called ‘‘soap project” study in Dielmo and Ndiop villages in order to evaluate the effectiveness of body hygiene in the prevention of infectious diseases . | Infectious diseases are one of the leading causes of morbidity and mortality in the world . They kill nearly 17 million people worldwide each year , mainly in developing countries . They are transmitted through four main channels: air , oral , parenteral and contact . The prevention of infectious diseases requires an understanding of the population's way of life and the knowledge of pathogenic microorganisms in circulation . Many diseases are transmitted through contact with soiled hands . This study allowed us to explore the pathogenic bacteria carriage on the skin in a rural population , following skin swabs made on the palms of the hands . Previous studies have shown that hands play an important role in the transmission and the prevention of infectious diseases such as diarrhea and pneumonia . One of the main results is the detection in high proportion of pathogenic bacteria such as Staphylococcus aureus and especially Streptococcus pneumoniae . Life in rural areas exposes the population to a lot of pathogenic microorganisms . These results could be used to implement prevention strategies against certain infectious diseases , by raising awareness of the importance of body hygiene and specifically hands hygiene , in order to improve the health of the population . | [
"Abstract",
"Introduction",
"Materials",
"and",
"method",
"Results",
"Discussion",
"Conclusion"
] | [
"dermatology",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"pneumococcus",
"pathogens",
"microbiology",
"staphylococcus",
"aureus",
"age",
"groups",
"streptococcus",
"pyogenes",
"skin",
"infections",
"body",
"limbs",
"bacteria",
... | 2018 | Asymptomatic carriage of Streptococcus pneumoniae detected by qPCR on the palm of hands of populations in rural Senegal |
The earliest stages of animal development are controlled by maternally deposited mRNA transcripts and proteins . Once the zygote is able to transcribe its own genome , maternal transcripts are degraded , in a tightly regulated process known as the maternal to zygotic transition ( MZT ) . While this process has been well-studied within model species , we have little knowledge of how the pools of maternal and zygotic transcripts evolve . To characterize the evolutionary dynamics and functional constraints on early embryonic expression , we created a transcriptomic dataset for 14 Drosophila species spanning over 50 million years of evolution , at developmental stages before and after the MZT , and compared our results with a previously published Aedes aegypti developmental time course . We found deep conservation over 250 million years of a core set of genes transcribed only by the zygote . This select group is highly enriched in transcription factors that play critical roles in early development . However , we also identify a surprisingly high level of change in the transcripts represented at both stages over the phylogeny . While mRNA levels of genes with maternally deposited transcripts are more highly conserved than zygotic genes , those maternal transcripts that are completely degraded at the MZT vary dramatically between species . We also show that hundreds of genes have different isoform usage between the maternal and zygotic genomes . Our work suggests that maternal transcript deposition and early zygotic transcription are remarkably dynamic over evolutionary time , despite the widespread conservation of early developmental processes .
Most early developmental processes , such as rapid cleavage cycles and the establishment of body axes , are shared across multicellular animals , but the extent to which the mechanisms and the genes involved are also shared remains an open question . Throughout the animal kingdom , the first stages of development are controlled by mRNA transcripts and proteins deposited by the mother during oogenesis . Genetic control is subsequently transferred from the maternal genome to the zygotic genome . This is accomplished through a precise and elegant series of regulatory steps , in which the zygotic genome is transcriptionally activated while maternal transcripts are degraded , in a process known as the maternal to zygotic transition ( MZT; [1 , 2] . This handoff between mother and zygote has the appearance of a functional logic that dictates which genome is in control . Genes involved in processes unique to the earliest stages of development , such as rapid cleavage cycles , are necessarily transcribed by the mother . Genes that control processes such as patterned gene expression in the developing embryo require zygotic transcription from specific subsets of cells . And genes performing essential housekeeping functions required at all stages of life are transcribed by both the mother and the zygote . For these genes , the maternal and zygotic genomes are able to coordinate during the MZT to such a degree that the transcript levels of these genes can be relatively constant , despite the transition between genomes of origin for these transcripts . In general , the logic underlying the partitioning of gene expression between the maternal and zygotic genomes is unclear . While we have examples of particular genes that are transcribed by either ( or both ) the maternal and zygotic genomes , in accordance with the requirements discussed in the previous paragraph , it is unknown whether these requirements play out to shape maternal and zygotic gene expression genome-wide . One way to address this question is to analyze evolution of transcript pools at these stages on short to moderate timescales . If the genome of origin is a constraining factor for many genes , we would expect to see a high degree of conservation of maternal or zygotic expression for those genes . Alternatively , if transcripts of some genes may be supplied by either mother or zygote , we might expect to see control of expression vary between the two genomes across species . The current evidence for conservation of maternal and early zygotic regulatory factors is mixed . One of the most critical maternal genes in the fruit fly Drosophila melanogaster , bicoid [3] , is of relatively recent evolutionary origin . Bicoid controls axis formation ( determines the anterior pole of the egg ) , and is not found outside of higher Diptera [4] , having resulted from a gene duplication . This demonstrates that conserved early developmental processes can incorporate new genes . Some theory and empirical studies suggest that maternal genes might be expected to evolve more quickly than zygotic genes , as selection will be less effective since these are ( largely ) autosomal genes that are expressed in only one sex [5–7] . These studies examined coding region changes in a limited number of genes , and might not fully account for the significant developmental constraint imposed by the need to build functional offspring . However , recent genomic studies demonstrate a high degree of maternal transcript level conservation relative to zygotic gene expression [8 , 9] . Mechanistically , maternal deposition and zygotic expression are subject to different constraints . Prior to zygotic genome activation , deposited maternal gene products are the only mRNA transcripts available , and transcript level cannot be dynamically increased to respond to the rapidly changing environment of the earliest developmental stages . Perhaps because of this , post-transcriptional control mechanisms play an especially important role in the levels of gene products produced from maternal transcripts [10–15] . On the other hand , since the products of many zygotic genes are needed soon after transcription is activated , the accumulation of sufficient transcripts in a short period of time ( especially in species with rapid development such as Drosophila ) can be difficult . For this reason , genes expressed in Drosophila at the early zygotic stage tend to be short in length and contain few or no introns [9 , 16 , 17] , allowing them to be transcribed quickly . Since both the maternal and zygotic phases of early development have unique constraints , predicting the level of conservation of each stage is challenging . In order to assess the extent of transcript level conservation in early development across evolutionary time , we sequenced transcriptomes from early embryonic stages from 14 Drosophila species . These species span divergence times of approximately 250 , 000 to close to 60 million years . We determined mRNA levels both before and after zygotic genome activation , analyzing the conservation and divergence of transcript representation across species over evolutionary time . Our findings show strong levels of evolutionary conservation of both maternal and zygotic transcripts . However , genes that are represented at only one stage ( either before or after zygotic genome activation ) show strikingly high degrees of transcript level evolution compared to genes represented at both stages . The results suggest that in most cases robust transcript levels may be achieved through regulatory mechanisms that rely on both the maternal and zygotic genomes . Nevertheless , when all transcripts at the maternal stage are compared to all the genes at the post-genome activation stage , the maternal stage has a higher degree of conservation . This is in contrast to the pattern observed with stage-specific transcripts , where transcripts present at the maternal stage that are entirely degraded at the MZT evolve faster than zygotic genes with no maternal contribution . Furthermore , we find that expression levels of a small proportion of zygotic genes with no maternal contribution are tightly regulated , and the use of stage-specific isoforms may be a hitherto unrecognized method of partitioning the contributions of two genomes . Finally , combining our results with data from the mosquito Aedes aegypti , we show the conservation of a core set of zygotic genes over 250 million years . Our study demonstrates the power of transcriptomic phylogenetics to identify the key players regulating core developmental processes .
Our results confirm our previous finding that single-embryo RNA-seq is highly reproducible [26–28] . Spearman rank correlation coefficients of transcript abundance ( in fragments per kilobase per million reads mapped , or FPKM ) for replicates from the same stage are high at both developmental stages ( S1 Table ) , though slightly lower at stage 5 ( post-zygotic genome activation ) than stage 2 ( maternal transcripts only ) . For stage 2 , the correlation coefficients within species range from 0 . 94–0 . 98 across 13 of the 14 species ( Fig 1A; S1 Table ) , with a mean of 0 . 96 . D . erecta stage 2 transcriptomes have slightly lower correlations ( 0 . 91–0 . 96 ) . Spearman coefficients for stage 5 replicates ( Fig 1A; S1 Table ) are always lower than their stage 2 counterparts and are also more variable , ranging from 0 . 80 to 0 . 95 , with a mean of 0 . 90 . They are still higher , however , than correlation coefficients of replicates from different stages ( for any given species ) , which range from 0 . 67 to 0 . 83 , with a mean of 0 . 76 ( Fig 1C; S1 Table ) . There are a few reasons why stage 5 transcript levels may be expected to be less correlated than stage 2 , even within replicates . First , stage 5 transcript levels are the result of both zygotic transcription and maternal transcripts that have yet to degrade . Approximately 50% of maternal transcripts are still present at this timepoint [1 , 17 , 29] , and degradation of maternal genes may vary between species [28] . As full activation of the zytogic transcriptome begins during stage 5 , small differences in developmental timing between replicates in this stage may also produce differences in zygotic transcripts present . To address this , our stage 5 timepoint is a precise point at the end of this stage , when cellularization has completed , but prior to gastrulation ( see Methods ) . Finally , embryo sex may begin to play a role by the end of stage 5 . As the sex determination pathway has been activated by this stage , it is possible to distinguish males from females by comparing levels of known female- and male-specific transcripts ( Sxl and msl-2 ) [30] . While sex-specific differences in stage 5 transcript abundance across species have been described previously [28] , there are no consistent differences between Spearman rank sum comparisons of same-sex vs . opposite-sex stage 5 embryos ( S1 Table ) , suggesting that these differences are overwhelmed by transcript levels from non-X-linked genes when comparing whole transcriptomes . Since replicate FPKM levels are highly correlated , our analysis in the remainder of this paper focuses on mean replicate FPKM values for a given species and stage . While transcript levels within each stage are highly correlated across species ( Fig 2; S2 Table ) , they diverge as evolutionary distance increases . Spearman rank correlation coefficients decrease when comparing a species from within the melanogaster subgroup ( e . g . D . melanogaster ) with more distantly related flies , but only drop to around 0 . 7 for divergence times of approximately 57 million years ( e . g . D . melanogaster to D . virilis , Fig 2C ) . As was found with replicates from the same species , stage 5 correlation coefficients are usually slightly lower than the equivalent stage 2 coefficients ( Fig 2B and 2C; S2 Table ) . In contrast , cross-species comparisons of different stages show striking differences . Clustering the transcriptomes from all species and timepoints ( S1 Fig ) , the two stages fall out as the first two distinct clusters , demonstrating that the biological distinctiveness of these developmental stages in Drosophila transcends interspecific variation . While divergence in transcript levels generally increases with evolutionary distance , transcriptome comparisons do not recapitulate the phylogeny ( S1A Fig ) . In particular , the obscura group shows strong transcriptomic divergence in excess of its phylogenetic distance from other species . The obscura group has experienced a number of fusions of sex chromosomes , resulting in a larger proportion of the genome being sex-linked [31] . To determine if this could be driving the pattern we observe , we performed the clustering analysis on autosomal genes , removing all gene groups where one or more orthologs were located on a sex chromosome in any of the species with chromosomal-level annotations . This required removing gene groups with an ortholog on the X chromosome , or Muller element A , in D . melanogaster , D . simulans , D . mauritiana , D . yakuba , D . pseudoobscura , and D . miranda , in addition to gene groups with an ortholog on Muller D in D . pseudoobscura or D . miranda or an ortholog on Muller C in D . miranda [32–34] . We found that the obscura group still clusters together outside the rest of the species when sex-linked genes were removed from the analysis ( S1B and S1C Fig ) . As this group represents the only clade in our study adapted to a temperate environment ( D . virilis is also considered a temperate species , but has no close relatives in our dataset ) , some of the differences in this group may be ecologically relevant . We return to this observation below . We then focused on stage-restricted genes , either maternal-only ( maternally deposited and entirely degraded by stage 5 ) or zygotic-only ( not maternally deposited , transcribed from the zygotic genome and present at stage 5 ) ; see Methods for further definitions . The number of maternal-only genes is relatively small; on average this group represents ~6% of transcripts present at either or both of these stages ( S3 Table ) . In comparison , zygotic-only transcripts represent ~20% of those present at these stages ( S3 Table ) . When comparing species pairs , we found that as evolutionary distance increases , the number of orthologs that are restricted to the same stage in both species declines , particularly in the maternal-only set ( S2 Fig ) . If the transcript levels of genes that are restricted to a given stage in either of the two species are compared , we find that correlation coefficients for both types of stage-restricted genes are markedly lower than those for all stage 2 and stage 5 genes ( S3 Fig ) . However , if we only compare the transcript levels of genes that are stage-restricted in both species ( Fig 3 ) , we see a striking contrast between the maternal-only group , which shows strong transcriptomic divergence as evolutionary distance increases , and the zygotic-only group , which shows remarkable conservation among even the most distant species in our analysis . This suggests that while both groups of stage-restricted genes show rapid evolution relative to non-stage restricted genes present at these stages , the transcript levels of a subset of zygotic-only genes are more tightly regulated across Drosophila clades . To investigate evolution of transcript pools in early development over longer periods of evolutionary time , we incorporated data from the previously published developmental transcriptomic time series of a basal Dipteran , the mosquito Aedes aegypti , which separated from Drosophila 170 to 250 million years ago [35] . Since Aedes aegypti shares the long-germ band mode of development with Drosophila [36] , where the body plan is established simultaneously rather than sequentially [37] , early developmental processes are largely conserved . In the Aedes aegypti dataset , all transcripts are maternal at the earliest stage assayed ( 0–2 hours; the equivalent of Drosophila stage 2 ) , while the approximate equivalent of Drosophila stage 5 is around 10 hours , when Aedes cellularization is complete [38] , covered by the 8–12 hour window in this time series . We will refer to these stages as Aedes stage 2 and Aedes stage 5 , respectively . As shown in hierarchical clustering analysis of the Drosophila and Aedes transcriptomes ( Fig 4A , S1C Fig ) , the Aedes transcriptomes from both developmental stages cluster together , rather than clustering with the comparable Drosophila stages , suggesting that evolutionary divergence of this magnitude is more significant than divergence between developmental stages . However , despite this divergence , and in spite of the differences in the experimental protocols that generated the data ( the Drosophila transcriptomes were generated from single embryos , while the Aedes time series used pooled embryos ) , there are still moderately strong and highly significant correlation coefficients between transcript levels of orthologous genes from the respective stages ( Fig 4B , S2 Table ) . We were particularly interested in analyzing the classes of genes that showed the greatest change in representation within Drosophila ( maternal-only and zygotic-only ) and determining whether we could find conserved orthologs between Drosophila and basal Diptera . In doing so , we found considerable change in stage-specific genes at this evolutionary distance , but also a core set of highly conserved zygotic-only genes likely to have important functions . Considering first the zygotic-only genes , we examined two sets of genes: 1 ) genes that displayed any evidence of conservation between Aedes and a Drosophila species , and 2 ) a subset of these genes that also showed strong evidence of conservation within the Drosophila clade . Looking first for any conservation between Aedes and Drosophila , we found 173 genes ( from a set of 4619 orthologs identified across both stages ) that were zygotic-only in both Aedes and at least one of the 14 Drosophila species ( S4 Table ) . This set of 173 genes represents approximately 10% of the mean number of zygotic-only genes in a Drosophila species ( S3 Table ) . A given gene in this set was , on average , also zygotic-only in 6 . 5 other Drosophila species ( S4A Fig ) . By contrast , a random gene from the larger dataset of 4619 orthologs was on average zygotic-only in less than one Drosophila species . To identify a more stringent set of genes that are also conserved within Drosophila , we required genes to be be zygotic-only in Aedes aegypti and two of the earliest-diverging Drosophila species ( D . willistoni , the outgroup to Old World Sophophora , and at least one of the two Drosophila subgenus species ) . With this set of criteria , we identified a core set of 61 genes , which show even greater evolutionary constraint ( S4B Fig ) , and are , on average , zygotic-only in 10 other Drosophila species ( a total of 13 of the 15 species ) . This number represents ~4% of the average number of zygotic-only genes in a Drosophila species ( S3 Table ) . Our results indicate that although most zygotic-only genes show rapid change in representation over stages , a core set can be identified where the zygotic-only state is highly conserved ( Fig 4C , S5 Table ) . These 61 genes represent key players in D . melanogaster embryonic patterning , with gap genes , pair-rule , segment polarity and homeotic genes in the anterior-posterior pathway represented , in addition to numerous genes in the dorsal-ventral patterning pathway . A gene ontology ( GO ) analysis using DAVID [39 , 40] comparing this set to the larger set of stage 5-represented genes shows that it is enriched approximately 8-fold in transcription factors ( S4C Fig ) , with terms related to early development ( S4D Fig ) strongly overrepresented . These results suggest that a naïve analysis based on conservation of transcriptional state can yield remarkable insight into the set of genes that are functionally significant . In contrast to the zygotic-only genes , we were unable to identify a significant core set of maternal-only ( maternally deposited , completely degraded at the MZT ) genes that were conserved between Aedes and Drosophila . Indeed , out of the approximately 4000 orthologs we examined , we only found 3 genes ( beaten path Ia , Phospholipase D and Sclp ) that were maternal-only in both D . melanogaster and A . aegypti . These results , which are consistent with our findings in Drosophila , suggest that this subset of the maternal genes ( or the maternal-only status ) may not be essential for key developmental processes that are conserved across Diptera . One possibility is that maternal deposition could be a process with considerable developmental noise , and that degradation might be a method for compensating for non-functional transcripts that are dumped by nurse cells during oogenesis . Alternatively , these transcripts may have clade-specific functions limited to the earliest stage of development . The previous analysis showed that transcriptomic conservation during early embryonic development is the norm , with RNA levels tightly correlated at up to 60 million years of evolutionary divergence . However , by reconstructing ancestral states ( presence or absence of gene transcripts ) using the Bayesian phylogenetics package MrBayes [41] ( Fig 5; S6 Table ) , we were also able to identify genes that change in their representation in either the maternal or zygotic transcript pools . For this analysis , a gene was considered represented at a given stage if the FPKM level of its isoforms was at least 1 . The ancestral stage 2 and stage 5 states of 7092 genes with one-to-one orthologs in at least 12 of the 14 species were reconstructed ( see Methods for more details ) . Even when using relatively liberal parameters for identifying transitions ( see Methods for details ) , only 245 of these genes , or less than 4% , showed evidence of a stage 2 gain or loss at any node on the phylogeny ( Fig 5A ) . Stage 5 transitions were approximately twice as common , with 499 observed ( Fig 5A ) . By far the greatest number of transitions at both stages were seen in the obscura group ( see Fig 5B for examples ) , in support of the hierarchical clustering results ( Fig 4A , S1 Fig ) suggesting that the early embryonic transcriptome in species of this group is exceptional . An additional finding of note is that loss of representation does not occur with greater frequency than gain of representation ( Fig 5A ) . Our phylogenetic analysis allowed us to address two major questions: 1 ) what are the evolutionary patterns of changes in transcript representation relative to maternal , zygotic , or maternal and zygotic representation; and 2 ) what types of transcripts undergo gains or losses along the tree ? To address the former , we compared patterns of gains and losses along the phylogeny . Observing changes in transcript representation at both developmental stages , we rarely find genes that transition from an entirely maternal ( maternal-only ) state to an entirely zygotic ( zygotic-only ) state ( S7 Table ) . Indeed , the only gene in S7 Table to show a simultaneous gain of representation at both stages is quasimodo ( qsm ) , and there is no gene that shows a simultaneous loss at both stages . Primarily , we observe transcripts that are present at both stages losing representation at one stage , or transcripts that are present at one stage gaining representation at the other . This means , for example , that to transition from maternal-only to zygotic-only , the gene would pass through an evolutionarily intermediate state where it is represented at both stages . Next , we examined which classes of genes change in transcript representation at either stage . We performed gene ontology analysis ( see Methods ) of genes that change transcript representation when compared with all transcripts present at that stage . In the case of genes that gain stage 5 representation , there is significant enrichment for genes that are involved in transport of small molecules through membranes ( ion channel activity , substrate-specific channel activity , ion transport , ion channel complex , channel activity , etc . ; see S5 Fig , S8 Table ) . We might expect transcription of these genes to evolve quickly [42] , as their products permit adaptation to new environments on a cellular level . Additionally , some types of these transport mechanisms have been implicated in genetic conflict [43 , 44] and could thus be expected to evolve quickly . The other categories of transcript abundance changes ( stage 5 losses , stage 2 gains , stage 2 losses ) had no significant enrichment of gene ontology categories compared to the rest of the genes with transcripts present at that stage . We highlight a few examples of individual gains of transcript representation in Fig 5B . These particular genes were chosen because they are dramatic examples of changes in transcript representation at different stages and in different species . Tpi ( Triose phosphate isomerase ) and Nhe2 ( Na+/H+ hydrogen exchanger 2 ) , show gains at stage 5 and stage 2 respectively , in the obscura lineage . Tpi functions in glycolysis , and is a classic example of clinal allele frequencies in D . melanogaster populations [45] . Nhe2 is a Na+/H+ exchanger that increases cellular pH , and has been shown to be upregulated in cold-acclimated D . melanogaster [46] . Consistent with this previous evidence of a role in environmental adaptation , both genes showed gains in our cold adapted ( temperate ) obscura group species . In most cases , however , it is difficult to interpret the significance of the change in representation . Ect4 ( Ectoderm-expressed 4 , Fig 5 , third panel ) , was one of number of genes that showed gains of expression in the melanogaster group . In this case , there was a gain in stage 2 representation in this lineage , while stage 5 was also represented in the outgroup . The role of Ect4 has largely been determined relative to its function in axon degeneration for the purposes of repair after injury , which is not directly relevant to its role in the early embryo . Recently , Ect4 was shown to be a target of DPP signaling in the embryo [47] , but its precise function is unknown . We also found genes , such as DNAJ-H , that showed highly variable transcript levels across the phylogeny . Transcripts of this gene , which is part of the Heat Shock Protein 40 family of co-chaperones ( essential factors in the Hsp70 chaperone cycle ) , may be represented at both stages or at either of the two stages , with no apparent pattern relative to the phylogeny . These changes provide evidence that the early embryonic transcriptome is evolutionarily dynamic within Drosophilids . While the mRNA levels of some genes with key roles in development are highly conserved , a substantial subset of genes whose functions are not as well-understood show evidence of lineage-specific changes that are potentially adaptive . Our results from previous sections suggest relatively rapid evolution of both maternal-only and zygotic-only transcripts ( S3 Fig ) . Transcript levels of a subset of genes where the zygotic-only state is conserved across Drosophila species , however , are highly correlated ( Fig 3 ) , and a core set of zygotic-only genes with critical developmental roles are conserved with basal Diptera ( Fig 4 ) . To extend our analysis , we looked at two special situations: cases of species-specific representation in the early embryo , and the transcriptional profile of a small group of genes that were previously unannotated . We were first interested in knowing whether there were instances of unique representation: cases where transcripts of a gene were present at a given stage in a species , while transcripts of one-to-one orthologs from all other species assayed were absent from that stage ( S9 Table ) . We only considered genes with one-to-one orthologs in at least 12 of the 14 species . In order to eliminate cases were the transcript level of a gene happened to be slightly above our threshold of 1 in a single species , we focused on instances where a gene had a transcript level over three times the threshold ( FPKM > 3 ) in a single species but an FPKM of less than 1 in all other Drosophila species ( S6 Fig; S9 Table ) . We found many more cases of species-specific representation at the stage 2 than stage 5 , for every species except D . melanogaster ( which has low numbers of both ) . There are an exceptionally high number of cases of unique representation of maternal deposition in both D . virilis and D . erecta , while D . virilis also has an unusually high number of instances of unique representation at the zygotic stage ( over twice as many as the next highest species , see S9 Table ) . We also examined a different situation: unannotated genes identified during analysis by Cufflinks . These genes were not found in the reference annotations ( see Methods for annotations used for each species ) provided to the software . The number of unannotated genes ranged from a low of 280 in D . melanogaster , the most thoroughly annotated genome , to a high of 1905 in D . mauritiana ( S10 Table ) . In many cases , unannotated genes may simply reflect the limitations of previously generated annotations . However , as has previously been found for de novo genes [48] , unannotated non-D . melanogaster genes show low complexity , having significantly fewer exons ( 2–3 vs . the mean from all genes of 4 . 5–5 . 5; Wilcoxon rank-sum test , p = 7 . 4 x 10−6 ) , as well as shorter exons ( mean of 1240bp vs 2547bp , Wilcoxon test , p = 1 . 9 x 10−5 ) and introns ( mean of 2105bp vs . 5256bp , Wilcoxon test , p = 5 x 10−8 ) than annotated genes ( S7 Fig; S10 Table ) . It is possible , therefore , that some of the unannotated genes identified here are de novo genes . We note that these properties of de novo genes are shared with annotated zygotic genes , which also tend to have shorter and fewer exons and few or no introns [9 , 16 , 17] . The unnanotated genes expressed by the zygotic genome identified here , and reported in the statistics above , are less complex as compared to annotated zygotic genes ( S7 Fig; S10 Table ) . We found that unannotated genes were strongly biased towards being zygotic-only ( Fig 6 ) . When considering all genes that are transcribed during this early period of development , only 16–25% are zygotic-only , depending on the species ( Fig 6 , “all genes” ) . For the unannotated genes represented by transcripts in our dataset , 65–80% are zygotic-only . This is a 3- to 4-fold increase that is highly significant using a Fisher exact test ( p<0 . 0001 ) . The finding that these unannotated genes are more frequently zygotically transcribed than maternally deposited could indicate that early zygotic transcription evolves rapidly for putatively novel genes . This contrasts with our results for conserved genes ( S6 Fig ) , discussed above , which were found to be more likely to evolve species-specific cases of maternal deposition than zygotic representation . In other words , a potentially novel gene of unknown function will often evolve early zygotic transcription , while a gene with conserved orthologs is more likely to evolve a unique instance of maternal representation . The generation of multiple isoforms for a given gene , through mechanisms such as alternate promoters or alternative splicing , has been recognized as a critical form of genetic regulation [49] . Organisms deploy these isoforms in a context-dependent manner to meet the varied challenges of both development and adult physiology . Little is known , however , about the role of alternative isoforms in the maternal to zygotic transition . In lineages such as Drosophila , flies that are adapted for rapid development , it has previously been shown that zygotic transcripts are shorter and have fewer introns than maternally deposited ones [9 , 16 , 17] . It is reasonable , therefore , to hypothesize that certain isoforms of the same gene may be better suited for maternal deposition or zygotic transcription . Specific examples of such cases are limited , however , and few studies have looked at the extent to which stage-specific isoforms are evolutionarily conserved . Our data suggest that different isoforms may be used stage-specifically , presenting an additional layer of subtlety when comparing early embryonic transcriptomes across species . Consider , for example , the gene headcase ( hdc ) , whose function has been characterized at later developmental stages , is expressed in all imaginal lineages , and is involved in trachea , head , and neuroblast development [50] . In the early embryo , transcript levels of orthologs of this gene are present at a higher level at stage 5 than stage 2 in both D . simulans and D . sechellia ( Fig 7C ) . However , while both species show zygotic enrichment for the predominant isoform of its ortholog ( Fig 7D , labeled “isoform 1” on the figure in both cases ) , the second most highly expressed isoform of the D . simulans ortholog ( labeled “isoform 2” ) is higher at the maternal stage , while the second most highly expressed D . sechellia isoform shows no significant difference in transcript levels between stages . A third isoform is also present at low levels in D . sechellia . To look more closely at the isoforms that are present at each stage , we classified each isoform into one of six categories: maternal ( M–only present at stage 2 ) , zygotic ( Z–only present at stage 5 ) , predominantly maternal ( PM–present at both stages , but at least twice as high at stage 2 than stage 5 , with the differences being statistically significant ) , predominantly zygotic ( PZ–present at both stages , but at least twice as high at stage 5 than stage 2 , with the differences being statistically significant ) , maternal-zygotic ( present at both stages , but predominant at neither ) and not present at either stage ( N ) . When looking across all isoforms , we find that in all species the mean exon number and the mean exonic length are slightly greater for the combined class of M and PM isoforms than for Z and PZ isoforms ( S10 Table ) . The mean intronic length showed much starker differences , and was from 1 . 5 to 2-fold greater in the M/PM isoforms . This extends the findings from previous studies that zygotic genes are shorter and have fewer introns [9 , 16 , 17] to across the genus Drosophila , but our data shows much stronger support for differences in intron length than exon number or length between maternal and zygotic transcripts . We looked specifically for genes that had at least one isoform that was maternal or predominantly maternal , and another that was zygotic or predominantly zygotic , and referred to these as the “alternative” ( ALT ) set ( S10 and S11 Tables ) . The number of identified ALT genes varied from a low of 307 ( ~4% of genes ) in D . ananassae to a high of 936 ( ~ 10% of genes ) in D . virilis ( Fig 7A , S10 Table ) , likely partially a consequence of statistical power . Regardless , the identification of hundreds of genes with stage-specific isoforms ( Fig 7A ) suggests that this could be an important regulatory strategy in early embryonic development . In contrast to the patterns we observe for all genes , the ALT isoforms do not differ consistently in exonic number and exonic length ( S8 Fig ) . Across isoforms , the number of exons and exonic length are always greater for isoforms of ALT genes than for those of all genes ( S10 Table ) . These results may indicate that ALT gene products are required to be long and potentially complex to produce stage-specific isoforms , and that a lower proportion of intronic/exonic sequence may be selectively favored in isoforms that are zygotically transcribed . To explore the functions of these genes with stage-specific or stage-enriched isoforms , we performed a GO analysis . When compared to the entire gene set , we found that these genes are moderately enriched in GO categories of genes that regulate development , morphogenesis and cell differentiation , in addition to sexual reproduction and gamete generation , among several other categories ( S12 Table ) . This is consistent with the fact that one of the most celebrated cases of alternative splicing is in Drosophila sex determination , and other the key players in the pathway ( Sex-lethal , transformer , doublesex , fruitless ) being regulated through alternative splicing . Regardless of their precise functions , the ALT gene state is frequently conserved over long evolutionary distances ( S11 Table; S13 Table ) , with 67 genes showing stage-specific alternative isoforms in orthologs of at least 2/3 of the species . As one example ( Fig 7B ) , consider the case of cap-n-collar ( cnc ) . This transcription factor , which is essential for viability in D . melanogaster , has one or more long isoforms ( 11 exons , 6500–7000 bp of exonic sequence in D . melanogaster ) , one or more medium-length isoforms ( 5 or 7 exons , approximately 4000 or 5000 bp in D . melanogaster ) and one or more short isoforms ( 3 exons , 3500 to 4000 bp in D . melanogaster ) . One-to-one orthologs were identified for 12/14 species ( S11 Table ) , and all of these orthologs except D . mauritiana were classified as ALT , with the same pattern of at least one long maternal-only , one medium-length zygotic-only , and one short maternal-only isoform found in each case except D . miranda . The broad conservation of this pattern across 12 species spanning over 55 million years of evolution provides strong evidence for a functional role for both the multiple isoforms and their maternal or zygotic transcription .
The strong conservation of maternal transcript levels across Drosophila demonstrates that the mother’s vast RNA endowment is regulated with a precision that has withstood sixty million years of evolution . This conservation is all the more remarkable given the divergent ecologies and life histories of the species analyzed [19] , and the extensive role played by post-transcriptional regulation of maternal transcripts [10–15] . Additional study will be necessary to determine protein abundance in each species , as post-transcriptional regulation may play an additional role in buffering the effects of differential transcript abundance . Since only a small minority of these transcripts are transcription factors , however , it is clear that the function of maternal deposition extends far beyond jumpstarting transcription in development . Theoretical comparisons of selective efficacy notwithstanding [7] , maternal genes such as bcd that are recent evolutionary innovations [51 , 52] should be considered the exception rather than the rule . From another perspective , however , the conservation of stage 5 transcript levels , while somewhat reduced relative to levels of maternal deposition ( stage 2 ) , is arguably even more remarkable . Transcript levels at stage 5 are a function of multiple processes: maternal RNA deposition that occurred during oogenesis , multiple embryonic degradation pathways , which themselves may be activated either maternally or zygotically [53 , 54] , and early zygotic transcription . These processes must be tightly coordinated to generate zygotic levels that are reproducible not only between individuals but also between species , a finding that is all the more impressive given the fact that they are regulated in two separate genomes ( that of the mother and the zygote ) . Transcript levels of two categories of genes show more rapid divergence ( S3 Fig ) : Maternal genes with transcripts that are entirely degraded by stage 5 ( maternal-only ) , and zygotic genes with no maternal contribution ( zygotic-only ) . These results support the hypothesis that the combination of maternal deposition and zygotic transcription is important for achieving the robust transcript levels that might be generally required for early embryonic development . However , when conducting pairwise comparisons in which only the set of genes where the specific stage-restricted state ( maternal-only or zygotic-only ) is conserved in both species is considered ( Fig 3 ) , there are distinct differences in the evolutionary trajectories of these two gene classes . Maternal-only genes , those maternal genes that are degraded at the MZT and absent by stage 5 evolve quickly . Interspecific correlation coefficients for transcript levels of maternal-only genes drop off rapidly with evolutionary distance , even when only the genes where the maternal-only state is conserved in both species are considered ( Fig 3 ) . The number of shared maternal-only orthologs also decreases rapidly with evolutionary distance ( S2 Fig ) , to the extent that there only three D . melanogaster–Aedes aegypti orthologs that are maternal-only in both clades . How we view these results depends on our interpretation of maternal-only genes . If both the deposition and the degradation are presumed to be functional , these genes would be cases where the transcripts are necessary in the very early , syncytial , embryo ( or previously in oogenesis ) but are strongly detrimental after cellularization . The finding that an ortholog of such a gene is not maternal-only in a related species would signify that either the gene was no longer necessary in the syncytium or that it was no longer detrimental at later stages . However , our results showing rapid divergence of transcript levels in cases where orthologs are maternal-only in both species , implies that if they have an early function they belong to a class of genes where tight regulation of transcript level may not be necessary . This would distinguish them from the vast majority of genes represented at stage 2 , where transcript levels are highly conserved . Alternatively , maternal-only genes may represent developmental noise , with degradation as a method of compensating for the noise . During oogenesis , vast numbers of transcripts and proteins are deposited by the nurse cells in the egg , and it is possible that not all of them are necessary or beneficial to the embryo . Finally , we must consider that the post-transcriptional regulation of maternal transcripts might have a larger impact on the maternal-only transcripts , and that translational control may buffer against any phenotypic consequences of differences in transcript level . This would explain the variation in transcript levels for these genes , but not how quickly the maternal-only status is gained or lost . The sharp decrease in shared orthologs of maternal-only genes as evolutionary distance increases lends weight to the interpretation that maternal deposition may be noisy , and argues against post-transcriptional regulation having a larger role for the maternal-only class of genes as discussed above . The transcript levels of most genes that are zygotic-only diverge rapidly in pairwise comparisons of interspecific orthologs ( S3 Fig ) . However , if we limit our analysis to the smaller group of shared zygotic-only orthologs ( genes that are zygotic-only in both species being compared; S2 Fig ) , we see a very different pattern ( Fig 3 ) , with transcript levels highly correlated , even among distantly related Drosophila species . Looking across much greater evolutionary distances , we were able to identify a core set of genes that are zygotic-only in both Aedes and early-diverging Drosophila species . These genes are strongly biased towards retaining the zygotic-only state across the Drosophila lineage . Our finding that this set is highly enriched in transcription factors with known functions in embryogenesis shows the power of evolutionary transcriptomics to identify key players in development . Functionally , we might expect these genes to be cases where zygotic-only expression is necessary since maternal deposition may be mechanistically deleterious prior to cellularization . Across species , an overwhelming majority of unannotated genes have zygotic-only transcript representation in the early embryo , while only about a fifth to a quarter are zygotic-only in the annotated set . Many of the unannotated genes we identified display the hallmarks of newly-evolved genes , with relatively few isoforms and low expression levels , possibly suggesting that novel genes are biased towards zygotic expression ( more work will need to be carried out on this gene set to determine whether the genes are indeed taxonomically-restricted ) . The idea that zygotic representation , on the whole , evolves more readily than maternal deposition is also consistent with our phylogenetic analysis , where a strong majority of the gains were stage 5 . The maternal-only genes discussed in the previous section are an exception , as they consist of the minority of maternal transcripts that are entirely degraded by stage 5 . Conversely , maternally deposited RNA differs from its zygotically transcribed counterpart in that it can be used during the earliest syncytial stages of embryonic development . If these early stages are highly conserved , the evolution of new genes with a function in this period may rarely be necessary . Maternal deposition may instead evolve to increase the overall robustness of RNA levels during post-syncytial development . Or , perhaps the earliest stages of development where maternal mRNAs act require different gene products to undergo the conserved developmental processes in differing environments [55] . The phylogenetic pattern of evolution of transcript representation at the maternal and zygotic stages speaks to both a regulatory logic and to the relative roles of maternal and zygotic genomes in early development . We found that maternal-only genes hardly ever become zygotic-only ( or vice versa ) between closely related species . Instead , we see genes transcribed by both the maternal and zygotic genomes losing either maternal deposition or zygotic transcription , or stage-restricted genes gaining transcript representation at the other stage ( e . g . a maternal-only gene gains zygotic transcription as well ) . This pattern can potentially be explained by the logic of regulation , since gaining or losing regulatory binding sites ( or regulatory factors ) at one stage may be a much more common occurrence than simultaneous evolution to both gain one set of binding sites or factors associated with transcription at one stage and lose binding sites or factors for the other stage . At the same time this provides evidence against compensatory evolution over these two stages as loss at one stage is not associated with gain at the other . While gene number does not appear to correlate with any measure of organismal complexity , it is often claimed that isoform number might [49 , 56 , 57] . Alternative splicing is particularly common in vertebrates , although the record for isoform number is currently held by the Drosophila Dscam1 gene , where isoforms vary across individual neurons [58] . Drosophila also famously uses alternative splicing in sex determination [59] . In addition to tissue-specific and sex-specific alternative isoforms , stage-specific isoforms allow for complex temporal regulation [60] . In the early embryo , where both the mother and zygote provide RNA , the logic of utilizing alternative isoforms stems from the differing constraints of each of these players . For example , the rapidity of transcription is a strong selective pressure in the zygote where cell divisions are extremely rapid , leading to zygotic transcripts with fewer introns [16] , while maternally supplied RNA is under no such constraint . Additionally , there are transcripts provided by both maternal and zygotic transcription where the maternal transcripts are to be selectively degraded at the MZT . Then the maternal genome could use isoforms with appropriate motifs to direct degradation in their untranslated regions ( e . g . miRNA target sites ) , and the zygotic isoform , without these motifs , will persist . Our discovery of hundreds of cases of alternative stage-specific isoforms ( ASIs ) in the 14 species we examined validates the potential utility for using different isoforms at these different stages . Furthermore , multiple cases of strong conservation of isoform structure ( number of exons , overall exonic length ) for stage 2 or stage 5 isoforms across 60 million years of evolution suggests functionality for this process . Future research will aim to determine if the localization of transcripts differs between the maternally and zygotically predominant isoforms , how these isoforms are differentially regulated , and whether they are functionally equivalent . We have demonstrated the remarkable ability of two genomes to collaborate in the regulation of early development , leading to RNA transcript levels in the embryo that are highly stable over tens of millions of years of evolution . Yet , we also find considerable variation in the transcripts present during the earliest stages of development , despite expectations that early development is highly conserved . A large remaining question is how much of this variation is functional . It is plausible that some fraction of it is , and this would imply either that the processes of early development are not as conserved as commonly regarded , or that different complements of transcripts are necessary across different environments and genomes to maintain these conserved early developmental processes . Alternatively , it could be that the processes of early development are remarkably robust , and that considerable variation in transcript abundance or transcript representation may have minimal phenotypic consequences .
Single embryos were collected from 3–8 day old females of each species . Genome lines from the original 12 Genomes study [21] were used for 11 of the species ( D . melanogaster , D . simulans , D . sechellia , D . yakuba , D . erecta , D . ananassae , D . pseudoobscura , D . persimilis , D . mojavensis , D . virilis ) . The lines for the additional species were as follows: D . mauritiana ( Dmau/[w1]; 14021–0241 . 60 ) , D . santomea ( STO-CAGO 1402–3; 14021–0271 . 01 ) , D . miranda ( MSH-22 ) . Embryos were dechorionated , and imaged on a Zeiss Axioimager , under halocarbon oil , to determine stage . Since embryos were collected from a large number of mothers , it is unlikely that multiple samples came from the same mother . Stage 2 and late stage 5 embryos were identified based on morphology . Stage 2 embryos were selected based on the vitelline membrane retracting from both the anterior and posterior poles , prior to when pole cells become visible . Late stage 5 embryos were chosen based on having completed cellularization , but not yet having gastrulated . Embryos were then removed from the slide with a brush , cleaned of excess oil , placed into a drop of Trizol reagent ( Ambion ) , and ruptured with a needle , then moved to a tube with more Trizol to be frozen at -80° C until extraction . RNA and DNA were extracted as in the manufacturer’s protocol , with the exception of extracting in an excess of reagent ( 1 mL was used ) compared to expected mRNA and DNA concentration [26–28] . Extracted total RNA from single embryos was treated with the TurboDNA-free kit ( Ambion ) prior to library construction . Embryo mRNA-Seq libraries were generated for at least 3 replicate individuals per stage and per species , producing a total of 68 stage 2 and 76 stage 5 libraries , or 144 overall . More detail about sampling and replication is available in S14 Table . mRNA-Seq libraries were constructed using TruSeq RNA sample preparation kits ( Illumina ) , using standard protocols , and indexed to pool 12 samples ( embryos ) per lane . Library concentration was measured using the Qubit fluorometer ( Life Technologies ) and the qPCR-based Library Quantification kits ( KAPA biosystems ) , and size was measured using the Bioanalyzer ( Agilent ) . Libraries were sequenced on an Illumina HiSeq 2000 DNA Sequencer . mRNA-Seq libraries were constructed using poly ( A ) selection . This creates a potential source of bias , as poly ( A ) -tail length is highly regulated during oogenesis and early embryogenesis , especially for maternally deposited transcripts [10–12 , 14 , 15] However , there is previous evidence that the use of oligo ( dT ) -based poly ( A ) selection does not bias the transcripts recovered . A study [61] measuring both poly ( A ) tail length and transcript level during this period of development found that the dynamic changes in poly ( A ) tail length had minimal impact on the transcript abundance levels measured . To determine if use of oligo ( dT ) -based poly ( A ) selection may have biased the transcript level measurements in our experiment , we examined our mRNA-Seq data from D . melanogaster relative to two datasets of poly ( A ) tail length during early development in the same species [61 , 62] . In comparing the poly ( A ) -tail length of sequenced transcripts in our experiment to the total distribution of poly ( A ) -tail lengths each of these two experiments , we find no difference in distributions ( Wilcox test , p = 0 . 74 compared to [61] , p = 0 . 99 compared to [62] ) . While we cannot rule out that we are recovering a biased subset of transcripts due to oligo ( dT ) enrichment , it seems unlikely that this method produces a substantial bias . Genome and annotation files for the 12 previously sequenced species [21] , downloaded from Flybase [63] , are listed in S15 Table . The D . mauritiana genome and assembly [22] were accessed from a website maintained by the Christian Schlötterer lab at the University of Veterinary Medicine Vienna . The D . miranda genome assembly ( DroMir_2 . 2 ) [20 , 64 , 65] was downloaded from Pubmed and an annotation file was provided by the Doris Bachtrog lab at the University of California , Berkeley . A draft version of the D . santomea genome ( using the non-inbred STO4 line ) , based on data from David Stern’s lab [66] was provided by Peter Andolfatto ( Columbia University ) . We used an annotation of this genome , generated for us by Kevin Thornton ( University of California , Irvine ) , to help construct the phylogeny of the 14 species ( see “Phylogenetic analysis” , below ) . However , since the D . santomea genome was generated using non-inbred flies , we decided to map our D . santomea reads using the flybase D . yakuba assembly and annotation . Reads were pre-processed using Cutadapt [67] to remove adapter contamination . Mapping and differential expression analysis were carried out using the Tuxedo suite [68] , which allows for the discovery of novel isoforms and genes . Briefly , Tophat2 , [69 , 70] , which leverages Bowtie2 [71 , 72] , was used to align reads to the reference and discover new splice junctions for each replicate of each stage and species . An assembly for each stage and replicate was generated using Cufflinks , where upper-quantile normalization was performed using the–N option , and all assemblies for each species were merged with CuffMerge . Using the merged assembly , FPKM levels were calculated with Cuffnorm , and differential expression between stages was assayed using CuffDiff . With the aid of the output files from CuffDiff and CuffNorm , gene FPKM levels were calculated using the total of the FPKM levels for all isoforms of each gene . Assignment of orthologs relied on an orthology table from Flybase ( “gene_orthologs_fb_2014_06_fixed . tsv” ) and the D . mauritiana and D . miranda annotations described above . A table was generated ( S16 Table ) consisting of all genes with one-to-one orthologs in at least 12 of 14 species . If there was no known D . melanogaster ortholog for a gene in a given species , or multiple orthologs , that entry in the table was left blank and was not included in any of the analyses . Spearman rank sum correlation coefficients were calculated using the R statistical environment [73] . When comparing species , only calculated FPKM values for genes with one-to-one orthologs were considered . Correlation plots and hierarchical clustering were generated using the R heatmap2 package . Data from an Aedes aegypti transcriptomic time course [35] were compared to our Drosophila results . We used FPKM values from the 0–2 hour time period ( the earliest available , and the one which is most likely to represent maternal RNA ) and the 8–12 hour period ( which corresponds to the completion of cellularization in Aedes [38] , approximately equivalent to stage 5 in Drosophila ) in our comparison . The data was gleaned from Supplemental S9 Table in Akbari et al . 2013 . Using Inparanoid [74] we found a total of 4619 genes from this dataset that had one-to-one orthologs in Drosophila melanogaster . We expanded S16 Table to include orthologs from Aedes , using the data we had gathered from the Akbari et al . , 2013 study . The phylogeny was constructed using 21 loci , listed in S17 Table . These loci were selected from a previously published list [75] of 250 candidate genes for a Drosphila phylogenetic analysis ( chosen based on low codon usage bias , availability of one-to-one orthologs , and other criteria ) . In selecting the loci for our study , we were limited by the quality of the available D . santomea genome and annotation , which was generated using the non-inbred STO4 line . The loci were aligned using MUSCLE , and regions of low quality in the alignment were removed using trimal . A total of 35 , 829 base pairs were used in generation of the phylogeny , of which 19 , 176 were informative . MrBayes3 . 2 was used to generate the phylogeny and infer ancestral stage 2 and stage 5 states for 8 , 075 genes with orthologs in 12 of the 14 species . Each gene was assigned a binary of state of 1 ( present ) if the FPKM level at a given stage was greater than or equal to the chosen threshold and 0 if it was below the threshold . If there was no one-to-one ortholog in a species , the state was considered unknown . Following previous transcriptomic analyses [35] , an FPKM threshold of 1 was selected . Our dataset thus consisted of 16 , 150 states ( two per species per gene ) in addition to the 35 , 829 nucleotides of DNA . For the binary data , the frequency of state 1 was 0 . 715 , the frequency of state 0 was 0 . 219 , and the frequency of unknown states was 0 . 0656 . MrBayes was run separately 12 times to reconstruct ancestral states on each internal node of the phylogeny ( a MrBayes file for reconstructing the ancestral state for the obscura group is found in S1 Table ) . For the nucleotide data , we used a GTR model and a gamma distribution to model rate variation across sites . Each chain was run for 200 , 000 generations with a burn-in fraction of 0 . 25 . To study changes along the phylogeny , a gain in representation of a gene at a given stage was recorded if an inferred state changed from 0 ( with at least 90% posterior probability ) in the ancestral node to 1 ( was at least 90% posterior probability ) in the derived node , while a change from 1 to 0 ( with 90% posterior probability in each case ) was designated as a loss . Unannotated genes were categorized as those given numbers by the Tuxedo suite but not found in the reference genome annotations ( download from flybase or another source , as described above ) that we provided to the software pipeline . Isoforms were classified as maternal ( M ) if they were present at stage 2 and absent ( below the FPKM threshold of 1 ) at stage 5 , while they were considered zygotic ( Z ) if they were only present at stage 5 . Isoforms that were present at both stages were filtered to select those that showed significant differences between stages ( q value less than 0 . 05 in the CuffDiff output ) . From this set , those where the level at one stage was at least twice that of the other stage were categorized as primarily maternal ( PM ) if stage 2 was higher or primarily zygotic ( if stage 5 ) was higher . All other isoforms that were present at the two stages were classified as maternal-zygotic ( MZ ) . Genes where at least one isoform was primarily maternal and the other was primarily zygotic show evidence of stage-specific isoform usage and were given the ALT classification . Using custom Perl scripts , these genes were identified and the extent of the conservation of the ALT state ( across the 14 species ) was assessed . | Genetic control of embryonic development in all animals requires precise coordination between mother and zygote . The mother provides gene products to the egg to drive the earliest stages of development , until the zygote is able to transcribe its own genome . Many processes of early development are highly conserved over evolutionary time , and are critical for organism survival . Here , we determine how conserved the pools of transcripts provided by the mother and transcribed by the zygote are over evolutionary time , by sequencing all the transcripts at each stage in 14 species of fruit flies at different stages of development . We find a substantial conservation of transcripts represented , with the transcripts deposited by the mother being especially highly conserved . However , we also find considerable variation in these early transcript pools between species , suggesting that while early developmental processes are highly conserved , the gene products driving them may not be . This may be necessary for carrying out the processes of development in different environments . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] | [
"taxonomy",
"invertebrates",
"medicine",
"and",
"health",
"sciences",
"animals",
"invertebrate",
"genomics",
"animal",
"models",
"developmental",
"biology",
"phylogenetics",
"drosophila",
"melanogaster",
"model",
"organisms",
"data",
"management",
"experimental",
"organism"... | 2018 | Evolution of maternal and zygotic mRNA complements in the early Drosophila embryo |
That closely related species often differ by chromosomal inversions was discovered by Sturtevant and Plunkett in 1926 . Our knowledge of how these inversions originate is still very limited , although a prevailing view is that they are facilitated by ectopic recombination events between inverted repetitive sequences . The availability of genome sequences of related species now allows us to study in detail the mechanisms that generate interspecific inversions . We have analyzed the breakpoint regions of the 29 inversions that differentiate the chromosomes of Drosophila melanogaster and two closely related species , D . simulans and D . yakuba , and reconstructed the molecular events that underlie their origin . Experimental and computational analysis revealed that the breakpoint regions of 59% of the inversions ( 17/29 ) are associated with inverted duplications of genes or other nonrepetitive sequences . In only two cases do we find evidence for inverted repetitive sequences in inversion breakpoints . We propose that the presence of inverted duplications associated with inversion breakpoint regions is the result of staggered breaks , either isochromatid or chromatid , and that this , rather than ectopic exchange between inverted repetitive sequences , is the prevalent mechanism for the generation of inversions in the melanogaster species group . Outgroup analysis also revealed evidence for widespread breakpoint recycling . Lastly , we have found that expression domains in D . melanogaster may be disrupted in D . yakuba , bringing into question their potential adaptive significance .
“Eventually the story of the chromosomal mechanisms and its evolution will have to be entirely rewritten in molecular terms” [1] . Over the last century , very detailed studies have been made by cytogeneticists of the intra- and interchromosomal changes that characterize genome evolution in groups as different as mammals ( e . g . , [2] ) and flies ( e . g . , [3]; see [1 , 4] for reviews ) . Chromosome rearrangements are thought to play an important role in reproductive isolation between species [5–7] and in the adaptation of species to their environments [8–10] . These rearrangements may affect fitness by effectively reducing recombination in heterozygotes , thereby preserving co-adapted gene complexes [11 , 12] , or by exerting position effects on loci neighboring breakpoints by modifying gene expression [13] . Only now , with the availability of “complete” genome sequences , can these structural changes in genomes be studied in the molecular detail , as foreseen by Michael White [1] over 30 years ago ( e . g . , [14–16] ) . Genomic sequence data are beginning to reveal a remarkable diversity of patterns of genome rearrangement in different taxa ( [17–21]; reviewed in [22] ) . For example , we see evidence for the recurrent presence of repetitive sequences near breakpoints [23–25] and evidence for the nonrandom distribution of genome breakpoints [16 , 26 , 27] . Moreover , there is evidence that large-scale gene expression domains are maintained as syntenic regions , perhaps because of a functional co-dependency of the genes that reside in these domains [20 , 28 , 29] . Comparative genomic data allow us to reconstruct the state of ancestral genome arrangements at key phylogenetic nodes [17 , 30] and to identify genomic regions conserved during the process of adaptation and divergence [31 , 32] . The genus Drosophila has long been a model for cytogenetic studies of genome evolution . Charles Metz's pioneering comparative studies of metaphase karyotypes in the genus [33] , combined with subsequent comparative genetic studies , led Muller [34] to conclude that the integrity of chromosome arms is largely preserved in the genus Drosophila , despite a 2-fold variation in haploid chromosome number ( see also [35] ) . The maintenance of the gene content of chromosomal arms is due to the paucity of inter-arm rearrangements ( i . e . , pericentric inversions and translocations ) ( [36 , 37]; see [38] for why this is so ) . Sturtevant and Dobzhansky [39] first showed how chromosome inversions can be used to study the evolutionary history of a species group , such as has been shown subsequently in the case of the endemic Hawaiian picture-winged group [3] or in the cactophilic repleta species group of the Americas [40] . Drosophila is a species-rich genus—about 1 , 500 species have been described [41]—and has an evolutionary history of perhaps over 120 million years ( Myr; Figure S1; [42] ) . The wealth of information on genome rearrangement in the genus Drosophila can now be studied at the molecular level , using the genome sequences of 12 different species of Drosophila that are available ( http://rana . lbl . gov/drosophila/ ) . Hitherto , the breakpoint regions of ten well-defined inversions have been characterized in Diptera: eight in Drosophila [25 , 43–49] , and two in Anopheles [50 , 51] . Here we investigate the genome-wide patterns of rearrangement among three closely related species: D . melanogaster , D . simulans , and D . yakuba . D . melanogaster , D . simulans , and D . yakuba are all members of the melanogaster species subgroup , a collection of nine species of Afrotropical origin [52] . D . melanogaster and D . simulans are cosmopolitan sibling species that split from a common ancestor about 5 . 4 Myr ago [42] and can form ( normally infertile ) hybrids . Their polytene chromosome banding patterns are very similar , differing by only one large , and four small , paracentric inversions [53 , 54] . By contrast , D . yakuba , a species of the African savanna , is completely isolated reproductively from D . melanogaster and D . simulans . These three species shared a common ancestor about 12 . 8 Myr ago [42] . The polytene chromosomes of D . yakuba differ from those of D . melanogaster by at least 28 fixed inversions [54] . The combination of prior cytological knowledge of inversion history and the close evolutionary distance among species in this group provides an unparalleled opportunity to reconstruct the detailed molecular events underlying genome rearrangements between animal genomes . We studied the first interspecific inversion ever to be documented , In ( 3R ) 84F1;93F6–7 , which differentiates chromosome 3 of D . melanogaster and the species of the simulans clade [55 , 56] . We characterized its breakpoint regions at the molecular level , i . e . , the genomic regions that encompass both the sites of chromosome breakage and adjacent sequences . We detected inverted duplications of sequences present in the breakpoint regions , a pattern also shown by the breakpoint regions of other chromosomal rearrangements recently characterized [49 , 51 , 57] . One of the breakpoint regions associated with this inversion overlaps that of another inversion that took place on the lineage to D . yakuba , suggesting that some genomic regions are repeatedly broken over time . By a large-scale comparison of the molecular organization of the genomes of D . melanogaster and D . yakuba , we asked if the features associated with inversion In ( 3R ) 84F1;93F6–7 reflect a recurrent pattern of genome rearrangement in the melanogaster species subgroup . We found that approximately 59% ( 17/29 ) of the inversions fixed between D . melanogaster and D . yakuba show evidence of inverted duplication of protein-coding genes or other nonrepetitive sequences present at the breakpoint regions . The prevalence of inverted duplications at inversion breakpoint regions suggests a mechanism of staggered breaks , either isochromatid or chromatid , as the most parsimonious explanation for their origin . Computational analyses failed to find support for the generalized presence of dispersed , repetitive sequences in co-occurrent breakpoint regions , i . e . , those that set the limits of a particular inversion . We conclude that the generation of chromosomal rearrangements in the lineages studied is not necessarily linked to ectopic recombination events between repetitive sequences . We also find evidence for the independent breakage of the same genomic region in different lineages , i . e . , fragile regions [16 , 25–27] , and in one case , we are able , for the first time in Diptera , to reconstruct the reuse of a breakpoint region .
In a remarkable study , Sturtevant and Plunkett [56] deduced from genetic evidence that the chromosomes of D . melanogaster and D . simulans differed by an inversion on the right arm of chromosome 3 . This inversion was later confirmed by an analysis of the polytene chromosomes of the interspecific hybrids ( [58];see also [53] ) . We have directly cloned the breakpoints of this inversion from the genome of D . simulans and , by a combination of experimental and computational methods , characterized the breakpoint regions in the genome sequences of D . melanogaster , D . simulans , and D . yakuba . The structure of the two breakpoint regions of this inversion is illustrated in Figure 1 . To clone the In ( 3R ) 84F1;93F6–7 breakpoints , we performed in situ hybridizations to polytene chromosomes of D . simulans ( and to those of D . melanogaster OR-R as a control ) , using five D . melanogaster bacterial artificial chromosomes ( BACs ) that we expected to cross the breakpoints of the major D . simulans inversion at 84F1 ( BACR07M14 and BACR45A07 ) and at 93F6–7 ( BACR16N15 , BACR42I20 , and BACR08K01 ) [54] . A BAC that includes an inversion breakpoint must necessarily yield two hybridization signals on chromosome arm 3R of D . simulans , but only one on that of D . melanogaster . We determined that BACR07M14 contains the proximal breakpoint and that BACR16N15 contains the distal breakpoint of this inversion . The breakpoints within these BACs were narrowed down by in situ hybridization with probes of genes selected from the predicted cytological coordinates of the breakpoints [54] . We determined that the limits of this inversion were between the protein-coding genes CG2708 and CG7918 , proximally , and CG31176 and CG34034 , distally . The gene pairs CG2708-CG7918 and CG31176-CG34034 delimit two breakpoint regions in D . melanogaster of 22 . 6 and 17 . 8 kilobases ( kb ) long at 84E9–10 and 93E10-F2 , respectively ( Figure 1 ) . Neither region contains any annotated protein-coding genes in the Drosophila genome Release 4 . 3 annotation ( http://chervil . bio . indiana . edu:7092/annot/ ) , with only the non-LTR retrotransposons BS and Cr1a in the region at 84E9–10 as identifiable features [59] . We further characterized the inversion breakpoint regions in D . melanogaster by BLAST analysis and found the presence of four putatively expressed sequences [60] and a sequence said to be related to the mammalian proto-oncogene c-fos ( pfd800 ) . The order of the sequences at these breakpoint regions is , from centromere to telomere: HDC14862-pfd800-HDC12400-Cr1a-BS-HDC14862 at 84E9–10 , and HDC14860-HDC14861-HDC12400-pfd800-HDC14862 at 93E10-F2 ( Figure 1 ) . Notably , three of these sequences ( HDC14862 , pfd800 , and HDC12400 ) are present at both breakpoint regions in an inverted orientation with respect to each other ( Figure 1 ) . The nucleotide identity between duplicated stretches is about 95% across approximately 6 . 3 kb of aligned sequence . Their divergence is greater than the divergence of the Cr1a and BS sequences from the consensus sequences of these elements , 3 . 2% and 0 . 5% , respectively . This suggests that the transposable elements ( TEs ) inserted more recently than the duplication event . The location of the inverted duplicated sequences at both breakpoint regions was confirmed by in situ hybridization . Sequences in this duplicated interval are not found elsewhere in the genome of D . melanogaster , as shown both computationally by BLAST analysis and experimentally by in situ hybridization with appropriate probes . Using probes for the HDC14862 , pfd800 , and HDC12400 sequences , we found that the duplication is also present in the Zimbabwe 2 strain of D . melanogaster , which is from an ancestral population relative to cosmopolitan and laboratory strains [61] , suggesting the duplication is widespread or fixed in D . melanogaster . Furthermore , BLAST analysis against the D . simulans and D . yakuba genomes suggested ( see Materials and Methods ) , and interspecific in situ hybridization confirmed , that the region duplicated in D . melanogaster is present as a single copy in both the D . simulans and D . yakuba genomes . This analysis indicates that the duplication of sequences associated with the breakpoint regions in D . melanogaster represents the derived state relative to that of D . simulans . A similar pattern of inverted duplicated sequences at breakpoint regions has been reported for the polymorphic inversion In ( 3R ) P in D . melanogaster [49] , the polymorphic inversion In ( 2L ) a in Anopheles gambiae [51] , and for the pericentric inversion fixed between Pan troglodytes chromosome 10 and the homologous Homo sapiens chromosome 12 [57] . The comparison of the molecular organization of the breakpoint regions of In ( 3R ) 84F1;93F6–7 between D . melanogaster , D . simulans , and the outgroup species D . yakuba revealed that a second inversion fixed in the lineage that leads to D . yakuba reused one of the In ( 3R ) 84F1;93F6–7 breakpoint regions . In D . yakuba , the CG2708-CG31176 breakpoint region is identical in molecular organization to that of D . simulans , further supporting the hypothesis that In ( 3R ) 84F1;93F6–7 is derived , occurring on the D . melanogaster lineage . In contrast , the gene CG7918 remains adjacent to CG34034 , but in a different chromosomal location from that of CG5849 , which is in turn adjacent to a second copy of CG34034 . In D . simulans , D . erecta , and other distantly related species ( Table 1 ) , the genes CG7918 , CG34034 , and CG5849 are collinear and CG34034 is present in a single copy . In D . yakuba , the gene pairs CG7918-CG34034 and CG34034-CG5849 are found close to the genes CG1315 and CG31286 , respectively . CG1315 and CG31286 are adjacent in D . melanogaster , D . simulans , and other Drosophila species ( Table 1 ) , indicating this to be the ancestral organization for this region . Therefore , the CG7918-CG34034-CG5849 interval has been independently disrupted by another inversion on the D . yakuba lineage , although the precise breakpoints differ from those associated with In ( 3R ) 84F1;93F6–7 . This inversion on the D . yakuba lineage is associated with inverted duplications of CG34034 and CG31286 ( Figure 1; see below ) . The reuse of the breakpoint region CG7918-CG34034 is the second example in Drosophila of recurrent breakage , demonstrated at the molecular level [25] , and is the first in which the associated inversion events can be unambiguously deciphered . The association of inverted duplications with these breakpoint regions is not consistent with a model of inversion origin by recombination between two copies of the same TE [62] . We propose a model of staggered breaks . These breaks may either be isochromatid ( Figures 2 and S2 , see also [57] ) , occurring during premeiotic mitosis , or chromatid , occurring during meiotic prophase ( Figure S3 ) . A potential difficulty of the isochromatid model is the length of DNA that would need to be unwound , presumably by helicase activity . Alternative mechanisms , such as multiple rearrangements or recombination between two independent , but similar , inversions [38] , cannot be ruled out , but they are less parsimonious . In either case , the frequent presence of duplications at co-occurrent breakpoint regions argues against a simple “cut-and paste” mechanism of inversion formation [44] . An important implication of our model is that the presence of inverted duplications at co-occurrent breakpoint regions allows the unambiguous determination of the polarity of chromosome change [49 , 51] . Traditionally , phylogenetic trees of Drosophila based on inversion analysis have been unrooted ( e . g . , [3 , 54] ) . Outgroup analysis can allow the determination of ancestral and derived states , as realized for polytene chromosome inversion phylogenies ( [63]; see also [64] ) , but the widespread signature of inverted duplications provides another independent source of data for polarizing inversion history ( see below ) . In the case of In ( 3R ) 84F1;93F6–7 , four breaks ( a , b , c , and d in Figure 1 ) would have occurred in an ancestral chromosomal arrangement that is now best represented in the D . simulans genome . The breakpoint pairs a-c and b-d ( which have been confirmed by resequencing; Figure 1 ) would each represent staggered breaks within a single chromatid in Figure 2 . CG2708 and HDC14862 overlap by 56–59 base pairs ( bp ) in D . simulans . Breakpoint a occurred at the 5′ end of this overlap , duplicating this region in D . melanogaster . Breakpoint b occurred in the region between HDC12400 and HDC14861 . Breakpoint c occurred downstream of the “exon” 2 of the distal , partial copy of HDC14862 in D . simulans , which roughly corresponds to the intron between “exons” 2 and 3 of the “complete” copy of HDC14862 of D . melanogaster ( roughly upstream of the start of the overlapping region with CG2708 ) . The fourth breakpoint , d , is found 1 , 760–1 , 764 bp downstream of breakpoint c in D . simulans , at 25 bp from the start of the “exon” 1 of HDC14862 . End-filling followed by nonhomologous end joining in the inverted orientation ( Figure 2 ) would result in both the inversion In ( 3R ) 84F1;93F6–7 , the duplication of the region including HDC14862 , pfd800 , and HDC12400 , and the fortuitous formation of what is considered a “complete” copy of the putatively expressed sequence HDC14862 . We used a computational approach to identify genome-wide disruptions in gene order between the chromosomes of D . melanogaster and D . yakuba . Each D . melanogaster transcript was used as a query in a high stringency ( E < 10−30 ) BLASTN search against the genomic sequence of D . yakuba . This allowed us to map unambiguously 12 , 690 genes ( 94 . 4% of those of Release 4 . 1 ) of D . melanogaster on the genome sequence of D . yakuba . A comparison of the gene orders of the two species identified 55 gene-order disruptions between them , which appear as discontinuities in the coordinates of neighboring genes in one species relative to the other ( Tables 1 and S1 ) . All predicted gene-order disruptions identified using this gene-based BLAST approach are also identified as termini of whole-genome global alignments at the University of California , Santa Cruz ( UCSC ) [65] . These 55 gene-order disruptions define 59 syntenic blocks between these species ( since both species have four chromosomes ) ( Table S2 ) . The location and relative orientation of the syntenic blocks for chromosome 2 of D . melanogaster and D . yakuba are shown in Figure 3; similar data are shown for chromosomes X and 3 in Figure S4 . We do not show the small chromosome 4 ( syntenic block 59 ) , since our results indicate that this chromosome is wholly collinear in the two species over the sequenced region [66] . Syntenic blocks 13 , 26 , and 46 include the centromeric heterochromatic regions for chromosomes X , 2 , and 3 , respectively . We are unable , given the present sequence data , to detect any chromosome rearrangements within these heterochromatic regions or those on chromosome 4 . To obviate possible artifacts of the assembly process ( see Material and Methods ) on our results , and directly confirm our predictions of the gene order around the D . yakuba breakpoint regions relative to those of D . melanogaster , we cloned and sequence verified a sample of 27 of the predicted breakpoint regions from D . yakuba , each containing the transition between adjacent syntenic blocks ( see Materials and Methods ) . In every case , our predictions were directly confirmed ( Table S1 ) . This result is consistent with the fact that all predicted gene-order disruptions are found in high-quality , contiguous ( i . e . , ungapped ) regions of the D . yakuba assembly . In fact , breakpoint regions in D . yakuba are sequenced to an average depth of 8× and are supported by an average of 14 clone pairs . These results demonstrate that the gene-order disruptions inferred between the D . yakuba and D . melanogaster genomes are not assembly artifacts . Approximately 117 . 8 megabases ( Mb ) of the D . melanogaster genome and about 118 . 9 Mb of the D . yakuba genome are included in the 59 syntenic blocks as defined by their outermost markers or reference genes . The amount of nonheterochromatic DNA not included in these syntenic blocks is 542 kb of the D . melanogaster genome and 674 kb of the D . yakuba genome . This is an upper estimate because in some cases , there is noncoding homology between the reference genes that define two consecutive syntenic blocks ( see below ) . The median size of syntenic blocks is 1 . 66 Mb in D . melanogaster , and 1 . 61 Mb in D . yakuba . Excluding the syntenic blocks that contain centromeric heterochromatin ( blocks 13 , 26 , and 46 ) , the largest ( syntenic block 57 ) is just over 6 Mb ( ~5 . 2% of the genome in both species ) , and the smallest is 161 kb ( syntenic block 22 , 0 . 08% of the D . melanogaster genome; and syntenic block 25 , 0 . 08% of the D . yakuba genome ) . The length of genomic regions in each syntenic block is highly correlated across species ( Spearman ρ = 0 . 997 , p = 3 . 78 × 10 −61; blocks 13 , 26 , and 46 not included ) , and in only two cases ( blocks 26 and 43 ) , do they differ by more than 10% . The DNA content per syntenic block does not differ significantly between D . melanogaster and D . yakuba ( Wilcoxon signed rank test , Z = −1 . 273 , p = not significant [n . s . ]; blocks 13 , 26 , and 46 not included ) . A departure of the observed distribution of the lengths of syntenic blocks from that expected if the breakpoints were randomly distributed across the genome ( a truncated negative exponential distribution ) would allow us to discard the random breakage model of chromosome evolution [26 , 67] . Based on the comparison of the empirical and theoretical distributions , we cannot reject the random breakage model ( Kolmogorov-Smirnov test , D = 0 . 2 , p = n . s . ; blocks 13 , 26 , and 46 not included ) . Despite the conservative criteria used in our BLAST analysis , its resolution is sufficient to detect gene sequences that may have “escaped” synteny by transposition , as has been observed in Drosophila both experimentally , e . g . , [68] , and by genomic analyses [69–71] . We detected 22 potential transposition events between D . melanogaster and D . yakuba , with 12 occurring unambiguously between chromosome arms and eight events within chromosome arms ( Tables S3 and S4 ) . This number is likely to be an underestimate because we used stringent criteria for paralogy . Of the 22 events that we detected , 20 are duplicative transpositions and two are conservative transpositions . Muller [34] defined the six fundamental elements of the karyotype of the genus Drosophila ( now referred to as Muller's elements A–F , each corresponding to a chromosome arm of D . melanogaster ) . The overall gene content of these elements has been conserved during the evolution of the genus as witnessed by the very few inter-element rearrangements ( i . e . , pericentric inversions and translocations ) that have been reported . Previous analysis of inversion differences between D . melanogaster and D . yakuba based on polytene chromosome revealed 28 inversions , of which only one , on chromosome 2 was pericentric [54] ( Table 2 ) . We established which pairs of breakpoint regions define particular inversions by taking into account the contiguity relationships in both species of the outermost genes of syntenic blocks between D . melanogaster and D . yakuba ( Figures 3 and S4; Table S1 ) . In general , our computational analysis of the genome sequences of these two species is broadly compatible with previous results based on polytene chromosomes [54] . We inferred that 29 inversions distinguish the chromosomes of D . melanogaster and D . yakuba , of which 28 are paracentric and one corresponds to the pericentric inversion on chromosome 2 ( Table 2 ) . The total number of inversions inferred computationally is just one more than that suggested by polytene chromosome analysis [54] , although the greater resolution of the sequence analysis increases the number of breakpoints from 48 to 55 and refines their positions ( Tables 1 and 2 ) . Our analysis shows many discrepancies in detail when compared to previous work ( [54]; Tables 1 and 2 ) . This is especially true on the X chromosome , where the banding pattern has diverged greatly in the melanogaster species group . On chromosome 2 , there is what Lemeunier and Ashburner [54] interpreted as a single pericentric inversion , which distinguishes D . yakuba and its relatives , D . teissieri , D . erecta , and D . orena , from D . melanogaster and the three species of the D . simulans clade . As shown in Figure 3 , there is a complex mosaic of syntenic blocks between the two arms of chromosome 2 . In good agreement with the previous work [54] , a single pericentric inversion , 2LR ( 5 ) , is sufficient to explain this pattern . This inversion has identical limits in both D . yakuba and D . erecta . Inverted duplications at the breakpoint regions in both species ( Table S5 , see below ) and information on gene order in other outgroup species ( Table 1 ) strongly suggest that this inversion occurred in the common ancestor of D . yakuba and D . erecta after this lineage split from that leading to the melanogaster-simulans complex . Figure S6 illustrates one of the most parsimonious scenarios that explains the evolution of chromosome 2 . We characterized in detail the sequences of the 55 breakpoint regions of D . yakuba because genomic and phylogenetic evidence suggested that virtually all inversion events between D . melanogaster and D . yakuba occurred on the D . yakuba lineage ( Table 1; see below ) . Remarkably , in 34 of 55 ( approximately 62% ) breakpoint regions , we detected the presence of duplications of sequences that are only present once in the genome of D . melanogaster . In each case , these duplications are specifically associated with the pair of breakpoint regions that limit a particular inversion ( Table S5; see below ) . These duplications are not repetitive in the D . yakuba genome ( by BLAST analysis ) , nor do they match any identifiable Drosophila TE . In a control experiment , the genomic regions of D . melanogaster that correspond to the co-occurrent breakpoint regions of D . yakuba were compared to each other . Repetitive sequences were found in six cases; in no case other than that of In ( 3R ) 84F1;93F6–7 ( see Figure 1 ) were duplications of unique sequences found . In total , 18 of 29 inversions ( approximately 62% ) fixed between D . melanogaster and D . yakuba are associated with duplications of sequences included at co-occurrent breakpoint regions . These duplicated sequences are in opposite orientations in the co-occurrent breakpoints of 17 inversions; 3R ( 6 ) is the only exception , potentially as a result of a subsequent microinversion [72] . These sequence duplications include 22 full or partial duplications of protein coding genes . Most of these ( exceptions are CG14817 at Xy ( 1 ) and Xy ( 4 ) , CG6081 at 2y ( 15 ) and 2y ( 18 ) , and CG34034 at 3y ( 46 ) and 3y ( 53 ) ) have accumulated many point and indel mutations , and are presumed to be nonfunctional . The average nucleotide identity ( ± the standard deviation [SD] ) between duplicates is approximately 88% ± 5 . 4% . For six of the inversions , sequences from both breakpoint regions are present as inverted duplications at each breakpoint . For the remaining 12 inversions , sequences from only one of the two breakpoint regions are duplicated . This may be due either to the evolutionary loss , by sequence change , of one of the copies of an original duplication , or to the fact that only one of the pair of single-stranded breaks was significantly staggered ( Figure S5A and S5B , respectively ) . The size of the duplications varies significantly in D . yakuba ( median = 321 bp , coefficient of variation [CV] = 81% counting only one of the copies when in tandem; Table S5 ) , but in no case do they involve more than about 1 . 9 kb of aligned sequence ( the shortest duplication is 46-bp long ) . In many taxa , repeated sequences have been found to be associated with rearrangement breakpoints and have been implicated in mediating chromosomal rearrangements by a process of ectopic exchange . This has been the case for tRNAs and ribosomal protein genes in yeasts [73 , 74] , segmental duplications in the human-mouse [24] and human-primate lineages [75–78] , and TEs in many organisms [46 , 79–81] . In D . melanogaster , there is abundant experimental evidence that exchange between TEs can result in chromosome rearrangement ( e . g . , [82] ) . Comparative sequence data also indicate that TEs are abundant at interspecific breakpoint regions between Diptera species [25 , 69] , and there is strong evidence implicating TE-mediated ectopic exchange events in four [25 , 46 , 47 , 51] of the ten well-defined inversions whose breakpoint regions have been characterized at the molecular level ( Table 3 ) . We analyzed the breakpoint regions of D . yakuba for TE sequences using RepeatMasker with the Release 4 . 2 TE annotation of the D . melanogaster genome [83] and by BLAST2 analysis using as a query TEs sequences from species other than D . melanogaster . Over 45% of breakpoint regions ( 25/55 ) include repetitive sequences in D . yakuba ( Table S6 ) , but only five co-occurrent pairs of breakpoint regions ( involving inversions 2LR ( 5 ) , 2L ( 6 ) , 2LR ( 8 ) , 3L ( 3 ) /3L ( 4 ) , and 3R ( 6 ) ) include a similar repetitive sequence ( Table S6 ) . These analyses would fail to detect any repetitive sequence absent from the RepeatMasker library ( as would be those exclusive to D . yakuba ) or not yet characterized in D . yakuba . For this reason , we manually extracted from the D . yakuba breakpoint regions a set of sequences , each corresponding to the precise transition region between syntenic blocks , and used them as BLAST queries to the entire D . yakuba genome . Similar repetitive sequences were found at the co-occurrent breakpoints of the inversions X ( 1 ) , 2L ( 6 ) , 3L ( 5 ) , and 3R ( 7 ) , although only in the case of 2L ( 6 ) and 3R ( 7 ) are the copies of the repetitive sequence inverted with respect to each other . The average length of these sequences was 685 bp and the range 49–3 , 037 bp . Unfortunately , we can neither date the insertion of these repetitive sequences ( with respect to the time of occurrence of the inversion ) , nor can we assert that the absence of repetitive sequences at other pairs of co-occurrent breakpoint regions is not due to their decay or loss subsequent to the occurrence of an inversion . Nevertheless , these data provide little direct evidence for the presence of TEs in generating fixed inversions between D . melanogaster and D . yakuba and , combined with the recurrent presence of inverted duplications of nonrepetitive sequences , suggests that ectopic recombination between TEs has not been the dominant mechanism of generating inversions in this lineage . These results contrast with the presence of inverted TEs at co-occurrent breakpoints of well-defined inversions ( Table 3 ) . We mapped the derived state of the 29 inversions between the two genomes to the D . melanogaster or D . yakuba lineages , using several independent criteria ( Table 1 ) : ( 1 ) by determining the arrangement of each gene pair disrupted by an inversion in D . melanogaster versus D . yakuba in five other sequenced Drosophila species; ( 2 ) by the presence of inverted duplications associated with co-occurrent breakpoints , as discussed above; and ( 3 ) by the disruption of a tandem array of related genes , or of a pair of genes whose transcripts show 3′-overlap ( see below ) , which we also consider to be a derived state . In all cases in which we can use more than one of these criteria , all are consistent . Our analyses show that of 29 inversions , 28 have been fixed in the lineage leading to D . yakuba , and only one ( 3R ( 8 ) , also known as In ( 3R ) 84F1;93F6–7 ) on the lineage leading to D . melanogaster ( eight of the former inversions occurred before the D . erecta/D . yakuba split ) . This difference is highly significant ( one-tailed binomial p = 5 . 59 × 10−8 ) and agrees well with previous interpretations [64] , demonstrating that rates of chromosomal evolution can vary by over an order of magnitude even among closely related species . The origin of this very asymmetric rate of fixation cannot stem from differences in the degree of intraspecific polymorphism , as has been proposed for D . pseudoobscura and D . subobscura [84] , because D . melanogaster is substantially more polymorphic for inversions than D . yakuba [54] . Rather , it might reflect different effective population sizes between the African populations of the immediate ancestors of D . melanogaster and D . yakuba [85 , 86] . We used the number of breakpoints per Mb per Myr to correct for differences in chromosomal size in a comparison of rates of chromosomal evolution between species pairs of different Drosophila groups ( Table 4 ) in which we assumed a constant rate of evolution as a null hypothesis . In view of the pericentric changes in chromosome 2 ( Muller's elements B+C ) , we combined the data for these elements . The overall rate of breakage in the D . melanogaster/D . yakuba lineage is 0 . 0183/Mb/Myr . This is slower than that seen in the D . pseudoobscura/D . miranda ( Gadj = 38 . 9; d . f . = 1; p < 4 . 4 × 10−10 ) and D . pseudoobscura/D . subobscura ( Gadj = 48 . 5; d . f . = 1; p < 3 . 4 × 10−12 ) comparisons , comparable with the rate seen in the comparison D . virilis/D . montana ( Gadj = 0 . 5; d . f . = 1; p = n . s . ) and accelerated with respect to that in the repleta species group ( Gadj = 4 . 3; d . f . = 1; p < 4 . 3 × 10−9 ) . Across Muller's elements , the rank order of the rate of chromosome evolution is A > ( B+C ) > E > D , which agrees well with the genus-wide pattern of rates of evolution A > E > D proposed by [87] , based on the comparisons of D . melanogaster and D . repleta [21 , 87] and of D . virilis , D . montana , and D . novamexicana [88] . Nevertheless , Muller's elements B+C appear to have evolved faster in the D . melanogaster/D . yakuba lineage than in D . melanogaster/D . repleta , in which element B was the slowest evolving [87] . Thus , in addition to rate variation among lineages , rates of chromosomal evolution may vary across Muller's elements in different groups of Drosophila , in good agreement with , for example , the fast evolution of the Muller's element E across the repleta species group [40] . Breakpoint reuse has been reported at the cytological [54 , 89–91] and the molecular level [16 , 25–27 , 92] . Based on our phylogenetic reconstruction of the chromosomal rearrangements of the species considered here ( Table 1 ) , it is clear that some ancestral gene configurations have been disrupted independently more than once during the evolution of the subgenus Sophophora . Using sequences from D . ananassae , D . persimilis , and D . pseudoobscura as outgroups to the D . melanogaster species subgroup , we found evidence for breakage in 17 out of the 55 ( ~31% ) regions disrupted in the D . melanogaster/D . yakuba lineage . We also see evidence for nonrandom breakage in the D . melanogaster/D . yakuba complex , i . e . , at a relatively short phylogenetic distance . For each of the three pairs of inversions 3L ( 3 ) /3L ( 4 ) , 3R ( 7 ) /3R ( 8 ) , and 3R ( 10 ) /3R ( 11 ) , three , instead of four , breakpoint regions are involved . This recurrent breakage might denote structural instability of particular genomic regions . For example , CG9579 , one of the genes adjacent to the breakpoints of the inversion X ( 5 ) , is also linked to a remarkable set of molecular reorganizations associated with the birth of a multigene family of a chimeric gene , Sdic , on the D . melanogaster lineage [93] . Additional support for structural instability of inversion breakpoint regions comes from the fact that one breakpoint region of inversion 2LR ( 4 ) , which occurs on the D . yakuba lineage , uses the same genomic interval that has independently permitted the recent evolution of an unusually high TE density in the D . melanogaster lineage ( HDR13 in [94] ) . A related issue to breakpoint reuse is the possibility that the same inversion can arise twice . The unique origin of inversions has been challenged ( see [37] for discussion ) , but in the two cases considered to be the most convincing , experimental evidence has not supported a polyphyletic origin of inversions [51 , 92] . Fourteen breakpoint regions are associated with shared inversions between D . yakuba and D . erecta ( Table 1 ) , which indicates that the same gene pairs have been disrupted and reorganized in the same way , suggesting a common origin in the ancestor of D . erecta and D . yakuba . Comparative sequence analysis at the nucleotide level for those 14 junctions failed to find evidence of an independent origin of these inversions in the lineages that lead to the D . yakuba and D . erecta , although it must be noted that our power of detection can be compromised by the time elapsed since D . yakuba and D . erecta shared an ancestor . Expression profiling of the genomes of several species has shown that co-expressed genes tend to co-locate in the genome ( for review , see [28] ) . The biological significance of co-expression clustering is still poorly understood , but if these “transcriptional territories” represent functional associations among neighboring genes , natural selection should prevent their disruption . Conservation of clusters across lineages differentiated by the accumulation of multiple chromosomal rearrangements has been interpreted as support for the functional association of clusters of co-expressed genes in mammals [95] and flies [29] . In D . melanogaster , the preferential clustering of genes , by the time or place of their expression , has been reported based on both expressed sequence tag ( EST ) and microarray data [96–99] . In a study of the distribution of sex-biased gene expression [97] , 75% of the genes on Release 3 . 1 of the D . melanogaster genome were assayed . Fifteen gene clusters that are expressed either in testis , in ovary , or in the soma were found . Despite the relatively small number of gene-order interruptions between D . melanogaster and D . yakuba , one of the clusters identified by Parisi et al . [97] , containing the Try multigene family , is broken in the lineage of D . yakuba by inversion 2LR ( 8 ) . At least eight out of ten members of the disrupted gene cluster are highly expressed in the soma . The disruption of this transcriptional territory may be related to the fact that the chromosomal breakage occurred between a member of the cluster , CG12388 ( kappaTry ) , which is soma-biased in expression , and CG12387 ( zetaTry ) , which is not . Transcriptional territories have been found to be correlated with the DNA replication program in D . melanogaster [100] . Specifically , 7 . 5% of the D . melanogaster genome , distributed in 52 well-defined regions , is under-replicated in polytene chromosomes , and 50 of these regions also replicate late during the S period in cultured Kc cells; other regions present a non-delayed replication status in at least one of the two tissues . Sixty percent ( 30/50 ) of these late or under-replicating regions are associated with previously defined transcriptional territories; these domains account for 20% of the D . melanogaster genome [98] . Globally , transcriptional territories with a delayed pattern of DNA replication seem to be enriched for genes expressed in the testis and during pupal development , and depleted of genes expressed in the ovary and embryonic development [100] . Are the 55 gene pairs disrupted by inversion breakpoints in the D . melanogaster/D . yakuba lineages randomly distributed across the genome with regard to their replication status ? We did not find a significant deviation from the random expectation ( Gadj= 5 . 29; d . f . = 3; p = 0 . 15 ) ; however , we did find that three out of the 53 ancestral gene pairs disrupted in D . yakuba ( Xm ( 8 ) , 2m ( 19 ) , and 3m ( 45 ) ) are embedded in regions that are under-replicated in salivary glands and late replicated in Kc cells . These results show that at least some of the regions of the D . melanogaster genome , within which genes have a similar expression profile and/or replication program , are not necessarily conserved between this species and D . yakuba . This suggests that either those domains have little adaptive value , supporting the idea of accidental co-expression , or that their adaptive value has evolved recently , relative to the time of the divergence between D . melanogaster and D . yakuba . Some 1 , 027 pairs of genes in D . melanogaster have overlapping transcripts in opposite strands [101] . Antisense overlap can play an important role in regulating gene expression at the post-transcriptional level [102 , 103] . Five of these genes pairs are disjunct in D . yakuba , as a consequence of an inversion breakpoint . Comparison across lineages ( Table 1 ) indicates that the disruption in D . yakuba represents the derived state . The five inversions that disrupt antisense pairs are all associated with inverted duplications ( Table S5 ) . Our model for the origin of inversions ( Figure 2 ) can account for the conservation of sequences of decoupled antisense pairs of genes . At least in two of these cases ( CG9578-CG9579 and CG31142-CG5289 ) , the 3′ UTR sequences of the independent gene pairs of D . yakuba are very similar in sequence and in length to their corresponding 3′ UTRs in D . melanogaster . In the other three cases , the D . yakuba 3′ UTR of one of the members of each pair is truncated . This work unveils novel aspects of the evolution of the molecular organization of the Drosophila genome in particular and of the genomes of insects in general . The use of genome sequence data of D . melanogaster and D . yakuba has proven to be useful in reconstructing the history of genome rearrangements in these species . The lineage that leads to D . yakuba is evolving substantially faster at the chromosomal level than D . melanogaster ( 28:1 ) ; nevertheless , the mechanism that underlies the generation of many inversions ( ~59% ) in both lineages is the same , and it seems to be initiated by the presence of staggered breaks , which in turn enables the generation of duplications in inverted orientation of sequences at co-occurrent breakpoint regions . These duplications diverge mainly by both nucleotide substitutions and small deletions [104 , 105] , and can contribute , as do segmental duplications in mammals , to the diversification of gene function [106] . A model of inversion generation based on staggered breaks , either isochromatid or chromatid , contrasts with a model of ectopic recombination between repetitive sequences [46 , 75 , 76] . Our data also give clear evidence , at the molecular level , of the reuse of the same breakpoint region and that expression domains in D . melanogaster may be disrupted in other species , bringing into question their potential adaptive significance . The availability of complete sequences from 12 Drosophila species now offers the opportunity to extend the analysis of chromosome evolution at a molecular level . Several fundamental questions remain: whether or not mechanisms of inversion formation are general across taxa; and whether there are functional constraints on chromosomal evolution , and , if so , at what level do these operate .
The following species and strains were used: D . melanogaster ( OR-R from the Department of Genetics , University of Cambridge , and Zimbabwe 2 from D . L . Hartl's laboratory ) ; D . simulans ( Sim-1 from Chapel Hill , North Carolina ) ; and D . yakuba ( Tai18E2 from the Tucson Stock Center ) . In the case of Zimbabwe 2 and Tai18E2 , we checked whether they were homokaryotypic by visually examining salivary gland polytene chromosome preparations stained with orcein . In the case of Zimbabwe 2 , we detected two paracentric inversions in a sample of 20 autosomal genomes and 16 X chromosome genomes . No gross chromosomal polymorphisms were detected in a sample of 20 autosomal genomes and 16 X chromosome genomes of Tai18E2 . Five BACs and 11 genomic clones were used as molecular probes . The BAC clones ( BACR07M14 , BACR45A07 , BACR16N15 , BACR42I20 , and BACR08K01 ) were obtained from the Children's Hospital Oakland Research Institute . Genomic clones were PCR amplified using the primers described in Table S7 . The genomic DNA used for the PCR amplifications was from the sequenced strain of D . melanogaster: y; cn bw sp [107] . The genomic fragments generated correspond to the protein-coding genes CG2708 ( Tom34 ) , CG7918 , CG31176 , CG34034 , CG5289 , and CG6576 ( Glec ) ; the putatively transcribed genes HDC14860 , HDC14861 , HDC14862 , and HDC12400 [60]; and the sequence of pdf800 , which is said to be related to the mammalian proto-oncogene c-fos . Cloning of PCR products and preparation of DNA from recombinant clones was performed using conventional methods . In the case of BAC clones , we used the methods described at http://bacpac . chori . org/bacpacmini . htm . In situ hybridization of probes to polytene chromosomes was done as in [108] . Detection of the hybridization signals was done by phase contrast with a Zeiss Axioskop 2 ( Carl Zeiss , http://www . zeiss . com ) . Chromosomal localization was determined using the photographic polytene chromosome maps of D . melanogaster [109] . All the probes yielded one or two hybridization signals with the exception of those for HDC14860 and HDC14861 , which failed to generate a detectable hybridization signal in D . yakuba under the experimental conditions used . The sequencing and assembly of the D . yakuba genome will be described elsewhere ( D . J . Begun , A . K . Holloway , K . Stevens , L . W . Hillier , Y . -P . Poh , M . W . Hahn , P . M . Nista , C . D . Jones , A . D . Kern , C . Dewey , L . Pachter , E . Myers , and C . H . Langley , unpublished data ) . To create chromosomal assignments and ordering of “supercontigs” ( gapped scaffolds of ungapped contigs as defined by mate pairs ) along the chromosomes for the D . yakuba genome assembly , contigs from the D . yakuba assembly that uniquely aligned with the D . melanogaster genome were identified and then ordered by their positions along the assigned D . melanogaster chromosomes . This process resulted in some D . yakuba supercontigs with contigs that aligned to different regions of a D . melanogaster chromosome . To assemble supercontigs into chromosome arms in D . yakuba , reversals of the tiling path of mapped contigs were introduced to “rejoin” those supercontigs that had been split by the alignments to D . melanogaster . The overall goal was to minimize the total number of reversals required to rejoin all D . yakuba supercontigs previously assigned to disjoint chromosomal regions based on D . melanogaster alignments . We note that reversals were introduced only between contigs ( not within contigs ) and the process was not gene based . The complete set of transcripts of the D . melanogaster Release 4 . 1 annotation was downloaded from UCSC Genome Browser ( http://genome . ucsc . edu/ ) . This set represents 13 , 449 annotated genes . Each D . melanogaster transcript was used as a query against the assembly of the D . yakuba genome release 2 . 0 ( WUSTL November 2005 , the droYak2 assembly ) using BLASTN 2 . 2 . 2 with default settings and then filtered for the top hit for each transcript with a cutoff E-value of 10−30; the nonfiltered output can be found as Table S8 . This approach localized 12 , 690 genes on the genome sequence of D . yakuba with a best hit on the same chromosome arm ( with exceptions made for genes inside the pericentric inversion on chromosome 2 ) ; 320 genes had no BLASTN hit higher than 10−30 , and 429 genes hit unmapped scaffolds or gave multiple hits with equal E-value in more than one chromosome arm . Genes unambiguously localized were sorted into chromosome order ( centromere to telomere ) for the six Muller's elements of D . yakuba . The gene order in D . yakuba was compared with that of D . melanogaster , and gene-order interruptions between the two species were inferred; the two genes flanking each gene-order interruption were taken as the limits of different syntenic blocks . This method will not reliably detect very small rearrangements , although we know that these occur ( e . g . , Figure S7; see also [72] ) . For calculating the minimum number of inversions necessary to transform the gene order of D . melanogaster into that of D . yakuba , we used GRIMM [110] . Estimates on the size of syntenic blocks and regions between them in D . yakuba were obtained by taking into account the coordinates of the BLASTN hits of the outermost markers of each syntenic block . In the case of transposition events , we examined the nonfiltered output for genes whose BLAST hits were surrounded by different pairs of flanking genes in D . melanogaster and D . yakuba , especially those with unambiguous hits in different Muller's elements . One complicating factor in our analysis is that BLASTN of a region including 3R:3862326–3867817 was highly similar to two different regions of the D . yakuba assembly: one on Contig690 ( currently assembled into chromosome arm 3R ) , and one , with a slightly lower match , on Contig706 ( currently assigned to the “random” bin of chromosome arm 3R because it seemed to overlap Contig690 ) . Contig690 has a sequence coverage of 5 . 8–8 . 3× , Contig706 of 3–4 . 7× . The overall coverage of the genome is 9 . 4× , but the supercontigs of chromosome arms 2R and 3R have approximately 12× coverage . Were this region to be truly duplicated in the genome of D . yakuba , we would expect the sum of the coverage of Contigs 690 and 706 to be at the very least 18× , rather than ( at most ) 13× . In situ hybridization to polytene chromosomes of probes from this region shows only a single site , that expected on chromosome arm 3R . Residual heterozygosity for other regions of the D . yakuba sequence has been experimentally verified ( J . Comeron and C . Langley , personal communication ) , and we interpret these two hits as being the consequence of heterozygosity in the genome . To confirm the predicted gene-order interruptions between D . melanogaster and D . yakuba , we cloned and sequence verified the transition between adjacent syntenic blocks of 27 ( 49% ) of the breakpoint regions in D . yakuba , namely Xy ( 9 ) , Xy ( 10 ) , 2y ( 19–24 ) , 2y ( 26–28 ) , 3y ( 35–43 ) , 3y ( 46–51 ) , and 3y ( 53 ) ( Table S1 ) . We extracted genomic DNA from the sequenced strain Tai18E2 by conventional methods . We designed primers to amplify the sequence that spans the transition between syntenic blocks . In a few cases , either because of the size of the region between the neighboring reference genes or because of technical difficulties , we amplified sets of overlapping segments that ensured coverage of the transition between adjacent syntenic blocks . PCR products were cloned into a pCR2 . 1 Topo Vector ( Invitrogen , http://www . invitrogen . com ) . Sequencing reactions of the two ends of each clone were done , and the reads were aligned by BLAST against the D . melanogaster genome . Primers used are listed in Table S7 . Because not all the genes of D . melanogaster were mapped to the D . yakuba assembly , and because there may have been transpositions of regions during the evolution of these genomes , we extracted the sequences of the 55 genomic discontinuities of D . yakuba , relative to D . melanogaster , and aligned these by BLASTN against the D . melanogaster genome . This refined the limits of the syntenic blocks and allowed their ends to be precisely mapped . To identify duplicates at co-occurrent breakpoint regions , we used PipMaker [111] , and BLAST2 [112] with their default parameters . Sequences from all local alignments spanning more than 40 bp from PipMaker were used as queries in a BLASTN analysis against the D . melanogaster genome , thereby verifying their identities and genomic locations . We did the same with the BLAST2 output for those sequences with hits whose E-value were lower than 10−8 and were at least 40-bp long . Both approaches provided essentially the same results . Nucleotide identities between particular duplicates and their reference sequences were derived from the BLAST2 analysis . For genes that are adjacent to breakpoints and/or are affected by them , we did an additional BLAST2 analysis , using as queries the D . melanogaster sequences of their transcripts . Sequences that are now found as inverted duplications at co-occurrent breakpoint regions may not necessarily have been in this orientation immediately after the occurrence of the inversion , because subsequent events may have taken place . For this reason , we reconstructed the most parsimonious history of each inversion in an attempt to establish the sequence immediately after each had occurred . We analyzed the presence of TE sequences using the RepeatMasker track from UCSC ( RepBase libraries: RepBase Update 9 . 11 and RM database version 20050112 ) and subsequently by BLAST2 analysis using a collection of TE sequences that includes those in different Drosophila species other than D . melanogaster . All the significant hits found by our BLAST2 analysis correspond to footprints of TEs of D . melanogaster previously detected with RepeatMasker . For duplications that spanned noncoding regions , we did a BLASTN analysis against the D . yakuba genome , in order to determine that they did not include repetitive sequences . When necessary , we proceeded in an identical manner with breakpoint regions of D . melanogaster , D . simulans , and D . erecta . In order to determine whether the gene configuration in the breakpoint regions in D . melanogaster or in D . yakuba is ancestral or derived , i . e . , the result of a chromosomal rearrangement , we took D . melanogaster as a reference , and we determined whether or not the reference genes within a particular breakpoint region were adjacent in a set of species selected on the basis of their phylogenetic relationships with D . melanogaster and D . yakuba . Specifically , we used: D . melanogaster ( Release 4 . 1; FlyBase ) ; D . simulans ( release 1 . 0 Apr . 2005; UCSC ) ; D . yakuba ( droYak2 Nov . 2005 ) ; D . erecta ( droEre1 Aug . 2005; UCSC ) ; D . ananassae ( droAna2 Aug . 2005; UCSC ) ; D . persimilis ( droPer1 Oct . 2005 UCSC ) ; and D . pseudoobscura ( Release 1 . 0; S . W . Schaeffer , personal communication ) . We used PipMaker to analyze the breakpoint regions apparently shared between D . yakuba and D . erecta . If these breakpoint regions were of independent origin , then we would expect to see discontinuities and indels between them . In fact , in all cases , the evidence suggests that these “shared” breakpoints were the consequence of a single ancestral event .
The GenBank ( http://www . ncbi . nlm . nih . gov/Genbank ) accession number for the D . melanogaster DNA sequence pfd800 discussed in this paper is Z16407 . The accession numbers for the sequences generated in this paper are EF569486–EF569554 . | The organization of genes on chromosomes changes over evolutionary time . In some organisms , such as fruit flies and mosquitoes , inversions of chromosome regions are widespread . This has been associated with adaptation to environmental pressures and speciation . However , the mechanisms by which inversions are generated at the molecular level are poorly understood . The prevailing view involves the interactions of sequences that are moderately repeated in the genome . Here , we use molecular and computational methods to study 29 inversions that differentiate the chromosomes of three closely related fruit fly species . We find little support for a causal role of repetitive sequences in the origin of inversions and , instead , detect the presence of inverted duplications of ancestrally unique sequences ( generally protein-coding genes ) in the breakpoint regions of many inversions . This leads us to propose an alternative model in which the generation of inversions is coupled with the generation of duplications of flanking sequences . Additionally , we find evidence for genomic regions that are prone to breakage , being associated with inversions generated independently during the evolution of the ancestors of existing species . | [
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods",
"Supporting",
"Information"
] | [
"evolutionary",
"biology",
"drosophila",
"genetics",
"and",
"genomics"
] | 2007 | Principles of Genome Evolution in the Drosophila melanogaster Species Group |
Varicose veins of lower extremities ( VVs ) are a common multifactorial vascular disease . Genetic factors underlying VVs development remain largely unknown . Here we report the first large-scale study of VVs performed on a freely available genetic data of 408 , 455 European-ancestry individuals . We identified the 12 reliably associated loci that explain 13% of the SNP-based heritability , and prioritized the most likely causal genes CASZ1 , PIEZO1 , PPP3R1 , EBF1 , STIM2 , HFE , GATA2 , NFATC2 , and SOX9 . VVs-associated variants within these loci exhibited pleiotropic effects on several phenotypes including blood pressure/hypertension and blood cell traits . Gene set enrichment analysis revealed gene categories related to abnormal vasculogenesis . Genetic correlation analysis confirmed known epidemiological associations between VVs and deep venous thrombosis , weight , rough labor , and standing job , and found a genetic overlap with multiple traits that have not been previously suspected to share common genetic background with VVs . These traits included educational attainment , fluid intelligence and prospective memory scores , walking pace ( negative correlation with VVs ) , smoking , height , number of operations , pain , and gonarthrosis ( positive correlation with VVs ) . Finally , Mendelian randomization analysis provided evidence for causal effects of plasma levels of MICB and CD209 proteins , and anthropometric traits such as waist and hip circumference , height , weight , and both fat and fat-free mass . Our results provide novel insight into both VVs genetics and etiology . The revealed genes and proteins can be considered as good candidates for follow-up functional studies and might be of interest as potential drug targets .
Varicose veins ( VVs ) are one of the clinical manifestations of chronic venous disease posing both a cosmetic and medical problem . VVs can be found in different parts of the body , but most commonly occur in the lower extremities . Prevalence estimates of this condition vary across ethnic groups ranging from 2–4% in the Northern group of the Cook Islands to 50–60% in some countries of the Western world [1] . Increased age , female sex , number of pregnancies , obesity , history of deep venous thrombosis , and standing occupation are among other risk factors [2] . VVs not related to the post-thrombotic syndrome or venous malformations are defined as primary VVs . Pathogenesis of VVs is still not fully clarified . According to current understanding , key factors implicated in VVs development include changes in hemodynamic forces ( decrease in laminar shear stress and increase in venous filling pressure ) , endothelial activation , inflammation , hypoxia , and dysregulation of matrix metalloproteinases and their tissue inhibitors [3–5] . These alterations underlie pathological remodeling of the vascular wall and loss of its tone . Questions remain about the order of events and the primary stimulus triggering the set of disease-related changes . The cumulative evidence from epidemiological , family , and genetic association studies strongly indicates that there is a hereditary component in VVs etiology [6–8] . However , despite progress in this field [9–12] , current knowledge of the genetic basis of this pathology is far from being complete . Elucidating genes involved in susceptibility to VVs would help to identify key molecular players in the disease initiation , provide deeper insights into its pathogenesis , and eventually contribute to development of improved targeted therapy aimed at VVs treating and preventing . Large-scale biobanks linked to electronic health records open up unparalleled opportunities to investigate the genetics of complex traits . Today , UK Biobank is the largest repository that contains information on genotypes and phenotypes for half a million participating individuals [13] . This resource is open to all bona fide researchers , and access to data is provided upon approval of their application and payment of necessary costs . However , the need to incur high costs related to data access and computation can be an insurmountable obstacle for those who cannot afford these expenses . The Neale Lab ( http://www . nealelab . is/ ) and the Gene ATLAS [14] ( http://geneatlas . roslin . ed . ac . uk/ ) are two independent projects , which intend to remove this burden by generating genetic association data and sharing them with broader scientific community . These resources provide free open access to “quick-and-dirty” genome-wide association study ( GWAS ) summary statistics for a wide range of phenotypes measured in the UK Biobank . In our study , we aimed to employ state-of-the-art bioinformatics approaches to extract maximum possible information from these open resources with regard to the genetics of VVs of lower extremities . Our objectives were to ( 1 ) identify genetic loci reliably associated with VVs risk and prioritize the genes that account for the revealed associations , ( 2 ) elucidate pleiotropic effects of identified loci , ( 3 ) investigate genetic overlap between VVs and other complex traits , ( 4 ) gain etiological insights and explore cause-and-effect relationships by means of Mendelian randomization analysis .
Our study was designed as a one-stage GWAS followed by downstream bioinformatics analysis . The overall workflow of the study is depicted in Fig 1 . All calculations were entirely based on the UK Biobank data for white British individuals available in open access databases: the Neale Lab database and the Gene ATLAS database . These projects used different software , methods of analysis , and quality control approaches . The Gene ATLAS project applied less stringent filtering criteria and therefore had larger sample size . We calculated genetic correlations between VVs in both public databases and found out that these traits are almost genetically equivalent ( genetic correlation coefficient = 0 . 99 , S1 Table ) . The identification of VVs-associated loci and the search for functional variants was carried out using GWAS summary statistics provided by the Neale Lab . Since no replication was performed , significance threshold was raised from 5 . 0e-08 to a more conservative level of 5 . 0e-09 . Moreover , we filtered out loci associated with VVs in the Gene ATLAS dataset at P-values ≥ P-values in the Neale Lab dataset ( assuming that large sample size provides higher level of statistical significance ) . Thus , only signals with more convincing association data have been left for further analysis and interpretation . Functional bioinformatics analysis was conducted using the Gene ATLAS data . In carrying out this study , we had to face a number of challenges and limitations that must be acknowledged . The first limitation was intrinsic to the general approach to phenotype definition based on the electronic medical records system , which was employed in the UK Biobank study . Phenotype “VVs of lower extremities” was defined based on International Classification of Disease ( ICD-10 ) billing code “I83” present in the electronic patient record . The Neale Lab reported the phenotype prevalence of 2 . 1% . It is much lower than VVs prevalence estimated by European epidemiological studies . Despite the evidence of a “healthy volunteer” selection bias in the UK Biobank study [15] , such a low rate indicates that a proportion of individuals remained undiagnosed . This is in line with a recent primary healthcare register-based study reporting VVs prevalence rate of 3% in German general practice [16] . This phenomenon could be explained by a non-life-threatening nature of varicose veins , which might discourage patients from communication to the doctor . Given that individuals not diagnosed with I83 served as controls in our study , we could therefore expect an overall decrease in the statistical power to detect gene-disease associations . Another potential source of missing associations was immensely strict criteria used by the Neale Lab for single nucleotide polymorphisms ( SNPs ) quality control removing around 75% of SNPs initially provided by UK Biobank . The next important limitation arose from the lack of access to individual-level data resulting in the inability to control a possible selection or sampling bias . We suggested that traits related to VVs risk factors could potentially cause unequal representation of patients with different characteristics in the case and the control groups , and thereby induce spurious associations or effect modification . The Neale Lab analyses were adjusted for sex , but other factors were beyond our control . In particular , we did not know how body mass index ( BMI ) was distributed in the case and the control groups and what proportion of patients in each group had deep venous thrombosis ( DVT ) . In order to address this challenge , we performed an adjustment for these potential confounders by implementation of the method based on GWAS summary statistics [17 , 18] ( Supplemental Methods , Section 4 ) . Summarizing the above , we can state that the limitations of our study were mainly related to the loss of statistical power and a “quick-and-dirty” approach to summary statistics generation . Nevertheless , given a large sample size and a large number of associations tested , we assume that this obstacle could be at least partially compensated by a huge scale of the UK Biobank study itself . SMR/HEIDI analysis revealed 32 traits that shared the same casual variants with VVs ( traits for this and further analyses were obtained from the GWAS-MAP database , see Materials and Methods section and S10 Table for details ) . Pleiotropic effects were found for 6 out of 12 studied loci ( Fig 3 , S11 Table ) . Variants within loci tagged by rs11121615 and rs3101725 showed positive SMR beta coefficients ( the same direction of effect ) with predicted and fat-free mass of both legs and a whole body as well as with basal metabolic rate , and negative coefficients ( opposed effects ) –with leg fat percentage as well as impedance of both legs and a whole body ( parameter negatively correlated with fat-free mass ) . Loci tagged by rs3101725 and rs7773004 comprised SNPs with pleiotropic effects on red blood cell erythrocyte distribution width ( negative beta ) . Locus tagged by rs7773004 was also related to numerous blood traits , such as mean corpuscular haemoglobin concentration , mean platelet thrombocyte volume , and monocyte and reticulocyte count . Locus tagged by rs73107980 was associated with platelet crit . Three loci ( rs11135046 , rs28558138 , and rs28558138 ) were linked to blood pressure and hypertension with different SMR beta signs . Another interesting finding was identification of positive SMR beta for associations with vascular/heart problems ( rs28558138 locus ) , cellulitis ( rs3101725 locus ) , forced viral capacity , and height size at age 10 ( rs7773004 locus ) . Overall , our analysis revealed three main groups of traits: one cluster related to mass , basal metabolic rate , and cellulitis , one cluster of blood-related traits linked only to rs7773004 locus , and one cluster containing the remaining traits ( Fig 3 ) . A list of genetic correlation estimates ( rg ) between VVs and 861 complex traits is presented in S12 Table . Twenty five traits showed statistically significant correlation with VVs with absolute values of rg ≥ 0 . 2 . Correlation matrix for this subset is displayed as a heatmap in Fig 4 . We observed 5 main clusters: traits related to the type of job , intelligence , and qualification; traits related to height and mass ( including predicted and fat-free mass of both legs ) ; thrombosis-related traits; traits related to operations; and traits related to pain ( including leg pain and gonarthrosis ) and health satisfaction . VVs trait was closest to the thrombosis-related cluster ( positive correlation ) . It was also positively correlated with mass , operations , and pain-related traits as well as with lower levels of qualification and heavy manual/walking/standing job . For traits with less prominent correlation with VVs , we observed the same trend: pain and anthropometric traits ( sitting and standing height , BMI , mass , etc . ) showed positive correlations , whilst higher levels of education–negative ones . Interestingly , negative correlation was observed with usual walking pace , and positive–with current smoking . We calculated partial genetic correlations for the subset of 7 non-collinear traits with |rg | ≥ 0 . 2 . Two traits–DVT and height size at age 10 –were shown to share common genetic background with VVs after the adjustment for the remaining traits in the analyzed subset ( S3 Fig ) . In other words , their shared genetic components were demonstrated to be at least partially independent of other complex traits . Additionally , we estimated partial genetic correlations between VVs , standing height , and weight . Both standing height and weight had independent genetic components shared with VVs ( S4 Fig ) . We applied a 2-sample Mendelian randomization ( 2SMR ) [31] strategy to infer causal relationships between a broad range of “exposure” phenotypes and VVs as an outcome . In total , 39 complex traits were shown to be potential causative factors . Although only genome-wide associated SNPs from exposure GWAS were selected as instrumental variables , we did not require these loci to be replicated . Nevertheless , we checked the stability of our tests with regard to instruments selection by performing the robustness analysis . This test along with the Steiger test [32] for the correct direction of effect underpinned the exclusion of 2 out of 39 traits ( S13 Table ) . Further , we assessed violations of MR assumption of absence of horizontal pleiotropy ( influence of genetic instruments on the outcome only through the exposure , also known as “exclusion-restriction criterion” ) by means of sensitivity analyses [31] . We did not observe a statistically significant intercept in MR-Egger regression for any trait . However , only a small proportion of traits showed symmetry in Funnel plots and had no heterogeneity in causal effects amongst instruments . This provides evidence that , for the majority of traits , at least some of the selected instruments exhibit horizontal pleiotropic effects . Such traits mainly involved several hundred genome-wide SNPs that made leave-one-out analysis also uninformative . In order to correct for horizontal pleiotropy , we applied a straightforward approach having excluded all instrumental variables associated with VVs at the level of statistical significance higher than 0 . 01 . Our correction led to symmetry in 26 out of 33 asymmetrical Funnel plots and eliminated heterogeneity in causal effects for 28 out of 34 traits , although 6 traits lost the statistical significance of 2SMR coefficients ( S5 Fig , S13 Table ) . Removing potential sources of heterogeneity also reduced absolute values of 2SMR beta for all the corrected phenotypes . A graphical representation of our results , including 25 causal inferences that we consider the most reliable , is shown in Fig 5 . Twenty one traits were related to anthropometry and included standing and sitting height , weight , hip and waist circumference , fat-free , predicted , and fat mass of legs and arms , etc . One trait was a spirometry measurement associated with pulmonary function . Nonetheless , since it is positively correlated with height , we suppose that it has no independent effect on VVs development . Similarly , the reverse association between malabsorption/coeliac disease and VVs could actually be induced by a weight loss as one of the complications of these conditions . Moreover , although this trait has passed all the necessary tests , we avoid making strong claims about its causality since it was self-reported and involved only ~1 , 500 cases . All the above mentioned traits were derived from the Neale Lab repository and therefore had about 80% sample overlap with VVs dataset obtained from the Gene ATLAS . Shared participants between the “exposure” and the “outcome” GWAS can cause bias in the Mendelian randomization analysis when weak instruments are selected [33] . In our study , this limitation was mitigated by using only strong instruments associated with exposure traits at a high level of statistical significance ( P < 1 . 0e-08 ) . We estimated the relative bias and type 1 error rate inflation using analytical formulae provided by Burgess et al . [33] and a web application ( https://sb452 . shinyapps . io/overlap/ ) . The relative bias was shown to be small , and the type 1 error was close to nominal: for 80% sample overlap , F-statistics of at least 33 . 3 ( strong instruments selection ) , sample size of 337 , 000 and 408 , 000 for exposure and outcome traits , ordinary least squares estimate of 0 . 1 and a number of instrumental variables = 600 , the relative bias was 2 . 3% , and the type 1 error– 6 . 2% . For 300 instruments used for most UK Biobank traits , the relative bias was 2 . 4% , and the type 1 error– 5 . 6% . ( Supplemental Methods , Section 8 ) . Thus , we considered the application of 2SMR approach suitable for our settings . Nevertheless , in order to fully overcome the sample overlap problem and confirm our results , we searched for independent GWAS and performed a replication analysis . Data were obtained for height , weight , waist circumference , and body fat percentage . Only height reached Bonferroni-corrected level of statistical significance and remained significant after the correction for pleiotropy ( 2SMR P < 1 . 0e-07 , S14 Table ) . However , null results for other traits could be explained by a limited power of the analysis . For example , sample size for weight was 4 . 5 times smaller than in the Neale Lab study . Finally , the remaining two traits from the “reliable” set were plasma levels of MHC class I polypeptide-related sequence B protein and CD209 antigen . These phenotypes were not UK Biobank traits , therefore our results could not be confounded by a sample overlap .
In the present study , we utilized freely available GWAS summary data to unravel the genetic underpinnings of VVs of lower extremities . We used “quick-and-dirty” statistics provided by the Neale Lab and the Gene ATLAS projects , which aimed to help the scientific community query UK Biobank results for hundreds of human traits avoiding the need to incur high computational costs [34] . These data have been generated without deep insight into each phenotype . Lack of individual-level data made us be as rigorous as possible to avoid false positive discoveries , so we skimmed off the most apparent evidence . Nevertheless , as far as we are aware , our study is the largest and the most comprehensive study of VVs genetics to date , and huge size of the UK Biobank dataset is expected to overcome potential issues related to phenotypic noise . Beyond that , the availability of GWAS results makes our research fully reproducible . Overall , SNP-heritability on the liability scale was estimated at ~28% assuming the disease prevalence of 20–30% . Using a one-stage GWAS approach , we identified 12 susceptibility loci that explain 13 . 4% of the SNP-based heritability ( Table 1 , Fig 2 , S2 Table ) . For each revealed locus , we prioritized the most likely causal genes ( Table 2 , literature data are given in S4 Table ) . Nearly 27% of the variance explained by our top 12 SNPs was attributable to the polymorphism in the CASZ1 gene involved in blood vessel development . This strong signal has previously been revealed by “23andMe” [12] and subsequently replicated in our own sample of ethnic Russian individuals [10] . An especially interesting finding , in our opinion , is an association of VVs with SNPs in a recently discovered PIEZO1 gene . PIEZO1 encodes a pressure-activated ion channel which senses shear stress and controls vascular architecture [35] . Mice embryos lacking functional Piezo1 exhibit defects in vascular remodeling and die at midgestation [36] . Other prioritized genes were PPP3R1 , EBF1 , STIM2 , HFE , GATA2 , NFATC2 , and SOX9 . The HFE gene has been linked to the risk of VVs in our recent candidate-gene study [21] . Meanwhile , we could not prioritize any gene in the loci tagged by rs3101725 , rs236530 , and rs73107980 ( Table 2 ) . On the one hand , region near rs3101725 contains two genes ( SLC12A2 and FBN2 ) that could play a role in varicose transformation , and on the other hand , the causal polymorphism was shown to be eQTL for a nearby non-protein coding RNA LINC01184 with unknown function . KCNJ16 and KCNJ2 genes in the locus tagged by rs236530 encode potassium voltage-gated channels , and their role in vascular biology is unknown . RAPGEF3 gene in the rs73107980 locus regulates vascular permeability and promotes vascular smooth muscle cell migration , and the transcription of the COL2A1 gene ( encoding an extracellular matrix component ) is under direct control of the SOX9 gene product ( prioritized for the rs2241173 locus ) , so both these genes could potentially be causal . The revealed genes can be considered as good candidates for future follow-up functional studies . It is noteworthy that SLC12A2 , FBN2 , STIM2 , HFE , KCNJ16 , KCNJ2 , and COL2A1 were included in the druggable gene set by Finan et al . [37] , and the PIEZO1 and the KCNJ2 gene products belong to a “potential drug target” class according to the Human Protein Atlas ( https://www . proteinatlas . org/ ) . Gene set enrichment analysis involving both genome-wide and less strongly associated signals detected gene categories related to abnormal vascular development and morphology ( S8A and S9A Tables ) . This observation is consistent with the hypothesis that pathological changes in the vein wall are the primary event preceding VVs formation [38] . Furthermore , our genetic correlations analysis confirmed known epidemiological associations between VVs and DVT [1 , 16] as well as standing job [1 , 39 , 40] and rough labour [40 , 41] ( Fig 4 , S12 Table ) . Shared familial susceptibility with venous thromboembolism has already been shown by Zöller et al . [42] . Here we demonstrate that DVT and VVs share specific genetic components which are independent from other factors such as obesity or number of operations ( S3 Fig ) . Since none of the top GWAS hits were related to thrombosis , we can conclude that SNPs with less prominent associations are responsible for this genetic overlap . On the contrary , albeit several identified loci were also associated with blood pressure/hypertension and red and white blood cell traits , including mean corpuscular hemoglobin concentration ( Fig 3 , S4 and S11 Tables ) , we found no evidence for genetic correlation between VVs and these traits . We can therefore attribute these effects only to pleiotropy . Intriguingly , we observed small , but significant genetic overlap with smoking ( rg = 0 . 16 ) . Smoking is considered only as suggestive risk factor for VVs since epidemiological studies have mainly shown no association with this habit [1 , 39 , 41 , 43] . Other novel interesting findings include genetic links with prospective memory and fluid intelligence , level of education ( negative correlation ) , pain ( knee pain , pain all over the body , neck or shoulder pain , and leg pain on walking ) , usual walking pace , and gonarthrosis . Further , we obtained strong evidence for association between VVs and anthropometric traits such as weight , height , waist and hip circumference . We not only observed genetic correlations ( Figs 3 and 4 , S11 and S12 Tables ) , but also demonstrated these traits to be causative factors for VVs development ( Fig 5 , S13 Table ) . It is important to note that genetic overlap and causal relationships were inferred for both fat and fat-free mass . Thus , we can speculate that increased weight is a risk factor for VVs regardless of whether it was caused by excess body fat or a large mass of other tissues . According to some theories , association between overweight and VVs can be explained by greater concentrations of circulating estrogens or even by a confounding effect of parity [1] . Although this can be true , our results suggest that the cause may also be an increase in mass per se . Moreover , we found that height has common genetic component with VVs independent from weight and other traits ( S4 Fig ) and is also causally related to VVs ( Fig 5 ) . Causal inference was additionally confirmed using an independent dataset ( S14 Table ) . This is in agreement with the results of the Edinburgh Vein Study that has shown a significant relationship between increasing height and VVs [39] . However , the majority of epidemiological studies to date have been focused only on BMI and did not consider height or other body characteristics . Notably , the formula for BMI contains the square of height in the denominator . It is possible that the impact of height underlies the inconsistency in the results , when some studies show a positive association between BMI and VVs , while others reveal no effect [1] . We therefore recommend that future epidemiological studies collect and analyze data on height and weight along with data on BMI . Last but not least , we detected the causal effect of plasma levels of two proteins–MHC class I polypeptide-related sequence B protein ( MICB ) and CD209 antigen . Both molecules are involved in innate and adaptive immunity . MICB is a ligand for the activating receptor NKG2D present on the surface of natural killer and some other immune-related cells . CD209 , also known as DC-SIGN , is a C-type lectin receptor expressed on dendritic cells and macrophages . In our study , Mendelian randomization analysis for CD209 was performed using two genetic instruments–rs505922 and rs8106657 identified by Suhre et al . [44] . The first SNP is located in the ABO gene responsible for ABO blood group determination . Allele rs505922 C tags non-O group and is in linkage disequilibrium with the allele A of the neighboring SNP rs507666 ( D’ = 1 . 00 , r2 = 0 . 39 ) , which was found to be the top variant associated with VVs in the “23andMe” GWAS [12] . This association was validated in our previous study using data from UK Biobank [10] . At the same time , rs505922 explains nearly 40% of the variability of the plasma CD209 concentration ( our estimate based on the published data [44] ) with the allele C being linked to higher levels of CD209 . It is tempting to speculate that association between the ABO gene polymorphisms and VVs as well as between VVs and blood group A [45] is mediated by this protein . We conducted SMR/HEIDI analysis for the locus containing these SNPs and demonstrated that both associations with VVs and CD209 were related to the same causal polymorphism ( PSMR = 3 . 3e-06 , PHEIDI = 0 . 75 ) , that supports our hypothesis . Nevertheless , the role of CD209 as well as MICB in VVs etiology needs further experimental confirmation . Besides this , it should be taken into account that the level of CD209 in plasma does not necessarily correlates with its level on the surface of the cells , and circulating CD209 may possess its own specific function .
This study was based on genetic data provided by UK Biobank . All study participants provided written informed consent , and the study was approved by the North West Multi-centre Research Ethics Committee . All cases and controls analyzed in our study were derived from UK Biobank [13] . Sociodemographic , physical , lifestyle , and health-related characteristics of UK Biobank participants have been reported elsewhere [15] . In brief , enrolled subjects were aged 40–69 years , were less likely to be obese , to smoke , to drink alcohol , and had fewer self-reported health conditions compared with the general population . Genotyping was performed using the Affymetrix UK BiLEVE and the Affymetrix UK Biobank Axiom arrays . Details on DNA extraction , genotyping , imputation , and quality control ( QC ) procedures have been reported previously [46] . Regional association plots were generated with LocusZoom tool ( http://locuszoom . org/ ) for regions within 250 kb from the lead SNP . Conditional and joint ( COJO ) analysis was carried out by a summary statistics-based method described by Yang et al . [48] . Calculations were performed using the GCTA software [49] . LD ( linkage disequilibrium ) matrix was computed with PLINK 1 . 9 software ( https://www . cog-genomics . org/plink2 ) using 1000 Genomes data for 503 European individuals ( http://www . internationalgenome . org/data/ ) . We claimed the presence of one independent signal per locus if none of the polymorphisms except the lead SNP passed the significance threshold of P = 5 . 0e-08 . We used regional association plots to identify genes located within associated loci . These genes were further queried for potential involvement in the processes relevant to VVs pathogenesis . For each gene , we scanned an Online Mendelian Inheritance in Man database ( OMIM , https://www . omim . org/ ) , the NCBI Gene ( https://www . ncbi . nlm . nih . gov/gene ) , and the Pubmed database to inquire into their biological functions . Furthermore , we interrogated whether other hypothetical varicose-related traits were associated with these genes according to previously published GWAS . The Pubmed search was performed using the gene name as a keyword as well as combinations of the gene name and “GWAS” , or “genome-wide association study” , or “varicose” , or “venous” , or “vascular” . The information obtained was used for the literature-based gene prioritization . Additionally , we used a PhenoScanner tool [50] to make a list of complex traits associated at P < 5 . 0e-08 with lead SNPs or other polymorphisms in high LD ( r2 ≥ 0 . 8 ) with lead variants ( http://www . phenoscanner . medschl . cam . ac . uk/phenoscanner ) . 1000 Genomes-derived polymorphisms served as proxies for the analysis . For the VEP [23] analysis , we used a set of SNPs in high LD ( r2 > 0 . 8 ) with 13 SNPs identified by the COJO analysis ( in a 250kb window ) . LD proxies were selected from the 1000 Genomes phase 3 version 5 panel using a "proxysnps" R package for European population . Besides this , we selected all SNPs associated with VVs at P ≤ T , where log10 ( T ) = log10 ( Pmin ) + 1 , and Pmin is a P-value for the strongest association per locus ( ±250 kb from the independent hit ) . This additional criterion was applied since genotype data for the UK Biobank samples have been imputed using the Haplotype Reference Consortium ( HRC ) panel , and some HRC SNPs could possibly be missed in the 1000 Genomes panel . Analysis was performed using software available online ( https://www . ensembl . org/info/docs/tools/vep/index . html ) . SMR/HEIDI approach [24] was used to test for potential pleiotropic effects of identified loci on VVs and other complex traits including gene expression levels in certain tissues . SMR analysis provides evidence for pleiotropy but is unable to define whether both traits are affected by the same underlying causal polymorphism . The latter is specified by a HEIDI test that distinguishes pleiotropy from linkage disequilibrium . Summary statistics for gene expression levels was obtained from Westra Blood eQTL [51] ( peripheral blood , http://cnsgenomics . com/software/smr/#eQTLsummarydata ) and the GTEx [52] database ( 44 tissues , https://gtexportal . org ) . VVs summary statistics was obtained from the Gene ATLAS database . Summary statistics for other complex traits was derived from the GWAS-MAP database ( see below ) . Nominal P for SMR test was set at 3 . 54e-06 ( 0 . 05/14 , 117 , where 14 , 117 is the total of number of tests corresponding to all analyzed SNPs , probes , and tissues ) and 1 . 88e-06 ( 0 . 05/12*2 , 219 , where 12 is the number of loci and 2 , 219 is the number of non-VVs traits in the GWAS-MAP database excluding binary traits with the number of cases or controls less than 1000 ) . For HEIDI analysis in a gene expression study , we used a conservative threshold of P = 0 . 01 ( P < 0 . 01 corresponds to the rejection of pleiotropy hypothesis ) . For HEIDI in a complex traits analysis , we implemented a less conservative threshold of P = 0 . 001 since the number of independent test was much higher . Details of data processing are given in Supplemental Methods ( Section 5 ) . DEPICT analysis was performed using DEPICT [26] software version 1 . 1 , release 194 ( https://data . broadinstitute . org/mpg/depict/ ) with default parameters . GWAS summary statistics was obtained from the Gene ATLAS database . We employed DEPICT for both genome-wide significant SNPs ( P < 5 . 0e-08 ) as well as for loci associated with VVs at P < 1 . 0e-05 . As in previous analyzes , we defined locus as ±250 kb from the lead SNP . The major histocompatibility complex ( MHC ) region was eliminated . Significance threshold was set at FDR < 0 . 05 . The GWAS-MAP platform integrates a database which was created to study cardiovascular diseases and contains summary-level GWAS results for 123 metabolomics traits , 1 , 206 circulating proteins , 2 , 419 complex traits from the UK Biobank , and 8 traits related to coronary artery disease , myocardial infarction and their risk factors . We additionally added 5 VVs-related traits analyzed in the present study ( the Gene ATLAS data for VVs; the Neale Lab data for VVs , BMI , and DVT; the Neale Lab data for VVs adjusted for BMI and DVT ) as well as 33 traits from the Gene ATLAS database that we supposed to be biologically relevant to VVs . Description of all 3 , 794 traits is provided in S10 Table . The GWAS-MAP platform contains embedded software for LD Score regression [47] , 2-sample Mendelian randomization analysis ( MR-Base package [31] ) , and our implementation of SMR/HEIDI analysis [24] . Further details are given in Supplemental methods ( Section 6 ) . Genetic correlations ( rg ) between VVs and other complex traits were calculated using LDSC software ( https://github . com/bulik/ldsc/ ) . We applied a cross-trait LD Score regression technique as previously described by Bulik-Sullivan et al . [53] . This method requires only GWAS summary statistics and is not biased by a sample overlap . We analyzed 861 heritable non-VVs traits from the GWAS-MAP database . Only traits with a total sample size of ≥ 10 , 000 and ≥ 1 million SNPs tested were included in the analysis . GWAS summary statistics for VVs was obtained from the Gene ATLAS database . Statistical significance threshold was set at 1 . 16e-05 ( 0 . 01/861 ) . For 25 traits that passed this threshold with |rg| ≥ 0 . 2 , we calculated a matrix of genetic correlations . Partial genetic correlations were estimated using the inverse of the genetic correlation matrix . Significance threshold was set at 3 . 1e-05 ( 0 . 01/325 , where 325 is the number of pairwise combinations ) . For partial genetic correlations between VVs , standing height , and weight , nominal P was set at 3 . 3e-03 ( 0 . 01/3 ) . Clustering and visualization were performed by the “corrplot” package for the R language ( basic “hclust” function ) . Further details are provided in Supplemental Methods ( Section 3 ) . Casual relationships between 2 , 221 non-VVs phenotypes ( “exposures” ) from the GWAS-MAP database and VVs ( “outcome” ) were assessed by 2-sample Mendelian randomization ( 2SMR ) as previously described by the MR-Base collaboration [31] ( http://www . mrbase . org/ ) . All the details of our protocol are given in Supplemental Methods ( Section 7 ) . Binary traits with the number of cases or controls less than 1000 were not included in the study . GWAS summary statistics for VVs was obtained from the Gene ATLAS database . Analysis was performed on the GWAS-MAP platform . Two 2SMR approaches were used: an inverse variance weighted meta-analysis of Wald ratios ( IVW ) and MR-Egger regression . The nominal P was set at 1 . 1e-05 ( 0 . 05/2*2 , 221 , where 2 is the number of approaches ) . For traits that passed either IVW or MR-Egger test , we performed the Steiger test [32] for identifying the correct direction of effect , and conducted the robustness analysis ( our own approach ) . Traits that passed all the analyses were then subjected to sensitivity tests [31] , that included assessing heterogeneity in causal effects amongst instruments , horizontal pleiotropy test ( based on the intercept in MR Egger regression ) , leave-one-out analysis , and Funnel plots inspection . The nominal P for the Steiger test was set at 2 . 3e-05 ( 0 . 05/2 , 221 ) , and for the robustness analysis , horizontal pleiotropy test , and heterogeneity analysis–at 1 . 3e-03 ( 0 . 05/38 , where 38 is the number of traits that passed both 2SMR analysis and the Steiger test ) . Leave-one-out and Funnel plots were examined manually . Sensitivity analyses revealed the presence of horizontal pleiotropy for the majority of traits . To correct for this confounder , we omitted all instrumental variables associated with VVs with P < 0 . 01 , and then repeated IVW 2SMR and sensitivity analyses . Additionally , for the resulting set of traits , we searched for independent GWAS in the MR-Base database [31] and conducted confirmatory IVW 2SMR analysis ( where appropriate ) with the MR-Base default parameters . The nominal P was set at 0 . 013 ( 0 . 05/4 , where 4 is the number of traits ) . | Varicose veins of lower extremities ( VVs ) affect about 30% of adults in developed countries and cause both cosmetic and health problems . A strong body of evidence indicates that heredity plays an important role in the etiology of this condition . However , genetic basis of VVs remains poorly understood . Here , we present the results of the first large-scale genetic study for VVs . We identified genes which are the most likely involved in VVs pathogenesis . We show that VVs are correlated at a genetic level with numerous traits and phenotypes , including those already known from prior epidemiological studies ( deep venous thrombosis , body mass index , standing job , etc . ) as well as with those that have not been suspected to share common genetic background with VVs ( fluid intelligence and prospective memory scores , smoking , walking pace , pain all over the body , and other traits ) . Finally , using genetic variants as instruments , we demonstrate direct causal effects of the traits related to anthropometry , such as height and weight , and plasma levels of immune-related proteins MICB and CD209 . Our study provides novel insight into both VVs genetics and etiology . The revealed genes ( CASZ1 , PIEZO1 , PPP3R1 , EBF1 , STIM2 , HFE , GATA2 , NFATC2 , and SOX9 ) and proteins ( MICB and CD209 ) can be considered as good candidates for follow-up functional studies and might be of interest as potential drug targets . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] | [] | 2019 | Varicose veins of lower extremities: Insights from the first large-scale genetic study |
The current Ebola outbreak poses a threat to individual and global public health . Although the disease has been of interest to the scientific community since 1976 , an effective vaccination approach is still lacking . This fact questions past global public health strategies , which have not foreseen the possible impact of this infectious disease . To quantify the global research activity in this field , a scientometric investigation was conducted . We analyzed the research output of countries , individual institutions and their collaborative networks . The resulting research architecture indicated that American and European countries played a leading role regarding output activity , citations and multi- and bilateral cooperations . When related to population numbers , African countries , which usually do not dominate the global research in other medical fields , were among the most prolific nations . We conclude that the field of Ebola research is constantly progressing , and the research landscape is influenced by economical and infrastructural factors as well as historical relations between countries and outbreak events .
No other infectious disease event has captured the attention of the international health community in recent years like the Ebola outbreak . The current epidemic started in December 2013 leading to over 26 . 000 infected patients and more than 10 . 000 deaths [1] The outbreak reached global dimension as hospitals in the United States of America ( US ) and Europe are now treating patients returning from health missions in Ebola affected countries [2] . 28 outbreaks were documented since 1976 , which all , except the recent one , occurred in isolated regions . During the first epidemic in the Democratic Republic of Congo and the Sudan the Ebola virus was identified as a non-segmented , negative-strand RNA virus and placed within the Filoviridae family . Since then five distinct virus species were distinguished . Although the pathogenesis of the Ebola virus is intensively investigated worldwide , the undisputed identification of the natural reservoir has not been successful yet . Bats were implicated as a possible host for the Ebola as well as the related Marburg virus [3] . Despite the short time span since the discovery of the Ebola virus , scientists have authored a substantial body of related scientific literature worldwide . Nevertheless , it should be an ethical obligation of all industrialized countries to invest future capacities in research of this life-threatening disease and vaccination strategies as one of the most effective means to fight infectious diseases [4 , 5] . In order to cast a first light on the question of global research activity in this field since 1976 , we present a combined scientometric and density equalizing study . It encompasses scientometric tools and advanced visualizing techniques such as density equalizing mapping [6] and draws a sketch of the global Ebola research architecture over the past 40 years . Scientometric analysis of the scientific output of individuals , institutions and countries is represented in the number of publications as well as citations and their bibliometric parameters . Density equalizing map projections ( DEMP ) are used as a state of the art technique to demonstrate the global architecture on the research output via distorted maps .
The data was analyzed using scientometric methods developed in the NewQIS project as previously described [7] . The analyzed data was retrieved from the Thomson Reuters Web of Science database ( WoS ) using the search term “Ebola” in the Science Citation Index Expanded ( SCIE ) and the Social Sciences Citation Index ( SSCI ) ( time frame between the first description of the virus in 1976 and 2014 ) . To limit our search to the original research articles , we used the WoS´s option for selecting the document type and included only “articles” in the analysis . Data was processed as previously described using a combination of scientometric tools with density equalizing mapping [8] . For the generation of density equalizing mapping , the Gastner and Newman’s algorithm was employed [6] . As parameters , citation quantities were determined using the “citation report” function ( number of citations per article , the citation rate of countries , and authors ) , H-indices [9] along with other general operating figures ( year of publication , country of publication , co-operations between different countries , language of publication , document types , subject areas , and journals ) . Also , author analysis , subheading-terms , and individual subject areas were examined . To evaluate the quality of a country’s publication output , we assessed the citation rate and the modified H-index [9] .
The total numbers of publications in the database added up to 2482 ( search term “Ebola” only as title word ) and 3081 ( search term “Ebola” also in keywords and abstract ) starting from 1976 with a steady yearly increase in publication activity until 2014 . Ebola research originated from 78 countries . Research groups based in the USA published most research with 1367 articles ( 44 , 4% of all determined articles ) , followed by groups from Germany ( 272 articles , 8 . 8% ) , Canada ( 202 articles , 6 . 6% ) , France ( 179 articles , 5 . 8% ) , United Kingdom ( 167 articles , 5 . 4% ) , Japan ( 157 articles , 5 . 1% ) , Russia ( 84 articles , 2 . 7% ) , Gabon ( 70 articles , 2 . 3% ) , Belgium ( 58 articles , 1 . 9% ) and Switzerland ( 51 articles , 1 . 7% ) . Each percent value stands for the part of the overall Ebola research output that was retrieved via the WoS . The rest of articles originated from the remaining 68 countries that are involved in Ebola research . Overall , North American and European countries took a lead position . The density equalizing mapping of the world shows that research activity translated into a distorted global architecture ( Fig 1 ) . African countries affected by Ebola cases exhibited a relatively low activity but were present . The continents South America , and Asia almost disappeared from the cartogram . When relating these operating figures to population numbers ( Fig 2 ) , we found that—besides the most active nations ( US , European ) —smaller African countries such as Gabon , Republic of Congo , Central African Republic and Cameroon gained importance . Uganda , Republic of Congo , Gabon and South Africa reached increased ratios regarding their research activity adjusted to Gross Domestic Product ( GDP , Fig 3 ) . We did not find any association between the death rate and the research output on a regional scale . With reference to the citation rate ( CR = average ratio of citations per publications ) , Gabon was ranked first ( CR 43 . 3 ) followed by western countries ( Switzerland = CR 34 . 4; Germany = CR 33 . 2; France = CR 32 . 9; US = CR 32 . 6; and UK = CR 30 . 1 ) ( Fig 4 ) . Focusing on the modified h-index ( country specific and thematically related ) , the US exhibited the highest h-index ( 102 ) followed by Germany ( 55 ) , France ( 46 ) , Canada ( 41 ) , UK ( 37 ) , Japan ( 32 ) , Gabon ( 26 ) , Switzerland ( 24 ) , Belgium ( 23 ) and Russia ( 20 ) ( Fig 5 ) . Ebola research has been characterized by strong international cooperations since its beginnings . Although most global research involved cooperation with US-American research groups ( 449 collaborative articles ) , the overall rate of US collaborative publications in relation to the total US output activity with 1367 articles in total ( 33% ) was relatively low compared to other countries ( Fig 6 ) . Most frequent research partners of the USA were Canada ( 121 US-Canadian collaborative articles ) , Japan ( 99 US-Japan collaborative articles ) and Germany ( 88 US-German collaborative articles ) . Canada also played an important role with 148 collaborative articles altogether conducted with a total of 11 countries ( 73% of all Canadian articles ) , followed by Japan with 121 collaborative articles in total ( 77% of all Japanese articles ) , Germany with a total of 175 collaborative articles ( 64% ) , France with 123 collaborative article all in all ( 69% from all French articles ) , the UK with 87 collaborative articles in total ( 52% of all UK articles ) and Belgium with 50 collaborative articles ( 71% of all Belgian articles ) . Due to the fact that most Ebola outbreaks were geographically defined to the African continent we found particularly strong cooperations of France with Gabon with 38 collaborative articles ( USA and Gabon have 19 common articles; Germany and Gabon have 14 common articles ) . Gabon has worked out 87% of the overall publication output together with other countries . The collaboration between the USA and Uganda followed with 21 collaborative articles . Uganda has written 26 together with other countries ( 72% of all Ugandan articles ) . South Africa published 94% of the overall research output as collaborative articles and Congo nearly 97% . On an institutional level most cooperate publications were produced between the Canadian Science Center of Human and Animal Health and the University of Manitoba . Both located in Winnipeg and have an overlap of staff working in both facilities . Also , numerous cooperations were found between research institutions within the US . Laboratories with the highest biosafety level clearance were available in 13 of the leading 33 cooperating institutions ( Fig 7 ) . Subject categories are defined classes of themes indicating a general area of science . For the field of Ebola research , they were determined and depicted in four year intervals ( Fig 8 ) . The research interest shifted from a predominant focus on general and internal medicine to a much more diversified picture covering subject categories in immunology , cell biology , pharmacology , experimental medicine , biochemistry and molecular medicine . We did not find a remarkable attribution to the category of public health .
In order to give an overview over the global Ebola research architecture , the current study employed the Web of Science and conducted the first analysis of the research output for the entire field since 1976 . Other than the majority of infectious diseases that have been discussed in the scientific literature for more than a century ( e . g . yellow fever or dengue [10 , 11] ) the advent of Ebola research is recent and started from its first description in 1976 [12] . Therefore , Ebola research offers a rare opportunity to observe the entire development of a scientific field since all related publications are achieved in electronically accessible scientific databases . Overall , we found that the number of publications was constantly low in the years after 1976 and remained on a level of up to 20 publications per year . This is surprisingly low for a dangerous infectious disease but understandable since a biosafety level 4 laboratory is needed to carry out the research . Coinciding with the reemergence of the virus in Kikwit ( located in the Democratic Republic of Congo ) [13] , research activity was evidently enforce dafter 1995 . With regard to the landscape of Ebola research analyzed in this study , US-American institutions contributed the largest amount of international research . This finding again demonstrates the leading role of the USA in science , as it is present in almost all other areas of biomedical research [14] . We also showed a particular strong involvement of German institutions in Ebola research . This might be explained by the local availability of numerous suitable research facilities that were established after the first European outbreak of the Marburg virus . This virus is termed the “forgotten cousin” of Ebola and also causes life threatening hemorrhagic fever [15] . The application of the number of the total population and socio-economic parameters changed the ranking of the nations: Whereas the apparent superiority of the USA research was put in perspective , the African countries that are traditionally most affected by the disease such as Gabon , Republic of Congo , Central African Republic , Cameroon and Uganda increased their ranking . Only a few studies evaluated the worldwide research efforts in the field of tropical diseases . These also found a predominance of North American and European countries regarding overall research activity . As examples of countries where particular diseases play a significant role for the health of the public , only Brazil , where yellow fever is endemic [16] , as well as Brazil and India , which are affected by Leishmaniasis , were among the 10 most prolific countries [17] . Even high prevalent infections such as malaria have not lead to a increased global impact of research originating from African countries [18] . From its first appearance in scientific databases Ebola research was characterized by a high percentage of collaborative studies as demonstrated in our study . It has been shown numerously that involved scientists benefits from the international cooperation [19] . Although the US is the favorite global cooperation partner for scientists from other countries , it had the lowest cooperation ratio in regard to its own overall publication activity . This might be caused by the fact that American scientists cooperate to a large extend with national colleagues due to the efficient and well-funded academic structure that is present in the US . When focusing on the collaboration of institutions committed to Ebola research , we found a network that favors only a limited amount of institutions . This might be explained by the necessary prerequisite of Ebola research . The risks involved in handling the virus require the maximum biosafety level [20] . Research facilities that provide these resources are sparsely distributed throughout the world and its highest numbers are found in the US–the leading country of Ebola research output . Subject categories in health research represent the interest of scientist in different aspects of a disease . In the beginning of Ebola research , publications dominated that were attributed to the categories of microbiology , internal medicine and virology . Then the field became more diverse including other subjects such as immunology , cell biology , pharmacology , and biochemistry . Research in the field of public health encompasses society-based measures to combat diseases . We want to point out that a lack of publication output regarding public health topics is apparent in the field of Ebola research , which is in sharp contrast to other tropical diseases [16] . In conclusion , we here present a first detailed analysis of the global Ebola research landscape . The collected data indicated that the efforts in scientific research have been constantly increasing since the time of discovery in 1976 . The USA was identified as being the leading country and a total of more than 3000 publications . However , the danger of the virus , the change in pattern of distribution and the neglect to put more emphasis on the development vaccines before the outbreak of 2013–2015 clearly point to the need that 1 ) research in the field of hemorrhagic fevers needs to be strengthened , 2 ) vaccine development should also be enforced for other neglected tropical diseases in order to prevent similar catastrophes in the future , and 3 ) research endeavors should be focused on the area of public health since we could identify a neglect in Ebola related public health research efforts . | For the first time in the history of the disease , the Ebola virus left its local setting and affected people not only in isolated rural areas , but reached larger towns and cities leading to worldwide repercussions . This development prompted a joint global response to this health threat . This encompassed not only immediate relief efforts , but also an up search in global research work . In this study , the scientific output in Ebola research available in one of the mayor medical search platforms was characterized . We studied among others the origin of research , the collaboration between countries and the research topics . Partly , the obtained data was weighted against economic parameters . We attained a detailed map of the research activities from the discovery of Ebola in 1976 up to today . Our research provides the first overview of the worldwide Ebola research output . It might help stakeholders in Ebola research to better plan investigations with a global perspective . | [
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] | [] | 2015 | Ebola and Its Global Research Architecture—Need for an Improvement |
Tropheryma whipplei is a bacterium commonly found in feces of young children in Africa , but with no data from Asia . We estimated the prevalence of T . whipplei carriage in feces of children in Lao PDR ( Laos ) . Using specific quantitative real-time PCR , followed by genotyping for each positive specimen , we estimated the prevalence of T . whipplei in 113 feces from 106 children in Vientiane , the Lao PDR ( Laos ) . T . whipplei was detected in 48% ( 51/106 ) of children . Those aged ≤4 years were significantly less frequently positive ( 17/52 , 33% ) than older children ( 34/54 , 63%; p< 0 . 001 ) . Positive samples were genotyped . Eight genotypes were detected including 7 specific to Laos . Genotype 2 , previously detected in Europe , was circulating ( 21% of positive children ) in 2 kindergartens ( Chompet and Akad ) . Genotypes 136 and 138 were specific to Chompet ( 21% and 15 . 8% , respectively ) whereas genotype 139 was specific to Akad ( 10 . 55% ) . T . whipplei is a widely distributed bacterium , highly prevalent in feces of healthy children in Laos . Further research is needed to identify the public health significance of this finding .
Tropheryma whipplei is a common bacterium found in human feces; its prevalence depends mainly on age , exposure to saliva and/or human feces , and geographical area [1–8] . The prevalence in feces is estimated to be between 1 and 11% of European adults , but reaches 75% in children <4 years old in rural Senegal and 38% in relatives of those with Whipple’s disease or chronic carriers of the bacterium in France [1 , 3 , 5 , 7] . T . whipplei carriage was estimated to be 12 . 9% in homeless people in France and between 12 to 26% among European sewerage workers [1 , 6 , 9] . In France , this bacterium was also detected in the feces of 15% of children aged 2 to 4 years with gastroenteritis but not in control subjects of the same age [8] . T . whipplei is viable in culture on Specific Axenic Medium ( SAM ) using the feces of patients with Whipple's disease [10] and data strongly suggest that it is contagious [3 , 5 , 6 , 8 , 11 , 12] . Indeed , specific clones were observed with a high prevalence in children from the same city with gastroenteritis , among individuals from the same village in Senegal , among relatives of patients with Whipple’s disease and chronic carriers and among homeless people living in the same temporary visitor centre [3 , 5 , 6 , 8 , 11] . Recent studies suggest that T . whipplei infection is probably contracted during childhood [12] . There is a high prevalence of T . whipplei in rural Senegalese children <4 years old , up to 75% , that decreases to 30% among the 5–10 years old [12] . Moreover , in Senegal for the majority of children <6 years old , but not for those >6 years old , the presence of an immune response against the bacterium is consistently associated with the presence of the bacterium in their feces . This suggests that primary infection occurs before 5 years of age . T . whipplei is widespread and many individuals are exposed to this bacterium in Africa and Europe [1–3 , 5–9 , 11–14]; almost 72% of individuals in Senegal and 50% in France carried antibodies against the bacterium [3 , 5] . T . whipplei is involved in multifaceted conditions , including acute and chronic infections [15] . Although many are exposed to T . whipplei , few develop disease , possibly due to as yet undelineated predisposing immuno-genetic factors [2 , 3 , 5–7 , 11 , 12 , 15 , 16] . Data about T . whipplei in Asia are few , although T . whipplei has been reported in the saliva of healthy Koreans [17–20] . We therefore sought to determine the prevalence of T . whipplei carriage and identify genotypes circulating among children in Laos .
The initial objective of this study was to look for enterovirus carriage in Lao kindergarten children . It is only secondarily that the carriage of T . whipplei has been investigated in children . For this purpose , we analyzed the feces from healthy children collected between 2010 and 2012 in the framework of enterovirus study [21] in 3 kindergartens in Vientiane City , Lao PDR ( Laos ) , located in 3 different villages ( Sailom , Chompet , and Akad ) ( Fig . 1 ) . Although children were present at the kindergarten the day of sampling and thus considered healthy , they were also examined for conjunctival suffusion , pharyngeal erythema , rhinitis or runny nose , diarrhea , vomiting , mouth ulcer , vesicle on hands , and vesicles on feet . Fresh feces were collected in sterile containers from children aged 1–7 years whose guardians/parents gave informed written consent . After collection , each specimen was stored at+4°C up to 24 hours . Each specimen was vortexed in MEM medium ( Minimum Essential Medium , Invitrogen , Carlsbad , CA , USA ) , that is a commonly used in Lao laboratory as viral transport medium as recommended by WHO , with glass beads for 1 minute . Then , after 30 minutes centrifugation at 3 , 000g at +4°C ( adaptation of the protocol from the polio laboratory manual 2004 WHO/IVB/04 . 10 with the replacement of chloroform by antibiotics and fungizone ) , the supernatant was recovered and stored at -80°C for culture of enterovirus but the presence of antibiotics in the transport medium prevented the culture of T . whipplei . After enterovirus screening in Laos , 113 feces , from 106 children , were shipped on dry ice to virology laboratory in Marseille , France , for enterovirus confirmation ( 94/113 , 83% were confirmed to be positive for enterovirus ) , all were positive for β-actin and included in the present study . The study was granted ethical approval by the Lao National Ethics Committee for Health Research ( 331/NECHR ) and by the Oxford Tropical Research Ethics Committee ( OXTREC 020–10 ) , with amendments from both committees for feces collection for T . whipplei . DNA was extracted from 100μl of each sample using Machery Nagel DNA Mini Kit ( Machery Nagel Gmbh & Co . , Düren , Germany ) in accordance with the manufacturer’s instructions . Quantitative real-time PCR ( qPCR ) was performed using CFX96TM Real-Time PCR Detection System ( Bio-Rad Laboratories , Hercules , CA , USA ) with the FAST qPCR Master mix No ROX ( Eurogentec , Liege Science Park , Belgium ) . DNA extraction quality was verified for all specimens by qPCR targeting a gene encoding β-actin using primer pair ActinF ( 5’-CATGCCATCCTGCATCTGGA-3’ ) and ActinR ( 5’-CCGTGGCCATCTCTTGCTCG-3’ ) associated with a specific TaqMan probe ( 6-FAM-CGGGAAATCGTGCGTGACATTAAG-TAMRA ) . Negative β-actin specimens were excluded from analysis as previously described [22 , 23] . For T . whipplei detection , specimens were first tested using qPCR targeting a repeated sequence with the Twhi3F ( 5'-TTGTGTATTTGGTATTAGATGAAACAG-3’ ) and Twhi3R ( 5'-CCCTACAATATGAAACAGCCTTTG-3’ ) primer pair and the specific TaqMan probe Twhi3 ( 6-FAM-GGGATAGAGCAGGAGGTGTCTGTCTGG-TAMRA ) . When a sample was positive with this assay , the result was confirmed using a second qPCR targeting another repeated sequence with the Twhi2F ( 5'-TGAGGATGTATCTGTGTATGGGACA-3’ ) and Twhi2R ( 5'-TCCTGTTACAAGCAGTACAAAACAAA-3’ ) primer set and the Twhi2 probe ( 6-FAM-GAGAGATGGGGTGCAGGACAGGG-TAMRA ) . To validate each assay , qPCR mix was systematically used as a negative control and T . whipplei DNA as positive control , as previously reported [24 , 25] . Genotyping of T . whipplei was attempted for each positive specimen and performed as described [26] . Briefly , four highly variable genomic sequences ( HVGSs ) of T . whipplei were amplified and sequenced . The genotype was obtained after the analysis of each four HVGSs obtained from each specimen and compared with those available in both the GenBank database and our internal laboratory database in order to determine their corresponding genotype . PASW statistics 17 software ( SPSS , Chicago , IL , USA ) was used for data analysis and non-parametric values were compared using χ2 or the Fisher’s exact tests . Statistical significance was defined as p<0 . 05 .
We analyzed 113 feces from 106 children ( 59 [56%] females ) aged 1–7 years ( mean 4 standard deviation [SD] ± 1 . 22 years ) . Fifty three of 106 children ( 50% ) were from Chompet , 35 ( 33% ) from Akad , and 18 ( 17% ) from Sailom kindergartens . Among 106 children , 51 ( 48% , 95% CI 39–58 ) had a positive qPCR for T . whipplei . The prevalence of positive samples in Chompet was 27/53 ( 51% , 95% CI 38–64 ) , that was not statistically different ( p = 0 . 506 , OR 1 . 2 , 95% CI 0 . 58–2 . 57 ) to those from Akad and Sailom with 17/35 ( 49% , 95% CI 32–65 ) and 7/18 ( 39% , 95% CI 19–62 ) positive , respectively . Seven children provided more than one fecal sample; 3 were negative in the first year but positive the following year . Two were positive for 2 years and 2 were negative for 2 consecutive years . Children aged ≤4 years were significantly less frequently positive for T . whipplei ( 17/52 , 33% ) than older children ( 34/54 , 63%; p<0 . 001 , OR 0 . 12 , 95% CI 0 . 29–0 . 68 ) . The prevalence of T . whipplei carriage in feces from Lao children and those from rural Senegal are summarized in Fig . 2 [5] . Lao children aged ≤4 years were significantly less frequently positive for T . whipplei than Senegalese children ( 33% versus 75% , p<0 . 001 , OR 0 . 16 , 95% CI 0 . 07–0 . 37 ) whereas Senegalese children aged >4 years were significantly less frequently positive for T . whipplei than Senegalese children ( 30 . 3% versus 63% , p<0 . 001 , OR 4 . 25 , 95% CI 2 . 04–8 . 92 ) . Among 51 Lao children T . whipplei positive , genotypes were obtained for 19 samples . We obtained 8 different genotypes . Genotype 2 has been detected in France and Germany but 7 were new genotypes described from Laos for the first time ( genotypes 136 , 137 , 138 , 139 , 140 , 141 , and 142; Fig . 3 ) . Genotype 2 was detected in 4/19 ( 21% ) children , 3 were from Chompet and one from Akad kindergarten . New genotypes 136 and 138 were both detected in Chompet kindergarten in 4/19 ( 21% ) and 3/19 ( 15 . 8% ) , respectively , whilst genotype 139 was identified in Akad kindergarten in 2/19 ( 10 . 5% ) . The new genotypes 140 , 141 , and 142 were each detected in one child in each kindergarten ( Fig . 3 ) . Only one genotype ( 137 ) occurred in all three kindergartens , but was only identified once in each . From the two children positive on two different occasions , we obtained successive genotypes from only one child; in whom the same genotype ( 138 ) was found after an 8 month interval . Among the children included in this study , only 1 presented pharyngeal erythema and 15 rhinitis . All the other symptoms , including diarrhea , were not observed in any of the children . Among the 51 children positive for T . whipplei , 5 exhibited rhinitis ( 9 . 8% ) .
We describe T . whipplei in the stools of apparently healthy children for the first time in Asia . The data are based on strict experimental procedures , including positive and negative controls to validate each assay . The positivity of the β-actin gene was checked and confirmed for each included sample [2] and each T . whipplei qPCR was systematically confirmed by the successful amplification of another specific DNA sequence . Overall , the prevalence of T . whipplei carriage in feces from Lao children aged <7 years is 48% [5] . The only symptom observed among children positive for T . whipplei was rhinitis ( 9 . 8% ) but this symptom was also observed in 16 . 1% of the negative children . Thus , no link between the presence of rhinitis and that of the bacterium can be established . Lao children were contaminated with T . whipplei significantly later than those of rural Senegal . In France T . whipplei DNA was not detected in feces of healthy children <6 years old [8] . As far as we aware no one has investigated T . whipplei fecal carriage in Asia , but it has been detected in Korea among 1 . 5% of healthy peoples’ saliva , but not in any of 108 small-bowel biopsies in Malaysia or in any of ten marginal supragingival plaque samples from Chinese patients with gingivitis [17 , 18 , 20] . In rural Senegal , the first contact with the bacterium occurred mainly in children before 4 years of age , with the prevalence of T . whipplei in feces declining from 75% in those ≤4 years to 30 . 3% among those >4 years [5] . This trend was confirmed by a recent study performed in Gabon , with a high prevalence of T . whipplei in children <5 years old , up to 40% , with a lower prevalence ( 36 . 4% ) among those aged 5–10 years and 12 . 6% among those of 11–20 years [27] . In contrast , in Laos we observed a significantly lower prevalence in children ≤4 years ( 33% ) than in children >4 years ( 63% ) . A current hypothesis is that T . whipplei is transmitted between humans according to hygiene conditions [5 , 7] , through saliva ( oral-oral transmission ) if hygiene conditions are good , or feces ( fecal-oral transmission ) if hygiene poor [28] . T . whipplei has rarely been identified in environmental samples and humans appear to be the main reservoir and source of this bacterium in rural Senegal [7 , 12] . Poor sanitary conditions and close contacts between children from the earliest age could explain the difference in exposure to T . whipplei between Senegalese and Gabonese children versus Laotian children for whom contacts with other children are probably much less common before entering kindergarten [7 , 27] . In Senegal , households with high prevalence of T . whipplei carriage had also significantly less access to toilets , supporting the inter-human transmission of the bacterium being from human feces through hand transmission . In Laos , toilets were present at home for all included children and also at kindergarten . However , no information was available about the hand washing practice . Finally , close contact could play a role in transmission of T . whipplei as well as the lack of hand hygiene [6] . In Europe and Africa , 135 different genotypes have been identified from 359 people positive for T . whipplei . Eight genotypes of T . whipplei , including 7 newly described , were identified in feces from healthy Lao children . Previous work showed that T . whipplei genotypes found in Senegal and in Europe were different [2 , 4–6] . Genotype 2 was detected in 4/19 ( 21% ) Lao children , 3 were from Chompet and one from Akad kindergarten . This genotype had been previously detected in 12 out of 359 other people ( 3 . 34% versus 21%; p<0 . 001 ) in Europe , but not in Senegal . The presence of highly prevalent clones specific to some kindergartens , such as genotypes 136 and 138 in Chompet kindergarten , suggest that these clones are probably epidemic and support the contagiousness of T . whipplei . Furthermore , it is important to underline that currently no specific genotypes have been associated with disease versus asymptomatic carriage . Besides , a same genotype may be observed in acute infections , chronic infections and/or asymptomatic carriage [4] . Overall , T . whipplei genotyping is a useful tool to differentiate a relapse of Whipple’s disease to a reinfection with another genotype but also to detect circulating or epidemic clones [3–6 , 8 , 26 , 29 , 30] . Chronic infections due to T . whipplei such as classic Whipple’s disease or endocarditis have been scarcely reported in Asia and no such patients have been reported from Laos or adjoining North East Thailand [31] . To the best of our knowledge , only two Japanese patients and three individuals from India with Whipple's disease have been reported in the literature [31 , 32] . These data reinforces that although many people are exposed to T . whipplei , only few individuals develop chronic infection [4] . The prevalence of T . whipplei carriage is high in areas where poor sanitary conditions and close contacts are observed but T . whipplei seems to be an opportunistic bacterium that causes chronic infections probably linked to an , as yet unknown , specific immunological defect . Indeed , several dysregulated T-cell functions , a persistent deficiency in , or the absence of , a T . whipplei-specific T-helper cell-1 response have been observed in the patients with classic Whipple’s disease [33] . Furthermore , the susceptibility of patients to Whipple’s disease , with relapses and reinfections with new strains , along with a lack of , or a weak , serological response supports also a specific T . whipplei immunodeficiency [30 , 34] . Several observations support a genetic predisposition for Whipple’s disease . Indeed , the majority of patients reported to date are Caucasian , mainly males [4 , 29 , 35 , 36] . It is also remarkable for a rare disease that six familial proven cases of Whipple’s disease have been reported in the literature [15 , 36 , 37] . A higher frequency of the HLA alleles DRB*13 and DQB1*06 in patients with classic Whipple’s disease has been observed [38] . All these data may explain why T . whipplei carriage is so high in Africa and Asia relative to Europe , whereas there are so few or no confirmed cases of chronic Whipple's disease in these populations . The high prevalence of T . whipplei carriage in the feces of Lao children , confirms that this bacterium occurs in South East Asia and that infection occurs in childhood . Further research is needed to better understand the natural history , public health significance , and effect of T . whipplei on the health in Laos . | Tropheryma whipplei is a common bacterium carried in feces of young children . Here , using specific PCR , we estimated the prevalence of T . whipplei in 113 feces from 106 children in Vientiane , the Lao PDR ( Laos ) . T . whipplei was detected in 48% ( 51/106 ) of children . Eight genotypes were detected , including 7 specific to Laos . Genotype 2 , previously detected in Europe , was circulating ( 21% of positive children ) in 2 kindergartens ( Chompet and Akad ) . Genotypes 136 and 138 were specific to Chompet ( 21% and 15 . 8% , respectively ) , whereas genotype 139 was specific to Akad ( 10 . 55% ) . Long regarded as a rare bacterium , now we can affirm that T . whipplei is a widely distributed bacterium , highly prevalent in feces including those from children in Vientiane . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [] | 2015 | High Prevalence of Tropheryma whipplei in Lao Kindergarten Children |
Stochasticity in gene expression affects many cellular processes and is a source of phenotypic diversity between genetically identical individuals . Events in elongation , particularly RNA polymerase pausing , are a source of this noise . Since the rate and duration of pausing are sequence-dependent , this regulatory mechanism of transcriptional dynamics is evolvable . The dependency of pause propensity on regulatory molecules makes pausing a response mechanism to external stress . Using a delayed stochastic model of bacterial transcription at the single nucleotide level that includes the promoter open complex formation , pausing , arrest , misincorporation and editing , pyrophosphorolysis , and premature termination , we investigate how RNA polymerase pausing affects a gene's transcriptional dynamics and gene networks . We show that pauses' duration and rate of occurrence affect the bursting in RNA production , transcriptional and translational noise , and the transient to reach mean RNA and protein levels . In a genetic repressilator , increasing the pausing rate and the duration of pausing events increases the period length but does not affect the robustness of the periodicity . We conclude that RNA polymerase pausing might be an important evolvable feature of genetic networks .
The stochastic fluctuations in the expression level of a gene under constant environmental conditions [1] , arises from the stochasticity of the chemical reactions and other steps comprising transcription and translation [2] . This is further enhanced by the low amounts of RNA polymerases ( RNAPs ) and transcription factors present in cells . This stochasticity affects cellular functioning [3]–[4] , differentiation [5]–[6] and adaptability of organisms to the environment [7]–[8] , besides having implications in pathological processes [2] , [9]–[10] . Better insight into the sources of this stochasticity helps in understanding cellular dynamics and generation of phenotypic diversity of genetically identical cells . Most previous studies have focused on the noise in transcriptional initiation [5] , [11]–[13] . However , transcriptional elongation has recently been shown to be an important source of noise in transcript levels [12] , [14]–[16] . Transcription elongation is not a constant forward process . The noisy stepwise progress of RNAP through the DNA template is further affected by pauses , arrests , pyrophosphorolysis , misincorporations and editing [17] . RNAP pausing is an important regulator of transcription in both prokaryotes and eukaryotes , including in genes associated with human breast cancer [18]–[20] . A pause is defined here as an event where the RNAP is halted at a nucleotide , according to the definition in [20]–[21] . We distinguish such pauses , sometimes referred to as “ubiquitous pauses” , from other means of delaying elongation , such as arrests or backtracking [20] , [22]–[23] . Pausing is spontaneously reversible , after which the RNAP resumes movement [14] . Its duration varies , following an exponential distribution [14] . Longer pauses , over ∼20 s , appear to occur at specific DNA template points , while most pauses last less than 10 s [14] . Given their high frequency of occurrence , pauses ought to be explicitly included in models of transcription at the single nucleotide level [15] . This is of particular importance if multiple RNAPs are on the DNA strand , as pauses enhance the probability of collisions between RNAPs . Promoter-proximal pausing has been estimated to occur at above average rates in 10–20% of promoters in Escherichia coli , suggesting that it is a commonly used regulator of gene expression [24]–[25] . Dynamically , a pause is a kinetic pathway that competes with elongation and other events at each nucleotide , and the elongation-competent state to which an RNAP returns after pausing is always the same [14] . Measurements suggest that pauses are independent of factors such as the length of the growing RNA [14] . In E . coli , the average rate of pausing is 0 . 55 s−1 ( i . e . , approximately once in every 100 bases ) [17] , [26] and their average duration is 3 s [17] . Values vary widely from gene to gene , as pause densities and lifetimes are sequence-dependent [14] , suggests that the pausing mechanism is evolvable at the single gene level , e . g . , by selecting in or out pause prone sequences . While the high propensity of some sites to pauses is sequence-dependent , pause propensity in other sites appears to be regulated by molecules such as GreA and elongin complex that can suppress pausing [27]–[28] . Such elongation factors might regulate the timely expression of many genes , e . g . , during development [25] , [29]–[30] . If so , these regulatory molecules might allow fast changes in pausing propensity , e . g . , as a response to environmental stress . It should be stressed that not all pauses are sequence-dependent . They can be random in the sense that they can arise solely due to the probabilistic nature of stepwise elongation , or be rare but unavoidable ( i . e . certain to occur ) such as when due to DNA lesions [31] . Recently , a model of transcription in prokaryotes [15] that includes explicitly the promoter open complex formation step and models elongation at the nucleotide level was proposed and successfully confronted with measurements of gene expression at the single molecule level [32]–[33] . This model [15] is based on the model proposed in [12] but additionally includes several alternative pathways to elongation , namely pausing , arrest , misincorporation and editing , pyrophosphorolysis , and premature termination [15] , [17] . The analysis of the dynamics of this model suggested that pausing is , potentially , one of the major enhancers of the occurrence of collisions between RNA polymerases on the DNA template [15] thus , of transcriptional bursting [32] , [34]–[35] . Collisions between RNAP molecules affect nonlinearly RNA production intervals by enhancing what we refer to as “microbursts” , that is when two or more RNA molecules are completed within an interval much smaller than the expected minimum interval between consecutive transcription initiations [15] . While the stochasticity of stepwise elongation causes some microbursts , we show that pauses , within realistic parameter values intervals , can significantly vary the probability of occurrence of these events . Microbursting may affect cellular development , if used to cause RNA levels to overcome thresholds for short time periods , so as to , e . g . , initiate differentiation cascades [25] . Since there are between one and a few copies of most mRNAs in cells and since several cellular processes can be initiated given a single or very few molecules [36]–[37] , pausing might be a viable mechanism for cells to reach such thresholds . Using the delayed stochastic model of transcription at the single nucleotide level proposed in [15] we investigate how pauses' average duration and rate of occurrence affect the dynamics of transcription , translation , and a small gene network , the repressilator . Focusing on the “mean rate and duration of pause” of DNA sequences and on sequence specific long pauses , we address the following questions . Can the pausing rate and the average duration of pauses , when varied within biologically realistic values , be used to affect the transcriptional and translational dynamics ? Which features of transcriptional dynamics are affected by pauses ? Are the effects at the single gene level relevant in the dynamics of genetic networks ? First , we describe the model of transcription at the single nucleotide level . Next , we present our results regarding the effects of varying pausing rate and average duration in the transcriptional and translation dynamics of a gene . Finally , we present the effects of RNAP pausing on the dynamics of the 3-gene negative feedback loop; the repressilator [38] . In the end , we measure the effects of specific long-pause sites on the dynamics of transcription . We show that RNAP pausing , with biologically realistic values , has important effects on the single gene and at the gene network level , and therefore needs to be accounted for in models of transcription .
Besides the stochasticity , another important feature of the dynamics of gene expression is the time that some steps in transcription and translation take to be completed once initiated . E . g . , the promoter open complex formation can take from a few seconds to several minutes [39] , and affects significantly the dynamics of gene networks [13] . For that reason , stochastic algorithms have been proposed to simulate chemical reactions with time delays . In [40] , a delay Stochastic Simulation Algorithm ( SSA ) was proposed ( from which the delayed SSA [12] was later developed ) that allows explicit delays in protein production . A similar algorithm was independently proposed in [41] . The algorithm proposed in [12] differs from these , in that it can handle more than one delayed generating event for one reacting event . Thus , we use the delayed SSA [12] , which uses a waitlist to store delayed output events and proceeds as follows [42]: Delayed events in reactions are represented as , e . g . : A→B+C ( τ ) . When this reaction occurs at moment t , B is instantaneously produced at t and C placed on a waitlist until it is released , at t+τ seconds . τ can be drawn from a distribution each time the reaction occurs . Transcription is the reading of a gene in the DNA strand by an RNA polymerase ( RNAP ) and forming it into an RNA molecule . The RNAP unwinds and reads the DNA , producing the RNA by adding matching nucleotides while going through the DNA strand . Transcription has three main phases: initiation , elongation , and termination . In initiation , the RNAP attaches to the promoter and unwinds a portion of the DNA double helix to expose the template DNA strand ( promoter open complex formation ) . After that , the RNAP starts moving on the DNA strand and elongation , forming of the RNA molecule , begins . Behind the region where ribonucleotides are added , the RNA chain is displaced and the DNA double helix is reformed . In termination , a single-stranded RNA molecule is released , ending the transcription process . Recent models of transcription at the single nucleotide level were proposed in [12] , [16] . The model proposed in [16] includes backtracking and was used to study the distribution of elongation times , showing the relevant role of backtracking . These and other models [43] do not include , besides the promoter open complex formation , several alternative pathways to elongation that have been shown to play a role in transcription regulation ( e . g . , arrest ) [17] , [20] . For that reason , we use the delayed stochastic model of transcription at the single nucleotide level proposed in [15] that incorporates the promoter occupancy time , pausing , arrest , misincorporation and editing , pyrophosphorolysis , premature termination , and accounts for the range occupied by an RNAP when on the DNA template [17] , [44] . As most measurements of transcriptional dynamics are from E . coli , all parameter values in the model are from E . coli . This model of transcription is described in detail in [15] . Here we present explicitly the reactions modeling the promoter open complex formation ( reaction ( 1 ) ) , stepwise elongation ( reactions ( 2 ) and ( 3 ) ) where nucleotides are added one at a time to the growing RNA molecule , pause events ( 4 ) , and pause release ( reactions ( 5 ) , ( 6 ) and ( 7 ) ) which can occur by various means . A time delayed reaction ( 1 ) models the formation of the promoter-RNAP complex [39] , to account for the time during which the RNAP is not moving and occupies the promoter , preventing further transcription initiations . In this reaction , RNAP . Pro , which represents the complex of the RNAP bound to the promoter , has a delay τoc , represented by RNAP . Pro ( τoc ) , meaning that it takes τoc seconds for RNAP . Pro to be produced after the reaction occurs . Each time the reaction occurs , the delay τoc on the promoter release is randomly drawn from a Gaussian distribution with a mean of 40 s and standard deviation of 4 s , according to measurements on an active unrepressed lacZ promoter [45]–[46] , in agreement with previous measurements [39] . In ( 1 ) , Pro stands for the promoter while kinit is the stochastic rate constant of the reaction which is set to 0 . 0148 s−1 [15] . We assume at all times 28 RNAP molecules available for initiating transcription [47]: ( 1 ) After the delay elapses and if the first 13 nucleotides are unoccupied ( due to the steric hindrance of a possible preceding RNAP molecule ) , the RNAP can initiate elongation . When it does , it occupies the first nucleotide and the promoter becomes available for future reactions . As mentioned , in elongation , at each nucleotide , the forward movement of the RNAP is in constant kinetic competition with other regulatory pathways [17] , namely pausing and other mechanisms that act at this stage [21] , [25] ( e . g . , arrests ) . Each pathway has a propensity of occurrence and the choice is probabilistic , biased by the propensities . The most likely event is stepwise elongation if the RNAP is on a given nucleotide , in an activated state . Transcription stepwise elongation has two stages . First , the RNAP moves from an activated nucleotide An ( already transcribed ) to occupy the next nucleotide , providing there is no steric hindrance from the succeeding RNAP ( reaction 2 ) ( where Δ = 12 is half the number of nucleotides occupied by an RNAP ) [15] . In ( 2 ) , the rate kmove is 150 s−1 ( to achieve an average elongation speed 75 nucleotides/s [17] ) . Let n be a nucleotide such that n = 1 , … , N , where N is the total number of nucleotides that the RNAP goes through during elongation . Reaction ( 2 ) models one of the possible chemical pathways that can be followed by the RNAP , namely moving from nucleotide to the next nucleotide , once activated: ( 2 ) Once the RNAP occupies nucleotide On+1 ( and frees nucleotide U ( n-Δ ) ) , the most probable pathway is activation ( reaction 3 ) , after which the RNAP can again move forward . In this step , a complementary nucleotide is added to the growing RNA [15] . ( 3 ) We set the activation rate , kact , to 150 nt/s , to attain an elongation rate of 75 nt/s ( the sum of kmove and kact ) on average [17] , [44] . The elongation rate can vary , e . g . , with the growth rate of E . coli [48] . The value assumed here is consistent with a duplication time of 55 minutes of E . coli [33] . Elongation is frequently interrupted by pauses [14] , [20] , [49] ( reaction 4 ) , where the RNAP is halted at a nucleotide [21] . Pause durations vary . For instance , longer pauses last over 20 s , and are reported to be more sequence-specific than shorter ones . This class of pauses can also be driven by the secondary structure , such as the hairpin loop from the his operon . Most “ubiquitous” pauses last less than 10 s [14] . The average pausing rate is kpause = 0 . 55 s−1 [17] . Note that , in this model , reaction ( 4 ) competes with ( 3 ) , which is reflective of the “kinetic partitioning” of active and paused RNAP in the cell . The relative value between their rates determines the fraction of times each occurs [50] . Since kpause∼kact/136 , a pause event occurs , on average , every 136 activation events which , in a template of 2445 nucleotides ( tsr-venus gene [33] ) is significant , causing collisions between RNAP molecules at high expression rates . ( 4 ) The paused complex is usually spontaneously released after a certain time duration which follows an exponential distribution [13] via reaction ( 5 ) ( on average , after dpause = 3 seconds [14] ) . It can also be released due to a collision ( reaction 6 ) with the next elongating RNAP [51] . The collision can instead cause the next RNAP to pause as well ( reaction 7 ) [51] . This is set empirically to occur in 20% of collision events ( reaction 7 ) . ( 5 ) ( 6 ) ( 7 ) This model of pausing , comprising four reactions [15] , is similar to the one proposed in [14] , which matched experimental measurements , but the events modeled by reactions ( 5 ) and ( 6 ) are there modeled in a single reaction , not specifying the mechanisms for the end of the paused state . Other events , such as arrest , misincorporation and editing , pyrophosphorolysis , and premature termination , can also occur at any nucleotide and are modeled similarly to pauses , with rate constants extracted from measurements . A complete description of the model ( and necessary references ) can be found in [15] . Here , in Table 1 we show the reactions modeling each of these events , and the rate constants used in the simulations . When the termination sequence is reached , the transcription bubble collapses as the RNA-DNA hybrid disrupts , releasing both the RNAP and the completed RNA molecule ( “R” ) . Reaction ( 8 ) models termination . When the last nucleotide is activated and the mature R is released , the RNAP is also released , unoccupying ( U ) the last 12 nucleotides . The rate for the transcript release , kf , is 2 s−1 [52]: ( 8 ) In our model , translation is modeled as a multi-delayed reaction ( 9 ) that accounts for variable time needed to complete a functional protein , P , due to the time taken by translation , folding , activation , etc [13] , [53] . The delay τ3 associated to the production of a protein follows a normal distribution ( we choose the normal distribution since the distribution has not yet been experimentally assessed , only mean and variance have [13] ) . ( 9 ) In ( 9 ) Rib is a ribosome and the values for the delays were extracted from measurements [33] . The length of the gene tsr-venus driven by a Lac promoter studied in E . coli is 2445 nucleotides [33] . The post-translational protein assembly process was observed to take 420±140 s in [33] , thus τ3 was set in accordance . The time of the R clearance in translation initiation , τ1 , is set to 2 s [54] , as translation can begin again as soon as the ribosome binding site is available . The average translation rate is 15 amino acids/s , thus we set τ2 = τ1+2445 nt/ ( 45 nt/s ) = 56 s [13] . A note is needed regarding how translation is modeled ( reaction 9 ) . We use a multi-delayed reaction ( from [53] ) instead of a set of reactions similar to transcription , at the single nucleotide level . Because of this , translation only starts when a complete RNA molecule is produced , rather than when the ribosome binding site is complete . The use of the multi-delayed reaction is necessary due to the computational complexity of having a translation model at the single nucleotide level but hampers the possibility of initiating translation when the ribosome binding site region of the RNA is complete . However , it is noted that in our approximate model , pauses still directly affect the bursting dynamics of proteins , and similarly to how they would in a more detailed model . Namely , pauses in transcription will enhance the broadening of the time intervals between the completions of consecutive proteins as shown in the results section . In our model of translation , the delay ( τ3 ) associated to the completion and release of the protein varies from one translation event to the next . Thus , the model copes with variability in the speed of translation and consequent different durations of translation events in normal conditions . However , if many collisions occur between ribosomes the model loses accuracy . One case where , therefore , the model becomes less accurate is if , during translation , long pauses occur . Thus , our model assumes that there are no long pauses in the process of translation or , at least , that these are very rare , in agreement with the measurements from which the mean duration and variability of τ3 were extracted [33] . If , for some specific gene sequence , such pauses do occur frequently , it is likely that they will enhance the bursting in protein levels . When simulating the model , both RNA and proteins are subject to degradation , modeled as first order reactions . Reactions ( 10 ) and ( 11 ) model degradation of RNA and proteins , respectively: ( 10 ) ( 11 ) Given that ktr = 0 . 00042 s−1 [13] , that there are 100 ribosomes available for translation [13] and that , on average , there are 1 . 2 R available for translation ( given that R goes into the ‘waitlist’ when reacting in a translation event ) , we set the protein degradation rate degP to 0 . 0003 s−1 ( 55 min−1 , using in vivo parameter values [13] ) so that the mean level of proteins at equilibrium is ∼150 . Similarly , degR ( ∼0 . 1 s−1 ) is set so as to impose a mean level of ∼5 transcripts .
The transcription model at the single nucleotide level used here [15] exhibits transcriptional bursting as reported in [32] ( defined as the periods during which RNAs are produced , versus what appear to be relatively long periods of inactivity of the promoter ) . It was observed [15] , [32] that during the periods of activity , there are sudden increases in the amounts of RNA molecules . These ‘microbursts’ were shown to be due to the completion of two or more RNA molecules within intervals shorter than the average duration of the promoter open complex formation [15] , which in the model was set to follow a Gaussian distribution with a mean of 40 s and a standard deviation of 4 s [39] , [45] . Using the same model , we explore how the occurrence and duration of pauses contribute to transcriptional microbursting . The movement of an RNAP molecule on the strand is stochastic [2] , [39] , thus , two or more consecutive RNAPs may shorten their initial distance in the strand and complete transcription within an interval shorter than the duration of the promoter open complex formation ( leading to RNA microbursts , as defined here ) . Several events can enhance these bursts . For example , pyrophosphorolysis can cause a gradual shortening of the distance between consecutive RNAs , or the arrest of an RNAP can cause several RNAPs to accumulate behind the halted one . Pauses were shown be to a major enhancer of microbursts [15] . While a microburst is expected to , on average , transiently increase the amount of RNA by only 2 or 3 units , this can affect a cell's functioning since , for many genes , the RNA level range from 1 to a few [36] . Transient increases can affect , e . g . , differentiation [2]–[3] by overcoming thresholds that lead to a cascade of events . We measured the interval between consecutive completions of transcription events as we varied the pausing rate of occurrence and average duration . Results are shown in Figures 1–3 . Transcription initiation rate was set to kinit = 0 . 0148 s−1 [15] . We simulated 500 independent cells , each for 3300 seconds ( the lifetime of E . coli [33] ) , measuring the RNA level at a sampling frequency of 0 . 1 s . While the DNA template is initialized without RNAP molecules on it , the transient to reach a steady state flow of RNAP molecules on the DNA is negligible in comparison to the simulation time ( ∼150 s , i . e . 4 , 5% of the simulation time ) . Nevertheless , such transient has no effect on the results on the interval between transcription completions . Figures 1A and 1B show the time intervals between each pair of consecutive transcription completions for , respectively , three values of kpause and three values of dpause ( within experimentally observed ranges ) . It is noted that these intervals depend of the value for kinit , of the number of RNAP available at each moment ( here kept constant for simplicity ) , and of τoc . Namely , one expects the mean time between transcription initiation events to be ∼ ( ( RNAP*kinit ) −1+τoc ) , which equals , given our parameter values , ∼42 . 4 s ( this approximation neglects the first elongation step of the RNAP , which releases the promoter , as it takes negligible time [39] ) . From Figure 1A , as kpause increases , the distribution of intervals between completions changes from “Gaussian-like” , to “exponential-like” . Increasing dpause causes similar but stronger effects ( Figure 1B ) . This change implies that more pairs of RNAPs complete transcription unevenly , separated by much shorter or longer intervals than the promoter delay and interval between transcription initiations , a consequence of the stochastic pause events . We next measured the number of microbursts as we increase kpause and dpause ( Figure 2 ) . From the time series of the number of RNAs measured at a sampling frequency of 1 second , we calculated the fraction of times that two or more consecutive RNAs are produced in an interval smaller than 5 seconds . This interval is defined arbitrarily , excepted that in that it needs to be smaller than the average duration of the promoter open complex formation , according to the definition of microburst . We did not find qualitative differences in the results using other interval lengths . We set kpause to 0 , 0 . 1 , 0 . 2 , 0 . 5 , 1 , 2 , 5 , 10 , 20 , 50 , and 100 s−1 ( kpause can range from 0 . 1 to ∼1 s−1 in vitro [17] ) , and then we set dpause to 0 , 0 . 1 , 0 . 2 , 0 . 5 , 1 , 2 , 5 , 10 , 20 , 50 , and 100 s ( in vitro dpause ranges from 0 . 1 to ∼25 s [14] ) . When varying kpause we set dpause to 3 s , and when varying dpause we set kpause to 0 . 55 s−1 ( mean values [17] ) . We tested values of kpause above those experimentally observed to better examine the decrease in microbursting . For each set of parameter values , we simulated 10 independent cells , each for 50 000 seconds , sampled every second ( this unrealistically long life time provides better statistics , but one can equivalently measure more cells with shorter lifetimes as one is approximately measuring “steady state statistics” ) . From Figure 2A , for 0<kpause<10 , it is visible that the number of microbursts increases with kpause . For a pause to occur , it has to compete with several events such as arrests . The most probable is stepwise elongation ( kmove = 150 s−1 ) and subsequent activation ( kact = 150 s−1 ) . Since kpause is much smaller than these rates , it is unlikely that two consecutive RNAPs will both pause . Interestingly , for kpause>10 s−1 the number of microbursts decreases . This is due to kpause having the same order of magnitude as elongation , leading to most RNAPs constantly pausing at each nucleotide , without significant variation in the distance between consecutive RNAPs . Importantly , this suggests that there similarly is a maximum noise level in transcription attainable via selecting for sequences prone to pauses . Also , as the reason for the decrease lies in the relationship of the magnitudes of kpause and kmove , this result , and the value of kpause for which it occurs , is independent of the value of dpause , as the propensity of occurrence and duration of pauses is identical in all nucleotides . From Figure 2B , the effect of increasing dpause on microbursting is different from varying kpause . Confronting the y-axis scales of Figures 2A and 2B , one concludes that increasing dpause causes significantly more microbursts and that this increase is not limited as when increasing kpause . Notably , the average time between completions does not vary with either kpause or dpause , since increasing the number of microbursts necessarily is accompanied by an increase in the number of consecutive RNAPs separated by longer time intervals ( Figures 1A and 1B ) . Thus , varying kpause and dpause may tune the noise level of the RNA and proteins , but mean levels are left unaffected . We measured the number of RNAs in the largest microburst in each simulation for each value of kpause and dpause and averaged it over all cells with the same values of kpause and dpause ( Figure 3 ) . From Figures 2 and 3 one can conclude that there is a strong correlation between the number of microbursts and the size of largest microburst . The size of the largest microburst increases with dpause , while for kpause the result is more complex . Namely , for 0<kpause<10 , the size of the largest microburst increases with increasing kpause , and beyond these values ( kpause>10 ) the size of the largest microburst decreases with the increase of kpause . Interestingly , for kpause>50 , the maximum size is actually smaller than for kpause<1 , meaning that an increased frequency of pauses can , in principle , be used as a means to decrease the occurrence of microbursts . An important dynamical aspect of gene expression in a genetic network is the time that it takes for a gene , initially repressed , to reach its steady state protein expression level , once activated . This transient time is a measure of the “speed of response” of that gene to either an externally or an internally induced activation or halting of repression . We measured this transient as a function of kpause and dpause . We ran 100 simulations , each for 5000 s with a sampling rate of 1 s , for each set of parameter values of kpause and dpause described , except that for kpause the maximum value was 10 s−1 . The initial transient is defined here as the time it takes for the protein level to be equal or higher , for the first time , than its mean level over the total simulation time . We then averaged the results of the 100 simulations for each set of parameter values of kpause and dpause . The mean RNA level is ∼5 in all simulations and the mean protein level is ∼150 . The average transient length with one standard deviation error bars is shown in Figures 4A and 4B . The results suggest that increasing kpause only affects the transient for values beyond 0 . 5 s−1 . Similarly , the increase in dpause only increases the transient significantly for values beyond 5 s ( importantly both values are within realistic intervals ) . This effect on the transient has , as shown later , consequences on the dynamics of the repressilator . Notably , the variance of the initial transient does not vary significantly in the range of values tested of kpause , while for dpause it only increases significantly for dpause>20 s explaining why , later on , we observed that the robustness of the genetic repressilator is not significantly affected by varying kpause and dpause . We next study the effects of pauses on the noise of the RNA and protein levels of a single gene , given that both RNA and proteins are subject to degradation . Noise is quantified by the coefficient of variation ( CV ) , defined as the standard deviation over the mean level over time . The increase in kpause from 0 to 10 s−1 causes a 20% increase in RNA noise level ( Figure 5A ) and 15% in protein noise level ( Figure 5C ) . The increase in dpause from 0 to 100 s causes increases of ∼50% in RNA ( Figure 5B ) and of ∼110% in protein noise levels ( Figure 5D ) . Thus , increases in the frequency and duration of pausing leads to substantial increases in noise . It is interesting to speculate , given these results , that pauses may be a regulatory mechanism of transcriptional and translational noise . Variations of this order of magnitude are likely to affect the dynamics of genetic circuits . Nevertheless , it is noted that while the increase in fluctuations of RNA levels is clear and in agreement with studies on the effects of varying the distribution of time intervals between transcription completions [56] , one should be careful when drawing conclusions regarding effects of pauses in the protein noise levels , as many more variables and processes are involved . E . g . , proteins levels are also affected by post-translational regulatory mechanisms such as phosphorylation or dephosphorylation that are , in some cases , used to regulate degradation [57] , and that would affect protein noise level . Nevertheless , afterwards , when observing effects on the dynamics of the repressilator , we observe significant effects as kpause and dpause are varied in the same range of values . The results agree with the effects of pausing on microbursting . To investigate how kpause and dpause alter the dynamics of genetic circuit , we model a repressilator [58] Additionally to the reactions ( and parameter values ) described for gene expression , additional reactions are needed to model binding and unbinding of monomeric repressor proteins to the promoter regions of genes ( reactions 12 ) , to define the topology of the repressilator , and for protein degradation when bound to the promoter ( reactions 13 ) and when free ( reactions 14 ) : ( 12 ) ( 13 ) ( 14 ) In reactions ( 12 ) and ( 13 ) , i = {1 , 2 , 3} and j = i-1 , except for i = 1 , in which case j = 3 . In reaction ( 14 ) , j = {1 , 2 , 3} . The repression rate kr is 0 . 1 s−1 , the unrepression rate ku is 10−4 s−1 , and protein degradation rate ( degp_r ) is 0 . 01 s−1 . Importantly , setting kpause = 0 . 55 s−1 and dpause = 3 s ( the mean observed values ) causes the repressilator to have a period of ∼7 . 000 s , similar to measurements [58] . A precise matching can be achieved by , e . g . , tuning the protein degradation rate . Both kpause and dpause affect the period length , but not mean protein levels or period robustness ( Figures 6A and 6B ) . Increasing either kpause or dpause increases the mean period , due to the increase in transient to reach maximum expression level , as in the case of individual genes . Robustness of the periodicity was assessed by the 3-tuple information-entropy ( H ) of the time series of ( P1 , P2 , P3 ) , binarized with k-means [59] , from a time series of 107 s sampled every 100 s . Measures of periodicity robustness cannot be chosen according to any fixed criteria , thus , in each case the measure yielding the most plausible results should be selected [60] . We aimed to measure the robustness of the periodicity of the levels of the three proteins of the repressilator . This behavior is robust if the period length is constant over time and if there are no disruptions in the periodic increase and decrease of each protein levels . Note that changes in the period length , disruptions in the periodicity of the protein levels fluctuations , and more high-frequency noise in each protein level cause the 3-tuple H to be higher than otherwise . Let ( P1 , P2 , P3 ) ( t ) be the 3-tuple binarized states of the proteins levels at moment t . There are 8 possible states ( P1 , P2 , P3 ) , namely , states ( 0 , 0 , 0 ) to ( 1 , 1 , 1 ) . From the entire time series , one can assume the probability of being in each state to be , in approximation , the normalized fraction of times that that state occurs . Let i = 1 , … , 8 be the index of the state and Pri be the probability to be in state i . The 3-tuple information-entropy of the time series of the proteins is then given by ( 15 ) [61]: ( 15 ) The 3-tuple information entropy of the binarized states is ∼1 . 2 for all values of kpause and dpause tested ( same range of values as in the previous cases ) , indicating that the repressilator is robust to the increase of noise in the temporal levels of each protein . In accordance , in long time scales ( 107 s ) , the number of disruptions in the periodic behavior is identical in the three models . In Figures 7A , B and C , we show time series of the protein levels of three repressilators: ( A ) with kpause = 0 . 55 s−1 and dpause = 3 s , ( B ) kpause = 10 s−1 and dpause = 3 s , and ( C ) kpause = 0 . 55 s−1 and dpause = 100 s . As in the case of the individual gene's protein time series , the increase of kpause and dpause cause stronger fluctuations in the protein levels in case ( B ) , and even more in case ( C ) . We also measured the noise level ( CV ) of the protein time series ( Figures 8A and B ) . The effect of the periodic oscillation on CV is approximately removed by summing , at each time step , the amounts of P1 , P2 and P3 into a single quantity , here referred to as Ptotal , of which we measure the mean and standard deviation of the time series ( CV ) . Interestingly , an increase in noise level at the single gene level does not significantly affect the robustness of the repressilator's periodicity . This is because the repressive interactions between the genes via their proteins act as ‘noise filters’ . The ‘tunability’ of genetic clocks might be of key importance in varying environments , and the results suggest that pausing is a good candidate for an evolvable mechanism to adapt to environmental changes by tuning the period without affecting the robustness ( Figures 6A and 6B ) . The length of the initial transient of a gene to reach its mean expression level ‘at steady state’ increases with the increase of kpause and dpause ( Figures 4A and 4B ) . In a repressilator , the expression level of each gene goes to zero periodically . The increase in transient time ( via increased kpause and dpause ) of each gene to reach its maximum expression level causes the length of the period of the repressilator to increase . So far , we have focused on the effects of short duration pauses on gene expression . For simplicity , we have assumed an identical probability of pause occurrence and an identical distribution of pause duration at each nucleotide , irrespective of the sequence . In this section , we examine the effects on gene expression of longer pauses , which are known to exist in numerous organisms [20] , [22] , [62] , and that can last from 30 seconds to several minutes [20] , [62] . Such long pauses are sequence dependent , thus , occur at specific sites . One class of long pauses is stabilized by the formation of a “pausing hairpin” in the newly transcribed RNA . Analysis of the his leader pause site showed that the pause it causes has a half-life of 47 seconds and occurs with a probability of 80% , and it was suggested that it facilitates the synchronization of the RNAP and ribosome movements during transcription of the his operon [20] , [22] . Interestingly , it has been shown also that , depending on the spacing of the hairpin loop from the RNA 3′ end , and the nature of the intervening RNA sequence , the hairpin can prolong pausing or vary the chance of premature transcriptional termination [22] ( thought to be modulated by a direct interaction between a flexible loop on RNAP and the hairpin ) . This effect , as it is sequence dependent , is also likely to be subject to selection . Finally , hairpin pausing is also known to play a key role in termination of transcription , by halting RNAP at terminators until appropriate factors , such as Rho-factor , are recruited and the elongation complex is dissociated [63] . To study the effects of such long-duration pause sites in transcriptional dynamics , we examine three hypothetical sequences of 400 nucleotides . These , referred to as A , B and C , are in all ways identical except that we introduce , at nucleotide 200 , a long-pause in B and C with a 50% probability of occurrence for each RNAP that reaches that nucleotide , and with a mean duration of 60 s . Additionally , in case C , there is a 25% chance of premature termination at nucleotide 200 , if a long pause occurs . Note that each RNAP can pause only once at the long-pause site . In Figure 9 , the distribution of time intervals between transcription completion events in the three scenarios A , B , and C are shown . The effects of the probabilistic long-pauses and premature termination at nucleotide 200 are visible comparing the figures . We set a high transcription initiation rate so that , given the promoter open complex delay , transcription events are separated on average by 40±4 s intervals . When comparing cases A and B , it is apparent that the pause site causes the distribution to convert from Gaussian like ( case A ) to tri-modal ( case B ) . By introducing a long duration pause with a 50% probability of occurrence two new peaks emerge in the distribution . One is due to microbursting , and the other corresponds to the pairs of consecutive RNAPs separated by long time intervals . This separation of peaks was not observed when examining short duration pauses ( Figure 1 ) , as the reduced pause duration would not cause a significant interval of RNAP separation . Note that a single pause event causes an increase in both peaks , since the existence of a long interval demands the existence of a short interval , given the approximately constant rate of transcription initiation . A 50% probability of pause occurrence explains the heights of the peaks at ∼40 s and ∼80 s , which are half the height of the peak at ∼40 for case A , since approximately 50% of the intervals between consecutive RNAPs are doubled due to the pause site . Given the change in the distribution of intervals between completions , one can conclude that , assuming a first order degradation rate of RNA , the existence of the long-pause site causes higher noise in the RNA levels , due to the increase of microbursting . The effects of premature termination ( case C ) are also of interest . A 25% chance of premature termination following a long pause causes the number of consecutive RNAP pairs separated by short intervals ( <5 ) to decrease significantly but does not affect the number of pairs of consecutive RNAP separated by long intervals ( ∼80 s , in comparison to the normal ∼40 s ) . This is due to the fact that a premature termination cannot cause two consecutive RNAPs to shorten the distance between them , but decreases the number of pairs of RNAPs separated by short distances in the template as one of them falls off . Note that Figure 9 shows a probability mass , not the total number of cases , which decreases given the premature terminations by approximately 10% ( given equal simulation times ) . Interestingly , the premature terminations diminish the noise level in comparison to case B , as it decreases significantly the occurrence of microbursts . However , the noise is still higher in case C than in case A . As seen , the broadening of the distribution of intervals between transcription completions results in higher noise in RNA levels and , consequently , protein levels . Thus , sequence-specific long-duration pause sites are likely to lead to increasing RNA and protein noise levels .
Several recent studies have focused on the stochasticity arising from transcription initiation . Importantly , in elongation there are several events also contributing to transcriptional noise , such as pauses , arrests , or premature terminations . We studied the effect of pauses in elongation on transcriptional dynamics using a delayed stochastic model of transcription at the single nucleotide level that includes the promoter open complex formation , pausing , arrest , misincorporation and editing , pyrophosphorolysis , and premature termination . Our results show that varying pauses rate of occurrence and duration , within realistic parameter value intervals , affects the dynamics of transcription and protein levels , namely , bursting dynamics and the noise in transcripts and proteins levels . As noise in gene expression is subject to selection [64] , and while there are other mechanisms by which noise in RNA and protein levels can be tuned , e . g . transcription initiation rate [8] , it can be speculated that the existence or absence of sequence-specific pauses is subject to selection as they are a viable mean to regulate the noise level at the single gene level and , consequently , in gene regulatory networks . Interestingly , in agreement with our predictions that pauses lasting more than 10 seconds significantly increase noise in transcripts levels , measurements in E . coli of sequence-dependent pauses dynamics suggest that most of these pauses last less than 10 seconds [14] . It should be noted that the measurements in [14] were made in vitro , and unknown mechanisms may alter some of these pauses lifetime in vivo . Furthermore , there is evidence that cells use stochasticity in gene expression to cope with fluctuating environments [7] and that fluctuations in the levels of dosage-sensitive genes can be harmful [65]–[66] . Given that RNAP pausing affects the noise in gene expression and thus , the dynamics of genetic networks we suggest that pauses are an evolvable mechanism by which cells adapt the transcriptional noise of specific genes to cope with environmental stresses and changes . Pause rate of occurrence and duration affect size and number of microbursts in transcription . Size of the largest microburst and number of microbursts might have different and important roles in cellular metabolism . While increasing the number of microbursts increases noise of transcripts levels , increasing the size of the microbursts allows overcoming thresholds in RNA levels otherwise not reachable . The ability of RNAP pausing to regulate microbursting in RNA levels suggests that it might be a regulatory mechanism of cells' sensitivity to external stresses , and of probabilistic decision-making processes such as in cell differentiation and phenotypic variability . Initiation of differentiation usually requires reaching a protein concentration threshold to switch between pathways , as depicted by the French flag model [67] or the competence decision circuit of Bacillus subtilis [4] , [6] . The ability of a gene to produce strong but sparse bursts is of importance in this context . In agreement with this hypothesis , it has been suggested that transcriptional promoter proximal pausing , is crucial in the embryonic development of Drosophila , by being a source of transcriptional bursts [25] . Sequence-specific long pauses were shown here to be an ideal regulatory mechanism of bursts . Not only a single long-pause site can drastically alter the distribution of bursts , but it can do so without changing mean expression levels . Further , combining the long pause site with higher premature termination rate , allows making the distribution between completion of RNA molecules sparser without increasing the number of bursts . There are several evidences that noise in gene expression is subject to selection [1]–[7] , [64] and that bursts in gene expression play a key role in allowing the overcoming of thresholds in protein concentrations otherwise unreachable [11] , [25] . The fact that long-pause sites are tentative candidate regulators of transcriptional noise might be one of the reasons for the widespread occurrence of promoter proximal pausing in prokaryotes and eukaryotes [25]–[26] . Another possible reason might be its ability to coordinate transcription elongation with pre-mRNA processing [25] , but one usage does not exclude others , i . e . , pauses might be used for multiple purposes , one of these being the regulation of transcriptional noise . Our results further suggest that pauses are a likely regulatory mechanism of gene networks dynamics . For example , altering the rate and duration of pauses in the genes composing a repressilator enables tuning the proteins' time series period length . Interestingly , even for rates of pausing exceeding biologically observed values , the robustness of the periodicity was not affected , unlike when using other methods to alter the period length ( e . g . decreasing transcription initiation ) . Also , pausing may be used to , e . g . , tune the switching frequency of a genetic switch as switches are noise-driven , due to the effect on individual genes' expression noise . Importantly , both the pausing rate and expected duration are sequence-dependent [14] , implying that this regulatory mechanism of transcriptional dynamics is evolvable . The additional dependence of the propensity to pause on regulatory molecules suggests that pausing may also be a mechanism able to respond to changes in the cellular environment . In this context , it of interest to note that essential genes exhibit , in general , lower noise levels than nonessential ones [68] , suggesting evolvability in the noise level of individual genes [64] . Due to the effects of pauses in transcriptional noise and its sequence dependence , it is likely that this is one of the evolvable mechanisms , to tune individual genes' noise level as a function of the gene's task . Finally , while the values of pausing rate and duration tested here are within the range of biologically observed values , extending our studies to values beyond these ranges might provide insights into the potential applications of pausing in synthetically engineered genetic networks . | Investigation on how phenotypic diversity of genetically identical organisms is generated and regulated has focused on noise in gene expression . It is unknown to what extent noise in gene expression and genetic networks is evolvable , and by which mechanisms it evolves . The noise has several sources , e . g . , noise in transcription initiation and during elongation . We focus on RNA polymerase ( RNAP ) pausing and show that it can regulate , to some extent , noise in gene expression . RNAP frequently pauses during elongation . The pausing frequency and average duration are sequence-specific , thus evolvable . The dependency of pause propensity on regulatory molecules makes pausing a mechanism adaptable to rapidly changing environments . We study , in a stochastic model of bacterial transcription at the single nucleotide level that includes the promoter open complex formation , pausing , arrest , misincorporation and editing , pyrophosphorolysis , and premature termination , how pausing affects the dynamics of gene expression and gene networks . In a model of a genetic clock , with periodic dynamics , pauses affect the period length but do not disrupt the periodicity . We conclude that RNAP pausing is an important evolvable feature of gene regulatory networks , that can be used by organisms to adapt to changing environments and regulate phenotypic diversity . | [
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] | [
"computational",
"biology/synthetic",
"biology",
"molecular",
"biology/transcription",
"initiation",
"and",
"activation",
"computational",
"biology/transcriptional",
"regulation",
"computational",
"biology/molecular",
"genetics",
"molecular",
"biology/transcription",
"elongation"
] | 2010 | Effects of Transcriptional Pausing on Gene Expression Dynamics |
Protein misfolding is common across many neurodegenerative diseases , with misfolded proteins acting as seeds for "prion-like" conversion of normally folded protein to abnormal conformations . A central hypothesis is that misfolded protein accumulation , spread , and distribution are restricted to specific neuronal populations of the central nervous system and thus predict regions of neurodegeneration . We examined this hypothesis using a highly sensitive assay system for detection of misfolded protein seeds in a murine model of prion disease . Misfolded prion protein ( PrP ) seeds were observed widespread throughout the brain , accumulating in all brain regions examined irrespective of neurodegeneration . Importantly , neither time of exposure nor amount of misfolded protein seeds present determined regions of neurodegeneration . We further demonstrate two distinct microglia responses in prion-infected brains: a novel homeostatic response in all regions and an innate immune response restricted to sites of neurodegeneration . Therefore , accumulation of misfolded prion protein alone does not define targeting of neurodegeneration , which instead results only when misfolded prion protein accompanies a specific innate immune response .
Many chronic neurodegenerative diseases , such as Alzheimer disease , Parkinson disease and Transmissible Spongiform Encephalopathies ( TSE ) or prion diseases , are characterised by accumulation of misfolded proteins [1] . The appearance of detectable misfolded proteins and their relationship to neurodegeneration have been the major focus for defining disease mechanisms . However , it remains unclear what precise role misfolded proteins have in disease pathogenesis . The prion diseases provide valuable model systems to examine this relationship and define mechanisms of neurodegenerative disease . The host protein is an absolute requirement for disease because , in the absence of prion protein ( PrP ) , mice have been demonstrated to be resistant to disease [2 , 3] . Furthermore , PrP-/- mice that have PrP+/+ tissue grafts into the brain have demonstrated deterioration of PrP+/+ graft tissue but preservation of host PrP-/- neurons when experimentally infected with a prion disease [4] . At clinical stages , misfolded protein aggregates are usually detected in brain regions undergoing overt neurodegeneration . The accumulation and aggregation of misfolded proteins precede detectable neurodegeneration [5–9] . Recent studies have suggested that the spread of a number of different misfolded protein species between distinct brain regions occurs in a “prion-like” mechanism and is thought to determine specific brain regions that undergo neurodegeneration ( reviewed in [10 , 11] ) . For example , in Alzheimer disease , the initial detection of tau-neurofibrillary tangles in the locus correleus and entorhinal cortex spreads in a pattern resembling interconnected brain regions [12] , which correlates with cognitive decline in patients . Initial detection of Amyloid-beta ( Aβ ) is in the neocortex before spreading to allocortex and subsequently to subcortical regions in a pattern that broadly corresponds to anatomical connections . These studies can lead to the conclusion that accumulation , spread , and distribution of misfolded proteins predict regions that ultimately undergo neurodegeneration and thus define disease outcome . However , a number of findings question the direct relationship between protein misfolding and neurodegeneration . For example , misfolded PrP accumulates in the brain , in some situations unaccompanied by other typical neuropathological changes or any clinical signs of disease [13–16] . This relationship is not clear in the other protein misfolding diseases because the accumulation of Aβ as amyloid plaques , for example , has been detected in the brains of cognitively normal , aged individuals [17] . Furthermore , the detection of Aβ in Alzheimer disease patients does not always correspond to the anatomical regions accumulating neurofibrillary tangle lesions , which , as mentioned above , correlate strongly with cognitive decline [12] . Taken together , this raises the question as to the relationship between the appearance of misfolded protein aggregates and neurodegeneration . Exquisitely sensitive methods have been developed for detection of misfolded PrP [18–20] . These methods test whether samples contain misfolded PrP isoforms , defined by their ability to convert recombinant PrP to abnormal isoforms . The products are detectable using fluorescent amyloid fibril-binding compounds , such as thioflavin-T ( ThT ) [18 , 19] , therefore testing the capability of a sample to act as a “prion seed . ” This therefore represents the detection of only those misfolded PrP isoforms that can seed the conversion of normally folded PrP to misfolded isoforms that may be just a portion of the total misfolded PrP present . Although seeding activity is unlikely to be the only important characteristic of pathological forms of PrP , it is an important feature in as much as it reflects the fundamental principle of the self-propagating potential of prions . The increased sensitivity of this assay , termed the real-time quaking-induced conversion ( RT-QuIC ) assay , allows a novel approach to study the role of misfolded protein , which can act as seeds , in relation to microglial and astrocytic responses and neurodegeneration . Moreover , the prion models allow precise time course studies to be conducted to assess the presence of misfolded protein in specific regions of the brain from initiation of disease through to neurodegeneration . We have used the RT-QuIC assay system to examine the distribution of prion seeds in selected regions of the brain in a murine model of prion disease and compared this with the spread of misfolded PrP using other detection systems such as immunohistochemistry ( IHC ) , neurodegeneration , and glial cell responses . We have established that the misfolded protein alone is insufficient for neurodegeneration , and a complex and heterogeneous microglia response is associated with disease .
In this study , the GSS/101LL model of prion disease was used [14] . This involves an intracerebral ( i . c . ) inoculation of a human Gerstmann-Sträussler-Scheinker ( GSS ) brain homogenate into mice that have a proline to leucine alteration at codon 101 of the murine PrP—henceforth termed GSS/101LL . For controls , a normal brain homogenate ( NBH ) was inoculated i . c . into age-matched mice ( NBH/101LL ) . Animals were assessed daily for 2 wk by animal staff independent to this study after inoculum injection to ensure condition and health of mice did not deteriorate . The important feature of the GSS/101LL model for this study is the distinct and restricted regions of pathology in the brain at the terminal stage of disease at 291 ± 5 . 3 days post inoculation ( dpi ) . To define the earliest detectable accumulation of misfolded PrP and its relationship to neurodegeneration , serial sections were taken throughout the brain of GSS/101LL and NBH/101LL mice at several points within the incubation period: 150 , 200 , 220 , and 240 dpi , and terminal stage ( 291 ± 5 . 3 dpi ) . Granular deposits of misfolded PrP were detected in the midbrain at 150 dpi by IHC ( Fig 1 ) . Specifically , the staining observed is associated with the interpeduncular nuclei and substantia nigra , pars compacta ( SNc ) . No staining could be observed in any other brain region ( Fig 1 ) ; therefore , the initial IHC-detectable misfolded PrP targets specific midbrain nuclei . At 200 dpi , IHC-detectable misfolded PrP remained restricted only to midbrain nuclei . At later time-points ( >220 dpi ) , additional IHC-detectable PrP could also be detected in brain stem regions such as the medial and dorsal raphe nuclei in GSS/101LL animals . No IHC-detectable PrP was observed in any other brain region at this time-point . At clinical onset of disease , IHC detection of misfolded PrP could be observed restricted to specific neuronal populations in three major brain regions: the midbrain , brain stem , and thalamus ( Fig 1 ) . No IHC-detectable misfolded PrP was observed in other brain regions . We hypothesised that misfolded prion seeds would also be restricted to the specific brain regions associated with IHC-detectable misfolded PrP and that these brain regions would specifically undergo neurodegeneration , whereas those with no immunopositive PrP deposits would contain no prion seeds and remain free of neurodegeneration . In order to test this hypothesis , four brain regions were assessed for the presence of prion seeds as defined by their ability to act as seeds in the RT-QuIC assay and generate a ThT-positive signal . These were two IHC-positive regions the brain stem ( between Bregma -6 to -8 ) and thalamus ( between Bregma -1 to -3 ) and two IHC-negative regions the cerebellum ( between Bregma -6 to -8 ) and cerebral cortex ( between Bregma -1 to -3 ) ( henceforth referred to as cerebellum , brain stem , thalamus , and cortex ) . All of these brain regions from GSS/101LL mice , when used in the RT-QuIC assay , elicited an increased ThT fluorescence not observed in uninfected NBH controls ( Fig 2A ) . Thus , the increased level of fibril formation was specific to prion infection and demonstrated the presence of prion seeds in each brain region tested . To assess whether the prion seeds detected in these brain regions represented a protease-resistant conformational rearrangement of PrP [21] , proteinase K ( PK ) digestion was performed on all samples of brain regions prior to their inclusion in the RT-QuIC assay . In all regions from prion-infected brains exposed to PK , prion seeds remained detectable , but PK-resistant prion seeds were not observed in any region of age-matched uninfected NBH controls ( Fig 2B ) . To confirm the widespread appearance of prion seeds , we have tested these same brain regions that gave positive RT-QuIC detection in a different assay: the protein misfolded cyclic amplification ( PMCA ) assay [22] . We show that , similar to RT-QuIC , all brain regions from GSS/101LL tested were capable of seeding the PMCA reaction ( S1 Fig ) , conclusively demonstrating the widespread accumulation of misfolded PrP , which can act as prion seeds in all brain regions tested in GSS/101LL mice . We performed an IHC analysis on the four regions from the prion-infected brains that were positive for prion seeds using the RT-QuIC ( Fig 2; brain stem , thalamus , cortex , and cerebellum ) . Previous studies have demonstrated morphological changes associated with microglial activation and astrogliosis as valuable histological markers for early pathology , as both occur early in the course of disease before other early pathological changes as observed via histology , such as synaptic degeneration [23–27] . At clinical disease stage , we observe activated microglial and astrocyte glial cell responses . These are characterised by an up-regulation of glial-fibrillary acidic protein ( GFAP ) , indicative of reactive astrocytes , or the hypertrophy of microglial cell bodies and thickening of microglial cell processes detected with Ionised calcium-binding adapter molecule 1 ( Iba1 ) indicative of activated microglia . These glial cell responses were observed specifically restricted to brain stem and thalamus in GSS/101LL animals ( Fig 3 ) but were not seen in cortex or cerebellum of GSS/101LL animals or in any NBH/101LL brain region ( Fig 3 ) . Midbrain neurons were assessed by their expression pattern of tyrosine hydroxylase , and no change in its staining pattern could be detected until clinical onset of disease ( Fig 4A ) , at which point significant cell loss was observed ( Fig 4E ) . Cell loss was also quantified in specific neuronal populations in the brain stem and cortex and showed substantial cell loss in the brain stem but not in the cortex ( Fig 4E ) . Brain stem , thalamus , cortex , and cerebellum neurons were assessed using antibodies against microtubule associated protein 2 ( MAP2 ) , isoforms a+b . No change in MAP2 staining could be observed in cerebellum or cortex neurons at clinical stages of disease ( Fig 4B and 4F ) . Substantial changes in MAP2 staining were observed in brain stem nuclei , such as the gigantocellular reticular nuclei ( Fig 4B , 4C and 4F ) . This encompassed a general loss in precise cell body–associated staining compared to region- and age-matched NBH/101LL controls , which is indicative of neuronal degeneration in GSS/101LL brain stem . A reduction of MAP2 staining associated with dendritic processes was observed in the thalamus at clinical stages of disease , which was particularly prevalent in the ventral-medial parts of the thalamus ( Fig 4B and 4F ) . To further assess specific neuronal populations of the cerebellum , antibodies specific to the calcium-binding protein parvalbumin were used . Parvalbumin is highly expressed in Purkinje and stellate and basket neurons of the cerebellum and has been shown to be lost as a result of pathology in the cerebellum [28] . No change in the pattern of staining or intensity was observed in the GSS/101LL cerebellum compared to uninfected controls , even at clinical stages of disease ( Fig 4D ) . Finally , we analysed the presence of spongiform vacuolation of the neuronal parenchyma , which is a pathology commonly associated with prion disease and is quantified using well-established protocols , which we followed in our study [29] . We observed very low levels of vacuolation in most areas of the brain that were analysed , with the highest levels of vacuolation being observed in the medulla ( brain stem ) and negligible levels found in cerebellum or cortical regions ( S2 Fig ) . Taken together , these data show that neurodegeneration and morphological glial cell responses are restricted to the brain stem and thalamus , whereas cortex and cerebellum appear to remain pathologically “unaffected , ” even at clinical stages of disease . Several arguments could be made to explain the apparent lack of neurodegeneration in regions exposed to prion seeds , such as the following: ( i ) the amount of prion seeds may vary between regions undergoing neurodegeneration and regions that appear unaffected; ( ii ) tissue undergoing neurodegeneration may have been exposed to prion seeds for comparatively longer periods of time than regions that remain unaffected; or ( iii ) prion seeds are necessary , but not sufficient , for neurodegeneration . To address scenario ( i ) , we ran the RT-QuIC assay with decreasing concentrations of homogenate from each brain region to determine at what dilution detection of prion seeds is lost . Prion seeds were observed to support generation of a ThT signal at a concentration of at least 10−5 of original brain homogenate , or 0 . 001% ( w/v ) of original sample mass ( Fig 5A ) . Thalamus , cerebellum , and cortex all showed an increased ThT fluorescence at a concentration of 10−6 ( 0 . 0001% w/v of original sample mass ) . These data show no relationship between quantity of detectable prion seeds and neurodegeneration , as defined in Figs 3 and 4 . To address scenario ( ii ) , we tested for prion seeds at all time-points used in this study ( 150 , 200 , 220 , and 240 dpi and terminal illness ) . At 200 dpi onwards , an increased ThT fluorescence was observed in brain stem and thalamus , which eventually succumb to neurodegeneration ( Fig 5B ) , and in cerebellum , which does not undergo neurodegeneration ( Fig 5B ) . This is reaffirmed at later time-points , with increased ThT fluorescence also observed in GSS/101LL cortex samples ( Fig 5B ) . The data show that detectable prion seeds become widespread at relatively early but specific stages of disease progression but appear not to be associated with resilience or susceptibility of a particular brain region to neurodegeneration . A transcriptomics analysis was performed on the four brain regions to help define host responses to misfolded PrP in the presence or absence of neurodegeneration . Mice were separated into three groups , with two mice per-group; RNA was extracted from individual brain regions , and each group was pooled together . To interpret the microarray data , we used the network analysis tool BioLayout Express3D [30] in combination with statistical methods . Following data normalisation , the software calculates a pairwise Pearson correlation matrix comparing the expression profile of each transcript represented on the array to the expression profile of all other transcripts . All correlations above a user-defined threshold are used to construct a network graph visualised in 3-D space . When groups of genes are similarly expressed , they are tightly correlated and give rise to highly connected cliques within the network . This structure is used by the Markov clustering algorithm to divide the graph into clusters of co-expressed transcripts [31] . The network topography of our data produced three groups of genes ( components A–C ) , which do not share direct correlation between one another . Within each component , there are several distinct clusters of highly correlated genes ( Fig 6A ) . The major clusters of genes in component A ( e . g . , clusters 1 , 3 , and 6 ) show differential gene expression between brain regions but have no obvious relationship to disease ( Fig 6 ) and therefore were not examined further . However , components B and C both contain disease-specific clusters ( Fig 6 ) . Component B contained the largest group of disease-associated genes , and further analysis concentrated on this group . Data from component B were filtered stringently by first removing transcripts not annotated to known or predicted protein coding regions of the genome . Annotated genes were then filtered by Mann–Whitney U test ( p ≤ 0 . 05 ) to define significantly altered genes by comparison of GSS/101LL brain regions to their respective NBH/101LL-matched control brain regions . Furthermore , of those statistically significant genes , only those that exhibited a >1 . 5-fold change between control and disease were analysed further . This group of genes exhibited an increase in gene expression across all brain regions , with the greatest increases observed in regions of neurodegeneration ( brain stem and thalamus ) and lower but significant increases in gene expression in brain regions that do not show neurodegeneration ( cerebellum and cortex ) ( Fig 7A ) . Within component B , only a few significant gene expression changes were noted in the cortex ( 11 genes ) compared with a much larger number in the other regions , and therefore the cortex was not included in further analyses in the current study . To determine the cellular origins of component B , the filtered gene list was overlaid onto previously published expression datasets of isolated cell populations . This included a variety of murine myeloid populations ( including microglia ) [32] and brain cell populations ( neurons and glial cells ) [33] . Using these datasets , in many cases we could attribute expression of specific genes to cell type of origin ( Fig 7B ) . The majority of up-regulated genes in the brain stem , thalamus , and cerebellum could be attributed to cells from a myeloid lineage and therefore most likely derived from the resident microglia population ( Fig 7C ) . Thus , the microglial response appears to be an important host response in disease , with expression changes evident in regions with and without IHC detectable glial cell changes ( Fig 3A ) . The specific genes attributed to microglia were sorted according to gene overlap ( Fig 7D ) , showing distinct groups of genes that have differential expression in specific brain regions . To broaden our analysis of this pattern of distinct groups of genes , we next examined the total microarray dataset , filtered according to the same statistical and fold-change parameters outlined above ( p ≤ 0 . 05; >1 . 5-fold change ) and attributed to predicted cellular origin ( S1 Table ) . Gene ontology ( GO ) enrichment of the microglial gene lists was assessed in two groups: ( 1 ) genes not directly associated with neurodegeneration ( i . e . , all microglial genes identified as up-regulated in the cerebellum ) and ( 2 ) genes associated with neurodegeneration ( i . e . , all microglial genes up-regulated in brain stem and thalamus but not differentially expressed in cerebellum ) . Amongst the microglial genes up-regulated but not directly associated with neurodegeneration , the major GO terms relate to metabolism and regulation of homeostasis ( S2 Table ) . This demonstrates a response of microglia , which they are known to exhibit , but has never , to our knowledge , been observed in chronic neurodegenerative diseases . In genes up-regulated in brain regions undergoing neurodegeneration , significantly enriched GO terms are related to activation of the innate immune response , complement activation , and antigen processing and presentation ( S3 Table ) , which concurs with previous data , suggesting the innate immune response as an important and necessary component of the neurodegenerative process .
We have demonstrated that , during the evolution of prion disease , prion seeds are widespread , accumulating in brain regions both with and without overt neurodegeneration . We show that the quantity or time of exposure to prion seeds was not directly related to the development of neuropathology . Therefore , these pathological differences are not due to a lack of spread and accumulation of prion seeds , indicating that prion seed accumulation is not itself sufficient for neurodegeneration , at least within the lifetime of the mouse . This extensive and widespread distribution of prion seeds has not been previously described , and the highly sensitive assay developed for the detection of misfolded PrP has allowed us to further investigate our understanding of the relationship of misfolded PrP and neurodegeneration . In this study , we performed an i . c . inoculation that results in widespread dispersion of the homogenate and its rapid degradation and clearance [34] . RT-QuIC detection of prion seeds from GSS/101LL mice was not increased over NBH/101LL age-matched controls at 150 dpi; therefore , the RT-QuIC assay is detecting specific murine replication and accumulation of misfolded protein at later time-points and not the initial human inoculum . As a result , the RT-QuIC provides a highly sensitive assay system for examining the prion seed accumulation and its interactions over the course of disease . We show that detection of misfolded PrP using IHC is restricted to specific brain regions undergoing neurodegeneration and associated morphological glial responses . Specific brain regions , for example , the thalamus , only accumulate IHC-positive misfolded PrP at late stages of disease ( 291 ± 5 . 3 dpi ) , whereas prion seeds are detected three months earlier in this region ( 200 dpi ) . Some authors have asserted that misfolded protein aggregates are responsible for causing neurodegenerative diseases and that the targeting of neurodegeneration is influenced by the restricted distribution of misfolded protein between brain regions [35 , 36] . These studies rely on detection of misfolded proteins using techniques such as IHC or basic histology staining with silver staining ( reviewed in [10] ) . In this study , by using an alternative and highly sensitive method for the detection of misfolded PrP , which act as prion seeds , we find the distribution of misfolded PrP seeding material does not predict the regions of neurodegeneration . IHC has been used extensively in prion disease as a terminal marker of the pathology in the brain . Although , in time course studies , PrP IHC can be detected earlier in the disease process , we believe the more sensitive assay systems ( RT-QuIC and PMCA ) are capable of detecting the earlier events that ultimately lead to the IHC-detectable protein . Thus , the sensitive assays both demonstrate that the pathways of protein misfolding are activated in all brain regions . Although uncertainties remain about the exact nature and toxicity of the abnormal PrP that is detected by IHC , the same can be said of RT-QuIC and PMCA . The use of sensitive techniques has uncovered the novel finding that these pathological differences are not due to a lack of spread and accumulation of prion seeds; rather , they correlate with local cellular differences in the responses to those seeds that govern whether those seeds lead to bulk PrP accumulation and pathological lesions . Although seeding activity is not likely to be the only important characteristic of pathological forms of PrP , it is likely to be one of the most important features in as much as it reflects the self-propagating potential of prions . The question remains as to why only specific regions of the brain succumb to neurodegeneration . Previous studies have shown that the same prion strain inoculated into different murine genetic backgrounds , which exhibit comparable quantities of detectable prion seeds , can result in different sized aggregates of misfolded protein [37] . Just as large prion particles have been shown to have lower infectivity per unit mass of PrP , i . e . , specific infectivity [38] , so might larger prion seed particles be expected to have lower seeding activity per mass than smaller particles . If so , then two regions could have the same seed concentration but divergent total loads of abnormal PrP . In addition , if the seeds are more clustered in one brain region than another , they would also be more likely to be detectable by IHC . If different types of aggregates are accumulating across brain regions , this is likely to impact on production and clearance of aggregates in each region , with the type of misfolded protein being the driver of the response . An alternative explanation is that different regions of the brain have differing abilities to respond to the same aggregates due to intrinsic variability in gene expression of cells present . A specific example would be the underlying differences in microglial gene expression between brain regions in healthy young and aged brains , which might underlie the specificity of microglial response during disease [39] . Thus , regional differences in protein aggregate and/or cell signalling could influence targeting of neurodegeneration . We analysed transcriptional responses in each brain region and showed that the major response in each brain region tested could be attributed to microglia , similar to previous microarray studies on prion disease cases in human and animals [40–44] . Functional annotation of the microglial response to misfolded protein , but in the absence of neurodegeneration , revealed a large proportion of genes related to regulation of homeostasis . This highlights a disease-specific response of microglia up-regulating a core set of genes , the key function of which may be to respond to changes associated with protein misfolding . However , differential activation of the innate immune response is also observed in brain regions undergoing neurodegeneration . These data suggest at least two distinct microglial responses occurring during disease . In one , microglia respond to either the protein misfolding and/or the consequences of protein misfolding by attempting to maintain homeostasis . In the other , in regions of neurodegeneration , microglia up-regulate an innate immune response , which includes pro- and anti-inflammatory genes , complement activation , and antigen processing and presentation . The data presented here demonstrate that the seeding and distribution of misfolded protein are more widespread at earlier time-points than previously described . Importantly , this seeding occurs in brain regions that do not undergo neurodegeneration . We show that there is a significant host response in all brain regions examined either in the presence or absence of neurodegeneration . These responses are predominantly associated with microglial cells , and functional annotation demonstrates distinct responses from these cells between brain regions accumulating misfolded protein seeds either in the presence or absence of neurodegeneration . Therefore , microglia appear to be on the one hand attempting to restore homeostasis introduced by misfolded proteins that act as seeds , but in regions of neurodegeneration , the microglia enter into a cycle of innate immune activation . Previous studies have demonstrated that , by altering specific aspects of the innate immune response by KO of specific pro- or anti-inflammatory genes , the severity of disease can be altered [45–51] . Furthermore , by inhibiting microglial proliferation during disease , and thus reducing the number of microglia exhibiting an activated innate immune response , other studies have shown a prolongation of incubation period in prion disease [52] . Although it remains unclear what signals microglial cells are specifically responding to , specific misfolded protein isoforms or neurodegeneration , data from previous studies together with data presented here could point to a role for the activation of the innate immune response in defining severity of disease , which may contribute to the destruction of cells in specific brain regions . Therefore , manipulation of the activation state of microglia could represent a therapeutic target for suppressing neurodegeneration during disease ( Fig 8 ) . In summary , we find that the pathological differences between brain regions during chronic neurodegeneration are not due to a lack of spread and accumulation of prion seeds; rather , they correlate with local cellular differences in the responses to those seeds that govern whether those seeds lead to bulk PrP accumulation and pathological lesions . A combined approach characterising the host responses to the misfolded protein accumulation and distribution , therefore , is required to more precisely determine the relevance of particular misfolded protein species to disease outcome .
All experiments were approved by the Roslin Institute Ethical Review committee and in accordance with the United Kingdom Home Office Regulations ( Animals [Scientific Procedures] Act of 1986 ) . Ethical consent for the use of human materials for research was obtained and approved by the Lothian National Health Service Board Research Ethics Committee ( reference: 2000/4/157 ) . The “101LL” transgenic mouse line contains an amino acid alteration from proline to leucine at codon 101 of the 129/Ola ( 129/OlaHsd , Harlan , UK ) murine prion gene by gene-targeting [53] . 129/Ola mice homozygous for the targeted allele were crossed and bred over multiple generations to generate progeny homozygous for the PrnpP101L ( 101LL ) allele , which can be used for experimental purposes . All mice were bred at the Roslin Institute under a temperature-controlled , 12 h light/12 h dark cycle . Mice were housed with wood chip bedding and a wood chew stick for environmental enrichment . Food and water were available ad libitum . To confirm the presence of the targeted allele alteration , mice were genotyped before and after studies . DNA extraction was performed using DNeasy Blood and Tissue kit ( Qiagen ) on ear clips . Presence of the 101LL mutation was determined by PCR analysis using a primers specific for positions 107–130 and 871–848 of the PrP gene; the 5′ primer used was ( 5′-ATGGCGAACCTTGGCTACTGGCTG–3′; DDBJ/EMBL/GenBank accession number M18070 ) , and the 3′ primer used was ( 5′–TCATCCCACGATCAGGAAGATGAG–3′; DDBJ/EMBL/GenBank accession number M18070 ) . PCR was set up using a Type-it mutation detection kit ( Qiagen ) . PCR cycle conditions were as follows: 94°C for 3 min , followed by 30 cycles of 94°C for 30 sec , 62°C for 30 sec , and 72°C for 1 min . A final extension phase of 72°C for 10 min was performed , and samples were subsequently stored at 4°C . The induced alteration of proline to leucine in 101LL mice forms a DdeI restriction enzyme cut site , which allows for identification of 101LL homozygous mice . The PCR product was incubated for 3 h at 37°C with DdeI restriction enzyme ( Promega ) before being run on a 1 . 5% agarose gel ( Invitrogen ) and imaged . Human frontal cortex from a patient with GSS carrying the PRNP-P102L mutation was obtained from the CJD Brain and Tissue Bank ( Edinburgh Brain and Tissue Banks ) . The tissue was homogenised in 0 . 9% sterile saline ( Martindale Pharmaceuticals , UK ) to a 1% ( w/v ) homogenate , and mice were injected i . c . into the right hemisphere with 20 μL of 1% homogenate under anaesthesia . Due to the difficulty in obtaining uninfected human tissue as a negative control , an i . c . injection of 20 μL uninfected hamster brain homogenate ( 1% w/v ) was performed instead . These protocols were used to replicate those of previous studies using this model of prion disease [14] . These form the GSS/101LL and NBH/101LL groups , respectively , which are referred to throughout the text . All animals were aged-matched and were injected between 10–14 wk of age . A time-course analysis was set up in this study , ranging from 150 dpi ( GSS/101LL n = 12; NBH/101LL n = 12 ) , 200 dpi ( GSS/101LL n = 12; NBH/101LL n = 12 ) , 220 dpi ( GSS/101LL n = 6; NBH/101LL n = 6 ) , 240 dpi ( GSS/101LL n = 6; NBH/101LL n = 6 ) , and upon clinical symptoms of disease ( GSS/101LL n = 9; NBH/101LL n = 12 ) . All animals were assigned to groups and assessed daily from around 100 dpi by an independent researcher to this study using parameters that have been previously described [54] . All mice were killed by CO2 asphyxiation according to Schedule 1 of the Animals ( Scientific Procedures ) Act of 1986 . The brains of 101LL ( GSS/101LL and NBH/101LL ) mice were removed and halved along the midline between the right and left brain hemispheres . The left hemisphere was flash frozen in liquid nitrogen and stored at -80°C for later use . The right hemisphere , which contains the injection site , was immersion fixed in 10% formal saline for 48 h before being exposed to 98% formic acid for 90 min to minimise the infectious titre of the sample . The tissue was subsequently re-washed in 10% formal saline for at least 24 h to remove residual formic acid . The tissue was cut into five coronal sections for vacuolation scoring [29] , which encompasses nine grey matter ( GM ) regions ( medulla , cerebellum , superior colliculus , hypothalamus , thalamus , hippocampus , septum , rerospinal cortex , and cingulate and motor cortices ) . Tissue was then paraffin embedded . Haematoxylin and eosin ( H&E ) staining of 6-μm sections were taken for vacuolation ( spongiform ) severity scoring , which was performed blind by a researcher independent to this study . For further pathological analysis , serial 10-μm sections were cut through the brain . Paraffin-embedded tissue was dewaxed by immersing in xylene and re-hydrated through a series of decreasing alcohol concentrations at room temperature . For PrP immunostaining , slides were immersed in citric buffer ( 0 . 1 M citric acid , 0 . 1 M Sodium Citrate , pH 6 . 4 ) and autoclaved at 121°C for 15 min . Slides were then cooled in running water for 5 min before immersing in 98% formic acid for 10 min . Subsequently , slides were thoroughly washed in running water for 20 min . To block for endogenous peroxidase , slides were immersed in 1% H2O2/methanol before washing in running water for 5 min followed by PBS/1% BSA wash buffer for 5 min . Sections were subsequently incubated with Normal Goat Serum ( Stratech ) for 20 min before application of either BH1 ( [55] used at 0 . 02 μg/mL ) or 6H4 ( Prionics used at 3 μg/mL ) anti-PrP antibodies . The primary antibodies were incubated overnight before washing with PBS/1% BSA wash buffer . Goat anti-mouse secondary antibody ( Jackson ImmunoResearch ) was applied for 1 h , washed in PBS/1% BSA , and ABC kit ( Vector Laboratories ) was applied for 30 min then washed . Peroxidase activity was visualised using diaminobenzidine ( DAB ) :H2O2 and slides were counterstained in Harris’ haematoxylin . For the immunohistochemical detection of other proteins , sections were either given no antigen-retrieval step ( e . g . , GFAP ) or were immersed in citric buffer ( pH 6 . 0 ) and autoclaved for 15 min at 121°C . The protocol replicated that of PrP detection , with exception of 10 min incubation in 98% formic acid . Ten serial sections from three GSS/101LL and three NBH/101LL mice at clinical stage pathology from three anatomically distinct brain regions , the gigantocellular reticular ( Gi ) nucleus of the brain stem , the retrosplenial granular cortex ( RSGc ) , and the SNc were cut at anatomically equivalent regions , approximately Bregma -6 mm ( brain stem ) , Bregma -1 . 8 mm ( cortex ) , and Bregma -3 mm ( midbrain ) . Digital images were captured at x20 magnification using a Nikon E800 bright-field microscope . A calibrated grid ( 100 x 100 μm ) was overlaid onto each image , and the number of cell bodies was determined by counting . The total number of cells counted in GSS/101LL brain regions was then normalised to the region matched NBH/101LL controls and presented as the percentage proportion of cells present in GSS/101LL compared to NBH/101LL . DAB quantification was performed on ten serial low-magnification images ( x4 ) of brain stem , thalamus , cerebellum , and cortex , examining the total quantity of positive staining in each section and averaging across the serial sections . DAB quantification was performed using Image J and the colour deconvolution plug-in . Prnp DNA sequences encoding Syrian hamster residues 23 to 137 followed by sheep residues 141 to 234 of the R154 Q171 polymorph ( accession number AY907689 ) ( Ham/Shp chimeric PrP ) were prepared according to previous methods [20] . RT-QuIC buffer composition was as follows: 10 mM phosphate buffer ( pH 7 . 4 ) , 130 mM NaCl , 10 μM Thioflavin T ( ThT ) , 10 μM EDTA , and a final concentration of 0 . 1 mg/mL recPrP . Ninety-eight μL of this master mix were loaded onto black 96-well clear bottom plates ( Nalgene Nunc International ) . Correspondingly , 2 μL of diluted brain homogenate were added to each well for a final reaction volume of 100 μL . Each sample was run in triplicate , and included in each reaction were standard positive controls ( 79A murine scrapie prion disease at a concentration of 0 . 1% [w/v] of the original brain weight ) and negative controls ( uninfected murine brains at a concentration of 0 . 1% [w/v] of the original brain weight and RT-QuIC master mix–only samples ) . Plates were then sealed with a plate sealer film ( Nalgene Nunc International ) . A PolarSTAR Omega ( BMG Labtech ) plate reader was used to incubate the samples at 42°C for 60 h with cycles of 1 min rest and 1 min 700 rpm double orbital shake . ThT fluorescence was then measured ( 450 nm excitation/480 emission ) every 15 min during the 60 h incubation . PMCA experiments were performed following the amplification procedure described previously [56] . Briefly , aliquots of PMCA substrate , derived from transgenic mice P101L brains , were incubated with PMCA seeds in 0 . 2-mL PCR tubes to a final volume of 120 μL . Serial cycles of sonication and incubation were performed for 48 h at 37°C , comprising 20 sec of sonication ( at an amplitude of 38 , wattage: 278–299 ) followed by 29 min 40 sec of incubation for each cycle ( Qsonica , model Q-700 ) . Detection of PrPres was assessed by PK treatment and western blotting methodology previously described [56] . Brain stem , thalamus , cerebellum , and cortex were dissected from 6 GSS/101LL animals at terminal illness and from 6 NBH/101LL age-matched mice . The tissue was weighed then homogenised in 1 mL Trizol ( Life Technologies ) per 100 mg of tissue . The homogenate was thus centrifuged at 11 , 500x g for 10 min at 4°C . The pellet was discarded . Chloroform ( 0 . 2 mL/mL Trizol ) was then added to cause phase separation , whereby protein constitutes the organic phase , DNA the interphase , and RNA the aqueous phase after a 11 , 500x g centrifugal step for 15 min at 4°C . The aqueous phase was transferred to fresh RNase-free tubes ( Life Technologies ) . Isopropanol ( 0 . 5 mL/mL Trizol ) was added and incubated for 10 min at room temperature . The RNA sample was then centrifuged for 10 min at 11 , 500x g at 4°C . The RNA pellet was washed in 75% ethanol ( 1 mL/mL Trizol ) then centrifuged for 5 min at 8 , 000x g at 4°C . The RNA pellet was then resuspended in Nuclease Free dH2O ( Life Technologies ) , aliquoted , and stored at -80°C . RNA was extracted from brain regions of individual animals . To remove non-disease–specific inter-animal variation , each sample consisted of a pool of two animals for an individual brain region . This resulted in 24 samples . Samples were subjected to transcriptomics analyses using Mouse Gene 2 . 0 array run on a GeneTitan instrument ( Affymetrix ) by Edinburgh Genomics ( www . genomics . ed . ac . uk ) . Data were normalised using Affymetrix Expression Console software and saved as an ‘ . expression’ file containing a unique identifier for each transcript ( gene annotation concatenated to probe ID ) . In subsequent columns , gene and GO annotations were included for assigning class-sets for the analysis of information contained in the network graph , followed by the RMA-normalised raw data . The ‘ . expression’ file is subsequently loaded into the network analysis tool BioLayout Express3D [30] , in which a pairwise Pearson correlation matrix is calculated as a measure of similarity between transcript profiles . A network was created using a threshold of r ≥ 0 . 95 and layout performed using modified Fruchterman–Rheingold algorithm [57] . In this paradigm , each node represents a single transcript , which are connected by weighted , undirected edges representing correlations above the threshold . Groups of highly correlated transcripts were then “clustered” using the Markov cluster algorithm [31] . To confirm the changed expression of transcripts between GSS and NBH/101LL brain regions , data were filtered to include only transcripts annotated to known or predicted genes . Each transcript was then tested for statistical significance . As data were not of equal variance , as determined by Kolmogorov–Smirnov test , nonparametric Mann–Whitney U test ( p < 0 . 05 ) was performed to assess the change of each transcript intensity in a GSS/101LL brain region compared to its respective NBH/101LL control brain region . Data were then further filtered to include only those transcripts that showed a >1 . 5-fold change . Further analyses were performed to include a “total” gene list for each GSS/101LL brain region using the same statistical parameters outlined above . Genes in the resulting lists were then assigned likely cell types of origin by inspecting their profile on an expression atlas of cell types on published microarray analyses [32 , 33] , including various types neuronal , glial , and myeloid cell populations . These data were loaded into BioLayout Express3D at a Pearson correlation cut-off of ( r ≥ 0 . 7 ) , and genes were assigned a putative cell type of origin based upon inspection of their expression pattern across the cell atlas . Genes attributable to a microglia origin up-regulated in all regions , regardless of the presence of pathology , were subjected to GO enrichment analysis . All the primary microarray data have been deposited in GEO: GSE74079 . | Normal brain function requires tight regulation of protein folding; when this goes wrong , proteins can fold into abnormal conformations , which have severe impacts on brain performance , leading to conditions like dementia . Previous studies show that abnormally folded proteins are found in restricted parts of the brain , and neuronal cells in these specific brain regions have been shown to undergo degeneration . Recent technological advances have enhanced the detection of abnormally folded prion protein ( PrP ) during disease; we used these technologies to test whether distribution of abnormally folded proteins is indeed restricted to regions of the brain undergoing degeneration . Surprisingly , we observed abnormally folded proteins throughout the brain , demonstrating that these proteins can accumulate in parts of the brain that do not show degeneration . Thus , the distribution of abnormally folded protein , by itself , is not sufficient for neuronal degeneration . In addition , we found that microglia ( one of the nonneuronal cell types in the brain ) change their response during prion disease in two different ways; one response is associated with resilient brain regions , and the second , an inflammatory response is associated with regions susceptible to degeneration . Thus , the microglial response appears to be important in determining the outcome of disease . | [
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] | [
"medicine",
"and",
"health",
"sciences",
"brain",
"neuroscience",
"microglial",
"cells",
"cerebellum",
"infectious",
"diseases",
"zoonoses",
"animal",
"cells",
"gene",
"expression",
"glial",
"cells",
"thalamus",
"brain",
"diseases",
"brainstem",
"cellular",
"neuroscienc... | 2016 | Distribution of Misfolded Prion Protein Seeding Activity Alone Does Not Predict Regions of Neurodegeneration |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.